The Effect of Ruboxistaurin on Visual Loss in Patients With Moderately Severe to Very Severe Nonproliferative Diabetic Retinopathy : Initial Results of the Protein Kinase C β Inhibitor Diabetic Retinopathy Study (PKC-DRS) Multicenter Randomized Clinical Trial

The trial enrolled 252 participants (170 men and 82 women) between the ages of 20 and 84 years with type 1 or type 2 diabetes and HbA_1c (A1C) values from 5.1 to 13% inclusive (Table 1). Participants were excluded if they had 1) a history of significant heart disease (including unstable angina, acute coronary syndrome, myocardial infarction, or history of coronary revascularization procedure) within 6 months before visit 1; 2) significant hepatic disease (defined as aspartate transaminase, alkaline phosphatase, or total bilirubin more than twice the upper limit of normal), renal disease (defined as serum creatinine >2.5 mg/dl, history of renal transplant, or undergoing dialysis at screening), or anemia (defined as hemoglobin <10 g/dl); 3) a systolic blood pressure ≥180 mmHg or a diastolic blood pressure ≥105 mmHg; or 4) undergone a major surgery within the previous 3 months.

Participants had to meet all the following ocular entry criteria in at least one eye: 1) an Early Treatment Diabetic Retinopathy Study (ETDRS) retinopathy severity level between 47B and 53E inclusive (moderately severe to very severe nonproliferative diabetic retinopathy [NPDR]), 2) best-corrected visual acuity of >44 letters using ETDRS visual acuity protocol (Snellen equivalent of 20/125 or better), 3) no history of scatter (panretinal) photocoagulation for DR, and 4) no evidence of glaucoma. Eyes meeting these criteria at enrollment and fellow eyes that had less than proliferative retinopathy without a history of PRP were designated “study eyes.” Therefore, some patients had two study eyes and other patients had only one study eye.

The PKC-DRS was a multicenter, double-masked, placebo-controlled study in which subjects were randomized to one of four treatment groups: 1) placebo (n = 61), 2) RBX (LY333531; Eli Lilly and Co., Indianapolis, IN) mesylate 8 mg/day (n = 60), 3) RBX 16 mg/day (n = 64), or 4) RBX 32 mg/day (n = 67). A total of 617 patients were screened to obtain the 252 patients randomized (Fig. 1). Randomization was performed at the site level and stratified based upon diabetes type with a block size of 8. The sample size of the study was estimated based on an anticipated 2-year primary outcome event rate in the placebo group of 0.64 obtained by interpretation of the ETDRS results (21) and was estimated to have 80% power to detect a clinically significant 50% reduction in event rate. Failure to meet ocular criteria accounted for 317 (87%) of the 365 patients not entered into the study.

Eligibility was based on two screening visits that occurred within 6 weeks of the randomization visit. Following randomization, visits occurred at 1 and 3 months, every 3 months through 24 months, then every 6 months thereafter. All patients were followed until the last patient randomized completed 36 months of follow-up.

Decisions regarding application of photocoagulation resided with individual study investigators and patients, but study policy discouraged initiation of scatter (panretinal) photocoagulation in study eyes before the development of ETDRS level ≥65 proliferative retinopathy. Focal/grid photocoagulation for DME was initiated at the investigators’ discretion.

All serious and nonserious adverse events were analyzed regardless of the investigators’ assessments of causality. Adverse events that resulted in death, hospitalization, cancer, permanent disability, or threat to life were classified as serious. The Medical Dictionary for Regulatory Activities (MedDRA) was used to categorize reported adverse events. Laboratory evaluations were performed at each visit. Physical examinations and electrocardiograms (ECGs) were performed at screening and every 6 months thereafter. ECGs were also performed 1 month postrandomization. To better assess safety data and account for less-common adverse events, safety data from the PKC-DRS trial were pooled with data from the PKC-Diabetic Macular Edema Study (PKC-DMES), a trial of similar duration with a similar patient population. Both studies were conducted in parallel at the same investigative sites. With the exception of the ophthalmologic entry criteria (PKC-DMES study eyes had macular edema >300 μm from the center, ETDRS retinopathy severity level between 20 and 47A inclusive, best-corrected ETDRS visual acuity of >74 letters, and no history of scatter or focal/grid photocoagulation for DR), the inclusion and exclusion criteria were similar for each study. At baseline, no significant differences in the patient characteristics were evident between the two study groups, and there was no clearly evident disparity in reported adverse events. Therefore, in this report, pooled safety data from these 937 patients followed for periods ranging from 30 to 52 months are presented. Subjects from both studies were questioned regarding adverse effects of treatment at each visit. All categories of adverse events for which the frequency was at least 1% and was significantly different (P < 0.05) across treatment groups are reported.

Ophthalmologic examination was performed at screening and every visit during the treatment period. Ophthalmologic examinations included a best-corrected visual acuity assessment at each visit using the ETDRS protocol (22), intraocular pressure, and Age-Related Eye Disease Study (AREDS) (AREDS Manual of Operations; EMMES, Rockville, MD; available from http://spitfire.EMMES.com/study/AREDS/mop.htm) clinical lens grading every 3 months. DR and DME were assessed by masked grading of ETDRS seven-field stereoscopic color fundus photographs (23,24) performed at the University of Wisconsin Fundus Photograph Reading Center. Photograph sets were obtained at screening, 3 months, 6 months, every 3 months through 24 months, and every 6 months thereafter and were graded using the ETDRS protocol, modified to include estimates of area of retinal thickening in each subfield of the ETDRS grid and of proximity of retinal thickening to the center of the macula. Gradings of right and left eyes from various visits were performed independently of each other.

The primary end point for this study was progression of DR defined as a greater than or equal to three-step worsening in the ETDRS retinopathy person severity scale (patients with two study eyes), a greater than or equal to two-step worsening in the ETDRS retinopathy eye severity scale (patients with only one study eye), or a scatter (panretinal) photocoagulation for DR in a study eye. The person scale is an adaptation of the eye scale for individuals who have two simultaneously eligible eyes (23,25). Both scales were used to allow assessment of patients with either one or two eligible eyes. Patients with both eyes of less than proliferative retinopathy without previous PRP had both eyes classified as study eyes, whereas patients with only one such eye were deemed to have only one study eye. In nearly all cases, retinopathy progression meeting the study end point would have resulted in patients developing PDR from their baseline NPDR state.

The secondary study outcome of MVL was defined as a decrease from baseline in ETDRS visual acuity score of 15 or more letters, corresponding to a decrease of three or more lines on the ETDRS chart (doubling or more of the visual angle). Sustained MVL (SMVL) was defined as a decrease from baseline of 15 or more letters observed at each of two consecutive visits 6 or more months apart. For patients with two study eyes, MVL occurring in either eye counted as an event for the patient. Additional subgroup analyses were performed to determine the effect of DME or DR severity on SMVL. Analyses of change in visual acuity included all study eyes, and no adjustments were made for correlation between eyes.

Baseline patient characteristics were compared among treatment groups by categorical tests or ANOVA. All analyses were done using the intent to treat principle. P values from pairwise comparisons between RBX (Eli Lilly) doses and placebo were not adjusted for multiple comparisons. The unadjusted effect of treatment on the occurrence of events (progression of DR, MVL, and SMVL) was analyzed by Kaplan-Meier time-to-event curves, Cox proportional hazard models adjusted for diabetes type, and observed proportions of patients or eyes with events. The impact of treatment effect by baseline DR and DME severity levels was examined by performing stratified analysis of proportions of patients with SMVL. Treatment effect (32 mg RBX vs. placebo) on MVL was assessed using a Cox proportional hazard model adjusting for important baseline covariates identified among potential effect modifiers including ACE inhibitor (ACEI)/angiotensin receptor blocker (ARB) use, age, alcohol use, antibiotic use, antihypertensive use, BMI, urine protein level, DME level, DR level, duration of diabetes, elevated LDL cholesterol or triglycerides, A1C level, insulin use, mean arterial blood pressure, nitrate use, race, sex, tobacco use, type of diabetes, and visual acuity score. Covariates that were either clinically relevant or statistically significant were included in the final Cox proportional hazard (adjusted) model.

Adverse event data were analyzed using categorical (χ² or Fishers exact) tests. Continuous safety parameters (e.g., laboratory data, intraocular pressure) were analyzed using ANOVA.

Baseline demographic characteristics by treatment group are summarized in Table 1. Presence of sites enrolling fewer than eight patients led to some numerical imbalances among treatment groups due to the randomization block size. Participants were predominantly Caucasian with type 2 diabetes, mean age 56 ± 12 years (range 20–84), mean A1C 8.8 ± 1.4% (5.1–13%), and mean duration of diabetes 16 ± 8 years (1–38). BMI was significantly different among the treatment groups (P = 0.046 overall). However, placebo and 32 mg/day RBX groups had similar BMI. There were no other statistically or clinically significant differences observed at baseline among treatment groups for demographic characteristics, laboratory values, or concomitant medications. Notably, A1C was comparable across all groups at baseline, and the change from baseline to end point was not statistically significantly different among treatment groups (range −0.2 to −0.4%).

There were no statistically or clinically significant differences in baseline ophthalmic characteristics among treatment groups, although severe nonproliferative retinopathy (level 53) tended to be less frequent and DME within 500 μm of the center of the macula or involving it tended to be more frequent in the 32 mg/day group than in the placebo group (Table 2). Mean best-corrected visual acuity score was 80 letters in both placebo and 32 mg/day groups.

The primary end point for this study was progression of DR by either three or more steps in the ETDRS retinopathy person severity scale (patients with two study eyes), two or more steps in the ETDRS retinopathy eye severity scale (patients with only one study eye), or application of scatter (panretinal) photocoagulation for DR in a study eye. As shown in Fig. 2, there were no statistically significant differences among treatment groups in the time to progression of DR or in the cumulative percentage of patients who reached this end point (log-rank test of difference in survival curves P = 0.535). Similarly, there were no statistically significant differences among treatment groups when the components of the composite primary end point were considered individually (e.g., only photographically documented progression of retinopathy, P = 0.616).

As demonstrated in Fig. 3, the rate of occurrence of MVL was lower in the 32-mg RBX group (log-rank P = 0.038). Similarly, the percentage of patients with MVL at each visit (except 42 months) was lower in the 32-mg/day RBX group. The highest RBX dose (32 mg/day) showed the strongest suggestion of a beneficial effect, followed by the 8-mg/day dose.

Similar findings were observed for SMVL (Fig. 4A) (log-rank P = 0.226). As shown in Fig. 4B, the percentage of patients with SMVL at most visits was lower in the 32-mg/day RBX group than in the placebo group (2 vs. 15% ([P = 0.031], 4 vs. 20% [P = 0.027], and 11 vs. 20% [P = 0.267] at 12, 24, and 30 months, respectively).

SMVL tended to occur more frequently in eyes with more severe retinopathy and in those with definite DME at baseline (Fig. 5). In these subgroups, it occurred less frequently in the 32-mg/day RBX group than in the placebo group (10 vs. 25% [P = 0.017] for definite DME and 8 vs. 27% [P = 0.024] for level 53).

There were no statistically significant differences observed among treatment groups for the development of ETDRS level 65 or greater PDR, development of PDR with high-risk characteristics (ETDRS levels 71 and 75), application of focal photocoagulation, development of DME involving or threatening the center of the macula, doubling or halving of the area of DME, or change in visual function assessed using the Visual Function Questionnaire (VFQ)-25 (26).

The final Cox proportional hazard model (adjusted) included these significant or clinically relevant baseline covariates: ACEI/ARB use, age, antibiotic use, urine protein level, DME level, DR level, elevated LDL cholesterol or triglycerides, nitrate use, race, sex, tobacco use, type of diabetes, and visual acuity score. The strongest predictors are shown in Fig. 6. Treatment with 32 mg/day RBX reduced the risk of MVL compared with placebo (hazard ratio 0.37 [95% CI 0.17–0.80], P = 0.012). Male sex and ACEI/ARB use at baseline were also associated with a reduced risk of MVL, while tobacco use, age, and severe DME at baseline were associated with increased risk of MVL

Patient discontinuations were similar among treatment groups in the PKC-DRS trial (Fig. 1). There were no differences among treatment groups in the rate of discontinuation or the reasons for discontinuation. This balance reduces potential confounding induced by patient discontinuation. From the pooled PKC-DRS and PKC-DMES populations (n = 937), there were 14 treatment-emergent adverse events with an incidence >1% and a statistically significant difference in incidence among treatment groups (P < 0.05) (Table 3). There was no consistent pattern of adverse events to suggest a causal relationship between RBX and any spontaneously reported adverse event. There were no serious adverse events reported more frequently in RBX treatment groups. Finally, there were no statistically significant, clinically relevant differences among treatment groups for baseline to end point change in laboratory values, vital signs, BMI, cataracts (as measured by AREDS lens grading), or intraocular pressure.

The PKC-DRS was designed to test the primary hypothesis that RBX, a β-isoform–selective PKC inhibitor, would delay either the progression of DR on the ETDRS retinopathy severity scale or the application of laser photocoagulation in patients with moderately severe to very severe NPDR, 20/125 or better visual acuity, and no prior scatter (panretinal) photocoagulation for DR. We found no statistically significant effect of RBX in this regard among any of three treatment doses after a minimum 3 years of follow-up (Fig. 2).

There are cellular and animal data suggesting that the β-isoform of PKC may be involved in mediating diabetes-induced retinopathy (18), retinal vascular permeability (11,27), and retinal neovascularization (13,28). The apparent lack of efficacy of RBX in preventing progression of retinopathy to the proliferative stage could occur for several reasons. Study patients had moderately severe to very severe NPDR at baseline. It has been well established that PKC β is activated very early in diabetes, well before clinically apparent retinopathy (29,30). Thus, in our study participants, significant biochemical and pathologic retinal changes that are no longer amenable to PKC β inhibition may have already occurred before enrollment. Alternatively, RBX may not be potent enough to overcome these effects. It is also known that the majority of the neovascular response in the retina is mediated by VEGF (31–33). Although activation of PKC β is involved in mediating VEGF-induced intracellular signaling (11,28,34,35), RBX is not primarily a VEGF inhibitor, and in cellular and animal models, its antiproliferative effect is weaker than its antipermeability activity (11,13,28). Over time, different growth factors or other molecules, including different isoforms of PKC, could be compensating any potential beneficial effects of PKC β inhibition. In addition, it is possible that PKC β activation is not critical for the progression of DR to the proliferative stage in humans.

However, there did appear to be a beneficial effect of RBX on the secondary study outcomes of MVL and SMVL (Figs. 3–5). MVL, defined as a loss of 15 letters or more of best-corrected visual acuity on the ETDRS visual acuity chart, and SMVL, defined as MVL at each of two consecutive visits 6 or more months apart, are clinically meaningful end points (36,37) equivalent to a doubling of the visual angle (e.g., deterioration from 20/20 to 20/40). Although reduction in SMVL by RBX treatment did not reach statistical significance, the relative risk reductions for SMVL and MVL were in the same direction and generally of similar magnitude (0.62 vs. 0.37, respectively). Since SMVL occurs less frequently than MVL, there is less chance of observing a statistical difference in SMVL in this study. However, reduction of SMVL by RBX in eyes with DME at baseline did achieve statistical significance (P = 0.017, Fig. 5). Thus, these findings are consistent with a beneficial effect of RBX on the occurrence of SMVL.

The trends for better visual acuity outcome in the RBX 32-mg/day group were not apparent in the VFQ scores. This is not surprising since VFQ results reflect visual acuity in the patient’s better-seeing eye, and while 22% of placebo-treated patients experienced SMVL, this occurred in the better-seeing eye in only 7% (four patients).

SMVL and MVL are often caused by DME involving the center of the macula. In the PKC-DRS, 79% of patients had DME in at least one study eye at baseline and 34% had DME involving the center of the macula in at least one study eye at baseline. The rate of SMVL was higher in eyes with definite DME at baseline, and it was among these eyes that there was a trend for a beneficial effect of 32 mg/day RBX (Fig. 5). This study was not designed to demonstrate an effect of RBX on DME or visual function, and patients were enrolled regardless of their DME status at baseline. The broad range of baseline DME severity in study eyes (Table 2) makes it difficult to detect an effect of RBX on change in distance of edema from the center of the macula.

DME results from increased leakage of plasma components from the retinal vasculature and has been postulated to occur via two different PKC β–mediated mechanisms. Diabetes-induced de novo synthesis of diacylglycerol activates PKC β in the retina (30), which increases vascular permeability through various mechanisms (38,39) including phosphorylation of junctional proteins and dissolution of tight junctions (40,41). In addition, VEGF is a potent vasopermeability factor that may be involved in DME, the intracellular signaling pathway of which involves activation of PKC β (11). RBX prevents diabetes- and VEGF-induced retinal vascular leakage in animals (11). RBX is more effective at preventing diabetes-induced retinal vascular leakage in these models than in preventing retinal neovascularization (13,28). Treatment of DME patients for 3 months with a multitargeted kinase inhibitor, which also acts as a nonspecific PKC inhibitor, led to reductions in some measures of retinal thickening as evaluated by optical coherence tomography (42). Systemic applicability of this nonselective compound was limited by gastrointestinal side effects and dose-related problems with tolerability, glycemic control, and liver toxicity.

A beneficial effect of RBX on MVL might also be the result of improved retinal cell viability resulting from PKC β inhibition. PKC β activation occurs very early in diabetes, before the onset of clinically relevant retinopathy (29,30). Considerable evidence suggests that the activation of the PKC β isoform is responsible for subsequent diabetes-induced ocular and nonocular microvascular complications (8). Reduction of PKC β activity might therefore result in greater resistance of retinal vascular and neural cells to the pathologic stresses of hyperglycemia. This may result in greater resistance to cellular damage, reduced cellular dysfunction and/or loss, and prevention of subsequent visual decline. Such a mechanism would be consistent with the prevention of visual decline noted in those patients with DME at baseline (Fig. 5B) even though resolution of DME was not observed. Even if the mechanism by which RBX reduces vision loss involves decreasing edema, the degree of vision loss from involvement of the center of the macula by edema depends on the degree and duration of such involvement, as well as other factors that are not well understood (43). Not all DME progression to the center of the macula results in vision loss, and numerous patients with substantial central retinal thickening from DME may retain excellent visual acuity (L.P.A., personal communication, http://www.drcr.net/ preliminary data) (44,45).

When considering systemic therapy, the safety profile of a compound is of substantial importance. This is particularly true when inhibiting a key signaling enzyme such as PKC, where substantial toxicity might be expected. As indicated above, treatment of DME patients with an inhibitor of multiple kinases and PKC isoforms resulted in significant liver enzyme elevations, nausea, vomiting, and diarrhea (42). In contrast, RBX is selective for the β-isoform of PKC and highly selective for PKC as compared with other kinases (46). This selectivity for a single PKC isoform involved in mediating diabetes-induced ocular complications would be expected to result in limited side effects. Indeed, the safety profile in 937 patients derived from two studies of ocular complications in individuals with diabetes demonstrates that RBX is well tolerated without significant adverse effects over 30 to 52 months of treatment.

Only nine adverse events occurred, with an incidence exceeding 1%, that were statistically different among groups (Table 3). No serious adverse events were reported more frequently in RBX treatment groups. Although the nonserious adverse event occurrence frequency of diarrhea, flatulence, nephropathy, proteinuria, and coronary artery disease was highest among patients in the 16-mg/day RBX treatment group (Table 3), there did not appear to be RBX dose-response effects. In addition, the small number of events makes it likely that any disparity in the 16-mg group was due to chance. However, safety has been carefully monitored in numerous completed and ongoing trials assessing DME as well as other diabetes complications.

Patients taking the highest RBX dose (32 mg/day) did not experience these same events more frequently than placebo patients. First-degree atrioventricular block, asthma, and dysuria were statistically different among treatment groups, with the highest rate of occurrence in the 32-mg/day RBX group. These events represent spontaneous reports from investigative sites where test abnormalities were not uniformly acquired. When ECGs of all 937 patients in both the PKC-DRS and PKC-DMES were examined, there was no evidence of cardiac conduction interval prolongation in any RBX treatment group. For the events of asthma and dysuria, there were no concomitant increases in the occurrence of related conditions such as bronchospasm, wheezing, or urinary tract infection, and there were no observed differences among treatment groups in serious adverse events in the system organ classes related to these events. Furthermore, the same adverse events were not seen with greater frequency in RBX-treated patients in analyses of five nonocular trials of 6–12 months’ duration. To date, over 1,400 patients have been exposed to RBX, and no clinically significant increase in adverse effects has been identified. Careful monitoring of adverse effects is continuing.

The PKC-DRS is the first clinical trial evaluating the effect of a PKC isoform-selective inhibitor on ocular complications in patients with diabetes. These data demonstrate that, unlike less selective compounds, oral administration of RBX at doses up to 32 mg/day was well tolerated without significant adverse effects over 30 to 52 months of treatment. In patients with moderately severe to very severe NPDR at baseline, RBX did not prevent retinopathy progression to proliferative disease. However, compared with placebo, treatment with 32 mg/day RBX was associated with less visual loss, especially in patients with DME at baseline and when accounting for important retinopathy covariates. These data support further evaluation of RBX to prevent MVL in diabetes.

Canada: Philip Hooper, London, Ontario; S.A. Ross, Calgary, Alberta. Denmark: Henrik Lund-Andersen, Herlev. Netherlands: Bettine Polak, Amsterdam. U.K.: John V. Forrester, Aberdeen; Eva M. Kohner, London; J. Vora, Liverpool. U.S.: Everett Ai, San Francisco, CA; Lloyd M. Aiello, Boston, MA; Raijv Anand, Dallas, TX; Mark Blumenkranz, Menlo Park, CA; Alexander J. Brucker, Philadelphia, PA; Thomas Chandler, Austin, TX; Lawrence Chong, Los Angeles, CA; Doug Dehning, Independence, MO; Dan Finkelstein, Baltimore, MD; Robert Frank, Detroit, MI; Charles Garcia, Houston, TX; Thomas W. Gardner, Hershey, PA; Karen M. Gehrs, Iowa City, IA; Roy A. Goodart and David Faber, Salt Lake City, UT; Justin Gottlieb, Madison, WI; Craig M. Greven, Winston-Salem, NC; William E. Jackson, Denver, CO; James L. Kinyoun, Seattle, WA; Michael L. Klein, Portland, OR; Hilel Lewis, Cleveland, OH; Helen K. Li, Galveston, TX; Colin Ma and Richard Dreyer, Portland, OR; Raymond Margherio, Royal Oak, MI; Daniel F. Martin, Atlanta, GA; Philip Y. Paden, Medford, OR; George Sharuk, Boston, MA; Lawrence J. Singerman, Beachwood, OH; William E. Smiddy, Miami, FL; Michael Trese, Royal Oak, MI; James P. Tweeten, Boise, ID; Andrew Vine, Ann Arbor, MI.

Lloyd P. Aiello, Boston, MA (Chair); Matthew D. Davis, Madison, WI; Roy C. Milton, Rockville, MD; Matthew J. Sheetz, Indianapolis, IN; Vipin Arora, Indianapolis, IN; Louis Vignati, Indianapolis, IN.

This study was funded by Eli Lilly.

The authors thank all the subjects who participated in the study and Rocky Johnson, Keri Kles, Tim Mason, and Kuolung Hu for their valuable assistance with the preparation of the manuscript.