In total, 823 carefully characterised type 1 diabetic patients (Table 1) from the Finnish Diabetic Nephropathy Study (FinnDiane) were investigated. Patients were required to have onset of diabetes before 35 years of age, permanent insulin treatment initiated within 1 year of diabetes diagnosis and a fasting C-peptide level <0.3 nmol/l. Demographic data and blood and urine samples for determination of HbA1c, lipids and AER were collected. Renal status was based on the AER values in at least two of three consecutive overnight or 24-h urine collections; patients were classified as having no diabetic nephropathy (AER<20 μg/min or <30 mg per 24 h) or diabetic nephropathy (AER >200 μg/min or >300 mg per 24 h). Patients with end-stage renal disease were not included. Patients without diabetic nephropathy were required to have had a diabetes duration of >15 years to ensure normal renal status. Informed written consent was obtained from all patients, and the local ethics committee of each participating centre approved the study protocol, which was conducted in accordance with the principles of the Declaration of Helsinki.
The SNPs for ACE2 were determined from DNA isolated from whole blood using phenol extraction or a PUREGENE DNA Purification Kit (Gentra Systems, Minneapolis, MN, USA). The SNPs were chosen from public databases and were required to be evenly distributed over the gene. The following SNPs were investigated in the present study (listed 5′–3′): rs2285666, rs2048684, rs879922, rs714205 and rs5978731. Even if these are not tagSNPs, they capture 15 of the 17 SNPs in HapMap with a minor allele frequency of over 0.05, and thus most of the common variants across the gene (Fig. 1). The fact that five SNPs can capture most of the information across 40 kb is probably explained by the regions of high linkage disequilibrium (LD) on the X-chromosome. The promoter region contains several repeats that are problematic for PCR; thus, no SNP was successfully genotyped in this region. Genotyping of the following SNPs in the promoter regions failed: rs4646112, rs4646114 and rs12009805. Genotyping was performed using TaqMan chemistry with ABI Prism 7000 and 7900HT Sequence Detection Systems (Applied Biosystems, Foster City, CA, USA). This genotyping method is based on fluorogenic 5′-nuclease allelic discrimination chemistry, with no post-PCR processing steps necessary, making it more reliable and safe to use . Furthermore, genotyping quality was ensured by using 2% internal controls in each run, for which we demanded 100% accuracy. The success rate with this system is >99%.
Statistical analyses were performed using SPSS, Version 11.5 for Windows (SPSS, Chicago, IL, USA). Associations with categorical variables were analysed using the χ
2-test; associations with continuous variables were investigated with ANOVA. Mantel–Haenszel statistics were used to combine male and female p values. A p value of less than 0.05 was considered significant. For LD analysis we used Linkage Disequilibrium Analyzer 1.0 (available at http://www.chgb.org.cn/lda/lda.htm, last accessed in August 2005) . PHASE, Version 2.0.2 (available at http://www.stat.washington.edu/stephens/software.html, last accessed in August 2005)  was used to estimate haplotypes, which were further analysed with a specially designed C-language-based program that finds all haplotypes in the material and tests their statistical significance using the χ2-test. A power calculation was performed using the Genetic Power Calculator, case–control for discrete traits (available at http://www.statgen.iop.kcl.ac.uk/gpc/, last accessed in August 2005) . In order to calculate power, the following assumptions were made: disease prevalence 0.003; marker allele frequency 0.3; D′ with disease allele 0.9; high-risk allele frequency 0.3; genotype AA relative risk 2.3; genotype Aa relative risk 1.0001. Using these assumptions, 365 cases and 456 control subjects would give a power of 80% with a type I error rate of 0.05.