Enhanced Excretion of Vitamin D Binding Protein in Type 1 Diabetes: A Role in Vitamin D Deficiency?

Abstract

Context: Vitamin D deficiency is an increasingly recognized comorbidity in patients with both type 1 (T1D) and type 2 diabetes, particularly associated with the presence of diabetic nephropathy.

Objective: Because we have previously reported enhanced excretion of megalin in the urine of T1D patients with microalbuminuria, we hypothesized that concurrent urinary loss of the megalin ligand, vitamin D binding protein, might contribute mechanistically to vitamin D deficiency.

Design and Participants: Examining a study cohort of 115 subjects with T1D, aged 14–40 yr, along with 55 age-matched healthy control subjects, we measured plasma and urine concentrations of vitamin D binding protein (VDBP) along with serum concentrations of total calcium, parathyroid hormone, 25-hydroxyvitamin D, and 1, 25-dihydroxyvitamin D; these results were compared between groups and investigated for relationships with metabolic control status or with albuminuria.

Main Outcome Measure: Between-group differences in urinary VDBP concentration were the main outcome measures.

Results: A marked increase in the urinary excretion of VDBP was apparent in subjects with T1D, compared with control subjects. Using multivariate regression modeling, significant correlates of urinary VDBP excretion included microalbuminuria (P = 0.004), glycosylated hemoglobin (P = 0.010), continuous glucose monitoring system average capillary glucose (P = 0.047), and serum 1,25(OH)₂D concentrations (P = 0.037). Vitamin D deficiency or insufficiency was slightly more prevalent in diabetic subjects with albuminuria, coincident with the increase in urine VDBP excretion.

Conclusions: These findings suggest that, theoretically, exaggerated urinary loss of VDBP in T1D, particularly in persons with albuminuria, could contribute mechanistically to vitamin D deficiency in this disease.

Exaggerated urinary loss of vitamin D binding protein occurs in type 1 diabetes, particularly in association with poorer glycemic control and albuminuria.

An increased incidence of vitamin D deficiency and/or insufficiency in patients with either type 1 diabetes (T1D) or type 2 diabetes (T2D) has been reported both in the United States (1,2) and elsewhere (3,4) in both children and adults (5). Moreover, patients with T1D and microalbuminuria have been reported to have lower vitamin D levels than those with normoalbuminuria (6). Recent studies have also demonstrated that the presence of vitamin D deficiency or insufficiency in diabetes is independently associated with the presence of diabetic nephropathy (7). Despite these findings and implications, mechanisms contributing to vitamin D deficiency in T1D remain unclear.

Megalin, or low-density lipoprotein-related protein 2 (LRP2) is an approximately 600-kDa single-spanning transmembrane glycoprotein belonging to the low-density lipoprotein receptor family (8). This endocytic receptor, expressed in polarized epithelial cells including proximal tubule cells of the kidney, is a multiligand receptor, with an extensive repertoire of ligands, which include albumin, vitamin-binding proteins, lipoproteins, hormones, enzymes, and drugs (8). One megalin ligand, vitamin D binding protein (VDBP), is the major transport protein for vitamin D metabolites in plasma, with approximately 88% of 25-hydroxyvitamin D (25OHD) in the circulation bound to VDBP (9). The VDBP + 25OHD complex is freely filtered across the glomerulus allowing transport to the proximal tubule (PT). In the PT, reabsorption of the VDBP + 25OHD complex by megalin facilitates the generation of 1, 25-dihydroxyvitamin D [1,25(OH)₂D] via 1α-hydroxylase activity in PT epithelial cells (10,11).

We have previously demonstrated enhanced excretion of megalin in the urine of patients with microalbuminuria secondary to T1D (12). In vivo, megalin-deficient mice exhibit increased urinary excretion of VDBP (11), vitamin D deficiency, and osteomalacia (13). Clinically, increased urinary excretion of VDBP has also been demonstrated in association with tubular dysfunction, such as occurs in patients with nephrotic syndrome (14), renal Fanconi syndrome (15), and cadmium-toxicity induced tubular dysfunction (16). Because vitamin D deficiency has been reported in patients with T1D (1), we examined the relationship between urinary loss of the megalin ligand, VDBP, with albuminuria and vitamin D status in 115 subjects with T1D compared with 55 healthy control subjects. We hypothesized that, due to megalin excretion, urinary loss of VDBP would be increased in T1D and particularly accentuated in those subjects with T1D and albuminuria; moreover, we hypothesized that the loss of VDBP could relate to systemic vitamin D concentrations.

Subjects and Methods

Study population

Subjects with T1D and age-matched healthy control subjects, aged 14–40 yr, were recruited from clinics at the University of Arkansas for Medical Sciences (UAMS), Arkansas Children’s Hospital, and surrounding communities. Approval was obtained from the Institutional Review Board of UAMS, and all subjects provided informed consent (18–40 yr) or assent (14–17 yr). Exclusion criteria included: 1) T2D; 2) history of other chronic systemic inflammatory or autoimmune disease or malignancy; 3) pregnancy; 4) concurrent ketonuria; or 5) evidence of active infection at the time of baseline evaluation.

For all subjects, two first-morning, fasting outpatient evaluations were conducted 3–5 d apart to obtain the following: 1) medical history and demographic information; 2) two separate venipuncture laboratory measurements of fasting plasma glucose (FPG), glycosylated hemoglobin (HbA1c), C-peptide, and serum creatinine (Cr); 3) a 3- to 5-d-interval recording of continuous glucose monitor sensor data (using the Minimed continuous glucose monitoring system (CGMS), MMT-7102; Medtronic Northridge, CA), indicative of concurrent glycemic control; and 4) an interim 24-h urine specimen to be used for the determination of a timed albumin excretion rate (UAE) and the measurement of urine VDBP concentration. Glomerular filtration rate (GFR) was calculated using the Modification of Diet in Renal Disease Study equation (17) for subjects 18 yr of age or older and using the equation of Schwartz and colleagues (18) for subjects younger than 18 yr of age. Serum from the visit 2 specimen was used for measurement of total calcium, 25OHD, 1,25(OH)₂D, and intact PTH. All blood specimens were obtained in the fasting state. No effort was made to control for or assess dietary calcium or vitamin D intake. However, concurrent medication use, including calcium or vitamin D supplements, was recorded.

Assays

FPG and C-peptide were measured by the UAMS General Clinical Research Center Core Laboratory. HbA1c, serum Cr, and urinary albumin and creatinine concentrations (24 h urine specimen) were measured by LabCorp (Dallas, TX). Serum measurements of total calcium (reference range 8.4–10.2 mg/dl), intact PTH (reference range 15–75 pg/ml; electrochemiluminescent immunoassay), 25OHD (reference range, optimal level 30–80 ng/ml; insufficiency 20–29 ng/ml; deficiency < 20 ng/ml; competitive immunoluminometry) and 1,25(OH)₂D (reference range 15–75 pg/ml; RIA) were performed at ARUP Laboratories (Salt Lake City, UT). VDBP was assayed in EDTA plasma (10 μl, diluted 1:40,000) and in aliquots of 24-h urine collections (100 μl, diluted 1:10) by ELISA (ALPCO Diagnostics; Salem, NH; enzyme immunoassay no. 30-2314). Urine albumin and VDBP values were normalized to urine Cr concentrations and expressed as albumin to Cr ratio (ACR; milligrams per gram) and as VDBP to Cr ratio (VDBPCR; nanograms per gram), respectively, to control for variable urine concentrations, particularly resulting from diabetic polyuria. In addition, the total daily excretion of VDBP was calculated using the 24-h urine specimen total volume (total VDBP; nanograms per day).

Statistical analyses

Statistical analysis was performed using SAS software (version 9.2; SAS Institute Inc., Cary, NC) and SPSS statistical software (version 18.0; SPSS Inc., Chicago, IL). Results for FPG, HbA1c, C-peptide, and serum Cr obtained from the two study visits were averaged. Serum concentrations of calcium, PTH, 25OHD, and 1,25(OH)₂D were measured only once per subject using visit 2 specimens. Transformations of variables were used to ensure that normality assumptions were satisfied. If necessary, nonparametric tests were used. Data are presented as mean ± sd or as median and range for nonnormally distributed variables. Kruskal-Wallis nonparametric tests were used to ascertain differences between multiple groups, whereas Mann-Whitney U tests were used for specific two-group comparisons. Spearman’s rank correlation coefficients were assessed as measures of correlation between variables of interest. We also used a linear regression model, incorporating the pertinent interaction term [PTH, 25OHD, 1,25(OH)₂D, total VDBP, or VDBPCR] to identify group difference of associations between dependent and independent variables. Additionally, possible predictor and confounding variables were modeled in a multivariate logistic regression for vitamin D deficiency and in a linear regression for VDBP excretion to identify those factors that best predicted the outcome variable. For all analyses, statistical significance was defined as P ≤ 0.05.

Results

Four of 59 enrolled control participants (two females and two males) were incidentally found to have albuminuria or proteinuria; hence, data from these individuals were excluded from the analyses. The final study population included 55 control subjects and 115 subjects with T1D. Aspects of this study population have also been reported elsewhere (19). Control and T1D groups were comparable with respect to gender (56 vs. 51% female, respectively); racial distribution (86 vs. 93% Caucasian, respectively) and baseline body mass index (BMI) (25.2 ± 4.7 vs. 24.9 ± 4.4 kg/m², respectively). No subjects were receiving oral calcium supplements, whereas eight subjects (two control and six T1D) reported daily intake of a generic multivitamin. Date of enrollment was used to define seasonality, as fall/winter (September through February) and spring/summer (March through August). Recruitment occurred during the fall/winter for 34.6% of control and 45.4% of T1D subjects; recruitment occurred during the spring/summer for 65.4% of control and 54.5% of T1D subjects. Seasonal categorization was not statistically different between groups (Season by Disease P = 0.174).

The mean UAE for all T1D subjects (n = 115) was 41.7 ± 213.4 mg/g Cr, compared with 10.1 ± 5.7 mg/g for the control group (Table 1: group 1). To assess specifically the effect of albuminuria on vitamin D status, T1D subjects were further categorized as normoalbuminuric (group 2: n = 99; UAE ≤ 30 mg/g Cr) or albuminuric (group 3: n = 16; >30 mg/g Cr) for between-group comparisons. Of the 16 albuminuric patients, UAE concentrations were within a microalbuminuria range in 14 subjects (UAE > 30–299 mg/g Cr) and within a macroalbuminuria range (UAE ≥ 300 mg/g Cr) in two subjects. For the subset of albuminuric patients, the mean UAE was 226.5 ± 551.0 mg/g (median value 50.8, minimum value 30.4; maximum value 2250.1 mg/g).

Table 1.

Demographic and metabolic characterization of study subjects, grouped by UAE

Condition	Group 1 Control	Group 2 T1D + UAE ≤30 mg/g	Group 3 T1D + UAE >30 mg/g	1 vs. 2 P value	1 vs. 3 P value	2 vs. 3 P value
n	55	99	16
Female: male	31:24	53:46	8:8
Age (yr)	24.3 ± 7.6	21.1 ± 7.8	19.4 ± 7.0	0.001	0.006	NS
BMI (kg/m2)	25.2 ± 4.7	25.0 ± 4.4	25.3 ± 4.5	NS	NS	NS
Systolic BP (mm Hg)	118.8 ± 14.3	121.1 ± 12.4	125.5 ± 13.9	NS	NS	NS
Diastolic BP (mm Hg)	69.2 ± 8.6	70.8 ± 7.7	69.4 ± 4.6	NS	NS	NS
Duration of DM (yr)		9.6 ± 8.0	10.5 ± 9.4			NS
HbA1c (%)	5.0 ± 0.3	8.1 ± 1.5	9.8 ± 2.7	<0.001	<0.001	0.008
CGMS glucose (mg/dl)	88.3 ± 9.9	165.8 ± 43.5	208.5 ± 64.2	<0.001	<0.001	0.01
C-peptide	0.88 ± 0.53	0.16 ± 0.19	0.12 ± 0.13	<0.001	<0.001	NS
UAE (mg/g)	7.8 (3.3, 27.1)	10.1 (3.5, 28.4)	50.8 (30.4, 2250.1)	0.02	<0.001	0.001
GFR (ml/min per 1.73 m2)	86.9 (62.6, 165.2)	108.3 (58.2, 220.0)	134/0 (76.9. 232.7)	<0.001	<0.001	NS

Comparisons are made between control subjects (group 1) and T1D subjects without (group 2) or with (group 3) albuminuria. Data are generally expressed as mean ± sd; for UAE and GFR, data are presented as median and range (minimum, maximum). P values are presented for between-group comparisons. BP, Blood pressure; NS, not significant. Statistically significant comparisons are designated in bold. 

Demographic and metabolic parameters for groups 1–3 are presented in Table 1. These groups were comparable with respect to gender, racial distribution, BMI, and blood pressure measurements; groups 2 and 3 were comparable for duration of disease. Both T1D groups were characterized by higher values for HbA1c and CGMS average capillary glucose and by lower C-peptide concentrations compared with controls, with a statistically significant worsening of glycemic control observed in group 3 (higher HbA1c and CGMS glucose) compared with group 2. UAE was significantly higher in groups 2 and 3 when compared with controls. GFR was also significantly higher in both T1D groups, when compared with control subjects; GFR was not statistically different, however, between groups 2 and 3.

Vitamin D deficiency, defined as a 25OHD level less than 20 ng/ml, was present in 12.7, 14.4, and 21.4% of subjects in groups 1, 2, and 3, respectively; vitamin D insufficiency, defined as a 25OHD level between 20 and 29 ng/ml, was present in an additional 25.5, 39.2, and 35.7% of subjects in groups 1, 2, and 3, respectively. Hence, optimal levels of 25OHD were seen in only 61.8, 46.4, and 42.9% of groups 1, 2, and 3 subjects. When the entire T1D cohort (n = 115) was compared with the control cohort, 25OHD concentrations were lower in participants with T1D (30.8 ± 12.2 vs. 36.2 ± 16.2 ng/ml, respectively; P = 0.059); only 47% of T1D participants were vitamin D sufficient, whereas 62% of control participants were vitamin D sufficient (P = 0.069). PTH concentrations were significantly higher in participants with T1D (18.5 ± 9.5 vs. 13.2 ± 8.4 pg/ml; P < 0.001), whereas 1,25(OH)₂D levels were not significantly different between the groups (49.7 ± 18.9 vs. 52.4 ± 25.6, respectively).

Table 2 presents data for vitamin D and related analytes when T1D subjects were subcategorized as normoalbuminuria (group 2) or albuminuric (group 3). Again, serum PTH concentrations were statistically higher in groups 2 and 3, compared with control subjects. Serum concentrations of 25OHD and 1,25(OH)₂D were not statistically different between groups in this subgroup analysis, although an increased prevalence of vitamin D deficiency + vitamin D insufficiency was noted in groups 2 and 3. Moreover, a marked increase in the urinary excretion of VDBP was apparent in the T1D groups, compared with the control group, whether reported as a urine VDBPCR or as total VDBP excretion per day. VDBP concentrations in plasma, however, were not statistically different across the three groups.

Table 2.

Comparison of vitamin D and related analytes in study subjects, grouped by UAE

Condition	Group 1 Control	Group 2 T1D + UAE ≤30 mg/g	Group 3 T1D + UAE >30 mg/g	1 vs. 2 P value	1 vs. 3 P value	2 vs. 3 P value
n	55	99	16
S-total calcium (mg/dl)	9.5 (8.4, 10.9)	9.5 (7.0, 11.0)	9.8 (8.9, 10.2)	NS	NS	NS
S-PTH (pg/ml)	11 (2, 37)	18 (1, 60)	20 (7, 35)	0.001	0.02	NS
S-25OHD (ng/ml)	34.0 (12, 75)	27.0 (8, 66)	28.5 (14, 44)	NS	NS	NS
Percent of group with 25OHD
≥30 ng/ml	61.8	46.4	42.9
20–29 ng/ml	25.5	39.2	35.7
<20 ng/ml	12.7	14.4	21.4
S-1,25(OH)2D (pg/ml)	43 (20, 146)	49 (13, 124)	46 (33, 76)	NS	NS	NS
P-VDBP (mg/dl)	69.4 (13.4, 276.1)	61.3 (11.1, 237.9)	33.4 (12.6, 161.1)	NS	NS	NS
U-VDBPCR (ng/g)	1949 (429, 387,636)	7043 (572, 146,745)	68844 (1333, 2,914,421)	<0.001	<0.001	<0.001
U-total VDBP (ng/d)	2727 (553, 385,310)	12707 (585, 372,110)	49,665 (1261, 5,976,750)	<0.001	<0.001	0.004

Comparisons are made between control subjects (group 1) and T1D subjects without (group 2) or with (group 3) albuminuria for concentrations of vitamin D and vitamin D-related parameters, measured in serum (S), plasma (P), or urine (U). Data are presented as median and range (minimum, maximum) for these nonnormally distributed variables. NS, Not significant. Statistically significant comparisons are designated in bold. 

Correlations between paired variables of interest are shown in Table 3. In many instances, significant relationships noted in groups 1 or 2 or in the overall study populations were not statistically significant in group 3, possibly due to insufficient power provided by the small sample size of this subgroup. For example, consistent with reports of others (20,21), 25OHD levels were inversely correlated with BMI (Table 3, no. 1), in both T1D subjects specifically and the study population overall. This relationship was not, however, statistically significant in group 3. Similarly, 1,25(OH)₂D levels were also negatively related to BMI in control subjects, T1D subjects, and in the population overall, although results in group 3 were not in agreement (Table 3, no. 3).

Table 3.

A comparison of observed correlations across groups 1, 2, and 3

	Paired comparison, showing Spearman’s Correlation (P value)	Group 1 Control (n = 55)	Group 2 T1D + UAE ≤30 mg/g (n = 99)	Group 3 T1D + UAE >30 mg/g (n = 16)	Group 2 + 3 All T1D (n = 115)	Overall (n = 170)
1	25OHD vs. BMI	−0.210 (0.123)	−0.262 (0.009)	−0.244 (0.400)	−0.276 (0.003)	−0.239 (0.002)
2	25OHD vs. HbA1c	−0.179 (0.191)	−0.017 (0.872)	−0.517 (0.058)a	−0.066 (0.484)	−0.149 (0.052)a
3	1,25(OH)2D vs. BMI	−0.329 (0.014)	−0.338 (<0.001)	0.211 (0.468)	−0.291 (0.002)	−0.301 (<0.001)
4	1,25(OH)2D vs. 25OHD	0.451 (<0.001)	0.190 (0.062)a	0.096 (0.744)	0.165 (0.078)	0.282 (<0.001)
5	1,25(OH)2D vs. PTHb	−0.356 (0.008)	0.027 (0.795)	0.051 (0.863)	0.018 (0.853)	−0.095 (0.220)
6	1,25(OH)2D vs. total VDBP	0.154 (0.270)	0.210 (0.043)	0.537 (0.047)	0.221 (0.021)	0.175 (0.027)
7	P-VDBP vs. 25OHD	0.232 (0.088)	0.044 (0.672)	0.284 (0.325)	0.113 (0.229)	0.162 (0.034)
8	P-VDBP vs. 1,25 (OH)2D	0.225 (0.099)	0.096 (0.349)	0.361 (0.204)	0.097 (0.300)	0.148 (0.054)a
9	P-VDBP vs. PTH	−0.146 (0.287)	−0.277 (0.006)	−0.064 (0.828)	−0.249 (0.008)	−0.215 (0.005)
10	VDBPCR vs. UAE	0.457 (<0.001)	0.475 (<0.0001)	0.321 (0.226)	0.539 (<0.001)	0.537 (<0.001)
11	VDBPCR vs. CGMS Glucose	0.025 (0.858)	0.019 (0.054)a	0.450 (0.080)	0.255 (0.006)	0.428 (<0.001)
12	VDBPCR vs. HbA1c	0.037 (0.790)	0.272 (0.007)	0.512 (0.043)	0.363 (<0.001)	0.464 (<0.001)
13	VDBPCR vs. C-peptide	0.106 (0.451)	−0.194 (0.056)a	−0.116 (0.668)	−0.178 (0.058)a	−0.356 (<0.001)
14	VDBPCR vs. PTH	0.059 (0.674)	0.113 (0.277)	0.051 (0.863)	0.079 (0.413)	0.185 (0.019)
15	VDBPCR vs. 25 OHD	0.054 (0.702)	0.029 (0.780)	−0.228 (0.318)	−0.028 (0.771)	−0.067 (0.393)
16	VDBPCR vs. 1,25 (OH)2D	0.202 (0.147)	0.157 (0.128)	0.414 (0.141)	0.164 (0.087)	0.147 (0.061)a

Spearman’s correlation coefficients and associated P values (in parentheses) are shown for paired variables of interest. Statistically significant relationships are designated in bold, including comparisons with P values between 0.05 and 0.06. No other interaction terms for any of these correlations were statistically significant. Total VDBP, Total urine VDBP per day. Additional abbreviations are defined in Tables 1 and 2. 

^aStatistically significant comparisons with P values between 0.05 and 0.06. 

^bA significant group difference of association between 1,25(OH)₂D and PTH was confirmed by using a linear regression (P = 0.04). 

A strong positive relationship between 1,25(OH)₂D and 25OHD was observed in control subjects; this relationship was less significant among subjects with T1D; moreover, this relationship was not present in group 3. Similarly, a strong inverse relationship between 1,25(OH)₂D and PTH was observed in control subjects but lost in T1D subjects. A significant group difference of association between 1,25(OH)₂D and PTH was confirmed by using a linear regression (Table 3).

Urinary excretion of VDBP (VDBPCR) was strongly positively related to UAE (analyzed as ACR) in all groups, again with the exception of group 3. Within the T1D subjects, VDBPCR was also positively related to indices of worsening glycemic control (higher CGMS average glucose, HbA1c) and negatively related to fasting C-peptide concentration, an indirect reflection of endogenous insulin secretion. Again, however, the inverse relationship between VDBPCR and C-peptide was not statistically significant in group 3. Interesting, both in the overall population and within group 3, an inverse relationship between serum 25OHD concentrations and HbA1c was also evident, although the statistical significance of this correlation was marginal. Among group 3 subjects specifically, a statistically significant positive association between urinary VDBP excretion and HbA1c or 1,25(OH)₂D was noted (Table 3, items 13 and 6; VDBPCR or total VDBP, respectively).

To examine the role of urine VDBP excretion in vitamin D deficiency, we performed multivariate analyses, using two categories of 25OHD concentration, specifically 30 ng/ml or greater (vitamin D sufficient or optimal range) and less than 30 ng/ml (vitamin D insufficient + deficient), as the outcome variable while using gender, race, age, season, disease, UAE, and urine VDBPCR as covariates. After multivariate logistic modeling, season (P = 0.003) and disease (T1D; P = 0.03) were identified as significant factors affecting vitamin D deficiency. The relative risk for vitamin D insufficiency/deficiency in T1D was 2.29 (95% confidence interval 1.06, 4.92). A significant linear relationship between VDBP excretion and vitamin D deficiency was not identified.

To determine those factors affecting urinary excretion of VDBP, we performed multivariate regression modeling, initially using gender, disease, age, PTH, HbA1c, CGMS glucose, 25OHD, 1,25(OH)₂D, GFR, and UAE as covariates. After backward elimination from the full model set, significant correlates of urinary VDBP excretion included HbA1c (P = 0.010; ß-coefficient = 0.208), CGMS capillary glucose (P = 0.047; β-coefficient =0.931), serum 1,25(OH)₂D concentrations (P = 0.037; β-coefficient = 0.607), and microalbuminuria (ACR > 30 mg/g; P = 0.004; β-coefficient = 1.312).

Conclusions

A number of studies demonstrate that osteopenia and osteoporosis are frequent complications of T1D (22,23,24,25) in both children (26,27,28) and adults (29). T1D is associated with decreased bone density (30,31,32,33) and a state of low bone turnover (31); lower bone mass usually develops within the first few years of T1D (26,29,34). Multiple factors may contribute to the development of this comorbidity, including vitamin D deficiency.

Although several studies report the presence of vitamin D deficiency or insufficiency in persons with T1D (1,2,5,35), not all do. A study by Bierschenk et al. (36) suggested that 25OHD deficiency was not specifically associated with T1D, when studied in individuals residing in a solar rich environment like Florida. Whereas mean 25OHD levels in our T1D population were very nearly significantly lower, this study groups reflects predominantly southern U.S. latitudes, perhaps counterbalancing other mechanisms for vitamin deficiency in diabetes. Nonetheless, the observed prevalence of both vitamin D insufficiency and deficiency was higher in groups 2 and 3 subjects, compared with control subjects.

It is also worth noting that 38.2% of persons in our control cohort (group 1) had vitamin D levels below the optimal concentration range of 30–80 ng/ml. This prevalence of vitamin D deficiency/insufficiency is somewhat less than the 56.4% prevalence recently reported in a population of 559 adolescents (45% African-American) living in the southeastern United States (37) and much lower than the 82% prevalence reported in a population of older adults (mean age ∼60 yr; 20% African-American), living in Georgia (38). The better vitamin D status in our population may reflect the higher percentage of Caucasians in this T1D-oriented study or may be age dependent.

Because megalin (LRP2) is involved in the constitutive reuptake of albumin and many smaller proteins from the glomerular filtrate, diminished functioning or availability of this scavenger receptor in diabetes could contribute to proteinuria. For example, about 95% of the filtered albumin is reabsorbed in the proximal tubule; data suggest that the majority of this reabsorption is mediated through the megalin-cubilin complex (39). Megalin mutations in humans, responsible for the Donnai-Barrow and faciooculoacousticorenal syndrome (40), result in renal disease and proteinuria with increased excretion of VDBP. Similarly, megalin-knockout mice exhibit proteinuria; characterization of the urine demonstrates an abundance of low-molecular-weight proteins, including VDBP and retinol binding protein (41,42). Furthermore, due to urinary losses of the VDBP + 25OHD complex, megalin deficient mice display skeletal abnormalities and 1,25(OH)₂D deficiency (13,41).

In T1D, previous studies have shown that certain megalin/cubilin ligands, such as retinol binding protein and transferrin, correlate with the degree of albuminuria or are elevated in T1D urine (43,44). Herein we report that VDBP concentrations are also increased in the urine of subjects with T1D, and these concentrations correlate with UAE. Moreover, highly significant positive relationships were identified between VDBP excretion and indices of worsening glycemic control, an independent risk factor for the development of diabetic nephropathy (45). In addition, VDBPCR concentrations were inversely correlated with fasting C-peptide concentrations, a marker of endogenous insulin secretion.

Our findings are also in keeping with data from the National Health and Nutrition Examination Survey, which demonstrated an inverse association between circulating 25OHD concentrations and HbA1c (46). In particular, among our albuminuric subgroup of T1D subjects, a similar inverse correlation between 25OHD concentrations and HbA1c was observed (R = −0.517, P = 0.058).

Exaggerated urinary loss of VDBP in T1D, particularly in persons with diabetic nephropathy, could contribute mechanistically to vitamin D deficiency in this disease. Ordinarily, the VDBP + 25OHD complex within the glomerular filtrate is endocytosed in the proximal tubule by megalin and its coreceptor, cubilin. Once within the epithelial cell, 25OHD is released, hydroxylated by 1α-hydroxylase to 1,25(OH)₂D, and returned to the circulation (11). As seen in control subjects (group 1), the significant positive correlation between plasma levels of 25OHD and 1,25(OH)₂D and the lack of a significant urinary loss of VDBP suggests that this process is very efficient under normal circumstances. In contrast, megalin null mice and mice in which megalin is not fully functional (i.e. CLC-5 knockout mice) experience increased urinary losses of both VDBP and 25OHD along with a marked increase in the renal expression of the 1α-hydroxylase gene (47); however, systemic concentrations of 1,25(OH)₂D are actually depressed in both conditions, compared with control animals (13,47). In the present study, we noted the emergence of a positive linear relationship between 1,25(OH)₂D and total urinary VDBP among the albuminuric group of T1D subjects. This finding could reflect decreased intracellular 25OHD due to urinary losses of the VDBP + 25OHD complex, leading to a compensatory up-regulation of renal 1α-hydroxylase activity in an attempt to maintain vitamin D homeostasis (48). VDBP binds both 25OHD and 1,25(OH)₂D; thus, urinary loss of VDBP complexes could also contribute to the loss of both forms of vitamin D.

Limitations of this study relate primarily to the fact that this study was conducted as a retrospective analysis of a preexisting specimen and data set. As such, we have no quantifiable information on the following: 1) dietary vitamin D or calcium intake other than self-reported use of vitamin or nutritional supplements; 2) sunlight exposure; or 3) physical activity levels. However, reported supplement use was minimal. And within this age group and geographic location, other contributing factors would very likely be equivalent across the control and T1D subgroups. Racial diversity of the study population is also limited, although, again, the racial distribution is comparable across the control and T1D study groups and not appreciably inconsistent with the demographic expectations for T1D (49). Finally, the subgroup of subjects with albuminuria is small, limiting the depth of analyses that can be conducted.

We also acknowledge that this analysis represents a cross-sectional, single point-in-time assessment of vitamin D status. Hence, we cannot discern between two possibilities: 1) that vitamin D insufficiency preceded and potentiated the progression of diabetic nephropathy in subjects with T1D or 2) that progressive renal involvement, characterized by increased VDBP excretion, contributed to the vitamin D insufficiency. In support of the first possibility, combined therapy with an angiotensin-1 receptor antagonist and an activated vitamin D analog has been shown to ameliorate diabetic nephropathy in animal models of both T1D (50) and T2D (51). However, a recent cross-sectional analysis of the 2001–2006 National Health and Nutrition Examination Survey reported a high prevalence of vitamin D deficiency (53.2 vs. 47.0%) and vitamin D insufficiency (37.2 vs. 36.4%) in diabetic adults, both with and without nephropathy, respectively (7). Moreover, in our study, increased excretion of VDBP along with a higher prevalence of vitamin D insufficiency was apparent in both groups 2 and 3.

In summary, we have identified a marked increase in urinary loss of VDBP in subjects with T1D, worsened in the presence of albuminuria. Because VDBP is critically involved in the delivery of 25OHD to the renal epithelial cell of the proximal tubule for activation to 1,25(OH)₂D, interruption of this delivery pathway would be expected to alter the intrarenal handling and processing of vitamin D. We question, therefore, whether urinary loss of VDBP might contribute, mechanistically, to vitamin D deficiency in this disease.