Low HDL Cholesterol and the Risk of Diabetic Nephropathy and Retinopathy: Results of the ADVANCE study

The study design of the ADVANCE study is reported in detail elsewhere (24). In brief, 11,140 patients with type 2 diabetes, aged at least 55 years at study entry and with at least one other cardiovascular risk factor, underwent factorial randomization to 1) the fixed combination of perindopril and indapamide (4 mg/1.25 mg) or matching placebo and 2) intensive (gliclazide MR-based, to an HbA_1c target of ≤6.5%) or standard (usual care) glucose control. Patients were followed-up for a median of 5.0 years. The ADVANCE trial enrolled subjects from 20 countries in Asia, Australasia, Europe, and North America. A total of 11,126 patients with HDL-C measurements at baseline were included in these analyses. From this population, 1,602 patients were included in the ADVANCE Retinal Measurement (AdRem) study, which involved serial retinal photography (1,241 had assessable images). Study design and the methods used in AdRem are published in detail elsewhere (25). Approval for the study was obtained from each center’s institutional ethics committee, and all participants provided written informed consent.

At baseline, venous blood was taken for a fasting lipid profile, HbA_1c, and creatinine. Blood pressure and BMI also were recorded. Creatinine was remeasured at 4 months and then annually. In addition, urine samples were collected at baseline, 24 months, 48 months, and at the end of the trial for determination of the urinary albumin-to-creatinine ratio (ACR). Fundoscopy also was performed at these times. HDL-C level was repeated at 24 and 48 months or at the end of the study. All samples were analyzed by local laboratories, and all assessments could be repeated at physician discretion. All nonstudy medication was at the discretion of the local physician. Participants in AdRem had retinal stereoscopic photographs taken after the ADVANCE trial randomization visit and at the final visit of the blood pressure arm of the trial. All images were graded centrally according to the Early Treatment of Diabetic Retinopathy Study classification as modified in the UK Prospective Diabetes Study (26).

The main outcomes for this analysis were microvascular events, defined as a composite of total renal and retinal events. Total renal events were defined as the development of new microalbuminuria (urinary ACR 30–300 μg/mg), new macroalbuminuria (urinary ACR of >300 μg/mg), a doubling of creatinine to at least 200 μmol/L, the need for renal replacement therapy, or death as a result of renal disease. Total retinal events were defined as the development of proliferative retinopathy (new blood vessels or fibrous proliferations on the disc or elsewhere, vitreous hemorrhage, or preretinal hemorrhage), macular edema, diabetes-related blindness, or the use of retinal photocoagulation therapy. Participants could record more than one event relating to renal and retinal disease, but only the first event was analyzed for total microvascular events.

Differences in variables at baseline between subgroups of the study population were tested using the Student t test, the Mann-Whitney test, or the χ² test, as appropriate. Participants were censored at their date of death or, for those still alive at the end of follow-up, the date of their last visit. The regression dilution bias in HDL-C was assessed using a linear mixed model, with HDL-C during follow-up as the outcome and baseline HDL-C as the predictor. To calculate the correction factor, HDL-C measurements after microvascular events were excluded (27). The risks of events associated with baseline HDL-C level were estimated using Cox proportional hazards models, with adjustment for potential confounding baseline covariates including age (continuous), sex (male/female), ethnicity (white/Asian/other), treatment groups (standard vs. intensive glucose control and placebo vs. fixed-dose blood pressure–lowering treatment), history of microvascular disease (yes/no), smoking status (current/previous/never), current alcohol consumption (yes/no), HbA_1c (continuous), BMI (continuous), systolic blood pressure (continuous), diabetes duration (continuous), total cholesterol (continuous), creatinine (continuous), and statin use (yes/no). The selection of variables was based on identifying all measured clinical variables of known or suspected prognostic importance for the outcomes of interest. The assumption of the proportional hazards was checked graphically using the log cumulative hazard plot for all variables included in the Cox model. The primary analyses compared patients in the highest with those in the lowest third (or tertile group) of HDL-C. Additional sensitivity analyses were performed comparing the risks in those in the lowest with those in the highest fourths (or quartile groups), excluding patients with microvascular disease, renal disease, or retinal disease at baseline and examining subgroups defined by age, sex, ethnicity, systolic blood pressure, BMI, and age. All P values were calculated from two-tailed tests of statistical significance with a type I error rate of 5%. All analyses were performed using SAS version 9.1 (SAS Institute, Cary, NC).

The ADVANCE study enrolled patients with type 2 diabetes at high risk for macrovascular events. At baseline, 10% (n = 1,155) had evidence of microvascular disease. During a median follow-up period of 5 years, 32% (n = 3,585) of participants developed new or worsening microvascular disease. Almost one-third (n = 3,161 [28%]) developed a renal event: microalbuminuria was the most frequent event, followed by macroalbuminuria, doubling of creatinine to at least 200 μmol/L, renal-related death, and the need for renal replacement therapy (25, 3.4, 1.2, 0.3, and 0.2%, respectively). A retinal event occurred in 6% (n = 680): the need for laser therapy was the most frequent event, followed by macular edema, proliferative retinopathy, and diabetes-related blindness (3.5, 3.2, 2.6, and 0.4%, respectively).

The mean baseline HDL-C level was 1.3 mmol/L (SD 0.45 mmol/L [range 0.1–4.0]). Table 1 shows the distribution of variables by thirds of baseline HDL-C level. Compared with the lowest third of HDL-C, those subjects in the highest third were less likely to be male (44.8 vs. 70.8%) or taking a statin (24.9 vs. 31.1%) and more likely to have lower BMI (27.7 vs. 29.1 kg/m²), lower serum creatinine (82.5 vs. 91.0 µmol/L), and to have never smoked (65.3 vs. 49.6%). Low HDL-C also was associated with higher triglycerides and lower total cholesterol and LDL cholesterol. Age, diabetes duration, systolic blood pressure, HbA_1c, ethnicity, and nonstatin lipid-lowering medication use all were statistically significantly different between thirds; however, the absolute differences were very small. The relative differences between HDL groups and the baseline variables of statin use, nonstatin lipid-lowering medication use, systolic blood pressure, and HbA_1c remained stable over the course of study follow-up (Supplementary Table 1).

Table 2 shows the association, both adjusted and unadjusted, between baseline thirds of HDL-C and the risk of developing new or worsening microvascular disease. Compared with the highest third of HDL-C, those in the lowest third had an 11% higher risk of a microvascular event (unadjusted hazard ratio [HR] 1.11 [95% CI 1.02–1.20], P = 0.01). Following multivariable adjustment and taking into account regression dilution, the risk was 17% higher (1.17 [1.06–1.28], P = 0.001). This finding was driven by an adjusted 19% higher risk of a renal event (1.19 [1.08–1.32], P = 0.0005). There was a similar significantly higher risk of developing new microalbuminuria and macroalbuminuria (1.14 [1.03–1.27], P = 0.01, and 1.42 [1.07–1.87], P = 0.01, respectively). Furthermore, we observed that those patients in the lowest third of HDL-C were more likely to maintain or progress to a worse category of urinary ACR over time, compared with those in the middle and highest third of HDL-C (66.8 vs. 62.0 and 58.8%, respectively) (Supplementary Table 2).

The direction of the effect was similar, however not statistically significant following adjustment, to the other renal outcomes of doubling of creatinine to at least 200 μmol/L and renal-related death. In contrast to the other renal outcomes, the association with the need for renal replacement therapy (the end point with the smallest number of events) was nonlinear.

There was no association between baseline thirds of HDL-C and the development of retinopathy or any specific type of retinal event. Furthermore, there was no association between baseline HDL-C and a wide range of predefined retinal outcomes from the AdRem substudy involving 1,241 participants who underwent assessable serial retinal photography, which included progression by the Early Treatment of Diabetic Retinopathy Study classification and the development of the individual signs of retinopathy (data not shown).

Sensitivity analyses by fourths (or quartile groups) of HDL-C revealed similar findings, namely an inverse relationship between HDL-C level and total microvascular events that was driven primarily by renal events, with no relationship found between HDL-C and retinal events. Sensitivity analyses excluding patients with baseline microvascular disease also were similar to the main analysis. When stratified by subgroups defined by age, sex, ethnicity, systolic blood pressure, BMI, or HbA_1c, there was no evidence of heterogeneity in the association.

This study is the largest prospective analysis specifically addressing HDL-C level and risk of microvascular disease in patients with type 2 diabetes. The main finding is that lower baseline HDL-C level is a significant and independent predictor of the development and progression of diabetic nephropathy. Compared with patients in the highest third, those in the lowest third of baseline HDL-C had a 19% higher risk of nephropathy, after adjustment for a wide range of potential confounders and accounting for regression dilution. In contrast, there was no association between baseline HDL-C and the risk of diabetic retinopathy. These findings suggest that differences exist in the pathophysiology between the two types of microvascular disease.

Our findings provide the strongest evidence to date for a role of HDL-C in the development and progression of diabetic nephropathy in patients with type 2 diabetes. Several small prospective studies have shown that low HDL-C predicts progression of microalbuminuria (8–10); however, others have not identified such an association (28,29). The largest previous study evaluating this question included 2,193 patients with type 2 diabetes and normal renal function at baseline. In this study, each 0.26 mmol/L higher level of HDL was associated with a 24% lower risk of developing stage 3 chronic kidney disease (11). This was not a prespecified end point of the ADVANCE study; however, we are the first to show an inverse relationship between baseline HDL-C and total renal microvascular events.

We observed no significant association between HDL-C and a whole spectrum of retinal complications. Although some non-HDL lipid fractions previously have been reported to be associated with the severity and progression of retinopathy in patients with type 2 diabetes, no study has shown an association with HDL-C (16,20), and our study is concordant with this literature. Our findings may be explained by the nature of the retinal events examined as the most frequent retinal end point was laser photocoagulation. Because this procedural event is likely to be driven by health service availability and local usual practice in addition to biological risk in a multicenter, international study such as this, its inclusion may have diluted any causal link, if any such association existed. However, we also found no evidence of an association with any of the specific end points examined in the small proportion of ADVANCE trial participants who underwent serial retinal photography, a more sensitive and objective measure of retinal disease.

The Fenofibrate Intervention in End Point Lowering in Diabetes Trial showed that fenofibrate reduced the development of renal dysfunction and the need for retinal laser therapy (22,30). In support of our findings of an association between HDL-C and renal events but not retinal events, fenofibrate seemed to be more protective against the estimated glomerular filtration rate loss in those with baseline dyslipidemia (defined as low HDL-C and high triglycerides) (30), whereas the retinal benefit was independent of any lipid fraction (22). In much the same way, the more recent Action to Control Cardiovascular Risk in Diabetes eye study showed a benefit of fenofibrate added to statin therapy on the progression of diabetic retinopathy, without an appreciable change in HDL-C level (HDL-C at 1 year 1.03 mmol/L in the fenofibrate group compared with 1.01 mmol/L in the placebo group) (23). The salient conclusion from these two studies of fenofibrate therapy is that treatment reduced both microvascular complications in type 2 diabetes. This effect, at least in part, may be related to HDL-C level for renal but not retinal complications. On the horizon, there is the promising cholesterylester transfer protein inhibitor, anacetrapib, which increases HDL-C by 138% (31). Our findings suggest that investigating the effect of HDL-C–modulating agents on the development of diabetic nephropathy is warranted.

The pathogenesis of diabetic microvascular disease is complex and involves the interplay of endothelial dysfunction, advanced end-glycation products, oxidative stress, and the abnormal production of cytokines and growth factors (32,33). The pathobiology of diabetic nephropathy and retinopathy is heterogeneous, with no single characteristic lesion (33,34). Although retinopathy and nephropathy can coexist in the same patient, the association is less clear in type 2 rather than type 1 diabetes (35), perhaps an indication of differing pathophysiology. HDL-C itself is a complex of several lipoproteins, cholesterol, and triglycerides and has several postulated mechanisms of vascular action, notably reverse cholesterol transport as well as anti-inflammatory and antioxidant properties (36). Furthermore, there is emerging evidence that HDL-C improves glycemic control by modulating glucose uptake into skeletal muscle (37) and protects against islet β-cell dysfunction (38). Although the exact putative mechanism for protection is unknown, it is clear that HDL-C exerts beneficial effects on many of the pathways known to be detrimental to vascular and kidney biology, but how this could be different in retinopathy is unclear.

Following the development of microalbuminuria, 20–40% of patients with type 2 diabetes will progress to overt nephropathy (1). Many randomized trials have shown a clear benefit for the treatment of this asymptomatic condition; thus, early detection is critical (39). International guidelines recommend screening at the time of diagnosis of diabetes and thereafter annually for patients with normoalbuminuria (1). The optimum timing of creatinine clearance measurement is unknown. The use of HDL-C level, along with other established risk factors, may help identify patients at high risk of development and progression of nephropathy and therefore warrant more frequent testing.

The strengths of this study are its large sample size, with an ethnically diverse population and rigorous collection of data. This allowed precise estimation of the independent effect of HDL-C on the development and progression of predefined diabetic microvascular disease outcomes. Our study also has a number of limitations. Because it is a post hoc observational analysis of data from a randomized controlled trial, the results should be interpreted as hypothesis generating. Despite the size of this study, there were very few events for some of the end points such as the need for renal replacement therapy and renal death, which limited our ability to provide accurate estimates of the associations for these outcomes. Although the difference was small (<2.7%), adherence to the study ACE inhibitor (perindopril) was slightly lower in the lowest HDL-C third compared with the middle and highest third (77.1, 79.2, and 79.8%, respectively). We cannot exclude that this may have had an effect on renal outcomes. In addition, it is possible that not all subtypes of HDL-C have the same relationship with microvascular outcomes. However, we were unable to measure subtype and function of HDL-C in order to explore these relationships in our study.

In conclusion, in a large population of patients with type 2 diabetes and after adjustment for a wide variety of confounders, low HDL-C level was shown to be an independent risk factor for the development and progression of diabetic nephropathy. Measurement of this lipid fraction may be useful in tailoring screening and therapeutic strategies. Additional research is needed to explore the possible benefits of therapies that increase HDL-C in patients with type 2 diabetes.

The ADVANCE study was funded by grants from Servier (the major financial sponsor) and from the National Health and Medical Research Council of Australia (211086 and 358395). J.C. has received research grants from Servier, administered through the University of Sydney. J.C., A.P., M.W., and S.Z. have received honoraria from Servier for speaking at scientific meetings. No other potential conflicts of interest relevant to this article were reported.

J.M. holds a postgraduate scholarship from the National Health and Medical Research Council (NHMRC) (598402) and Heart Foundation (PC 09S 4757). A.P. holds a senior research fellowship from the NHMRC (632938). S.Z. holds a career development award from the Heart Foundation (CR 10S 5330).

J.M. researched data and wrote the manuscript. S.Z., A.A.P., J.C., M.W., D.S.C., P.G., and M.K.C.N. contributed to the discussion and reviewed and edited the manuscript. Q.L., J.W.J.B., and R.P.S. researched data and contributed to the discussion. S.Z. is the guarantor of this work and, as such, 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.

This research was presented in abstract form at the European Conference of Cardiology (ESC) Congress, Paris, France, 27 August 2011. All members of the ADVANCE Collaborating Group have been listed in full previously (40).