Genetic Risk Factors for Kidney Function in Individuals with Type 1 Diabetes

Visual Abstract

Abstract

Key Points

• Previous research has identified polygenic risk scores that are associated with low eGFR and albuminuria in the general population.

• We observed that these eGFR and albuminuria polygenic risk scores were associated with eGFR and albuminuria, respectively, in type 1 diabetes.

• Associations were independent of glycemic control and suggest shared genetic kidney risk factors between type 1 diabetes and the general population.

Background

Genetic risk factors underlying kidney disease in type 1 diabetes (T1D) remain poorly understood. We examined whether previously established polygenic risk scores (PRS) for eGFR and albuminuria are associated with these measures in adults with T1D in the Diabetes Control and Complications Trial (DCCT)/Epidemiology of Diabetes Interventions and Complications study.

Methods

We applied eGFR and albuminuria PRS derived in general population cohorts to 1304 DCCT/Epidemiology of Diabetes Interventions and Complications participants with genome-wide genotyping. We tested PRS associations with eGFR and urine albumin excretion rate (AER) as well as incident eGFR <60 ml/min per 1.73 m², AER ≥30 mg/24 h, and AER ≥300 mg/24 h. For consistency, PRS values were linearly transformed so higher scores corresponded to higher eGFR and AER. We also examined associations of kidney outcomes with rs55703767 in COL4A3, which has previously been associated with CKD in T1D.

Results

At DCCT baseline, participants had a mean age of 27 years; 53% were male. 49% of participants were randomized to intensive versus conventional glucose-lowering therapy. Participants were followed for median of (first–third quartiles) 35 (33–37) years. The eGFR PRS was significantly associated with continuous eGFR (per one SD higher PRS 2.72 ml/min per 1.73 m² higher [95% confidence interval (CI), 2.05 to 3.40]) and incident eGFR <60 ml/min per 1.73 m² (hazard ratio [HR]=0.82 [95% CI, 0.73 to 0.92]), but not consistently with albuminuria. There was no association with quantitative AER (2.42 mg/24 h [95% CI, −1.86 to 6.89]) or sustained AER ≥30 mg/24 h (HR=1.03; [95% CI, 0.94 to 1.14]). The albuminuria PRS was significantly associated with incident AER ≥30 mg/24 h (HR=1.12 [95% CI, 1.02 to 1.22]) but not continuous eGFR (0.49 ml/min per 1.73 m² higher [95% CI, −0.23 to 1.21]) or incident eGFR <60 ml/min per 1.73 m² (HR=0.96 [95% CI, 0.85 to 1.08]). Associations were similar in analyses stratified by DCCT treatment group assignment. rs55703767 was associated with lower incident macroalbuminuria in the overall cohort (HR=0.77 per minor allele [95% CI, 0.59 to 0.99]), and upon stratification by DCCT treatment group assignment, only within the conventional and not intensive glucose-lowering therapy group.

Conclusions

PRS associated with eGFR and albuminuria in the general population were associated with corresponding measures in adults with T1D. The results suggest shared genetic risk factors for kidney disease between T1D and the general population but different genetic risk factors for albuminuria and eGFR in T1D.

Clinical Trials Registration Numbers:

NCT00360893, NCT00360815.

Introduction

CKD, defined as persistent albuminuria ≥30 mg/g creatinine (Cr) or sustained eGFR <60 ml/min per 1.73 m², affects as many as 20%–30% of people with type 1 diabetes (T1D).¹ CKD in T1D is associated with a higher risk of cardiovascular events, morbidity, and mortality.^2,3 As such, identification of risk factors for CKD in T1D is important for improving long-term health outcomes.

Lifestyle and clinical features, such as poor glycemic control, longer diabetes duration, elevated systolic BP (SBP), and elevated triglycerides, have been identified as CKD risk factors in T1D.^4,5 In addition, familial clustering of CKD in T1D suggests an underlying genetic component to the disease.^6,7 Although numerous genome-wide association studies (GWAS) have been conducted identifying genetic polymorphisms associated with kidney function in people with and without diabetes, specific loci have varied across studies partly because of heterogeneity in kidney outcomes, study populations, and in underlying causes of CKD.^8–13 CKD broadly describes a common end point of a diverse set of pathologic processes, each potentially influenced by several genetic polymorphisms. A genetic risk factor identified generally for CKD may not be associated with heightened CKD risk in diabetes, with APOL1 risk variants serving as an example.¹⁴ Even T1D and type 2 diabetes (T2D) have distinct genetic susceptibility patterns differentially associated with CKD risk.^15–17

Here, we examine whether previously established polygenic risk scores (PRS) for eGFR and albuminuria from general population cohorts correlate with these measures in adults with T1D beyond established risk factors in the Diabetes Control and Complications Trial (DCCT)/Epidemiology of Diabetes Interventions and Complications (EDIC) study. Determining the applicability of genetic risk factors for kidney dysfunction from the general population to T1D is valuable for both understanding underlying disease mechanisms and for developing prognostic tools and targeted treatment strategies.

Methods

Study Participants

The DCCT and EDIC studies have been previously described.^18,19 Briefly, the DCCT randomized 1441 participants (age 13–39 years) with T1D to two treatment groups. Intensive glucose-lowering therapy (n=711) aimed to reduce glycemia to close to normal levels, whereas conventional glucose-lowering therapy (n=730) aimed to reduce the symptoms of hyperglycemia and hypoglycemia without specific glycemic targets. The primary prevention cohort included individuals with 1–5 years T1D duration, without retinopathy and albumin excretion rate (AER) <40 mg/dl. The secondary intervention cohort included individuals with 1–15 years T1D duration with mild to moderate nonproliferative retinopathy and AER <200 mg/dl. Individuals with hypertension, hyperlipidemia, cardiovascular disease, and severe retinopathy or kidney disease were excluded. At the end of DCCT (1993; mean follow-up 6.5 years), all participants were taught intensive therapy, referred to their personal health care providers for continuing care, and 97% enrolled in the observational EDIC follow-up study in 1994.

The present analyses employed available data through April 2024 (mean 33 years combined DCCT/EDIC follow-up) among the 1304 DCCT/EDIC participants with genome-wide genotyping data.

Kidney Function Measures

Serum Cr levels were measured annually in DCCT and EDIC; eGFR was calculated using the 2009 CKD Epidemiology Collaboration Cr equation because this was the equation used in the initial development of the eGFR PRS and to maintain consistency with previously published DCCT/EDIC analyses.^20,21 Cr was measured using an automated Jaffe kinetic method until May 2007, after which an isotope dilution mass spectrometry–traceable enzymatic method was used. Cr measurements obtained using the Jaffe method were recalibrated to the isotope dilution mass spectrometry–traceable enzymatic method. Reduced eGFR was defined as eGFR<60 ml/min per 1.73 m² or the onset of chronic kidney failure (dialysis or kidney transplantation). After the onset of ESKD, defined as chronic dialysis or kidney transplant, eGFR was imputed as 10 ml/min per 1.73 m² for all subsequent study visits. Sensitivity analyses were performed without imputed values by censoring follow-up at ESKD onset. Models without imputation are described as ESKD truncated.

Urinary AER was measured annually in DCCT and every other year in EDIC using 4-hour timed urine samples until EDIC year 18 (2012). Thereafter, AER was estimated using albumin-to-creatinine ratios (ACR) measured from spot urine samples.²² Sustained microalbuminuria was defined as AER ≥30 mg/24 h over two consecutive visits, whereas macroalbuminuria was defined as AER ≥300 mg/24 h in any one visit.²³ AER measurements below the limit of detection (n=259 [0.8%]) were set to 1 mg/24 h.

Clinical Variables

Demographic characteristics and clinical risk factors were obtained using standardized methods.^18,19 HbA1c was obtained quarterly in DCCT and annually in EDIC using high-performance liquid chromatography. BP, pulse, body weight, and insulin dose were assessed quarterly in DCCT and annually during EDIC during the annual physical examination. Fasting cholesterol and triglyceride levels were measured annually in DCCT and every other year in EDIC. Calcium channel blocker and β-blocker use was assessed annually in EDIC. Risk factors characterized by time-weighted mean values (e.g., mean HbA1c and mean SBP) were calculated by weighting each measurement by the time interval since the last measurement.

PRS

DCCT/EDIC samples underwent GWAS genotyping on the HumanCore BeadChip (Illumina, San Diego, CA), which contains 250,000 genome-wide tag single-nucleotide polymorphisms (SNPs) and over 200,000 exome-focused variants.¹¹ Ungenotyped SNPs were imputed using the TOPMed v2 reference panel. SNPs with low imputation quality (R² <0.3) were excluded (Supplemental Figure 1).²⁴ Genetic ancestry was assigned based on principal component analysis; all participants in this analysis were of European ancestry.¹¹

To evaluate genetic associations of eGFR in the DCCT/EDIC, we used an eGFR PRS (“LDpred PRS” signature) developed by Yu et al. incorporating information from 1.5 million SNPs from a meta-analysis of the UK Biobank GWAS and CKDGen Consortium GWAS of Cr-derived eGFR.²¹ The eGFR PRS was validated in the Atherosclerosis Risk in Communities (ARIC) study (https://www.pgscatalog.org/score/PGS000883/). The weights for the eGFR PRS were initially constructed such that a higher score corresponded with lower eGFR. To maintain consistency with our ACR analysis, we negatively transformed the eGFR PRS (multiplied by −1) such that a higher score corresponded to a higher eGFR.

For assessment of genetic albuminuria risk, we used an albuminuria PRS developed by Haas et al. composed of 46 SNPs independently associated with albuminuria in a GWAS conducted using data from 382,500 individuals of European ancestry from the UK Biobank, setting individuals with urine albumin concentrations below the lower limit of detection to the limit of detection of 6.7 mg/L (n=263,654 [69%]).²⁵ The ACR was then natural log-transformed and subsequently normalized to 1-SD (https://hugeamp.org/dinspector.html?dataset=GWAS_UKBiobankACR_eu). The albuminuria PRS was validated in ARIC and the Framingham Heart Study. Overall, a higher ACR PRS corresponded to a higher ACR in these cohorts.

A common missense variant in COL4A3: rs55703767 >T (XP_005246334.1:p.Asp326Tyr) which has previously been associated with diabetic nephropathy was directly genotyped and examined separately with the goal of confirming and extending previous findings.¹¹

Statistical Analysis

Clinical variables were described separately by eGFR or albuminuria PRS quintiles (first and fifth quintiles representing the lowest and highest PRS scores, respectively). Categorical variables were described as percentages, continuous variables were described using medians and interquartile ranges (IQR), and kidney complications were described by event rate per 1000 patient-years. Differences between quintiles were assessed using the Kruskal-Wallis test.

Associations of eGFR or albuminuria PRS with the risk of kidney outcomes (reduced eGFR, sustained microalbuminuria, or any macroalbuminuria) were assessed using Cox proportional hazards (PH) models, whereas linear generalized estimating equation models for repeated values over time were used to evaluate associations of PRS on continuous eGFR and AER. The PH assumption was tested using the Schoenfeld global tests and inspection of the residuals (Supplemental Figures 2–5). Departures from the PH assumption were not detected for exposures of interest (PRS) but were detected for some covariates (SBP, pulse, mean HbA1c, or duration of diabetes) for some outcomes, and these were addressed by including interaction terms of the covariates with functions of time. PRS were included as continuous covariates, and estimates were reported per one SD higher risk score, whereas the other risk factors were included either as fixed covariates (sex and DCCT treatment group assignment) or as time-dependent covariates (time-weighted mean SBP and HbA1c). Models were (1) unadjusted; (2) minimally adjusted for age and time-weighted mean HbA1c; and (3) fully adjusted using risk factors previously associated with kidney disease in the DCCT/EDIC, which include mean HbA1c, age, sex, mean triglycerides, pulse, SBP, duration of diabetes, use of any β-blockers or calcium channel blocker, any hypertension, current HbA1c, and mean weight.⁵ Models additionally assessed the associations of eGFR or albuminuria PRS with kidney outcomes separately within the DCCT intensive and conventional glucose-lowering therapy groups.

To determine whether adjusting for PRS attenuated associations between risk factors for kidney disease and reduced eGFR and sustained microalbuminuria, we compared hazard ratios from models with and without PRS adjustment.

Associations of the COL4A3 rs55703767 variant were examined with eGFR and AER outcomes using Cox PH models with stratification by intensive versus conventional glucose-lowering therapy group.

All analyses were conducted using SAS 9.4. P values < 0.05 were considered nominally significant.

Results

Participant Characteristics

At DCCT baseline, participants had a median age of 27 years and 53% were male, with a median diabetes duration of 52 months, HbA1c 8.6%, eGFR 125 ml/min per 1.73 m², and AER 11.5 mg/24 h (Table 1). The distributions of eGFR PRS (median [IQR], 1.76 [1.61–1.92]) and albuminuria PRS (median [IQR], 0.57 [0.54–0.61]) were approximately normally distributed (Supplemental Figure 6); participants were divided into quintiles based on PRS scores. Fewer participants in the second eGFR quintile were randomized to intensive versus conventional glucose-lowering therapy compared with other quintiles (43% versus 49%–52%); other characteristics were similar across eGFR PRS quintiles. Baseline characteristics were generally similar across albuminuria PRS quintiles. Compared with participants without genotyping data, those included in this analysis had lower baseline HbA1c (mean 9.1% versus 8.6%), otherwise baseline characteristics were similar (Supplemental Table 1).

Table 1.

Participant characteristics at Diabetes Control and Complications Trial baseline by eGFR and albuminuria polygenic risk score quintile (n=1304)

DCCT Baseline Characteristics	eGFR PRS					Albuminuria PRS
	First Quintile (n=260)	Second Quintile (n=261)	Third Quintile (n=261)	Fourth Quintile (n=261)	Fifth Quintile (n=261)	First Quintile (n=259)	Second Quintile (n=262)	Third Quintile (n=262)	Fourth Quintile (n=260)	Fifth Quintile (n=261)
Randomization to intensive glucose-lowering therapy	51%	43%	49%	52%	49%	51%	50%	47%	48%	48%
Secondary cohort	46%	57%	50%	50%	48%	47%	54%	50%	50%	48%
Male	53%	54%	57%	54%	49%	52%	56%	56%	55%	48%
Age (Yr)	27 (22–33)	27 (22–32)	27 (21–33)	27 (23–33)	27 (21–32)	28 (23–33)	27 (21–32)	27 (22–32)	28 (22–33)	27 (22–32)
T1D duration (Mo)	50 (28–107)	59 (27–117)	55 (32–107)	51 (26–111)	48 (29–108)	50 (30–103)	54 (28–108)	53 (32–111)	54 (26–116)	49 (27–115)
SBP (mm Hg)	114 (106–122)	114 (108–122)	114 (108–120)	114 (108–122)	112 (106–122)	114 (106–122)	114 (108–120)	114 (108–122)	112 (108–122)	114 (106–122)
DBP (mm Hg)	74 (68–80)	72 (68–78)	72 (68–80)	74 (68–80)	70 (64–78)	72 (66–80)	74 (68–80)	72 (68–78)	72 (68–80)	70 (66–80)
Pulse (bpm)	68 (61–75)	68 (60–75)	68 (60–77)	69 (60–75)	69 (61–77)	68 (60–77)	66 (60–73)	69 (62–76)	69 (58–77)	68 (61–75)
Weight (kg)	70 (61–78)	69 (61–79)	70 (61–79)	69 (60–77)	67 (60–76)	69 (60–77)	71 (62–78)	70 (61–78)	68 (60–77)	67 (60–78)
Triglycerides (mg/dl)	71 (57–98)	71 (55–93)	71 (56–98)	70 (54–93)	69 (54–89)	71 (56–95)	68 (53–92)	71 (56–94)	72 (56–99)	68 (54–93)
Insulin dose (units/kg per d)	0.65 (0.50–0.80)	0.63 (0.51–0.76)	0.61 (0.49–0.78)	0.66 (0.47–0.83)	0.65 (0.52–0.84)	0.63 (0.50–0.79)	0.62 (0.48–0.81)	0.66 (0.51–0.79)	0.63 (0.46–0.80)	0.64 (0.51–0.82)
AER (mg/24 h)	10 (6–16)	10 (7–17)	13 (7–22)	12 (7–19)	10 (7–19)	12 (7–17)	11 (6–20)	12 (7–19)	12 (6–20)	10 (7–17)
eGFR (ml/min per 1.73 m2)	123 (115–129)	124 (117–132)	123 (117–134)	126 (119–134)	127 (121–136)	124 (118–132)	126 (118–134)	125 (119–135)	125 (118–133)	124 (117–133)
HbA1c (%)	8.8 (7.9–9.8)	8.5 (7.7–9.8)	8.9 (8.0–10)	8.4 (7.5–9.5)	8.6 (7.7–9.8)	8.7 (7.9–9.9)	8.4 (7.7–9.7)	8.6 (7.8–9.9)	8.9 (7.8–10)	8.4 (7.6–9.6)

AER, albumin excretion rate; DBP, diastolic BP; DCCT, Diabetes Control and Complications Trial; HbA1c, hemoglobin A1c; PRS, polygenic risk score; SBP, systolic BP; T1D, type 1 diabetes.

Associations of eGFR and Albuminuria PRS with Kidney Outcomes

Reduced eGFR (<60 ml/min per 1.73 m²) occurred in 291 participants (22%), with an incidence rate of 7.1 events per 1000 person-years. Fewer reduced eGFR events were observed within higher eGFR PRS quintiles (Figure 1A and Supplemental Table 2). This effect was primarily driven by lower incidence of reduced eGFR in the fifth eGFR PRS quintile compared with other quintiles (5.0 events versus mean 7.7 events per 1000 person-years for quintiles 1–4; P value = 0.01 for differences across quintiles). Sustained microalbuminuria and incident macroalbuminuria events occurred in 470 (36%; 14.5 events per 1000 person-years) and 194 participants (15%; 4.9 events per 1000 person-years), respectively. More microalbuminuria events were observed with increasing albuminuria PRS quintiles, although differences across quartiles did not reach statistical significance (Figure 1B and Supplemental Table 2). By contrast, macroalbuminuria event rates were similar across albuminuria PRS quintiles. Kidney outcomes for participants with and without genotyping data are presented in Supplemental Table 1.

Figure 1.

Cumulative incidence of kidney outcomes in DCCT/EDIC PRS study subcohort. (A) Cumulative incidence of reduced eGFR by eGFR PRS quintile. (B) Cumulative incidence of sustained microalbuminuria by albuminuria PRS quintile. DCCT, Diabetes Control and Complications Trial; EDIC, Epidemiology of Diabetes Interventions and Complications; PRS, polygenic risk score.

Associations of eGFR PRS and albuminuria PRS with kidney outcomes are presented in Table 2. When identified, significant departures from the PH assumption were addressed using interaction terms between functions of time and covariates other than the eGFR PRS and the albuminuria PRS (Supplemental Table 3). In fully adjusted models, each one SD higher eGFR PRS was associated with 2.72 ml/min per 1.73 m² (95% confidence interval [CI], 2.05 to 3.40) higher eGFR as well as 18% lower risk of developing reduced eGFR (hazard ratio [HR], 0.82; [95% CI, 0.73 to 0.92]). The results were similar in unadjusted and minimally adjusted models, as well as when stratifying by DCCT intensive or conventional glucose-lowering therapy. In general, risk factor adjustment had minor effects on the hazard ratios for PRS (percent change in HR between −12% and 16%, data not shown), indicating a robust association of eGFR PRS with eGFR outcomes. Notably, higher eGFR PRS was also associated with a higher risk of incident macroalbuminuria in fully adjusted, but not unadjusted or minimally adjusted models; also, no association with quantitative AER or sustained microalbuminuria was apparent.

Table 2.

Associations of quantitative polygenic risk score with kidney parameters in the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications Study

Outcome	eGFR PRS			Albuminuria PRS
	Unadjusted	Minimally Adjusted	Fully Adjusted	Unadjusted	Minimally Adjusted	Fully Adjusted
Overall (n=1304)
Quantitative eGFR (ml/min per 1.73 m2)	2.8 (2.2–3.5)a	2.6 (2.1–3.2)a	2.7 (2.1–3.4)a	0.3 (−0.4 to 1.0)	−0.1 (−0.8 to 0.6)	0.5 (−0.2 to 1.2)
Reduced eGFR (eGFR <60 ml/min per 1.73 m2)	0.81 (0.73–0.91)a	0.80 (0.72–0.90)a	0.84 (0.75–0.94)b	1.01 (0.90–1.14)	1.01 (0.90–1.13)	0.96 (0.85–1.08)
Quantitative AER (% increase)	1.7 (−3.3 to 6.9)	2.0 (−2.6 to 6.7)	2.4 (−1.9 to 6.9)	5.0 (−0.1 to 10.3)	5.6 (0.7–10.6)c	3.4 (−0.9 to 7.9)
Sustained microalbuminuria (AER ≥30 mg/24 h)	1.02 (0.93–1.13)	1.03 (0.93–1.13)	1.03 (0.94–1.14)	1.13 (1.03–1.24)b	1.12 (1.03–1.22)c	1.12 (1.01–1.23)c
Macroalbuminuria (AER ≥300 mg/24 h)	1.06 (0.92–1.22)	1.10 (0.96–1.26)	1.17 (1.02–1.34)c	1.05 (0.91–1.20)	1.01 (0.88–1.16)	1.01 (0.88–1.17)
Intensive therapy group (n=637)
Quantitative eGFR (ml/min per 1.73 m2)	3.1 (2.3–3.9)a	2.7 (2.1–3.4)a	2.8 (1.9–3.7)a	0.0 (−1.0 to 1.1)	−0.6 (−1.5 to 0.4)	0.4 (−0.7 to 1.5)
Reduced eGFR (eGFR <60 ml/min per 1.73 m2)	0.76 (0.64–0.90)b	0.78 (0.66–0.92)b	0.80 (0.67–0.95)c	1.04 (0.87–1.24)	1.17 (0.98–1.41)	1.08 (0.90–1.29)
Quantitative AER (% increase)	1.8 (−3.7 to 7.6)	2.2 (−2.9 to 7.6)	3.8 (−1.1 to 9.0)	5.0 (−1.2 to 11.6)	7.0 (1.3–13.0)c	5.3 (0.3–10.5)c
Sustained microalbuminuria (AER ≥30 mg/24 h)	1.04 (0.90–1.20)	1.05 (0.91–1.21)	1.06 (0.91–1.22)	1.10 (0.96–1.27)	1.18 (1.02–1.35)c	1.16 (1.01–1.34)c
Incident macroalbuminuria (AER ≥300 mg/24 h)	0.97 (0.77–1.23)	1.01 (0.80–1.27)	1.04 (0.82–1.32)	1.03 (0.81–1.30)	1.12 (0.88–1.43)	1.11 (0.86–1.44)
Conventional therapy group (n=667)
Quantitative eGFR (ml/min per 1.73 m2)	2.5 (1.5–3.5)a	2.5 (1.5–3.5)a	2.6 (1.6–3.7)a	0.6 (−0.3 to 1.5)	0.3 (−0.6 to 1.2)	0.5 (−0.5 to 1.5)
Reduced eGFR (eGFR <60 ml/min per 1.73 m2)	0.87 (0.74–1.02)	0.84 (0.71–0.98)c	0.86 (0.73–1.01)	0.99 (0.84–1.16)	0.90 (0.77–1.06)	0.90 (0.76–1.06)
Quantitative AER (% increase)	1.8 (−6.1 to 10.3)	1.7 (−5.7 to 9.7)	1.0 (−5.7 to 8.3)	4.6 (−3.3 to 13.0)	4.1 (−3.4 to 12.2)	1.6 (−5.2 to 8.8)
Sustained microalbuminuria (AER ≥30 mg/24 h)	1.01 (0.88–1.15)	1.01 (0.89–1.16)	1.01 (0.89–1.16)	1.14 (1.01–1.29)c	1.10 (0.98–1.23)	1.11 (0.99–1.25)
Incident macroalbuminuria (AER ≥300 mg/24 h)	1.12 (0.94–1.33)	1.15 (0.97–1.37)	1.22 (1.03–1.45)c	1.05 (0.89–1.25)	0.97 (0.82–1.15)	1.01 (0.85–1.20)

After the onset of ESKD, eGFR was imputed as 10 ml/min per 1.73 m² for all subsequent study visits. Estimates for quantitative albumin excretion rate and eGFR are presented as change in albumin excretion rate or eGFR per SD change in the polygenic risk score (with 95% confidence interval) and are obtained from linear generalized estimating equation models. Estimates for binary complications (e.g., macroalbuminuria and sustained microalbuminuria) are presented as hazard ratios (with 95% confidence interval) per SD increase in the polygenic risk score and are obtained from Cox proportional hazards models. Risk factors included in the minimally adjusted models are age and mean updated hemoglobin A1c, whereas in the fully adjusted models are mean updated hemoglobin A1c, sex, mean triglycerides, pulse, systolic BP, duration of diabetes, any β-blockers, any hypertension, current hemoglobin A1c, and mean weight. AER, albumin excretion rate; PRS, polygenic risk score.

^aP value < 0.001.

^bP value < 0.01.

^cP value < 0.05.

Each one SD higher albuminuria PRS was associated with 12% higher risk of incident sustained microalbuminuria in fully adjusted models (HR, 1.12; [95% CI, 1.02 to 1.22]), with similar results in unadjusted and minimally adjusted models. This association was also present among participants randomized to DCCT intensive glucose-lowering therapy (HR, 1.16; [95% CI, 1.01 to 1.34]), with a trend toward significance among participants randomized to conventional glucose-lowering therapy (HR, 1.11; [95% CI, 0.99 to 1.25]; P value = 0.08). No associations between albuminuria PRS and quantitative AER, incident macroalbuminuria, quantitative eGFR, or reduced eGFR were observed.

Associations of kidney disease risk factors with reduced eGFR and sustained microalbuminuria remained similar after adjustment for eGFR and albuminuria PRS, respectively (Supplemental Table 4). Specifically, in models for reduced eGFR, hazard ratios for current duration and SBP were unchanged with the addition of eGFR PRS, whereas hazard ratios for mean HbA1c, mean triglycerides, history of hypertension, and use of β blockers changed by <5%. Current HbA1c was no longer significantly associated with reduced eGFR after the addition of eGFR PRS. In sustained microalbuminuria models, after addition of albuminuria PRS, hazard ratios for current HbA1c, mean HbA1c, mean triglycerides, pulse, and SBP were unchanged, whereas hazard ratios for β-blockers and male versus female patients changed by <6%.

Sensitivity analyses performed with ESKD truncation yielded similar results to those performed with eGFR imputation (Supplemental Table 5).

Associations of COL4A3 rs55703767 with Kidney Outcomes

Overall, 60% of participants had no copies of T (minor allele) at rs55703767, whereas 33% had one copy and 5% had two copies (Supplemental Figure 6). In unadjusted analyses, incidence of reduced eGFR, sustained microalbuminuria, or incident macroalbuminuria did not differ by rs55703767 allele copies (Supplemental Figure 7). Adjusted for age and mean HbA1c, each additional copy of the rs55703767 minor allele was associated with a 21% higher risk of reduced eGFR (HR, 1.21; [1.00–1.46]; P value = 0.047); however, in the fully adjusted model, this association was no longer apparent (Table 3).

Table 3.

Associations between COL4A3 rs55703767 and kidney parameters in the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications Study

Outcome	Unadjusted	Minimally Adjusted	Fully Adjusted
Overall (n=1304)
Quantitative eGFR (ml/min per 1.73 m2)	−0.1 (−1.3 to 1.0)	0.1 (−1.0 to 1.2)	0.3 (−0.9 to 1.5)
Reduced eGFR (eGFR <60 ml/min per 1.73 m2)	1.17 (0.96–1.41)	1.21 (1.00–1.46)a	1.20 (0.99–1.46)
Quantitative AER (% increase)	−3.0 (−10.8 to 5.5)	−2.5 (−9.8 to 5.4)	−5.0 (−11.5 to 2.0)
Sustained microalbuminuria (AER ≥30 mg/24 h)	0.99 (0.84–1.16)	0.99 (0.84–1.16)	0.95 (0.80–1.11)
Macroalbuminuria (AER ≥300 mg/24 h)	0.84 (0.65–1.08)	0.82 (0.64–1.05)	0.77 (0.59–0.99)a
Intensive therapy group (n=637)
Quantitative eGFR (ml/min per 1.73 m2)	−1.0 (−2.5 to 0.5)	0.2 (−1.0 to 1.5)	−0.2 (−1.7 to 1.3)
Reduced eGFR (eGFR <60 ml/min per 1.73 m2)	1.32 (1.00–1.75)a	1.27 (0.97–1.66)	1.25 (0.94–1.66)
Quantitative AER (% increase)	2.8 (−7.5 to 14.2)	3.5 (−6.0 to 14.0)	0.4 (−8.4 to 10.0)
Sustained microalbuminuria (AER ≥30 mg/24 h)	1.10 (0.86–1.41)	1.15 (0.90–1.47)	1.09 (0.85–1.40)
Macroalbuminuria (AER ≥300 mg/24 h)	1.05 (0.70–1.57)	1.00 (0.67–1.50)	0.98 (0.65–1.47)
Conventional therapy group (n=667)
Quantitative eGFR (ml/min per 1.73 m2)	0.7 (−1.0 to 2.4)	−0.1 (−1.8 to 1.6)	0.8 (−0.9 to 2.5)
Reduced eGFR (eGFR <60 ml/min per 1.73 m2)	1.05 (0.81–1.36)	1.14 (0.88–1.47)	1.18 (0.90–1.53)
Quantitative AER (% increase)	−8.4 (−19.3 to 4.0)	−7.2 (−17.7 to 4.7)	−9.2 (−18.3 to 1.0)
Sustained microalbuminuria (AER ≥30 mg/24 h)	0.90 (0.72–1.12)	0.90 (0.72–1.13)	0.85 (0.68–1.06)
Incident macroalbuminuria (AER ≥300 mg/24 h)	0.73 (0.53–1.00)	0.73 (0.53–1.00)	0.65 (0.47–0.91)a

Estimates for quantitative eGFR presented as change in eGFR per rs55703767 allele (with 95% confidence interval) and are obtained from linear generalized estimating equation models. Estimates for binary outcomes (e.g., reduced eGFR) are presented as hazard ratios (with 95% confidence interval) per rs55703767 allele and are obtained from Cox proportional hazards models. Risk factors included in the minimally adjusted models are age and mean updated hemoglobin A1c, whereas in the fully adjusted models are mean updated hemoglobin A1c, sex, mean triglycerides, pulse, systolic BP, duration of diabetes, any β-blockers, any hypertension, current hemoglobin A1c, and mean weight. AER, albumin excretion rate; PRS, polygenic risk score.

^aP value < 0.05.

By contrast, each additional copy of the rs55703767 minor allele was associated with a 23% reduced risk of macroalbuminuria in the fully adjusted model (HR, 0.77; [0.59–0.99]). In participants randomized to conventional glucose-lowering therapy, each additional copy of the minor allele was associated with a 35% lower risk of macroalbuminuria (HR, 0.65; [0.47–0.91]), whereas no association with macroalbuminuria was present in the DCCT-intensive glucose-lowering therapy group. However, formal tests for interaction did not reveal statistically significant heterogeneity by glucose-lowering therapy group (fully adjusted model, P value = 0.14).

Discussion

We evaluated associations of population-based eGFR and albuminuria PRS with kidney outcomes in adults with T1D from the DCCT/EDIC. The eGFR PRS was significantly associated with both continuous eGFR and reduced eGFR <60 ml/min per 1.73 m², and the albuminuria PRS was significantly associated with sustained microalbuminuria (AER ≥30 mg/24 h). We also reaffirmed the effect of rs55703767 on reduced macroalbuminuria risk in T1D,¹¹ particularly in the setting of hyperglycemia. Estimated GFR and albuminuria PRS associations with respective kidney outcomes were present even after adjusting for CKD risk factors in T1D. Notably, PRS associations with kidney outcomes were independent of glycemic control. These results demonstrate the applicability of population-based eGFR and albuminuria PRS in T1D, suggesting common genetic risk factors across distinct etiologies of kidney disease.

The search for genetic risk factors for CKD in diabetes has yielded mixed results, implicating multiple genes with varying effect sizes. Several large GWAS for CKD have been conducted in diabetes populations.^10,12,13,15 Although most primarily have focused on individuals with T2D, dedicated, although smaller, studies in T1D populations have also been performed.^9,11 These studies have identified several common loci associated with kidney dysfunction in diabetes, some of which are also observed with CKD in nondiabetes populations (e.g., UMOD encoding uromodulin).^10,12 These studies have added important insights into the mechanisms underlying CKD in diabetes. However, few single genetic variants have emerged as strongly and consistently associated with diabetic CKD, underscoring the disease’s complexity. Restricting the search for genetic CKD risk factors to diabetes-specific populations may limit detection of risk variants shared more broadly across various CKD etiologies, particularly those occurring at lower population frequencies. Notably, in a large GWAS (n=approximately 1.4 million) of eGFR stratified by diabetes status, although several loci had differential effects by diabetes status, the majority of associations did not differ across groups.¹²

Large GWAS in general population cohorts have enhanced our understanding of the genetic underpinnings of eGFR, identifying hundreds of associated genetic variants. A GWAS involving 1.5 million people from the CKDGen consortium, the Million Veteran Program, and the UK Biobank identified 878 loci associated with eGFR.²⁶ Other large GWAS including these cohorts have been conducted in cross-sectional studies of eGFR^8,21 and longitudinal cohorts of eGFR decline (n=approximately 340,000).²⁷ Overall, the heritability of eGFR is estimated to be as high as approximately 40%.^26,28,29 However, the absence of robust monogenic associations emphasizes the multifactorial nature of eGFR, characterized by polygenetic inheritance and strong environmental influences. PRS address the genetic complexity of eGFR by incorporating data across multiple SNPs. The Yu et al. eGFR PRS used in this study incorporates data from 1.5 million SNPs from combined UK BioBank and CKDGen Consortium GWAS analyses of eGFR.²¹ Among adults from the ARIC study, who at baseline had a mean eGFR of 100 ml/min per 1.73 m² and 9% of whom had diabetes (T1D versus T2D not specified), 1-SD lower eGFR PRS was associated with a 19% greater risk of incident CKD (HR, 1.19; 95% CI, 1.15 to 1.24) over 30 years of follow-up, adjusting for demographic and clinical covariates, including baseline eGFR. Our findings in the DCCT/EDIC were comparable, with 1-SD lower eGFR PRS corresponding to 21% higher risk of sustained eGFR <60 ml/min per 1.73 m² (HR, 1.21; 95% CI, 1.08 to 1.36) in fully adjusted models. Recently, another study investigating associations of eGFR PRS derived from previously published GWAS likewise found an association between the PRS and age of kidney failure onset.³⁰

Large meta-GWAS have identified gene loci associated with albuminuria primarily in general population cohorts, such as the CKDGen Consortium and the UK Biobank.^25,31,32 Although these cohorts are predominantly comprised individuals of European ancestry, smaller studies in Asian and Hispanic/Latino populations have also been conducted validating existing studies and identifying novel loci associated with albuminuria.^33–38 Overall, the heritability of microalbuminuria based on common variants captured by genotyping arrays is estimated at <10%.^28,31 Although some genetic variants are linked to specific glomerular basement membrane disorders, such as Alport syndrome, other single genes have consistently emerged in population-level GWAS as having strong associations with albuminuria independent of monogenic disease, for example CUBN (encoding cubulin, a proximal tubular membrane protein involved in albumin reabsorption).³⁹ Combining composite PRS data from albuminuria loci, such as the albuminuria PRS score from Haas et al. evaluated here, have also been valuable for capturing the polygenic nature of albuminuria risk. We observed that this derived albuminuria PRS from a general population cohort was also associated with albuminuria risk in the DCCT/EDIC cohort.

Notably, each evaluated PRS score mapped only to its respective phenotype—the eGFR PRS was only associated with eGFR-based outcomes, and the albuminuria PRS was only associated with albuminuria-based outcomes. Unexpectedly, higher eGFR PRS was associated with a higher risk of incident macroalbuminuria in addition to a lower risk of reduced eGFR. This finding is most likely spurious because the association was statistically significant only after full adjustment, and no associations with quantitative AER or sustained microalbuminuria were present. Overall, our findings highlight eGFR and albuminuria as separate clinical presentations of kidney disease with distinct underlying genetic risk factors and molecular mechanisms.

Moreover, there was no heterogeneity by glycemic control, implying that these PRS reflect common mechanisms of eGFR decline and albuminuria development that are independent of hyperglycemia in diabetes. At the same time, previous studies have linked albuminuria PRS to hypertension, hyperlipidemia, and other comorbidities, and conversely, have linked genetic factors associated with cardiometabolic risk factors, including body mass index, hypertension, and β-cell dysfunction, with kidney complications.^{25,31,40–42} These associations may represent genetic pleiotropy or the impacts of gene–environment interactions on kidney dysfunction. Collectively, these findings underscore the biologic complexity and heterogeneity of CKD in diabetes.

In contrast to the population-based eGFR and albuminuria PRS, which represent global CKD risk factors, the rs55703767 COL4A3 variant has been identified as a risk factor for CKD in T1D.¹¹ Specifically, in T1D, this variant has been associated with thinner glomerular basement membranes and protection from multiple albuminuria-based outcomes, including diabetic nephropathy, macroalbuminuria, and any albuminuria. Notably, this protective effect on CKD development was most strongly associated with poor glycemic control.¹¹ Here, using additional follow-up data and continuous measures, we again demonstrated a protective effect of the COL4A3 variant on macroalbuminuria development in the overall DCCT/EDIC cohort. Moreover, in stratified analyses, we show this effect is more strongly present with poorer glycemic control (although results of analyses using an interaction term were NS, likely due to limited power). Despite a protective impact on macroalbuminuria development, we did not observe a similar effect on eGFR decline. Because albuminuria often precedes eGFR decline in T1D, a detectable impact on eGFR may require longer follow-up or be more modest in magnitude. Of note, COL4A3 is linked to autosomal forms of Alport syndrome, which is characterized by thinning of the glomerular basement membrane and presents clinically with albuminuria and hematuria.⁴³

Strengths of this study include use of a large, well-characterized, T1D cohort with long-term follow-up extending over a mean of 33 years, standardized serial eGFR and albuminuria measurements, and genotyping data. This study has several limitations. Study participants were predominantly of White race; the eGFR and albuminuria PRS used here were developed in cohorts of primarily European ancestry and while appropriate for use in the DCCT/EDIC, may not be applicable to other racial or ethnic groups. Event rates for binary outcomes (reduced eGFR, sustained microalbuminuria, and incident macroalbuminuria) were relatively low for discovery but are robust for external validation of existing genetic markers.

In conclusion, we found that genetic factors that predict eGFR and albuminuria in the general population are similarly associated with these measures in adults with T1D, even after adjusting for CKD risk factors. Moreover, long-term glycemic exposure did not affect the associations of eGFR and albuminuria PRS with their respective kidney outcomes, suggesting underlying common pathologic pathways across etiologies of CKD. At the same time, we confirmed the impact of rs55703767 on macroalbuminuria risk, the effects of which may be amplified by hyperglycemia. Together, these findings point toward a diverse set of genetic risk factors for CKD in T1D, some of which impart risk independent of glycemic control and others which may mediate the effect of hyperglycemia on kidney disease. Further research is needed to investigate the potential for clinical risk stratification of kidney outcomes based on genetic variants.

Supplementary Material

cjasn-21-1554-s001.pdf ^{(3MB, pdf)}

cjasn-21-1554-s002.pdf ^{(635.8KB, pdf)}

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/CJN/C800.

Author Contributions

Conceptualization: Ionut Bebu, Ian H. de Boer, Christine P. Limonte, Gayle M. Lorenzi, Andrew D. Paterson, Bruce A. Perkins.

Data curation: Ionut Bebu, Xiaoyu Gao, Delnaz Roshandel.

Formal analysis: Xiaoyu Gao, Andrew D. Paterson, Delnaz Roshandel.

Funding acquisition: Gayle M. Lorenzi, Bruce A. Perkins.

Investigation: Ionut Bebu, Ian H. de Boer, Xiaoyu Gao, Amy B. Karger, Christine P. Limonte, Gayle M. Lorenzi, Andrew D. Paterson, Bruce A. Perkins, Delnaz Roshandel.

Methodology: Ionut Bebu, Xiaoyu Gao, Amy B. Karger, Christine P. Limonte, Gayle M. Lorenzi, Andrew D. Paterson, Bruce A. Perkins, Ian H. de Boer.

Resources: Gayle M. Lorenzi, Andrew D. Paterson, Bruce A. Perkins.

Supervision: Ionut Bebu, Andrew D. Paterson, Bruce A. Perkins, Ian H. de Boer.

Writing – original draft: Christine P. Limonte.

Writing – review & editing: Ionut Bebu, Ian H. de Boer, Xiaoyu Gao, Amy B. Karger, Christine P. Limonte, Gayle M. Lorenzi, Andrew D. Paterson, Bruce A. Perkins, Delnaz Roshandel.

Funding

The DCCT/EDIC has been supported by cooperative agreements (1982-1993, 2012-2017, 2017-2022, 2022-2027), and contracts (1982-2012) with the Division of Diabetes, Endocrinology, and Metabolic Diseases of the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK; current grant numbers U01 DK094176 and U01 DK094157), and through support by the National Eye Institute, the National Institute of Neurologic Disorders and Stroke, the General Clinical Research Centers Program (1993-2007), and Clinical Translational Science Center Program (2006-present), Bethesda, MD. The sponsor of this study is represented by the NIDDK Project Scientist who serves as part of the DCCT/EDIC Research Group and plays a part in the study design and conduct as well as the review and approval of manuscripts. The NIDDK Project Scientist was not a member of the writing group of this paper. The opinions expressed are those of the investigators and do not necessarily reflect the views of the funding agencies.

Data Availability Statements

https://repository.niddk.nih.gov/studies/edic/ Data collected for the DCCT/EDIC study through June 30th, 2022, are available to the public through the NIDDK Central Repository (https://repository.niddk.nih.gov/studies/edic/). Data collected in the current cycle (July 2022–June 2027) will be available within 2 years after the end of the funding cycle.

Supplemental Material

This article contains supplemental material online, published as provided by the authors, at http://links.lww.com/CJN/C801.

Supplemental Table 1. Baseline characteristics of DCCT/EDIC participants with and without eGFR or albuminuria PRS.

Supplemental Table 2. Cumulative incidence of kidney outcomes by eGFR and albuminuria PRS quintile.

Supplemental Table 3. Modifications to models to address violations of the PH assumption.

Supplemental Table 4. Effect of PRS on the association between risk factors and the risk of kidney outcomes.

Supplemental Table 5. Associations of quantitative PRS with kidney parameters (with and without imputation) in the DCCT/EDIC Study—sensitivity analysis showing results of analyses performed with imputed values and ESKD truncated results.

Supplemental Figure 1. Workflow of eGFR and AER PRS analyses.

Supplemental Figure 2. Schoenfeld residual plots for reduced eGFR versus eGFR PRS in the overall cohort.

Supplemental Figure 3. Schoenfeld residual plots for reduced eGFR versus eGFR PRS in the intensive treatment group.

Supplemental Figure 4. Schoenfeld residual plots for reduced eGFR versus eGFR PRS in the conventional treatment group.

Supplemental Figure 5. Schoenfeld residual plots for sustained AER>=30 versus albuminuria PRS in the overall cohort.

Supplemental Figure 6. Distribution of albuminuria PRS, eGFR PRS, and COL4A3 rs55703767 minor (T) allele count.

Supplemental Figure 7. Cumulative incidence of kidney outcomes by COL4A3 rs55703767 minor allele count.

Statistical Analysis Plan.

List of members of the DCCT/EDIC research group.