NDT Advance Access published online on March 27, 2008
Nephrology Dialysis Transplantation, doi:10.1093/ndt/gfn149
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Effects of albuminuria and renal dysfunction on development of dyslipidaemia in type 2 diabetes—the Hong Kong Diabetes Registry
1 Department of Medicine and Therapeutics 2 Hong Kong Institute of Diabetes and Obesity 3 Li Ka Shing Institute of Health Sciences 4 Department of Chemical Pathology, The Chinese University of Hong Kong, Hong Kong, China
Correspondence and offprint requests to: Wing Yee So, 9/F Department of Medicine and Therapeutics, The Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong SAR, China. Tel: +852-2632-3138; Fax: +852-2632-3108; E-mail: wingyeeso{at}cuhk.edu.hk
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and methodsResultsDiscussionReferences |
|---|
Background. It is uncertain whether albuminuria precedes the future development of high total cholesterol (TC > 6.2 mmol/l) and high LDL-C (>4.1 mmol/l) while renal dysfunction precedes the future development of low HDL-C (<0.9 mmol/l) in type 2 diabetes.
Methods. A prospective cohort of 2761 type 2 diabetic patients without significant dyslipidaemia and having at least one measurement of TC, LDL-C and HDL-C during 2.8 years of follow-up was analysed. The spline Cox regression model was used to derive hazard ratio (HR) curves of the spot urinary albumin:creatinine ratio (ACR) and the estimated glomerular filtration rate (eGFR) for dyslipidaemia, followed by standard Cox models to confirm the findings from the HR curves.
Results. Seven percent of the cohort developed high TC, 4.6% developed high LDL-C and 5.7% developed low HDL-C during follow-up. In multivariate analysis, the HR of ACR for high TC and high LDL-C increased rapidly and linearly from zero with no apparent threshold. Patients with macroalbuminuria (ACR 
25 mg/mmol) were, respectively, 1.6- and 2.4 folds more likely to develop high TC and high LDL-C than those with normoalbuminuria at baseline. The HR of eGFR for low HDL-C increased rapidly with declining eGFR at <110 ml/min/ 1.73 m2. Subjects with eGFR <60 ml/min/1.73 m2 and 60–<110 ml/min/1.73 m2, respectively, had 3.0-fold and 1.8-fold risks of low HDL-C compared to those with eGFR 110–<140 ml/min/1.73 m2.
Conclusions. In type 2 diabetes, macroalbumninuria predicts high TC and high LDL-C, while reduced renal function, even within normal range, predicts low HDL-C.
Keywords: albuminuria; Chinese; dyslipidaemia; Hong Kong; renal function
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and methodsResultsDiscussionReferences |
|---|
A large cross-sectional study [1] shows that increasing albuminuria was associated with a graded increase in the likelihood of high total cholesterol (TC) and low-density lipoprotein cholesterol (LDL-C) levels. Although several prospective studies suggest that dyslipidaemia may predict progression of albuminuria and renal dysfunction in type 2 diabetes [2], there are experimental studies suggesting that proteinuria and renal dysfunction per se may promote an atherogenic dyslipidaemia pattern [3], a phenomenon well documented in nephrotic syndrome. Adding to this complexity is the predictive role of albuminuria for end-stage renal disease (ESRD) [4] and the independent effect of reduced renal function on cardiovascular disease (CVD). In-depth analysis suggests that in patients with ESRD, low HDL-C rather than high LDL-C is the predictor for CVD [5]. Small cross-sectional studies have shown that increased serum creatinine is associated with decreased HDL-C, independent of albuminuria [6]. These findings thus call for better understanding of the natural progression of dyslipidaemia and the nature of frequent associations amongst albuminuria, dyslipidaemia and renal dysfunction before embarking upon major interventional studies.
In this analysis, we used a large prospective database to test the hypothesis that albuminuria precedes the future development of high TC and high LDL-C while renal dysfunction precedes the future development of low HDL-C in type 2 diabetes.
|
Subjects and methods |
|---|
|
TopAbstractIntroductionSubjects and methodsResultsDiscussionReferences |
|---|
Patients
The study patients have been described in detail in previous studies [4,7]. Briefly, the Hong Kong Diabetes Registry was established in 1995, at a regional hospital, the Prince of Wales Hospital, that serves a 1.2 million local population. All the patients at enrolment to the registry underwent a 4-h assessment of complications and risk factors on an outpatient basis, which is modified from the European DIABCARE protocol [8]. Patients in the registry would be followed up till death. The study was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong. The study complied with the Declaration of Helsinki and written and informed consent was obtained from all the patients.
All laboratory results on follow-up were retrieved from the Hospital Authority (HA) Central Computer System. For patients who continued to be followed up at the HA hospital clinics or who have been admitted to the HA hospitals after enrolment, laboratory results and details of hospital admissions would be available from the HA Central Computer System using the Hong Kong Identity Card required of all citizens. In addition, causes of death were also obtained from the Hong Kong Death Registry. All dispensary data of drugs of interest from 1 December 1996 to 30 July 2005, including dates of commencement and discontinuation, were extracted from the HA Central Computer System. All medications of patients attending HA hospitals are dispensed on site.
High TC (>6.2 mmol/l), high LDL-C (>4.1 mmol/l) and low HDL-C (<0.9 mmol/l) were used as the three endpoints of the study [1]. Follow-up times were calculated as the number of years from the date of enrolment to the date of the first event of high TC, high LDL-C or low HDL-C during follow-up, or the date when the last measurements of TC, LDL-C and HDL-C were available.
From 1995 to 2005, 7920 diabetic patients were enrolled in this Registry. Since detailed information on drug use became available after 1 December 1996, we limited the analysis to 7387 patients who were enrolled into the registry and had at least one visit registered in this registry after this date. We excluded 323 patients with type 1 diabetes defined as acute presentation with ketoacidosis, heavy ketonuria (>3+) or continuous requirement of insulin within 1 year of diagnosis. Sequentially, five patients with uncertain type 1 diabetes status, 45 non-Chinese patients and 1913 patients who met the definitions of high TC, high LDL-C or low HDL-C at baseline were also excluded. Of the remaining 5101 patients, 2761 patients who had all the three lipid measurements, TC, LDLC and HDL-C during follow-up, were used in the analysis.
Clinical measurements
Clinical measurements including assessment methods, definitions and laboratory assays have been published elsewhere [4,7]. Apart from documentation of demographic data and clinical assessment of complications, fasting blood samples were taken for the measurement of plasma glucose, glycosylated haemoglobin (HbA1c), lipid profile (TC, HDL-C, triglycerides (TG) and calculated LDL-C), renal and liver functions. A sterile, random-spot urine sample was used to measure the albumin:creatinine ratio (ACR). We used the modification of diet in renal disease (MDRD) equation re-calibrated for Chinese [9] to estimate eGFR expressed in ml/min/1.73 m2:
|
|
Statistical analyses
The statistical analysis system, SAS Release 9.10 (SAS Institute Inc., Cary, USA) was used to perform the statistical analysis. Compared to patients without repeat lipid results in the HA system, patients with repeat lipid measurements had poorer metabolic profile at enrolment and were more likely to receive lipid-lowering drugs and had more clinical events (Table 1). Thus, the cohort with repeat lipid data on follow-up over-represented patients with poor metabolic profile at enrolment. To correct for this bias, a propensity score of the likelihood of having lipid measurements on follow-up [10] was calculated using a logistic regression model with an availability of repeat lipid measurements (Yes/No) as the dependent variable and baseline covariates as predictors. The latter included age, sex, HbA1c, waist circumference, mean arterial pressure (MAP), urinary albumin creatinine ratio (ACR), eGFR, smoking status (current/ex), TC, LDL-C, HDL-C, duration of diabetes, sensory neuropathy, retinopathy, peripheral arterial disease, history of coronary heart disease (CHD), heart failure and stroke and drug use (statin, fibrate, other lipid-lowering drugs, ACE inhibitors, angiotensin II receptor blockers and insulin) at baseline. The propensity score model had a moderate model fit (c-statistic: 0.671 and P value for the Hosmer and Lemeshow test: 0.562). Then, stratified Cox regression analysis on decile of the propensity score was used to obtain the hazard ratio (HR) in the group with repeat lipid measurements while adjusting for other covariates and correcting for the over-representation of patients with poorer metabolic profile. By applying this model, patients within each stratum had a similar probability of having follow-up lipid measurements.
|
Cox proportional hazard regression combined with restricted cubic spline (RCS) with four knots (i.e. one term decomposed into three terms: x, x1 and x2) [11] was used to calculate HR curves over the full ranges of baseline risk factors as described before [7,12]. This allows us to check linearity and correctly stratify ACR and eGFR according to their predicting powers for dyslipidaemia. The hazard ratio between two points of variable Xi can be estimated by exp(Y2–Y1), where Y1 and Y2 are the corresponding spline function values of two Xi points. Here, Y1 and Y2 were calculated by the formula: the spline function value of Xi = βx + βx1 + βx2, where β, β1 and β2 were estimated by applying x, x1 and x2 as covariates in Cox models. In this study, the 10th-percentile point for ACR and the 90th-percentile point for eGFR were arbitrarily chosen as the reference points for calculating HR of other points of ACR and eGFR, respectively.
In cohort analysis, it is critically important to adequately adjust for all the potential confounders. To better control for covariates, we directly used the spline function of Xi, i.e. x, x1, x2, to control for potential confounding effects of all continuous covariates. Drug use, especially, statin, fibrate and other lipid-lowering drugs, are important confounders when examining the independent effects of ACR and eGFR on lipid endpoints. We assume that (1) drug effects are related to durations of drug use, i.e. cumulative effects, (2) the cumulative effects need to be adjusted for the total follow-up time and (3) cumulative effects will have diminished since termination of drug use [13]. Therefore, we calculated the proportions of time of drug usage from enrolment to the follow-up date of different lipid endpoints. The calculated time proportions of drug use were further divided by the sum of 1+ the number of years from termination of drug use to the endpoint or censored date. The adjusted time proportions of drug use were used in Cox models to adjust for their potential confounding effects.
Confirmatory Cox regression analyses were performed by recoding ACR and eGFR in categorical variables based on the HR plots versus their full-range values to verify the findings in the HR plots.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and methodsResultsDiscussionReferences |
|---|
Population characteristics
The cohort had a median age of 57 (IQR: 48–68) years and a median duration of diabetes of 7 (IQR: 2–12) years. The follow-up duration was 2.83 (IQR: 1.43–4.78) years for the high TC endpoint, 2.83 (IQR: 1.43–4.77) years for the high LDL-C endpoint and 2.83 (IQR: 1.42–4.77) years for the low HDL-C endpoint. During the follow-ups, 7.9% (n = 219) of the cohort developed high TC, 4.6% (n = 126) developed high LDL-C and 5.7% (n = 156) developed low HDL-C.
Patients who developed high TC and high LDL-C during follow-up had a longer duration of diabetes, higher HbA1c, ACR, TC and LDL-C and were more likely to use statins, fibrate and insulin during follow-up than patients who did not have significant dyslipidaemia on repeat testing. Patients who developed low HDL-C during follow-up were older, had longer disease duration, higher waist circumference, systolic BP, HbA1c and ACR, but lower eGFR, HDL-C and TG at baseline and were more likely to receive statins and ACE inhibitors. These patients were also more likely to develop CHD than those who had repeat HDL-C 0.9 mmol/l (see online appendix tables 1, 2 and 3).
Effects of ACR and eGFR on new onset of dyslipidaemia
After adjusting for covariates, the HRs of ACR for high TC, high LDL-C and low HDL-C increased rapidly and linearly from 0 to 90 mg/mmol (Figure 1a). The increasing HR curves of ACR for high TC and high LDL-C became less steep for values >90 mg/mmol while the curve gradient for low HDL-C remained more or less constant. For the same value of ACR, the HR for high LDL-C was higher than HRs for high TC and low HDL-C (Figure 1a). Patients with micro- and macroalbuminuria were, respectively, 1.1-fold (P > 0.05) and 2.4-fold (P < 0.05) more likely to have high LDL-C than those with normoalbuminuria after controlling for eGFR and other covariates. The HR of macroalbuminuria for low HDL-C was significant after adjusting for age and sex, but the significance did not persist after adjusting for eGFR and other covariates (Table 2).
|
|
After adjusting for covariates, the HR of eGFR for low HDL-C started to rise as eGFR declined from 110 ml/ min/1.73 m2 with the gradient of the curve accelerating from 60 ml/min/1.73 m2 (Figure 1b). Patients with eGFR <60 ml/min/1.73 m2 and those with
60–<110 ml/ min/1.732 were, respectively, 3.0-fold and 1.8-fold more likely to develop low HDL-C than subjects with eGFR 110–<140 ml/min/1.73 m2 after adjusting for ACR and other covariates (Table 2). |
Discussion |
|---|
|
TopAbstractIntroductionSubjects and methodsResultsDiscussionReferences |
|---|
In this prospective analysis, in type 2 diabetic patients without significant dyslipidaemia at baseline, macroalbuminuria was a significant risk factor for the new onset of high LDL-C and high TC, independent of renal function and use of lipid-lowering drugs, while eGFR was a significant risk factor for low HDL-C, independent of albuminuria and use of lipid-lowering drugs. Looked at another way, macroalbuminuria preceded the development of high LDL-C and high TC while reduced eGFR (<110 ml/min/1.73 m2) preceded the development of low HDL-C.
These findings lend support to results from several cross-sectional studies showing the association of albuminuria with the increased risk of high TC and high LDL-C [1,14] and that of renal dysfunction with low HDL-C [6,14]. The third National Health and Nutrition Examination Survey (NHANES III) of USA (1) (n = 17 702) reported that the odds ratio of developing high TC (>6.2 mmol/l) and high LDL-C levels (>4.1 mmol/l) were, respectively, 1.04 and 1.32 for those with microalbuminuria (4.0–36.9 mg/mmol) and 1.96 and 1.69 for those with macroalbuminuria (>39.6 mg/mmol), while albuminuria was not associated with low HDL-C level (<0.9 mmol/l). Two smaller cross-sectional studies examined the relative impacts of albuminuria and renal function on lipid abnormalities [6,14]. Shoji et al. [6] reported that elevated serum creatinine, but not urinary ACR, was an independent factor associated with increased intermediate density lipoprotein cholesterol (IDL-C) and low HDL-C in diabetic patients. In another study [14], eGFR was positively associated with HDL-C, but not with TC or LDL-C. Proteinuria was positively associated with TC and LDL-C and inversely associated with HDL-C.
Traditional factors of CHD such as age, hypertension, hyperglycaemia, smoking and obesity are associated with albuminuria [15]. To date, LDL-C has emerged as one of the strongest predictors for CHD [16]. On the other hand, albuminuria is the most powerful predictor for all-cause and CVD mortality in diabetic patients [17]. In the RENAAL Study [18] which enrolled 1513 type 2 diabetic patients with nephropathy and renal insufficiency, albuminuria reduction was the only predictor for CVD outcome. For every 50% reduction in albuminuria, there was an 18% risk reduction in CVD risk and 27% risk reduction in heart failure. Our findings strongly suggest albuminuria as the key linking factor for the frequent associations of LDL-C and non-traditional risk factors (such as eGFR and inflammatory markers) in the development of CHD in type 2 diabetes.
Moorhead et al. [3] first hypothesized that renal loss of low-molecular-weight protein such as albumin and lipoproteins might promote increased hepatic secretion of atherogenic lipid particles such as very low-density lipoprotein cholesterol (VLDL-C), leading to nephrotoxicity. Reduced GFR may lead to peripheral hyperinsulinaemia and insulin resistance that reduce the lipoprotein lipase activity, thus resulting in reduced catabolism of VLDL-C and chylomicrons. TG-rich lipid particles promote the formation of small dense LDL and small dense HDL particles. Small dense HDL particles are prone to catabolism, leading to reduced cholesterol efflux as well as attenuated antioxidant and anti-inflammatory capacity of the altered HDL particles [19]. Against this background, the novelty of our analysis lies in the quantification of the inter-relationships between dyslipidaemia and renal function and the independent determinants of different lipid parameters. Our findings emphasize the critical importance of renal dysfunction and abnormal lipid metabolism for CVD risk in type 2 diabetes. It has been pointed out that non-traditional risk factors other than low HDL-C and high LDL-C may explain the markedly increased CHD risk in ESRD [14]. With the onset of mild to moderate renal dysfunction, there are progressive changes in metabolic milieu including increased oxidative stress and microinflammation that may increase CVD risk independent of lipid abnormalities [20]. A recent trial failed to reduce CHD rates in patients with established renal diseases by means of using atorvastatin to lower LDL-C [21]. In a similar vein, low HDL-C reflects a more generalized disturbance in metabolism that collectively contributes to increased CVD risk.
The study had several limitations. First, there were a relatively large number of subjects who did not have lipid measurements available in our system. However, the use of propensity score in our multivariate models should largely correct the potential bias due to over-representation of patients with poorer metabolic profile at enrolment in the present prospective cohort with repeat lipid measurements. Second, only one urine sample was considered to classify a patient as normo-, micro- or macroalbuminuric, and the high variability of ACR measurement is well recognized. Third, the findings were derived from Chinese patients with type 2 diabetes and cannot be readily extrapolated to other populations.
In conclusion, using a prospective study design and a careful adjusting scheme for confounders, we confirmed that increased albuminuria is a risk factor for high TC and high LDL-C without apparent thresholds, while reduced eGFR, commencing from 110 ml/min/1.73 m2, is associated with the increased risk of low HDL-C. Our findings provide further evidence supporting the critical importance of reducing albuminuria and preserving renal function to attenuate CVD risks in type 2 diabetes.
Supplementary material is available at NDT Journal online.
|
Acknowledgments |
|---|
This study was supported by the Hong Kong Foundation for Research and Development in Diabetes, established under the auspices of the Chinese University of Hong Kong.
Conflict of interest statement. None declared.
|
References |
|---|
|
TopAbstractIntroductionSubjects and methodsResultsDiscussionReferences |
|---|
- Shankar A, Klein R, Moss SE, et al. The relationship between albuminuria and hypercholesterolemia. J Nephrol (2004) 17:658–665.[Web of Science][Medline]
- Ravid M, Brosh D, Ravid-Safran D, et al. Main risk factors for nephropathy in type 2 diabetes mellitus are plasma cholesterol levels, mean blood pressure, and hyperglycemia. Arch Intern Med (1998) 158:998–1004.
[Abstract/Free Full Text] - Moorhead JF, Chan MK, El-Nahas M, et al. Lipid nephrotoxicity in chronic progressive glomerular and tubulo-interstitial disease. Lancet (1982) 2:1309–1311.[Web of Science][Medline]
- Yang XL, So WY, Kong AP, et al. End-stage renal disease risk equations for Hong Kong Chinese patients with type 2 diabetes: Hong Kong Diabetes Registry. Diabetologia (2006) 49:2299–2308.[CrossRef][Web of Science][Medline]
- Kaysen GA. Dyslipidemia in chronic kidney disease: causes and consequences. Kidney Int (2006) 70:S55–S58.
- Shoji T, Emoto M, Kawagishi T, et al. Atherogenic lipoprotein changes in diabetic nephropathy. Atherosclerosis (2001) 156:425–433.[CrossRef][Medline]
- Yang X, Ma RC, So WY, et al. Impacts of chronic kidney disease and albuminuria on associations between coronary heart disease and its traditional risk factors in type 2 diabetic patients—the Hong Kong diabetes registry. Cardiovasc Diabetol (2007) 6:37.[CrossRef][Medline]
- Piwernetz K, Home PD, Snorgaard O, et al. For the DiabCare Monitoring Group of the St. Vincent Declaration Steering Committee. Monitoring the targets of the St. Vincent declaration and the implementation of quality management in diabetes care: the DiabCare initiative. Diab Med (1993) 10:371–377.[Web of Science][Medline]
- Ma YC, Zuo L, Chen JH, et al. Modified glomerular filtration rate estimating equation for Chinese patients with chronic kidney disease. J Am Soc Nephrol (2006) 17:2937–2944.
[Abstract/Free Full Text] - Joffe MM, Rosenbaum PR. Invited commentary: propensity scores. Am J Epidemiol (1999) 150:327–333.
[Abstract/Free Full Text] - Harrell F. Regression Modelling Strategies: With Applications to Linear Models, Logistic Regression, and Survival Analysis. (2001) New York: Spinger.
- So WY, Yang X, Ma RC, et al. Risk factors in V-shaped risk associations with all-cause mortality in type 2 diabetes—The Hong Kong Diabetes Registry. In: Diabetes Metab Res Rev (2008) 24:238–246.[CrossRef][Medline]
- Yang X, Kong AP, So WY, et al. Effects of chronic hyperglycaemia on incident stroke in Hong Kong Chinese patients with type 2 diabetes. Diabetes Metab Res Rev (2007) 23:220–226.[CrossRef][Web of Science][Medline]
- Sarnak MJ, Coronado BE, Greene T, et al. Cardiovascular disease risk factors in chronic renal insufficiency. Clin Nephrol (2002) 57:327–335.[Web of Science][Medline]
- Parving HH, Lewis JB, Ravid M, et al. Prevalence and risk factors for microalbuminuria in a referred cohort of type II diabetic patients: a global perspective. Kidney Int (2006) 69:2057–2063.[CrossRef][Web of Science][Medline]
- Genser B, Marz W. Low density lipoprotein cholesterol, statins and cardiovascular events: a meta-analysis. Clin Res Cardiol (2006) 95:393–404.[CrossRef][Web of Science][Medline]
- Ko GT, So WY, Chan NN, et al. Prediction of cardiovascular and total mortality in Chinese type 2 diabetic patients by the WHO definition for the metabolic syndrome. Diab Obes Metab (2006) 8:94–104.[CrossRef]
- de Zeeuw D, Remuzzi G, Parving HH, et al. Albuminuria, a therapeutic target for cardiovascular protection in type 2 diabetic patients with nephropathy. Circulation (2004) 110:921–927.
[Abstract/Free Full Text] - Kaysen GA, Eiserich JP. The role of oxidative stress-altered lipoprotein structure and function and microinflammation on cardiovascular risk in patients with minor renal dysfunction. J Am Soc Nephrol (2004) 15:538–548.
[Abstract/Free Full Text] - Sarnak MJ, Levey AS, Schoolwerth AC, et al. Kidney disease as a risk factor for development of cardiovascular disease: a statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Circulation (2003) 108:2154–2169.
[Free Full Text] - Wanner C, Krane V, Marz W, et al. Atorvastatin in patients with type 2 diabetes mellitus undergoing hemodialysis. N Engl J Med (2005) 353:238–248.
[Abstract/Free Full Text]
Accepted in revised form: 25. 2.08
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  G. T.C. Ko, W.-Y. So, P. C. Tong, W.-B. Chan, X. Yang, R. C. Ma, A. P. Kong, R. Ozaki, C.-Y. Yeung, C.-C. Chow, et al. Effect of interactions between C peptide levels and insulin treatment on clinical outcomes among patients with type 2 diabetes mellitus Can. Med. Assoc. J., April 28, 2009; 180(9): 919 - 926. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||