Genetic Variation at the ACE Gene Is Associated With Persistent Microalbuminuria and Severe Nephropathy in Type 1 Diabetes

RESEARCH DESIGN AND METHODS

The DCCT study design (21) is described in the online appendix (available at http://diabetes.diabetesjournals.org).

Phenotypic characterization of nephropathy.

Two nephropathy outcomes, persistent microalbuminuria and severe nephropathy, were defined using repeated 4-h AER measurements obtained yearly during the DCCT and every 2nd year during EDIC. Persistent microalbuminuria was defined as an AER ≥20.8 μg/min on two consecutive assessments. The definition of severe nephropathy is also based on repeated measures of AER and represents the progression from persistent microalbuminuria to an AER ≥208 μg/min or renal replacement therapy (dialysis or transplant). Time to outcome development or censoring was determined as number of visit years from DCCT baseline up to and including the 8th year of EDIC follow-up (2001); mean (±SD) follow-up from DCCT entry was 13.5 ± 2.6 years. Subjects meeting the criteria for persistent microalbuminuria at DCCT baseline and DCCT year 1 (n = 66) were excluded from subsequent analyses of that outcome. Due to the DCCT exclusion criteria, no subjects had severe nephropathy at baseline.

Determination of ACE genotypes.

Genetic variation at the ACE gene has been extensively studied, and haplotypes have been determined (15,22). Fifteen haplotypes were identified by measured haplotype analysis in families, but because of extensive linkage disequilibrium between markers, only four common haplotypes had frequencies >5% (15). In addition to the insertion/deletion polymorphism, we genotyped two SNPs that allowed us to capture the four most common haplotypes occurring with frequency >5% in Caucasian populations. The intermarker distance between the 5′ SNP, rs1800764 (T7715C) (15), and the insertion/deletion is 15.4 kb and between the insertion/deletion and the 3′ SNP, rs9896208 (T33569C) (15), is 10.4 kb (see online appendix for genotyping protocol).

Statistical analysis.

The primary analysis was to evaluate the association between each of the ACE markers and the risk of complications, taking into account the DCCT/EDIC study design and potential confounding factors, including all physiological traits known to be associated with variation in risk of renal complications. The relative risks associated with different marker genotypes were estimated by the hazard ratio (HR) in the Cox proportional hazards model, with and without including explanatory covariates in the model (see online appendix for further description of analysis).

RESULTS

Examination of the distribution of baseline covariates among genotype groups (Table 1) revealed a significant association between BMI and two of the ACE variants, as well as a weak association between triglyceride level and the insertion/deletion genotype. A significant association (P = 0.03) was also present between use of ACE inhibitors during EDIC and the genotype at rs1800764 (Table 1). Due to the clinical correlation of the development of microalbuminuria and the subsequent institution of an ACE inhibitor, we repeated this analysis after exclusion of the individuals who received an ACE inhibitor after the development of persistent microalbuminuria. There was no significant association of ACE inhibitor use with the rs1800764 genotype in subjects with persistent microalbuminuria (P = 0.30) or severe nephropathy (P = 0.07). There was also no association of non-ACE inhibitor antihypertensive medications with any of the markers (P = 0.59). As expected, many of these baseline covariates were significantly associated with persistent microalbuminuria and severe nephropathy (online appendix Tables A1 and A2).

During follow-up between DCCT entry and EDIC year 8, 312 subjects developed persistent microalbuminuria (of which 66 had events at DCCT baseline and were excluded from analyses). Of these individuals, 115 also developed a more severe renal phenotype, including clinical grade proteinuria with an AER >208 μg/min (n = 114) and/or a renal end point (n = 12 patients with either dialysis or transplant).

The quality characteristics of the genotype data were assessed by replicate Centre d’Etude du Polymorphisme Humain (CEPH) control samples, and of 498 replicate genotypes at the three markers, only two genotyping errors occurred corresponding to an observed genotyping error rate of 0.4%. The minor allele frequencies for rs1800764, insertion/deletion, and rs9896208 were 0.45, 0.47, and 0.37, respectively. No significant deviations from Hardy-Weinberg equilibrium were observed.

Proportional hazards analysis of the ACE polymorphisms adjusted for DCCT year of entry, cohort, and treatment strata indicate that genotype variation at each of the markers was significantly associated with the risk of developing persistent microalbuminuria (Table 2). Specifically, as depicted in Fig. 1, individuals with either the T/T, I/I, or C/C genotypes at rs1800764, insertion/deletion, and rs9896208, respectively, had a lower risk of progression to persistent microalbuminuria during a maximum of 17 years of follow-up from DCCT baseline. Individuals with the C/C genotype at rs9896208 had a significantly lower risk of development of persistent microalbuminuria (HR 0.70 [95% CI 0.53–0.93], P = 0.015) (Fig. 1 and Table 2) and severe nephropathy (0.60 [0.39–0.92], P = 0.020) (online appendix Fig. A1 and Table A2). Genotypes at rs1800764 and insertion/deletion, however, were not significantly (P > 0.05) associated with severe nephropathy in analyses that did not take any of the DCCT baseline physiological trait variables into account (Table 2).

Three-locus haplotypes were estimated using a Bayesian reconstruction algorithm implemented in the PHASE (version 2.1) software (see online appendix for haplotype construction). In the DCCT/EDIC cohort, 92% of chromosomes had one of the four common haplotypes previously reported. In univariate analysis, individuals homozygous for the TIC haplotype (which corresponded to the T, insertion, and C alleles at the three markers, rs1800764, insertion/deletion, and rs9896208, respectively) had a lower risk of persistent microalbuminuria and severe nephropathy when compared with individuals with the most common CDT/TIC haplotype pair (Table 2).

When we adjusted for the additional covariates (Table 3 legend), the association of each of the three ACE variants with the development of persistent microalbuminuria remained significant (Table 3). Values of R² indicating the proportion of explained variability in survival times to persistent microalbuminuria were calculated as 0.8, 0.6, and 0.8% for rs1800764, insertion/deletion, and rs9896208 markers, respectively. For severe nephropathy, adjustment for covariates in the multivariate model strengthened the genotype associations (Table 3). At rs1800764, individuals with the T/T genotype had a lower risk than C/T heterozygotes. Individuals with the I/I genotype at the insertion/deletion also had lower risk for severe nephropathy than the D/I heterozygotes, and similarly at rs9896208, individuals with the C/C genotype had a lower risk than those with the C/T genotype. As measured by the R² statistic, genotypes at rs1800764, insertion/deletion, and rs9896208 accounted for 0.4, 0.4, and 0.6%, respectively, of explained variability in risk for severe nephropathy.

In multivariate proportional hazards analysis, individuals homozygous for the TIC haplotype compared with individuals with CDT/TIC had a HR of 0.49 (95% CI 0.32–0.75, P = 0.0009) and 0.41 (0.22–0.78, P = 0.006) for development of persistent microalbuminuria and severe nephropathy, respectively (Table 3). These three marker haplotypes explained 1.2% of variability in risk for persistent microalbuminuria and 0.8% for severe nephropathy. It appears that homozygosity for the C allele at the last marker, rs9896208, alone does not confer the substantially lower risk associated with the TIC/TIC haplotype, as the haplotype pair CDC/TIC, which is also homozygous for the C allele at rs9896208, is not distinguishable from CDT/TIC in these models (P = 0.09). The significantly lower risk of the TIC haplotype therefore appears to reside primarily in the homozygosity of the T allele and the insertion allele of rs1800764 and insertion/deletion or other unmeasured variation on that haplotype.

DISCUSSION

Given the previous contradictory or weakly positive cross-sectional studies of the ACE gene polymorphisms, we examined their role in the development of microalbumin excretion and severe nephropathy in the DCCT/EDIC cohort. Because of the extensive linkage disequilibrium within this genomic region, there are relatively few common haplotypes in Caucasians. Using only the insertion/deletion and two SNPs, we are able to capture the genetic variation that is present in the four most common haplotypes occurring at frequencies >5% in Caucasian subjects. This approach is superior to measurement of variation at only a single marker locus (such as the insertion/deletion, on which many of the previous cross-sectional studies are solely based) because it captures the common variation spanning ∼25 kb of genomic sequence encompassing the ACE gene. Using this approach, we are less likely to miss an association between polymorphic variation at a gene and nephropathy outcomes if it exists. At the same time, we are spared the task of genotyping all polymorphic variants in each haplotype to capture this variation.

When the ACE polymorphisms were examined against our nephropathy outcomes during a period of up to 17 years of observation, we determined that subjects with certain ACE genotypes had a significantly lower risk in progression to persistent microalbuminuria as well as severe nephropathy, which included subjects with AER >208 μg/min, dialysis, or transplant. The effect of the ACE genotypes persisted even when controlling in multivariate models for known covariates that are associated with microalbumin excretion in the DCCT/EDIC cohort, which specifically included sex, age at diagnosis, diabetes duration, DCCT baseline BMI, mean blood pressure, triglycerides, total cholesterol, HDL cholesterol, time-dependent updated HbA_1c weighted mean, cohort, and treatment.

A major quantitative trait loci for circulating ACE activity lies within a 16-kb region between intron 5 and downstream of exon 26 of the ACE gene (15). We speculate that genetic variation at the markers that we have analyzed, or at other loci in strong linkage disequilibrium with these markers, affects ACE enzyme activity, which in turn contributes to the development of diabetic nephropathy. Supporting this hypothesis are data from studies of transgenic diabetic mice engineered to have three copies of the Ace gene with resultant ACE enzyme activity 153% relative to two copy diabetic mice (23). The three copy mice had increased blood pressure and overt proteinuria compared with the two copy mice after 12 weeks of streptozotocin-induced diabetes, supporting the hypothesis that a genetic increase in ACE activity was sufficient to accelerate diabetic nephropathy in animal models (23).

The current study provides more definitive support for the role of ACE polymorphisms in diabetic nephropathy than the many previous cross-sectional studies (18,20). Based on our multivariate models adjusted for covariate effects, the three-marker ACE haplotypes explain 0.8% of variability in persistent microalbuminuria and 0.5% of variability in severe nephropathy. The previous studies may therefore lack statistical power to detect small effects of the ACE insertion/deletion polymorphism on renal outcomes. The pooled odds ratios for elevated urinary albumin excretion for I/I versus I/D and D/D genotypes in a meta-analysis (including different ethnic groups) in type 1 diabetes was 0.72 (95% CI 0.51–1.01) with a borderline significant P value of 0.06 (20). Three type 1 diabetes studies (17,24,25) in that meta-analysis reported odds ratios with 95% CIs excluding unity that indicated a protective effect of the I/I genotype on nephropathy.

The DCCT clinical trial was not initially designed as a genetic study, and to guard against population stratification, a potential source of bias, analyses were limited to subjects self-reported as “white.” In addition, the modest number of severe renal end points limits our power to detect small genetic effects. Despite this, there are many strengths of this study. The DCCT/EDIC cohort represents one of the largest collections of genetic and clinical information on subjects with type 1 diabetes. Biochemical measures were performed in one central laboratory, and the cohort is phenotypically well characterized, with longitudinal prospective clinical data on complication status for up to 17 years. The attrition from the EDIC study has been very low, and clinical data continues to be acquired on >90% of those DCCT participants who took part in EDIC. Finally, our genotyping strategy is superior to the many previous cross-sectional studies that only genotyped one polymorphic variant (i.e., insertion/deletion) and subsequently were unable to capture the common variation spanning the entire ACE gene.

Our findings in the DCCT/EDIC cohort therefore offer definitive evidence that in Caucasians, genetic variation at the ACE gene is associated not only with the onset of microalbuminuria but also with severe diabetic nephropathy in subjects with type 1 diabetes.