The Effect of Interventions to Prevent Type 2 Diabetes on the Development of Diabetic Retinopathy: The DPP/DPPOS Experience

Several randomized clinical trials have shown that the onset of type 2 diabetes (T2D) can be prevented or delayed by intervention before the onset of diabetes (1), including lifestyle (2–5), medication (2,6–9), or surgical (10,11) interventions. It would be expected that such interventions would also prevent or delay the development of the micro- and macrovascular and neurological complications of diabetes, but evidence for this is sparse, as reviewed by Nathan et al. (12) We, therefore, evaluated this question with respect to retinopathy in the Diabetes Prevention Program (DPP) and the Diabetes Prevention Program Outcomes Study (DPPOS).

The DPP was a randomized controlled clinical trial of a lifestyle intervention targeting lower weight and increased physical activity or metformin to reduce progression to T2D in a cohort of adults at high risk for development of T2D (prediabetes) (13,14). DPP had a mean follow-up of 3.2 years (range 0.0–5.3 years). The DPPOS has followed these subjects for an additional 19 years as of 2020. Our initial results at DPP-end suggested that lesions consistent with diabetic retinopathy were present in 12.6% of those who had progressed to diabetes, based on oral glucose tolerance testing, and in 7.9% among a subgroup of participants who had not progressed to diabetes at the time of evaluation (14).

We have now examined the prevalence and severity of diabetic retinopathy by fundus photography in the entire cohort over 20 years of follow-up to determine whether the prevalence of diabetic retinopathy differed between treatment groups or use of metformin. We chose to report prevalence not incidence, because we did not have a determination of retinopathy status at the beginning of the DPP study and the cohorts measured at the four time points are not constant.

The design, implementation, and primary results of the DPP have been previously reported (13,14). In brief, the DPP was an National Institutes of Health-sponsored three-arm, randomized, placebo-controlled trial to determine whether, compared with a placebo control group, metformin or an intensive lifestyle intervention aimed at weight loss would reduce the incidence of diabetes in adults at high risk for developing diabetes. High-risk subjects, defined as having impaired glucose tolerance (“prediabetes”) plus elevated fasting glucose and BMI ≥24 kg/m², were randomly assigned to the placebo group (PLB), metformin group (MET), or intensive lifestyle intervention group (ILS). The primary results of the DPP Study have been reported (2). The DPP Outcomes Study (DPPOS) is a long-term longitudinal, observational follow-up of the DPP cohort. During both DPP and DPPOS, diabetes was diagnosed by American Diabetes Association criteria based on a 2-h oral glucose tolerance (OGTT) test done yearly, a fasting glucose done at 6 months between the OGTTs, or, more recently, an HbA_1c ≥6.5% confirmed with glucose-based testing.

One aim of DPPOS was to determine whether treatment group assignment during DPP had any impact on the subsequent development of microvascular complications. At the study visit at the beginning of DPPOS at a mean of 4.2 years (range 0.0–6.3) postrandomization, available subjects who had progressed to diabetes (n = 595; 68% of 876 who had progressed) and a subset of those who had not progressed (n = 304; 17% of 1,832 who had not progressed) underwent seven-field stereo fundus photography. Fundus photography was again performed on all available and consenting subjects, regardless of diabetes status, at DPPOS years 5, 11, and 16 (mean years since randomization 9.1 [range 7.3–11.8], 14.5 [range 12.9–16.7], and 20.0 [range 18.3–22.1], respectively). All fundus photographs were graded using the Early Treatment Diabetic Retinopathy Study (ETDRS) grading system (15) by the Fundus Photography Reading Center at the University of Wisconsin. Diabetic retinopathy was diagnosed by the presence of typical lesions (microaneurysms, exudates, or hemorrhage) or more advanced lesions in either eye (ETDRS grade of ≥20 in either or both eyes). Clinically significant macular edema (CSME) was also diagnosed based on reading of the fundus photographs.

Weight, height, BMI, blood pressure, HbA_1c, fasting glucose, 2-h glucose (for those who had not yet developed diabetes), fasting lipid profile (total cholesterol, triglycerides, HDL cholesterol, and calculated LDL cholesterol), urinary albumin-to-creatinine ratio, and glomerular filtration rate (eGFR; based on serum creatinine as estimated with the Chronic Kidney Disease Epidemiology Collaboration equation) (16) were determined annually during DPP and DPPOS. Hypertension was defined as a blood pressure >140/90 or using antihypertension medications. Dyslipidemia was defined as LDL cholesterol >130 mg/dL, triglycerides >200 mg/dL, HDL cholesterol <40 mg/dL, or using lipid-lowering medications. Total circulating adiponectin was measured with a latex particle-enhanced turbidimetric assay (Otsuka Pharmaceutical, Tokyo, Japan). The within-run and between-run coefficients of variation for this assay are 6.21% and 9.25%, respectively. Smoking history, pregnancy history, and history of gestational diabetes were also obtained at DPP baseline and throughout the follow-up. Insulin sensitivity (1/fasting insulin and HOMA-insulin resistance) and insulin secretion (HOMA-β and insulinogenic index [(ΔIns₁₂₀ − Ins₀)/(Glu_u120 − Glu₀)]) were determined annually based on the glucose and insulin levels during an OGTT. Potential risk factors considered included the baseline value of all candidate risk factors as well as including the average values of weight, fasting glucose, 2-h glucose, HbA_1c, and systolic and diastolic blood pressure up to and including the time points at which each set of stereoscopic fundus photographs were taken. These factors for the full cohort and for those who underwent fundus photography at each time point are shown in Supplementary Table 1.

We used compete case analysis using all available information because there were no data missing for the risk factors in Table 1. Intent-to-treat analyses were performed for all comparisons involving the original treatment assignments. Prevalences of retinopathy at the end of the follow-up across treatment groups were compared using the Pearson χ² test of independence. Sensitivity analyses stratified by sex, race/ethnicity groups, age at randomization categories (<45, 45–59, ≥60 years), obesity status (baseline BMI <30 kg/m² or ≥30 kg/m²), and diabetes status at the last fundus measurements were performed to examine associations between initial treatment randomizations and retinopathy risks in different subgroups. Percentages of three-step progression on a person-scale of the ETDRS or regression between two consecutive fundus examinations were compared across treatment groups using the Pearson χ² test at each time point. All calculations were done in SAS 9.4 software.

Of the 3,234 subjects randomized into DPP, 2,779 enrolled in DPPOS, and 2,499 of these underwent fundus photography at least once during DPPOS. The characteristics of these 2,499 participants at the time of each fundus photograph are shown in Table 1. Of these, retinal photography was performed in 899 (34% of those who completed that visit) at the first DPPOS visit (referred to as year 1) and was performed in 2,128 (84%), 2,086 (92%), and 1,563 (76%) at DPPOS visits years 5, 11, and 16, respectively. (The breakdown of subjects being followed and undergoing fundus photography at each time point is summarized in Supplementary Fig. 1.)

The prevalence of diabetic retinopathy in each treatment group at DPPOS years 1, 5, 11, and 16 is shown in Fig. 1. The numbers of participants with diabetic retinopathy across all three treatment groups were 99 of 899 (11.0%), 206 of 2,128 (9.7%), 238 of 2,086 (11.4%), and 145 of 1,563 (9.3%) at years 1, 5, 11, and 16, respectively, among those who had retinal photographs at each time point. There was no difference between the prevalence of retinopathy between treatment groups at any time point (8.9% vs. 12.1% vs. 11.6% for ILS vs. MET vs. PLB, respectively, at year 1; 8.7% vs. 10.5% vs. 9.7% at year 5; 11.4% vs. 11.0% vs. 11.9% at year 11; and 10.0% vs. 9.1% vs. 8.8% at year 16). Of the 145 with diabetic retinopathy at year 16, 96.5% had mild nonproliferative diabetic retinopathy (NPDR; ETDRS grade 20–43), 1.4% had moderate-severe NPDR (ETDRS grade 47–61), and 2.1% had PDR (ETDRS grade 65–85) (see Supplementary Table 2). The prevalence of any retinopathy and the level of retinopathy did not differ significantly (P = 0.44–0.85) between treatment groups at any of the time points (Fig. 1).

We evaluated the effect of treatment group assignment during DPP on the subsequent development of diabetic retinopathy within prespecified subgroups, including sex, race/ethnicity, BMI at entry into DPP (<30 kg/m² vs. ≥30 kg/m²), and age at entry into DPP (<40 years, 40–65 years, ≥65 years). As shown in Table 2, there was no difference in the prevalence of retinopathy between treatment groups at any time point overall or within subgroups by sex (P = 0.26–0.82 for women; 0.42–0.95 for men), race/ethnicity (P = 0.08–0.86 for NHW; 0.24–0.85 for Blacks; 0.42–0.82 for Hispanics; 0.30–0.98 for American Indians; 0.37–0.79 for Asians), BMI at entry in DPP (P = 0.43–0.80 for those <30 kg/m²; 0.19–0.83 for those ≥30 kg/m²) or age at entry into DPP (P = 0.62–0.99 for those <40 years; 0.25–0.43 for those 40–65 years; 0.39–0.96 for those >65 years).

As noted above, based on an intention-to-treat analysis using the DPP treatment group assignment, treatment group did not have a significant effect on the prevalence of retinopathy at any time point (1, 5, 11, or 16 years) in the entire cohort or in any of our subgroup analyses (sex, race/ethnicity, baseline age, baseline BMI). However, although the exposure to metformin was markedly greater in the MET group than in the ILS or PLB groups, some subjects in the ILS or PLB groups did receive metformin by their primary physician or health care provider during DPP (because they developed diabetes) or during DPPOS (either because they developed diabetes or in hopes of slowing or preventing the onset of diabetes). Therefore, we compared the prevalence of retinopathy for those who were taking any metformin to those who were not taking metformin at the time of outcome measurements, regardless of treatment group assignment. There was no difference in the prevalence of retinopathy between those who, regardless of treatment group assignment, had received metformin and those who had not (data not shown.) This was also true within all age-group categories (<45 years, 45–60 years, >60 years). This lack of difference remained true after adjusting for potential confounders, including duration of T2D up to the time of fundus measurements, average HbA_1c, weight, and diastolic blood pressure during follow-up and baseline adiponectin.

At the time of the last fundus photograph among the participants with at least one fundus photograph, 1,614 subjects had already developed diabetes, and 385 of these 1,614 (24%) had retinopathy; 885 still had not developed diabetes, and 127 of these 885 (14%) had developed retinopathy. The difference in retinopathy prevalence between subgroups with diabetes and prediabetes was significant at P < 0.001. Within the subgroups of those who had developed diabetes and those who had not yet developed diabetes, there was no difference in the prevalence of retinopathy between treatment groups. There was no difference in age, sex, race/ethnicity, or baseline BMI between those with or without T2D or between those with or without retinopathy.

Progression of retinopathy was defined as a three-step progression between successive time points among those who had fundus photography at two successive time points on the ETDRS grading system using both eyes, similar to what was done in Diabetes Control and Complications Trial (DCCT)/Epidemiology of Diabetes Interventions and Complications (EDIC) (15), and over the entire course of the follow-up among those who had fundus photography at both years 1 and 16. Three-step progression of retinopathy between examinations was low, occurring in 0.9%, 0.7%, and 2.5% between years 1 and 5, years 5 and 11, and years 11 and 16, respectively. There was no difference in progression between DPP treatment group assignments by the end of follow-up, with 3.95% in ILS, 3.35% in MET, and 4.26% in the PLB (P = 0.66). Regression to a lower ETDRS score did not differ by treatment groups and occurred in 8.12%, 6.55%, and 7.89%, between years 1 and 5, years 5 and 11, and years 11 and 16, respectively, in the cohort with fundus measures.

CSME based on fundus photography was present in 10 participants across all four time points (2 participants at DPPOS year 1, 1 participant at DPPOS year 5, 2 participants at DPPOS year 11, and 5 participants at DPPOS year 16), and there was no difference in the prevalence of macular edema between DPP treatment group assignment.

We believe this to be the largest prospective, long-term longitudinal follow-up with retinal photography of a cohort of subjects at increased risk for the development of T2D. In this study, we evaluated the prevalence of retinopathy in 2,086 subjects for 11 years and in 1,553 for 16 years after completion of their participation in the DPP Study. These subjects had prediabetes (elevated fasting glucose) and impaired glucose tolerance (IGT), and BMI ≥24 kg/m² at the time of enrollment in DPP. The time to onset of diabetes, by American Diabetes Association criteria, was known in these subjects within a 6-month time window. This enabled the determination of the presence of diabetic retinopathy in prediabetes and from the time of biochemical onset of T2D, rather than from the time of clinical diagnosis, as has been done in most studies.

Our results show that despite reduction in the incidence of T2D by a lifestyle intervention or metformin in subjects with prediabetes, these interventions did not result in a reduction in the prevalence or severity of nonproliferative diabetic retinopathy up to 20 years later. Although the prevalence of retinopathy was lower in those who did not develop diabetes than in those who did, there was no difference in the prevalence of nonproliferative retinopathy between treatment groups regardless of diabetes status. When we explored the potential effect of the interventions in subgroups based on sex, race/ethnicity, age at entry into the DPP, or BMI at entry into the DPP, there was still no treatment effect on the prevalence of diabetic retinopathy. In addition, the use of metformin, regardless of treatment group assignment, did not have an effect on the prevalence of diabetic retinopathy.

These results are in agreement with other published studies, reviewed by Nathan et al. (12), showing that interventions that reduced the development of diabetes did not always reduce long-term microvascular complications of diabetes. Numerous studies using various interventions to prevent the development of diabetes in subjects at risk for diabetes, usually based on the presence of IGT, did not report retinopathy outcomes (17–24). The Da Qing Diabetes Prevention Outcome Study (DQDPOS), which compared a 6-year lifestyle intervention to control subjects with IGT, evaluated retinopathy after 20 and 30 years in 540 of the 577 of the randomized subjects. In this study, there was a 47% reduction (16.2% vs. 9.2%; P = 0.03) of severe diabetic retinopathy (vision loss, photocoagulation, or proliferative diabetic retinopathy) after 20 years (25) and a 40% reduction after 30 years (26). However, at 20 years, based on retinal photographs using a nonmydriatic camera and graded by the ETDRS system, there was no difference (38.7% vs. 36.1%; P = 0.51) between the intervention group and the control group in nonproliferative retinopathy (25). In our study, only 2.1% had proliferative retinopathy at 16 years, the duration of the follow-up was shorter, which may, in part, explain the different findings related to severe retinopathy. The Outcome Reduction With Initial Glargine Intervention (ORIGIN) study (27) of glargine in subjects with dysglycemia, which included nearly 1,500 subjects with IGT at baseline, showed a reduction (hazard ratio 0.90; 95% CI 0.89–0.99) in advanced complications for those with an initial HbA_1c >6.4%, but not in those with prediabetes (HbA_1c ≤6.4%; hazard ratio 1.07; 95% CI 0.95–1.20). However, this study did not separate out the various components of the microvascular outcome, so it does not provide data specifically about retinopathy.

Our study has some limitations. First, since our participants were early in the disease course, the number with retinopathy was relatively small, and thus, the power to detect group differences may be limited. We can conclude, however, that if an effect of treatment group does exist in our study, it is minor and could not be detected in a study of this size and duration. In addition, the number of subjects with proliferative retinopathy or CSME is too few to be able to show a treatment group effect on sight-threatening retinal disease.

Second, our assessment of retinopathy was based on a single set of seven-field retinal photographs at each time point. Although in the DCCT/EDIC study, in which progression of retinopathy was the primary outcome, retinopathy was classified based on two consecutive sets of photographs, most studies have used a single set of fundus photographs to define retinopathy. Indeed, at 5, 11, and 16 years, 8.1%, 6.6%, and 7.9%, respectively, had regression to a lower retinopathy grade.

Third, retinopathy was assessed only at four time points separated by ∼5 years, and the populations at each time point were not the same, making comparisons between time points problematic. For example, fewer subjects were available for retinal photography at year 16 (n = 1,563) than at year 11 (n = 2,086); 587 who were evaluated for retinopathy at year 11 were not evaluated at year 16, and 64 who were evaluated at year 16 missed their year 11 measurements.

Fourth, although we determined the onset of diabetes within a 6-month window, we were not able to determine the exact onset of retinopathy other than within 5-year intervals.

Fifth, we do not have an ophthalmologic evaluation at the baseline visit of the DPP. Although at that time all subjects had prediabetes, without a baseline assessment, we cannot determine the incidence of retinopathy.

Lastly, although there appears to be a lower retinopathy prevalence at year 16 compared with year 11, fewer subjects were available for retinal photography at year 16 (n = 1,563) than at year 11 (n = 2,086). The percentages lost-to-follow-up and deaths do not differ across treatment groups (P = 0.5340).

In conclusion, in adults at risk for diabetes because of the presence of prediabetes and overweight/obesity, diabetic retinopathy begins to develop early during the course of dysglycemia, occurring during the prediabetic phase and before the diagnosis of diabetes based on currently accepted criteria. Although interventions during the prediabetic phase influence the incidence of developing diabetes, such treatments did not reduce the prevalence of sight-threatening retinopathy after 20 years of follow-up. Since interventions that reduce the development of diabetes do not appear to reduce the subsequent development of long-term diabetes-related retinopathy, and since retinopathy is invariably mild and of little clinical consequence, screening for retinal changes in persons with prediabetes does not seem to be warranted based on currently available data. Whether interventions to reduce plasma glucose or other metabolic abnormalities during the prediabetic phase will alter the course of long-term complications requires further study.

Acknowledgments. The DPP Research Group gratefully acknowledges the commitment and dedication of the participants of the DPP and DPPOS.

Funding. Research reported in this publication was supported by the National Institutes of Health (NIH), National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) under award numbers U01 DK048489, U01 DK048339, U01 DK048377, U01 DK048349, U01 DK048381, U01 DK048468, U01 DK048434, U01 DK048485, U01 DK048375, U01 DK048514, U01 DK048437, U01 DK048413, U01 DK048411, U01 DK048406, U01 DK048380, U01 DK048397, U01 DK048412, U01 DK048404, U01 DK048387, U01 DK048407, U01 DK048443, and U01 DK048400, by providing funding during DPP and DPPOS to the clinical centers and the Coordinating Center for the design and conduct of the study, and collection, management, analysis, and interpretation of the data. Funding was also provided by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the National Institute on Aging, the National Eye Institute, the National Heart Lung and Blood Institute, the National Cancer Institute, the Office of Research on Women’s Health, the National Institute on Minority Health and Health Disparities, the Centers for Disease Control and Prevention, and the American Diabetes Association. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The Southwestern American Indian Centers were supported directly by the NIDDK, including its Intramural Research Program, and the Indian Health Service. The General Clinical Research Center Program, National Center for Research Resources, and the Department of Veterans Affairs supported data collection at many of the clinical centers. Merck KGaA provided medication for DPPOS. DPP/DPPOS have also received donated materials, equipment, or medicines for concomitant conditions from Bristol-Myers Squibb, Parke-Davis, and LifeScan Inc., Health O Meter, Hoechst Marion Roussel, Inc., Merck-Medco Managed Care, Inc., Merck and Co., Nike Sports Marketing, Slim Fast Foods Co., and Quaker Oats Co. McKesson BioServices Corp., Matthews Media Group, Inc., and the Henry M. Jackson Foundation provided support services under subcontract with the Coordinating Center.

The sponsor of this study was represented on the Steering Committee and played a part in study design, how the study was done, and publication. All authors in the writing group had access to all data. The opinions expressed are those of the study group and do not necessarily reflect the views of the funding agencies. A complete list of Centers, investigators, and staff can be found in the Appendix.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. N.H.W. wrote the manuscript and researched data. Q.P. researched data. W.C.K., E.B.S., D.D., E.Y.C., B.B., R.B.G., X.P.-S., C.D., M.S., and D.M.N. reviewed/edited the manuscript. Q.P. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.