Klein BEK, Myers CE, Howard KP, Klein R. Serum Lipids and Proliferative Diabetic Retinopathy and Macular Edema in Persons With Long-term Type 1 Diabetes Mellitus: the Wisconsin Epidemiologic Study of Diabetic Retinopathy. JAMA Ophthalmol. Published online December 11, 2014. doi:10.1001/jamaophthalmol.2014.5108.
Klein BEK, Myers CE, Howard KP, Klein R. Serum Lipids and Proliferative Diabetic Retinopathy and Macular Edema in Persons With Long-term Type 1 Diabetes MellitusThe Wisconsin Epidemiologic Study of Diabetic Retinopathy. JAMA Ophthalmol. 2015;133(5):503-510. doi:10.1001/jamaophthalmol.2014.5108
Copyright 2015 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Total serum and high-density lipoprotein cholesterol have been considered risk factors for severe vascular outcomes in persons with type 1 diabetes mellitus.
To examine the long-term relationships between these 2 serum lipids and the incidence and prevalence of proliferative diabetic retinopathy and macular edema.
Design, Setting, and Participants
Nine-hundred three persons with younger-onset type 1 diabetes mellitus who participated in the Wisconsin Epidemiologic Study of Diabetic Retinopathy.
Serum total and high-density cholesterol and history of statin use during the course of 5 visits spanning approximately 30 years (April 10, 1984, to February 13, 2014).
Main Outcomes and Measures
Prevalence and incidence of proliferative diabetic retinopathy and macular edema.
A modest association was found for higher levels of high-density lipoprotein cholesterol and decreased prevalence of proliferative diabetic retinopathy (odds ratio per 10 mg/dL, 0.87; 95% CI, 0.82-0.93), adjusting for duration of diabetes mellitus, glycosylated hemoglobin A1c, statin use, and end-stage renal disease. While adjusting for covariates, no associations of serum total or high-density lipoprotein cholesterol and incident proliferative diabetic retinopathy or macular edema, nor of statin use with decreased incidence of proliferative diabetic retinopathy or macular edema, were identified.
Conclusions and Relevance
In the course of long-duration diabetes mellitus during a time of changing medical care, there appeared to be little effect of serum lipids or statins on the incidence of proliferative diabetic retinopathy and macular edema.
Proliferative diabetic retinopathy and macular edema are important causes of decreased vision in persons with type 1 diabetes mellitus.1 Serum lipids have been found to be associated with the incidence and progression of lesions of diabetic retinopathy2 and macular edema,3 although in some studies, the associations were no longer observed after adjustment for important covariates.4 Long-term estimates of these relationships are uncommon because type 1 diabetes mellitus is an uncommon disease and systematic long-term follow-up data of persons in the general population with this condition are usually not available. The Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR) documented the presence and severity of retinal lesions associated with diabetes mellitus for more than 30 years during a period of change in treatment and levels of glycemia, blood pressure, and serum lipids.5- 8 In this report, we investigated the prevalence and incidence of proliferative diabetic retinopathy (PDR) and macular edema in the WESDR cohort during the course of 5 examinations spanning approximately 30 years.
The study group for this investigation included all persons with type 1 diabetes mellitus who received primary care in an 11-county area in southern Wisconsin between July 1, 1979, and June 30, 1980.9- 16 Of 1210 such persons, 996 participated in the baseline examination (1980-1982),9 903 participated in the 4-year follow-up (1984-1986),12 816 participated in the 10-year follow-up (1990-1992),13 667 participated in the 14-year follow-up (1994-1996),14 567 participated in the 20-year follow-up (2000-2001),17 520 participated in the 25-year follow-up (2005-2007),16 and 335 participated in the 32-year follow-up (2012-2014). Reasons for nonparticipation and comparisons between participants and nonparticipants at each examination have been presented elsewhere.9- 14,16- 18 Analyses in this report are limited to persons who completed at least 1 examination phase beginning at the 4-year follow-up, had information regarding retinopathy level, and had serum total and high-density lipoprotein (HDL) cholesterol levels measured at the time of their examinations. Data from the 20-year follow-up examination were excluded because the determination of the outcome variables (PDR and macular edema) was not comparable with that of the other examinations.
The examinations were performed in a mobile examination van or clinic near the participant’s place of residence or in the participant’s residence. Written informed consent was obtained from participants before each examination and all examinations followed a similar protocol approved by the institutional Human Subjects Committee of the University of Wisconsin, which conformed to the tenets of the Declaration of Helsinki.
The study examinations and interviews were conducted by trained examiners. Quality control was monitored throughout each study examination phase. The pertinent parts of the ocular and physical examinations included measuring height, weight, and blood pressure19; dilating the pupils; and taking stereoscopic color fundus photographs of 7 standard fields.20,21 Owing to funding constraints, there were no photographs taken at the 20-year follow-up examinations. A structured interview was conducted that included questions about medication use, history of kidney transplant and dialysis, and smoking history. If there was any doubt regarding history of medication use, it was verified by a physician’s report. An aliquot of whole blood was used for determination of the glycosylated hemoglobin A1c level using affinity chromatography (Isolab Inc). The normal range for hemoglobin A1c was 4.6% to 7.9%. Its intra-assay coefficient of variation was 2.4%. Serum was used to measure total and HDL cholesterol.22,23
For each eye, the maximum grade of diabetic retinopathy in any of the 7 standard photographic fields was determined for each of the lesions using the Early Treatment Diabetic Retinopathy Study (ETDRS) classification scheme.13,24 Proliferative diabetic retinopathy was defined as having an ETDRS severity level of 60 or greater in either eye. Macular edema was defined as retinal thickening in the macular area in either eye according to the ETDRS classification scheme.25
Because serum cholesterol was not measured at the baseline examination, the prevalence of PDR and macular edema was defined at each examination beginning at the second WESDR examination in 1984-1986 (and excluding the fifth examination in 2000-2001). Incidence of PDR and macular edema were defined, beginning at the second WESDR examination and excluding the fifth, as the presence of PDR or macular edema in either eye in an individual who had been free of PDR or macular edema at all previous examinations.
Age was defined as the participant’s age at the time of each examination. Age at diagnosis of diabetes mellitus was defined as the participant’s age at the time the diagnosis was first recorded by a physician on the patient’s medical record or in a hospital record. The duration of diabetes mellitus was defined as the period between the participant’s age at diagnosis and his or her age at each WESDR examination. Systolic and diastolic blood pressures were defined as the average of the last 2 of 3 obtained measurements taken according to the protocol of the Hypertension Detection and Follow-up Program.19 Body mass index was calculated as weight in kilograms divided by height in meters squared. End-stage renal disease (ESRD) was defined as participant reports of receiving dialysis or a kidney transplant. Statins are competitive inhibitors of hydroxymethylglutaryl-coenzyme A reductase. Statin use was defined as the current use at the time of each study examination.
Cross-sectional analyses evaluated associations of serum total and HDL cholesterol to the prevalence of PDR and macular edema using proportional odds models.
Incidence analyses examined the associations of total and HDL cholesterol with the incidence of PDR and macular edema. Multilevel modeling was used to account for the varying follow-up times between examinations (6 years between the 1984-1986 and 1990-1992 examinations, 4 years between the 1990-1992 and 1994-1996 examinations, 11 years between the 1994-1996 and 2005-2007 examinations, and 7 years between the 2005-2007 and 2012-2014 examinations). For incidence analyses, the duration of diabetes mellitus was used as the time scale and the baseline hazard was assumed to be piecewise constant with duration of diabetes mellitus categorized as less than 20 years, 20 to 29 years, and 30 or more years of duration. The duration of diabetes mellitus was not reported as a covariate in incidence models because it was used as the time scale. Hazard ratio estimates were calculated by exponentiation of estimated coefficients. The PROC GENMOD and PROC NLMIXED of SAS version 9.1 were used for the cross-sectional and incidence analyses, respectively.
Multivariable analyses were performed by adding adjustment variables in a stepwise manner. Initial prevalence and incidence models included serum total or HDL cholesterol and duration of diabetes mellitus. Next, hemoglobin A1c, statin use, and ESRD (in prevalence models) were sequentially added to each model to evaluate the effect of each covariate on each retinal outcome as well as on each lipid and retinal outcome relationship. End-stage renal disease was not included as a covariate in incidence models because too few individuals with ESRD were at risk for incidence of PDR or macular edema. However, because ESRD in persons with diabetes mellitus is strongly associated with the severity of retinopathy and with serum lipids, we also evaluated the relationship of lipids to retinal outcomes in the subset of the cohort without ESRD.
There were 819 participants (400 women and 419 men) who contributed 2319 person-visits to the analyses for the prevalence of PDR or macular edema. There were 520 participants (269 women and 251 men) who contributed 1146 person-visits to the analyses for the incidence of PDR or macular edema. In the prevalence and incidence analyses, the mean durations of diabetes mellitus were 27.7 years and 22.8 years, respectively; mean hemoglobin A1c was 8.7%; and 8.8%, respectively; and regular statin use was reported by 20.5% and 7.4% of participants during all person-visits, respectively (Tables 1 and 2). The estimated prevalence of PDR was 41.0% per person-visit and 23.8% for macular edema. The estimated incidence of PDR was 1.21% per person-year and 0.77% for macular edema.
The proportion of persons reporting statin use increased at each visit, with a marked increase reported at the last 2 examinations. Reported statin use was associated with lower total cholesterol levels (162.7 mg/dL vs 194.6 mg/dL [to convert to millimoles per liter, multiply by 0.0259], respectively, for statin users vs nonusers; P < .01, adjusting for age). However, HDL cholesterol differed little by statin use status during all visits (55.1 mg/dL vs 51.0 mg/dL [to convert to millimoles per liter, multiply by 0.0259] for users vs nonusers; P = .75, adjusting for age).
A higher serum total cholesterol level was associated with a higher prevalence of PDR (odds ratio [OR], 1.06; 95% CI, 1.02-1.10; Table 3). Adding hemoglobin A1c to the model slightly decreased the OR for total cholesterol (OR, 1.03; 95% CI, 0.99-1.07). Adding statin use and ESRD to the model had little effect on the OR for serum total cholesterol (OR, 1.03; 95% CI, 0.99-1.08). This series of analyses was repeated for serum HDL cholesterol (Table 3). Higher HDL cholesterol level was associated with decreased odds of PDR when including hemoglobin A1c in the model (OR, 0.86; 95% CI, 0.81-0.92) and also when further including statin use and ESRD (OR, 0.87; 95% CI, 0.82-0.93).
Quiz Ref IDApplying a similar modeling approach (Table 3), we found that serum total cholesterol was directly associated with increased odds of prevalent macular edema (OR, 1.08; 95% CI, 1.03-1.12); adjusting for age and serum HDL cholesterol was inversely associated with prevalent macular edema when adjusting for the duration of diabetes mellitus, hemoglobin A1c, and statin use (OR, 1.04; 95% CI, 0.99-1.09 for serum total cholesterol; OR, 0.91, 95% CI, 0.85-0.98 for HDL cholesterol) but not in the full model including ESRD (OR, 0.94; 95% CI, 0.87-1.01). After including ESRD in each model, only serum HDL cholesterol remained associated with a decreased prevalence of PDR (OR, 0.87; 95% CI, 0.82-0.93).
Quiz Ref IDSerum total cholesterol was associated with an increased risk of incident PDR (hazard ratio per 20 mg/dL, 1.13; 95% CI, 1.06-1.21) and serum HDL cholesterol was associated with a decreased risk of incident macular edema (hazard ratio per 10 mg/dL, 0.83; 95% CI, 0.71-0.96; Table 4). However, neither of these associations remained after further adjustment for hemoglobin A1c and statin use (Table 4).
To examine the potential effect of mortality on the lipid analyses, we examined the hazard of death associated with total and HDL cholesterol levels. The hazard ratio for total cholesterol was 0.96 (95% CI, 0.76-1.22) and for HDL cholesterol, 0.88 (95% CI, 0.69-1.13).
End-stage renal disease is often associated with lipid levels and macrovascular complications of diabetes mellitus. In this study cohort across all visits, serum total cholesterol and HDL cholesterol levels were not higher in those with ESRD compared with those without (188.2 mg/dL vs 186.9 mg/dL, respectively, for serum total cholesterol and 51.9 mg/dL vs 51.2 mg/dL, respectively, for serum HDL cholesterol), adjusting for statin use. However, adjusting for cholesterol levels, individuals with ESRD were more likely to currently take a statin (P < .01) than those without ESRD. When ESRD status was considered in our models, it did not materially affect the strength of the association of the lipids with prevalence of PDR and macular edema. End-stage renal disease itself was associated with prevalent retinal outcomes.
There were 217 individuals with ESRD in the prevalence analyses and 10 individuals with ESRD in the incidence analyses. When the previous analyses were rerun excluding these individuals, serum total cholesterol was directly associated with an increased prevalence of macular edema but not PDR and HDL cholesterol was inversely associated with risk of both macular edema and PDR, adjusting for duration of diabetes mellitus, hemoglobin A1c, and statin use. After excluding individuals with ESRD, no associations were found between serum total or serum HDL cholesterol and incident PDR or macular edema.
Quiz Ref IDThis report provides long-term follow-up data on the association between measures of serum lipids and prevalent and incident PDR and macular edema. While the outcomes for prevalence relationships of total and HDL cholesterol level accounting for hemoglobin A1c were modest, neither serum lipid was associated with incidence of either retinal outcome. There are limited data supporting an independent role of serum lipids on microvascular complications of diabetes mellitus.26- 28 We reported a cross-sectional association between retinal hard exudates and total cholesterol in a previous publication.29 We are uncertain of the reason for the differences but they may be related to the fact that our prior study included a small number of participants (n = 299) and that the temporal order of the association could not be determined from a cross-sectional relationship. Romero and colleagues28 reported that high levels of low-density lipoprotein cholesterol were associated with incident diabetic retinopathy in 112 persons with type 1 diabetes mellitus who were followed up for 15 years but an association with HDL cholesterol was not identified. Information on the incidence of PDR or macular edema with regard to serum lipids was not presented (PDR was an uncommon outcome), so it was not possible to directly compare those results with ours. Cetin and colleagues30 examined the relationships of total serum cholesterol, low-density lipoprotein cholesterol, and triglycerides with severity of diabetic retinopathy in a university-based clinical medical record review of 191 patients. No associations were identified. While our study was a cohort study and our longitudinal data provided a potentially powerful model to detect associations, our results suggested a modest effect of serum lipids on PDR and macular edema. To evaluate the possibility that selective mortality affected our results, we examined the association of serum total and HDL cholesterol on survival. In these data, we did not observe an effect, so we doubt that this caused a spuriously diminished relationship between the lipid levels and retinal outcomes.
The effects of statins on lipid levels are well established and the resultant decrease in the risk of macrovascular disease associated with these medications is also well documented.31 While we found that serum total cholesterol was lower in the presence of statin use,8 we found no evidence to support a beneficial effect of statins on microvascular disease as reflected in retinopathy in our study, compatible with the interpretation of limited effect of serum total and HDL cholesterol on the retinopathy end points. We also noted that we did not identify an effect of the presence of ESRD on the relationship of serum lipids to the retinal end points.
Quiz Ref IDIn the current analyses of the WESDR data, the presence of ESRD was a stronger correlate of prevalent PDR and macular edema than serum lipids, hemoglobin A1c, or duration of diabetes mellitus. This may reflect common risk indicators and risk factors (eg, markers of oxidation, inflammation, endothelial dysfunction, uric acid, cytokines, and advanced glycation end products) as well as effects of other metabolic changes and medications to which persons with ESRD are exposed.32- 34 We were unable to address the possible effects of these exposures and potential effects of other unmeasured confounders.
While long-term follow-up permitted the opportunity to examine relationships of serum total and HDL cholesterol to severe retinal outcomes during the course of long-duration type 1 diabetes mellitus, this occurred during a time of great change in the medical care of persons with diabetes mellitus. This may have influenced our findings in that there has been a decreasing risk of progression of diabetic retinopathy to the severe lesions of interest in this article. In addition, treatment of triglyceride levels with fenofibrate in patients with a background of statin use has been found to decrease the risk of progression of diabetic retinopathy.35 It was not possible to test whether there might be synergy of high triglyceride levels with cholesterol because triglycerides were measured only at the fifth WESDR examination; the effect of fenofibrate on the retina in the WESDR could also not be measured because there were too few users of this medication in the cohort. Because of the many differences in lifestyle and health care among participants in these types of studies, it was not possible to reconcile disparate findings. Another limitation of our findings was that this cohort had little racial/ethnic diversity. If race/ethnicity influences the strength of lipid risk factors for PDR and macular edema, our data would be primarily relevant to a white European-derived American population.
We found modest associations between total serum and HDL cholesterol and the prevalence of PDR and macular edema, 2 vision-threatening complications of diabetes mellitus in persons with long-duration type 1 diabetes mellitus. The lack of an observed association of lipids and the incidence of these severe retinal outcomes was compatible with the possibility that these serum lipids are not important in the etiologic classification of PDR and macular edema.
Corresponding Author: Barbara E. K. Klein, MD, MPH, Department of Ophthalmology and Visual Sciences, University of Wisconsin–Madison, 610 N Walnut St, 409 WARF, Madison, WI 53726-2336 (firstname.lastname@example.org).
Submitted for Publication: September 3, 2014; final revision received October 17, 2014; accepted October 22, 2014.
Published Online: December 11, 2014. doi:10.1001/jamaophthalmol.2014.5108.
Author Contributions: Ms Howard had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: B. E. K. Klein.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: B. E. K. Klein.
Critical revision of the manuscript for important intellectual content: Myers,Howard, R. Klein.
Statistical analysis: Myers, Howard.
Obtained funding: B. E. K. Klein, R. Klein.
Study supervision: B. E. K. Klein, R. Klein.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.
Funding/Support: This research was supported by grant EY016379 from the National Institutes of Health (Drs R. Klein and B. E. K. Klein) and an unrestricted grant from Research to Prevent Blindness, New York, New York.
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily reflect the official views of the National Eye Institute or the National Institutes of Health.
Previous Presentation: An abstract of this study was presented at the 2014 Annual Meeting of the Association for Research and Vision in Ophthalmology (ARVO); May 7, 2014; Orlando, Florida.