Introduction
Dipeptidyl peptidase-4 (DPP-4) inhibitors are an emerging class of oral hypoglycemic agents increasingly being used in the treatment of diabetes. The primary function of DPP-4 is to enzymatically degrade peripheral incretin hormones, such as glucagon-like peptide (GLP)-1 and gastric inhibitory polypeptide (GIP).1,2 Thus, suppression of this enzyme helps regulate blood glucose levels by increasing circulating incretin levels, promoting insulin secretion from the pancreatic β cells and suppressing glucagon secretion from the αcells.1,2
DPP-4 inhibitors have been demonstrated to be highly effective in lowering hemoglobin A1c levels, either as monotherapy or in combination therapy.1–3 Furthermore, several large cardiovascular outcome trials have established their protective effect on major adverse cardiac events in type 2 diabetics with an established cardiovascular risk profile.4–6 The impact of this medication class on systemic microvascular complications, however, is less well understood. Limited, but growing evidence recently has suggested that systemic DPP-4 inhibitor use may be protective against the development or progression of diabetic retinopathy (DR) independent of its glucose-lowering effects.7–9
These studies did not control for the possible effects of DPP-4 inhibitors on therapies used to treat DR and/or diabetic macular edema (DME), such as vascular endothelial growth factor (VEGF) inhibitors and laser photocoagulation. Here, we conducted a post hoc analysis to evaluate whether patients with DME taking systemic DPP-4 inhibitors differed in baseline characteristics as well as if there was a protective effect on treatment outcomes in the VISTA and VIVID studies, two large phase 3 studies evaluating intravitreal aflibercept injection (IAI) (Eylea; Regeneron Pharmaceuticals, Tarrytown, NY).
Patients and Methods
Study Design
This was a post hoc exploratory analysis of previously completed prospective phase 3 clinical trials, VISTA (NCT01363440) and VIVID (NCT01331681). Both trials were similarly designed, randomized, double-masked, and active controlled and followed patients for 148 weeks. Each clinical site's respective institutional review board/ethics committee approved the study. All patients provided written informed consent prior to the enrollment. The studies were conducted in compliance with the International Conference on Harmonization guidelines and the Health Insurance Portability and Accountability Act of 1996.10,11
Study design and patient eligibility for VISTA and VIVID have been described previously by Korobelink et al.12 Briefly, eligible patients were adults with type 1 or type 2 diabetes mellitus who presented with central involved DME (defined as retinal thickening involving the central 1 mm, measured by spectral-domain optical coherence tomography central subfield thickness [CST]). Best-corrected visual acuity (BCVA) in the study eye was required to be between 73 and 24 Early Treatment Diabetic Retinopathy Study (ETDRS) letters (approximately 20/40 to 20/320 Snellen equivalent). Only one eye per patient was eligible for enrollment. Eyes were randomized 1:1:1 to receive IAI 2 mg every 4 weeks (2q4), IAI 2 mg every 8 weeks after five initial monthly doses (2q8), or macular laser photocoagulation at baseline and at visits in which patients met any of the laser retreatment criteria (laser control group). Study eyes in all treatment groups were assessed for laser retreatment if criteria were met beginning of Week 12.13 If criteria were met, study eyes in the 2q4 and 2q8 groups received sham laser and those in the laser control group received active laser, no more frequently than every 12 weeks.
Beginning at Week 24, study eyes in all treatment groups were assessed for additional “rescue” treatment based on protocol-defined criteria.13 Rescue treatment for study eyes enrolled in the 2q4 and 2q8 groups was active laser, and study eyes that were enrolled in the laser control group received five doses of 2 mg IAI every 4 weeks followed by dosing every 8 weeks.12,13
Outcome Measures
This post hoc analysis evaluated the impact of baseline use of systemic DPP-4 inhibitors on baseline characteristics and improvements in BCVA, CST, and Diabetic Retinopathy Severity Scale (DRSS) score through Week 100.
Statistical Methods
The analysis integrated data from VISTA and VIVID. The combined population was assigned to one of two subgroups, +DPP-4 inhibitor or –DPP-4 inhibitor, based on the status of reported DPP-4 inhibitor use at baseline, respectively. A list of DPP-4 inhibitors included in this analysis are listed in Table 1. Visual and anatomic outcomes were analyzed within each treatment group (2q4, 2q8, and laser control) every 4 weeks through Week 100. In patients receiving rescue treatment, data were censored from the time rescue treatment was given, and the last observation carried forward (LOCF) method was used to impute missing data. In this post hoc analysis, BCVA and CST were analyzed using one-way analysis of covariance. Odds ratios and Chi-squared tests were used to assess the association between DRSS improvement and subgroups. Confidence intervals (CIs) and P values were calculated based on the subgroup difference [+DPP-4 vs –DPP-4] within each treatment group of the least squares (LS) mean change. All statistical analyses were performed using SAS version 9.2 (SAS Institute, Cary, NC).
Results
Demographic and Baseline Characteristics
Of 865 eyes randomized and treated in VISTA and VIVID, 862 were included in this integrated post hoc analysis. Three eyes were excluded because of the lack of a post-baseline BCVA assessment. Overall, 12.2% (n = 35), 9.7% (n = 28), and 15.4% (n = 44) of patients in the laser control, 2q4, and 2q8 groups, respectively, reported DPP-4 inhibitor use at baseline (Figure 1). A small number of patients who were not using a DPP-4 inhibitor at baseline initiated DPP-4 therapy during the study: 15 (6.0%) in the laser control, nine (3.4%) in the 2q4 group, and 17 (7.0%) in the 2q8 group.
Baseline characteristics by baseline DPP-4 inhibitor use are summarized in Table 2 and were similar across treatment groups and DPP-4 inhibitor subgroups. In the +DPP-4 inhibitor subgroup, baseline mean BCVA was 62.3, 59.5, and 62.0 letters in the laser control, 2q4, and 2q8 groups, respectively. These results were similar to those observed for eyes in the –DPP-4 inhibitor subgroup with 60.0, 59.8, and 58.6 letters in the laser control, 2q4, and 2q8 groups, respectively. In addition, eyes in both DPP-4 inhibitor subgroups had similar CST and DR severity at baseline.
Impact of Baseline DPP-4 Inhibitor Use on Treatment Outcomes
Visual Acuity: In the +DPP-4 inhibitor subgroup, the mean changes from baseline BCVA in the laser control, 2q4, and 2q8 groups at Week 100 were 1.1, 13.4, and 9.7 letters, respectively. Similarly, for eyes in the –DPP-4 inhibitor subgroup, the mean changes from baseline BCVA at Week 100 were 0.7, 11.3, and 10.4 letters in the laser control, 2q4, and 2q8 groups, respectively (Figure 2A). Comparing BCVA outcomes after adjusting by differences in baseline values per treatment group, no differences in LS mean BCVA change were observed within each treatment group at Weeks 52 and 100 when comparing eyes in the +DPP-4 inhibitor subgroup with eyes in the –DPP-4 inhibitor subgroup (Figure 2B).
Central Subfield Thickness: In the +DPP-4 inhibitor subgroup, mean CST changes from baseline in the laser control, 2q4, and 2q8 groups at Week 100 were −81.2, −220.7, and −195.1 μm, respectively. Corresponding CST changes from baseline for eyes in the –DPP-4 inhibitor subgroup were −85.2, −198.9, and −193.0 μm, respectively (Figure 3A). No difference in LS mean CST change was observed within each treatment group at Weeks 52 and 100 when comparing eyes in the +DPP-4 inhibitor subgroup with eyes in the –DPP-4 inhibitor subgroup (Figure 3B).
Diabetic Retinopathy Severity Score
The proportion of eyes with a 2-or-greater-step DRSS score improvement are summarized in Figure 4A. In the +DPP-4 inhibitor subgroup, 19%, 38%, and 29% of eyes in the laser control, 2q4, and 2q8 groups, respectively, demonstrated a 2-or-greater-step improvement from baseline in DRSS score at Week 52. The corresponding outcomes for the –DPP-4 inhibitor subgroup were 12%, 36%, and 30%, respectively. The proportions of eyes in the +DPP-4 inhibitor subgroup with a 2-or-greater-step improvement from baseline in DRSS score at Week 100 were 19%, 44%, and 40%, respectively, in the laser control, 2q4, and 2q8 groups (Figure 4A). Corresponding proportions of eyes in the –DPP-4 inhibitor subgroup were 13%, 35%, and 36%, respectively. Although the odds of having a 2-or-more-step improvement in DRSS score were not significantly different between the DPP-4 inhibitor subgroups within each treatment group at Week 52 or 100 (Figure 4B), it should be noted that a larger proportion of eyes experienced an improvement in DRSS score in the IAI treatment groups from Week 52 to 100 in the +DPP-4 inhibitor subgroup, but not the –DPP-4 inhibitor subgroup.
Discussion
Intensive glycemic control has been well established as a method of mitigating the negative effects of type 1 and type 2 diabetes on both the microvasculature and macrovasculature.14,15 Although DPP-4 inhibitors are known to improve glycemic control in diabetes in general, their specific influence on microvascular tissues, such as the retina, remains the subject of ongoing research. This post hoc analysis of VISTA and VIVID demonstrated that baseline characteristics and functional and anatomical treatment outcomes, did not appear to differ significantly between patients who were or were not using a DPP-4 inhibitor at baseline. To our knowledge, this is the first study to evaluate whether there is a correlation between concomitant use of DPP-4 inhibitors and DME outcomes with anti-VEGF treatment; prior investigations have focused on the presence/absence or level of severity of DR.7–9
Earlier studies suggested that DPP-4 inhibitors may exacerbate, rather than improve, DR. In a murine model of DR, Lee et al. observed that DPP-4 inhibitors increased retinal vascular leakage through activation of the stromal cell-derived factor-1β (SDF-1α)–Src-family tyrosine kinase–vascular endothelial (VE)-cadherin signaling pathway, known to be increased in proliferative DR.16,17 Animals overexpressing SDF-1β were shown to be more prone to developing neovascularization in ischemic tissues.18 Separately, in the Trial Evaluating Cardiovascular Outcomes with Sitagliptin (TECOS) evaluating 14,671 patients with type 2 diabetes and cardiovascular disease, the incidence of DR increased by 21.4% in the DPP-4 arm compared with placebo.6
More recent experimental studies have suggested a protective effect of the DPP-4 inhibitor class on DR. Goncalves et al. demonstrated that the DPP-4 inhibitor sitagliptin ameliorated retinal endothelial cell dysfunction triggered by the proinflammatory cytokine TNFβ in vitro.19 Similarly, DPP-4 suppression reduced blood–retinal barrier breakdown, inflammation, and neuronal cell death in animal models of type 1 and type 2 diabetes.20,21 Dietrich et al. observed that a different DPP-4 inhibitor, linagliptin, exhibited a protective effect on the diabetic retinal vasculature in an experimental rat model, which they concluded was due to a combination of neuroprotective and antioxidative effects of the agent.22 One proposed mechanism is through the upregulation of GLP-1, which has been shown to block the overproduction of reactive oxygen species and downstream proinflammatory cytokines by endothelial cells exposed to hyperglycemic conditions.23 In diabetic animal models, GLP-1 promotes antioxidative effects on the vasculature, reducing cellular apoptosis and increasing cell survival.24,25 Intravitreal injection of GLP-1 in diabetic rats has also been shown to transiently improve neuronal function and reduce glutamate toxicity.26 In addition, in a mouse model of oxygen induced retinopathy, Kolibabka et al. showed that linagliptin exerted anti-angiogenic effects through the inhibition of VEGF receptor signalling.27
A limited number of large clinical studies confirmed the positive experimental findings in patients with diabetes. Ott et al. conducted a double-blind, controlled, cross-over trial of 50 patients with type 2 diabetes without clinical signs of microvascular disease, who were randomized to receive 5 mg saxagliptin or placebo for a 6-week study period.7 Retinal arteriolar structure and retinal capillary flow at baseline and during flicker-light exposure were measured and after 6 weeks, the vasodilatory capacity had increased two-fold with saxagliptin treatment, implying improved retinal circulation with DPP-4 therapy. Chung et al. retrospectively reviewed 82 patients with type 2 diabetes, investigating associations between baseline risk factors and progression of DR.8 The investigators found that treatment with DPP-4 inhibitors was the only independent protective factor against the progression of DR, aside from improving glycemic control. A comparison by Kolaczynski et al. of microvascular outcomes between vildagliptin and a traditional sulfonylurea found that treatment with vildagliptin was associated with a significantly lower incidence of DR.9
The above-mentioned studies provided the rationale to conduct this post hoc analysis of the VISTA and VIVID population; however, the hypothesis of a beneficial effect of DPP-4 agents on DR progression was not confirmed. DPP-4 inhibitors also did not increase leakage in eyes with DME. Our results are consistent with a population-based study in Korea by Kim et al., who reported that during a median follow-up of 28.4 months, DPP-4 inhibitor use at any time was not found to be associated with either beneficial or negative effects on the risk of composite DR events as compared with no prior use.28 Interestingly, a subanalysis did suggest that during the first year of DPP-4 inhibitor therapy (treatment < 12 months), there was a greater risk of DR events compared with other glucose-lowering agents over the same period.28
The relatively low number of patients taking DPP-4 inhibitors in VISTA and VIVID (9.7% to 15.4%) was notable given how prevalent their use is currently in the management of diabetes. The first agent of the class to obtain U.S. Food and Drug Administration approval was sitagliptin (Januvia; Merck, Kenilworth, NJ) in 2006; however, the subsequent members to gain marketing approval became available either during or towards the conclusion of the VISTA/VIVID study period (2011–2014). This could help to explain the low number of patients we observed using this medication class in these studies. It would be interesting to investigate baseline characteristics and treatment outcomes with concurrent DPP-4 inhibitor use from a more recent DR clinical trial. Alternatively, it cannot be ruled out that patients taking DPP-4 inhibitors may be less prone to develop visual impairment due to DME, which was the major eligibility criterion in VIVID and VISTA.
There are additional limitations to this study. First, as with any post hoc analysis, interpretation of these data is inherently limited. Second, we are unable to ascertain whether any potential impact of concurrent DPP-4 inhibitor use is generalizable to the entire therapy class, or whether effects may be directly attributable to a specific agent as some of the other reported studies discussed here have suggested. Finally, the impact of duration of DPP-4 treatment could not be analyzed. Self-reporting at the time of enrollment was dependent on patients among those using DPP-4 medications.
The strength of the post hoc analysis lies in the dataset acquired from two large, prospective, randomized clinical trials with fixed dosing. Masked personnel measured BCVA and anatomic parameters were graded by masked reading centers.
In summary, DPP-4 inhibitor use at baseline in VISTA and VIVID did not appear to influence the degree of visual gain, anatomical benefit as evidenced by CST reduction, or 2-or-greater-step improvement in DRSS scores in patients with DME treated with IAI or laser control through Week 100. The results of this study help strengthen the limited but growing body of literature regarding the interaction between DPP-4 inhibitors and microvascular complications of diabetes, such as DR. Further investigations are required to help better elucidate whether this class of medication does indeed exert a protective effect on DR progression.