Objective
To summarize the results from three phase II studies that evaluated the effects of dexamethasone compared with anti-VEGF therapies (ranibizumab, bevacizumab) for the treatment of adult patients with diabetic macular edema (DME) who are pseudophakic.
Findings
Study Design
A total of three phase II trials studied the effects of dexamethasone compared with anti-VEGF drugs in adult patients with diabetic macular edema (DME) who are pseudophakic. All trials were multi-centre; however, only two were multinational and none recruited patients from Canada. One study was a noninferiority study that compared 700 mcg dexamethasone injections to 0.5 mg ranibizumab injections (RAN study), and another superiority study compared 700 mcg dexamethasone injections to 1.25 mg bevacizumab injections (BEVORDEX). In addition to 0.3 mg ranibizumab, the other superiority study compared 700 mcg dexamethasone injections to sham (COMB Study). Detailed study characteristics are provided in .
Patients in the RAN study were randomized to a 1:1 ratio using IVRS/IWRS and stratified by BCVA score at baseline (≥ 34 to ≤ 49 or ≥ 50 to ≤ 70). Both the study personnel who measured BCVA and evaluators at the reading center were masked to the study treatment assignment. Patients who experienced a decrease of ≥ 10 BCVA letters from baseline or > 300 µm CRT (> 320 µm on Spectralis OCT) as measured by Cirrus OCT could also receive deferred laser therapy. A sample size of 149 patients in each group was determined to provide 80% power to detect a two-letter difference between treatments groups from baseline in BCVA using a standard deviation of 9.21 based on the RESOLVE trial. A noninferiority margin of five letters was used based on half the historical maximum treatment effect of ranibizumab. A 2-sided alpha of 0.05 and 95% CI for the least squares mean difference between treatments were used to establish statistical significance of efficacy outcomes using the ITT population and based on an ANOVA model with treatment group and baseline BCVA as the main effects. Safety analyses were based on the safety population.
Patients in BEVORDEX Study were randomized to a 1:1 ratio using computer-generated pseudorandom numbers in permuted block of variable size. Only the study personnel who measured BCVA were masked to the study treatment assignment. Patients with two eligible eyes had one treated with the randomly generated treatment assignment, while the fellow eye received the other treatment. A sample size of 35 eyes per group and a difference of at least 30% between treatment groups were required to provide 80% using a two-sided alpha of 0.05. Logistic regression with BCVA as a covariate and generalized correlation equation methods were used to model the primary outcome given that a correlation was possible between the same eyes of a patient using the ITT population and LOCF approach.
Initially, only pseudophakic eyes were eligible for enrolment in the COMB Study; however, due to recruitment difficulties, the eligibility criteria were broadened to include phakic eyes. During a 12-week run-in phase, patients were required to receive three additional anti-VEGF injections of ranibizumab (at enrolment week 4 and week 8) in addition to the minimum of three prior injection required for enrolment. Patients were randomized to a 1:1 ratio using a permuted block design and stratified improvement in BCVA and CRT during the run-in phase. A sample size of 150 eyes was determined to provide 90% power to detect a five-letter difference between treatments groups from baseline in BCVA using a standard deviation of nine. Patients with two eligible eye for enrolment had one eye randomly assigned to each group. In COMB, patients and investigators assessing AEs were masked to treatment. Refractionists, visual acuity testers, and OCT technicians were masked at the 24-week primary end point. Investigators and study coordinators were not masked. Between-group differences (95% CI and 0.05 of the two-sided alpha) were performed using a linear mixed model with visual acuity at randomization and randomization stratification as fixed effects using the ITT population and multiple imputation to account for missing data. For patients that had both eyes involved in the study, a random effect was included to account for any fellow eye correlation.
Table 26Details of the Phase II Studies
View in own window
| RAN Study | BEVORDEX Study | COMB Study |
|---|
| Designs and Populations | Study design | Active-control, multi-centre, phase II, RCT |
|---|
| Multinational, single-masked, noninferiority trial | Single-masked, superiority trial | Masked (patient and investigator), superiority trial |
| Locations | 60 sites in 12 countries Israel. South Africa, US, Western Europe | 4 sites in Australia | 40 sites in the US |
|---|
| Patients (N) | 363 (one eye per patient) | 88 eyes from 61 patients | 236 eyes from 203 patients |
|---|
| Eligibility | Adult patients with DME involving the center of the macula with mean CRT as measured by OCT ≥ 300 µm with Spectralis (Heidelberg) or ≥ 275 µm with Cirrus (Zeiss) at screening. Patients with BCVA > 34 and < 70 ETDRS letters (Snellen equivalent between 20/200 and 20/40). Patients with glycated hemoglobin > 12 %, IOP > 22 mm Hg at, glaucoma, a history of laser treatment within 3 months prior to screening, use of anti-VEGF treatment within 3 months prior to screening, use of intravitreal triamcinolone acetonide within 6 months prior to screening, and a history of vitrectomy were excluded. | Adult patients with DME involving the central fovea at least 3 months following at least 1 session of laser treatment. Patients for whom the investigator believed that laser treatment would be unhelpful, with Snellen equivalent between 20/400 and 20/40. Patients with uncontrolled glaucoma or glaucoma controlled with more than 1 medication, loss of vision because of other causes, intercurrent severe systemic disease, or any condition affecting follow-up or documentation were excluded. | Adult patients with DME with CRT as measured by OCT ≥ 290 in women, ≥ 305 in men with Cirrus (Zeiss) or ≥ 305 in women, ≥ 320 in men with Spectralis (Heidelberg). Patients with BCVA score of 78 to 24 (Snellen equivalent between 20/320 and 20/32). Patient received treatment with at least 3 anti-VEGF injections for DME (aflibercept, bevacizumab, or ranibizumab) within the previous 20 weeks. Patients with glaucoma loss of vision because of other causes, or any condition affecting follow-up or documentation were excluded. |
|---|
| Drugs | Intervention | Dexamethasone 700 mcg intravitreal injection at baseline month 5 and month 10. | Dexamethasone 700 mcg intravitreal injection at baseline. Patients were eligible for retreatment if retinal thickness as measured by OCT was ≥ 300 µm or visual acuity was 79 letters or better (Snellen equivalent 20/25). Study treatment procedure was not to be performed more often than every 16 weeks. | Dexamethasone 700 mcg intravitreal injection was to be administered no more than 8 days following background ranibizumab 0.3 mg administered at baseline. At weeks 4 and 8 only ranibizumab injections were permitted. At weeks 12 through 20, patients were eligible for retreatment with dexamethasone in combination with ranibizumab if the visual acuity letter score was less than 84 (Snellen equivalent of 20/25 or worse) or if the CRT as measured by OCT was at or above ≥ 290 in women, ≥ 305 in men with Cirrus (Zeiss) or ≥ 305 in women, ≥ 320 in men with Spectralis (Heidelberg) A maximum of 2 injections of dexamethasone were given in each eye. |
|---|
| Comparators | Ranibizumab 0.5 mg administered at baseline and monthly thereafter until the patients’ visual acuity was stable for three consecutive monthly assessments. If no improvements in visual acuity were observed, monthly injections could be suspended until a decrease in visual acuity was observed (i.e., reinitiation of monthly injections after decrease in visual acuity). No treatment could be administered at month 12 (i.e., last dose was at month 11). | Bevacizumab 1.25 mg injection at baseline. Patients were eligible for retreatment if retinal thickness as measured by OCT was ≥ 300 µm or visual acuity was 79 letters or better (Snellen equivalent 20/25) Study treatment procedure was not to be performed more often than every 4 weeks | Sham procedure using a needleless applicator pressed against the conjunctiva was to be administered no more than 8 days following background ranibizumab 0.3 mg administered at baseline. At weeks 4 and 8 only ranibizumab injections were permitted. At weeks 12 through 20, patients were eligible for retreatment with sham in combination with ranibizumab if the visual acuity letter score was less than 84 (Snellen equivalent of 20/25 or worse) or if the CRT as measured by OCT was at or above ≥ 290 in women, ≥ 305 in men with Cirrus (Zeiss) or ≥ 305 in women, ≥ 320 in men with Spectralis (Heidelberg). A maximum of 2 injections of sham treatment were given in each eye. |
|---|
| Duration | Phase |
|---|
| Pre-treatment | 2 week screening phase | None | 12 week run-in phase |
|---|
| Treatment | 12 months | Every 4 weeks for up to 50 weeks | Every 4 weeks for 24 weeks |
|---|
| Outcomes | Primary end point | Change from randomization at each visit over 12 months in the mean average BCVA as measured by ETDRS. | Proportion of eyes with BCVA improvement of 10 or more letters as measured by ETDRS at week 48 | Change from randomization to week 24 in the mean visual acuity letter score as measured by ETDRS |
|---|
| Other end points |
|
| |
|---|
| Notes | Publications | Callanan et al. 201695 | Gillies et al. 201496 Fraser-Bell et al. 201697 | Maturi et al. 201898 |
|---|
BCVA = best-corrected visual acuity; DME = diabetic macular edema; ETDRS = Early Treatment Diabetic Retinopathy Study; IOP = intraocular pressure; OCT = optical coherence tomography; RCT = randomized controlled trial; VEGF = vascular endothelial growth factor.
Source: Callanan et al. 2016,95 Gillies et al. 2014,96 Fraser-Bell et al. 2016,97 Maturi et al. 2017.98
Methods
Patient Disposition
Disposition data were only available for the full DME populations included in the trials. No disposition data were available for the subgroup of patients that were pseudophakic. Most patients in all trials (between 90% and 100%) completed the studies. Reasons for discontinuations were not transparently reported. Of those that were reported, the most common reasons were loss to follow-up (between 7% and 10%), personal reasons (3%), and non-ocular AEs (3%). Detailed disposition is provided in .
Table 27Patient Disposition (Full DME population)
View in own window
| RAN Studya | BEVORDEXc | COMB Studyc |
|---|
| DEX | RAN | DEX | BEV | DEX+RAN | SHAM+RAN |
|---|
| Full trial populationb |
|---|
| Randomized, N (%) | 181 | 182 | 46 | 42 | 65 | 64 |
|---|
| Discontinued study, N (%) | 16 (88) | 16 (88) | 3 (7) | 4 (10) | 2 (3) | 0 |
|---|
| Completed study, N (%) | 165 (91) | 166 (91) | 43 (93) | 38 (90) | 63 (97) | 64 (100) |
|---|
| Reasons for discontinuation, N (%) | | | | | | |
|---|
| Ocular AEs | 4 (2) | 0 | NA | NA | NA | NA |
|---|
| Non-ocular AEs | 6 (3) | 5 (3) | NA | NA | NA | NA |
|---|
| Lack of efficacy | 1 (<1) | 1 (<1) | NA | NA | NA | NA |
|---|
| Pregnancy | 0 | 1 (<1) | NA | NA | NA | NA |
|---|
| Lost to follow-up | 3 (2) | 1 (<1) | 3 (7) | 4 (10) | NA | NA |
|---|
| Personal reasons | 0 | 5 (3) | NA | NA | NA | NA |
|---|
| Protocol violations | 0 | 0 | NA | NA | NA | NA |
|---|
| Other | 2 (1) | 3 (2) | NA | NA | NA | NA |
|---|
| Pseudophakic subgroup | 54 (30) | 62 (34) | 16 (35) | 10 (24) | 26 (40) | 32 (50) |
|---|
AE = adverse event; BEV = bevacizumab; DEX = dexamethasone; NA = not available; RAN = ranibizumab.
- a
N represents the number of patients.
- b
Patients screened in the overall DME population.
- c
N represents the number of eyes.
Source: Callanan et al. 2016,95 Gillies et al. 2014,96 Fraser-Bell et al. 2016,97 Maturi et al. 2017.98
Baseline Characteristics
Baseline characteristics were only available for the full DME populations included in the trials. No baseline characteristics were available for the subgroup of patients that were pseudophakic. Overall, patients were between 61.4 and 66 years of age (SD 9.0 to 10.5) and were mostly male (52% to 65%) with the exception of the SHAM+RAN group in the COMB Study (44% male). Mean DME duration was only available for the RAN study in which patients had DME for approximately 30 months. The majority of patients in the RAN study had diabetes for more than five years (87%). Overall, the minority of patients included in the trials were pseudophakic, 24% in BEVORDEX, and up to 50% in one arm of the COMB Study. The minority of patients were previously treated with laser therapy (between 25% and 48%) in the COMB STUDY and the RAN study, whereas all patients were previously treated with laser therapy in BEVORDEX. The minority of patients were previously treated with anti-VEGF drugs in the RAN study (22%), whereas all patients were treated with prior anti-VEGF drugs in the COMB Study. BEVORDEX did not report on prior use of anti-VEGF drugs. In general, A1C was similar between all trials (between 7.1% and 7.8%) as was IOP (between 14.5 and 16 mm Hg). Both BCVA and CRT varied across trials ranging between 55.5 letters to 63 letters and 375 µm to 503 µm, respectively. Detailed baseline characteristics are provided in .
Table 28Summary of Baseline Characteristics (Full DME population)
View in own window
| Characteristics | RAN Studya | BEVORDEXc | COMB Studyc |
|---|
DEX N = 181 | RAN N = 182 | DEX N = 46 | BEV N =42 | DEX+RAN N = 65 | SHAM+RAN N = 64 |
|---|
| Age | | | | | | |
| Mean years (SD) | 63.4 (9.39) | 63.7 (10.05) | 61.4 (9.0) | 62.2 (10.5) | NR | NR |
| Median years (IQR) | NR | NR | NR | NR | 64 (59 to 69) | 66 (59 to 71) |
| Gender, n (%) |
|---|
| Male | 112 (61.9) | 116 (63.7) | 30 (65) | 26 (62) | 34 (52) | 28 (44) |
| Female | 69 (38.1) | 66 (36.3) | 16 (35) | 16 (38) | 31 (48) | 36 (56) |
| DME duration |
|---|
| Mean months (SD) | 36.3 (58.1) | 29.7 (33.3) | NR | NR | NR | NR |
| Diabetes duration, years |
|---|
| Mean (SD) | NR | NR | 16.7 (10.3) | 16.7 (10.7) | NR | NR |
| Median (IQR) | NR | NR | NR | NR | 15 (10 to 21) | 19 (10 to 26) |
| ≤ 6 months | 1 (0.6) | 1 (0.5) | NR | NR | NR | NR |
| > 6 months to1 year | 5 (2.8) | 7 (3.8) | NR | NR | NR | NR |
| >1 to5 years | 14 (7.7) | 15 (8.2) | NR | NR | NR | NR |
| > 5 years | 160 (88.4) | 158 (86.8) | NR | NR | NR | NR |
| Lens status of the study eye, pseudophakic, n (%) | 54 (29.8) | 62 (34.1) | 16 (35) | 10 (24) | 26 (40) | 32 (50) |
| Prior treatment, n (%) |
|---|
| Laser | 53 (29.3) | 47 (25.8) | 46 (100) | 42 (100) | 31 (48) | 31 (48) |
| Anti-VEGF | 40 (22.1) | 39 (21.4) | NR | NR | 65 (100) | 64 (100) |
| Intravitreal steroid injection | NR | NR | NR | NR | 9 (14) | 10 (16) |
| A1C |
|---|
| Mean, % (SD) | 7.7 (1.4) | 7.5 (1.3) | 7.7 (2.5) | 7.8 (2.1) | NR | NR |
| Median (IQR) | NR | NR | NR | NR | 7.1 (6.4 to 8.3) | 7.4 (6.6 to 8.2) |
| BCVA in the study eye at baseline |
|---|
| Mean (SD) | 60.2 (9.74) | 60.4 (9.34) | 55.5 (12.5) | 56.3 (11.9) | 63 (12) | 63 (13) |
| IOP in the study eye, mm Hg |
|---|
| Mean (SD) | 14.9 (2.9) | 14.9 (2.7) | 14.8 (3.0) | 14.5 (2.4) | NR | NR |
| Median (IQR) | NR | NR | NR | NR | 15 (13 to 17) | 16 (14 to 18) |
| OCT retinal thickness at center subfield, microns |
|---|
| Mean (SD) | 465 (136) | 471 (140) | 474.3 (95.9) | 503 (140.9) | 375 (97) | 396 (122) |
A1C = glycated hemoglobin; BCVA = best-corrected visual acuity; BEV = bevacizumab; DEX = dexamethasone; DME = diabetic macular edema; IOP = intraocular pressure; IQR = interquartile range; OCT = optical coherence tomography; RAN = ranibizumab; SD = standard deviation; VEGF = vascular endothelial growth factor.
- a
N represents the number of patients.
- b
Patients screened in the overall DME population.
- c
N represents the number of eyes.
Source: Callanan et al. 2016,95, Gillies et al. 2014,96 Fraser-Bell et al. 2016,97 Maturi et al. 2017.98
Results
Efficacy
The adjusted the least squares (LS) mean difference for the BCVA average change from baseline was not provided for the pseudophakic group in BEVORDEX or the RAN study. The adjusted LS mean change from baseline in the average BCVA was 4.6 letters in the DEX group and 6.6 letters in the RAN group in the RAN study, whereas it was 10.4.etters in the DEX group and 7.7 letters in the BEV group at year 1 and 8.9 letters (95% CI, 2.0 to 13.4) in the DEX group and 7.7 letters (95% CI, 3.03 to 14.8) in the BEV group at year 2 in BEVORDEX.
The adjusted LS mean difference for the BCVA average change from baseline between treatment groups was 3.1 letters (95% CI, −2.1 to 8.3) in the COMB Study. The adjusted LS mean difference between treatment groups for the CRT as measured by OCT average change from baseline was −78 µm (95% CI, −131 to −25).
Table 29Visual Acuity Efficacy Outcomes (Pseudophakic Subgroup)
View in own window
| Outcome | RAN Studya | BEVORDEXc | COMB Studyc |
|---|
| Year 1 | Year 2 | |
|---|
DEX N = 54 | RAN N = 62 | DEX N = 16 | BEV N =10 | DEX N = 16 | BEV N =10 | DEX+RAN N = 26 | SHAM+RAN N = 32 |
|---|
| BCVA average change from baselinea |
|---|
| Baseline, n (%) | NR | NR | NR | NR | NR | NR | NR | NR |
| Adjusted LS mean change from baseline, letters (SD) | 4.6 (NR) | 6.6 (NR) | 10.4 (NR) | 7.7 (NR) | 8.9 (95% CI, 2.0, 13.4) | 7.7 (95% CI, 3.03, 14.8) | 5.1 (9.7) | 2.0 (7.6) |
| Adjusted LS MD versus comparator (95% CI) | NR | NR P = 0.47 | NR P = 0.77 | 3.1 (−2.1 to 8.3) P = NR |
| CRT as measured by OCT average change from baselinea |
|---|
| Baseline, n (%) | NR | NR | NR | NR | NR | NR | NR | NR |
| Adjusted LS mean change from baseline, microns (SD) | NR | NR | NR | NR | NR | NR | −111 (86) | −49 (96) |
| Adjusted LS MD versus sham (95% CI) | NR | NR | NR | −78 (−131 to −25) P = NR |
BCVA = best-corrected visual acuity; BEV = bevacizumab; CI = confidence interval; CRT = central retinal thickness; DEX = dexamethasone; LS = least squares; MD = mean difference; OCT = optical coherence tomography; RAN = ranibizumab; SD = standard deviation.
- a
N represents the number of patients
- b
Patients screened in the overall DME population
- c
N represents the number of eyes
No adjustments for multiple statistical tests were made for any outcomes in the subgroup of adult patients with DME who were pseudophakic
Source: Callanan et al. 201695 Gillies et al. 201496 Fraser-Bell et al. 201697 Maturi et al. 201798
Safety
No safety data were provided for the pseudophakic subgroups in any of the phase II trials.
Limitations
There are several limitations to the phase II trials included in this section (RAN Study, BEVORDEX Study, and COMB Study). The phase II trials were active-controlled RCTs that used appropriate methods to randomize patients. However, these trials were originally designed to evaluate the effects of dexamethasone in the general DME population. This CDR report is based on the results of a subgroup (i.e., adults with DME who are pseudophakic) which only consisted of approximately 24% to 50% of overall enrolled patients. Subgroup analyses were performed in the pseudophakic populations; however, randomization was not stratified for this subgroup. Inadequate randomization introduces biases through the presence of confounders (known and unknown). Given that no baseline characteristics or disposition data were provided in any of the trials for the pseudophakic subgroup, it is difficult to assess adequacy of randomization and therefore appropriately interpret the validity of the results.
All phase II trials included in this section based their sample size calculations on the general DME population and not the subgroup. Given that the subgroup of patients who are pseudophakic would consist of a smaller subset of this population (24% to 50%) the subgroups are also likely underpowered to detect a statistically significant difference.
Therefore, the results of these phase II trials are susceptible to false negatives (i.e., finding no difference between treatments when a difference truly exists).
Although, patients and investigators assessing AEs were masked to treatment, as well as refractionists, visual acuity testers, and OCT technicians, some investigators and study coordinators were not masked. Both the RAN study and BEVORDEX were single-masked trials in which patients were not blinded to treatment allocation. Furthermore, the adverse event profile associated with intravitreal steroids (i.e., IOP) is well known, therefore some accidental unblinding may have occurred.12,19 Given that prior intravitreal steroid experience was not an exclusion criterion in any of the trials, some patients with prior experience may have surmised that the allocated treatment was dexamethasone. However, considering that the primary end point of the MEAD trials is relatively objective, the potential for bias is of lesser concern.
The Health Canada–approved indication is for the treatment of patients with DME who are pseudophakic. Therefore, the focus of this CDR review is based on a subset of the overall DME population. In addition, subgroups of interest included pseudophakic patients with DME who are either unsuitable for anti-VEGF therapy or have had inadequate response to prior anti-VEGF therapy. However, only approximately 20% of patients in the general DME population included in the trial in the RAN study were previously treated with anti-VEGF drugs; whereas, all patients in the general DME population included in the trial were previously treated with anti-VEGF drugs in the COMB Study. Prior use of anti-VEGF drugs were not provided in BEVORDEX. It is important to note that prior use of anti-VEGF drugs in the pseudophakic subgroups were not reported in any of the trials. Furthermore, it remains unclear if these patients responded to these treatments and were truly anti-VEGF refractory. It is also unclear if there were any patients included in the phase II trials that were considered unsuitable for anti-VEGF therapy. According to the clinical expert consulted for this review, the date of conduct of the trials (between February 2005 to June 2012) was prior to the adoption of anti-VEGF therapies and may therefore have influenced the number of patients having access to anti-VEGF therapy. It is therefore unclear if the results of the phase II trials can be generalized to patients who are unsuitable for anti-VEGF therapy or have had an inadequate response to anti-VEGF therapy.
All phase II trials were multi-centre; however they did not include any sites from Canada. Based on the characteristics of the MEAD studies, the phase II trials appear to have recruited patients with characteristics similar to those of the overall DME population in Canada, however, given that the baseline characteristics of the pseudophakic subgroup, it remains unclear whether the patient population included in these trials is truly representative of patients with DME who are pseudophakic in Canadian clinical practice.
With respect to the study duration, the FDA suggested that 36 months is considered short term. The FDA recommends that the treatment effect be demonstrated at a time point of at least 36 month or later for the indication of DME given that earlier treatment success is not necessarily a good indicator of a later success.52 Therefore, it is unclear if the results of the phase II trials would be representative of the long-term treatment effect given that they are considerably shorter in duration (i.e., between approximately six months and one year in duration).
Summary
The three phase II studies (RAN study, BEVORDEX and the COMB Study) summarized here were designed to evaluate the effects of dexamethasone in the general DME population, not the pseudophakic subgroup of patients which is of interest for this review. Some pseudophakic subgroup results were reported, however the lack of stratification at randomization based on this factor, as well as the absence of reporting on baseline characteristics for the pseudophakic population make it difficult to truly assess the comparative efficacy and harms between dexamethasone and anti-VEGF drugs (i.e., bevacizumab and ranibizumab). These studies were also likely underpowered to detect differences between treatments in the pseudophakic subgroup, there was no control for multiple statistical testing, study durations were short, and no Canadian sites were included. These limitations make it difficult to appropriately interpret the comparative efficacy and safety of dexamethasone versus other drugs used for the treatment of patients with DME who are pseudophakic.
