Higher Levels of HDL Cholesterol Are Associated With a Decreased Likelihood of Albuminuria in Patients With Long-Standing Type 1 Diabetes

Data on the presence of diabetic nephropathy, lipid levels, and other characteristics were obtained by chart review of patients seen in the routine office practice of one of the authors (M.E.M.). Charts of all patients with type 1 diabetes were reviewed and selected solely on the basis of a self-reported duration of diabetes of ≥20 years. Patients were designated as having type 1 diabetes on the basis of a history of diabetic ketoacidosis or the need to begin insulin therapy within 1 year of diagnosis.

Patients had urine albumin levels assessed on an annual basis, initially as part of timed 24-h urine collections but since about 1990 by measurement of urine albumin-to-creatinine ratios, and these were not required to be first- or second-voided morning specimens. Patients were designated as having albuminuria if they had a 24-h urine measurement >30 mg or an albumin-to-creatinine ratio >30 mg/g creatinine on more than one occasion (2). Lipid profiles were also generally done yearly, including measurements of total, HDL, and LDL cholesterol and triglyceride levels. Unless the patient had elevated triglyceride levels, these blood samples for lipid measurements were usually not fasting specimens. The levels of these lipid parameters and HbA_1c (A1C) used in this study were obtained before starting any hypolipidemic therapy. Albuminuria characterization was the last obtained before starting any hypolipidemic therapy or any ACE inhibitor treatment and were within 1 year of the lipid and A1C measurements. Angiotensin II receptor blockers were not used in these patients. Urine albumin and creatinine, serum lipids, and plasma A1C were performed by routine assays in the Northwestern Memorial Hospital Clinical Pathology Laboratory. Assessments of diabetic retinopathy were performed at least annually by ophthalmologists in most patients and were reported in the office records. The ophthalmic assessments reported here were within 1 year of the lipid, A1C, and urine microalbumin assessments.

Descriptive data are reported as means ± SD or proportions. For continuous and categorical variables, statistical comparisons were made using t tests and χ² tests, respectively. A series of logistic regression models with varying adjustment were used to calculate the odds of having albuminuria per SD higher HDL: model 1, unadjusted; model 2, adjusted for age and sex; and model 3, model 2 plus duration of diabetes and A1C. HDL cholesterol levels from this patient sample were also compared against an external referent population of people (with and without diabetes) from the National Health and Nutrition Examination Survey (NHANES) 1999–2002 (32). In a second set of logistic regression models, patients with HDL cholesterol levels 1 SD above the sex-specific NHANES population mean (men 1.53 mmol/l, women 1.84 mmol/l) were classified as having high HDL cholesterol. SAS version 9.0 (SAS Institute, Cary, NC) was used for all analyses. Statistical significance was denoted at P < 0.05.

Chart review identified 107 patients with type 1 diabetes who had diabetes of at least 20 years duration (Table 1). Of this group, 42 were identified who had increased urinary albumin excretion (designated as albuminuria); 28 had microalbuminuria (30–300 mg/g creatinine) and 14 had macroalbuminuria (>300 mg/g creatinine). Nine with macroalbuminuria had normal GFR and five had decreased GFRs. Sixty-five patients had normal urinary albumin excretion and GFRs. The mean age of the patients was 43.5 ± 10 years and the duration of diabetes was 30.8 ± 7.1 years; the two groups were virtually identical with respect to age and duration of diabetes. Approximately two-thirds of the patients were women, and this distribution did not differ between the two groups. The mean A1C was 7.9 ± 1.6% for the total population, and the group with albuminuria had a significantly higher level than those without albuminuria (8.5 ± 1.7 vs. 7.5 ± 1.4%, P < 0.01). The proportions with diabetic retinopathy also differed by group. Proliferative retinopathy was more common in individuals with albuminuria, whereas no or background retinopathy was more common in individuals without albuminuria. Sixteen patients (15%) had neither retinopathy nor albuminuria.

The total cholesterol level in these patients was 5.08 ± 1.07 mmol/l, with no difference between the two groups. Although LDL cholesterol and triglyceride levels in those with albuminuria were higher than in those without albuminuria (Fig. 1), these differences were not statistically significant (LDL cholesterol, P = 0.26; triglycerides, P = 0.19). In contrast, A1C was higher and HDL cholesterol levels were significantly lower in those who had albuminuria compared with those that did not have albuminuria (Fig. 1). By univariate analysis, a 0.26-mmol/l (10-mg/dl) increase in HDL cholesterol was associated with an unadjusted odds ratio (OR) of 0.74 (95% CI 0.59–0.93) for the development of albuminuria and an OR of 0.70 (0.54–0.90) after adjustment for age and sex.

In a multivariable model that included HDL cholesterol, A1C, age, sex, and diabetes duration, both A1C and HDL cholesterol were associated with a lower likelihood of having albuminuria (Table 2). For every 0.54-mmol/l (21-mg/dl) increase in HDL cholesterol, patients are half as likely to have albuminuria, even after controlling for A1C.

Categorical analysis showed that patients with HDL cholesterol levels >1 SD above the NHANES population mean are 74% less likely (OR 0.26 [95% CI 0.10–0.72]) to have albuminuria independent of the A1C levels. Sixteen percent of those with albuminuria and 46% of those without albuminuria had HDL cholesterol values >1 SD above the NHANES population mean.

Microalbuminuria is the earliest clinical indicator of diabetic nephropathy, and 25–80% of those with type 1 diabetes and microalbuminuria go on to progress to higher rates of urinary albumin excretion associated with progressive kidney disease (33). Most long-term epidemiologic studies have shown that generally only ∼30–40% of patients with type 1 diabetes will ever develop diabetic kidney disease even with poor diabetes control (8,34,35), with the peak incidence being reached by ∼20 years duration of diabetes. This information, along with the known familial clustering of diabetic kidney disease in those with type 1 diabetes (36–38), suggests that there must be some additional genetic or other factor(s) that either predispose individuals to this complication or, alternatively, protect some individuals from this complication (39–42).

In this study, we selected a time point of 20 years duration of diabetes for inclusion of patients in this analysis, a time at which the annual incidence of new development of microalbuminuria has already decreased markedly, and those who do not have microalbuminuria by this point are unlikely to ever develop it (8,34,35). In fact, the mean duration of diabetes in the two groups was 30 years. We have shown that elevated HDL cholesterol levels may indicate a partial protection from the development of nephropathy.

Although most other studies that have examined the effects of lipid levels on the development of nephropathy have focused on LDL cholesterol levels, several have actually shown that those individuals without nephropathy have HDL cholesterol levels higher than those with nephropathy (15,16,18,19,26,28). In the cohort of 400 patients with type 1 diabetes of >50 years duration studied by Bain et al. (43), the mean HDL cholesterol level was remarkably high at 1.84 mmol/l, even though 35.7% of them had elevated urinary albumin excretion. They did not report levels in those with and without albuminuria.

Our patients with long-term disease and no microalbuminuria also had less severe retinopathy, only 23% of them having proliferative disease compared with 81% of those with albuminuria. This clustering of microvascular complications has been demonstrated by a number of studies previously (6,7).

As indicated above, clustering of coronary artery disease with nephropathy has been shown previously in patients with type 1 and type 2 diabetes (9–12). Furthermore, a protective effect of HDL cholesterol levels on coronary artery disease has been known for many years (30,31), but its protective effect on other complications of diabetes has not been remarked upon previously. Whether protective effects of high HDL cholesterol levels on coronary artery disease and nephropathy will correlate with each other can only be determined by long-term studies. It is thought that HDL is not only involved in reverse cholesterol transport but also may have a number of other beneficial effects on the vascular endothelium (31). Therefore, it is possible these additional mechanisms may be the means by which high HDL levels may have a salutary effect on the glomerulus.

We have shown in this study that higher HDL cholesterol levels may be protective against the development of albuminuria in patients with type 1 diabetes. Whether this is due to the HDL cholesterol levels or whether they serve as a marker for some other mechanism remains to be determined.

This study was presented in part at the Scientific Sessions of the 65th Annual Meeting of the American Diabetes Association, San Diego, California, 10–14 June 2005 (44).