Is Undiagnosed CKD Stages 1–3 Sufficiently Prevalent?

Determination of how many individuals need to be screened to identify each new case of CKD in the population overall and within high risk groups will be a function both of the prevalence of undiagnosed CKD in these groups and the frequency with which such patients already are tested for CKD in usual practice.

As described earlier, approximately 11.1 percent (22.4 million) of U.S. adults age 20 or older have CKD stages 1–3.⁸ This estimate is derived from the NHANES population by using the CKD-EPI formula to estimate GFR and the urine albumin-creatinine ratio to estimate kidney damage. Of individuals with CKD stages 1–3, half have increased albuminuria only (nearly all with microalbuminuria), one-third have decreased GFR only, and the remainder have both abnormalities. Of individuals with albuminuria, nearly 85 percent have microalbuminuria, with the remainder (approximately 1 percent of NHANES participants) having macroalbuminuria. In another population-based sample, prevalence of macroalbuminuria among adults aged 28 to 75 years was 0.6 percent.⁴¹ Compared with the overall population, prevalence of CKD stages 1–3 is higher among older adults, including 17.6 percent in those aged 60-69, and 44.4 percent among adults aged 70 years or older⁸ Also based on NHANES data, prevalence of CKD stages 1–3 is 39.0 percent in patients with diabetes, 27.8 percent in patients with hypertension, and 37.9 percent in those with cardiovascular disease⁴² (Table 1) Combining these risk factors, NHANES data have been used to stratify individuals into different groups with respect to their likelihood of having CKD⁹ (Tables 2 and 3). For example, only 5 percent of individuals less than 52 years old and without diabetes, hypertension, or obesity were estimated to have CKD compared with 68 percent of those aged 81 years or older.

Table 1

Percentage of U.S. adult population age 20 years or older with each stage of CKD, overall and within subgroups defined by age, gender, race, and comorbidities using the creatinine based CKD-Epi formula for estimating GFR.

Table 2

Sensitivity and specificity of different population characteristics for identifying individuals who would have one-time eGFR <60 ml/min/1.73 m²: using creatinine and CKD-Epi formula.

Table 3

Sensitivity and specificity of different population characteristics for identifying individuals who would have one-time UACR ≥30 mg/g.

Other data suggest that most individuals with CKD stages 1–3 are not clinically recognized to have this diagnosis. In one study, among patients with GFR <60 ml/min/1.73m², just 26.5 percent were documented to have a clinical diagnosis of CKD.⁴³ In 2008 data from the VA system, even in patients with CKD stages 3-5, only 33 percent had a provider-coded ICD-9 diagnosis for CKD.⁴⁴ Awareness of CKD appears even lower in patients. According to the CDC CKD Surveillance Project 2009 Report, among NHANES participants in 1999-2006, fewer than 5 percent with proteinuria and an estimated GFR ≥60 ml/min/1.73m² (based on a single measurement) reported being aware of having CKD, and only 7.5 percent of participants with a GFR between 30-59 ml/min/1.73m² were aware of having CKD.⁴⁴

Most patients without CKD, even those in high risk groups, do not appear to be undergoing CKD testing in usual clinical care. Based on 2007-2008 Medicare data, among patients without CKD who had diabetes, the annual probability of urine microalbumin testing was just over 30 percent.¹² In those without CKD who had hypertension, the annual probability of urine microalbumin testing was 4 percent. Based on 2004 Medicare data, among patients without CKD who had either diabetes or hypertension, the annual probability of serum creatinine measurement was less than 20 percent.⁴⁵

Is CKD Stages 1–3 Associated With Sufficient Adverse Health Consequences?

As described earlier, early stage CKD is usually asymptomatic. However, data from many studies indicate that a GFR 30-59 mL/min/1.73 m² (stage 3 CKD) is associated with an increased risk of mortality,^3,13 cardiovascular disease,¹⁴ fractures,¹⁵ bone loss,¹⁶ infections,¹⁷ cognitive impairment,¹⁸ and frailty. Similarly, albuminuria and proteinuria (stage 1–4 CKD) are associated with an increased risk of mortality,^3,20 ESRD,²¹ and cardiovascular disease,²² with risk increasing according to the severity of albuminuria or proteinuria. Further, the risk for adverse outcomes conferred by reduced GFR and increased albuminuria or proteinuria appear independent and multiplicative.^3,21

Are There Valid, Reliable, and Clinically Available CKD Screening Tests?

Serum creatinine is measured from a simple blood test. Formulas to estimate GFR are now automatically reported in many clinical labs from serum creatinine and are highly correlated (i.e., >0.9)⁴⁶ with direct GFR measurement based on urinary clearance of ¹²⁵I-iothalamate. At present, the Modification of Diet in Renal Disease (MDRD) formula is the one most commonly used in clinical practice. A large external validation study indicated that compared with measured GFR the CKD-EPI formula had a small median bias (measured GFR minus estimated GFR) of +/-4 ml/min/1.73m² or less at all levels of measured GFR.⁴⁷ This represents a significant improvement in accuracy compared with the MDRD formula for measured GFR ≥30 ml/min/1.73m², which is known to underestimate measured GFR above this level, particularly in individuals with GFR ≥60 ml/min/1.73m². However, the precision of both formulas are limited in that the percentage of their estimates that diverge by more than 30 percent from measured GFR exceeds 15 percent.⁸ Framed differently, the sensitivity and specificity of a one-time estimate of GFR <60 mL/min/1.73m² for detection of a one-time direct measurement of GFR <60 mL/min/1.73m² were 91 percent and 87 percent according to the CKD-EPI equation and 95 percent and 82 percent according to the MDRD Study equation.⁸ These data correspond to a false-positive rate of 13 percent and 18 percent for GFR estimation with CKD-EPI and MDRD, respectively. We did not identify studies that compared estimated GFR with directly measured GFR based on two or more measurements three or more months apart as would be consistent with the definition of CKD. It would be expected that when compared with persistently abnormal measured GFR, the false-positive rate of one-time estimated GFR would be higher.

There are many sources of variability in measurement of urinary albumin excretion. Intra-individual variability is high, with many published coefficients of variance estimates clustering around 30 to 50 percent.⁴⁸ Factors that can impact urinary albumin excretion include body position, exercise, and fever.⁴⁸ While most groups recommend use of spot tests and calculation of the urine albumin-creatinine ratio, methodology for its collection and for measurement of both urinary albumin and creatinine has yet to be standardized. Although these are additional sources of variation, they appear considerably smaller in magnitude than the intra-individual variability.^48-50 Impacted by these issues, among individuals with one-time microalbuminuria and GFR ≥60 ml/min/1.73m² in the NHANES study, only 63 percent had either microalbuminuria or macroalbuminuria on repeat testing two months later.⁵¹ Further, even in a diabetic population with persistent microalbuminuria, as defined by repeated UACR measurements during a 2-year period, regression of the microalbuminuria to normal occurred in 59 percent patients over a subsequent 6-year evaluation period.⁵²

Unfortunately, we did not identify any population-based studies that tested the sensitivity or specificity of one-time screening using both estimated GFR and albuminuria for diagnosis of CKD as defined by persistence of impaired GFR and/or albuminuria for at least 3 months (the current “gold standard”). We also did not identify any data on the validity and reliability of repeated screening for CKD.

Do Treatments for Screen-Detected CKD Patients Improve Important Clinical Outcomes?

We did not identify RCTs involving treatment of CKD patients identified through systematic screening, but did systematically review the RCT evidence on the effectiveness of treatments of CKD patients identified more generally in the Results section for Key Question 5.

Psychological Effects of Screening

We did not identify any studies that reported on the psychological effects of screening tests for CKD.

Adverse Physical Effects of Screening Tests and of Followup Tests To Evaluate Abnormal Screening Test

Phlebotomy required to measure serum creatinine may be associated with a small degree of bruising or discomfort. In a small number of patients, postscreening evaluation will include a renal biopsy, which has an associated risk of pain, bleeding, and infection.

Misclassification/False Positive Test for CKD

We did not identify any studies that reported on the effects of a false positive result from tests used to screen for CKD. False positive results may be common with tests for microalbuminuria. As described above, intra-individual variability in albuminuria is high. In one study, more than one-third of individuals with microalbuminuria and normal GFR on first testing regressed to normoalbuminuria on repeat testing two months later.⁵¹ Raising questions about the sufficiency of the requirement that albuminuria be persistent for at least 3 months to diagnose CKD, in a second study, 59 percent of individuals with persistent microalbuminuria over a 2 year period regressed to normal during a subsequent 6 year evaluation period.⁵² We did not identify any studies that reported the specificity of a single measurement of GFR estimated from serum creatinine for a diagnosis of CKD defined by abnormalities in kidney function or damage that persist for at least 3 months.

Labeling of an Individual With CKD

We did not identify any studies that reported on the effects of labeling an individual with CKD.

Increased Clinic Visits to Primary and/or Specialist Providers

We did not identify any studies that reported on the effect of CKD screening tests on subsequent patient visits to primary or specialist providers. However, to the extent that their provider is aware of it, individuals who have an abnormal result on CKD screening, seem likely to be seen more frequently in primary and specialty clinics. These visits may be for further evaluation to confirm the abnormal screening test, or providers may follow and treat these patients under the assumption that they have diagnosed CKD. According to recent U.S. Renal Data System (USRDS) data, in the year following a claim-documented CKD diagnosis, approximately 90 percent of individuals have at least one physician visit and 30 percent have a visit with a nephrologist.¹²

Adverse Effects Associated With Treatment

We systematically reviewed the RCT evidence on adverse effects of treatments of CKD patients in the Results section for Key Question 6.

Impact on Insurance Coverage

We did not identify any studies that reported the effects of being diagnosed with CKD on obtaining or keeping health insurance coverage.

Is Undiagnosed Worsening of Kidney Function and/or Damage Sufficiently Frequent in Patients With CKD Stages 1–3?

Determination of whether and how frequently individuals with CKD stages 1–3 need to be monitored to identify patients with CKD progression, overall and within high risk groups, will be a function both of the incidence of undiagnosed CKD progression in these groups, the incidence of CKD regression (e.g., to normoalbuminuria), and the frequency with which these patients already have their level of kidney function and/or damage tested in usual practice.

In patients with CKD, reported rates of CKD progression vary widely. Mean annual GFR decline may range from approximately 1 to >10 ml/min/1.73m².²⁷ Factors shown in at least some studies to predict faster decline include diabetes, proteinuria, increased blood pressure, older age, obesity, dyslipidemia, smoking, male gender, and etiology of primary kidney disease. The high intra-individual variation in albuminuria makes it harder to estimate rates at which albuminuria increases in CKD. However, in several RCTs that randomized individuals with diabetes and microalbuminuria to either ACEI or ARB versus placebo,^54-59 the average annual progression rate to macroalbuminuria was approximately 5 to 9 percent (Table 5). A lower annual conversion rate of 2.8 percent was reported in the United Kingdom Prospective Diabetes Study.⁶⁰ However, these estimates of progression in albuminuria from RCTs are limited both in that being derived from RCTs they may not be representative of all patients with microalbuminuria, and in that a substantial portion of individuals with microalbuminuria also will regress (i.e., to normoalbuminuria) over time.

Table 5

Rate of progression from microalbuminuria to macroalbuminuria.

Contrasted to the lower frequency of testing among individuals who do not carry a CKD diagnosis, most patients with CKD stages 1–3 appear to be undergoing at least some CKD testing in usual clinical care. Based on 2008 data, the annual probability that patients with CKD stages 1–3 receive serum creatinine testing is about 95 percent in the Medicare population and about 80 percent in a younger privately insured population.¹² By comparison, the annual probability that patients with CKD stages 1–3 get albuminuria measured is between 30 and 40 percent.

In Patients With CKD Stages 1–3, Is CKD Progression Associated With Sufficient Adverse Health Consequences?

As described earlier, data from many studies indicate that a GFR 30-59 mL/min/1.73 m² (stage 3 CKD) is associated with an increased risk of mortality,^3,13 cardiovascular disease,¹⁴ fractures,¹⁵ bone loss,¹⁶ infections,¹⁷ cognitive impairment,¹⁸ and frailty. Similarly, albuminuria and proteinuria are associated with an increased risk of mortality,^3,20 ESRD,²¹ and cardiovascular disease,²² with risk increasing according to the severity of albuminuria or proteinuria. Further, the risk for adverse outcomes conferred by reduced GFR and increased albuminuria or proteinuria is independent and multiplicative.^3,21

We did not identify studies that longitudinally recalibrated risk of adverse health consequences among individuals with CKD stages 1–3 as their CKD progressed. However, a large, recent meta-analysis of prospective cohort studies reported risk of all-cause and cardiovascular mortality for different strata defined by baseline eGFR and albuminuria as follows:³

• Within individuals who had albuminuria and GFR >60 ml/min/1.73m² (CKD stages 1–2):
  • Mortality risk was higher in those with macroalbuminuria than in those with microalbuminuria.
  • A lower GFR within this range was not associated with a higher mortality risk.
  • Mortality is increased for each lower level of eGFR below 60 ml/min/1.73m², higher for 45–59 (CKD stage 3), still higher for 30–44, and higher for GFR <30 ml/min/1.73m² (CKD stage 4).
• Within individuals with GFR <60 ml/min/1.73m2 (CKD stage 3):
  • Mortality risk is increased for each lower level of eGFR, lowest for 45–59, higher for 30–44, and higher for GFR <30 ml/min/1.73m2 (CKD stage 4).
  • Mortality risk is lowest in those without albuminuria, higher in those with microalbuminuria, and highest in those with macroalbuminuria.

Are There Valid, Reliable, and Clinically Available Tests To Monitor CKD Progression in Patients With CKD Stages 1–3?

Tests used to monitor CKD progression in patients with CKD stages 1–3, most typically quantitative measures of albuminuria and estimates of GFR calculated from serum creatinine, are derived from simple blood and urine tests that are widely available in primary care settings.

As described earlier in the section on screening, formulas to estimate GFR are automatically reported in many clinical labs from serum creatinine and are highly correlated with direct GFR measurement.⁴⁶ Compared with measured GFR, the CKD-EPI formula to estimate GFR has only a small bias at all levels of measured GFR,⁴⁷ which represents an improvement in accuracy compared with the MDRD formula, particularly in individuals with GFR ≥60 ml/min/1.73m². Both formulas suffer from some imprecision, however, as more than 15 percent of their estimates diverge from measured GFR by at least 30 percent.⁸ Still, they appear to perform well for one-time classification of individuals as either having CKD or not. The sensitivity and specificity of estimated GFR <60 mL/min/1.73m² for detection of directly measured GFR <60 mL/min/1.73m² were 91 percent and 87 percent according to the CKD-EPI equation and 95 percent and 82 percent according to the MDRD Study equation.⁸ Unfortunately, we did not identify data regarding the accuracy and precision of these formulas for assessing change in GFR within individuals over time, or their sensitivity and specificity for detecting change in GFR category over time (e.g., a decline from a GFR of 30 to 59 ml/min/1.73m2 to one of <30 ml/min/1.73m²).

Also as described in the section on screening, inter-assay and intra-assay coefficient of variance for urinary albumin is less than 5 percent.^49,50 However, as is the case for individuals without CKD, intra-individual variation of urinary albumin excretion is high in individuals with CKD. The impact of hydration can be addressed by accounting for urine output (e.g., using urine albumin-to-creatinine ratio), but nonhydration factors that may impact estimates of urinary albumin excretion include body position, exercise, certain medications, fever, and urinary tract infections.²⁷ As an illustration of this variability, based on NHANES data, among individuals with one-time microalbuminuria and GFR ≥60 ml/min/1.73m², only 63 percent had either microalbuminuria or macroalbuminuria on repeat testing two months later.⁵¹ Further, even in a diabetic population with persistent microalbuminuria over a 2-year period, regression of the microalbuminuria to normal occurred in 59 percent patients during a subsequent 6-year evaluation period.⁵² This variability makes it more difficult to determine whether longitudinal changes in measured albuminuria represent progression of CKD.

In Patients With CKD Stages 1–3 Whose CKD Has Progressed, Do Treatments Improve Important Clinical Outcomes?

For monitoring to improve clinical outcomes, changes in CKD status (such as the patient reaching a specific threshold or rate of change in kidney function or damage) would need to impact patient behavior or provider treatment in ways that improve these outcomes. RCT evidence that certain treatments had differential effects on clinical outcomes between patients with CKD stages 1–3 and those with CKD stage 4, or differential effects between different categories of patients within CKD stages 1–3 might suggest that treatment should be modified when change in CKD status is identified. While RCT data on CKD treatments are reviewed in greater detail elsewhere in this report, there is limited evidence to suggest that some treatments may have such differential effects based on CKD stage. For example, in RCTs comparing ACEI versus placebo treatment, a significant 40 percent reduction in relative risk of ESRD with ACEI is evident in trials comprised of patients with macroalbuminuria. By comparison, in ACEI versus placebo trials comprised of patients with microalbuminuria only, with very low power to detect changes in ESRD events, the pooled effect size suggests no difference between treatments. In a post hoc analysis in CHF patients with CKD, tests for interaction between study participants with GFR of >60, 45 to 60, and 30 to 44 ml/min/1.73m² suggest that benefit of beta blocker treatment versus placebo may be greater in the lower GFR group for reducing risk of hospitalizations due to CHF (p=.038), of two composite outcomes including all-cause mortality and hospitalization (both p <.05), and may be borderline significant with regard to all-cause mortality (p=.095).

Psychological Effects of Monitoring

We did not identify any studies that reported on the psychological effects of monitoring tests for CKD.

Adverse Physical Effects of Monitoring Tests and of Followup Tests To Further Evaluate Monitoring Tests

Phlebotomy required to measure serum creatinine may be associated with a small degree of bruising or discomfort. In a small number of patients, postscreening evaluation will include a renal biopsy, which has an associated risk of pain, bleeding, and infection.

Incorrect Reclassification of CKD Severity

We did not identify any studies that reported on the effects of testing that incorrectly reclassifies patients with CKD stage 1–3 as having worse or improved CKD, or even no CKD. Limitations in the precision of formulas that estimate GFR means there is a reasonable likelihood that any one test will suggest that a patient's CKD has changed or remained stable when this isn't the case. However, the small bias, in particular of the CKD-EPI formula, suggests that multiple GFR estimates will cluster accurately around true measured GFR. The high intra-individual variability of albuminuria in the absence of changes in underlying disease means there is at least a modest likelihood that findings of any one quantitative test will be inaccurate, whether it indicates that a patient's albuminuria is improving, stable, or worsening. As an example, in one study cited above, more than half of individuals with persistent microalbuminuria during a 2-year period regressed to normal over a subsequent 6-year evaluation period.⁵²

Labeling of an Individual With More Advanced CKD Stage

We did not identify any studies that reported on the effects of labeling an individual with a more advanced CKD stage.

Increased Clinic Visits to Primary and/or Specialist Providers

We did not identify any studies that reported on the effect of CKD monitoring tests on subsequent patient visits to primary or specialist providers. However, individuals whose monitoring tests indicate progression of their CKD seem likely to be seen more frequently in primary and specialty clinics. These visits may be for further evaluation to confirm the abnormal monitoring test, or providers may follow and treat these patients under the assumption that they have more severe CKD.

Adverse Effects Associated With Treatment

We systematically reviewed the RCT evidence on adverse effects of treatments of CKD patients in the Results section for Key Question 6.

Impact on Insurance Coverage

We did not identify any studies that reported the effects of being diagnosed with worsening CKD on obtaining or keeping health insurance coverage.

Overview

In patients with CKD, compared with placebo, we found moderate strength of evidence that ACEI treatment does not reduce risk of all-cause mortality more than placebo, and low strength of evidence that ACEI treatment does reduce risk of ESRD. Compared with placebo, ACEI treatment did not appear to reduce risk of MI or stroke, but significantly reduced risk of doubling serum creatinine and risk of progression from microalbuminuria to macroalbuminuria.

Description of Studies

Seventeen trials met all eligibility criteria and randomized participants with CKD (n=11,661, range 52 to 4,912) to an ACEI versus placebo (n=16 trials).^{55,57-59,61-72} or no treatment (n=1 trial).⁷³ Two of the included reports were post hoc analyses performed within subsets of participants with CKD from larger trial populations that were not originally limited to subjects with CKD.^61,66 Detailed baseline characteristics are presented in Appendix Tables C1 and C2.

Among eligible trials, 7,537 participants were randomized to ramipril versus placebo (n=7 trials),^{58,63,65,66,68,69,71} 1,757 to perindopril versus placebo (n=1 trial),⁶¹ 864 to fosinopril versus placebo (n=1 trial),⁶² 665 to captopril versus placebo (n=4 trials)^57,64,67,72 583 to benazepril versus placebo (n=1 trial),⁷⁰ 108 to enalapril versus placebo (n=1 trial),⁵⁹ 97 to lisinopril versus placebo (n=1 trial),⁵⁵ and 52 to enalapril versus no treatment (n=1 trial).⁷³ The mean age of subjects was 60 years (range 33 to 70; n=16 trials), and men constituted 66 percent (range 35 to 82; n=15 trials) of all patients randomized. Among the five trials reporting ethnicity, the patients were mostly of white race (77 percent).^{57,61,62,67,72} Most trials were conducted in Europe (including North Africa and Israel), three were conducted primarily or partially in the United States, and two were conducted in Japan. Mean or median study duration ranged from 6 months to 5 years. Seven trials had a followup of 3 years or longer and 12 trials had a followup of at least 2 years. Only one trial had a study duration of less than 1 year.⁷¹ One trial was conducted in a subset of individuals who previously had responded to an effort to screen all city residents aged 28 to 75 years for albuminuria.⁶²

Renal Function

One of the two post hoc analyses restricted inclusion to participants with GFR <60 ml/min/1.73m², by definition CKD stage 3 or worse.⁶¹ Otherwise, no trial based study eligibility on CKD stage or reported baseline distribution of participants by CKD stage. In 15 of 17 trials, participants were required to have albuminuria or proteinuria. In 10 of these trials, participants must have been microalbuminuric,^{55,57-59,62,65-67,71,73} most commonly with a urinary albumin excretion rate of 20 to 200 μg/minute. In three of the 15 trials, they were required to have overt proteinuria, with minimum thresholds ranging from ≥500 mg/day,⁷² to ≥1 but ≤3 g/day,⁶⁸ and to >3 g/day.⁶⁹ In the last two of the 15 trials, both microalbuminuric and macroalbuminuric participants were allowed,^63,64 with approximately three-quarters of the participants in one of these trials being microalbuminuric,⁶³ but no similar data reported for the other trial. Among the 15 trials requiring participants to have albuminuria or proteinuria, seven required that participants also have normal creatinine, creatinine clearance or GFR,^{55,57,58,62,67,71,73} three allowed some participants with abnormal levels for these renal function measures but mandated a maximally abnormal limit,^63,66,72 and the remaining five trials did not specify an eligibility requirement with respect to these measures.^{59,64,65,68,69} Finally, inclusion in two of 17 studies was based strictly on elevated serum creatinine, or reduced creatinine clearance or GFR.^61,70

Among the 10 trials restricted to microalbuminuric patients, mean baseline urinary albumin excretion rate was reported as 61.0 μg/min (range 53 to 71.5) in five trials^{55,57,58,67,71} and as 25.6 mg/24 hour (range 23 to 72) in two trials,^62,73 and mean urinary protein excretion rate was 133 mg/24 hours in one trial.⁵⁹ Among the three trials restricted to patients with overt proteinuria, mean urinary protein excretion was 3.0 g/day (range 1.7 to 5.3).^68,69,72 In the two trials that permitted inclusion of both microalbuminuric and macroalbuminuric patients, one reported mean baseline urinary albumin excretion rate of 711 mg/24 hours.⁶⁴ One of two trials that did not require albuminuria for inclusion nevertheless had an elevated mean baseline urinary protein excretion rate of 1.8 g/day,⁷⁰ while the other did not report baseline albuminuria or proteinuria.⁶¹ In trials reporting, mean baseline serum creatinine was 1.0 mg/dL (range 0.8 to 2.4; n=10 trials),^{55,57,59,62-64,68-70,72} mean creatinine clearance was 64.1 ml/min/1.73m² (range 43 to 114; n=8 trials),^{55,57,64,68-70,72,73} and mean GFR was 68.5 ml/min/1.73m² (range 39 to 114; n=5 trials).^55,58,67-69

Baseline Comorbidities

Twelve of 17 studies were restricted to patients with diabetes, including seven limited to those with type 1 diabetes,^{55,57-59,64,65,72} four limited to those with type 2 diabetes,^63,67,71,73 and one analysis that was open to both types of diabetics.⁶⁶ Among the five trials that did not report restricting enrollment solely to diabetics,^61,62,68-70 two nevertheless excluded participants with type 1 diabetes,^69,70 and three reported no data on baseline prevalence of diabetes^68-70 Mean glycosylated hemoglobin was 8.2 percent (range 7.1 to 11.7, n=10 trials).^{55,57-59,63-65,71-73}

Seven trials excluded participants with hypertension,^{55,57-59,62,65,73} including five that mandated that blood pressure be controlled without antihypertensive medications.^57,58,62,73 Four additional trials excluded participants only for severe hypertension.^68-71 In addition, though information on hypertension was not available for all participants from two studies, prevalence was at least 35 percent⁷¹ and 53 percent⁶¹ in these two trials. Prevalence of hypertension across all trials excluding these two with incomplete information was 49.8 percent (n=14 trials). Mean systolic and diastolic blood pressures at baseline were 144 mm Hg (range 126 to 149) and 83 mm Hg (range 74 to 92), respectively.

One trial reported data on prevalence of “cardiovascular disease,” at 24 percent.⁶³ Another trial was comprised entirely of participants with a history of cerebrovascular disease, including 71 percent with ischemic stroke, 10 percent with hemorrhagic stroke, and 7 percent with a stroke of unknown type.⁶¹ Prevalence of specific cardiovascular conditions was reported in few trials, including coronary artery disease (18.5 percent, range 0 to 20, n=2 trials),^57,61 myocardial infarction (5.1 percent, range 0 to 6, n=3 trials),^57,62,63 and stroke (3.5 percent, range 0.8 to 4; n=2 trials).^62,63 Participants with CHF were excluded from four trials,^57,62,63,66 and prevalence of CHF was not reported in other trials.

Study Quality

(Appendix Table C140)

Among the 17 studies, five were rated good quality and 12 were rated fair quality. Allocation concealment was adequate in seven trials and unclear in the remaining studies. All 16 placebo-controlled trials were double blinded. Nine trials reported outcomes assessment by blinded adjudication committees. Analysis by intention-to-treat principle was reported in nine trials. All trials adequately described reasons for study withdrawal except for the two reports that were post hoc subgroup analyses from larger trials. Percentages of study withdrawals ranged from 7 to 32 percent, including nine trials with withdrawal rates greater than 20 percent.^{55,57,58,62,64,68-70,72} No data were reported on withdrawals in the two studies that were post hoc analyses of CKD subsets from larger trial populations not limited to CKD.^61,66

Results

Mortality

(Table 7, Appendix Table C3, and Appendix Figure C1)

Table 7

Pooled clinical and renal outcomes, ACEI monotherapy versus control treatment trials.

All-Cause Mortality

Patients with CKD randomized to ACEIs did not have a significantly reduced risk of all-cause mortality compared with those assigned placebo (RR 0.94, 95% CI, 0.80 to 1.12; n=16 trials, 11,536 patients). In two trials reporting, effect of ACEI versus placebo on mortality risk appeared similar in patients with and without CKD. In the HOPE trial,^66,74 relative risk of mortality was 0.77 [95% CI, 0.64 to 0.93] in patients with microalbuminuria and 0.90 [95% CI, 0.78 to 1.04] in patients without microalbuminuria (p=0.20 for interaction). In a second trial,⁶¹ relative risk of mortality was 1.04 [95% CI, 0.83 to 1.31] in patients with creatinine clearance <60 ml/min and 0.84 [95% CI, 0.68 to 1.04] in patients with creatinine clearance ≥60 ml/min (p=0.1 for interaction).

Cardiovascular Mortality

Compared with placebo treatment, trial participants assigned to ACEIs also were not at lower risk for cardiovascular mortality (RR 1.03, 95% CI, 0.86 to 1.23).^61-63 Effect of treatment appeared similar in patients with and without CKD.⁶¹

Subgroup Results

Albuminuria or Impaired eGFR

(Figures 5 and 6)

Figure 5

ACEI versus placebo: All-cause mortality by albuminuria subgroups.

Figure 6

ACEI versus placebo: End-stage renal disease by albuminuria subgroups.

In trials restricted to patients with overt proteinuria at baseline, there was no significant difference in risk between those assigned ACEI versus placebo for all-cause mortality (RR 0.71, 95% CI, 0.33 to 1.54; n=3 trials, 761 patients). However, there was a significant 40 percent relative reduction in risk of ESRD (12.0 versus 20.7 percent; RR 0.60, 95% CI, 0.43 to 0.83; n=861 patients).^68,69,72 In trials restricted to patients with microalbuminuria, mortality risk was significantly reduced in the ACEI group versus placebo (9.3 versus 12.1 percent; RR 0.79, 95% CI, 0.66 to 0.96; n=3,440 patients), with similar results in the diabetic (RR 0.78, 95% CI, 0.61 to 1.00; n=1,140 patients) and nondiabetic (RR 0.75, 95% CI, 0.55 to 1.02; n=816 patients) microalbuminuria subgroups. However, there was no significant reduction in risk of ESRD between ACEI and placebo groups (0.8 versus 0.9 percent; RR 0.88, 95% CI, 0.27 to 2.88; n=1,234 patients).^59,66 In trials restricted to patients with microalbuminuria or worse, there was no significant difference between treatment groups in risk of ESRD (0.4 versus 0.5 percent; RR 0.93, 95% CI, 0.42 to 2.03; n=5,495 patients)^63,70 or mortality (RR 0.92, 95% CI, 0.74 to 1.15; n=9,192 patients). However, the two trials that together contributed more than 95 percent of the deaths for the ACEI versus placebo albuminuria subgroup analyses presented contrasting results, with a significant reduction in mortality risk in the HOPE trial^66,74 (15.7 versus 20.3 percent; RR 0.77, 95% CI, 0.64 to 0.93) but not the DIABHYCAR trial⁶³ (13.7 versus 13.1 percent; RR 1.04, 95% CI, 0.90 to 1.20). In the overall HOPE study population, 80 percent of participants had a history of coronary artery disease, including 52 percent with a history of MI, and 38 percent had diabetes, though comorbidity data were not reported for the subset with CKD. HOPE study participants were randomized to ramipril 10 mg per day versus placebo, and those with CKD were defined as having microalbuminuria. In the DIABHYCAR trial, prevalence of cardiovascular disease was lower at 24 percent, with only 6 percent having a history of MI, and 100 percent had diabetes. Participants were randomized to ramipril 1.25 mg per day versus placebo. Those with CKD were defined as having either microalbuminuria or macroalbuminuria. In results from two trials restricted to patients with impaired eGFR, there was no significant difference between treatments in risk of mortality (RR 2.14, 95% CI, 0.34 to 13.39).

Diabetes

In 12 trials restricted to patients with diabetes, there was no significantly reduced risk with ACEI versus placebo for mortality (RR 0.91, 95% CI, 0.70 to 1.18, n=11 trials), ESRD (RR 0.73, 95% CI 0.48 to 1.10, n=4 trials), MI, stroke, or doubling of serum creatinine (RR 0.69, 95% CI, 0.44 to 1.09, n=5 trials). With respect to mortality risk, results appeared clinically heterogeneous between diabetic participants from the HOPE trial^66,74 (16.3 versus 20.8 percent; RR 0.78, 95% CI, 0.61 to 1.00) and those from the DIABHYCAR trial⁶³ (13.7 versus 13.1 percent; RR 1.04, 95% CI, 0.90 to 1.20). In contrast, diabetic participants randomized to ACEI had a significant reduction in risk of conversion from microalbuminuria to macroalbuminuria (RR 0.38, 95% CI, 0.18 to 0.84, n=7 trials). In four trials reporting a composite vascular outcome, risk was significantly reduced in the ACEI group in one trial.^66,74

Hypertension

No trials were restricted to patients with hypertension, but in seven trials that excluded patients with hypertension, there was a significantly reduced risk of stroke (RR 0.10, 95% CI, 0.01 to 0.78, n=1 trial), doubling of serum creatinine (RR 0.15, 95% CI, 0.04 to 0.65), and conversion from microalbuminuria to macroalbuminuria (RR 0.29, 95% CI, 0.13 to 0.64), but no significant treatment group difference in mortality (RR 1.87, 95% CI, 0.65 to 5.37; n=7 trials, 1,454 patients) or other clinical vascular outcomes.

Congestive Heart Failure

No trials were restricted to patients with CHF, but in four trials that excluded patients with CHF, there was no significant difference between ACEI and placebo treatment groups in risk of mortality (RR 1.07, 95% CI, 0.52 to 2.18; n=4 trials, 1,192 patients), ESRD, or any other vascular or renal outcome.

Cerebrovascular Disease

In one trial restricted to patients with a history of cerebrovascular disease, risk of stroke was significantly reduced with ACEI versus placebo (RR 0.71, 95% CI, 0.57 to 0.89).However, there was no significant difference in risk of mortality (RR 1.07, 95% CI, 0.87 to 1.32), and no other vascular or renal outcomes were reported. Otherwise, no trials were limited to or excluded participants with cardiovascular disease.

Summary

In patients with CKD stages 1–3, compared with placebo, ACEI monotherapy did not significantly reduce risk of all-cause mortality in results overall. However, results appeared discordant between the two trials that together reported nearly all the deaths. In a study comprised of patients with microalbuminuria, and a high prevalence of cardiovascular disease who were treated with ramipril 10 mg per day versus placebo, mortality risk was significantly reduced. Results were similar in study subsets with and without diabetes. In a study comprised of patients with diabetes, microalbuminuria or macroalbuminuria, and a low prevalence of cardiovascular disease who were treated with ramipril 1.25 mg per day versus placebo, mortality was not significantly reduced. Because the latter trial appeared to be comprised of participants at only slightly lower absolute mortality risk and had a large total number of deaths, the difference between ramipril treatment doses seems the most likely explanation for the difference in outcomes. There was no significant difference between ACEI and placebo in risk of cardiovascular mortality, or MI. Overall, there was no significant reduction in risk of stroke, though results appeared heterogeneous between trials, with two moderate-sized trials reporting a significant reduction in stroke risk and another one finding no difference. Two of seven trials reporting found a significantly reduced risk in a composite vascular outcome in participants randomized to ACEI. Overall, subjects assigned to ACEIs had a significant 35 percent reduction in risk of ESRD compared with patients assigned to placebo. This risk reduction appeared restricted to studies that enrolled only patients with overt proteinuria. CKD patients assigned to ACEIs had a significant 40 percent reduction in risk of doubling serum creatinine, 62 percent reduction in risk of converting from microalbuminuria to macroalbuminuria, and approximately 50 percent reductions in all composite renal outcomes reported. Overall study withdrawals were not significantly different between ACEI and placebo groups. ACEIs increased risk of cough, but there was little apparent difference from placebo subjects in hyperkalemia. Results were limited in that few trials were of sufficient size to assess mortality or clinical vascular outcomes.

Overview

In patients with CKD, we found low strength of evidence suggesting that there is no difference between ACEI and ARB treatment for the outcome of all-cause mortality. We found insufficient evidence regarding whether there is a difference between these treatments for ESRD.

Description of Studies

Six trials met all eligibility criteria and randomized participants (n=4,799, range 90 to 4,046) to ACEI monotherapy versus ARB monotherapy.^67,75-79 Detailed baseline characteristics are presented in Appendix Tables C1 and C2.

Among eligible trials, 4,046 participants were randomized to ramipril versus ARB (n=1 trial), 353 were randomized to enalapril versus ARB (n=2 trials), 309 were randomized to lisinopril versus ARB (n=2 trials), 91 were randomized to captopril versus ARB (n=1 trial), 4,515 were randomized to telmisartan versus ACEI (n=3 trials), 181 were randomized to valsartan versus ACEI (n=2 trials), and 103 were randomized to losartan versus ACEI (n=1 trial). While five of the six trials maintained the ACEI versus ARB comparison throughout the entire treatment period, in a single partial crossover trial, after 24 weeks patients initially assigned to ACEI were randomized to ACEI plus ARB versus continued ACEI monotherapy, and patients initially assigned to ARB monotherapy were randomized to ARB plus ACEI versus continued ARB monotherapy.⁷⁷ By far the largest study, comparing ramipril versus telmisartan, was a post hoc analysis performed within the subset of ONTARGET trial participants with CKD (n=4,046 out of 25,620).⁷⁵ The mean age of subjects was 59 years (range 56 to 61; n=5 trials) and men constituted 62 percent (range 37 to 81; n=5 trials) of all patients evaluated. The ethnicity of patients in the three trials reporting was nearly all white race (96 percent).^67,78,79 Two trials were conducted exclusively in Canada, two exclusively in Europe, one in Turkey, and one trial included sites in the United States, Canada, and Europe, as well as Asia, Africa, and Australia. Mean or median study duration was 1 year in three trials, 2.5 years in one trial, and about 5 years in two trials.

Renal Function

The single post hoc analysis restricted inclusion to participants with GFR <60 ml/min/1.73m², by definition CKD stage 3 or worse,⁷⁵ and a second trial required participants to have microalbuminuria and a GFR ≥60 ml/min/1.73m², by definition CKD stages 1–2.⁶⁷ Otherwise, no trial based study eligibility on CKD stage and no trial reported baseline distribution of participants by CKD stage. Among the six trials, five required participants to be albuminuric, including four restricted to patients with microalbuminuria,^67,76,77,79 and one that allowed subjects to have either microalbuminuria or macroalbuminuria.⁷⁸ Among the five trials requiring participants to have albuminuria, two required that they also have normal creatinine or GFR,^67,78 and three allowed some participants with abnormal levels for these renal function measures but mandated a maximally abnormal limit.^76,77,79 One trial determined eligibility based only on impaired GFR.⁷⁵ Within trials, measures of baseline renal function were inconsistently reported. The ONTARGET post hoc analysis reported no data on baseline renal function in its CKD population.⁷⁵ In other trials, the most commonly reported measure was urinary albumin excretion rate (UAER), with mean UAER 62 μg/min in two trials^67,79 and 260 mg/24 hours in one trial,⁷⁷ and median UAER 46 μg/min (ACEI treatment arm) to 60 μg/min (ARB treatment arm) in one trial.⁷⁸ The mean GFR was 92 ml/min/1.73 m² (range 91 to 96, n=3 trials).^67,78,79 In two trials, mean baseline serum creatinine was 1.0 mg/dL in both trials.^77,78 Mean creatinine clearance was 101 ml/min/1.73 m² (range 97 to 112, n=2 trials).^76,77

Baseline Comorbidities

The study within the subset of ONTARGET participants with impaired GFR did not report any data on their baseline characteristics,⁷⁵ though the main study required subjects to have established atherosclerotic vascular disease or diabetes associated with end-organ damage. In the main study, prevalence of comorbidities included diabetes 37.3 percent, hypertension 68.3 percent, and MI 48.7 percent.⁸⁰ Within the five other trials, prevalence of diabetes was 97 percent, including four trials comprised entirely of subjects with type 2 diabetes^67,77-79 and another that excluded type 1 diabetics and had a prevalence of type 2 diabetes of 74 percent.⁷⁶ Nearly all study participants were hypertensive at baseline (94 percent; range 33 to 100), including four trials that enrolled only patients with hypertension.^76-79 Five trials excluded patients with severe hypertension,^75-79 and mean baseline systolic and diastolic blood pressure measurements were 151 and 87 mm Hg, respectively. Nearly half the enrollees from one trial had cardiovascular disease,⁷⁸ a history of non-MI cardiac disorder was reported in 19 percent of subjects in another trial,⁷⁶ and two trials excluded participants with CHF.^76,79 Otherwise, studies reported no data on baseline cardiovascular disease.

Study Quality

(Appendix Table C140)

Among the six trials, two were rated good quality and four were rated fair quality. Allocation concealment was adequate in three trials^75,76,78 and unclear in the remaining trials. All trials were double blinded except one open-label study.⁷⁷ Analysis by the intention-to-treat principle was reported in two trials.^75,78 All trials adequately described reasons for study withdrawals. No data on study withdrawals were reported in one trial.⁷⁵ Otherwise, withdrawals were 33 percent in one trial,⁷⁸ and ranged between 11 and 14 percent in the other trials.

Results

Summary

In trials comparing ACEI and ARB treatments individuals with CKD, there were very few vascular events reported, based on which there was no significant difference between treatments for the outcomes of all-cause mortality, cardiovascular mortality, MI, or CHF. No data were reported for stroke, ESRD, or any composite vascular outcome. Results from the CKD subset of the ONTARGET study population, whether defined by GFR <60 ml/min/1.73m² or by albuminuria, appeared to show no difference in the risk of the composite renal outcome of doubling creatinine, dialysis, renal transplant, or death. Results were limited by small sample sizes in all but one trial, and by the small number of events among trials reporting them. Because no trial provided followup beyond 5 years, longer term effects of ACEI monotherapy versus ARB monotherapy in CKD patients could not be determined from these trials. Overall study withdrawals were not significantly different between ACEI and ARB treatment groups, though cough was significantly more likely in participants assigned to ACEI.

Overview

In comparing ACEI versus CCB for treatment of patients with CKD, we found low strength of evidence, suggesting that there was no difference in risk of all-cause mortality or ESRD. We found no significant difference between treatment groups in risk of cardiovascular mortality, stroke, or halving of GFR.

Description of Studies

Six trials met all eligibility criteria and randomized participants (n=4,357, range 88 to 3,049) to ACEI monotherapy versus CCB monotherapy.^55,81-88 Baseline characteristics are presented in Appendix Tables C1 and C2.

Among eligible trials, 3,137 participants were randomized to lisinopril versus CCB (n=2 trials), 653 were randomized to ramipril versus CCB (n=1 trial), 446 were randomized to fosinopril versus CCB (n=2 trials), 121 were randomized to captopril versus CCB (n=1 trial), 3,907 were randomized to amlodipine versus ACEI (n=3 trials), and 450 were randomized to nifedipine versus ACEI (n=3 trials). By far the largest eligible study was a post hoc analysis performed in the subset of 3,049 individuals with GFR <60 ml/min/1.73m² from the larger Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial (ALLHAT) (N= 42,418).^81-83 In the AASK trial, designed as a 3×2 factorial study, besides randomizing 653 participants to ACEI versus CCB, an additional 441 were randomized to beta blocker, and all participants also were randomized to one of two blood pressure target groups.^89,90 The CCB treatment arm was stopped early by recommendation of the data and safety monitoring board, with patients switched to open label medication. The mean age of study participants was 66 years (range 37 to 71; n=6 trials) and men constituted 51 percent (range 48 to 69) of all subjects studied. In the two trials that reported race/ethnicity,^81-83,85 48 percent of participants were white and 38 percent were African American, including one trial comprised entirely of African American participants.⁸⁵ Two trials were conducted primarily in the United States,^81-83,85 three trials were conducted in Italy, and one was performed in Spain. Mean or median study duration ranged from 3 to approximately 5 years.

Renal Function

The single post hoc analysis restricted inclusion to participants with GFR <60 ml/min/1.73m², by definition CKD stage 3 or worse,⁸¹ and a second trial required participants to have microalbuminuria and a GFR ≥80 ml/min/1.73m², by definition CKD stages 1-2.⁵⁵ Otherwise, no trial based study eligibility on CKD stage and no trial reported baseline distribution of participants by CKD stage. Among the six trials, two required that participants have microalbuminuria to be included,^55,84 while four determined eligibility based only on impaired creatinine or GFR.^81-83,85-88 Within included participants, there was no single measure of renal function or damage that was reported in every trial. The most commonly reported measure of baseline renal function was serum creatinine, with a mean of 2 mg/dL (range 0.96 to 2.8, n=5 trials)^55,84-88 The mean baseline GFR, reported in three trials, was 50 ml/min/1.73m2 (range 46 to 120),^55,81-83,85 and the mean baseline creatinine clearance concentration was 66 ml/min/1.73m² (range 36 to 109, n=3 trials).^55,84,86 Mean proteinuria was 0.94 gm/24 hours (range 1.7 to 1.8, n=3 trials),^85-88 and mean urinary albumin excretion rate was 89 μg/min (range 61 to 97, n=2 trials).^55,84

Baseline Comorbidities

Thirty percent of study participants had diabetes, which included two trials that restricted enrollment to participants with diabetes^55,84 and three trials that excluded patients with diabetes.^85-88 In two trials reporting data, mean baseline hemoglobin A_1c was 7.2.^55,84 Ninety-nine percent of study participants had hypertension, which included five trials that restricted enrollment to participants with hypertension and one small trial that excluded patients with hypertension.⁵⁵ Mean baseline systolic and diastolic blood pressure measurements were 149 and 87 mm Hg, respectively. One trial excluded any participants with a history of coronary artery disease,⁸⁴ three excluded participants with either recent^55,86 or severe^87,88 cardiovascular events but provided no data on past history of coronary artery disease, while the remaining two trials reported that 29 percent^81-83 and 52 percent⁸⁵ of randomized participants, respectively, had a history of coronary artery disease.

Study Quality

(Appendix Table C140)

Among the six trials, two were rated good quality and four were rated fair quality. Allocation concealment was adequate in three of six trials and three trials were double blinded. Analysis by the intention-to-treat principle was reported in four trials^81-83,85-88 All trials, except the single post hoc analysis, adequately described reasons for study withdrawal. Withdrawals across studies ranged from 0 to 37 percent.

Results

Summary

In patients with CKD, there was no apparent difference between treatment with ACEI monotherapy and CCB monotherapy for the outcomes of all-cause mortality, cardiovascular mortality, stroke, CHF, any composite vascular endpoint, or ESRD. Relative risk of MI could not be determined. While results for the composite renal outcome indicated significant benefit for ACEI treatment compared with CCB in one trial,⁸⁶ there was no between-treatment group difference in the composite renal endpoints reported in two other trials.^81-83,85 Results were limited in that several studies were not designed for and reported no clinical outcomes data, and the modest number of clinical events overall may have limited power to detect differences between treatment groups. Further, no trial provided followup beyond 5 years; therefore, longer term effects of ACE-inhibitor monotherapy versus CCB monotherapy cannot be determined from these data. Withdrawals appeared similar between treatment groups, with cough appearing more common in patients assigned ACEI and edema more common in patients assigned CCB.

Overview

In comparing ACEI versus beta blocker treatment in patients with CKD, there was low strength of evidence that there is no difference in risk of all-cause mortality and ESRD. We found no significant difference between treatments for risk of cardiovascular mortality, stroke, or heart failure.

Description of Studies

Three trials met all eligibility criteria and randomized participants (n=1,080, range 100 to 877) to ACEI versus beta blocker monotherapy.^90,92,93 Baseline characteristics are presented in Appendix Tables C1 and C2.

Among eligible trials, 877 participants were randomized to ramipril versus metoprolol (n=1 trial),⁹⁰ 103 were randomized to enalapril versus atenolol (n=1 trial),⁹² and 100 were randomized to enalapril versus either atenolol or acebutelol (n=1 trial).⁹³ The mean age of study participants was 54 years, and men constituted 61 percent of patients studied. In the single trial that reported race/ethnicity, 100 percent of participants self-identified as African American.⁹⁰ One trial was conducted in the United States,⁹⁰ while two trials were performed in Europe.^92,93 Mean or median study durations were three years or greater in all trials.

Renal Function

No trial based study eligibility on CKD stage and no trial reported baseline distribution of participants by CKD stage. All three trials based eligibility on impairment in GFR (20 to 65 ml/min/1.73m²),⁹⁰ creatinine clearance (30 to 90 ml/min),⁹² or creatinine (2.3 to 5.2 mg/dL).⁹³ None based inclusion on the presence of albuminuria, though one excluded patients with urinary protein-creatinine ratio >2.5,⁹⁰ and another excluded participants with nephrotic syndrome.⁹³ Mean serum creatinine, reported in all three trials, was 2.0 mg/dL (range 1.8 to 3.0). Mean baseline GFR was 47 ml/min/1.73m² (range 46 to 53, n=2 trials).^90,92 Urinary protein excretion ranged from 0.5 g/24 hour⁹⁰ to 2.2 g/24 hour⁹³ in two trials. In the third trial, median urinary protein excretion was 3.3 g/24 hour.⁹²

Baseline Comorbidities

All three trials excluded individuals with diabetes. Approximately 51 percent of participants had a history of heart disease in one trial,^90,91 patients with coronary artery disease were excluded from one trial,⁹³ and no data were reported regarding cardiovascular disease in the third trial.⁹² While two trials were limited to patients with hypertension,^90,93 more than half of the participants in the third trial were reported to have diastolic blood pressure less than 90 mm Hg off antihypertensive medications.⁹² Overall, 96 percent of participants in the three trials had hypertension. Mean baseline systolic and diastolic blood pressure measurements were 152 and 95 mm Hg, respectively.

Study Quality

(Appendix Table C140)

Among the three trials, one was rated good quality and two were rated fair quality. Allocation concealment was adequate in two trials,^90,93 two trials were double blinded,^90,92 and analysis was performed by intention-to-treat in two trials.^90,93 All trials adequately described reasons for study withdrawal. Percentages of study withdrawals ranged from 0 to 23 percent. The AASK trial reported that no participants withdrew from treatment or were lost to followup.⁹⁰

Results

Summary

In patients with CKD, there was no significant difference between ACEI and beta blocker treatment for risk of all-cause mortality, cardiovascular mortality, stroke, heart failure, or either of two composite vascular endpoints. Overall, there was no difference between ACEI and beta blocker treatment for risk of ESRD, but results were heterogeneous between trials. However, ACEI treatment was associated with a significantly lower risk of the composite renal outcome of >50 percent reduction in GFR, need for dialysis or transplant, or death. With respect to adverse effects, ACEI treatment was associated with a significantly higher rate of cough and angioedema. Results were limited in that only one study, the AASK trial, was designed to evaluate the effect of ACEI and beta blocker treatment on clinical cardiovascular outcomes. The two smaller trials reported few or no events for most vascular endpoints and had very limited power to detect differences in these outcomes between treatment groups. No trial provided mean or median followup beyond 5 years; therefore, longer term effects of ACEI monotherapy versus beta blocker monotherapy cannot be determined from these study results.

Overview

In patients with CKD there was insufficient strength of evidence that there was no difference in risk of all-cause mortality risk between those assigned to ACEI and those allocated to diuretic treatment. There was low strength of evidence that there was no difference between ACEI and diuretic in risk of ESRD. There was no significant difference between treatment groups in risk of stroke or multiple composite cardiovascular outcomes, but there was a significantly increased risk of CHF in the group assigned to ACEI. Our confidence in these estimates is limited because they are based almost entirely on results reported from a post hoc analysis in a single large trial.

Description of Studies

Two trials met all eligibility criteria and randomized participants (n=4,716, range 570 to 4,146) to ACEI monotherapy versus diuretic monotherapy.^81-83,94 Detailed baseline characteristics are presented in Appendix Tables C1 and C2). One of the included reports was a post hoc analysis performed within a subset of participants with CKD from the ALLHAT trial, a population that was not originally limited to subjects with CKD.

In one study, from the ALLHAT trial, 4,146 participants were randomized to lisinopril versus chlorthalidone,^81-83 while in the second study, the NESTOR trial, 570 participants were randomized to enalapril versus indapamide.⁹⁴ The mean age in these two trials was 70 years (range 60 to 71), and men comprised slightly over half of all patients studied (51 percent; range 49 to 65). The most common race/ethnicity of patients in the two trials was white (61 percent), followed by black (23 percent).^81-83,94 Hispanics comprised 11 percent of participants in the ALLHAT study.^81-83 The NESTOR trial was conducted in Europe,⁹⁴ while the ALLHAT study was performed primarily in the United States.^81-83 Study durations were 1 year⁹⁴ and 4.9 years, respectively.^81-83

Renal Function

The single post hoc analysis restricted inclusion to participants with GFR <60 ml/min/1.73m², by definition CKD stage 3 or worse, and reported a mean baseline GFR of 50 ml/min/1.73m².⁸¹ The second study, the NESTOR trial, did not base eligibility on CKD stage and neither trial reported baseline distribution of participants by CKD stage. In the NESTOR trial, participants were required to have microalbuminuria for inclusion, and the mean baseline urinary albumin excretion rate was 58 μg/min, the urinary albumin/creatinine ratio was 6.2 mg/g, and the creatinine clearance was 92 ml/min/1.73m².⁹⁴ The two studies excluded subjects with baseline creatinine levels exceeding 1.7 mg/dL⁹⁴ and 2 mg/dL,^81-83 respectively.

Baseline Comorbidities

Both studies were limited to patients with hypertension, with mean blood pressures at baseline being 147/83 mm Hg in the ALLHAT study^81-83 and 161/94 mm Hg in the NESTOR trial.⁹⁴ In the ALLHAT study, 61 percent of participants reported cardiovascular disease, 31 percent reported coronary artery disease, and 31 percent were diabetic. In the NESTOR trial, however, prevalence of type 2 diabetes was 100 percent (mean hemoglobin A_1c 7.6 percent), but no information was reported regarding history of any cardiovascular disease.

Study Quality

(Appendix Table C140)

Of the two eligible trials, one was rated good quality and one was rated fair quality. Allocation concealment was adequate in the ALLHAT study and unclear in the NESTOR trial. Both trials were double blinded. Analysis by the intention-to-treat principle was reported in ALLHAT. However, the NESTOR trial excluded one randomized participant from analyses who was reported to not have been exposed to study drug.⁹⁴ The NESTOR trial reported an 11 percent withdrawal rate and adequately described reasons for study withdrawal. By contrast, the ALLHAT study reported no data regarding withdrawals.

Results

Summary

Within the two eligible trials of patients with CKD, there was no apparent difference between the ACEI and diuretic monotherapy treatment groups in risk of all-cause or cardiovascular mortality, MI, stroke, ESRD, or other composite clinical vascular or renal outcomes. There was a statistically significantly greater risk of heart failure among CKD patients allocated to ACEI therapy versus diuretic treatment. Results were limited in that one trial was a 1 year bioequivalence study not designed to evaluate the effect of these treatments on clinical events⁹⁴ and that the second study was a post hoc subgroup analysis. The large ALLHAT study also did not provide mortality data based on CKD status. Also, since mean followup did not extend beyond 5 years, longer term effects of ACE-inhibitor monotherapy versus diuretic monotherapy cannot be determined from these data. Withdrawals were not significantly different between treatment groups in the one trial reporting, and no adverse events data were available.

Overview

In patients with CKD, we found high strength of evidence that ARB treatment reduces risk of ESRD compared with placebo. These results are based entirely on data from trials enrolling CKD patients with overt albuminuria. We found high strength of evidence that ARB treatment does not reduce risk of all-cause mortality compared with placebo. While patients with CKD randomized to ARB versus placebo had a significantly lower risk of progression from microalbuminuria to macroalbuminuria, we found no statistically significant difference between treatment groups for risk of MI, and mixed results regarding risk for CHF hospitalization. Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events in some trials, and the heterogeneity of the study populations.

Description of Studies

Five trials met all eligibility criteria and randomized participants with CKD (n=5,769, range 527 to 1,991) to an ARB versus placebo.^95-99 Detailed baseline characteristics are presented in Appendix Tables C9 and C10. One of the included reports was a post hoc analysis⁹⁹ performed within a subset of participants with CKD from the TRANSCEND trial, a population that was not originally limited to subjects with CKD.

Among eligible trials, 1,738 participants were randomized to irbesartan versus placebo (n=2 trials), 1,513 participants (n=1 trial) to losartan versus placebo, and 2,518 participants (n=2 trials) to telmisartan versus placebo. The mean age of subjects was 62.7 years (range 58 to 68.7; n=5 trials), and men constituted 60.0 percent (range 51 to 69; n=4 trials) of all patients randomized. Four trials reported race/ethnicity, within which 64 percent of subjects were white. One trial was conducted in the United States,⁹⁷ one in Japan,⁹⁵ and three were multinational.^96,98,99

Renal Function

In four trials, patients were required to have albuminuria or proteinuria. In two of these trials, patients must have been microalbuminuric, with a urinary albumin excretion rate between 20-200 μg/min in one study⁹⁶ and a urinary albumin-creatinine ratio between 100-300 mg/g in the second study.⁹⁵ Both of these trials also were restricted to participants with a normal serum creatinine, defining them as CKD stages 1-2. In the other two trials that required albuminuria or proteinuria for entry, patients must have been overtly albuminuric or proteinuric, with either an albumin-creatinine ratio ≥300 mg/g or urinary protein excretion rate ≥0.5 g/day,⁹⁸ or a 24 hour urine protein excretion ≥900 mg.⁹⁷ These two trials required participants to be within a range that included both normal and moderately elevated serum creatinine values (e.g., between 1.0 and 3.0 mg/dL), so that it was not possible to determine CKD stage.^97,98 The fifth trial, a post hoc analysis of TRANSCEND study participants who had either impaired eGFR or albuminuria, excluded participants with serum creatinine >3.0 mg/dL. This study categorized participants with eGFR <60 ml/min/1.73m² versus ≥60 ml/min/1.73m², and with normoalbuminuria versus microalbuminuria versus macroalbuminuria, and all possible combinations so that they further could be categorized as either CKD stages 1-2 or CKD stages 3-4. At baseline, serum creatinine was the measure of renal function most frequently reported, with a mean of 1.5 mg/dL (range 1.2 to 1.9; n=4 trials). Two trials reported urinary albumin excretion rate, with results of 55.5 μg/min (0.08 g/day)⁹⁶ and 1.9 g/day,⁹⁷ respectively. Median urinary albumin-creatinine ratio reported in one trial was 1,250 mg/g.⁹⁸ One trial reported no baseline data on renal function/damage for its participants.⁹⁵

Baseline Comorbidities

Four studies were restricted to subjects with type 2 diabetes, and three further specified exclusion of patients with any nondiabetic kidney disease.^95,96,98 Two trials also were limited to subjects with hypertension,^96,97 while in two trials 81 and 93.5 percent of participants had a diagnosis of hypertension.^98,99 In the fifth trial, the prevalence of hypertension was not reported, though patients with severe hypertension (>180/100 mm Hg) were excluded and mean baseline blood pressure was 137/77 mm Hg.⁹⁵ Across all five trials, mean baseline blood pressures were 149/83 mm Hg (range 137/77 to 159/90 mm Hg). In one trial 28 percent of participants reported a history of cardiovascular disease.⁹⁷ In a second trial, cardiovascular disease was more common, including 73 percent of participants with coronary artery disease and 22 percent with stroke.⁹⁹ However, in the two other trials reporting data,^96,98 cardiovascular disease was uncommon, including myocardial infarction (8.9 percent, range 3 to 11.2, n=2 trials), coronary artery disease (4.5 percent, n=1 trial), stroke (0.9 percent, range 0.1 to 3.1, n=2 trials). Two trials explicitly excluded patients with CHF,^96,99 and another excluded patients with an indication for ACEIs or ARBs, likely indicating an exclusion of patients with CHF.⁹⁶ Finally, in one trial no entrance criteria related to cardiovascular disease were listed and no baseline data on cardiovascular disease were reported.⁹⁵

Study Quality

(Appendix Table C140)

Of the five eligible trials, three were rated good quality and two were rated fair quality. Allocation concealment was adequate in three trials^97-99 and unclear in two trials.^95,96 All trials were double blinded. All but one trial⁹⁵ performed analyses using the intention-to-treat principle. All trials adequately described study withdrawal and reasons for withdrawals, with withdrawals ranging from 0.8 to 13.1 percent of randomized participants.

Results

Mortality

(Table 8, Appendix Table C11, and Appendix Figure C2)

Table 8

Pooled clinical and renal outcomes, ARB monotherapy versus control treatment trials.

All-Cause Mortality

Among these CKD patients studied, overall incidence of death in trials reporting this outcome ranged from less than 1 percent of study participants in one trial,⁹⁶ to between 16 and 20 percent in the other three trials.^97-99 Nevertheless, no individual trial results suggested a difference in risk of death among CKD patients randomized to ARB versus those allocated to placebo. In pooled results, there was no between-treatment difference in mortality risk (RR 1.04, 95% CI, 0.92 to 1.18, n=4 trials, 5,242 patients). Only one trial reported results stratified by baseline category of albuminuria in patients with and without an eGFR <60 mL/min/1.73m². There was no difference in mortality stratified by baseline category of albuminuria in patients with and without an eGFR <60 mL/min/1.73m².⁹⁹

Cardiovascular Mortality

In a single trial that reported data on cardiovascular mortality, there was no significant difference in risk between study participants randomized to ARB versus placebo (RR 1.03, 95% CI, 0.80 to 1.31).⁹⁹

Renal Outcomes

(Table 8, Appendix Tables C14 and C15, and Appendix Figure C2)

End-Stage Renal Disease

In three trials reporting incident ESRD, subjects with CKD assigned to ARB treatment were 22 percent less likely to progress to ESRD than those allocated to placebo treatment, a statistically significant result (10.0 versus 12.9 percent; RR 0.78, 95% CI, 0.67 to 0.90; n=4,652 patients)^97-99 (Figure 7). Two of these trials were comprised entirely of participants with proteinuria, whereas the third trial reported results for risk of ESRD stratified by albuminuria groups.⁹⁹ It reported no interaction between category of albuminuria (normal, microalbuminuria, or macroalbuminuria) and the relative reduction in risk of ESRD with ARB treatment versus placebo.

Figure 7

ARB versus placebo: End-stage renal disease by albuminuria subgroups.

Other Renal Outcomes

In three trials reporting, CKD patients randomized to ARB treatment were significantly less likely to develop a doubling of their baseline serum creatinine (11.8 versus 15.2 percent; RR 0.78, 95% CI, 0.68 to 0.90; n=4,652 patients).^97-99 Risk of conversion from microalbuminuria to macroalbuminuria was 58 percent lower in CKD patients assigned to ARB compared with those allocated to placebo (13.2 versus 31.2 percent; RR 0.42, 95% CI, 0.33 to 0.52; n=2 trials, 1,104 patients).^95,96 One or more composite renal outcomes were reported in three trials (Appendix Table C15),^97,98 with all suggesting that assignment to ARB reduces risk of the composite outcome compared with placebo, though not all differences were statistically significant.

Summary

In individuals with CKD, compared with placebo, assignment to ARB treatment was associated with significant reductions in risk of ESRD (reported only in patients with macroalbuminuria), and of doubling of serum creatinine and conversion from microalbuminuria to macroalbuminuria (both reported only in patients with microalbuminuria at baseline). Assignment to ARB treatment also was associated with reduction in risk in one of two composite renal outcomes, and in risk of CHF hospitalization. There was no significant difference between treatment groups for the outcomes of all-cause mortality, MI, or any reported composite vascular outcomes. No trials reported results for stroke. Results were limited in that several outcomes were reported in only one trial or not at all, in particular with neither of the studies that limited enrollment to microalbuminuric patients reporting results for MI, stroke, CHF, ESRD, or a composite vascular or renal endpoint. Though withdrawal and adverse event reporting were limited, individuals with CKD allocated to ARB were significantly more likely to experience hyperkalemia requiring discontinuation of study medication. In one trial that reported results stratified by baseline albuminuria category, there was no significant difference between these groups in the relative risk between ARB and placebo for any outcome or adverse event.

Overview

In patients with CKD, we found low strength of evidence that ARB treatment does not reduce risk of all-cause mortality or ESRD relative to CCB. We found that patients assigned ARB treatment were significantly less likely to experience doubling of baseline creatinine, but that there was no significant difference between treatment groups for risk of stroke, or conversion from microalbuminuria to macroalbuminuria. Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events in some trials, and the heterogeneity of the study populations.

Description of Studies

Three trials met all eligibility criteria and randomized participants with CKD (n=3,924 patients, range 58 to 2,720) to an ARB versus CCB.^97,100,101 Detailed baseline characteristics are presented in Appendix Tables C9 and C10.

Among eligible trials, one compared candesartan to amlodipine (n=2,720 patients),¹⁰⁰ one compared irbesartan to amlodipine (n=1,146 patients),⁹⁷ and one compared candesartan to nifedipine (n=58 patients).¹⁰¹ In total, there were 2,778 participants randomized to candesartan versus a CCB and 3,866 participants randomized to amlodipine versus an ARB. The mean age of subjects was 63.2 years (range 59 to 65; n=3 trials), and men constituted 55.4 percent (range 46.6 to 64.3; n=3 trials) of all patients randomized. Just one trial reported race/ethnicity, in which 72.1 percent of subjects were white.⁹⁷ Two other trials were conducted in Japan.^100,101 Median study durations ranged from 1.8 to 3.2 years.

Renal Function

In two trials, the initial study design specified restriction to patients with albuminuria (repeated urinary albumin-creatinine ratio 100-300 mg/g)¹⁰¹ or proteinuria (urinary protein excretion ≥900 mg/24 hours).⁹⁷ In the third study, the current report¹⁰⁰ was a secondary analysis conducted in patients with either GFR <60ml/min/1.73m² or a positive dipstick test for proteinuria from among a larger trial population with either serum creatinine >1.3 mg/dL or undefined proteinuria.^101,102 Among the 2,720 participants enrolled in this study, 330 were reported in combined CKD stages 1 or 2, 2,265 were CKD stage 3, and 125 were CKD stage 4. No other study based inclusion on or reported distribution of participants by CKD stage. The only measure of renal function reported in more than one trial was serum creatinine, which ranged from 0.74 to 1.66 mg/dL in two trials reporting,^97,101 and was by definition >1.3 mg/dL in all participants in the third trial.¹⁰⁰ The baseline level of albuminuria differed considerably in two trials reporting, from an albumin-creatinine ratio of 237 mg/g¹⁰¹ to a 24 hour urinary albumin excretion of 1.9 g.⁹⁷ Neither baseline GFR nor creatinine clearance was reported in any trials.

Baseline Comorbidities

All three studies included only subjects with hypertension and type 2 diabetes. Mean baseline systolic and diastolic blood pressures was 162/90 mmHg. Two trials excluded subjects with severe hypertension (systolic >200mmHg and/or diastolic >110 or 120 mmHg).^100,101 Trials provided little information on participant history of cardiovascular disease. In one trial, 28.7 percent of subjects had a history of cardiovascular disease,⁹⁷ while in a second trial 4.8 percent of participants had a history of MI.¹⁰⁰ One trial excluded subjects with “severe cerebral or cardiovascular diseases,”¹⁰¹ while a second trial excluded participants with MI or stroke ≤6 months before screening, coronary angioplasty or bypass ≤6 months before screening or currently scheduled, current treatment for class II-IV CHF or ejection fraction <40 percent, or coronary artery disease requiring BB or CCB.¹⁰⁰

Study Quality

(Appendix Table C140)

Among the three eligible trials, one was rated good quality and two were rated fair quality. Allocation concealment was adequate in two trials⁹⁷ and unclear in two trials.^100,101 One trial was double blinded.⁹⁷ Two trials were single blinded, one to the assessors only¹⁰⁰ and one to the patients only.¹⁰¹ Analysis by the intention-to-treat principle was performed in two trials^97,100 and was unclear in one trial.¹⁰¹ Two trials adequately described reasons for study withdrawals, with withdrawals ranging from 0.6 to 3.4 percent of randomized participants.^97,101 The third trial did not report any data on withdrawals.¹⁰⁰

Results

Summary

In individuals with CKD, compared with CCB, assignment to ARB treatment was associated with a significant 33 percent reduction in risk of doubling serum creatinine, but no significant difference in risk of all-cause mortality, MI, stroke, ESRD, or at least two defined composite vascular outcomes. Risk for a composite renal outcome including doubling creatinine, ESRD, or death was significantly lower with ARB in one trial that enrolled CKD patients with substantial baseline proteinuria. In another study of CKD patients at lower risk for these renal outcomes, there was no significant reduction in risk. Results were limited in that most outcomes were reported in only one trial or were uncommon. Evaluated CKD study populations appeared heterogeneous with respect to risk of clinical events. However, small sample sizes and few clinical events in some studies, and the limited reported data on baseline vascular disease and renal function/damage, limited evaluation as to whether there are differences in the relative effect of ARB and CCB treatment according to these patient characteristics.

Overview

In patients with CKD, we found moderate strength of evidence that there is no difference between ACEI plus ARB combination therapy versus ACEI monotherapy for the outcome of all-cause mortality. We found insufficient strength of evidence that there is no difference between these treatments for ESRD. We found no significant difference between treatment groups in risk of stroke, CHF, doubling of serum creatinine, or progression from microalbuminuria to macroalbuminuria. Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events in some trials, and the heterogeneity of the study populations.

Description of Studies

Six trials met eligibility criteria and randomized participants with CKD (N=7,233, range of 54 to 3,988) to combination therapy with an ACEI plus an ARB versus ACEI therapy alone.^{75-77,103-105} One of the included reports was a post-hoc analysis performed within a subset of participants with CKD from a larger trial population that was not originally limited to subjects with CKD,¹⁰⁵ while a second report was a post hoc analysis from a larger trial that evaluated outcomes in multiple participant subgroups, including impaired GFR and albuminuria.⁷⁵ Detailed baseline characteristics are presented in Appendix Tables C17 and C18.

The mean age of study subjects was 65 years (range of study means 51–66; n=5 trials), and men constituted 83 percent (range 37 to 88; n=5 trials) of all participants randomized. Among the three trials that reported race/ethnicity, one was entirely comprised of Japanese participants,¹⁰⁴ one reported only that 91 percent of participants were white,¹⁰⁵ and, in third trial, 45 percent of participants were Hispanic, 34 percent were black, and 19 percent were white.¹⁰³ Two studies were conducted solely in the United States,^103,105 one study was conducted in Japan¹⁰⁴ one study was conducted in Turkey,⁷⁷ and two studies were multinational.^75,76 The mean or median study duration ranged from 30 weeks to 3.1 years. All studies but two^76,103 had followup durations of at least 1 year.

Renal Function

One post hoc analysis restricted inclusion to participants with GFR <60 ml/min/1.73m², by definition CKD stage 3 or worse,¹⁰⁵ while a second post hoc analysis reported results for a subgroup defined by GFR <60 ml/min/1.73m² as well as for a subgroup defined by albuminuria, the latter by definition could have included CKD stages 1–4.⁷⁵ Otherwise, no trial based study eligibility on CKD stage or reported baseline distribution of participants by CKD stage. Of the six trials, two required that participants have microalbuminuria,^76,77 two required that participants have macroalbuminuria or overt proteinuria,^103,104 and one reported a post hoc analysis of participants with either microalbuminuria or macroalbuminuria.⁷⁵ Among these trials that required participants to have albuminuria or proteinuria, one required that participants also have a normal creatinine,⁷⁷ three allowed participants to have abnormal levels for creatinine or creatinine clearance but mandated a maximally abnormal limit,^76,103,105 while one required that participants also had an elevated creatinine between 1.2 and 5 mg/dL.¹⁰⁴

Overall, four studies reported on some measure of proteinuria at baseline.^{75-77,103,104} One reported a mean 24 hour proteinuria of 1.7 g/d,¹⁰⁴ one reported a mean urinary albumin: creatinine ratio of 9.4 mg/mmol,⁷⁶ one reported a mean 24 hour albumin excretion rate of 260 mg,⁷⁷ and one study reported a mean urinary albumin:creatinine ratio of 907 mg/g.¹⁰³ The study by Anand reported only on dipstick proteinuria.¹⁰⁵ Several measures of renal function were reported by the studies, including a mean serum creatinine 1.5 mg/dL (range 1 to 3, n=3 trials),^77,103,104 a mean creatinine clearance of 96 ml/min/1.73m2 (range 65 to 112, n=3 trials),^76,77,103 and a mean eGFR of 50 (range 48 to 51, n=2 trials).^75,105

Baseline Comorbidities

Two of six trials were restricted to patients with diabetes,^77,103 including one limited to participants with type 2 diabetes.⁷⁷ Among the remaining trials, only two report data on diabetes prevalence, with 29 percent¹⁰⁵ and 74 percent⁷⁶ of study participants, respectively.

Three trials were restricted to participants with hypertension,^76,77,104 two trials excluded participants with hypertension,^75,103 while prevalence of hypertension in the remaining study was 15 percent.¹⁰⁵ Mean baseline blood pressures was 127/76 mmHg.

Three trials excluded participants with heart failure,^76,103,104 while one included only participants with heart failure.¹⁰⁵ Four trials excluded participants with a recent stroke or ischemic cardiac event. Prevalence of other cardiovascular disease was reported only in that heart failure was attributed to ischemic disease in 36 percent of participants in one trial,¹⁰⁵ and a history of MI or coronary artery procedure was reported in fewer than 10 percent of participants in a second study.^103,104

Study Quality

(Appendix Table C140)

Among six eligible trials, two were rated as good quality and four were rated as fair quality. Allocation concealment was adequate in three trials and unclear in the remaining studies. Four trials were double blinded.^{75,76,103,105} Two studies were not blinded.^77,104 For the outcomes presented here, only two studies analyzed results according to the intention-to-treat principle.^75,105 All studies adequately described reasons for study withdrawals. Withdrawals ranged from 5 to 24 percent (n=4 trials).

Results

Mortality

(Table 9, Appendix Table C19, and Appendix Figure C3)

Table 9

Pooled clinical and renal outcomes, ACE inhibitor plus ARB versus ACE inhibitor trials.

All-Cause Mortality

Overall, there was no significant difference in risk of all-cause mortality between CKD patients randomized to ACEI+ARB versus those allocated to ACEI alone (RR 1.03, 95% CI, 0.91 to 1.18). More than 99 percent of events occurred in only one trial.¹⁰⁵

Cardiovascular Mortality

No study reported data for cardiovascular mortality.

Summary

In patients with CKD, compared with ACEI monotherapy, assignment to combination ACEI+ARB therapy did not significantly reduce risk of all-cause mortality but was associated with significant reductions in risk of the one composite vascular outcome reported and in risk of progression from microalbuminuria to macroalbuminuria. Results suggested that combination treatment might reduce risk of doubling creatinine, but they did not achieve statistical significance. Too few events were reported for all other outcomes for the results to be clinically meaningful, including for stroke, MI, and ESRD. Reporting on study withdrawals and adverse effects was limited. No trial provided followup beyond 4 years.

Overview

In patients with CKD, we found insufficient evidence regarding whether there is a difference between ACEI+ARB combination therapy versus ARB monotherapy for all-cause mortality (no events) or ESRD (no data reported). Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events, and the heterogeneity of the study populations.

Description of Studies

Three trials met eligibility criteria and randomized participants with CKD (n=approximately 4,300) to combination therapy with an ACEI+ARB versus ARB therapy alone.^75-77 Baseline characteristics are presented in Appendix Tables C17, C18, and C25.

Among eligible trials, the mean age of study subjects was 57 years (range of study means 57 to 58; n=2 trials reporting), and men constituted 46 percent (range 37 to 69; n=2 trials reporting) of all participants. Ethnicity was not reported by any study. Two studies were multinational^75,76 and one was conducted in Turkey.⁷⁷ The mean or median study duration ranged from 30 weeks to 4.7 years. Two studies had a duration of followup of 1 year or greater.^75,77

Renal Function

One post hoc analysis reported results for a subgroup defined by GFR <60 ml/min/1.73m² as well as for a subgroup defined by albuminuria, the latter by definition could have included CKD stages 1–4 but did not state the total number of participants with CKD.⁷⁵ Otherwise, no trial based study eligibility on CKD stage or reported baseline distribution of participants by CKD stage. Of the three trials, two required that participants have microalbuminuria,^76,77 and one reported a post hoc analysis of participants with either microalbuminuria or macroalbuminuria.⁷⁵ Of the two trials that required participants to have albuminuria or proteinuria, one required that participants also have a normal creatinine,⁷⁷ and one allowed participants to have abnormal levels for creatinine or creatinine clearance but mandated a maximally abnormal limit.⁷⁶

Two studies reported on some measure of proteinuria at baseline.^76,77 One reported a mean urinary albumin:creatinine ratio of 9.4 mg/mmol,⁷⁶ and the other reported a mean 24-hour albumin excretion rate of 260 mg.⁷⁷ Two studies reported a mean creatinine clearance of 101 ml/min/1.73m2 (range 97 to 112, n=2 trials),^76,77 one reported a mean serum creatinine of 1.0 mg/dL,⁷⁷ and one reported a mean eGFR of 50 mL/min/1.73m2.⁷⁵

Baseline Comorbidities

One trial was restricted to patients with type 2 diabetes.⁷⁷ In the only other study that reported data, diabetes prevalence was 74 percent.⁷⁶ Two trials were restricted to participants with hypertension,^76,77 and one trial excluded participants with hypertension.⁷⁵ Mean baseline blood pressures was 152/90 mmHg. One trial excluded participants with heart failure,⁷⁶ but otherwise the presence of cardiovascular disease at baseline was not reported in any study.

Study Quality

(Appendix Table C140)

Among the three trials, one was rated good quality and two were rated fair quality. Allocation concealment was adequate in two studies and unclear the third study. Two studies were double blinded.^75,76 The other study was not blinded.⁷⁷ For the outcomes presented here, only one study analyzed results according to the intention-to-treat principle.⁷⁵ All studies adequately described reasons for study withdrawals. Withdrawals ranged from 12 to 14 percent (n=2 trials).

Results

Mortality

(Table 10, Appendix Table C19, and Appendix Figure C4)

Table 10

Pooled clinical and renal outcomes, ACE inhibitor plus ARB versus ARB trials.

Of the three studies, only one reported on mortality during the trial.⁷⁶ In this study of 86 patients with CKD there were no deaths.

Summary

In individuals with CKD, trials comparing ACEI+ARB combination therapy versus ARB alone reported few or no clinical outcomes, including no deaths in one trial reporting this outcome. No trials reported data on MI, stroke, CHF, ESRD, doubling of serum creatinine, or any composite vascular or renal outcome. Though trials reported data for progression from microalbuminuria to macroalbuminuria, all had either few events or errors in reporting that impeded interpretation. Reporting on study withdrawals and adverse effects was limited. No trial provided followup beyond 5 years.

Overview

In patients with CKD, we found moderate strength of evidence that combination ACEI and ARB treatment did not reduce risk of all-cause mortality compared with ACEI or ARB monotherapy. We found low strength of evidence that that there was no difference in risk of ESRD between treatment groups. There was no significant difference between treatment groups for risk of cardiovascular mortality, a composite vascular outcome, doubling of serum creatinine, or a composite renal outcome defined as doubling of serum creatinine or ESRD. Our confidence in these estimates is limited as data were drawn from only one trial.

Description of Studies

One trial met all eligibility criteria and randomized participants with CKD to either ACEI monotherapy, ARB monotherapy, or combined ACEI plus ARB treatment. Detailed baseline characteristics are presented in Appendix Tables C17 and C18. From the larger ONTARGET trial (n=23,422), this post hoc analysis was limited to 8,933 participants with either eGFR ≤60 ml/min/1.73m² or albuminuria and reported results only for combination therapy versus the pooled monotherapy arms.

In this post hoc analysis,⁹⁹ 2,943 participants were randomized to ramipril 10 mg/d plus telmisartan 80 mg/d versus either ramipril or telmisartan monotherapy (n=5,990). The mean age of subjects was 68.2 years, and men constituted 68.0 percent of patients randomized. Race was reported as 70 percent European and 16 percent Asian. This trial was multinational.⁹⁹

Renal Function

Participants were excluded if they had a serum creatinine >3 mg/dL with no restriction on albuminuria. At baseline, mean serum creatinine was 1.1 mg/dL, mean eGFR was 73.6 ml/min/1.73m², and mean urinary albumin:creatinine ratio was 129.1 mg/g. In this post hoc analysis, 5,623 participants (62.9 percent) had eGFR <60 ml/min/1.73m², 2,631 (29.5 percent) had isolated microalbuminuria, and 679 (7.6 percent) had isolated macroalbuminuria.

Baseline Comorbidities

Participants were included if they were >55 years with established cardiovascular disease or with diabetes associated with end organ damage. Seventy-seven percent of participants had hypertension with a mean blood pressure of 144/82 mm Hg. Diabetes was present in 49 percent of participants. The prevalence of cardiovascular disease was 70 percent, including 45 percent with a previous myocardial infarction, 20 percent with a prior stroke and 17 percent with prior peripheral vascular disease.

Study Quality

(Appendix Table C140)

The study was rated good quality. It was double blinded, performed analyses using the intention-to-treat principle, and adequately described study withdrawals and reasons for withdrawals. Study withdrawals occurred in 29 percent of participants.

Results

Summary

In individuals with CKD, compared with ACEI or ARB monotherapy, assignment to ACEI plus ARB combination treatment was associated with a similar risk of all-cause mortality, cardiovascular mortality, a composite vascular outcome, ESRD, and doubling of serum creatinine. However, there was a significant interaction between baseline category of albuminuria and the association of treatment assignment on risk of all cause mortality. While those with normoalbuminuria had an increased risk of death with combination therapy, those with macroalbuminuria demonstrated a trend towards a decreased risk of mortality. Results were limited in that they are derived from only one trial. Adverse effects were more likely in patients randomized to combination ACEI and ARB therapy compared with monotherapy with either an ACEI or ARB.

Overview

In comparing ACEI plus ARB versus ACEI plus aldosterone antagonist, we found insufficient evidence regarding whether there is a difference between treatments in risk of mortality or ESRD. Our confidence in these estimates is limited because data are drawn from only one trial and because of the small number of clinical events.

Description of Study

One trial met all eligibility criteria and randomized 54 participants with CKD and taking an ACEI (lisinopril 80 mg/day) to the addition of either an ARB (losartan) or of an aldosterone antagonist (spironolactone).¹⁰³ A third arm of the trial, discussed elsewhere, involved addition of placebo to lisinopril. Detailed baseline characteristics are presented in Appendix Table C26.

The mean age of trial participants was 52 years, and males constituted 49 percent of study subjects. Most patients (55 percent) were Hispanic, with an additional 28 percent black, 15 percent white, and 2 percent Native American. The study duration was 48 weeks.

Renal Function

Patients were included if they had macroalbuminuria, defined as a urinary albumin to creatinine ratio of 300 mg/g or higher despite treatment with an ACEI or ARB for at least 3 months prior to study entry. Females with a serum creatinine greater than 3.0 mg/dl and males with a serum creatinine greater than 4.0 mg/dl were excluded. Baseline renal function for trial participants included mean urine albumin to creatinine ratio of 997.4 mg/g, mean baseline serum creatinine of 1.8 mg/dl, and mean creatinine clearance of 58.0 ml/min.

Baseline Comorbidities

All study participants were required to have hypertension, with a systolic blood pressure on antihypertensive treatment of greater than 130 mm Hg. Mean baseline blood pressure was 134.0/72.5 mm Hg. Trial participants also were required to have diabetes, but with an HbA_1c at or below 11 percent. The mean HbA_1c at baseline was 7.5 percent. Patients with any history of heart failure, or with a stroke or MI in the past 12 months were excluded. A history of either MI, coronary artery bypass grafting, or percutaneous transluminal coronary angioplasty was reported by 7.5 percent of the patients.

Study Quality

(Appendix Table C140)

The trial was rated fair quality. The method used for treatment allocation was not clearly described. The study was double blinded; however, the analysis was not completed using intention-to-treat principles. From the 54 randomized participants, 35.2 percent withdrew, with reasons for withdrawals adequately explained.

Results

Summary

(Appendix Table C140)

In this trial of diabetic, hypertensive CKD patients already on ACEI, there appeared to be no difference between subjects randomized to additional ARB versus additional aldosterone antagonist for the outcome of doubling of baseline creatinine. Few or no results were reported with respect to risk of all-cause mortality, MI, stroke, CHF, or ESRD. Withdrawals due to adverse events appeared possibly were more likely with ACEI combined with aldosterone antagonist. Results were limited in that they are based on only one small trial that reported few clinical endpoints. Further, the withdrawal rate was high and followup duration was less than 1 year.

Overview

In patients with CKD, we found insufficient evidence regarding whether there is a difference between ACEI monotherapy, CCB monotherapy, and ACE+CCB combination therapy for reducing risk of mortality or any clinical vascular or renal outcome.

Description of Study

We identified one trial that met all eligibility criteria. Patients with CKD were randomized to receive ACEI and CCB combined, ACEI alone, or CCB alone.⁸⁴ Detailed baseline characteristics are presented in Appendix Table C30.

After randomization to CCB (amlodipine at 5 to 15 mg/day), ACEI (fosinopril at 10 to 30 mg/day), or the combination, participants began a three month dose titration phase to a goal diastolic blood pressure less than 90 mm Hg for the monotherapy groups and less than 85 mm Hg for the combination therapy group. Patients judged nonresponders or who complained of side effects during the titration phase were withdrawn (n=144 overall, with no data reported by treatment group) and were not entered into the subsequent treatment phase. Study followup during the treatment phase was 4 years.

Renal Function

Study participants were required to have microalbuminuria, defined by UAER 30 to 300 mg/24 hours. For the patients entered in the treatment phase into either the ACEI plus CCB group or the ACEI alone group, the mean baseline serum creatinine was 1.0 mmol/L, mean creatinine clearance was 89.9 mg/min, and mean UAER was 97.9 μg/min. For the patients entered into either the ACEI plus CCB group or the CCB alone group, baseline characteristics were similar. Mean serum creatinine was 1.0 mg/dL, creatinine clearance was 89.3 mg/min, and UAER was 96.6 μg/min.

Baseline Comorbidities

Study participants were required to have hypertension (diastolic blood pressure 90 to 110 mm Hg) and type 2 diabetes (well controlled without insulin). Patients with a history of coronary heart disease, CHF, MI, or stroke were excluded. For patients entered in the treatment phase into either the ACEI plus CCB group or the ACEI alone group, mean baseline blood pressure was 160/99 mm Hg and baseline HbA_1c was 7.1 percent. For the patients entered into either the ACEI plus CCB group or the CCB alone group, mean baseline blood pressure was 161/99 mm Hg and HbA_1c was 7.0 percent.

Study Quality

(Appendix Table C140)

The trial was rated fair quality. Concealment of treatment allocation was adequate. This open-label study did not perform analysis according to the intention-to-treat principle. In addition to participants excluded during the dose titration phase, additional participants withdrew during treatment, resulting in 47 percent total withdrawals.

Results

Summary

In one study of patients with CKD, hypertension, and diabetes without a history of cardiovascular disease, few participants died or experienced clinical vascular or renal events. There was no significant difference for any of these outcomes between ACEI plus CCB versus either ACEI monotherapy or CCB monotherapy groups, but wide confidence intervals around all estimates could not exclude large between-group differences. Adverse events were infrequent and risk did not appear significantly different between treatment groups. There were no data on clinical renal outcomes. The study was limited by its exclusion of one-third of randomized participants from the analyses, the large number of additional withdrawals during treatment, the small number of clinical vascular events, and the absence of any reported clinical renal outcomes.

Overview

In patients with CKD we found insufficient evidence regarding whether there is any difference between combination therapy with an ACEI and a diuretic and combination therapy with an ACEI and CCB for risk of mortality or ESRD. Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events, and the heterogeneity of the study populations.

Description of Study

Two trials met all eligibility criteria. One trial randomized 332 patients with CKD to ACEI plus diuretic versus ACEI plus CCB.¹⁰⁶ The second trial reported results from 1,093 patients with CKD enrolled in a trial of 11,506 patients with hypertension and randomized to ACEI plus diuretic or ACEI plus CCB.^107,108 Detailed baseline characteristics are presented in Appendix Table C34.

In the first trial, patients randomized to the ACEI plus CCB group received benazepril and amlodipine. Those randomized to the ACEI plus diuretic group received benazepril and hydrochlorothiazide (HCTZ). Doses were titrated to reach a blood pressure target below 130/80 mm Hg, and additional antihypertensives were added as needed with the exception of ACEI, ARB, or aldosterone receptor antagonists. The mean age of study participants was 58 years, and men constituted 65 percent of subjects. Patients were mostly white race (60 percent), with blacks comprising another 26 percent of participants. Study followup duration was 12 months.¹⁰⁶

The protocol was similar in the second trial. The mean age of the subgroup with CKD was 70.9 years and 67 percent were men. Approximately 77 percent of the patients were white; 20 percent were black. The followup period was 2.9 years.¹⁰⁷

Renal Function

In the first study, participants were required to have either microalbuminuria or macroalbuminuria (UACR 20 to 500 mg/g), and to have serum creatinine ≤1.3 mg/dl for women and ≤1.5 mg/dl for men. In data available only for the 304 patients who completed followup, the median UACR was 60.6 mg/g, and the median estimated GFR was 90.6 ml/min/1.73m².¹⁰⁶

Patients eligible for the second trial had hypertension and were at high risk for cardiovascular events based on evidence of cardiovascular or renal disease or target organ damage. Criteria for renal disease included serum creatinine >1.5 mg/dL for women or >1.7 mg/dL for men or macroalbuminuria (UACR > 300 mg/g or > 200 mg/g if receiving an ACEI or aldosterone receptor blocker).¹⁰⁷

Baseline Comorbidities

Trial participants in the first study were required to have hypertension (mean systolic blood pressure 130 to 179 mm Hg and diastolic blood pressure 80 to 109 mm Hg). Mean baseline blood pressure was 151/88 mm Hg. Individuals with CHF, type 1 diabetes or uncontrolled type 2 diabetes were excluded, as were those with a cardiovascular disease event in the past 6 months (MI, stroke, transient ischemic attack, coronary artery bypass grafting, or percutaneous transluminal coronary angioplasty).¹⁰⁶

The second study enrolled patients age 60 and older with systolic blood pressure of 160 mm Hg or higher (or currently on antihypertensive therapy). Within the participants with CKD, mean systolic blood pressure was 145 mm Hg and 59 percent had diabetes. Vascular and hypertension-related reasons for study exclusion included: current evidence for angina pectoris, history of symptomatic heart failure or evidence of left ventricular ejection fraction <40 percent, acute MI or revascularization in the prior month, stroke or other ischemic cerebrovascular events in the prior 3 months, or hypertension that is severe, refractory to treatment, or known to have a secondary cause.¹⁰⁷

Study Quality

(Appendix Table C140)

The first study was rated fair quality. Treatment allocation concealment was adequate. The study was double blinded. Analysis was not according to the intention-to-treat principle. Withdrawals were 19 percent.¹⁰⁶ The second study was rated good quality. Treatment allocation was adequate, the study was double-blinded, analysis was by intention to treat, and withdrawals were adequately reported.¹⁰⁷

Results

Summary

In one study of patients with CKD defined by albuminuria, hypertension with no recent cardiovascular events, and no heart failure, there was no significant difference between patients allocated to ACEI plus diuretic versus ACEI plus CCB in risk of mortality, or of unspecified “cardiac disorders” or “vascular disorders.” However, there were very few events for any of these outcomes, and confidence intervals around risk estimates were wide. No data were reported for MI, stroke, CHF, ESRD, or any transparently defined composite vascular or renal outcome. The risk of progression from microalbuminuria to macroalbuminuria appeared similar between treatment groups, but again confidence intervals were wide. A second study reported no difference between treatment groups for two composite renal outcomes assessed in patients with CKD and diabetic nephropathy. Risk of edema appeared somewhat higher in the ACEI plus CCB group in the first study and was significantly higher in the second study. Cough and dizziness appeared somewhat higher in the ACEI plus CCB groups in both studies.

Overview

In patients with CKD, we found insufficient evidence regarding whether treatments differ for risk of mortality or ESRD. There was a significantly reduced risk of achieving the composite vascular outcome in the ACEI plus diuretic group. Our confidence in these estimates is limited because data are drawn from only one trial and there were few reported clinical events.

Description of Study

One trial met all eligibility criteria and randomized 481 patients with CKD to receive either an ACEI and diuretic or a different ACEI alone.¹⁰⁹ Detailed baseline characteristics are presented in Appendix Table C40 for the 457 patients who took at least one dose of study medication and who had albuminuria measured at least once during treatment.

The ACEI plus diuretic group received perindopril and indapamide, while the ACEI monotherapy group received enalapril. Both groups were titrated to achieve a blood pressure goal of less than 140/90 mm Hg. The mean age of subjects was 59 years and 61 percent of study participants were male. Ninety-one percent of the study participants were white. Mean followup duration was 10.7 months.

Renal Function

Study participants were required to have albuminuria (UAER of 20 to 499 μg/min) and to have a serum creatinine less than approximately 1.6 mg/dl. Mean UAER was 92.1μg/min and the mean urine albumin/creatinine ratio was 8.5 mg/mmol.

Baseline Comorbidities

Study participants were required to have type 2 diabetes with HbA_1c less than 9 percent in the 3 months prior to the study. Mean baseline HbA_1c was 7.2 percent. Participants also were required to have hypertension (systolic blood pressure 140 to 179 mm Hg and diastolic blood pressure less than 110 mm Hg). Mean baseline blood pressure was 158/93 mm Hg.

Study Quality

(Appendix Table C140)

Study quality was rated fair. Though the study was double blind, allocation concealment was unclear and analysis was not intention-to-treat. Withdrawals were 23 percent and were not adequately described.

Results

Summary

In patients with CKD, hypertension, and type 2 diabetes, a combination of ACEI and diuretics was associated with a significant reduction in risk of serious cardiovascular events compared with treatment with ACEI monotherapy. The risk of adverse events was similar in the two groups. Results were limited because there were no data on mortality or specific cardiovascular or renal outcomes. Further, analysis was not based on intention-to-treat principles and the study withdrawals or dropouts were not adequately described. Mean followup for this study was 10.7 months.

Overview

In patients with CKD, we found a low strength of evidence that combination therapy with an ACEI and a diuretic did not significantly reduce mortality compared with placebo. We found insufficient evidence that there was no difference between treatments in risk of ESRD.¹¹⁰ We found no significant difference between treatment groups for risk of cardiovascular mortality; risk of major cardiovascular events, major coronary events, or major cerebrovascular events; or risk of a composite renal outcome. Our confidence in these estimates is limited because data are drawn from only one trial and there were few reported clinical events for several outcomes.

Description of Study

A subgroup analysis of a larger trial of patients with type 2 diabetes met all eligibility criteria. The analysis included 2,482 patients with stage 1 or 2 CKD and 2,044 patients with stage 3 CKD who had been randomized to receive either an ACEI and diuretic or placebo¹⁰⁹ Detailed baseline characteristics are presented in Appendix Table C44).

The ACEI plus diuretic group received perindopril and indapamide. The mean age of subjects in the subgroup analysis was 67 years and 53 percent of study participants were male. Race/ethnicity data were not reported. Mean followup duration was 4.3 years.

Renal Function

For inclusion in this post hoc analysis, study participants were required to have CKD. As noted above, 2,482 participants had CKD stages 1–2, and 2,044 had CKD stage 3 or worse. Mean urine albumin-creatinine ratio was 48.1 μg/mg, and mean eGFR was 70.7 ml/min/1.73m².

Baseline Comorbidities

Study participants were required to be age 55 or older, diagnosed with type 2 diabetes at age 30 or older. Mean baseline HbA_1c was 7.7 percent. A history of macrovascular disease was reported in 34.7 percent of participants, and 12.8 percent reported a history of MI, and 10.8 percent reported a history of stroke. Mean baseline blood pressure was 148/81 mm Hg.

Study Quality

(Appendix Table C140)

Study quality was rated as good. Treatment allocation concealment was adequate, the study was double-blind, the study was analyzed as intention to treat, and withdrawals were adequately reported.

Results

Summary

In a subgroup of patients with CKD, type 2 diabetes, and at high risk for vascular events, a combination of ACEI and diuretic was not associated with a significant reduction in risk of mortality or clinical cardiovascular or renal events compared with treatment with placebo. Risk for the composite renal outcome were significantly reduced with ACEI plus diuretic in patients with CKD stages 1-2, but given that this finding was not observed in analyses stratified by eGFR or albuminuria, and tests for interaction by CKD strata were not significant, the probability of this being a chance finding is substantial.

Overview

We found insufficient strength of evidence regarding whether there is any difference between higher and lower dose ARB in risk of mortality or ESRD. We found that higher dose ARB significantly reduces risk of conversion from microalbuminuria to macroalbuminuria. Our confidence in these estimates is limited because data are drawn from only one trial and there were few reported clinical events.

Description of Studies

Three trials met eligibility criteria and randomized participants with CKD to at least two different fixed doses of ARB treatment. One trial randomized 269 participants to three different doses of candesartan (16 mg/day, 64 mg/day, and 128 mg/day.¹¹¹ A second trial randomized participants to telmisartan 40 mg/day versus 80 mg/day).⁹⁵ A third trial randomized participants to irbesartan 150 mg/day versus 300 mg/day.⁹⁶ Detailed baseline characteristics are presented in Appendix Tables C51 and C52.

Mean age of study participants was 58.5 years. Men constituted 73 percent of all study participants (2 trials reporting). In two studies reporting race/ethnicity, 91 percent of participants were white. The studies were conducted in Canada, Japan, and multinational. Median followup durations ranged from 7 months to 2 years.

Renal Function

Two trials required participants to have microalbuminuria, defined in one study by urinary albumin:creatinine ratio of 100-300 mg/g and in the other study by repeated urinary albumin excretion rate of 20-200 micrograms/minute. Both of these studies excluded participants with serum creatinine greater than 1.5 mg/dl in men and either greater than 1.1 or 1.3 mg/dl in women. The third study required repeated urinary protein excretion of at least 1 g/day and excluded participants with serum creatinine >3.4 mg/dl. One trial reported a mean serum creatinine of 1.44 mg/dl, eGFR of 52.0 ml/min/1.73m² and proteinuria of 2.83 g/day. A second trial reported a mean serum creatinine of 1.05 mg/dl, creatinine clearance of 109 ml/min/1.73m² and albuminuria of 55.9 micrograms/day. The third trial provided no data on baseline renal function. Mean serum creatinine in two trials reporting was 1.21 mg/dl.

Baseline Comorbidities

All participants in two trials were required to have hypertension, and all participants in two trials were required to have diabetes. One trial each provided no information on prevalence of diabetes or hypertension. Mean baseline blood pressure was 142/82 mm Hg. In one trial reporting, prevalence of cardiovascular disease was rare.

Study Quality

(Appendix Table C140)

Two trials were rated as fair quality and one was rated as good quality. All trials were double blinded but only one reported using adequate concealment methods for treatment allocation. Results were analyzed according to the intention-to-treat principle in two studies and withdrawal and dropouts were adequately described in all trials. Withdrawals ranged from 12 to 14 percent in two trials reporting.^96,111

Results

Summary

In these three small trials of CKD patients with albuminuria, high dose ARB treatment was associated with a significant reduction in risk of conversion from microalbuminuria to macroalbuminuria. Trials reported very few deaths and no other vascular or renal outcomes. Withdrawals and adverse events did not appear higher in the higher dose ARB groups compared with the low dose group in either trial reporting these data.

Overview

In patients with CKD, we found a low level of evidence that telmisartan significantly reduces risk of all-cause mortality compared with losartan and a low level of evidence that there is no difference in risk of all-cause mortality between telmisartan and valsartan. In addition, we found a low level of evidence that there is no difference in risk of ESRD between telmisartan and losartan and insufficient (no) evidence regarding whether risk for ESRD differs between telmisartan and valsartan. Our confidence in these estimates is limited by the small number of trials reporting different outcomes and the small number of clinical events.

Description of Studies

Two trials met all eligibility criteria and randomized 1,745 participants with CKD to treatment comparing two different ARBs.^112,113 Detailed baseline characteristics are presented in Appendix Tables C51 and C52.

One trial randomized 860 participants to telmisartan versus losartan.¹¹² The second trial randomized 885 participants to telmisartan versus valsartan.¹¹³ The mean age of study participants was 61 years (range 60 to 61) and men constituted 63 percent (range 62 to 64) of all participants studied. Both trials reported race/ethnicity, and 63 percent of participants were white, 30 percent were Asian, and 7 percent were black. Both were multinational studies and followup duration ranged from 10.7 to 12 months.

Renal Function

Both trials required that participants have overt proteinuria, and allowed subjects with either normal or elevated serum creatinine levels, setting an upper abnormal limit. In one trial, participants had to have proteinuria of 900 mg/24 hours or greater and a serum creatinine of 3.0 mg/dl or less.¹¹³ This study reported a mean baseline proteinuria of 2.78 g/day. In the second trial, patients had to have a urine protein-creatinine ratio at least 700 mg/g and a serum creatinine <3.0 mg/dl in women and <3.2 mg/dl in men.¹¹² At baseline, this study reported a mean urine protein-creatinine ratio of 1,991 mg/g, mean urine albumin-creatinine ratio of 1,394 mg/g, and a mean serum creatinine of 1.55 mg/dl. For both trials considered together, the mean baseline GFR was 53.2 ml/min/1.73m² (range 49.6 to 56.6).

Baseline Comorbidities

Both trials were restricted to patients with type 2 diabetes and hypertension. At baseline, mean HbA_1c was 7.85 percent and mean blood pressure was 146/81 mm Hg. One trial excluded patients with a history of “clinically significant” heart disease or stroke, which was presumed to exclude patients with a history of coronary artery disease, MI, or congestive heart failure.¹¹² The second study excluded patients with any history of congestive heart failure and those with a “recent acute cardiovascular event.”¹¹³ It did not report data on prevalence of coronary artery disease, MI, or stroke.

Study Quality

(Appendix Table C140)

Both trials were rated as fair quality. Allocation concealment was unclear in both studies. Both were double blinded. One study analyzed results according to the intention-to-treat principal and adequately described the 19.1 percent of subjects who withdrew from the study.¹¹³ The second study did not include an intention-to-treat analysis and did not adequately describe the 18.4 percent of participants who withdrew.¹¹²

Results

Mortality

(Table 11, Appendix Table C53, and Figure C12)

Table 11

Pooled clinical and renal outcomes, ARB versus ARB trials.

All-Cause Mortality

Among these patients with CKD, those randomized to telmisartan had a significant 84 percent reduction in risk of all-cause mortality compared with those randomized to losartan (0.5 versus 2.9 percent; RR 0.16, 95% CI, 0.04 to 0.71).¹¹² However, the risk of all-cause mortality was higher, although not significantly so, for patients assigned telmisartan versus valsartan (3.5 versus 1.9 percent; RR 1.88, 95% CI, 0.81 to 4.39).¹¹³ Results from these trials were not pooled as the results suggested large differences in the direction of the effect of losartan and valsartan compared with telmisartan. This was reflected in the I² of 75 percent.

Cardiovascular Mortality

One study reported no significant difference between the telmisartan or valsartan treatment groups for cardiovascular mortality (RR 1.34, 95% CI, 0.47 to 3.82).¹¹³

Summary

In individuals with CKD, type 2 diabetes mellitus, and hypertension, compared with losartan, telmisartan was associated with a significant 84 percent reduction in all-cause mortality and a borderline significant 40 percent reduction in cardiovascular morbidity or cardiovascular mortality. In addition, telmisartan was associated with a nonsignificant 41 percent reduction in risk of the composite endpoint of doubling of serum creatinine, ESRD, or death, and with fewer serious adverse events. However, compared with valsartan, CKD patients randomized to telmisartan appeared to have a nonsignificantly higher risk of all-cause and cardiovascular mortality, stroke, and CHF, but a lower risk of MI. There was little difference in the composite vascular outcome or in any of the adverse event measures recorded. Results were limited by relatively small sample size and number of clinical events, with most outcomes reported only in one trial, and heterogeneity in comparison groups and outcomes that prevented statistical pooling. This resulted in there being low statistical power to determine if even large differences in outcomes between treatment groups were statistically significant. Results also were limited in that there were no studies that directly compared losartan and valsartan. Because no trial was longer than 1 year, it was not possible from these studies to determine the longer term effects of telmisartan versus losartan or valsartan.

Overview

In patients with CKD, we found insufficient evidence regarding whether there is a difference between ACEI plus aldosterone antagonist versus ACEI alone for risk of all-cause mortality or ESRD. Our confidence in these estimates is limited because data are drawn from only one trial and there were few reported clinical events.

Description of Study

We identified one trial that met all eligibility criteria and randomized 54 patients with CKD being treated with ACEI to either additional aldosterone antagonist or placebo.¹⁰³ Detailed baseline characteristics for this comparison are presented in Appendix Table C59. Data regarding a third treatment arm, the addition of ARB to ACEI are discussed separately.

Mean age of randomized participants was 51 years, and men constituted 46 percent of the subjects. Fifty-four percent of patients were Hispanic, 32 percent were black, 11 percent were non-Hispanic white, and 3 percent were Native American. Mean study followup duration was 11.1 months.

Renal Function

For inclusion, participants were required to have macroalbuminuria (UACR at least 300 mg/g) despite run-in treatment. Women with serum creatinine above 3.0 mg/dL and men with creatinine above 4.0 mg/dL were excluded from the study. Among randomized participants, mean baseline UACR was 1,006 mg/g, mean serum creatinine was 1.6 mg/dL, and mean creatinine clearance was 62 ml/min.

Baseline Comorbidities

All study participants were required to be hypertensive prior to screening but were treated with diet and ACEI during a pre-randomization 3 month run-in period to a target systolic blood pressure of less than 130 mm Hg. Mean blood pressure at randomization was 132/74 mm Hg study participants also were required to have diabetes, and mean HbA_1c was 7.8 percent. Patients with a history of heart failure, and those with a stroke or MI within 12 months were excluded from the trial. A history of either MI, CABG, or PTCA was reported by 9.3 percent.

Study Quality

(Appendix Table C140)

Study quality was rated fair. The trial was double blinded, but allocation concealment was unclear. While the overall study analysis was not by intention-to-treat, this pertained to exclusion from analyses of a single participant randomized into the ACEI plus ARB treatment group that is not the focus of this section of the report. Withdrawals were 30 percent.

Results

Summary

In this small, short duration study of CKD patients with macroalbuminuria, hypertension, and diabetes, but with no history of heart failure and with a low prevalence of other cardiovascular disease, there were no deaths and very few cardiovascular outcomes. Differences in individual cardiovascular outcomes were not statistically significant. Participants in the ACEI plus aldosterone group appeared to be at higher risk for adverse events leading to discontinuation of treatment and study withdrawal. Results were limited by the short study duration, and small number of individual clinical vascular events. Also, no clinical renal outcomes data were reported.

Overview

In patients with CKD, there was insufficient evidence regarding whether, in comparison to treatment with ACEI or ARB plus placebo, treatment with ACEI or ARB plus aldosterone antagonist reduces mortality or ESRD. Our confidence in these estimates is limited because data are drawn from only one trial and there were few reported clinical events.

Description of Study

One trial met all eligibility criteria and randomized 59 participants with CKD and taking ACEI or ARB at baseline to the addition of an aldosterone antagonist versus addition of placebo.¹¹⁴ Detailed baseline characteristics are presented in Appendix Table C63.

Participants using an ACEI or ARB in recommended dosages for at least 1 year were randomized to addition of the aldosterone antagonist, spironolactone, 50 mg daily versus placebo. Mean age in study participants was 52 years and 66 percent of subjects were men. Study followup duration was 1 year.

Renal Function

Eligible participants were required to have albuminuria, defined as either 24 hour urinary albumin excretion greater than 300 mg or UACR greater than 20 mg/mmol. Mean serum creatinine was 98.2 μmol/l, mean UACR was 81 mg/mmol, and mean protein-to-creatinine ratio was 128.5 mg/mmol. The mean estimated GFR was 70.5 ml/min/1.73m².

Baseline Comorbidities

All participants had type 2 diabetes. Patients with MI or stroke within the past 3 months or with unstable angina pectoris were excluded. Mean blood pressure was 146/81 mm Hg and mean HbA_1c was 8.1 percent.

Study Quality

(Appendix Table C140)

Study quality was rated fair. Treatment allocation concealment was adequate. The study was double blinded. Analysis was not performed according to the intention to treat principle. Withdrawals were 11.9 percent and were adequately described.

Results

Summary

In one trial in patients with CKD and diabetes, already on ACEI or ARB, there was no significant difference in risk of all-cause or cardiovascular mortality between those randomized to addition of aldosterone antagonist versus placebo. No data were reported for other vascular or clinical renal outcomes. Results were limited in that they were based on only one small study with low statistical power for clinical events, which do not appear to have been a priori study outcomes.

Overview

We found low strength of evidence that in patients with heart failure and CKD who are on optimal medical therapy for their heart failure, treatment with BB significantly reduced risk of all-cause mortality. We found insufficient evidence in this population regarding whether there is a difference between BB and placebo regarding risk of ESRD as neither trial reported ESRD outcomes. Participants with CKD assigned to BB had a significantly lower risk of CHF complications, MI, or cardiac death. Our confidence in these estimates is limited by the small number of trials reporting different outcomes and the small number of clinical events.

Description of Study

We identified two trials that met all eligibility criteria and randomized participants with CKD to BB versus placebo.^115,116 One study was a post hoc subgroup analysis of 1,469 subjects with eGFR ≤60 ml/min/1.73m² from the larger MERIT-HF heart failure trial (n=3,991)¹¹⁵ The second study was a post hoc subgroup analysis of 704 patients with eGFR <55.5 ml/min/1.73m² from the larger SENIORS heart failure trial (n=2,135). Detailed baseline characteristics are presented in Appendix Table C66.

All participants in MERIT-HF were required to have been on optimum heart failure therapy consisting of any combination or diuretics and an ACEI, with hydralazine, long acting nitrate, or ARB if an ACEI was not tolerated. Patients then were randomized to the BB, metoprolol XL/CR versus placebo. At baseline, 88 percent of the patients were taking an ACEI and 94 percent were taking diuretics. The mean age of study participants was 68 years and 68 percent of subjects were male. No data were reported on race/ethnicity in this multinational study. Study followup duration was 1 year.

Participants in the SENIORS trial had a documented clinical history of heart failure and were receiving optimal standard therapy. Patients were randomized to receive either BB (nebivolol) or placebo. Baseline use of other medications was not reported according to eGFR levels but it was noted that among participants with poorer renal function, more were taking diuretics and ARB and fewer were taking ACEI. The mean age of the CKD subgroup was 77 years and 59 percent were male. The study was conducted in 11 European countries, but no race/ethnicity data were reported. Mean followup was 21 months.

Renal Function

For inclusion in the post hoc analysis of the MERIT-HF trial, participants were required to have eGFR ≤60 ml/min/1.73m². There were 976 participants with eGFR 45 to 60 ml/min/1.73m², and 493 with eGFR <45 ml/min/1.73m². In these two strata combined, mean GFR was 48 ml/min/1.73m² and mean serum creatinine was 1.5 mg/dL. For the post hoc analysis of the SENIORS trial, tertiles of eGFR were created. In the tertile with eGFR <55.5 ml/min/1.73m², the mean eGFR was 43 ml/min/1.73m² and mean serum creatinine was 1.6 mg/dL.

Baseline Comorbidities

All participants in the MERIT-HF and SENIORS trials were required to have symptomatic or documented heart failure. Among participants in the MERIT-HF post hoc analysis, diabetes was reported for 29 percent, a history of hypertension for 49 percent, and a history of myocardial infarction for 55 percent. Mean baseline blood pressure was 130/77 mm Hg. Among participants in the SENIORS post hoc analysis, diabetes was reported for 29 percent, and 46 percent had a history of myocardial infarction. Mean baseline blood pressure was 134/78 mm Hg.

Study Quality

(Appendix Table C140)

Study quality was rated as good for one trial and fair for one trial. Concealment of treatment allocation in both double-blind trials was adequate. Analyses were performed according to the intention-to-treat principle in the MERIT-HF trial; however, 23 randomized patients were excluded from the SENIORS trial post hoc analysis. No data on withdrawals were reported for the CKD subgroups.

Results

Summary

In two post hoc analyses, patients with well controlled heart failure and CKD who were randomized to BB versus placebo had a significantly lower risk of all-cause mortality. One trial also reported significantly lower risks of hospitalizations for CHF, CHF deaths, and of the composite vascular outcomes of all cause mortality or CHF hospitalization and of cardiac death or nonfatal MI. Analyses stratified by eGFR subgroup suggested that the relative benefit of BB versus placebo may be greatest in patients with eGFR <45 ml/min/1.73m² (MERIT-HF) or eGFR < 55.5 ml/min/1.73m² (SENIORS), though the statistical tests for interaction by eGFR strata did not approach statistical significance. Results were limited in that both studies were post hoc subgroup analyses, there were no measures of albuminuria available, and no clinical renal outcomes and little adverse events data were reported. Because trial followup was a mean of 21 months or less, longer term effects of BB monotherapy versus placebo in this population cannot be determined from these data.

Overview

In patients with CKD, we found low strength of evidence that there is no difference in risk of all-cause mortalityor ESRD between participants randomized to CCB versus placebo. In participants randomized to CCB versus placebo, there was a statistically significant reduction in risk of MI and conversion from microalbuminuria to macroalbuminuria, but there was no significant difference between treatment groups for the outcomes of cardiovascular mortality, stroke, doubling of baseline creatinine, or any composite vascular or renal outcomes. Our confidence in these estimates is limited by the small number of trials reporting different outcomes and the small number of clinical events.

Description of Studies

Two trials met all eligibility criteria and randomized 1,226 participants (range 90 to 1,136) with CKD to CCB versus placebo.^55,97,117 Detailed baseline characteristics are presented in Appendix Tables C71 and C72.

The larger trial, IDNT, randomized 1,136 hypertensive, type 2 diabetic individuals to amlodipine versus placebo.^97,117 This trial also included an ARB treatment arm discussed elsewhere in this report. Mean age of study participants was 59 years, 67 percent of all subjects were male and 71 percent of participants were white. The study was multinational and followup duration was 2.6 years.

A second trial randomized 90 normotensive, type 1 diabetic subjects to nifedipine versus placebo,⁵⁵ and also included an ACEI treatment arm discussed elsewhere in this report. After randomization, 22 participants were excluded for having UAER outside the 20 to 200 μg/min range and an additional seven for adverse clinical events. Baseline data were only reported on these 61 participants. Within these participants, mean age was 37 years and 70 percent of all subjects were male. No information was reported on race/ethnicity, though the study was conducted in Italy. Followup duration was 3 years.

Renal Function

For inclusion in the IDNT trial, participants were required to have both elevated serum creatinine (1.0 to 3.0 mg/dL for women and 1.2 to 3.0 mg/dL for men) and proteinuria >900 mg/day. At baseline, mean serum creatinine was 1.7 mg/dL, mean proteinuria was 2.9 g/day, and mean albuminuria was 1.9 g/day.

For inclusion in the smaller trial, participants were required to have microalbuminuria, with a UAER of 20 to 200 μg/min, a GFR of 80 ml/min/1.73m² or greater, and a serum creatinine <10 percent higher than the upper limit of normal. After randomization, 22 participants were excluded for having UAER outside the 20 to 200 μg/min range. Within participants not withdrawn after baseline, baseline median UAER was 80.2 μg/min, mean serum creatinine was 0.97 mg/dL, mean creatinine clearance was 107.8 mL/min, and mean GFR was 111.8 ml/min/1.73m².

Baseline Comorbidities

In the IDNT trial, all participants were required to have hypertension, Mean baseline blood pressure was 159/87 mm Hg. All participants also were required to have diabetes, and mean baseline HbA_1c was 8.2 percent. Thirty percent of study subjects had a history of cardiovascular disease.

In the smaller trial, participants with hypertension were excluded and no information on baseline blood pressure was reported. All participants were required to have type 1 diabetes. Baseline HbA_1c was not reported, though those with HbA_1c 11 percent or greater were excluded. The study did not report any information on the prevalence of cardiovascular disease, though patients with an MI in the prior 3 months were excluded.

Study Quality

(Appendix Table C140)

Of the two studies, one was rated good quality and one was rated fair quality. The IDNT reported adequate concealment of treatment allocation, while concealment was unclear for the other study. Both trials were double blinded. The IDNT trial performed analyses according to the intention-to-treat principle, but the other study excluded 24 percent of participants after randomization from analyses. Withdrawals ranged from 0.5 percent in the IDNT trial to 32 percent in the other study.

Results

Mortality

(Table 12, Appendix Table C73 and Appendix Figure C16)

Table 12

Pooled clinical and renal outcomes, CCB versus placebo trials.

All-Cause Mortality

In the IDNT trial,^97,117 there was a nonsignificant reduction in risk of all-cause mortality (14.6 versus 16.3 percent; RR 0.90, 95% CI, 0.68 to 1.18). In the smaller study, only one death occurred, in an individual assigned to the CCB group.⁵⁵ In pooled results, risk with CCB treatment was nonsignificantly decreased for all-cause mortality (RR 0.90, 95% CI, 0.69 to 1.19).

Cardiovascular Mortality

In the IDNT trial,^97,117 there was a nonsignificant reduction in risk of cardiovascular mortality (6.5 versus 8.1 percent; RR 0.81, 95% CI, 0.53 to 1.22). In the smaller study, only one cardiovascular death occurred, in an individual assigned to the CCB group.⁵⁵ In pooled results, risk with CCB treatment was nonsignificantly decreased for cardiovascular mortality (RR 0.83, 95% CI, 0.55 to 1.25).

Summary

In two trials of patients with CKD and diabetes, treatment with CCB as compared with placebo was associated with nonsignificant reductions in risk of all-cause mortality, cardiovascular mortality, MI, stroke, and two different composite vascular outcomes. Risk of congestive heart failure was nonsignificantly higher for patients with CKD. The risk between treatment groups appeared similar for ESRD, doubling of creatinine, and a composite renal outcome, including both of these events as well as death. The rate of withdrawals in the smaller study was high. In both trials, adverse event rates were difficult to interpret due to incomplete reporting. Results were limited in that nearly all were derived from only one trial. The multiple post-randomization exclusions from the smaller trial and its apparent nonsystematic reporting of outcomes lowered our confidence in its reported results. Because the followup of the IDNT trial was 2.6 years, it is not possible to determine from these results the longer term effects of CCB versus placebo in patients with CKD.

Overview

In patients with CKD, we found low strength of evidence that there is no difference between treatments in risk of all-cause mortality. We found insufficient evidence regarding whether treatments differ for risk of ESRD. There was a statistically significant reduction in risk of stroke in the diuretic group versus placebo. Our confidence in these estimates is limited because data are drawn from only one trial and there were few reported clinical events.

Description of Study

One trial met all eligibility criteria and randomized CKD patients (n=393) to diuretic versus placebo.¹¹⁸ Detailed baseline characteristics are presented in Appendix Table C79.

The single eligible study was a subgroup analysis in patients with CKD from within the larger SHEP study (n=4,736), a randomized trial comparing chlorthalidone versus placebo in older patients with hypertension. Mean subject age was 74 years, and men constituted 76 percent of participants. Seventy-six percent of study participants were white, 20 percent were black, and 3 percent were Asian. The study was performed in the United States and followup duration was 5 years.

Renal Function

Participants included in this post hoc analysis were the subgroup from the larger study with a baseline creatinine of 1.35 mg/dL or higher, the level considered to represent the upper threshold of normal in the SHEP trial. Within this subgroup, no measures of baseline renal function were reported.

Baseline Comorbidities

For inclusion in the SHEP trial, participants were required to have isolated systolic hypertension, with a systolic blood pressure of 160 to 219 mm Hg, and a diastolic blood pressure less than 90 mm Hg. Mean baseline blood pressure within patients with CKD was 172/77 mm Hg. A history of myocardial infarction was reported by 5 percent, a history of stroke by 4 percent, and a history of diabetes by 12 percent. Patients were excluded from participation in SHEP for any recent myocardial infarction or stroke or for insulin-treated diabetes.

Study Quality

(Appendix Table C140)

Study quality was rated as good. Concealment of treatment allocation was adequate and the study was reported to be double blind, though it is not clear whether open-label potassium supplementation for potassium levels <3.5 mmol/L could have compromised blinding. Analysis was performed according to intention-to-treat principles. Study withdrawals were not reported for the CKD subgroup, but were adequately reported for the overall SHEP trial.

Results

Summary

In this analysis of a subgroup of patients with CKD from a larger trial of older patients with systolic hypertension, diuretic treatment compared with placebo significantly reduced risk of stroke and of one of two composite vascular outcomes. There was no significant difference between treatment groups in all-cause mortality. Results were limited by the small number of patients with CKD, with insufficient statistical power to determine whether large magnitude differences in risk for clinical outcomes were statistically significant. Results also were limited in that this was a post hoc subgroup analysis without confirmation of findings in another study population. Further, results were not reported for several vascular events of interest, including cardiovascular mortality, MI and heart failure, and no clinical renal outcomes were reported.

Overview

In patients with CKD, we found insufficient evidence that ACEI therapy as compared with non-ACEI antihypertensive therapy is associated with a reduced risk of all-cause mortality and low level of evidence that ACEI therapy compared with non-ACEI antihypertensive therapy does not significantly reduce the risk of ESRD. There was no statistically significant difference between treatment groups for risk of halving of GFR or for one reported composite renal outcome. Our confidence in these estimates is limited because data are drawn from only one trial and there were few reported clinical events.

Description of Study

We identified one trial that met all eligibility criteria and randomized 131 participants with CKD to ACEI versus non-ACEI hypertension treatment.¹¹⁹ Detailed baseline characteristics are presented in Appendix Table C84.

Randomized subjects assigned to ACEI were treated with lisinopril or lisinopril in combination with another antihypertensive agent versus a non-ACEI antihypertensive treatment regimen. Prior to randomization, 139 hypertensive patients underwent a run-in period, during which they were to follow a 0.8 g/kg protein and 3–4 g salt intake per day, and non-ACEI antihypertensive agents were used to obtain diastolic blood pressure of 90 mm Hg or less. Only patients achieving this target on two or fewer drugs, judged compliant, and with stable renal function were eligible to proceed to randomization.

Mean age of randomized study participants was 51 years, and men constituted 66 percent of all subjects. Race/ethnicity of study participants was not reported, though the study was conducted in Italy. Mean followup was 1.9 years.

Renal Function

Participants were required to have creatinine clearance between 20 and 50 ml/min/1.73m² and were excluded if they had proteinuria of ≥1 gram/day. Among those enrolled, mean creatinine clearance was 36 ml/min/1.73m², mean GFR was 36 ml/min/1.73m², mean creatinine was 2.4 mg/dL, and mean proteinuria was 512 mg/day.

Baseline Comorbidities

All study participants had hypertension, with an untreated diastolic blood pressure of ≥95 mm Hg prior to run-in, and a stable treated diastolic blood pressure <90 mm Hg prior to randomization. Patients with malignant hypertension were excluded. Among subjects randomized, mean baseline blood pressure was 142/86 mm Hg. Patients with diabetes, heart failure or another major (undefined) cardiac disease, or a recent history of MI or stroke were excluded from study entry.

Study Quality

(Appendix Table C140)

Study quality was rated as fair. This study was open-label and concealment of treatment allocation was unclear. Analyses were conducted according to the intention-to-treat principle. No information was reported regarding withdrawals.

Results

Summary

In a single study of patients with hypertension and CKD, antihypertensive treatment with ACEIs in comparison to that without ACEIs was associated with nonsignificant reductions in the risk for MI, ESRD, halving of GFR, and a composite renal outcome including ESRD and GFR. Mortality data and other cardiovascular or renal outcomes were not reported, nor were study withdrawals or serious adverse events. Results were limited by the small sample size, small number of clinical events, and short followup duration.

Overview

In patients with CKD, we found low strength of evidence that treatment with CCB does not significantly reduce the risk of all-cause mortality compared with BB, and low strength of evidence that there is no difference between treatments in risk of ESRD. Participants assigned CCB were statistically significantly less likely to experience one composite vascular outcome. Our confidence in these estimates is limited by the small number of trials reporting different outcomes and the small number of clinical events.

Description of Studies

Three trials met all eligibility criteria and randomized participants with CKD (n=12,766, range 34 to 12,074) to CCB versus BB.^{89,90,120,121} Detailed baseline characteristics of patients enrolled in the three trials are presented in Appendix Tables C80 and C81.

Among eligible trials, most data were derived from a subgroup analysis reported in a subset of 12,074 patients with undefined “renal dysfunction” from the larger ASCOT-BPLA trial (n=19,257).¹²¹ In this study, participants were randomized to amlodipine versus atenolol. As needed to meet blood pressure targets (<140/90 mm Hg for patients without diabetes and <130/90 mm Hg for patients with diabetes), participants randomized to amlodipine could have had an ACEI added and subjects randomized to atenolol could have had a diuretic added. In the AASK trial, designed as a 3×2 factorial study, besides randomizing 658 participants to amlodipine versus metoprolol, an additional 436 were randomized to an ACEI, and all participants also were randomized to one of two blood pressure target groups as described elsewhere in this report.^89,90 In this trial, the amlodipine treatment arm was stopped early by recommendation of the data and safety monitoring board with patients switched to open label medication. Results presented here compare outcomes including followup until the time blinded amlodipine was discontinued. In the smallest trial,¹²⁰ 34 participants were randomized to one of two CCBs (verapamil or diltiazem) versus atenolol. This study also included an additional ACEI treatment arm that is reviewed elsewhere in this report.

The mean age of study participants across all three trials was 55 years (range 55 to 62) and men constituted 60 percent (range 44 to 61, n=2 trials) of all subjects studied. In the two trials that reported race/ethnicity,^89,90,120 98 percent of participants were African American, including 100 percent of subjects in the AASK trial.^89,90 Two studies were conducted in the United States, and the large subgroup analysis was conducted in Europe. Median study duration ranged from 3 to 5.5 years.

Renal Function

Among eligible trials, one required that participants have impaired GFR (20 to 65 ml/min/1.73m²)^89,90 and reported a mean baseline GFR of 46 ml/min/1.73m², a mean creatinine of 2.0 mg/dL, and mean proteinuria of 0.5 g/day. A second trial required that participants have both impaired creatinine clearance (<70 ml/min) and at least 2 g/day proteinuria, and reported a mean baseline creatinine clearance of 61 ml/min/1.73m², a mean creatinine of 1.9 mg/dL, and mean proteinuria of 4.4 g/day.¹²⁰ The third study reported no information on the baseline renal function in its “renal dysfunction” subgroup.

Baseline Comorbidities

In all three studies, all participants were required to be hypertensive. In two trials reporting, mean baseline blood pressure was 150/95 mm Hg.^89,90,120 Patients with heart failure were excluded from all three trials, and patients with a history of MI¹²¹ or of any documented coronary artery disease¹²⁰ were excluded in two trials. While one trial required that participants be diabetic,¹²⁰ a second trial excluded diabetic patients,^89,90 and the third study reported no information on participants' diabetes status.¹²¹

Study Quality

(Appendix Table C140)

Study quality was rated good for two trials and fair for one trial. Two of the trials reported adequate treatment allocation concealment.^89,90,121 One study was open-label,¹²¹ a second study was double blind with respect to medication assignment but not to blood pressure target.^89,90 Both reported that endpoint adjudicators were blinded to treatment allocation. The third study provided no information with respect to blinding.¹²⁰ Two of the three studies performed analyses according to the intention-to-treat principle. Withdrawals ranged from 0 to 11.5 percent between studies.

Results

Mortality

(Table 13, Appendix Table C91, and Appendix Figure C19)

Table 13

Pooled clinical and renal outcomes, CCB versus BB trials.

All-Cause Mortality

In two trials of CKD patients reporting mortality data, those randomized to CCB versus BB had a nonsignificant 38 percent reduction in risk of all-cause mortality (6.0 versus 9.2 percent; RR 0.62, 95% CI, 0.31 to 1.22; n=692 patients).^89,90,120

Cardiovascular Mortality

One study reported cardiovascular deaths per patient year of followup (CCB 0.9 percent, BB 0.8 percent) but did not report the number and percentage of participants with this outcome by treatment group.^89,90 A second study reported cardiovascular deaths (9.6 percent) but did not report these outcomes by treatment group.¹²⁰

Summary

In patients with CKD and hypertension, there was a nonsignificant 38 percent reduction in all-cause mortality with CCB compared with BB treatment. One of two trials reported a significant reduction in a composite vascular outcome, but this was the only one of six composite vascular endpoints collected in this trial that was reported for patients with renal dysfunction, raising uncertainty regarding whether this risk reduction is a consistent finding within this study. There was no significant difference between CCB and BB treatment groups in risk of ESRD or in risk of the composite renal outcome of ESRD, death, or greater than 50 percent decline in GFR. Both the composite outcome of ESRD or death and the risk of doubling creatinine appeared less likely in patients randomized to CCB, though results were not statistically significant. Results were limited in that most outcomes were not reported by treatment group in more than one study, and by the uncertainty regarding whether the patients in the ASCOT-BPLA study with “renal dysfunction” meet criteria for CKD.

Overview

In patients with CKD, there was insufficient evidence regarding whether there is a difference between CCB and diuretic treatment for risk of all-cause mortality and low strength of evidence that there was no difference between CCB and diuretic treatment for risk of ESRD. There was no statistically significant difference between CCB and diuretic treatment groups in risk of stroke, CHF, or in multiple composite vascular or renal outcomes. Our confidence in these estimates is limited because they are based entirely on results reported from a post hoc analysis from a single large trial.

Description of Study

One study met all eligibility criteria and randomized 4,129 participants to CCB monotherapy versus diuretic monotherapy.^81-83 Detailed baseline characteristics are presented in Appendix Tables C97 and C98. The eligible study was a post hoc analysis performed within a subset of participants with CKD from the ALLHAT trial, a study of 23,261 subjects that was not originally limited to individuals with CKD, contained two additional antihypertensive treatment arms and, as part of a factorial design, also randomized participants to pravastatin versus control.

The CCB and diuretic utilized in this trial were amlodipine and chlorthalidone, respectively. The mean age among the 4,129 study participants assigned to CCB versus diuretic was 71 years, and men constituted 47 percent all study subjects. The most common race/ethnicity of trial participants was white non-Hispanic (57 percent), black (25 percent), and Hispanic (12 percent). The ALLHAT trial was performed primarily in the United States. The study duration was 4.9 years.

Renal Function

Patients with a baseline creatinine level >2 mg/dL were excluded from the main ALLHAT trial. Inclusion in the post-hoc analysis was limited to ALLHAT participants with a GFR <60 ml/min/1.73m². Within subjects in the CKD subgroup, mean baseline GFR was 50 ml/min/1.73m². No baseline data on albuminuria was reported.

Baseline Comorbidities

Enrollment was limited to patients with hypertension, with the mean blood pressure at baseline 147/83 mm Hg. Thirty-four percent of participants reported diabetes at baseline, 60 percent reported cardiovascular disease, and 30 percent reported coronary artery disease.

Study Quality

(Appendix Table C140)

Study quality was rated as good. Allocation concealment was adequate. The trial was double blinded and analysis by the intention-to-treat principle was reported. No data regarding withdrawals was reported.

Results

Summary

Within the one eligible trial of patients with CKD, there was no apparent difference between the CCB and diuretic monotherapy treatment groups in risk of stroke, ESRD, or other composite clinical vascular or renal outcomes. Results were limited in that the study was a post hoc subgroup analysis. The ALLHAT study also did not report results for risk of mortality or risk of MI in the subgroup of CKD patients. In addition, mean followup did not extend beyond 5 years, so longer term effects of CCB monotherapy versus diuretic monotherapy cannot be determined from these data.

Overview

In patients with CKD, we found a low strength of evidence regarding whether antihypertensive treatment targeting stricter blood pressure targets reduces risk of all-cause mortality compared with treatment targeting standard blood pressure control targets. We found a low strength of evidence regarding whether there was a difference between treatments for risk of ESRD. Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events, and heterogeneity between studies.

Description of Studies

Six trials met all eligibility criteria and randomized participants with CKD (n=2,520, range 77 to 1,094) to treatment aimed to reach different target blood pressures, i.e., “strict” versus “standard” blood pressure targets.^90,122-127 One study was not limited to individuals with CKD but presented subgroup results for the approximately 3 percent of participants whose baseline creatinine was >1.7 mg/dL.¹²⁶ Detailed baseline characteristics are presented in Appendix Tables C104 and C105.

In general, studies established blood pressure targets for their strict control group about 10-15 mm Hg lower than for their standard control group, though there was variability between trials in the absolute blood pressure targets selected. The most common treatment target, used in three trials, was a mean arterial blood pressure (MAP) of ≤92 mm Hg versus a MAP of 100 to 107.^90,123,125 Two trials set diastolic blood pressure (DBP) targets, <90 mm Hg for the strict target versus >90 mm Hg for the standard target in one trial,¹²⁶ and 65 to 80 mm Hg for the strict target versus 85 to 95 mm Hg for the standard target in the second trial.¹²⁴ The most recent trial compared treatment to achieve blood pressure <130/80 mm Hg versus a DBP target of <90 mm Hg.¹²² The specific antihypertensive agents utilized to achieve these blood pressure targets varied between trials. The oldest trial, published in 1989,¹²⁶ used diuretics, adrenergic receptor blockers, and vasodilators, while all three trials published in the 1990s used ACEIs with or without diuretics as first-line treatment.^123-125 A trial published in 2002, structured as a 3 × 2 factorial design, assigned participants to initial treatment with either an ACEI, beta blocker or calcium channel blocker.⁹⁰ Finally, the most recent trial, published in 2005,¹²² titrated all participants with an ACEI prior to randomization and then used a long-acting CCB to compare strict versus standard blood pressure control.

The mean age of study subjects was 53 years (range of study means 37 to 56; n=5 trials), and men constituted 63 percent (range 47 to 75; n=6 trials) of all patients evaluated. Among five trials reporting race/ethnicity, three were predominately^124,126 or entirely⁹⁰ comprised of blacks/African Americans. In two other trials, more than 85 percent of participants were white.^123,125 All trials were conducted in the United States, except for one performed in Italy.¹²² Mean or median study duration ranged from 19 months to 5 years, with all but one trial having a followup duration of at least 2 years.

Renal Function

Among the six trials, two required that participants have proteinuria to be included,^122,123 and one trial excluded participants with proteinuria.⁹⁰ Five trials required decreased GFR or creatinine clearance or elevation in serum creatinine for entry,^{90,122,124-126} including one study that was a subgroup analysis of participants from a larger trial with baseline creatinine >1.7 mg/dL.¹²⁶ Measures of baseline renal function were reported in all but one trial.¹²⁶ Mean GFR was 43 ml/min/1.73m² (range 35 to 63), mean serum creatinine was 2.0 mg/dL (range 1.3 to 2.7), and mean proteinuria was 1.0 gm/day (range 0.36 to 2.85). Creatinine clearance, reported in only two trials, averaged 46.2 ml/min/1.73m².^122,125

Baseline Comorbidities

In five trials reporting data, approximately 95 percent of study participants had a history of hypertension. In the sixth trial, though information on history of hypertension was not reported, approximately two-thirds of the subjects were receiving blood pressure lowering drugs at baseline.¹²² Mean blood pressures at baseline were 142/89 mm Hg (MAP 106 mm Hg). Overall, few study participants had diabetes, though among individual trials one included only patients with type 1 diabetes,¹²³ two excluded all diabetic patients,^90,124 one had about 15 percent diabetic patients¹²⁶ and two studies provided no information regarding whether participants had a history of diabetes.^122,125 While only one trial reported baseline prevalence of cardiovascular disease, at 36 percent,¹²⁴ several reported exclusions of such participants, including exclusion of all subjects with recent MI or stroke,^122,124 exclusion of all participants with a history of any past MI or stroke,¹²³ and exclusion of any participants with clinical or overt heart failure.^90,122 One further trial documented enrollment of individuals with cardiovascular disease but did not report baseline prevalence.¹²⁶

Study Quality

(Appendix Table C140)

Among the six trials, study quality was rated as good in one trial and as fair in five trials. Allocation concealment was adequate in three trials and unclear in the remaining studies. Three trials were not blinded,^90,122,126 one was double blinded,¹²⁴ and blinding was unclear for two trials.^123,125 For the outcomes presented here, four of six trials analyzed results according to the intention-to-treat principal.^90,123-125 Three trials adequately described reasons for study withdrawal.^90,122,125 Percentages of study withdrawals ranged from 0 to 16 percent (n=4 trials).

Results

Mortality

(Table 14, Appendix Table C106, and Appendix Figure C21)

Table 14

Pooled clinical and renal outcomes, strict versus standard blood pressure target treatment trials.

All-Cause Mortality

Compared with standard blood pressure control, there was no significant reduction in risk of all-cause mortality with strict blood pressure control (RR 0.86, 95% CI, 0.68 to 1.09; n=4 trials, 1,803 patients). These results were driven almost entirely by two trials that, though they each reported a 12 to 15 percent relative reduction in mortality with strict compared with standard blood pressure control, differed markedly in other respects. In the trial by Shulman, 35 percent of participants assigned strict blood pressure control versus 41 percent assigned standard control died during a 5-year followup period, compared with 6.9 percent versus 7.8 percent, respectively, in an approximately 4-year followup period, in the trial by Wright.^90,126 Other differences between these trials included the substantially higher baseline blood pressure, most of which had been untreated, among participants in the Shulman trial,¹²⁶ and lower blood pressure targets for both treatment groups, and use of ACEIs and BB only in the trial by Wright.⁹⁰

Cardiovascular Mortality

Compared with standard blood pressure control, there was no significant reduction in risk of cardiovascular mortality with strict blood pressure control (RR 0.83, 95% CI, 0.54 to 1.26). Nearly all the weight contributing to this pooled estimate was derived from one trial, in which 20.1 percent versus 23.9 percent of participants experienced a cardiovascular death (RR 0.84, 95% CI, 0.55 to 1.29),¹²⁶ while fewer than 1 percent of participants died due to cardiovascular causes in the only other trial reporting this outcome.¹²²

Summary

(Appendix Table C140)

In individuals with CKD, compared with targeting standard blood pressure control, assignment to targeting strict control was associated with 14 percent and 17 percent relative reductions in risk of all-cause mortality and cardiovascular mortality, respectively, which were not statistically significant. There were no significant differences between treatment groups for the outcomes of MI, stroke, ESRD, or, in individual trials, for several composite renal outcomes. Findings for the mortality and ESRD outcomes were driven mostly by a single trial conducted more than 20 years ago that may have limited generalizability to current patient populations and available antihypertensive treatment options. Results for MI and stroke in particular were limited by small sample sizes and could not exclude either clinically meaningful benefits or harms. Overall results were further limited by heterogeneity in patient populations (i.e., baseline level of renal function, comorbidities), and heterogeneity in blood pressure targets. Subgroup analyses, though limited, did not identify any comorbid conditions or category of renal function in which there was a significant difference between strict and standard blood pressure control for any clinical outcome. Reporting on study withdrawals and adverse effects was limited. Finally, no trial provided followup beyond 5 years; therefore, longer term effects of different blood pressure targets cannot be determined from these studies.

Overview

In patients with CKD, we found a low level of evidence that, compared with usual protein diets, a low protein did not significantly reduce risk of all-cause mortality nor increase risk of ESRD. We found no statistically significant difference between treatment groups in risk of doubling baseline creatinine, or halving GFR. Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events, and heterogeneity between studies.

Description of Studies

Six trials met all eligibility criteria and randomized 1,480 (range 63 to 585) participants with CKD to a low protein diet (typically 0.6 or 0.8 g protein per kg of ideal/lean body weight per day) versus a usual diet (typical protein intake less than 1.3 g/kg/day).^125,127-135 Detailed baseline characteristics are presented in Appendix Tables C112 and C113.

Mean participant age was 52 years (range 49 to 58; n=5 trials), and men constituted 59 percent (range 54 to 83; n=5 trials) of all study participants. In the one trial reporting ethnicity, 85 percent of participants were white. One study was conducted in the United States.^{125,127,130,131} Of the remaining studies, two were conducted in Italy^132,133 and one each was conducted in Japan,¹²⁸ France,¹²⁹ and the United Kingdom.^134,135 Followup periods ranged from 2 to 3.5 years.

Renal Function

Two of the six eligible trials required that participants have albuminuria, with some limitation in the severity of their renal function. In one of these studies, all subjects had to have macroalbuminuria (UAER >200 μg/min) or proteinuria (urine protein excretion rate, i.e., urine protein excretion rate (UPER), >1 g/day) and serum creatinine <2.0 mg/dl (i.e., CKD stages 1–3).¹²⁸ In the second study, patients were required to have UAER >30 mg/day (i.e., at least microalbuminuria) and GFR of at least 15 ml/min (i.e., CKD stages 1-4).¹²⁹ The remaining four trials required that participants have either an elevated serum creatinine or a reduced GFR or creatinine clearance, and three also imposed limits on UPER. Thresholds for eligibility in these trials included creatinine 1.2 to 7.0 mg/dl in women and 1.4 to 7.0 mg/dl in men, with UPER less than 10 g/day (i.e., CKD stages 1-5);¹³¹ creatinine 1.35 to 7.0 mg/dl in women and 1.5 to 7.0 mg/dl in men, with GFR <60 ml/min, and UPER <3 g/day (i.e., CKD stages 3-5);¹³³ creatinine clearance 15 to 70 ml/min and UPER <3 g/day (i.e., CKD stages 1-4);¹³² and GFR 10 to 60 ml/min (i.e., CKD stages 3-5).^134,135 Among all six eligible trials, baseline mean UPER ranged from 0.28 g/day/1.73m² to 1.5 g/day (n=3 trials), mean UAER was reported in only one trial (366 mg/day), mean serum creatinine was 1.7 mg/dl (range 1.1 to 1.9, n=3 trials), and mean GFR was 45 ml/min/1.73m² (range 39 to 86, n=3 trials).

Baseline Comorbidities

Two trials enrolled only patients with diabetes,^128,129 two trials excluded patients with diabetes,^132,133 and two trials did not report baseline prevalence of diabetes^131,134,135 Among two trials reporting,^128,131 mean blood pressure was 132/80 mm Hg. One trial excluded all patients with CHF,¹²⁸ another trial excluded patients with either class III or IV CHF,¹³¹ and two trials excluded participants with a recent MI^128,133 or stroke.¹²⁸ However, no additional information on baseline cardiovascular morbidity was reported in any trial.

Study Quality

(Appendix Table C140)

Study quality was rated as fair in all six trials. Allocation concealment was adequate in three studies and unclear in three studies. One trial reported that measures of GFR were blinded,¹³¹ but the other trials were not reported as blinded. Five trials did not perform analyses using intention-to-treat principles, and it was unclear in one study.¹³¹ Withdrawals were adequately described in all but two studies.^132,134,135 Study withdrawals ranged from 2 to 25 percent.

Results

Mortality

(Table 15, Appendix Table C114, and Appendix Figure C22)

Table 15

Pooled clinical and renal outcomes, dietary intervention trials.

All-Cause Mortality

In the CKD patients studied in eligible trials, low protein diets were associated with a nonsignificant 42 percent reduction in risk of all-cause mortality compared with usual protein diets (1.9 versus 3.3 percent; RR 0.58, 95% CI, 0.29 to 1.16; n=4 trials).^{128,131,133,135} All individual trials suggested a lower mortality risk with low protein diets, but the difference was not statistically significant in any trial.

Cardiovascular Mortality

Only one trial reported cardiovascular mortality, in which there were four such events (1.4 percent) in the low protein diet group and five cardiovascular deaths (1.7 percent) in the usual protein diet group.¹³¹

Summary

In six trials conducted in patients with CKD, low protein diets were associated with a nonsignificant 42 percent reduction in all-cause mortality compared with usual protein diets. In one trial reporting, the small number of cardiovascular deaths appeared similar in both diet intervention groups. No other vascular outcome was reported in more than two cases in any trial. In three trials reporting, all with fewer than 10 cases of ESRD, low protein diets were associated with a nonsignificant 62 percent increase in risk of ESRD compared with usual protein diets. One trial reported a significant 37 percent lower risk of the composite renal outcome of dialysis or doubling of plasma creatinine concentration in CKD subjects randomized to low protein diet versus usual protein diet. Applicability to patients with CKD stages 1–3 may be limited since at least four of six trials also included individuals in CKD stages 4–5. Withdrawals ranged widely between trials but did not appear greater in the low protein diet group in any trial. Results were limited by small sample sizes, few trials reporting clinical vascular or renal outcomes, and almost no events in the trials that reported these outcomes. Judging applicability was limited because of the variability in renal function reported between trials and scant data reported on comorbid conditions. Trials did not systematically report adverse events.

Overview

In patients with CKD, we found a low level of evidence that, compared with a low protein diet, the CR-LIPE diet reduced risk of mortality or ESRD. There was a low level of evidence that diets altering phosphate intake impacted risk of ESRD and insufficient evidence regarding whether it impacted risk of mortality. There was insufficient evidence regarding whether a low triglyceride diet and pharmacological treatment to lower triglycerides differ regarding risk of mortality or ESRD. Our confidence in these estimates is limited because for each comparison data are drawn from only one trial and there were few reported clinical events.

Description of Studies

Three trials met all eligibility criteria and randomized participants with CKD to a diet intervention versus a control treatment group. Detailed baseline characteristics are presented in Appendix Tables C112 and C113.

Among the three trials, one randomized 191 participants to a carbohydrate restricted, low-iron-available, polyphenol-enriched diet (CR-LIPE) versus a low protein diet.¹³⁶ Mean age of study participants was 60 years and 53 percent of study participants were men. The study was conducted in the United States. Followup duration was 3.9 years.

A second trial randomized 57 participants to a triglyceride lowering diet versus gemfibrozil, a triglyceride lowering medication.¹³⁷ Mean age of study participants was 51 years and 75 percent of study subjects were men. The study was conducted in Sweden. Followup duration was 1 year.

The third trial randomized 98 participants to either a low protein-low phosphate diet, a low phosphate diet with phosphate binders, or an unrestricted diet.¹³⁸ Mean age of study participants was 45 years and 66 percent of study subjects were men. The study was conducted in the United Kingdom. Followup duration was 1.6 years.

Renal Function

For inclusion in the CR-LIPE versus low protein diet trial, participants were required to have GFR 15 to 75 ml/min and UPER 0.35 to 12 g/day. Baseline renal function was reported as mean GFR 63 ml/min, mean UPER 2.47 g/day, and mean creatinine 1.84 mg/dl.¹³⁶ For inclusion in the triglyceride lowering diet versus gemfibrozil trial, participants were required to have a GFR of 10 to 70 ml/min/1.73m². Baseline renal function was reported as mean GFR of 35.5 ml/min/1.73m², mean serum creatinine of 2.4 mg/dl, and mean UAER of 0.95 g/day.¹³⁷ For inclusion in the low protein low phosphate diet, low phosphate diet, or unrestricted diet trial, participants were required to have a serum creatinine between 1.7 and 10.2 mg/dl. At baseline, mean serum creatinine was 4.5 mg/dl, mean UPER was 3.15 g/day, and mean creatinine clearance was 26.8 ml/min/1.73m2.¹³⁸

Baseline Comorbidities

For inclusion in the CR-LIPE versus low protein trial, participants were required to be diabetic.¹³⁶ No additional information was reported on comorbid conditions. For inclusion in the triglyceride lowering diet versus gemfibrozil trial, participants were required to be nondiabetic. No additional information was reported on comorbid conditions.¹³⁷ For the third trial, no information was reported on comorbid conditions.¹³⁸

Study Quality

(Appendix Table C140)

All three trials were rated as fair quality. Allocation concealment was adequate in one trial¹³⁸ and unclear in the other two studies. One study reported that study personnel were blinded to the aims of the study, but it was unclear if the outcome assessment was blinded.¹³⁶ The remaining two studies were unblinded. None of the studies analyzed by the intention-to-treat principle. Withdrawals ranged from 5.3 to 15.8 percent, and reasons for withdrawals were adequately explained in two of the three trials.^136,137

Results

Summary

In one trial, CKD patients randomized to a carbohydrate-restricted, low iron available, polyphenol-enriched diet (CR-LIPE) appeared to have lower all-cause mortality, lower risk of ESRD, and lower risk of the composite endpoint of ESRD or death compared with participants assigned to a low protein diet. In a second trial, study participants allocated to a low phosphate-phosphate binding diet appeared to have a higher risk of all-cause mortality than did patients assigned to either a low protein-low phosphate diet or to a control diet. There was no apparent difference between these three diet groups in risk of ESRD. In the third trial, results suggested that CKD patients randomized to a low triglyceride diet may have a lower risk of ESRD and fewer gastrointestinal side effects than patients assigned to gemfibrozil. Results were limited in that all trials were small, reported few clinical outcomes, and did not conduct their analyses according to an intention-to-treat principle.

Overview

In diabetic patients with CKD, we found insufficient evidence regarding whether there is a difference between treatments in risk of mortality or ESRD. Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events, and heterogeneity between studies.

Description of Studies

Two trials met all eligibility criteria and randomized participants with diabetes and CKD to intensive versus standard glycemic control.^139,140 Detailed baseline characteristics are presented in Appendix Tables C119 and C120.

In the first study, conducted in 70 patients, those assigned intensive diabetes control (treatment targets HbA_1c ≤7.5percent, fasting blood glucose 72 to 108 mg/dL, and 2 hour postprandial blood glucose ≤180 mg/dL) were treated using insulin by continuous infusion or multiple daily injections. Frequent visits and medication adjustment were made as needed, and 24 hour/day consultation was available. Participants assigned to conventional therapy (no glycemic targets) generally were treated using two daily insulin injections, adjusted only for symptoms, along with conventional education about diet, exercise, and blood glucose monitoring. Standard control patients were seen every 3 months. No changes were made to the usual diabetic diet of participants in either treatment group, and all patients were treated to keep their blood pressure <160/95 mm Hg.¹⁴⁰ Mean age was 37 years, 73 percent of participants were male, and no data was reported on ethnicity.

In the second study, a subgroup analysis within 491 patients with microalbuminuria from a larger diabetes treatment trial (n=1,791), trial participants allocated to the intensive control group were started on maximal doses of oral therapy, and insulin was added as needed to achieve a target HbA_1c <6 percent.¹³⁹ Participants assigned to standard control were started on one-half of maximal doses of oral therapy and insulin was added as needed to achieve a target HbA_1c <9 percent. No data on age, gender, or ethnicity was reported for the subgroup with microalbuminuria.

Renal Function

Both trials were restricted to participants with microalbuminuria. In one study this was defined as UAER between 30 and 200 μg/min.¹⁴⁰ In the second trial, in which subgroup results were reported for conversion from microalbuminuria to macroalbuminuria, no definition of microalbuminuria was reported. Participants in this second trial further were restricted to those with serum creatinine ≤1.6 mg/dl.¹³⁹ Participant mean baseline GFR was 116.7 ml/min/1.73m² and mean UAER was 47.9 μg/min.

Baseline Comorbidities

Both trials were restricted to participants with diabetes. The first trial enrolled patients with baseline blood pressure <160/95 mm Hg and no evidence of diabetic macrovascular complications.¹⁴⁰ Participant mean HbA_1c was 10.1 percent. The second trial enrolled individuals with type 2 diabetes who had HbA_1c >7.5 percent, despite maximal doses of oral agents or on insulin.¹³⁹ In this second trial, no baseline characteristics were reported for the subgroup of participants with microalbuminuria, though overall study excluded patients with a cardiovascular event during the previous 6 months, advanced CHF, or severe angina.

Results

Renal Outcomes

(Table 16, Appendix Table C122)

Table 16

Pooled clinical and renal outcomes, glycemic control trials.

End-Stage Renal Disease

There were no reports of ESRD.

Other Renal Outcomes

(Appendix Figure C23)

One trial reported that one patient experienced acute renal failure (2.7 percent) but did not report the group assignment of the patient.¹⁴⁰ In data from both trials, compared with conventional treatment, participants allocated to intensive diabetes treatment had a nonsignificant 31 percent relative reduction in risk of conversion from microalbuminuria to macroalbuminuria (8.7 versus 12.8 percent; RR 0.69, 95% CI, 0.42 to 1.12; n=561 patients). Neither study reported on any other renal outcomes.

Summary

In diabetic individuals with CKD, compared with conventional treatment, assignment to intensive diabetes treatment was associated with an approximately 31 percent relative reduction in risk of conversion from microalbuminuria to macroalbuminuria that was not statistically significant. There were no data regarding the relative risk between these treatment strategies for all-cause or cardiovascular mortality, MI, stroke, ESRD, doubling of serum creatinine, halving of GFR, or any composite vascular or renal endpoint. Reporting on withdrawals and adverse effects associated with these treatment regimens in CKD patients was limited. Results were limited by the small number of trials, few outcomes reported, the heterogeneity of patient populations, and the heterogeneity in intensity of treatment regimens.

Overview

In patients with CKD defined by impaired GFR, in comparison to control treatment, there is a high level of evidence that statins reduce risk of all-cause mortality. There is a low level of evidence that there is no difference between statins and control treatment for risk of ESRD. Compared with participants assigned to control, those randomized to statin had a statistically significantly lower risk of MI, stroke, and most composite cardiovascular outcomes, but no statistically significant difference in risk of cardiovascular mortality, CHF hospitalization, or a single composite renal outcome. Our confidence in these estimates is limited by the heterogeneity between studies and because all outcomes data are drawn from post hoc analyses.

Description of Studies

Twelve trials met all eligibility criteria and randomized participants with CKD (n=17,460, range 304 to 4,491) to statin therapy versus control.^{62,83,141-149} One study, the CARE trial, was reported both by itself¹⁴⁸ and as part of a pooled subject-level meta-analysis of three trials of pravastatin versus placebo.¹⁴⁹ All but one study⁶² were post hoc analyses performed within subsets of participants with CKD from larger trial populations not originally limited to subjects with CKD. Detailed baseline characteristics are presented in Appendix Tables C124 and C125.

Among eligible trials, 9,890 participants were randomized to pravastatin versus control, including 5,355 versus placebo (n=2 trials),^62,149 2,978 versus diet (n=1 trial),¹⁴² and 1,557 versus usual care (n=1 trial).⁸³ In addition, 4,902 participants were randomized to rosuvastatin versus placebo (n=2 trials),^146,150 and 1,549 participants were randomized to atorvastatin versus control, including 970 versus placebo (n=1 trial)¹⁴³ and 579 versus usual care (n=1 trial).¹⁴⁴ There also were several smaller studies, in which 505 participants were randomized to simvastatin versus placebo (n=1 trial),¹⁴⁵ 310 participants were randomized to fluvastatin versus placebo (n=1 trial),¹⁴⁷ and 304 participants were randomized to lovastatin versus placebo (n=1 trial).¹⁴¹ The mean age of subjects was 65 years (range 51 to 71; n=10 trials), and men constituted 53 percent (range 24 to 82; n=10 trials) of all patients randomized. Among the six trials that reported race/ethnicity, 79 percent of participants were white. The majority of trials were multinational.^{83,143,145-147,149,150} Mean or median study duration ranged from 1.9 to 5.4 years, with most trials having a followup of at least 4 years.

Renal Function

All studies except one were post hoc analyses from large statin trials, performed in subsets of participants with decreased GFR or creatinine clearance from the larger trial populations. Most of these analyses defined impaired GFR or creatinine clearance as <60 ml/min/1.73m²,^{83,141-144,149,150}(i.e., CKD stage 3 or worse) or at least provided data for patients under this threshold.¹⁴⁵ No other trials based study eligibility on CKD stage or reported baseline distribution of participants by CKD stage. Instead, individual studies defined impairment as GFR <51 ml/min/1.73m²,¹⁴⁶ and creatinine clearance of <55.9 ml/min¹⁴⁷ and <75 ml/min.¹⁴⁸ In trials reporting, mean baseline serum creatinine was 1.3 mg/dL (range 1.0 to 1.5, n=9 trials), mean GFR was 54 ml/min/1.73m² (range 50 to 56; n=10 trials), and mean creatinine clearance was 59 ml/min (range 47 to 61; n=2 trials). Most of the larger trials on which the post hoc analyses were based excluded at least some patients with impaired renal function, with exclusion thresholds ranging from creatinine >1.5 to 1.8 mg/dL,^{142,143,147,149} >2.0 mg/dL,¹⁵⁰ >2.5 mg/dL,^146,148 to >4.5 mg/dL.¹⁴⁹ One trial excluded participants with creatinine clearance less than 60 percent of the age-based normal.⁶² Just one trial required participants to have microalbuminuria for inclusion,⁶² and this was also the only trial to report mean baseline urinary albumin excretion (23 mg/24 hours).

Baseline Comorbidities

One trial was restricted to patients with diabetes,¹⁴³ and the mean prevalence of diabetes among 11 trials reporting was 22 percent (range 2 to 100). One trial was restricted to patients with hypertension,⁸³ while hypertensive patients were excluded from one trial.⁶² The mean prevalence of hypertension was 49 percent among nine trials reporting. Mean systolic blood pressure was 136 mm Hg (range 131 to 146; n=10 trials) and mean diastolic blood pressure was 80 mm Hg (range 75 to 84; n=9 trials). Six trials were restricted to patients with coronary artery disease (secondary prevention studies),^144-149 including one restricted to patients with a history of myocardial infarction,¹⁴⁸ while patients with coronary artery disease were excluded from five trials (primary prevention studies).^{141-143,149,150} The mean prevalence of coronary artery disease was 46 percent in 12 trials reporting. One trial was restricted to patients with CHF,¹⁴⁶ while patients with CHF were excluded from one trial.⁶² The mean prevalence of CHF in four trials reporting was 39 percent. Mean baseline total cholesterol and low density lipoprotein (LDL) cholesterol were 220 mg/dL (range 189 to 265) and 142 mg/dL (range 109 to 192), respectively (n=12 trials).

Study Quality

(Appendix Table C140)

Study quality was rated as good in eight trials and as fair in four trials. The method of allocation concealment was adequate in nine trials^{83,142-146,148-150} but was unclear in three studies.^83,145,149 Nine trials were double blinded, of which eight explicitly stated that outcomes were adjudicated by blinded assessors.^{62,141,143,145-147,149,150} Three trials were open label studies,^83,142,144 though one stated that outcomes were adjudicated by blinded assessors.⁸³ Analysis was by intention to treat in eleven studies^{62,83,141-144,146-150} and not by intention to treat in one study.¹⁴⁵ None of the post hoc analyses reported the number of withdrawals within the subgroup of participants with impaired GFR. Only the trial that enrolled participants on the basis of microalbuminuria reported withdrawals (23 percent).⁶²

Results

Mortality

(Table 17, Appendix Table C126, and Appendix Figure C24)

Table 17

Pooled clinical and renal outcomes, anti-lipid agents versus control trials.

All-Cause Mortality

In CKD patients assigned to statins versus control, there was a significant 20 percent reduction in all-cause mortality (7.1 versus 8.7 percent; RR 0.80, 95% CI, 0.68 to 0.95; n=8 trials, 13,964 patients).^{62,142-145,147,149,150} However, in three trials that limited entry to patients with coronary artery disease, the 11 percent reduced all-cause mortality among patients randomized to statin treatment was not statistically significant (12.8 versus 14.3 percent; RR 0.89, 95% CI, 0.68 to 1.15; n=1,394 patients)^144,145,147 (Figure 8). In three trials that limited enrollment to patients without coronary artery disease, the 37 percent relative reduction in all-cause mortality was statistically significant (2.1 versus 3.4 percent; RR 0.63, 95%CI, 0.44 to 0.90; n=7,215 patients)^142,143,150 (Figure 8).

Figure 8

Statins versus placebo: Mortality by coronary artery disease (CAD) subgroups.

Cardiovascular Mortality

In the four trials reporting data for cardiovascular mortality, there were few events (2.4 versus 3.4 percent for statin and control groups, respectively), with a nonsignificant 29 percent relative risk reduction in this outcome in participants randomized to statin treatment versus control (RR 0.71, 95% CI, 0.43 to 1.71; n=2,057 patients).^{62,141,144,147} There also was no significant difference in risk of cardiovascular mortality between subjects assigned to statin versus control groups in two trials limited to patients with coronary artery disease (4.6 versus 6.6 percent; RR 0.69, 95% CI, 0.40 to 1.19; n=889 patients)^144,147 or in one trial limited to patients without coronary artery disease (0 versus 0.6 percent).¹⁴¹

Summary

In individuals with CKD, statin treatment, as compared with control, was associated with significant relative reductions in risk of all-cause mortality (20 percent), MI (28 percent), and stroke (38 percent). Results appear to favor statin in both patients with and without a history of coronary artery disease, though results were statistically significant only for patients without coronary artery disease for mortality and stroke outcomes. Overall results were statistically nonsignificant but in favor of statin versus control for risk of hospitalization due to congestive heart failure. Risk for most composite vascular outcomes was significantly lower in CKD patients assigned statin treatment. In results available from only one trial, there was no difference between statin and control treatment groups regarding risk of ESRD or a composite outcome of ESRD or GFR decline by at least 50 percent. Only one trial reported on doubling of serum creatinine, but it had very few events.

While the magnitude of effect sizes favoring statins for many vascular outcomes, if real, seemed large enough to be clinically meaningful, results were limited by the small number of events in many studies and by small sample sizes in some others, in particular for analyses evaluating results in separate primary and secondary prevention subgroups. Results also were limited because, with one exception, studies were post hoc subgroup analyses from large statin trials that were not originally designed to evaluate CKD patients and renal outcomes. Consequently, there are almost no data on any renal outcome or on any vascular or renal outcome as a function of baseline albuminuria. Because most trials excluded patients with moderate and/or severely impaired renal function, available results may not be generalizable to these populations. Another limitation was that though composite vascular endpoints were reported in nearly all trials, the variability in their definitions prevented statistical pooling. Finally, few studies provided data on withdrawals or adverse events, so there was little information available regarding the relative tolerability and safety of statins versus control treatments in this population.

Overview

In patients with CKD, there is a low level of evidence that there is no difference in risk of mortality between treatment with high versus low dose statin. There is insufficient evidence regarding whether there is a difference between high and low dose statin in risk of ESRD. Our confidence in these estimates is limited by the small number of trials reporting different outcomes and the small number of clinical events.

Description of Study

Two trials met all eligibility criteria and randomized participants with CKD to high versus low dose HMG-CoA reductase inhibitor treatment.^151,152 The first study was a post hoc analysis in 3,107 individuals with eGFR <60 ml/min/1.73m² from among the 10,003 enrolled in the TNT trial. The second study was a post hoc analysis in 1,686 individuals with a eGFR <60 ml/min/1.73m² from among the 12,064 enrolled in the SEARCH trial. Detailed baseline characteristics are presented in Appendix Tables C124 and C125.

Patients were randomized to atorvastatin 10 mg daily versus atorvastatin 80 mg daily in the TNT trial and to simvastatin 20 mg daily versus simvastatin 80 mg daily in the SEARCH trial. Only the TNT trial provided baseline characteristics for the CKD participants. The mean age of participants in the TNT post hoc analysis was 66 years, and men constituted 68 percent of patients. Ninety-five percent of study participants were white. Both trials were multinational. Mean or median study duration ranged from 5.0 to 6.7 years.

Renal Function

Study participants were required to have eGFR <60 ml/min/1.73m² (i.e., CKD stage 3 or worse), but did not otherwise report participant distribution by CKD stage. Mean eGFR was 53 ml/min/1.73m² in the TNT study. No other measures of renal function were reported.

Baseline Comorbidities

All participants in the TNT trial had coronary artery disease and hyperlipidemia while off cholesterol medications, while participants in the SEARCH trial had a history of myocardial infarction and either were taking or considered to have a clear indication for statin therapy. Additional comorbid conditions from TNT participants included hypertension (63 percent) and diabetes (18 percent). Mean baseline blood pressure was 133/78 mm Hg. In the TNT trial, all participants completed an 8 week open-label run-in of atorvastatin 10 mg daily, and only those with LDL cholesterol less than 130 mg/dL were considered for enrollment, mean baseline total cholesterol and LDL cholesterol were 176 mg/dL and 96 mg/dL, respectively. In the SEARCH trial, all participants completed a run-in period of treatment with 20 mg of simvastatin daily.

Study Quality

(Appendix Table C140)

Study quality was rated as fair in one trial and good in one trial. Both trials reported that they were double blind. However, in the TNT post hoc analysis it was unclear whether allocation concealment was adequate, and analysis was not by intention to treat. Only 0.4 percent of TNT participants withdrew from the overall study, but withdrawals in the CKD subset were not reported. In the SEARCH trial, allocation concealment was adequate, analysis was intention to treat, and withdrawals were adequately reported.

Results

Summary

In individuals with CKD defined by reduced eGFR, high dose statin did not reduce all-cause mortality but significantly reduced risk of hospitalization attributed to CHF and risk of all defined composite vascular endpoints in the TNT trial, but did not reduce risk of the single reported composite vascular outcome in the SEARCH trial. There were no data reported for the individual outcomes of MI, stroke, ESRD, doubling of serum creatinine, halving of GFR, progression from microalbuminuria to macroalbuminuria, or for any composite renal outcome. Results were limited because they were based on two post hoc analyses, there were no data comparing treatment results in patients with albuminuria, and results were not reported for many vascular outcomes or any renal outcomes of clinical interest.

Overview

In patients with CKD, there is insufficient evidence regarding whether there is any difference between these treatments for the outcomes of mortality or ESRD. Our confidence in these estimates is limited because data are drawn from only one trial and there were few reported clinical events.

Description of Study

One trial met all eligibility criteria and randomized 86 participants with CKD to an HMG-CoA reductase inhibitor versus a bile acid sequestrant.¹⁵³ Detailed baseline characteristics are presented in Appendix Tables C124 and C125.

Participants were randomized to simvastatin versus cholestyramine. The mean age of subjects was 62 years. No data on gender or race/ethnicity was reported. Followup duration for this study, based in a single site in Italy, was 4 years.

Renal Function

The study did not base eligibility on CKD stage or report baseline distribution of participants by CKD stage. All participants were required to have microalbuminuria (urine albumin-to-creatinine ratio between 30 and 300 μg/mg) and at least a small measurable decline in GFR in the past 3 years. Mean GFR was 91 ml/min/1.73m2. Mean urine albumin/creatinine ratio was 83 μg/mg

Baseline Comorbidities

Eligible patients had treated hypertension and type 2 diabetes. Mean systolic blood pressure was 131 mm Hg and mean diastolic blood pressure was 76 mm Hg. No information on other comorbid conditions was reported.

Study Quality

(Appendix Table C140)

Study quality was rated as fair. Though the adequacy of treatment allocation concealment was unclear, the study was double blinded and analysis was conducted using the intention-to-treat principle. Withdrawals were adequately described and five percent of participants withdrew from the study.

Results

Summary

In patients with CKD defined by microalbuminuria, hypertension and diabetes, simvastatin significantly reduced risk of conversion to macroalbuminuria as compared with cholestyramine. There were no between-treatment group data for the endpoints of mortality, MI, stroke, CHF, ESRD, doubling of serum creatinine, halving of GFR, or for any composite vascular or renal outcome. Results were limited because they were based on a single small trial, and there were no between-treatment results for any vascular outcome or any other renal outcomes of clinical interest.

Overview

In CKD patients defined by impaired GFR, we found a low level of evidence that there is no difference between gemfibrozil and placebo for risk of mortality. There was insufficient evidence regarding whether gemfibrozil and a low triglyceride diet differ for risk of mortality. There was insufficient evidence regarding whether gemfibrozil differs from either placebo or a low triglyceride diet for risk of ESRD. Our confidence in these estimates is limited because for each comparison data are drawn from only one trial and there were few reported clinical events.

Description of Studies

Two trials met all eligibility criteria and randomized participants with CKD to gemfibrozil versus a control treatment.^137,149,154 The largest of the two trials involved a post hoc analysis involving 470 participants with GFR <60 ml/min/1.73m² from the larger (n=2,531) VA-HIT trial. Detailed baseline characteristics are presented in Appendix Tables C124 and C125.

Participants in the post hoc VA-HIT trial analysis were randomized to gemfibrozil versus placebo. The mean age of participants in this analysis was 67 years, all participants were male U.S. veterans, and 91 percent of study participants were white. Followup for this multinational study was 5.3 years.^149,154 The second study randomized 57 nondiabetic patients to gemfibrozil versus a low triglyceride diet. The mean age of study participants was 51 years, and men constituted 75 percent of study participants. No data on race/ethnicity were reported. Followup for this single-site Swedish study was 1 year.¹³⁷

Renal Function

In the VA-HIT post hoc analysis, participants were required to have GFR <60 ml/min/1.73m² (CKD stage 3 or worse).^149,154 All participants in the larger VA-HIT study had been required to have baseline serum creatinine ≤2.0 mg/dL. Mean GFR was 52 ml/min/1.73m². Mean creatinine clearance was 60 ml/min/1.73m². The second study did not base eligibility on CKD stage. The second study enrolled patients with impaired GFR (10 to 70 ml/min/1.73m²).¹³⁷ Mean GFR was 36 ml/min/1.73m² and mean albuminuria was 0.95 g/24 hours. Mean serum creatinine was 2.4 mg/dL. Neither study reported baseline distribution of participants by CKD stage

Baseline Comorbidities

All participants in the VA-HIT trial had coronary heart disease, LDL cholesterol ≤140 mg/dL, and HDL cholesterol ≤40 mg/dL. Additional comorbid conditions included hypertension (67 percent) and diabetes (30 percent). Mean baseline systolic blood pressure was 134 mm Hg, and mean diastolic blood pressure was 77 mm Hg. Mean total and LDL cholesterol were 176 mg/dL and 111 mg/dL, respectively. The second study excluded individuals with diabetes.¹³⁷ Mean baseline systolic blood pressure was 137 mm Hg, and mean diastolic blood pressure was 84 mm Hg. Mean total and LDL cholesterol were 244 mg/dL and 170 mg/dL, respectively. No other comorbidity data were reported.

Study Quality

(Appendix Table C140)

Study quality of the VA-HIT post hoc analysis was rated as good. The adequacy of treatment allocation concealment in the first study was clear. The study was double blinded, including outcome adjudication by a blinded endpoint committee. Analysis was performed using the intention-to-treat principle. No study participants were reported as lost to followup.^149,154 Study quality of the second study was rated as fair. The adequacy of treatment allocation concealment in the second study was unclear. The study was open label, and analysis was not performed using the intention-to-treat principle. Withdrawals were adequately described, and 11 percent of study participants withdrew from the study.¹³⁷

Results

Summary

In male veterans with CKD defined by impaired GFR, coronary artery disease, LDL ≤140 mg/dL and HDL ≤40 mg/dL, gemfibrozil did not reduce all-cause mortality compared with placebo. In the one composite vascular endpoint reported of the two the study defined, gemfibrozil significantly reduced risk of fatal CHD, nonfatal MI, or stroke. In both studies, too few (or no) patients developed ESRD to effectively compare risk between gemfibrozil and either placebo or low triglyceride diet. The gemfibrozil versus diet study suggested an increased risk of gastrointestinal side effects with gemfibrozil, but the VA-HIT CKD study reported no information on the incidence of adverse gastrointestinal symptoms. Results were limited because they were based on small studies, with few reported outcomes and small numbers of clinical events. Results from the VA-HIT study were limited because they were a post hoc analysis from a larger trial not designed to look at CKD patients or renal outcomes. Studies are limited in that they do not also report results based on baseline albuminuria.

Overview

In patients with CKD, we found a low strength of evidence that there is no difference between intensive, multicomponent treatment and conventional treatment for risk of all-cause mortality or ESRD. We found no statistically significant difference between treatment groups in risk of cardiovascular mortality, MI, or stroke. Risk of conversion from microalbuminuria was statistically significantly lower in the intensive treatment group. Our confidence in these estimates is limited by the small number of trials reporting different outcomes, the small number of clinical events, and heterogeneity between studies.

Description of Studies

Five reports of four unique trials met all eligibility criteria and randomized participants with CKD (n=892 patients, range 90 to 437) to an intensive multicomponent treatment intervention versus usual care.^155-159 Detailed baseline characteristics are presented in Appendix Tables C132 and C133.

In all eligible trials, the intensive treatment arm was implemented by a multidisciplinary research team, comprised of at least a physician, a nurse, and a dietitian. In three of these trials, conducted entirely in patients with diabetes, the research team met with the patients at least every three months and directly intervened in their care, treating them to achieve explicit targets for blood pressure (systolic <130 to 140 mm Hg, diastolic <80 to 85 mm Hg), diabetes (HbA_1c targets ranged from <6.5 to <8 percent), and lipid control (cholesterol <154 to 193 mg/dL, LDL <100 mg/dL, HDL >42 mg/dL, triglycerides <66 to 75 mg/dL).^155,156,158 The interventions were introduced in a stepwise fashion, including behavior modification and pharmacologic therapy, as necessary. In the fourth trial, the research team implementing the intensive treatment arm met with patients every 3 to 6 months and utilized a mix of direct intervention and letters sent with management recommendations to the patients' primary care providers.¹⁵⁹ In this latter trial, while improved blood pressure control was a stated aim, no explicit blood pressure target was reported, and the study did not discuss management of diabetes or lipids. An emphasis was placed on improving medication compliance and decreasing nephrotoxic drug exposure. In three trials, ACEIs or ARBs were to be initiated in all intensive treatment group participants,^155,156,159 and although the fourth trial did not state that these medications were mandated, it reported ACEI use in 95 percent of enrolled participants at followup.¹⁵⁸ Within the intensive treatment intervention groups, dietary recommendations in three trials included low protein,^155,158,159 with low potassium recommended in two trials.^155,159 Low fat¹⁵⁶ and low sodium¹⁵⁸ each were part of diet recommendations in one trial.

By comparison with the intensive intervention arms, all study participants assigned to control treatment groups were managed by their primary physician. In two trials, their management was left entirely to the discretion of their primary physician.^155,159 However, in two other trials their doctors were to target explicit goals for blood pressure, diabetes, and lipid control, aiming either for the same thresholds being used for treatment of the intensive treatment group,¹⁵⁸ or following national guidelines that were modestly less strict than the thresholds targeted for the intensive treatment group.¹⁵⁶

The mean age of subjects was 65 years (range 55 to 68; n=4 trials) and men constituted 52 percent (range 34 to 74; n=4 trials) of all patients randomized. In the only trial that reported data on ethnicity/race, 80 percent of participants were African American.¹⁵⁹ Two trials were conducted in Europe (including Scotland and Denmark),^156,158 one was conducted primarily in the United States,¹⁵⁹ and one was conducted in China.¹⁵⁵ Mean or median study durations ranged from 2 to 7.8 years.

Renal Function

No study based eligibility on CKD stage or reported baseline distribution of participants by CKD stage. Among the four eligible trials, two based participant eligibility on presence of albuminuria,^156,158 while two others determined eligibility based on impaired creatinine clearance and/or elevated serum creatinine.^155,159 In trials reporting these data, mean baseline creatinine clearance was 37.6 mL/min (range 34 to 55, n=2 trials),^158,159 mean baseline serum creatinine was 1.8 (range 0.9 to 2.1, n=2 trials),^156,159 and urinary albumin excretion rate ranged from a mean of 73.5 mg/24 hours in one trial¹⁵⁶ to a median of 755 mg/24 hours in a second trial.¹⁵⁸ In addition, one trial reported a baseline mean GFR of 117 ml/min/1.73m²,¹⁵⁶ and another reported a mean albumin-to-creatinine ratio of 79 mg/mmol.¹⁵⁸

Baseline Comorbidities

Hypertension prevalence was reported in three trials, within which 98 percent of participants had a diagnosis of hypertension.^155,158,159 Mean systolic and diastolic blood pressure measurements were 147 and 82 mm Hg, respectively (n=4 trials). Three trials were comprised entirely of type 2 diabetic patients,^155,156,158 with the fourth trial including 44 percent diabetic participants.¹⁵⁹ In the two diabetic trials reporting, mean baseline HbA_1c was 8.3 percent (range 7.9 to 8.6).^156,158 The prevalence of other comorbidities included coronary artery disease 35 percent (range 16 to 48, n=3 trials), CHF 30 percent (range 7 to 40, n=2 trials), MI 26 percent (range 2 to 37, n=2 trials), and stroke 16 percent (range 3 to 20, n=3 trials).

Study Quality

(Appendix Table C140)

Among the four eligible trials, study quality was rated as good for one trial and as fair for three trials. Allocation concealment was adequate in three trials and unclear in the remaining study. All of these intensive multicomponent intervention trials were open label. Analysis by the intention-to-treat principle was performed in two trials. Reasons for study withdrawal were adequately described in all reports, and 2.6 percent (range 0 to 17) of randomized participants withdrew from trials overall.

Results

Mortality

(Table 18, Appendix Table C134, and Appendix Figure C25)

Table 18

Pooled clinical and renal outcomes, INT versus control treatment trials.

All-Cause Mortality

Compared with control treatment, assignment of CKD patients to an intensive, multicomponent intervention did not significantly reduce risk of all-cause mortality (19.5 percent versus 23.3 percent; RR 0.86, 95% CI, 0.67 to 1.10; n=4 trials, 892 patients).

Cardiovascular Mortality

Assignment to the multicomponent treatment group was not associated with a significant difference in risk of cardiovascular mortality compared with control treatment (RR 1.07, 95% CI, 0.47 to 2.43; n=2 trials).

Summary

In individuals with CKD, compared with usual care, assignment to intensive, multicomponent intervention was not associated with a significant reduction in risk of all-cause mortality. Further, there was no significant association between treatment groups and risk of MI, fatal stroke, and ESRD. In data from single trials only, there was a significantly reduced risk with intensive, multicomponent treatment for the outcomes of nonfatal stroke, a composite vascular endpoint, and conversion from microalbuminuria to macroalbuminuria. Results for all outcomes, with the possible exception of all-cause mortality, were limited by small sample sizes and few events and could not exclude either clinically meaningful benefits or harms. Overall results were further limited by heterogeneity in patient populations and in treatment protocols, including those for both the intensive intervention groups and the usual care groups. Reporting on study withdrawals and adverse effects was limited. Finally, no trial provided followup beyond 5 years; therefore, longer term effects of intensive, multicomponent interventions cannot be determined from these studies.