25-Hydroxyvitamin D Deficiency and Diabetes Predict Reduced BMD in Patients With Chronic Kidney Disease†
The authors state that they have no conflicts of interest.
In this study of 242 patients with renal failure, women, patients with diabetes, and patients on peritoneal dialysis had the highest risk of 25-hydroxyvitamin D deficiency. Levels correlated positively to BMD Z scores, and hip BMD was inversely related to prevalent fracture. Increasing 25-hydroxyvitamin D levels may benefit these patients.
Introduction: 25-Hydroxyvitamin D deficiency (<37 nM) is common in patients with chronic kidney disease (CKD) stage 5 (glomerular filtration rate < 15 ml/min/1.73 m2 or on dialysis), but it is unclear if this deficiency is associated with bone disease and if supplementation is warranted.
Materials and Methods: Blood samples were collected on 242 patients with CKD stage 5 caused by type 1 diabetes (33%), type 2 diabetes (2%), and other causes (65%), who were about to undergo kidney or kidney pancreas transplantation. Prevalent spinal fracture was assessed by X-ray and BMD by DXA.
Results: 25-Hydroxyvitamin D deficiency was present in 28% of patients with diabetes versus 12% without (p < 0.0001). Patients on peritoneal dialysis (PD) had lower levels of 25-hydroxyvitamin D than patients on hemodialysis (HD; 49 ± 26 versus 77 ± 34 nM; p < 0.0001), and women had lower levels than men (51 ± 25 versus 77 ± 35 pM; p < 0.0001). BMD Z scores were within 1 SD of the mean at all sites, except in patients with diabetes (femoral neck Z score, −1.07 ± 1.2; p < 0.0001) and patients who had undergone parathyroidectomy (lumbar spine Z score, 1.03 ± 1.34, femoral neck Z score, 1.24 ± 1.35; p < 0.001 and p < 0.0001, respectively). In multiple stepwise linear regression analysis, levels of 25-hydroxyvitamin D correlated positively and intact PTH (iPTH) correlated negatively to Z scores at the lumbar spine and wrist. Time on dialysis correlated positively to Z scores at the femoral neck and lumbar spine. Diabetes and serum alkaline phosphatase levels correlated negatively with Z scores at the femoral neck. Lower femoral neck BMD was associated with an increased prevalence of vertebral fracture and fragility fracture at any site (p = 0.03 and p < 0.03, respectively).
Conclusions: This study of patients with CKD stage 5 identifies women, patients with diabetes, and patients on PD as being at particular risk of 25-hydroxyvitamin D deficiency. We describe positive associations of 25-hydroxyvitamin D levels and BMD Z scores and an association between femoral neck BMD and fragility fracture at any site. Treatment to improve 25-hydroxyvitamin D levels may benefit these patients.
Bone disease is observed in 75–100% of patients with chronic kidney disease (CKD) at glomerular filtration rates (GRFs) <60 ml/min.(1) Deficiency of 1,25-dihydroxyvitamin D contributes to the abnormal bone turnover and mineralization found in renal osteodystrophy (ROD), and active vitamin D sterols are commonly used in the treatment of patients with CKD stage 5 (GFR < 15 ml/min/1.73 m2 or on dialysis). 25-Hydroxyvitamin D deficiency is relatively common in predialysis patients, particularly those with diabetes, and has been associated with osteomalacia.(2) Patients on dialysis have also been reported to have low levels of 25-hydroxyvitamin D,(3–5) but the role of treatment is unclear because deficiency of renal 1α-hydroxylase activity results in reduced conversion of precursors to 1,25-dihydroxyvitamin D. One study of patients on dialysis has reported a positive correlation of 25-hydroxyvitamin D levels to histological features of ROD (osteomalacia and fibrosis),(6) whereas a negative correlation to PTH and radiological features of ROD (subperiosteal resorption and Looser's zones) has been reported in others.(6,7) For patients with CKD stage 5, it is unclear if 25-hydroxyvitamin D levels are associated with BMD or fracture, but in the general adult population, 25-hydroxyvitamin D levels are positively associated with BMD irrespective of age and sex.(8,9) Both 25-hydroxyvitamin D levels and fracture have been reported to show seasonal periodicity, including in temperate areas of southeastern Australia.(10)
Even in the presence of CKD stage 5, therapy to increase levels of 25-hydroxyvitamin D may have therapeutic potential. More than 20 years ago, a small study of patients on hemodialysis (HD) reported that 25-hydroxyvitamin D administration lead to a 3-fold increase in 1,25-dihydroxyvitamin D levels and increases in other vitamin D metabolites.(11) More recently, 1α-hydroxylase activity has been detected in a number of tissues including breast, prostate, colon, skin, osteoblasts, the parathyroids, and vascular smooth muscle cells, suggesting that 25-hydroxyvitamin D has paracrine roles independent of renal conversion to 1,25-dihydroxyvitamin D.(12–14)
We collected data on 242 patients with CKD stage 5 of whom most were on HD or peritoneal dialysis (PD) to assess the prevalence of 25-hydroxyvitamin D deficiency and whether particular patient groups were at increased risk. We also studied associations of 25-hydroxyvitamin D levels with biochemical parameters, BMD, and fracture.
MATERIALS AND METHODS
The study population consisted of 242 patients with CKD stage 5, caused by type 1 diabetes in 82, type 2 diabetes in 4, and other causes in 156 patients. Patients were admitted to Westmead Hospital, Sydney (the Australian National Renal Pancreas Transplant Center) between 2002 and 2005 for renal pancreas transplantation (n = 70) or for renal transplantation (n = 172). Most patients were on thrice weekly HD or on PD (continuous ambulatory peritoneal dialysis or automated peritoneal dialysis), but 15% had not started dialysis. Characteristics of the study population are shown in Table 1. Ethnicity was white 88%, Asian 6%, Arabic 3%, and 3% were of other ethnicity. Phosphate binders were taken by more than two thirds of the patients. Of these patients, calcium carbonate was taken solely or in combination by 64%. Approximately one third of patients were taking oral calcitriol, but no patients were being treated with ergocalciferol, cholecalciferol, cinacalcet HCl, or intravenous vitamin D analogs.
Table Table 1.. Characteristics of the Study Population
Blood was collected on the morning of all elective admissions, and ranges are those of the local laboratory. For patients on HD, the last dialysis varied from <1 to 3 days before venipuncture. In addition to general biochemistry including serum alkaline phosphatase (ALP; reference range: 30–115 U/liter), assays were performed for 25-hydroxyvitamin D (reference range: 37–131 nM; DiaSorin) and 1,25-dihydroxyvitamin D (reference range: 26–120 pM; DiaSorin). Deficiency of 25-hydroxyvitamin D was defined as a level below the assay lower range of 37 nM (15 ng/ml) and insufficiency as levels from 37 to 75 nM (15–30 ng/ml). Intact PTH (iPTH; reference range: 1–6.8 pM), estradiol (all patients), and total testosterone (men only) were assayed using the Immulite system (Diagnostic Products Corp.). Assays were performed for follicle-stimulating hormone (FSH) and leuteinizing hormone (LH) (Axsym System; Abbott Laboratories), and a two-site immunometric assay (recognizing the intact molecule) was used to assay osteocalcin (reference range for non-CKD patients: 1.1–7.2 ng/ml; Nichols Advantage).
DXA using a Norland XR36 densitometer was performed at the lumbar spine, femoral neck, and distal radius and ulna within 2 weeks of admission. Z scores were adjusted for age and sex using Geelong (Australia) data for women and Boston (MA, USA) data for men. Lateral X-rays of the thoracic and lumbar spine were assessed by an experienced clinician and a radiologist without prior knowledge of the patient, using semiquantitative criteria developed by Genant et al.(15) Any suspected fracture was assessed quantitatively and for inclusion was defined as a reduction in vertebral height of ≥20%.
Descriptive statistics, two-sample t-tests, bivariate correlations (using Pearson if both variables were approximately normally distributed and Spearman rank correlation otherwise), and multiple stepwise linear regression analyses were performed using SPSS 14.0 (SPSS) The level of significance and inclusion in regression analyses was set at p ≤ 0.05, and unless otherwise stated, values are given as mean ± SD. Because type 2 diabetes was uncommon in this group, all patients with diabetes were analyzed together.
Patients admitted to the transplant unit are informed that data are collected for clinical studies with approval of the local ethics committee. All investigations performed for this study were part of routine care and patients were followed in the Renal Metabolic Bone Clinic after transplantation.
The mean level of 25-hydroxyvitamin D was 67 ± 34 nM, and patients with diabetes had significantly lower levels than other patients (54 ± 30 versus 75 ± 33 nM; p < 0.0001). 25-Hydroxyvitamin D deficiency was present in 28% of patients with diabetes versus 12% without (p < 0.0001), and insufficiency was present in a further 50% with diabetes versus 41% without. Patients on HD had higher levels of 25-hydroxyvitamin D than patients on PD (77 ± 34 versus 49 ± 26 nM; p < 0.0001) or patients who had not started dialysis (56 ± 20 nM; p = 0.001) and levels of 25-hydroxyvitamin D were lower for women than men (51 ± 25 versus 77 ± 35 nM; p < 0.0001). In a multiple stepwise linear regression analysis, the presence of diabetes, use of PD, and female sex independently predicted lower levels of 25-hydroxyvitamin D (p < 0.0001). Blood samples for 25-hydroxyvitamin D levels were collected throughout the year (50% in spring and summer and 50% in autumn and winter). No association was detected between vitamin D levels and racial background (ANOVA).
Laboratory data are presented in Table 2. Mean levels of 1,25-dihydroxyvitamin D were in the low normal range. Patients taking calcitriol capsules had higher levels than patients not on supplementation (42 ± 34 versus 28 ± 30 pM; p < 0.05). Levels of iPTH and the calcium phosphate product did not differ significantly for patients taking or not taking vitamin D supplements. Whereas the mean level of iPTH was 6.5 times the normal upper range of the assay, 15% of patients had levels of iPTH <6.8 pM, which is generally associated with low bone turnover or adynamic bone disease. Lower iPTH levels were not associated with age, sex, the presence of diabetes, or dialysis modality. Levels of ALP and of osteocalcin correlated to iPTH (r = 0.43, p < 0.0001 and r = 0.7, p < 0.0001, respectively). No significant correlations were detected between levels of 25-hydroxyvitamin D or 1,25-dihydroxyvitamin D and iPTH for the group as a whole or when patients with and without diabetes were analyzed separately.
Table Table 2.. Laboratory Values
For the 242 patients, BMD Z scores were within 1 SD of the mean at all sites. Women had lower BMD (g/cm2) than men at the femoral neck (p < 0.01), lumbar spine (p = 0.08), and wrist (p < 0.0001). Z scores of patients with diabetes were significantly lower than the mean and Z scores of parathyroidectomized patients were significantly above the mean (Table 3). For the group as a whole, Z scores at the lumbar spine and femoral neck correlated to time on dialysis (r = 0.22, p < 0.001 and r = 0.27 p < 0.0001, respectively), but this association was not detected for patients with diabetes when analyzed separately.
Table Table 3.. Age- and Sex-Adjusted BMD Z Scores
Z scores correlated to levels of 25-hydroxyvitamin D at the lumbar spine (r = 0.24, p = 0.0005), femoral neck (r = 0.23, p ≤ 0.001) and wrist (r = 0.22, p < 0.01), and at the femoral neck were associated with phosphate levels (p < 0.05). Z scores were negatively correlated to levels of iPTH and ALP at all sites (PTH: lumbar spine p < 0.01, femoral neck p < 0.001, wrist p < 0.01 and ALP: lumbar spine p < 0.05, femoral neck p < 0.001, wrist p < 0.005). When patients who had undergone parathyroidectomy were excluded from the analysis, iPTH levels were negatively but nonsignificantly correlated to Z scores at the lumbar spine and femoral neck, with a significant negative correlation at the wrist (p < 0.03). Z scores were not correlated to the use of calcitriol, mode of dialysis, or serum levels of osteocalcin, corrected calcium, or 1,25-dihydroxyvitamin D. For men, Z scores at the wrist were positively correlated to testosterone levels (p < 0.005), but estradiol levels were not associated with BMD at any site. Women in this study were 43 ± 12 years of age, and hypogonadism (assessed by menstrual history and levels of estradiol, LH, and FSH) was not predictive of BMD.
To test the hypothesis of an association between levels of 25-hydroxyvitamin D and BMD, multiple stepwise linear regression analysis was used with age- and sex-adjusted Z scores as the dependent variables. The presence or absence of diabetes, time on dialysis, levels of 25-hydroxyvitamin D, iPTH, phosphate, and ALP were considered as possible predictors of Z scores (Table 4). Levels of 25-hydroxyvitamin D predicted Z scores at the lumbar spine and wrist, and iPTH levels correlated negatively with Z scores at these sites. Time on dialysis correlated positively with Z scores at the spine and hip but negatively at the wrist. The presence of diabetes and levels of ALP correlated negatively with femoral neck Z scores.
Table Table 4.. Results of Multiple Stepwise Linear Regression Analyses With Age- and Sex-Adjusted BMD Z Scores as Dependent Variables and Presence of Diabetes, Time on Dialysis, Levels of 25-Hydroxyvitamin D, iPTH, Phosphate, and ALP Considered as Possible Predictors of Z scores
Based on history and radiological assessment, 37% of patients had sustained at least one fragility fracture (one or more vertebral fractures in 28% and nonvertebral fragility fracture in 13%). Lower BMD at the femoral neck was associated with an increased prevalence of vertebral fracture or history of fragility fracture at any site (p = 0.03 and p < 0.03, respectively) with a similar trend for BMD at the lumbar spine. No significant associations were observed for vertebral fracture or fragility fracture at any site and levels of 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, ALP, osteocalcin, iPTH, time on dialysis, or presence of diabetes.
In this study of patients with CKD stage 5, levels of 25-hydroxyvitamin D were positively associated with BMD at all sites and at the wrist and lumbar spine after adjustment for covariates. We are not aware of other studies that have reported this finding in patients with CKD stage 5. In one study of patients on HD conducted during winter, a positive association was observed between 25-hydroxyvitamin D levels and BMD at the radius, but this association was not significant after adjustment for covariates.(16) The authors also reported an association of 25-hydroxyvitamin D levels and calcaneal broadband ultrasound attenuation, a technique that has been used to predict overall fracture risk.(17,18) The median level of 25-hydroxyvitamin D (17 nM; range, 4.1–78.5 nM) was considerably lower than in this study (median, 61 nM; range, 10–176 nM); therefore, it is possible that winter clustering of low 25-hydroxyvitamin D results may have obscured an interactions of 25-hydroxyvitamin D and BMD at other sites.
Patients in this study had normal range Z scores, with positive values at both the lumbar spine and wrist. In comparison, other studies have reported the lumbar spine and femoral neck Z scores of patients on PD to be normal or lower than the reference population,(19,20) and the Z scores of patients on HD to be normal or below the reference population at the hip, midshaft, and ultradistal radius.(21,22) However, we did find that patients with diabetes had significantly lower BMD at all sites, particularly the hip, with mean Z scores < −1. This finding is consistent with a small study of 19 patients undergoing kidney pancreas transplantation that reported lower T scores at the femoral neck than at the lumbar spine.(23) We found that parathyroidectomy was associated with higher levels of BMD and, because iPTH levels were inversely related to Z scores in nonparathyroidectomized patients, this was unlikely to be caused by an anabolic effect of PTH before parathyroidectomy. Together with an inverse relationship between Z scores at all sites and levels of ALP, these data suggest that high bone turnover impacts negatively on BMD in patients with CKD stage 5 and that parathyroidectomy facilitates remineralization, a conclusion supported by longitudinal and other cross-sectional studies.(22,24,25) Time on dialysis correlated positively to Z scores at the hip and lumbar spine, and the latter association remained significant after adjustment for covariates. This association was not detected for the subgroup of patients with diabetes. Other studies have reported variable associations of time on dialysis and BMD, including a weak positive association for time on PD and lumbar spine Z scores and a negative correlation for time on HD and midradius Z scores.(19,21) Together with our finding of a negative correlation with wrist Z scores, these data may indicate differential effects of time on dialysis at axial and appendicular sites.
The 28% prevalence of vertebral fracture in this study compares with a prevalence of 21% reported for 187 Japanese men on HD.(26) In that study, vertebral fracture prevalence for patients 40–49 years of age was three times that of healthy Japanese men. Analyzed by tertiles of iPTH, fracture was 2.4 times more likely if iPTH levels were <61 pg/ml (6.5 pM) and 1.6 times more likely if iPTH levels were >202 pg/ml (21 pM). A similar but weaker association of PTH and vertebral fracture was reported in an analysis of data on 9007 patients supplied by the U.S. Renal Data System (USRDS),(27) with the lowest fracture risk around a PTH concentration of 300 pg/ml (32 pM). In this study, no associations were detected between vertebral fracture and sex, presence of diabetes, or iPTH levels. However, a significant association was detected between femoral neck BMD and prevalent vertebral fracture or fragility fracture at any site, with a similar trend for lumbar spine BMD. In the elderly and in patients with CKD, degenerative changes and soft tissue calcification may result in spuriously elevated BMD values at the lumbar spine, which may explain inconsistent associations of lumbar spine BMD and vertebral fracture in other studies.(26,28) In contrast, measurement of BMD at the hip is a reliable site in elderly subjects and in people with severe lumbar osteoarthritis and has been reported to predict fracture at any site.(29,30)
The definition of an optimal level of 25-hydroxyvitamin D is controversial, even in people without CKD. Although the lower range for the assay used in this study was 37 nM (15 ng/ml), there is reasonable agreement that levels >50 nM (20 ng/ml) are necessary for bone health and fracture reduction. In the general population, suggested levels have varied from 50–80 nM (20–32ng/ml)(31) to levels >75–100 nM (30–40 ng/ml).(32) Several investigators have classified levels from 50–100 nM (20–40 ng/ml) as hypovitaminosis D. This definition is based on the inverse relationship of PTH and 25-hydroxyvitamin D levels observed within that range, and levels > 100 nM (40 ng/ml) have been suggested as optimal.(33,34) In early CKD, target values > 75 nM (30 ng/ml) have been suggested, based on an inverse relationship of PTH reported up to that level of 25-hydroxyvitamin D.(32) For patients on HD, Coen et al.(35) reported reduced mineralization and bone formation at levels <50 nM (20ng/ml) and reduced bone turnover at levels ≥100 nM (40 ng/ml). These findings, which were independent of PTH or 1,25-dihydroxyvitamin D levels, lead to a suggested optimal range for patients on HD of 50–100 nM (20–40 ng/ml). For patients with CKD stage 5 in this study, we did not detect a relationship between levels of PTH and vitamin D.
25-Hydroxyvitamin D levels below the normal lower range of the assay were common in this study, despite the patients being relatively young and healthy compared with the general dialysis population. These results are consistent with findings in the study of Coen et al., in which 28% of patients on HD had levels <37 nM (15 ng/ml)(35) and with a study conducted on predialysis patients with severe CKD, in which 17% were reported to have 25-hydroxyvitamin D insufficiency (defined in that study as levels of 25–70 nM).(4) A recent study of 69 patients on HD in Budapest reported a 59% rate of 25-hydroxyvitamin D deficiency <20 nM (8 ng/ml) and levels of 20–30 nM (8–12 ng/ml) in a further 20% of patients.(16) The fact that bloods were collected in midwinter may explain the high level of vitamin D deficiency. In the United States, modest seasonal changes in levels of 25-hydroxyvitamin D have been reported in patients with CKD.(4)
In this study, levels of 25-hydroxyvitamin D were lower in women, patients with diabetes, and patients on PD. Sex differences for 25-hydroxyvitamin D deficiency have not previously been reported for patients on dialysis, but in some epidemiological studies, inactive elderly people and young women have lower 25-hydroxyvitamin D levels than the rest of the population.(36) An association of diabetes and hypovitaminosis D has been reported in predialysis patients,(5) with likely contributing factors including poor nutrition, fat malabsorption, and catalysis of 25-hydroxyvitamin D to 24,25 dihydroxyvitamin D by extrarenal 24-hydroxylase. For patients on PD, loss of 25-hydroxyvitamin D and D binding protein has been reported to occur in the peritoneal dialysate effluent,(37) and severe deficiency was reported to affect 28 of 29 patients in a recent study.(38) However, this mechanism remains controversial, because in an earlier report, neither a loss of vitamin D nor D binding proteins was detected in patients on PD, despite large dialysate protein losses.(39) In addition to the factors listed above, patients with CKD are encouraged to avoid prolonged sun exposure, to wear hats, and to use effective sun block preparations to reduce the risk of skin cancer. Few foods in Australia are fortified with vitamin D and assiduously following such advice is likely to result in vitamin D insufficiency, despite relatively high ambient UVB radiation levels in locations such as Sydney at latitude 34° south.
This study has several limitations. The cross-sectional design means that associations between levels of vitamin D and BMD do not prove a causal relationship. A prospective, randomized, placebo-controlled trial of vitamin D therapy would be required to assess the influence of improved levels of 25-hydroxyvitamin D on BMD or fracture. Prospective studies would also be required to assess whether interventions to improve 25-hydroxyvitamin D levels might influence levels of PTH, 1,25-dihydroxyvitamin D, or patient-based endpoints. When renal bone disease is present, controversy exists regarding the interpretation of BMD measured by DXA, and for patients with CKD, abnormalities of bone turnover and architecture may also predispose to fracture.(40) The patients in this study were selected from the general dialysis population but tended to be younger with less comorbidity than patients who were not awaiting transplantation. Additionally, one third of these patients had type 1 diabetes, and there was a low incidence of CKD caused by type 2 diabetes.
In summary, this study of patients with CKD stage 5 identified women, patients on PD, and patients with diabetes as prone to 25-hydroxyvitamin D deficiency. Lower BMD Z scores were associated with the presence of diabetes, higher bone turnover (assessed by levels of iPTH and ALP), and lower 25-hydroxyvitamin D levels, and femoral neck BMD was inversely related to prevalent vertebral and previous fragility fracture. After consideration of standard risk factors predisposing to lower BMD, these data support special care being given to the at-risk patient groups we have identified. Treatment to improve 25-hydroxyvitamin D levels may benefit these patients.