Study design
The initial study population was 5,336 adult Greenlanders who had participated in population-based health surveys conducted in the years 1999-2001, 2005-2010 and 2017-2019 respectively. The Population Study in Greenland 1999 (B99(18), Inuit Health in Transition (IHIT(19)) and the Population Survey in Greenland 2018 (B2018 (2)) with nationwide sampling. A total of 3,820 participated once and 1,516 participated twice in B99 or IHIT at baseline and follow-up in either IHIT or B2018. A small sample participated in all three surveys, but missing data on key variables made it infeasible to make a three-point follow up.
Participants completed lifestyle questionnaires, clinical examinations and a majority contributed with paraclinical data from blood samples, oral glucose tolerance tests (OGTTs) and random spot urine samples. We therefore had information on age, sex, height, weight, systolic and diastolic blood pressure, smoking status, blood glucose levels (Hba1c, fasting and two-hour glucose values from the OGTT), serum creatinine, urinary albumin creatinine ratio (UACR), low density lipoprotein (LDL) cholesterol and triacyl glycerol (TG).
All participants in the health surveys provided oral and written informed consent. The health surveys were conducted in accordance with the Helsinki Declaration and were approved by the Ethics Committee for Medical Research in Greenland. Details of the B99 study(18), the IHIT study(19) and the B2018 study(2) are found elsewhere.
Diabetes
We defined diabetes according to 2006 World Health Organization (WHO) OGTT criteria of fasting plasma glucose ≥7.0 mmol/l, 2 hour plasma glucose ≥11.1mmol/l(20) or self-reported by questionnaire. In a sensitivity analysis diabetes was defined by Hba1c ≥48mmol/l as recommended by WHO in 2011(21).
Kidney function
Frozen samples were used to estimate kidney function. Blood stored at the laboratory at Steno Diabetes Center Copenhagen at -80°C from the years 1999-2001 and 2005-2010 were analyzed for creatinine levels using “Vitros 5600” Ortho Clinical Diagnostics(22). Estimated glomerular filtration rate (eGFR) was calculated using serum creatinine values expressed as milliliters per minute and adjusted for mean body surface area of 1,73 m2, age and sex according to the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) formula with CKD cut-off at eGFR < 60 ml/min/1.73m2. We used Danish guidelines(23) similar to the 2012 KDIGO guidelines (Kidney Disease: Improving Global Outcomes) defining albuminuria as urine albumin creatinine ratio in a random spot urine > 30 mg/g(24).
Genotyping
The TBC1D4 variant was genotyped using the KasPAR assay (LGC Genomics, Hoddesdon, UK) and European admixture proportions were estimated with a proportion of one equal to 100% Inuit ancestry and zero equal to 100% European ancestry, using data from the Illumina MetaboChip(17).
Analyses
For cross sectional analyses we used baseline data from individuals at the time of their first visit. We examined the effect of diabetes on kidney function expressed as dichotomous outcomes albuminuria yes/no and CKD yes/no using logistic regression. For both outcomes we performed a crude analysis of the effect of diabetes, and in model 1 we adjusted for age and sex. In model 2 we further adjusted for body mass index (BMI), systolic- and diastolic blood pressure, low density lipoprotein (LDL) cholesterol, triacylglycerol (TG) and smoking. In model 3 we further adjusted for the TBC1D4 variant and genetic admixture. We did a sensitivity analysis using Hba1c diabetes criteria instead of OGTT criteria and ran analyses again to quantify differences in associations between the two diagnostic measures. We also tested the effect of the TBC1D4 variant and European genetic admixture on microalbuminuria and CKD assuming a recessive effect comparing homozygous (HO) with wildtype (WT) and heterozygous (HT) carriers combined. In an unadjusted model we tested the effect of the TBC1D4 variant and genetic admixture, then in model 1 we adjusted for age and sex.
Using baseline and follow-up data for those who participated twice, we used linear regression to test predictors of continuous measures of kidney function, measured as changes in eGFR and urinary albumin creatinine ratio (UACR) from baseline to follow-up, adjusted for baseline values. For both outcomes we first tested the effect of diabetes on eGFR and UACR adjusted for baseline values. In model 1 we adjusted for age and sex. In model 2 we further adjusted for BMI, systolic and diastolic blood pressure, LDL cholesterol, TG, smoking and years between baseline and follow-up and in model 3 we further adjusted for the effect of the TBC1D4 variant and genetic admixture. We checked the normality of distributions of covariates and log10 transformed UACR to get a better fit. We back transformed for interpretation of parameter estimates that therefore reflect proportional changes in UACR.
Statistical significance level was set at 5% using complete cases. Data was managed and analyzed using SAS 9.4(25-27).