The Toxic Condition of Hypercalcemia and Hypercalciuria

Hypercalcemia occurs when serum calcium levels are 10.5 mg/dL (also expressed as 2.63 mmol/L) or greater depending on normative laboratory values. It can be induced by excess intake of calcium or vitamin D, but it is more commonly caused by conditions such as malignancy and primary hyperparathyroidism (Moe, 2008). Clinical signs and symptoms of hypercalcemia may vary depending on the magnitude of the hypercalcemia and the rapidity of its elevation; they often include anorexia, weight loss, polyuria, heart arrhythmias, fatigue, and soft tissue calcifications (Jones, 2008). When serum calcium levels rise above 12 mg/dL, the kidney's ability to reabsorb calcium is often limited; in turn, hypercalciuria can occur, particularly with increased calcium or vitamin D intake. Hypercalciuria is present when urinary excretion of calcium exceeds 250 mg/day in women or 275-300 mg/day in men. Often, urinary calcium excretion is expressed as the ratio of calcium to creatinine excreted in 24 hours (milligrams of calcium per milligram of creatinine). Values above 0.3 mg/mg creatinine are considered to be within the hypercalcuric range.

Hypercalcemia, in addition to leading to hypercalciuria, can cause renal insufficiency, vascular and soft tissue calcification including calcinosis leading to nephrocalcinosis, and nephrolithiasis. Nephrolithiasis, often referred to as kidney stones, can also be caused by hypercalciuria. Hypercalciuria may occur in the absence of hypercalcemia and is related to either hyperabsorption of calcium in the gut or a renal leak whereby calcium excretion is enhanced. Both etiologies can lead to nephrocalcinosis.

In the North American population, as many as 30 percent of persons ages 60 years or older have some degree of renal insufficiency (Coresch et al., 2005; Szczech et al., 2009). Decreased renal functioning may make persons more sensitive or susceptible to the effects of excess calcium or vitamin D intake. Urinary calcium excretion decreases in older adults, although it is not clear the extent to which this may be due to decreased calcium intake as compared to decreased renal function. However, if due to decreased renal function, older persons may be at higher risk for adverse effects derived from excess intakes. Moreover, decreased renal function simultaneously increases cardiovascular disease (CVD) risk and impairs calciuric responses and calcium and phosphate homeostasis. Likewise, those using thiazide-based diuretics—a sizeable proportion of older adults—are more readily challenged by excess calcium and vitamin D due to reduction in calcium excretion from the kidney (Medarov, 2009).

Excess Calcium and Hypercalcemia Leading to Renal Insufficiency

Prior to the introduction of histamine-2 blockers and proton pump inhibitors, liquid formulations that contained high calcium levels and absorbable alkali were used to treat gastric and duodenal ulcers. High intake of these formulations, however, caused a variety of adverse effects including hypercalcemia and renal failure. The syndrome became known as “milk-alkali syndrome” or MAS (Hardt and Rivers, 1923; Burnett et al., 1949) and was originally associated with men with peptic ulcer disease. In this context, hypercalcemia causes emesis and natriuresis, which result in significant drops in extracellular blood volume. This contraction worsens the hypercalcemia. Decreases in blood volume also induce an alkalotic state that causes an increase in proximal tubular bicarbonate resorption. Also, high serum calcium levels worsen the alkalosis through suppression of parathyroid hormone (PTH), which physiologically enhances bicarbonate excretion. Although the availability of absorbable alkali in the diet may enhance the alkalotic state, it is not the major pathogenic factor in MAS.

Recently, Patel and Goldfarb (2010) suggested renaming MAS as “calcium-alkali syndrome” to better reflect the current understanding of the disorder, which now has shifted to be more prevalent in other groups including postmenopausal women. The earlier MAS presented with hyperphosphatemia after prolonged ingestion of phosphorus-containing milk with cream (Patel and Goldfarb, 2010). In contrast, the modern version of the syndrome is associated with hypophosphatemia or low-normal serum phosphorus levels as a result of the phosphorus-binding properties of calcium carbonate. The hypophosphatemia is more pronounced in elderly patients or those with eating disorders, who tend to have relatively low consumption of protein and therefore phosphorus (Picolos et al., 2005; Felsenfeld and Levine, 2006; Medarov, 2009). Confounding this, however, is the chronic renal insufficiency that often accompanies MAS; in that case, serum phosphorus levels may be normal or high. Available case reports tend to provide only serum calcium levels and do not specify calcium intakes per se, or serum phosphate, associated with the condition.

As shown in Table 6-1, a number of recent case reports have been identified for calcium-alkali syndrome. For these individuals, a calcium intake of 3,000 mg/day was associated with the onset of hypercalcemia. However, in every case except one, all outcomes were found in individuals with impaired renal function and high serum creatinine levels. The one exception (Nabhan et al., 2004) was a patient who was using hydrochlorothiazide as a diuretic and was hypoparathyroid. Although these data cannot be applied directly to the normal, free-living population, they are informative and indicate that calcium levels of 3,000 mg/day are problematic for these compromised persons.

TABLE 6-1

Case Reports of Calcium-Alkali Syndrome.

Patel and Goldfarb (2010) suggested that the incidence of calcium-alkali syndrome is growing as a result of the widespread use of over-the-counter calcium and vitamin D supplements, particularly among older persons. The basis for the suggestion is that while healthy younger adults rely on the bone reservoir to buffer excess calcium, the net flux of calcium is out of the bone for older persons, thereby making the bone less functional as a reservoir. These older persons are more susceptible to the syndrome when they begin taking supplemental calcium. Patel and Goldfarb (2010) also noted that the excess ingestion of calcium with or without vitamin D is an integral feature of this syndrome, making it potentially relevant to the consideration of upper levels of calcium intake.

Excess Calcium and Soft Tissue Calcification

Excess Calcium and Nephrolithiasis (Kidney Stones)

More than 12 percent of men and 6 percent of women in the general population will develop kidney stones (Stamatelou et al., 2003). The morbidity of kidney stones is not limited to the pain of stone passage; stones increase the risk of renal and urinary tract infections as well as renal insufficiency. A contributing factor in stone formation is hypercalciuria from any cause; another is hyperabsorption of calcium from the gut. Hypercalciuria increases the risk for nephrolithiasis (Pak and Holt, 1976). Hypercalciuria can be present in the absence of hypercalcemia and may reflect routine excretion of excess calcium intake.

Incidence rates for kidney stones vary by age and gender. The rates are highest in men, rising after age 20, peaking between 40 and 60 years, and then beginning to decline (Johnson et al., 1979; Hiatt et al., 1982; Curhan et al., 1993). For women, incidence rates seem to be higher in the late 20s, decreasing by age 50, and then remaining relatively constant (Johnson et al., 1979; Hiatt et al., 1982; Curhan et al., 1997, 2004).

Although calcium is present in approximately 80 percent of kidney stones (Coe et al., 1992), the role of calcium and other nutrients, acting alone or in concert as risk factors, is not completely understood and may be a function of physiological context. Various dietary and non-dietary factors are associated with stone formation, making data difficult to interpret. Rodent models that have been used to explore the effect of dietary factors on the propensity to form calcium oxalate and calcium phosphate stones suggest that the role of supplemental calcium in determining risk for nephrolithiasis varies by interaction with a given dietary component. One study in rats compared renal oxalate crystallization relative to the consumption of calcium-supplemented or oxalate-rich diets as well as control diets. The study found that rats fed the calcium-supplemented diet had enhanced calcium and oxalate accumulation as well as crystallization in renal tissues, even though urinary oxalate and citrate excretion was not significantly different in rats fed the control diet (Mourad et al., 2006). In this study, measures of renal function, including glomerular filtration rate, fractional excretion of urea, and fractional reabsorption of water and magnesium were not affected by the calcium-supplemented diet, and calciuria was only slightly increased.

Excess Calcium and Prostate Cancer

The vast majority of the data relating to prostate cancer and calcium intake are derived from observational studies, and the ability to sort the effect of dairy products from that of calcium is challenging. Some observational data suggest a role for dairy products as a risk factor for prostate cancer (Tominaga and Kuroishi, 1997; Grant, 1999). A recent case–control study examined associations between dairy products and dietary calcium and prostate cancer risk among men ages 35 to 84 years with a histological diagnosis of prostate cancer (Raimondi et al., 2010). Intake of dairy products, in particular milk consumption, was associated with a two-fold increased risk for prostate cancer, whereas consumption of other dairy products (cheese, yogurt, and cream) suggested no association for increased risk. Total calcium intake was not significantly associated with risk for prostate cancer (p = 0.09).

Other observational studies evaluating associations between milk or dairy product intake and overall risk for prostate cancer have suggested that supplemental calcium intake may be a stronger risk factor for prostate cancer than calcium from foods, particularly for aggressive prostate cancer with high mortality. Studies of associations between calcium supplement use and risk for incident prostate cancer provide mixed results. A small case–control study assessed men ages 40 to 64 years with newly diagnosed prostate cancer for multivitamin and supplement use by questionnaire (Kristal et al., 1999). Although about a third of both cases and controls reported using multivitamins, only 5 percent reported taking calcium supplements. No association was found between calcium supplement use and risk for incident prostate cancer in this relatively young and low-risk population. A prospective study of male participants ages 50-74 years from the Cancer Prevention Study II examined associations between calcium and dairy product intake and risk for incident prostate cancer (Rodriguez et al., 2003). This analysis of 65,321 men found a small increase in overall prostate cancer risk for calcium intakes of 2,000 mg/day and higher compared with intakes less than 700 mg/day. High calcium intake (≥ 2,000 mg/day), however, was significantly associated with risk for advanced prostate cancer. When calcium supplements (≥ 500 mg/day) were analyzed, controlled for total calcium intake, a weak association was found for prostate cancer risk. Dairy product intake was not associated with risk for prostate cancer. The report from the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR, 2007) concluded that the relationship between prostate cancer and milk and dairy product intake is inconsistent from both cohort and case–control studies, and there is limited evidence suggesting that milk and dairy products are a cause of prostate cancer. A food-use questionnaire administered at baseline to participants in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) Study explored associations between intake of certain foods and nutrients and risk for incident prostate cancer in a large cohort of male smokers ages 50 to 69 years. This study analyzed intake of calcium and dairy foods and found no associations with development of prostate cancer (Chan et al., 2000). In a longitudinal follow-up of this cohort, Mitrou et al. (2007) found a graded positive association between increasing total calcium intake and total prostate cancer risk. A prospective study of male participants ages 40 to 75 years from the Health Professionals Follow-Up Study (HPFS) examined whether calcium and fructose intake were risk factors for prostate cancer (Giovannucci et al., 1998). Calcium intake exceeding 2,000 mg/day was found to be associated with higher risk for total, advanced, and metastatic prostate cancer. Further, supplemental calcium intake above 900 mg/day was associated with metastatic prostate cancer risk at all levels of total calcium intake. In a follow-up analysis of this cohort, Giovannucci et al. (2006a) found a significantly increased risk for advanced prostate cancer associated with increasing total calcium intake and for fatal prostate cancer associated with supplemental calcium intakes of 401 mg/day and above. The WCRF/AICR (2007) concluded that there is a probable association between diets high in calcium and prostate cancer.

In the case of intervention studies, one randomized controlled multicenter clinical trial based on 672 men (mean age 61.8 years) living in the United States examined risk for prostate cancer from supplemental calcium intake. Participants received either 3 g of calcium carbonate or placebo daily for 4 years and were followed for up to 12 years for prostate cancer diagnosis (Baron et al., 2005). Over the entire study period, risk for prostate cancer was lower in the calcium-supplemented group than in controls (relative risk [RR] = 0.83; 95% CI: 0.52–1.32) but was not statistically significant. For specific years in the study, increase in prostate cancer risk was statistically significant between baseline and year 6 (RR = 0.52; 95% CI: 0.28–0.98) and between years 2 and 6 (RR = 0.44; 95% CI: 0.21–0.94). No significant differences were found for total calcium intake and prostate cancer risk.

The 2009 analysis from the Agency for Healthcare Research and Quality (Chung et al., 2009; hereafter referred to as AHRQ-Tufts) examined 12 cohort studies that reported on the association between calcium intake and the risk of prostate cancer (Schuurman et al., 1999; Chan et al., 2001; Rodriguez et al., 2003; Baron et al., 2005; Tseng et al., 2005; Giovannucci et al., 2006a; Koh et al., 2006; Mitrou et al., 2007; Park et al., 2007a,b; Rohrmann et al., 2007; Kurahashi et al., 2008). One of the studies also provided a post hoc analysis of a randomized controlled trial on calcium supplementation. The incidence of prostate cancer in these studies ranged from 0.008 to 0.10. Most of the studies were conducted in Europe or North America, and one study was conducted in Japan. The mean age of the subjects ranged from 53 to 67 years, but did include men 51 to 70 years of age. No study specifically targeted men older than 70 years of age. Total calcium intake ranged from less than 500 mg/day to at least 2,000 mg/day. The time between dietary assessment and the diagnosis of prostate cancer varied from 1 to 17 years. AHRQ-Tufts rated the studies for methodological quality as follows: four studies were rated A, seven studies were rated B, and one study was rated C. The studies included participants in the age range of 51 to 70 years. Seven studies did not find an association between calcium intake and the risk of prostate cancer (Baron et al., 2005; Koh et al., 2006; Mitrou et al., 2007; Park et al., 2007a,b; Kurahashi et al., 2008). The remaining five studies found that the risk was higher in the groups that took more calcium compared with those that took a lower amount; the higher amount ranged from 921 to at least 2,000 mg of calcium per day.

Overall, data in this area are at best emerging. Although observational studies suggest that total calcium intake of 2,000 mg/day or higher may be associated with increased risk for prostate cancer and particularly with advanced and metastatic cancer, these data are not sufficiently robust to serve as an indicator for a UL. The one available trial was negative. The observations, however, are notable for levels of intake that are less than those that produce hypercalcemia and hypercalciuria.

Excess Calcium and Nutrient Interactions: Iron and Zinc

Despite the absence of clinically or functionally significant depletion of relevant mineral nutrients, calcium interaction with other minerals in the diet has been considered a potential risk related to high calcium intakes. The 1997 DRI report (IOM, 1997) specifically called for increased study in this area. However, data remain limited.

With respect to iron, Ilich-Ernst et al. (1998) carried out a placebo-controlled randomized trial assessing the effects of calcium supplementation on bone mass in adolescent girls ages 8 to 13 years (n = 354). A secondary analysis at year 4 of this 7-year trial found that girls in the supplemented group achieved a total calcium (food plus supplements) intake of 1,500 mg/day. When assessed for interactions between calcium and iron, measures of iron status—hemoglobin, hematocrit, and corpuscular indexes—were not significantly different from those of girls in the placebo group who reached a calcium intake of 800 mg/day. Ames et al. (1999) found no effect of a calcium intake of approximately 1,200 mg/day compared with 500 mg/day for 5 weeks on iron absorption in children 3 to 5 years of age.

With respect to zinc, McKenna et al. (1997) conducted a calcium and zinc balance study on a subset (n = 26) of participants in a longitudinal clinical trial of the effects of calcium supplementation on bone mass in girls with a mean age of 11 years. Trial participants received either 1,000 mg/day of supplemental calcium or a placebo. Mean calcium intake reached 847 ± 287 and 821 ± 224 mg/day from diet for placebo and intervention groups, respectively, at 6 months. With the additional supplement, the mean calcium intake in the intervention group exceeded 1,700 mg/day. The results of the balance study found no effect in the intervention group from intake of approximately 1,700 mg of calcium per day on net zinc absorption, zinc excretion, or zinc balance compared with intakes of approximately 800 mg/day in the placebo group.

Taken together, the studies suggest that calcium intakes of 1,500 to 1,700 mg/day do not interfere with iron or zinc absorption in adolescent girls. However, as calcium intakes among this age group could be higher than those studied, there is little evidence to shed light on the larger issue.

Excessive Calcium and Constipation

Calcium supplement intake has long been associated with constipation. In fact approximately 1 of every 10 participants in the WHI calcium–vitamin D supplementation trial reported moderate to severe constipation (Jackson et al., 2006). If a food source of calcium is the problem, the constipation is likely due to the components of dairy products (Anthoni et al., 2009) rather than to the calcium in food. Calcium supplements, which are regarded as “binding,” can cause side effects for some people, such as constipation and gas (Jackson et al., 2006; Prince et al., 2006), which varies greatly from person to person. Usually the constipation is alleviated by increasing intakes of water or fiber-rich foods, or by trying another form of supplement (calcium citrate may be less constipating than calcium carbonate, for example). Although such conditions warrant attention, the utility of constipation as an indicator for DRI development is doubtful.

Selection of Indicator for Calcium UL

The risk assessment framework, as described in Chapter 1, specifies that, in the case of ULs, the available data pertaining to adverse effects be first examined for evidence of a benchmark intake (BI). Alternatively, either a no observed adverse effect level (NOAEL) or a lowest observed adverse effect level (LOAEL) is considered. In the case of calcium, limited new information has become available since 1997. The indicators selected are calcium excretion for younger age groups, and kidney stone formation for older age groups. The calcium excretion data provide information for a group for which no data were available in 1997. The newer data on kidney stone formation form a basis for a UL that is more akin to conditions experienced by the normal, healthy population than is calcium-alkali syndrome, although the cautions expressed by Patel and Goldfarb (2010) concerning the vulnerability of older persons to calcium-alkali syndrome with the use of calcium supplements are worthy of note.

The available data could not offer a BI or be used to estimate a dose–response relationship. The basis for the ULs is a NOAEL for infants and a LOAEL for adults, as described further below for specific life stage groups.

ULs for Infants 0 to 12 Months of Age

In the previous 1997 DRI report (IOM, 1997), a UL for calcium for infants was not specified owing to lack of data. The 1997 report noted a small, randomized trial of 81 infants (103 at baseline, ages 2.5 to 5.0 months at entry) designed to examine the tolerance of calcium-supplemented infant formula through 9 months of age (Dalton et al., 1997). The data, as analyzed in 1997 (IOM, 1997), indicated only that infants fed calcium at up to approximately 1,750 mg/day experienced no adverse effect on iron status.

Using these same data, Sargent et al. (1999) later reported on calcium excretion measures, and this measure serves as the UL indicator for infants. This 1999 report has provided the ability to estimate a NOAEL for calcium intake for infants based on calcium excretion. Within the confines of the limitations of the data, they suggest that infants can tolerate approximately 1,750 mg of calcium per day with no noted adverse effects. A NOAEL of 1,750 mg/day is therefore established for infants on this basis.

ULs for Children and Adolescents 1 Through 18 Years

New data on adverse outcomes due to excess calcium intake among children and adolescents—specifically data that would identify a NOAEL or LOAEL—have not emerged since the last DRI report on calcium in 1997 (IOM, 1997). At that time, it was noted that the safety of excess calcium intake in children and adolescents had not been studied. A UL of 2,500 mg of calcium per day was established in 1997 for all children and adolescents in these life stage groups, largely on the basis of the UL established for adults (i.e., 2,500 mg/day) (IOM, 1997).

There is currently no evidence that the 1997 level is too low to provide public health protection for this group; further, when compared with the new UL set for infants, the level of 2,500 mg of calcium per day is a reasonable increase given the expected increases in body weight and metabolic capacities, especially for younger children between the ages of 1 and 8 years.

However, for older children it is also appropriate to take into account the likely increases in tolerated intakes as metabolic demands increase and the pubertal growth spurt associated with bone accretions sets in, primarily between 9 and 18 years of age. Again, there are no data to allow quantitative uncertainty factors to be developed to mathematically correct for the likelihood of increased capacities during the bone growth spurt, but to do so in some fashion is consistent with a general toxicological approach. An added level of 500 mg/day is reasonable, resulting in a UL of 3,000 mg of calcium per day for children 9 to 13 years of age and adolescents 14 to 18 years of age.

The UL for children 1 through 8 years of age is the same as that established for these life stage groups in 1997 (IOM, 1997). However, the UL has been increased by 500 mg/day for older children and adolescents compared with 1997 (IOM, 1997). This is based on a biologically reasonable adjustment intended to take into account increased need and therefore increased capacity to tolerate a slight increase in a UL value.

ULs for Adults 19 or More Years of Age

The onset of hypercalcemia is clearly an adverse outcome. However, it was not selected as an indicator for ULs for adults because it reflects an extreme pathological condition, and the ability to consider other adverse events associated with sustained, high levels of intake has emerged. Specifically, kidney stone formation is an adverse outcome, notably among post-menopausal women. Although there is also evidence related to calcium-alkali syndrome among adults, most of the data relate to those with compromised kidney function. Vascular calcification in postmenopausal women has emerged as an interesting hypothesis, but available data are conflicting, and threshold levels for intake are unknown. Evidence related to prostate cancer, although concerning, was too confounded to allow this disease risk to serve as an indicator for establishing ULs for calcium intake. Further, neither constipation nor nutrient interactions were associated with data to suggest that these outcomes would serve as indicators for UL development.

Given the size and quality of the WHI trial, its outcome relative to the incidence of kidney stones (Jackson et al., 2006) results in the selection of kidney stones as the indicator for adults for DRI purposes. The levels of calcium intake that may cause kidney stones within a normal population cannot be specified with certainty and are known to be variable depending upon a number of factors, including baseline renal function, pre-existing disease conditions, and interactions with drugs. Based on the findings of Jackson et al. (2006), and with the understanding that the data are derived from women between the ages of 50 and 79 years, there is a concern for kidney stone risk at total calcium intakes of approximately 2,000 mg/day. Underpinning the concern is the recognition that intakes of calcium from food do not readily result in excess intakes and are not associated with adverse effects; rather, the adverse effects appear to be a function of calcium supplementation added to baseline intake. The level of 2,000 mg of calcium per day is established as the LOAEL for adults, including men, more than 50 years of age. The very limited data available for adults 19 to 50 years of age do not allow the specification of a LOAEL or NOAEL for this younger group and the UL for this group is derived from considerations used for the UL for persons above 50 years of age. For this reason the ULs for older adults are discussed first below, followed by adults 19 to 50 years of age.

ULs for Pregnancy and Lactation

Hypercalciuria is often present during normal pregnancy as a consequence of the doubling of intestinal calcium absorption that occurs, and pregnancy itself increases the risk of kidney stones. Consequently, excess intakes of calcium during pregnancy will aggravate hypercalciuria and possibly increase the risk of kidney stones. During lactation, the serum calcium (both ionized and albumin-corrected total calcium) level rises and usually remains within the normal range (although hypercalcemia can occur during normal lactation), and urinary excretion of calcium is reduced to the low-normal range or below. Consequently, higher intakes of calcium during lactation could potentially increase the risk of hypercalcemia. However, there is no evidence to suggest that the risk manifests itself at intakes lower than the UL for non-pregnant or non-lactating women, although it is acknowledged that relevant studies have not been rigorously carried out for pregnancy and lactation. Given that available evidence suggests that requirements for calcium among pregnant and lactating females are similar to those of non-pregnant and non-lactating females, and lacking data to suggest a basis for a different UL, the ULs for calcium for pregnancy and lactation have been kept the same as those for their non-pregnant and non-lactating counterparts.

Vitamin D Intoxication and Related Hypercalcemia

Serum 25OHD Concentrations as Indicative of Toxicity

In the absence of well-controlled studies, the serum 25OHD level representing the vitamin D toxicity threshold in humans is not readily defined. Similarly, the vitamin D intakes required to trigger toxicity symptoms are not precisely known. Moreover, even though the physiological changes that occur with vitamin D toxicity are correlated to serum 25OHD levels, they may not be precisely aligned (Towler, 2009) and may vary from subject to subject and among sub-populations. Appendix G summarizes a number of human vitamin D toxicity case studies gathered from the scientific literature from early in the 20th century to the present. Some of these reports originated from a time before vitamin D metabolism was discovered and hence lack confirmation that the causative agent was vitamin D overdose. Table 6-3 contains case reports from the past 35 years in which the data are supported by vitamin D dose administered, serum calcium levels, and serum 25OHD levels. Also provided are data from several month-long studies with a range of vitamin D supplements in which fully documented vitamin D intoxication was not identified. It is concluded that occasional reports of hypercalcemia in these studies are not related to vitamin D.

TABLE 6-3

Case Reports of Vitamin D Intoxication: Intake and Plasma Measures.

As shown in Table 6-3, the literature contains evidence that a range of vitamin D supplements from 800 to 300,000 IU/day have been used for periods ranging from months to years. Doses below 10,000 IU/day are not usually associated with toxicity, whereas doses equal to or above 50,000 IU/day for several weeks or months are frequently associated with toxic side effects including documented hypercalcemia.

DeLuca (2009) concluded that, overall, the toxicity of hypercalcemia becomes evident at vitamin D intakes above 25,000 IU/day, corresponding to a serum 25OHD level of about 500 nmol/L. Hathcock et al. (2007), following an analysis of more than 20 publications, concluded that there was no association between harm and intakes of 10,000 IU/day. Although toxic effects associated with 400 IU/day seem implausible,³ the diverse range of intakes and serum 25OHD levels is notable. Most reports suggest that the toxicity threshold is between 10,000 and 40,000 IU of vitamin D per day. Also, most do not identify toxicity until serum 25OHD levels of 500 to 600 nmol/L or higher are reached; frank toxicity has been associated with a serum 25OHD level of 750 nmol/L (Jones, 2008; Deluca, 2009).

There have been no reports of vitamin D intoxication by ultraviolet B light alone (Webb et al., 1989). Davie et al. (1982), using high-performance liquid chromatographic analysis, found that skin irradiation reached a plateau after 5 to 6 weeks of exposure and achieved a plasma 25OHD₃ level of no more than 45 nmol/L. In a study of healthy men (n = 26) who had just completed a summer of extended outdoor activity, Barger-Lux and Heaney (2002) found that the median serum 25OHD level was 122 nmol/L in late summer and decreased to 74 nmol/L by late winter. Binkley et al. (2007) found that among subjects (n = 93) with habitually high sun exposure (~ 29 hours/week), the mean serum 25OHD level was 79 nmol/L, with the highest reported level of 155 nmol/L. In a study of subjects (n = 50) who used a tanning bed on a regular basis (at least once a week), serum 25OHD levels were 90 percent higher in tanners than in controls (n = 106) (115.5 ± 8.0 nmol/L vs. 60.3 ± 3.0 nmol/L) (Tangpricha et al., 2004).

Excess Vitamin D and Serum Calcium

The 1997 IOM report on DRIs for vitamin D used the effect of vitamin D intake on serum calcium level in humans as the basis for developing ULs for vitamin D (IOM, 1997). The work of Johnson et al. (1980) and most notably that of Narang et al. (1984) were taken into account. In the Narang et al. (1984) study, serum calcium levels in humans (with and without tuberculosis) were measured as a function of daily vitamin D doses of 400, 800, 1,200, 2,400, and 3,800 IU for 3 months. Thirty healthy men and women ranging in age from 21 to 60 years and without tuberculosis were included in the study. Hypercalcemia with vitamin D supplementation was reported in 63 percent of the patients with active tuberculosis, consistent with the known effect of granulomatous diseases on enhancement of 1α-hydroxylase activity. In the 30 subjects reported to be normal, statistically significant increases in serum calcium level were observed with vitamin D doses of 2,400 and 3,800 IU/day; however, only at the dose of 3,800 IU/day did the serum calcium level exceed the upper limits of normal (i.e., 10.5 mg/dL). Moreover, there were only five subjects in the highest dose group, the duration of the effect was not reported, and the heterogeneity within that subgroup was reflected by the large standard error.

Although increased serum calcium levels are of concern, the Narang et al. (1984) study is likely too small to allow any conclusions to be drawn beyond the potential risk of hypercalcemia during vitamin D supplementation in patients with tuberculosis. More recently, Aloia et al. (2008) conducted a 6-month dose–response study using 138 white and African American adults to determine the intake of vitamin D₃ needed to achieve a targeted plasma 25OHD level. Doses of vitamin D varied but reached means of 3,915 ± 840 IU/day for blacks and 3,040 ± 1,136 IU/day for whites. No patient presented with a serum calcium level above 265 mmol/L (or 10.6 mg/dL).

Excess Vitamin D and Measures in Infants

Jeans and Stearns (1938) found a retarded linear growth rate in 35 infants up to 45 weeks of age who received daily doses of 1,800 to 4,500 IU of vitamin D as supplements (without regard to sun exposure), for a minimum of 6 months, compared with infants receiving supplemental doses of 340 IU/day or less. Fomon et al. (1966), in a similar study, explored the effects of vitamin D on linear growth in infants (n = 13) ingesting 1,380 to 2,170 IU/day (mean = 1,775 IU/day) of vitamin D from fortified evaporated milk formulas as the only source of vitamin D, compared with infants receiving 350 to 550 IU/day (n = 11) from another batch of formula. No effect was found in infants who were enrolled in the study from the first 9 days after birth up to 6 months of age. Newer data to better elucidate the relationship between vitamin D and retarded linear growth in infants have not emerged in recent years.

Reports in Britain in the 1950s, when foods were being liberally fortified with vitamin D, indicated an unusually large number of cases of “idiopathic hypercalcemia” (British Paediatric Association, 1956). Given the number of foods fortified at the time, the British Paediatric Association (1956) estimated an intake of about 4,000 IU of vitamin D per day for an infant who consumed a typical diet of milk (1.5 pints), cereal (1 ounce), and cod liver oil (1 teaspoon). The outbreak of idiopathic hypercalcemia that took place was attributed to vitamin D supplementation, but the cause cannot be determined with certainty. Survey data apparently reported a marked decline in hypercalcemia in infants, from 7.2 cases per month in a 1953 to 1955 survey, to 3.0 cases per month in a 1960 to 1961 survey (British Paediatric Association, 1956, 1964). This change occurred at the time new guidelines were introduced for fortification of food products with vitamin D. Data from the British Paediatric Association (1956) and Bransby et al. (1964) also suggested that the estimated total vitamin D intake in infants at the 75th percentile declined from 4,000 IU/day to a range of 724 to 1,343 IU/day between the two surveys.

Other Adverse Effects of Excess Vitamin D: Mortality, Chronic Disease, Falls and Fractures

The committee reviewed the evidence emerging from observational/association studies and a limited number of clinical trials related to vitamin D intake and a diverse set of health outcomes, ranging from breast cancer to falls and fractures. The purpose was not to determine that certain levels of intake definitively cause harm, but rather to decide whether the emerging data were sufficiently compelling to warrant caution relative to vitamin D intakes and associated serum 25OHD concentrations that may be less than those associated with the more widely known acute toxicity but still associated with adverse effects that may occur as a result of chronic intake. The potential adverse effects are considered in alphabetical order.

All-Cause Mortality

All-cause mortality data emerging from the examination of national survey data as well as observational studies suggest adverse effects at serum 25OHD levels much lower than those associated with the toxicity demonstrated by hypervitaminosis D. The AHRQ-Tufts analysis identified four cohort studies (Sambrook et al., 2004, 2006; Visser et al., 2006; Jia et al., 2007; Melamed et al., 2008) that focused on the relationship between serum 25OHD level and all-cause mortality. In general, these studies, as expected, indicated that low serum 25OHD levels akin to deficiency states (< 30 nmol/L) are associated with an increased risk of mortality. Further, as serum 25OHD levels increase—up to a point—mortality is lowered.⁴ However, some, but not all, of the studies have observed a troubling U-shaped (or perhaps more appropriately a reverse-J-shaped) relationship. For example, Jia et al. (2007) found a statistically significant trend between increasing serum 25OHD levels and lower odds ratios for all-cause mortality (p = 0.03); however, a U-shaped or reverse-J-shaped relationship between serum 25OHD level and mortality was observed, with the lowest mortality at serum 25OHD levels below 50 nmol/L (see Figure 6-1). Visser et al. (2006) showed a similar pattern, with reduced mortality associated with higher than deficiency levels, but increased mortality at the highest blood 25OHD levels (see Figure 6-2). Melamed et al. (2008), using data from the Third National Health and Nutrition Examination Survey (NHANES III), also suggested a U-shaped or reverse-J-shaped risk curve with increasing risk at about 75 nmol/L (see Figure 6-3). The similar patterns emerging in these studies are of concern and are suggestive of at least a reverse-J-shaped curve, if not precisely a U-shaped curve for risk relative to serum 25OHD levels and all-cause mortality. Of note, Sambrook et al. (2004, 2006) found no relationship between mortality and the log of serum 25OHD levels in a sample (n = 842) of frail, institutionalized persons, most over the age of 80 years. Also, the committee identified another cohort study not included in the AHRQ-Tufts report (Semba et al., 2009) that did not observe a U-shaped relationship, but the highest exposure category in this Italian cohort was approximately 64 nmol/L. In addition to these published observational studies, a preliminary analysis of NHANES III data limited to data on non-Hispanic blacks with follow-up as of December 31, 2006, also saw a U-shaped relationship, although the suggested increase in risk was seen at a lower serum 25OHD concentration of approximately 60 nmol/L.⁵

FIGURE 6-1

Hazard ratios of risk of death according to baseline serum 25OHD level (subjects with serum 25OHD levels 39.1–82.0 nmol/L are the referent category). NOTE: Model 1 is adjusted for age and gender; model 2 is adjusted for model 1 and taking five (more...)

FIGURE 6-2

Risk of death in elderly people according to baseline serum 25OHD level in the Longitudinal Aging Study (subjects with serum 25OHD levels of 50.0–74.9 nmol/L are the referent category). NOTE: Model 1 is adjusted for gender, age, and education; (more...)

FIGURE 6-3

Rate ratios of all-cause mortality by serum 25OHD level in NHANES III (subjects with serum 25OHD levels above 80.3 nmol/L are the referent category). NOTE: Model 1 is unadjusted; model 2 is adjusted for age, gender, race, and season; model 3 is adjusted (more...)

Turning to evidence from vitamin D supplementation trials, AHRQ-Tufts calculated an overall relative risk (RR) for all-cause mortality of 0.97 (95% CI: 0.92–1.02), with no evidence of between-study heterogeneity. The doses studied included 400 and 880 IU of supplemental vitamin D per day, and one trial that gave a supplement of 100,000 IU every 3 months, which is roughly equivalent to 1,100/day.

Because the trials did not evaluate particularly high doses and as observational studies are subject to confounding, one cannot interpret conclusively whether or not this U-shaped relationship is real or causal. However, the data are clearly suggestive of a U-shaped or reverse-J-shaped risk curve between serum 25OHD level and all-cause mortality; increases in risk are suggested at thresholds in the range of 75 to 120 nmol/L for the white population, with lower levels for the black population.

Selection of Indicator for the UL: Vitamin D

The best available data set for establishing a UL for vitamin D is associated with the onset of hypercalcemia and related toxicity. This indicator is selected as the basis for the UL for all age groups except infants, with the caveat that while it serves as a starting point, it is to be subject to adjustment for uncertainty. The adjustment is based on: (1) the recognition of the goal of public health protection, which suggests that avoiding hypervitaminosis D is, of course, desirable, but not necessarily sufficient; and (2) the emerging data concerning other adverse effects at intakes lower than those associated with acute toxicity and at serum 25OHD levels previously considered to be at the high end of physiological values. Taken as a whole, the body of evidence suggests that there is reason to proceed cautiously in assuming that higher levels of vitamin D intake below those expected to cause hypervitaminosis D are harmless, especially in the absence of data to demonstrate benefit at such intake levels. For infants, the long-standing measures related to retarded linear growth serve as the indicator for the UL.

The available data could not offer a BI or be used to estimate a dose– response relationship. The basis for the ULs is a NOAEL as described further below for specific life stage groups.

ULs for Infants 0 to 12 Months of Age

The work of Fomon et al. (1966) forms the starting point for these life stage group, as it did in the 1997 IOM report (IOM, 1997). Given the small sample size used in the study, the NOAEL for infants is based on the mean intake in this study rather than the high end of the range. The NOAEL was rounded to 1,800 IU/day from 1,775 IU/day. The British Paediatric Association (1956) data and data reported by Bransby et al. (1964) suggested that hypercalcemia could be present at intakes of 4,000 IU/day, but appeared to decline at intakes between 700 and 1,300 IU/day, lending some support to the NOAEL of 1,800 IU/day as reasonable. However, considerable uncertainty surrounds this estimate, and newer data have not emerged regarding vitamin D intake and hypercalcemia in infants. Stearns (1968) commented that Fomon et al. (1966) did not study the infants long enough, because the greatest differences in the Jeans and Stearns (1938) study appeared after 6 months. Also, a report from 1959 (Graham, 1959) suggested that a serum calcium level obtained from a study in Glasgow of infants with hypercalcemia ages 3 weeks to 11 months was associated with an estimated vitamin D intake of 1,320 IU/day. Overall, on balance, 1,800 IU/day is reasonable as a NOAEL and offers an appropriate starting point.

ULs for Adults 19 or More Years of Age

The indicator of hypercalcemia for vitamin D toxicity is the starting point for the UL for adults. This condition is at an extreme end of an adverse outcome continuum and it may be appropriate to consider instead as a starting point for other measures, such as hypercalciuria. However, interpretation of measures such as hypercalciuria as a predictor of adverse outcomes is unclear. Therefore, the best available option as an indicator is hypercalcemia. In this case, an intake value of 10,000 IU/day reflects a NOAEL. This NOAEL is initially adjusted for uncertainty to establish a UL of 4,000 IU/day, as described below.

Initially, it should be noted that evidence pertaining to the levels of 25OHD in serum that are associated with adverse effects is less well established than that associated with benefit, and the available literature suggests considerable variability. As shown above in Table 6-3, frank toxicity has been reported to have occurred within a wide range of serum 25OHD levels, from as low as 60 nmol/L (Byrne et al., 1995) to values above 1,500 nmol/L (Rizzoli et al., 1994; Pettifor et al., 1995; Vieth et al., 2002), although the majority of available reports of toxicity involve serum 25OHD values above 350 nmol/L. The variability in the toxicity data may mean that toxicity can be affected by numerous mitigating factors or perhaps may be a function of the diversity in the nature of the available case reports. Reports on maximal sun exposure also described previously (Barger-Lux and Heaney, 2002; Binkley et al., 2007) suggest that serum 25OHD levels under these circumstances generally remain below 125 to 150 nmol/L, although the populations studied are not diverse and generally include younger men. The emerging data related to all-cause mortality, chronic disease risk, and falls would appear to suggest that adverse events may occur with serum 25OHD levels of approximately 75 nmol/L or above (Visser et al., 2006; Ginde et al., 2009), but ranging up to approximately 125 nmol/L (Melamed et al., 2008). The vagaries of serum 25OHD measures in general, the sparse data available, and the uncertainty as to the nature of the adverse effects preclude strong conclusions. On the basis of available reports, the committee considered that serum 25OHD levels above approximately 125 to 150 nmol/L should be avoided. Given the conclusion derived in Chapter 5 that bone health benefit is achieved by 97.5 percent of the population at 50 nmol/L, there is a range of serum 25OHD levels between 50 and 150 nmol/L that remains undescribed. The adjustment to the starting point of 10,000 IU/day reflects first data concerning adverse effects related to all-cause mortality, falls and fractures, and CVD risk, which, taken as a total body of evidence, provide reason for caution, as described earlier. More specifically, the evidence from the studies that focused on all-cause mortality, chronic disease, falls and fractures suggested that serum 25OHD levels between 75 nmol/L and approximately 120 nmol/L were associated with the adverse effect. There is considerable uncertainty surrounding such values, and—using information on the serum levels achieved during maximal sun exposure and to avoid being unnecessarily restrictive given the uncertainties—the committee determined that for the purposes of the UL, concern would be for levels above approximately 125 to 150 nmol/L.

Further, there are emerging data concerning the possible differences in adverse event response for African Americans and perhaps other dark-skinned population groups. The cross–sectional study from Freedman et al. (2010) reported a positive association between serum 25OHD levels and calcified atherosclerotic plaque in the aorta and carotid arteries of African Americans, and preliminary reports from NHANES suggest that the risk for all-cause mortality among non-Hispanic blacks compared with whites occurs at a lower serum 25OHD level (60 vs. 75 nmol/L).⁶ These data are limited, are not necessarily consistent with other findings, and may eventually be explained by factors other than serum 25OHD levels, but they are concerning. Although data on race and ethnic differences are much too sparse to justify providing different ULs for different racial or ethnic groups, they can be incorporated as a source of uncertainty.

To determine an adjustment from the starting point of 10,000 IU/day as a NOAEL taking into account the uncertainties introduced by reports concerning all-cause mortality and other chronic disease outcomes as well as the possibility that blacks in the North American population may experience adverse effects at lower serum 25OHD concentrations than whites, the committee considered the work of Heaney et al. (2003). As suggested by the study, vitamin D intakes of 5,000 IU/day achieved serum 25OHD levels that range between 100 and 150 nmol/L, but do not surpass 150 nmol/L after 160 days of administration. Almost no other studies have assessed the safety of long-term maintenance of serum 25OHD levels in this range in relation to chronic disease risk and all-cause mortality, so the information about the increases in serum levels is useful for the purposes of establishing a UL. Given the uncertainties surrounding the data and the reliance on a single report, the UL is set 20 percent below the level identified by Heaney et al. (i.e., 5,000 IU), specifically at 4,000 IU/day.

This value is greater than that set in 1997 by the previous IOM committee. A UL of 4,000 IU/day is still, however, a reference value that reflects the interest in providing public health protection, especially when existing data do not support benefit above such intakes. Intake values in the range of 4,000 IU/day would not appear to cause serum 25OHD levels to exceed 125 to 150 nmol/L,⁷ a concentration which is at the high end of the range of serum levels associated with nadir risk of outcomes such as all-cause mortality.

ULs for Children and Adolescents 1 Through 18 Years of Age

No specific data are available for age groups other than adults and infants. In 1997 it was determined that increased rates of bone formation in toddlers, children, and adolescents suggested that the adult UL is appropriate for these age groups (IOM, 1997). The present committee chose to scale down the adult UL for younger children—to 2,500 IU/day for 1- to 3-year-olds and 3,000 IU/day for 4- to 8-year-olds—so as to be more consistent with concepts of graded tolerances with maturity. Although the simulated dose–response relationship between vitamin D intake and serum 25OHD level described in Chapter 5 is not affected by age, the data available did not include any children younger than 6 years old. There is no quantitative basis for such scaling, but it reflects a cautious and prudent approach given current biological understandings. Children and adolescents between 9 and 18 years of age have ULs that are the same as that for adults. All the UL values for children are slightly higher than the values provided in 1997 (IOM, 1997).

ULs for Pregnancy and Lactation

The available data do not indicate a basis for deriving a UL for pregnant and lactating women or adolescents that is different from those for their non-pregnant and non-lactating counterparts.