Effect of Vitamin D Supplementation on Kidney Function in Adults with Prediabetes

A Secondary Analysis of a Randomized Trial

Abstract

Background and objectives 

Low serum 25-hydroxyvitamin D (25[OH]D) concentration has been associated with higher levels of proteinuria and lower levels of eGFR in observational studies. In the Vitamin D and Type 2 Diabetes (D2d) study, we investigated the effect of vitamin D supplementation on kidney outcomes in a population with prediabetes.

Design, setting, participants, & measurements 

Overweight/obese adults with high risk for type 2 diabetes (defined by meeting two of three glycemic criteria for prediabetes) were randomized to vitamin D₃ 4000 IU per day versus placebo. Median duration of treatment was 2.9 years (interquartile range 2.0–3.5 years). Kidney outcomes included (1) worsening in Kidney Disease: Improving Global Outcomes (KDIGO ) risk score (low, moderate, high, very high) on two consecutive follow-up visits after the baseline visit and (2) mean changes in eGFR and urine albumin-to-creatinine ratio (UACR).

Results 

Among 2166 participants (mean age 60 years, body mass index 32 kg/m², serum 25(OH)D 28 ng/ml, eGFR 87 ml/min per 1.73 m², UACR 11 mg/g, 79% with hypertension), 10% had moderate, high, or very high KDIGO risk score. Over a median follow-up of 2.9 years, there were 28 cases of KDIGO worsening in the vitamin D group and 30 in the placebo group (hazard ratio, 0.89; 95% confidence interval [95% CI], 0.52 to 1.52]). Mean difference in eGFR from baseline was −1.0 ml/min per 1.73 m² (95% CI, −1.3 to −0.7) in the vitamin D group and −0.1 ml/min per 1.73 m² (95% CI, −0.4 to 0.2) in the placebo group; between-group difference was −1.0 ml/min per 1.73 m² (95% CI, −1.4 to −0.6). Mean difference in UACR was 2.7 mg/g (95% CI, 1.2 to 4.3) in the vitamin D group and 2.0 (95% CI, 0.5 to 3.6) in the placebo group; between-group difference was 0.7 mg/g (95% CI, −1.5 to 2.9).

Conclusions 

Among persons with prediabetes, who were not preselected on the basis of serum 25(OH)D concentration, vitamin D supplementation did not affect progression of KDIGO risk scores and did not have a meaningful effect on change in UACR or eGFR.

Visual Abstract

Introduction

The American Diabetes Association (ADA) defines prediabetes not as a disease, but as a condition that increases risk for type 2 diabetes and cardiovascular disease (¹). Individuals with prediabetes commonly have comorbidities, including hypertension, which mediate risk for patient-relevant outcomes, including kidney disease. In addition, low circulating 25-hydroxyvitamin D (25[OH]D) concentration has been suggested as a risk factor for both type 2 diabetes (²) and kidney disease (^345–6).

Several lines of reasoning suggest a potential reno-protective role of activated vitamin D, 1,25(OH)₂D. Preclinical studies suggest that 1,25(OH)₂D helps regulate the renin-angiotensin system (⁷,⁸), enhances insulin sensitivity (⁹), and improves endothelial function (¹⁰), which can all affect BP and maintain vascular health (¹¹) of the kidney. In addition, knockout of vitamin D receptors in mice can aggravate hyperglycemia-induced kidney injury leading to worsening albuminuria and glomerulosclerosis (⁸).

Human studies have shown that low circulating 25(OH)D concentration can predict all stages of kidney disease, from albuminuria to kidney failure (^345–6), but some studies have found no association (¹²,¹³). In the Third National Health and Nutrition Examination Survey (NHANES III), prevalence of albuminuria increased with each decrease in quartile of blood 25(OH)D concentration (³). In an Australian population cohort, blood 25(OH)D concentration lower than 20 ng/dl was independently associated with prevalent albuminuria (⁴). In the same cohort, vitamin D deficiency (25[OH]D<15 ng/dl) was associated with incident albuminuria and reduced eGFR (eGFR<60 ml/min per 1.73 m²) (⁵); low circulating 25(OH)D concentration was also associated with progression to kidney failure in the NHANES III cohort (⁶).

Despite evidence for associations between low circulating 25(OH)D concentration and kidney disease, few clinical trials have evaluated the effect of vitamin D supplementation on kidney outcomes. The Vitamin D and Type 2 Diabetes (D2d) study was a randomized clinical trial of US adults with prediabetes to test the effect of vitamin D₃ supplementation versus placebo on diabetes risk (¹⁴). In the D2d study, vitamin D supplementation did not significantly decrease new-onset diabetes (hazard ratio, 0.88; 95% confidence interval [95% CI], 0.75 to 1.04); however, since publication of the main D2d results, aggregate meta-analyses have reported a significant 11%–12% reduction in diabetes risk with vitamin D supplementation among people with prediabetes (¹⁵,¹⁶). This study is a secondary analysis to determine prevalence of kidney dysfunction in the D2d prediabetes population and to examine the effect of vitamin D supplementation on incident kidney outcomes.

Materials and Methods

Overview of the D2d Study

This study is a secondary analysis of the D2d study focusing on kidney outcomes. The D2d study was a randomized, double-blind, placebo-controlled, multisite clinical trial to test whether vitamin D supplementation has an effect on new-onset diabetes in adults with prediabetes (clinicaltrials.gov NCT 01942694, registered 9/16/2013). Details of the protocol and results have been published (¹⁴,¹⁷). The institutional review board at each clinical site approved the protocol, and all participants provided written, informed consent. The study was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice.

Study Population

Participants were recruited from 22 medical centers across the United States (¹⁸). Eligible participants met at least two of three glycemic criteria for prediabetes as defined by the 2010 ADA guidelines: fasting plasma glucose 100–125 mg/dl (5.6–6.9 mmol/L); plasma glucose 2 hours after a 75-g oral glucose load 140–199 mg/dl (7.8–11.0 mmol/L); hemoglobin A1c 5.7%–6.4% (39–47 mmol/mol) (¹⁹). Other inclusion criteria were age greater than or equal to 30 years (25 years for American Indians, Alaska Natives, Native Hawaiians, or other Pacific Islanders) and body mass index (BMI) of 24–42 kg/m² (22.5–42 kg/m² for Asians). A low serum 25(OH)D level was not an eligibility criterion. Key exclusion criteria included: any glycemic criterion in the diabetes range; use of diabetes or weight-loss medications; history of hyperparathyroidism, nephrolithiasis, hypercalcemia, or bariatric surgery; or an eGFR—calculated by means of the Chronic Kidney Disease Epidemiology Collaboration equation (²⁰)—of less than 50 ml/min per 1.73 m² of body-surface area. For a complete list of eligibility criteria, see Supplemental Table 1. The recruitment and screening process has been described previously (¹⁷,¹⁸,²¹).

Intervention and Procedures

Participants were randomized to take either a single soft-gel that contained 4000 IU of vitamin D₃ (cholecalciferol) or matching placebo once daily. Randomization was computer generated and block-stratified according to trial site, BMI (<30 or ≥ 30 kg/m²), and race (White or non-White). Study staff used a web-based, interactive, D2d-specific study pill inventory and randomization system to enter the stratification information, and the system assigned the participant to vitamin D₃ or placebo. At randomization and every 6 months, the system generated a specific pill bottle number that study staff used to dispense study pills to participants in a blinded fashion. No unblinding took place during the study, and study staff and participants were notified of the assignment after the primary results were published. Participants were asked to limit the use of outside-of-study vitamin D to 1000 IU per day from all supplements and calcium supplements to 600 mg per day.

Follow-Up and Measurements

Follow-up visits occurred at month 3, at month 6, and twice per year thereafter. Blood and urine for kidney outcomes were collected after an 8-hour overnight fast at baseline, and at months 3 (blood only), 12, 24, 36, and 48. Serum for creatinine was analyzed on the same day of the visit at each site’s clinical laboratory (see Supplemental Table 2). Other blood and urine specimens were processed locally and shipped to the central laboratory for long-term storage at −80°C until analyses. Stored serum samples were used to measure 25(OH)D by liquid chromatography tandem mass spectrometry validated by a quarterly proficiency-testing program administered by the Vitamin D External Quality Assessment scheme (United Kingdom) (²²,²³). Stored urine was used to measure both albumin, using an immunoturbidimetric method, and creatinine, using an enzymatic-colorimetric method with calibration traceable to an isotope dilution mass spectrometry reference measurement procedure; both measurements were completed on the Cobas c311 analyzer (Roche Diagnostics, Indianapolis, IN).

Outcomes

We evaluated time to worsening in kidney function on the basis of the risk categories established by the Kidney Disease: Improving Global Outcomes (KDIGO) organization (Supplemental Figure 1) (²⁴). The KDIGO risk classification has four levels on the basis of eGFR (ml/min per 1.73 m²) and urine albumin-to-creatinine ratio (UACR, mg/g): (1) green—low risk (eGFR≥60 and UACR<30); (2) yellow—moderate risk (eGFR≥60 and UACR 30–300 or eGFR 45–59 and UACR<30); (3) orange—high risk (eGFR≥60 and UACR>300 or eGFR 45–59 and UACR 30–300 or eGFR 30–44 and UACR<30); and (4) red—very high risk (eGFR 45–59 and UACR>300 or eGFR 30–44 and UACR≥30 or eGFR<30 and any UACR). We defined worsening in kidney function as an increase by at least one KDIGO risk category from baseline on two consecutive visits (confirmed KDIGO worsening). Time to worsening in KDIGO risk category was chosen as a composite outcome in order to incorporate both eGFR and UACR, as prior studies have suggested an effect of vitamin D on both kidney measures (^345–6), and KDIGO categories predict risk of cardiovascular disease, progressive kidney disease, and mortality. In sensitivity analyses, we expanded the outcome to include any individual who had worsening in KDIGO classification at their last study visit only (unconfirmed KDIGO worsening).

We also examined the following kidney end points: change from baseline in eGFR and change from baseline in UACR (as continuous variables); time to meeting eGFR<60 ml/min per 1.73 m² (among participants with eGFR≥60 ml/min per 1.73 m² at baseline); and time to meeting UACR≥30 mg/g (among participants with UACR<30 mg/g at baseline).

Covariates

Age, race, and ethnicity were self-reported. Height, weight, and BP were measured using standardized procedures. All medications and dietary supplements being taken by the participant were brought in at the screening visit and recorded and reviewed at each follow-up encounter.

Statistical Methods

The sample size for the parent study was determined on the basis of a target of 508 diabetes events and a total sample size of 2382 participants randomized equally to the vitamin D and placebo groups. The rationale has been previously published (¹⁴). The analysis population for this study included all participants, regardless of follow-up status, who had available eGFR and UACR data at the baseline visit. All participants were analyzed according to their randomized treatment group, regardless of adherence to the assigned intervention (intention-to-treat population). Follow-up time was calculated as time from randomization until the occurrence of the kidney end point (e.g., KDIGO worsening), death, withdrawal, or last follow-up encounter.

We imputed missing values for eGFR and UACR at follow-up visits using Markov chain Monte Carlo simulations assuming the data were from a multivariable normal distribution (²⁵,²⁶). We generated five fully imputed datasets and combined estimated results using Rubin’s rules (²⁷). Missing values were imputed for scheduled visits until the end-of-study date for each participant, which varied from participant to participant given that the parent trial was designed as event-driven (in relation to the diabetes outcome) (¹⁷). We imputed 8.9% of the data.

Between-group differences for the change in continuous variables compared with baseline were determined using a linear mixed-effects model approach to account for within-participant correlation across the time points. The outcome of the linear mixed model is change compared with baseline with randomization assignment as the single independent fixed effect. All postbaseline observations are weighted equally. Average change within each group and mean differences between groups along with 95% CIs are presented. Time-to-event end points were evaluated with the use of Kaplan–Meier estimates and Cox proportional hazards models. All models included group assignment, as the main predictor variable, and the stratification variables (trial site, BMI, and race).

Variability of response to vitamin D supplementation for confirmed KDIGO worsening was assessed in subgroups defined by key baseline variables: age, race, BMI, 25(OH)D concentration, and, given their known renoprotection, use of angiotensin-converting enzyme inhibitor (ACEi) /angiotensin II receptor blockers (ARB). We also evaluated change in eGFR and UACR in those without imputed values and in two select subsets: (1) those not taking vitamin D supplements at baseline and (2) those not on any ACEi/ARB during the study. No adjustments were made for multiple comparisons; therefore, only point estimates and 95% CIs are presented without P values.

This analysis was not prespecified, and, therefore, considered post hoc, exploratory. However, the statistical analysis plan (including the definition of kidney outcomes) was reviewed and approved by all coauthors and the study’s publications and presentations committee before outcomes data were analyzed and presented by group. Statistical analyses were performed by an independent statistician using SAS version 9.4 (SAS Institute Inc.).

Results

Of 2423 participants randomized to vitamin D or placebo, 247 did not consent to the biorepository, and ten did not have baseline eGFR or UACR. After excluding these participants, a total of 2166 participants were included in the present analysis (see Supplemental Figure 2). Baseline characteristics of participants did not differ from the entire D2d cohort (Table 1 and Supplemental Table 3). The mean (±SD) age of the kidney cohort was 60 (±10) years and BMI was 32 (±5) kg/m². Over half (52%) reported use of antihypertensive medications; 33% were using ACEi/ARB (31% in the vitamin D versus 34% in the placebo). Forty-three percent reported some use of vitamin D supplementation at baseline; mean serum 25(OH)D was 28 ng/ml. Adherence to trial pills was high, with 83% of prescribed pills taken (83% in vitamin D versus 83% in placebo).

Table 1. - Baseline characteristics of the Vitamin D and Type 2 Diabetes kidney cohort^a

Variables	Overall (n=2166)	Vitamin D (n=1083)	Placebo (n=1083)
Characteristic
Age, years	60±10	60±10	61±10
Women, no. (%)	958 (44)	475 (44)	483 (45)
Race, no. (%) b
Asian	116 (5)	59 (5)	57 (5)
Black or African American	517 (24)	248 (23)	269 (25)
White	1477 (68)	746 (69)	731 (68)
Other	56 (3)	30 (3)	26 (2)
Hispanic or Latino ethnicity, no. (%) a , b	199 (9)	109 (10)	90 (8)
Family history of diabetes (first degree relative), no. (%)	1363 (63)	683 (63)	680 (63)
Smoking, no. (%)
Never	1245 (58)	622 (57)	623 (58)
Former	764 (35)	385 (36)	379 (35)
Current	139 (6)	67 (6)	72 (7)
Unknown or not reported	18 (1)	9 (1)	9 (1)
Dietary supplement use c
Vitamin D
Participants taking vitamin D supplements, no. (%)	934 (43)	454 (42)	480 (44)
Vitamin D intake among all participants, IU/d d	315±399	313±404	318±394
Vitamin D intake among participants using supplements, IU/d	732±255	746±257	718±252
Calcium
Participants taking calcium supplements, no. (%)	726 (34)	340 (31)	386 (36)
Calcium intake among all participants, mg/d d	105±177	102±178	109±177
Calcium intake among participants using supplements, mg/d	315±168	324±170	307±166
Physical activity, total MET hour/week
Mean ±SD	112±161	110±158	113±165
Median (IQR)	56 (26–128)	60 (26–129)	55 (27–125)
Body mass index, kg/m2	32±5	32±5	32±4
Body mass index category, kg/m2, no. (%)
30 kg/m2	781 (36)	393 (36)	388 (36)
30–34.9 kg/m2	808 (37)	410 (38)	398 (37)
≥35 kg/m2	577 (27)	280 (26)	297 (27)
Systolic BP, mm Hg	128±14	128±13	129±14
Diastolic BP, mm Hg	77±9	77±9	77±10
Hypertension, no. (%) e	1703 (79)	850 (79)	853 (79)
Antihypertensive medication use, no. (%)	1133 (52)	555 (51)	578 (53)
Angiotensin II receptor blockers or ACE inhibitors	706 (33)	337 (31)	369 (34)
Other	427 (20)	218 (20)	209 (19)
Prediabetes category, no. (%) f
Met all 3 glycemic criteria (IGT + iA1c + IFG)	775 (36)	384 (36)	391 (36)
Met two glycemic criteria only
IGT + IFG	136 (6)	69 (6)	67 (6)
IGT + iA1c	205 (10)	92 (9)	113 (10)
IFG + iA1c	1050 (49)	538 (50)	512 (47)
Laboratory
Fasting plasma glucose, mg/dl	108±7	108±7	108±8
2-hour post load plasma glucose, mg/dl	138±34	137±34	138±34
Hemoglobin A1c, %	5.9±0.2	5.9±0.2	5.9±0.2
Serum creatinine, mg/dl	0.9±0.2	0.9±0.2	0.9±0.2
eGFR, ml/min per 1.73 m2
Mean±SD	87±16	87±15	86±16
Median (IQR)	88 (76–97)	88. (76–98)	87 (75–97)
eGFR category, no. (%)
≥60 ml/min per 1.73 m2	2079 (96)	1046 (97)	1033 (95)
<60 ml/min per 1.73 m2	87 (4)	37 (3)	50 (5)
Urine albumin-to-creatinine ratio, mg/g
Mean±SD	11±48	11±52	12±4
Median (IQR)	3 (2–7)	3 (2–6)	3 (2–7)
Urine albumin-to-creatinine ratio category, no. (%)
<30 mg/g	2040 (94)	1028 (95)	1012 (93)
30–300 mg/g	118 (5)	52 (5)	66 (6)
>300 mg/g	8 (0.4)	3 (0.3)	5 (1)
KDIGO classification, no. (%)
Normal/low risk	1961 (91)	997 (92)	964 (89)
Moderate risk	191 (9)	79 (7)	112 (10)
High risk	12 (1)	5 (1)	7 (1)
Very high risk	2 (0.1)	2 (0.2)	0
Serum 25-hydroxyvitamin D, ng/ml	28.1±10.1	27.8±10.2	28.4±10.1
Serum 25-hydroxyvitamin D category, no. (%) g
<12 ng/ml	89 (4)	52 (5)	37 (3)
12–19 ng/ml	366 (17)	192 (18)	174 (16)
20–29 ng/ml	779 (36)	401 (37)	378 (35)
≥30 ng/ml	931 (43)	438 (40)	493 (46)

MET, metabolic equivalent of task ; IQR, interquartile range; ACE, angiotensin-converting enzyme; IGT, impaired glucose tolerance; iA1c, impaired hemoglobin A1c; IFG, impaired fasting glucose; KDIGO, Kidney Disease: Improving Global Outcomes.

^aPlus-minus values are mean±SD. Percentages may not add up to 100 because of rounding.

^bRace and ethnicity were reported by the participant. The category “other” includes American Indian or Alaska Native; Native Hawaiian or other Pacific Islander; or other race. Ethnicity includes any race.

^cData on vitamin D and calcium intake are derived from a specific question about use of dietary supplements, including multivitamins.

^dValue shown is among all participants regardless of whether they reported use of supplements or not.

^eHypertension is defined as one of the following: (1) self-reported or (2) use of antihypertensive medication or (3) systolic BP ≥130 mm Hg or diastolic BP ≥80 mm Hg.

^fIFG defined as fasting plasma glucose 100–125 mg/dl (5.6–6.9 mmol/L); IGT defined as 2-hour post load plasma glucose after a 75-g glucose load 140–199 mg/dl (7.8–11.0 mmol/L); or iA1c defined as hemoglobin A1c 5.7%–6.4% (39–47 mmol/mol).

^gCategories of serum 25-hydroxyvitamin D are on the basis of the 2010 Dietary Reference Intakes for calcium and vitamin D recommended by the Food and Nutrition Board of the National Academy of Medicine (³¹).

Kidney Function at Baseline in the D2d Kidney Cohort

Mean (±SD) eGFR was 87 (±16) ml/min per 1.73 m². eGFR was <60 ml/min per 1.73 m² in 4% of participants (3% in the vitamin D versus 5% in the placebo), and UACR≥30 mg/g was seen in 6%. Using the KDIGO classification, 10% of participants scored at or above moderate risk (yellow, orange, or red) (8% in the vitamin D versus 11% in the placebo).

Kidney Outcomes

Over a median follow-up of 2.9 years (interquartile range 2.0–3.5 years), there were 28 cases of confirmed worsening in KDIGO risk score (increase in KDIGO risk category on two consecutive visits) in the vitamin D group and 30 in the placebo group (hazard ratio for vitamin D, 0.89; 95% CI, 0.52 to 1.52) (Table 2). Reanalysis without imputed numbers showed similar results (data not shown). In sensitivity analysis in which the outcome included individuals who had worsening in KDIGO risk category at the last visit (and thus unconfirmed), there were a greater number of events in both groups but no differences between groups (hazard ratio, 1.04; 95% CI, 0.73 to 1.48). Table 2 also shows the incidence of eGFR<60 ml/min per 1.73 m² and UACR≥ 30mg/g, which also did not differ by group.

Table 2. - Kidney outcomes in the vitamin D and placebo groups

Kidney Outcomes	Vitamin D No. of events/total No.	Placebo No. of events/total No.	Vitamin D versus Placebo Hazard Ratio (95% CI)
Higher KDIGO risk (confirmed) a	28/1083	30/1083	0.89 (0.52 to 1.52)
Higher KDIGO risk (confirmed and unconfirmed) a , b	73/1083	70/1083	1.04 (0.73 to 1.48)
eGFR<60 ml/min per 1.73 m2	74/1046	60/1033	1.22 (0.86 to 1.72)
UACR≥30 mg/g	63/1028	75/1012	0.82 (0.58 to 1.17)

95% CI, 95% confidence interval; KDIGO, Kidney Disease: Improving Global Outcomes; UACR, urine albumin-to-creatinine ratio.

^aConfirmed higher KDIGO risk was defined as increase by at least one KDIGO risk category from baseline on two consecutive visits.

^bUnconfirmed higher KDIGO risk was defined as individuals who had worsening in KDIGO risk category at their last study visit only.

Mean change (95% CI) in eGFR and UACR at each study visit are shown in Table 3. Overall, there was a statistically significant greater decline in mean eGFR in the vitamin D group relative to the placebo group (mean difference −1.0 ml/min per 1.73 m²; 95% CI, −1.4 to −0.6, for vitamin D versus placebo). This decrease in eGFR represented the average change over the full period of follow-up with all postbaseline observations weighted equally. The mean difference in change in UACR between groups was not statistically significant (0.7 mg/g; 95% CI, −1.5 to 2.9). Restricting the analyses to those without imputed numbers showed similar results (Supplemental Table 4).

Table 3. -  Changes over time in eGFR and UACR in the vitamin D and placebo groups

Kidney Outcomes	Baseline	Month 3	Month 12	Month 24	Month 36	Month 48	Mean Difference Compared with Baseline (95% CI) a
eGFR ml/min per 1.73 m2 b
Vitamin D group	87±15	85±15	87±16	87±16	88±16	89±16
No. analyzed	1083	1075	1047	952	589	229
Difference compared with baseline		−2.2 (−2.7 to −1.6)	−1.0 (−1.5 to −0.4)	−0.52 (−1.1 to 0.1)	−0.3 (−1.1 to 0.4)	−0.2 (−1.6 to 1.3)	−1.0 (−1.3 to −0.7)
Placebo group	86±16	85±16	87±16	87±16	87±16	88±16
No. analyzed	1083	1072	1049	948	593	228
Difference compared with baseline		−1.8 (−2.4 to −1.3)	0.2 (−0.3 to 0.8)	0.4 (−0.2 to 1.0)	1.1 (0.4 to 1.9)	1.3 (0.0 to 2.6)	−0.1 (−0.4 to 0.2)
Between-group difference		−0.4 (−1.1 to 0.4)	−1.2 (−2.0 to −0.4)	−1.0 (−1.8 to −0.1)	−1.5 (−2.52 to −0.4)	−1.5 (−3.5 to 0.5)	−1.0 (−1.4 to −0.6)
P value c							< 0.01
UACR, mg/g d
Vitamin D group	11±52	NA	12±63	12±44	13±46	14±45
No. analyzed	1083		1047	952	589	229
Difference compared with baseline			2.6 (0.1 to 5.2)	2.1 (−0.6 to 4.7)	3.6 (0.3 to 7.0)	3.6 (−1.8 to 9.1)	2.7 (1.2 to 4.3)
Placebo group	12±42	NA	13±65	13±47	15±67	12±35
No. analyzed	1083		1049	948	593	228
Difference compared with baseline			0.8 (−1.8 to 3.3)	1.7 (−1.01 to 4.3)	3.8 (0.4 to 7.2)	4.6 (−1.1 to 10.3)	2.0 (0.5 to 3.6)
Between-group difference			1.9 (−1.7 to 5.5)	0.4 (−3.4 to 4.2)	−0.2 (−5.0 to 4.6)	−1.0 (−8.8 to 6.9)	0.7 (−1.5 to 2.9)
P value c							0.52

95% CI, 95% confidence interval; UACR, urine albumin-to-creatinine ratio.

^aWithin-group differences over time and P values comparing between-group difference were calculated with a mixed model for repeated measures. Between-group difference reflect average change (compared with baseline) over the full period of follow-up.

^bGFR was estimated using the Chronic Kidney Disease Epidemiology Collaboration equation. Serum creatinine was measured at the local laboratory in real time.

^cP value for the comparison between the unadjusted means for vitamin D versus placebo groups at each follow-up visit is on the basis of the Wilcoxon rank-sums test (because the distributions are skewed).

^dAll time points for UACR for a given participant were analyzed in the same analytic run. UACR was not measured at Month 3.

Subgroup Analyses

Among participants who did not use ACEi/ARB at baseline, the hazard ratio for confirmed KDIGO worsening was 0.76 (95% CI, 0.35 to 1.64) for vitamin D compared with placebo (Supplemental Figure 3). Among those on ACEi/ARB, the hazard ratio for vitamin D versus placebo was 1.05 (95% CI, 0.49 to 2.28). Among participants with baseline serum 25(OH)D level <20 ng/ml, the hazard ratio for confirmed KDIGO worsening was 0.37 (95% CI, 0.10 to 1.30) for vitamin D versus placebo. The P value for interaction was not significant for any of the subgroup analyses.

As 43% of the D2d population was taking some form of vitamin D supplementation at baseline, we evaluated mean changes in eGFR and UACR in the subset not taking vitamin D. There was no significant difference in eGFR change from baseline between groups (mean difference −0.3 ml/min per 1.73 m²; 95% CI, −1.0 to 0.5) (Supplemental Table 5). Mean change from baseline in UACR was 0.4 mg/g (95% CI, −2.0 to 2.9) in the vitamin D group and 4.0 mg/g (95% CI, 1.3 to 6.6) in the placebo group. The between-group difference at the end of 48 months was −3.5 mg/g (95% CI, −7.1 to 0.1).

Throughout the trial, there was a trend for less use of ACEi/ARB in participants on vitamin D compared with placebo (odds ratio, 0.88; 95% CI, 0.75 to 1.03; Supplemental Table 6). After excluding participants who were on ACEi/ARB at baseline and throughout the study, there was no statistically significant difference in eGFR change from baseline between groups (mean difference −0.5; 95% CI, −1.0 to 0.0) (Supplemental Table 7). There also was no statistically significant difference in UACR change from baseline between groups (mean difference −2.1 mg/g; 95% CI, −4.6 to 0.3; Supplemental Table 7).

There were no significant differences in adverse events, including hypercalcemia and nephrolithiasis, during the trial (¹⁴).

Discussion

In people with prediabetes, vitamin D₃ supplementation had no significant effect on several kidney outcomes including a composite kidney outcome on the basis of KDIGO risk category, which includes eGFR and UACR measurements. Vitamin D supplementation also appeared to have no benefit when examining eGFR and UACR as separate variables using a continuous scale. There was a statistically significant, albeit clinically less meaningful, decline in eGFR of about 1 ml/min per 1.73 m² in the vitamin D group relative to the placebo group. This decrease in eGFR represents the average change over the full period of follow-up with all postbaseline observations weighted equally. Some prior studies have reported a decrease in eGFR with 1,25(OH)₂D analogs, such as paricalcitol (²⁸,²⁹), which was speculated to be due to a change in creatinine metabolism rather than a true decline in GFR. Cholecalciferol, however, has not been reported to affect creatinine metabolism. Future studies with vitamin D may want to assess GFR using a non–creatinine-based method.

Although observational studies have reported associations between low circulating 25(OH)D concentration and kidney disease (^345–6), few trials have tested for an effect of vitamin D supplementation on kidney outcomes. Recently, the Vitamin D and Omega-3 Trial to Prevent and Treat Diabetic Kidney Disease (VITAL -DKD) study reported that supplementation with 2000 IU/d of vitamin D₃ compared with placebo had no significant effect on eGFR or UACR at 5 years in older persons with established diabetes (³⁰). Similar to our study, the majority of participants had normal kidney function (mean eGFR of 86 ml/min per 1.73 m²) and sufficient circulating 25(OH)D concentration at baseline by current guidelines (³¹). Interestingly, in both the VITAL-DKD study and ours, subgroup analyses in those with lower 25(OH)D seem to favor vitamin D intervention. In addition, in the subgroup not taking vitamin D at baseline, there was a trend to lower UACR in those randomized to vitamin D versus placebo. However, in both studies, confidence intervals were wide, and tests for interaction by 25(OH)D level were not significant. Thus, on the basis of both studies, vitamin D₃ supplementation is unlikely to appreciably affect kidney indices in a population with normal kidney function and adequate vitamin D status.

With the caveat that a clinical trial population may not represent the general population, our analysis informs on the prevalence and incidence of kidney dysfunction in a population with high risk for type 2 diabetes, using the modern ADA glucose criteria. In this cohort with over half on antihypertensive medications and about one third on ACEi/ARB, only 4% had kidney disease to begin with, defined as eGFR<60 ml/min per 1.73 m², and 6% had UACR 30 mg/g or greater. These proportions are somewhat lower than persons with prediabetes in NHANES diagnosed on the basis of either an elevated fasting glucose or hemoglobin A1c (³²). In NHANES (survey periods 2011–2014), 5% had eGFR<60 ml/min per 1.73 m², and 8% had UACR 30 mg/g or greater. The prevalence of microalbuminuria is similar to the Diabetes Prevention Program (DPP) cohort who were selected on the basis of high 2-hour glucose during an oral glucose tolerance test. In DPP, a similar proportion (6%) to the D2d study had UACR≥30 mg/g, with less than 10% of the DPP cohort using ACEi or ARBs (³³). Trial results, however, may not be generalizable to real-word settings.

Few studies have reported on incident kidney disease in the prediabetes population, and most have used fasting glucose to define prediabetes (³⁴,³⁵). In the current trial, 6% of individuals developed eGFR<60 ml/min per 1.73 m² and 7% developed UACR>30 mg/g after a median follow up of 2.9 years. Comparable populations reporting both eGFR and UACR were not found in the literature; thus, our analysis provides important information on the natural progression of kidney function in prediabetes.

The present analysis retains the strengths of the parent trial, including use of the gold standard assay for 25(OH)D, use of the latest ADA glycemic criteria to define prediabetes, and ascertainment of kidney function at yearly intervals. One study limitation is that although kidney outcomes were prespecified in the analytic plan, the trial was not designed for kidney outcomes and participants were not selected to be at particularly high risk for kidney disease. In addition, trial duration was relatively short and neither vitamin D nor placebo groups displayed clinically meaningful progression in kidney parameters. Although the vitamin D and placebo groups were well balanced, there also were some baseline differences in kidney indices and use of renin-angiotensin system inhibitors, which may have shifted the comparison between vitamin D and placebo toward null. Finally, participants were not selected on the basis of vitamin D status and, as a result, the cohort would be considered as having sufficient vitamin D status by current recommendations (³¹). Of interest, among participants not taking vitamin D at baseline, vitamin D supplementation lowered UACR compared with placebo.

In conclusion, the D2d study—the largest vitamin D diabetes prevention trial—showed no significant benefit of vitamin D₃ supplementation on kidney disease progression in individuals with high-risk prediabetes but low baseline risk for adverse kidney outcomes. As participants were not selected on the basis of baseline 25(OH)D concentrations, we cannot exclude a kidney benefit for those individuals with vitamin D deficiency.

Disclosures

R. Chatterjee has received research funding from Bristol Myers Squibb, Epigenomics, and Verily. S. Kashyap has served as a consultant for Fractyl Inc. and GI Dynamics, and has additionally received honoraria from GI Dynamics. S. Kim is an advisor for GI Dynamics and has received funding from Bayer. R. Pratley reports employment with AdventHealth Translational Research Institute. R. Pratley has served as a consultant for AstraZeneca, Glytec LLC, Janssen, Merck, Mundipharma, Novo Nordisk, Pfizer, Sanofi, Scohia Pharma Inc., and Sun Pharmaceutical Industries; except for services for Sanofi US Services Inc. on 2/12/2018 and 6/25/2018 (which were paid to R. Pratley personally), all payments are made directly to his employer (AdventHealth). R. Pratley has received research grants from Hanmi Pharmaceutical Co., Janssen, Metavention, Novo Nordisk, Poxel SA, and Sanofi; all payments are made directly to his employer (AdventHealth). He has also acted as a speaker for Merck and Novo Nordisk; all payments are made directly to his employer (AdventHealth). N. Rasouli has received research grants from Lilly and Novo Nordisk and has served as a consultant for Novo Nordisk. All payments are made directly to her employer (University of Colorado). M. Sarnak has acted as a consultant for Cardurian and as a steering committee member for Akebia (funds paid to Tufts Medical Center), and has attended an advisory board meeting for Bayer in May of 2019. All remaining authors have nothing to disclose.

Funding

The planning phase of the Vitamin D and Type 2 Diabetes (D2d) trial was funded by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) through a multicenter clinical study implementation planning grant to Tufts Medical Center in Boston (U34DK091958, principal investigator, A.G. Pittas). Planning was also supported, in part, by the Intramural Research Program of the NIDDK. The conduct of the trial was supported primarily by the NIDDK and the Office of Dietary Supplements of the National Institutes of Health through the multicenter clinical study cooperative agreement (U01DK098245, principal investigator, A.G. Pittas) to Tufts Medical Center, where the D2d Coordinating Center is based. The U01 grant mechanism establishes the NIDDK project scientist (Dr. Staten ) as a member of the D2d Research Group. The trial also received secondary funding from the American Diabetes Association to Tufts Medical Center (1-14-D2d-01, principal investigator, A.G. Pittas). A.G. Pittas is supported in part by generous donations to the Tupper Research Fund at Tufts Medical Center. Neither the funders nor the authors’ institutions had any role in study design, collection, analysis, and interpretation of data; writing the report; and the decision to submit the report for publication.