Printer Friendly

Low 25-hydroxyvitamin D and risk of type 2 diabetes: a prospective cohort study and metaanalysis.

The pathogenesis of type 2 diabetes mellitus involves the development of a relative deficiency in insulin secretion and insulin resistance (1). Deficient vitamin D status has been associated with decreased insulin secretion and increased insulin resistance in animals and humans (2-9). Moreover, substitution with vitamin D in the deficient state has been associated with improvement in insulin secretion and glucose tolerance (5, 7,9). These studies thus suggest a link between vitamin D deficiency and type 2 diabetes.

Observational and randomized studies on vitamin D concentrations or intake and risk of type 2 diabetes have been contradictory (10). In general, observational studies suggest that higher plasma 25-hydroxyvitamin D [25(OH)D] [4] concentrations and higher vitamin D intake are associated with lower risk of type 2 diabetes. However, randomized studies do not show an effect of vitamin D supplementation on low risk of type 2 diabetes. Several factors have been proposed to explain these seemingly contradictory results, such as residual confounding in observational studies and insufficient doses in randomized studies. It is thus unclear at present whether low plasma 25(OH)D concentrations are associated with increased risk of type 2 diabetes.

We tested the hypothesis that low plasma 25(OH)D is associated with increased risk of type 2 diabetes in the general population. For this purpose, we studied 9841 white individuals from the Copenhagen City Heart Study followed for up to 29 years. We used seasonally unadjusted clinical categories of [greater than or equal to] 20 [micro]g/L [[greater than or equal to] 50 nmol/L] (sufficient), 10-19.9 [micro]g/L [25-49.9 nmol/L] (insufficient), 5-9.9 [micro]g/L [12.5-24.9 nmol/L] (deficient), and <5 [micro]g/L [<12.5 nmol/L] (severely deficient), as well as concentrations adjusted for seasonal variation. Furthermore, the association of low plasma 25(OH)D concentrations with increased risk of type 2 diabetes was summarized in a metaanalysis including present and previous studies.

Materials and Methods


The Copenhagen City Heart Study is a prospective cohort study of the Danish general population initiated in 1976-1978 with follow-up examinations in 1981-1983, 1991-1994, and 2001-2003 (11). Individuals 20-100 years of age were drawn randomly from the national Danish Central Person Register and invited to participate; all inhabitants in Denmark are uniquely identified through their central person registration number that also holds information on date of birth and sex.

The present study included 9841 participants from the 1981-1983 examination (18 089 invited; 70% response rate) who were free of type 2 diabetes at baseline, had a nonfasting plasma glucose < 198 mg/dL [<11 mmol L] at baseline (fasting glucose concentrations were not available), and had available plasma samples for 25(OH)D measurement.

A Danish ethics committee approved the study (KF100.2039/91 and KF01-144/01). Participants provided written informed consent.


Plasma samples collected at baseline in 1981-1983 were stored at -20[degrees]C until 2009-2010, when 25(OH)D was measured with the DiaSorin Liaison 25(OH)D Total assay (12). Assay precision was tested daily, and assay accuracy was tested monthly with an external quality control program. The interassay CV was 10% for low-concentration controls [approximately 16 [micro]g/L (40 nmol/L)] and 8% for high-concentration controls [approximately 54 [micro]g/L (135 nmol/L)].


Variables were ascertained in 1981-1983, 1991-1994, and 2001-2003 (11) and used as time-varying variables in multivariable adjusted models. Information on smoking habits was obtained from self-reported questionnaires completed together with an examiner on the day of attendance. Participants also reported their level of income (high, medium, or low) and duration and intensity of leisure-time physical activities (h/week) in self-reported questionnaires reviewed together with an examiner on the day of attendance. Body mass index (BMI) was calculated as measured weight (kilograms) divided by measured height (meters) squared.


Incident type 2 diabetes was self-reported diabetes and use of antidiabetic medicine at follow-up examination (1991-1994 or 2001-2003), nonfasting glucose >198 mg/dL [>11 mmol/L] at follow-up examination, or information on incident diagnoses of type 2 diabetes (WHO, International Classification of Diseases, Revision 8, code 250, and Revision 10, codes E11, E13, and E14) collected and verified by reviewing hospital admissions and diagnoses entered in the national Danish Patient Registry and by reviewing the national Danish Causes of Death Registry. Follow-up time for each subject began at the day of blood sampling in 1981-1983 and ended at diagnosis of type 2 diabetes (n = 810), death (n = 5908), emigration (n = 54), or August 2010, whichever occurred first. The median follow-up time was 20 years (range 0.03-29). Follow-up was 100% complete; that is, we did not lose track of even a single individual.


We divided baseline 25(OH)D into the following a priori seasonally unadjusted clinical categories of [greater than or equal to] 20 [micro]g/L [[greater than or equal to] 50 nmol/L] (sufficient), 10-19.9 [micro]g/L [25-49.9 nmol/L] (insufficient), 5-9.9 [micro]g/L [12.5-24.9] nmol L (deficient), and <5 [micro]g L [<12.5 nmol L] (severely deficient). In addition, because concentrations of 25(OH)D were expected to vary according to time of year due to the high-latitude geographical position of Denmark, we used seasonally adjusted 25(OH)D concentrations. Two strategies were applied to adjust for the seasonal variation in vitamin D. First, we used unadjusted 25(OH)D concentrations in regression analyses, while adjusting for calendar month of blood draw. Second, we obtained calendar month-specific cutpoints by assigning subjects to quartile categories within the same month of sample collection (see Supplemental Table S1, which accompanies the online version of this article at content/vol59/issue2). For trend tests, individuals in each group were assigned the median value of their group, as either absolute values or percentiles. As a supplement to these analyses, we also compared participants with plasma 25(OH)D >30 [micro]g/L [>75 nmol/L] with participants with plasma 25(OH)D of 20-30 [micro]g/L [50-75 nmol/L], as it has been suggested that the non-calcemic benefits of vitamin D may be maximized when 25(OH)D is >30 [micro]g/L [>75 nmol/L] (13). We chose to carry out analyses using both clinical categories with absolute values and month-specific quartiles. Although month-specific quartiles may be more suitable for biological hypothesis testing, the clinical categories give information that facilitates comparability between studies, and absolute values are also those used clinically, making absolute values transferable to the everyday activities of clinicians.

To evaluate whether storage time was associated with median concentrations of plasma 25(OH)D, we also measured plasma 25(OH)D in 400 participants without diabetes, cancer, heart disease, or other chronic diseases participating in the 1981-1983, 1991-1994, and 2001-2003 examinations of the Copenhagen City Heart Study.

We estimated cumulative incidences using the competing risk proportional subhazard models by the method of Fine and Gray (14), in which competing risk of death was accounted for. The analyses were adjusted for age and year of birth to account for calendar effects. We used age as time scale. The cumulative incidence functions were plotted by seasonally unadjusted clinical categories and seasonally adjusted percentile categories.

We used Cox proportional hazards regression to estimate hazard ratios with 95% CI for incident type 2 diabetes. We used age as time scale with delayed entry (left truncation). Thus, age differences were automatically adjusted for, and analyses are referred to in text, tables, and figures as age adjusted. Multivariable adjusted Cox regression models included (a) risk factors for type 2 diabetes as age, sex, smoking status (never/ ever), BMI, and duration and intensity of leisure time physical activities, (b) income as a measure of social status, and (c) calendar month of blood draw (the latter only for models with clinical categories) as a confounder for 25(OH)D concentrations. We tested for interactions using likelihood ratio tests with Cox regression models including and excluding multiplicative 2-factor interaction terms, the latter nested in the former model. In interaction analyses and stratified analyses, we used [log.sub.2]-transformed values of plasma 25(OH)D, whereby a 1-unit decrease corresponds to a 50% lower concentration of plasma 25(OH)D. The proportional hazards assumption was assessed in Cox regression models graphically by plotting -ln[-ln-(survival)] vs ln(analysis time); we detected no violations of the proportional hazards assumption. The data were 99.8% complete in relation to the included variables (see online Supplemental Table S2); the missing data were imputed using multivariable chained imputation (mi impute chained) where age and sex were independent variables and BMI, duration and intensity of leisure time physical activities, and income were dependent variables in the model.

We analyzed the data with the statistical package Stata 12.1, including the metaanalysis described below.


We identified relevant peer-reviewed studies on the association between plasma 25(OH)D concentrations and risk of type 2 diabetes by an electronic search of published articles in PubMed up to June 30, 2012, using combinations of the following keywords: ("Vitamin D"[Mesh] OR "25-hydroxyvitamin D" [Supplementary Concept] or "serum 25-hydroxyvitamin D" or "25-hydroxyvitamin D3" or "vitamin D3") AND ("Diabetes Mellitus"[Mesh] or "diabetes"). Inclusion criteria were prospective design; only type 2 diabetes as an endpoint; a general population sample or subsample, not selected on the basis of presence of disease; and information on effect estimates of the association of 25(OH)D concentrations with risk of type 2 diabetes. In total 1335 studies were identified, 32 articles were retrieved for full-text review, and a further 19 articles were excluded after review due to wrong endpoint, no measurement of plasma 25(OH)D, and/or cross-sectional design (see online Supplemental Fig. S1). This search strategy identified 13 articles representing 15 studies on the association of 25(OH)D plasma concentrations with risk of type 2 diabetes (15-27).

Data from each study were extracted by SA and confirmed by BGN. The extracted data included first author; publication year; cohort size and source; reported follow-up time; design; method of vitamin D measurement; method of 25(OH)D categorization; estimates of the association between 25(OH)D concentrations and outcome; ascertainment of diagnosis; and adjustment for age, sex, overweight or obesity, smoking, and physical activity, as these variables are known risk factors for type 2 diabetes and vitamin D deficiency, and season of blood draw, which is associated with plasma vitamin D concentrations. We converted the risk estimates from individual studies to risk estimates for top vs bottom quartiles to obtain more robust synthesized risk estimates (28). For 3 studies, this conversion was not possible (15, 18, 21), and the corresponding authors were contacted to obtain the risk estimates. Some studies did not report mean or median concentrations of 25(OH)D, and in these studies mean concentrations were estimated from the reported distribution of 25(OH)D (17, 25, 26).

We performed the meta-analysis using fixed and random-effect models (29) and calculated random-effect weights using the DerSirmonian and Laird model. Heterogeneity was assessed by the Q statistic and its extent was quantified by [I.sup.2] (the fraction of between study variability due to heterogeneity) (30). Publication bias was evaluated by funnel plots, Begg rank correlation test, and Egger regression test.



Table 1 and online Supplemental Table S3 summarize baseline characteristics by plasma 25(OH)D concentrations. Low concentrations of 25(OH)D were associated with high age, smoking, high BMI, low income, low-duration leisure time physical activity, and blood sampling in winter. The association of 25(OH)D concentrations with BMI showed decreasing 25(OH)D in participants with increasing BMI (trend, P = 2 x [10.sub.-41]), but underweight participants had lower concentrations of 25(OH)D than normal-weight participants (see online Supplemental Fig. S2). The median 25(OH)D concentration was 16 [micro]g/L [41 nmol/L] among all participants and 14 [micro]g/L [36 nmol/L] among those who later developed type 2 diabetes. A total of 810 incident cases of type 2 diabetes occurred among 9841 participants during up to 29 years of follow-up. For 400 healthy participants, we had measurements of plasma 25(OH)D from 1981-1983, 1991-1994, and 2001-2003, which showed that median concentrations were relatively stable, i.e., storage time did not systematically associate with lower concentrations of 25(OH)D (see online Supplemental Fig. S3).

The cumulative incidence of type 2 diabetes increased with decreasing concentrations of baseline plasma 25(OH)D expressed in clinical categories (trend, P = 3 x [10.sup.-5]) and expressed in seasonally adjusted quartiles (P = 2 x [10.sup.-6]) (Fig. 1). Multivariable adjusted hazard ratios for type 2 diabetes increased with decreasing concentrations of 25(OH)D by clinical categories and seasonally adjusted quartiles, and were 1.22 (95% CI 0.85-1.74) for 25(OH)D <5 [micro]g/L [<12.5 nmol/L] vs [greater than or equal to] 20 [micro]g/L [50 nmol/L], and 1.35 (1.09-1.66) for lowest vs highest quartile (Fig. 2). Additional analyses including the clinical category of 25(OH)D > 30 [micro]g/L [> 75 nmol/L], consisting of 985 participants, showed multivariable adjusted hazard ratios for type 2 diabetes of 0.91 (0.67-1.25) for 25(OH)D >30 [micro]g/L [>75 nmol/L] vs 30 [greater than or equal to] 25(OH)D [greater than or equal to] 20 [micro]g/L [75 > 25(OH)D > 50 nmol/L]. The use of 25(OH)D > 30 g/L [ > 75 nmol/L] as the reference value showed results similar to those of the above analyses (see online Supplemental Fig. S4).

The multivariable adjusted hazard ratio for type 2 diabetes for a 50% lower concentration of 25(OH)D was 1.12 (1.03-1.21) (Fig. 3). A 50% lower concentration of 25(OH)D was associated with a hazard ratio >1.0 in most strata; however, not all individual risk estimates were significant. Nevertheless, as tests of interaction were nonsignificant for all stratifications, except age, after correction for 7 parallel tests using the Bonferroni correction, this implies that low 25(OH)D concentrations associate with increased risk of type 2 diabetes irrespective of category levels of other variables. Concerning age, the multivariable adjusted hazard ratio for type 2 diabetes for a 50% lower concentration of 25(OH)D was 1.50 (1.33-1.70) and 1.00 (0.88-1.15) for those [less than or equal to] 58 years and >58 years old, respectively (interaction, P = [10.sup.-8]).


A total of 14 studies representing 16 cohorts were included in the metaanalysis, with a total of 72 204 participants and 4877 type 2 diabetes events. The characteristics of the studies are summarized in Table 2 and Fig. 4. The odds ratios of type 2 diabetes comparing low vs high concentrations of 25(OH)D were 1.50 (95% CI 1.33-1.66, fixed effect) and 1.50 (1.33-1.67, random effect) (Fig. 4). Further analyses restricted to studies of the general population or studies with complete adjustment did not change the estimates appreciably. Analyses stratified according to study design likewise did not alter the associations substantially. There was no evidence of between-study heterogeneity ([I.sup.2] = 1.4%, P = 0.44) or publication bias (Begg rank correlation test, P = 1.00, and Egger regression test, P = 0.58) (see online Supplemental Fig. S5). The Anderson et al. study (15) differed from the other studies with regard to population, follow-up (mean 1.3 years), adjustment, and ascertainment of diabetes; thus the metaanalysis was repeated without this study resulting in a odds ratio for type 2 diabetes of 1.39 (1.21-1.58).


In the largest general population study to date, we observed an increasing risk of type 2 diabetes with decreasing plasma 25(OH)D concentrations. These findings were confirmed in a metaanalysis of prospective cohort and nested case-control studies published until July 2012.

Biologically, our results make sense, since vitamin D status has been implicated in 2 essential processes linked to type 2 diabetes, i.e., insulin secretion and insulin resistance. (1) Evidence supporting a role for vitamin D in insulin secretion: the vitamin D receptor and the 1-[alpha]-hydroxylase enzyme, the enzyme that converts 25(OH)D into the active hormone 1,25-dihydroxyvitamin D, are present in [beta]-cells (31, 32); in vitro and in vivo studies show that vitamin D receptor knockout or vitamin D deficiency impairs glucose-induced insulin secretion (5, 6, 8, 9, 33); and the insulin secretory response improves after vitamin D supplementation in both animals and humans (5, 6, 8, 9, 34). (2) Evidence supporting a role for vitamin D in insulin sensitivity: the vitamin D receptor is present in skeletal muscle cells (35); vitamin D stimulates insulin receptor expression and insulin-induced glucose transport in vitro (36, 37); vitamin D directly regulates pathways implicated in the regulation of fatty acid metabolism in skeletal muscle and adipose tissue (38); and low concentrations of vitamin D are associated with impaired insulin sensitivity, whereas substitution with vitamin D in the deficient state improves insulin sensitivity (2-4, 9, 39). However, several randomized studies have also shown contrasting results with no improvement in insulin secretion or sensitivity after vitamin D supplementation (10).

Our metaanalysis shows that low concentrations of 25(OH)D are robustly associated with increased risk of type 2 diabetes irrespective of population, level of adjustment, or study design. The estimate from the present metaanalysis is comparable to previous metaanalyses with fewer studies and not including the present study (10, 16). Interestingly, there were no signs of statistical heterogeneity or publication bias in our metaanalysis. Further studies should be randomized intervention studies or genetic epidemiological studies designed to establish causality rather than association as in the present study.

A potential limitation is that our cohort consists of whites of Danish descent living in Denmark (latitude 55-58 degrees north) with less sun exposure than closer to the equator; consequently, our findings would be most applicable to individuals with a similar skin color and a similar level of sun exposure. The delay in measurement from 1981-1983 to 2009-2010 could raise concern of potential decay of plasma 25(OH)D, but this seems unlikely to have distorted our analyses for several reasons: we noticed the expected seasonal variation of 25(OH)D concentrations; median concentrations of plasma 25(OH)D across plasma samples from 3 different examinations on the same healthy participants with storage times of 10, 20, and 30 years were similar; previous studies have shown high stability during storage (40); the median concentration observed in our study of 16 [micro]g/L [41 nmol/L] was similar to that in comparable populations (22, 26); and a low sample quality for the 25(OH)D measurement would tend to weaken rather than inflate an association. Similarly, the diagnoses were obtained from self-report, hospital discharge, and death registries, thus postponing diagnoses made by the general practitioner alone and leading to potential underreporting by participants. However, this potential underreporting would only tend to weaken rather than inflate an association.

Our study has several strengths: our population was homogeneous, we had up to 29 years of follow-up with no loss to follow-up, we could account for other major risk factors associated with risk of type 2 diabetes, and we had the highest statistical power to date to examine the associations of low plasma 25(OH)D concentrations with risk of type 2 diabetes. Furthermore, in Northern Europe, UV-B radiation from the sun is adequate for sufficient endogenous vitamin D production in the skin only during the summer months, and food has never been fortified with vitamin D in Denmark. Thus, this cohort from the Danish general population allows determination of the natural history of the association of vitamin D deficiency with risk of type 2 diabetes.

Clinical applications of the present study should be considered cautiously, as this is an observational study. Randomized interventional trials are needed before supplementation with vitamin D can be recommended for prevention of diabetes.

In conclusion, we observed an association between low plasma 25(OH)D and increased risk of type 2 diabetes in the general population. This finding was substantiated in a metaanalysis.

Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 3 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; and (c) final approval of the published article.

Authors' Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:

Employment or Leadership: None declared.

Consultant or Advisory Role: None declared.

Stock Ownership: None declared.

Honoraria: None declared.

Research Funding: The Danish Heart Foundation, Herlev Hospital, and Copenhagen University Hospital. DiaSorin Laison provided kits for measurement of 25(OH)D.

Expert Testimony: None declared.

Role of Sponsor: The funding organizations played no role in the design of study, choice of enrolled patients, review and interpretation of data, or preparation or approval of manuscript.


(1.) Stumvoll M, Goldstein BJ, van Haeften TW. Type 2 diabetes: principles of pathogenesis and therapy. Lancet 2005;365:1333-46.

(2.) Chiu KC, Chu A, Go VLW, Saad MF. Hypovitaminosis D is associated with insulin resistance and beta cell dysfunction. Am J Clin Nutr 2004;79: 820-5.

(3.) Forouhi NG, Luan J, Cooper A, Boucher BJ, Wareham NJ. Baseline serum 25-hydroxy vitamin D is predictive of future glycemic status and insulin resistance: the Medical Research Council Ely Prospective Study 1990-2000. Diabetes 2008;57: 2619-25.

(4.) Kayaniyil S, Retnakaran R, Harris SB, Vieth R, Knight JA, Gerstein HC, et al. Prospective associations of vitamin D with beta-cell function and glycemia: the PROspective Metabolism and ISlet cell Evaluation (PROMISE) cohort study. Diabetes 2011;60:2947-53.

(5.) Gedik O, Akalin S. Effects of vitamin D deficiency and repletion on insulin and glucagon secretion in man. Diabetologia 1986;29:142-5.

(6.) Cade C, Norman AW. Vitamin D3 improves impaired glucose tolerance and insulin secretion in the vitamin D-deficient rat in vivo. Endocrinology 1986;119:84-90.

(7.) Chertow BS, Sivitz WI, Baranetsky NG, Clark SA, Waite A, Deluca HF. Cellular mechanisms of insulin release: the effects of vitamin D deficiency and repletion on rat insulin secretion. Endocrinology 1983;113:1511-8.

(8.) Norman AW, Frankel JB, Heldt AM, Grodsky GM. Vitamin D deficiency inhibits pancreatic secretion of insulin. Science 1980;209:823-5.

(9.) Boucher BJ, Mannan N, Noonan K, Hales CN, Evans SJ. Glucose intolerance and impairment of insulin secretion in relation to vitamin D deficiency in east London Asians. Diabetologia 1995; 38:1239-45.

(10.) Mitri J, Muraru MD, Pittas AG. Vitamin D and type 2 diabetes: a systematic review. Eur J Clin Nutr 2011;65:1005-15.

(11.) Schnohr P, Jensen JS, Scharling H, Nordestgaard BG. Coronary heart disease risk factors ranked by importance for the individual and community: a 21 year follow-up of 12 000 men and women from the Copenhagen City Heart Study. Eur Heart J 2002;23:620-6.

(12.) Ersfeld DL, Rao DS, Body JJ, Sackrison JL Jr, Miller AB, Parikh N et al. Analytical and clinical validation of the 25 OH vitamin D assay for the LIAISON automated analyzer. Clin Biochem 2004;37:867 74.

(13.) Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA, Heaney RP, et al. Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab 2011;96: 1911-30.

(14.) Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496-509.

(15.) Anderson JL, May HT, Horne BD, Bair TL, Hall NL, Carlquist JF, et al. Relation of vitamin D deficiency to cardiovascular risk factors, disease status, and incident events in a general healthcare population. Am J Cardiol 2010;106:963-8.

(16.) Forouhi NG, Ye Z, Rickard AP, Khaw KT, Luben R, Langenberg C, Wareham NJ. Circulating 25-hydroxyvitamin D concentration and the risk of type 2 diabetes: results from the European Prospective Investigation into Cancer (EPIC)-Norfolk cohort and updated meta-analysis of prospective studies. Diabetologia 2012;55:2173-82.

(17.) Gagnon C, Lu ZX, Magliano DJ, Dunstan DW, Shaw JE, Zimmet PZ, et al. Serum 25-hydroxyvitamin D, calcium intake, and risk of type 2 diabetes after 5 years: results from a national, population-based prospective study (the Australian Diabetes, Obesity and Lifestyle study). Diabetes Care 2011;34:1133-8.

(18.) Gonzalez-Molero I, Rojo-Martinez G, Morcillo S, Gutierrez-Repiso C, Rubio-Martin E, Almaraz MC et al. Vitamin D and incidence of diabetes: a prospective cohort study. Clin Nutr 2012;31: 571-3.

(19.) Grimnes G, Emaus N, Joakimsen RM, Figenschau Y, Jenssen T, Njolstad I, et al. Baseline serum 25-hydroxyvitamin D concentrations in the Tromso Study 1994-95 and risk of developing type 2 diabetes mellitus during 11 years of follow-up. Diabet Med 2010;27:1107-15.

(20.) Hurskainen AR, Virtanen JK, Tuomainen TP, Nurmi T, Voutilainen S. Association of serum 25-hydroxyvitamin D with type 2 diabetes and markers of insulin resistance in a general older population in Finland. Diabetes Metab Res Rev 2012;28:418-23.

(21.) Husemoen LL, Thuesen BH, Fenger M, Jergensen T, Glumer C, Svensson J, et al. Serum 25(OH)D and type 2 diabetes association in a general population. Diabetes Care 2012;35:1695-700.

(22.) Knekt P, Laaksonen M, Mattila C, Harkanen T, Marniemi J, Heliovaara M, et al. Serum vitamin D and subsequent occurrence of type 2 diabetes. Epidemiology 2008;19:666-71.

(23.) Pilz S, van den Hurk K, Nijpels G, Stehouwer CD, Van't Riet E, Kienreich K et al. Vitamin D status, incident diabetes and prospective changes in glucose metabolism in older subjects: the Hoorn study. Nutr Metab Cardiovasc Dis 2012;22: 883-9.

(24.) Pittas AG, Sun Q, Manson JE, Dawson-Hughes B, Hu FB. Plasma 25-hydroxyvitamin D concentration and risk of incident type 2 diabetes in women. Diabetes Care 2010;33:2021-3.

(25.) Robinson JG, Manson JE, Larson J, Liu S, Song Y, Howard BV, et al. Lack of association between 25(OH)D levels and incident type 2 diabetes in older women. Diabetes Care 2011;34:628-34.

(26.) Thorand B, Zierer A, Huth C, Linseisen J, Meisinger C, Roden M, et al. Effect of serum 25-hydroxyvitamin D on risk for type 2 diabetes may be partially mediated by subclinical inflammation: results from the MONICA/KORA Augsburg study. Diabetes Care 2011;34:2320-2.

(27.) Deleskog A, Hilding A, Brismar K, Hamsten A, Efendic S, Ostenson CG. Low serum 25-hydroxyvitamin D level predicts progression to type 2 diabetes in individuals with prediabetes but not with normal glucose tolerance. Diabetologia 2012;55:1668-78.

(28.) Danesh J. Association of fibrinogen, C-reactive protein, albumin, or leukocyte count with coronary heart disease: meta-analyses of prospective studies. JAMA 1998;279:1477-82.

(29.) DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177-88.

(30.) Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21: 1539-58.

(31.) Johnson JA, Grande JP, Roche PC, Kumar R. Immunohistochemical localization of the 1,25(OH)2D3 receptor and calbindin D28k in human and rat pancreas. Am J Physiol 1994;267: E356-60.

(32.) Bland R, Markovic D, Hills CE, Hughes SV, Chan SL, Squires PE, Hewison M. Expression of 25-hydroxyvitamin D3-1alpha-hydroxylase in pancreatic islets. J Steroid Biochem Mol Biol 2004; 89-90:121-5.

(33.) Zeitz U, Weber K, Soegiarto DW, Wolf E, Balling R, Erben RG. Impaired insulin secretory capacity in mice lacking a functional vitamin D receptor. FASEB J 2003;17:509-11.

(34.) Mitri J, Dawson-Hughes B, Hu FB, Pittas AG. Effects of vitamin D and calcium supplementation on pancreatic beta cell function, insulin sensitivity, and glycemia in adults at high risk of diabetes: the Calcium and Vitamin D for Diabetes Mellitus (CaDDM) randomized controlled trial. Am J Clin Nutr 2011;94:486-94.

(35.) Simpson RU, Thomas GA, Arnold AJ. Identification of 1,25-dihydroxyvitamin D3 receptors and activities in muscle. J Biol Chem 1985;260:8882-91.

(36.) Maestro B, Campion J, Davila N, Calle C. Stimulation by 1,25-dihydroxyvitamin D3 of insulin receptor expression and insulin responsiveness for glucose transport in U-937 human promonocytic cells. Endocr J 2000;47:383-91.

(37.) Maestro B, Davila N, Carranza MC, Calle C. Identification of a Vitamin D response element in the human insulin receptor gene promoter. J Steroid Biochem Mol Biol 2003;84:223-30.

(38.) Dunlop TW, Vaisanen S, Frank C, Molnar F, Sinkkonen L, Carlberg C. The human peroxisome proliferator-activated receptor delta gene is a primary target of 1alpha,25-dihydroxyvitamin D3 and its nuclear receptor. J Mol Biol 2005;349: 248-60.

(39.) Nagpal J, Pande JN, Bhartia A. A double-blind, randomized, placebo-controlled trial of the short-term effect of vitamin D(3) supplementation on insulin sensitivity in apparently healthy, middle-aged, centrally obese men. Diabet Med 2009;26:19-27.

(40.) Ocke MC, Schrijver J, Obermann-de Boer GL, Bloemberg BP, Haenen GR, Kromhout D. Stability of blood (pro)vitamins during four years of storage at -20 degrees C: consequences for epidemiologic research. J Clin Epidemiol 1995;48: 1077-85.

Shoaib Afzal, [1] Stig E. Bojesen, [1,2,3] and Borge G. Nordestgaard [1,2,3] *

[1] Department of Clinical Biochemistry, Herlev Hospital, Copenhagen University Hospital, Copenhagen, Denmark; [2] The Copenhagen City Heart Study, Bispebjerg Hospital, Copenhagen University Hospital, Copenhagen, Denmark; [3] Faculty of Health Sciences, University of Copenhagen, Copenhagen, Denmark.

* Address correspondence to this author at: Department of Clinical Biochemistry, Herlev Hospital, Copenhagen University Hospital, Herlev Ringvej 75, DK-2730 Herlev, Denmark. Fax: 38683311; e-mail:

Received July 12, 2012; accepted October 31, 2012.

Previously published online at DOI: 10.1373/clinchem.2012.193003

[4] Nonstandard abbreviations: 25(OH)D, 25-hydroxyvitamin D; BMI, body mass index.
Table 1. Baseline characteristics according to clinical cutpoints
for plasma 25(OH)D concentrations. (a)

                                         Plasma 25(OH)D, ng/mL

                                     <5        5-9.9      10-19.9

n                                   458        1805        3932
Men                               209 (46)   797 (44)    1680 (43)
Age, years
  Median                             59         58          58
  Interquartile range              50-65       49-65       48-65
  Never                           62 (14)    308 (17)    857 (22)
  Ever                            396 (86)   1497 (83)   3075 (78)
Body mass index, kg/[m.sup.2]
  Median                            24.9       25.5        25.1
  Interquartile range              22-29       23-29       23-28
  Low                             206 (45)   663 (37)    1218 (31)
  Medium                          190 (42)   806 (46)    1828 (47)
  High                            57 (13)    306 (17)    834 (22)
Duration of leisure time
    physical activity, h/week
  [less than or equal to] 2       147 (32)   417 (23)    646 (16)
  2-4 (light activity)            198 (43)   887 (49)    1953 (50)
  [greater than or equal to] 4
    or 2-4 (heavy activity)       113 (25)   501 (28)    1328 (34)
  May-October (summer)            114 (25)   667 (37)    2016 (51)
  November-April (winter)         344 (75)   1138 (63)   1916 (49)

                                       Plasma 25(OH)D, ng/mL

                                   [greater than
                                  or equal to] 20   Trend, P (b)

n                                      3646
Men                                  1561 (43)          0.29
Age, years                                             <0.001
  Median                                57
  Interquartile range                  47-64
Smoking                                                <0.001
  Never                              850 (23)
  Ever                               2796 (77)
Body mass index, kg/[m.sup.2]                          <0.001
  Median                               24.2
  Interquartile range                  22-27
Income                                                 <0.001
  Low                                977 (27)
  Medium                             1700 (47)
  High                               936 (26)
Duration of leisure time
    physical activity, h/week                          <0.001
  [less than or equal to] 2          406 (11)
  2-4 (light activity)               1788 (49)
  [greater than or equal to] 4
    or 2-4 (heavy activity)          1450 (40)
Season                                                 <0.001
  May-October (summer)               2329 (64)
  November-April (winter)            1317 (36)

(a) Data are n (%) unless noted otherwise.

(b) Cuzick nonparametric trend test.

Table 2. Observational prospective studies of the association of
plasma 25(OH)D with risk of type 2 diabetes. (a)

                                          Mean        Mean
                                 Women,   age,        BMI,
Reference                 Year     %      years   kg/[m.sup.2]

Fourouhi et al. (16)      2008     58      64         NDc

Pilz et al. (23)          2012     61      68          27

Knekt et al. (22)         2008     54      ND          ND

Gonzalez-Molero           2012     57      50          ND
  et al. (18)
Grimnes et al.            2010     60      57         24.7
  (smokers only) (19)

Hurskainen et al. (20)    2012     54      63         27.8

Thorand et al. (26)       2011     47      52         27.1

Pittas et al. (24)        2010    100      56         27.8

Fourouhi et al. (16)      2012     58      58         26.0

Deleskog et al. (27)      2012     40      48         26.3

Husemoen et al. (21)      2012     52      46          26

Gagnon et al. (17)        2011     55      51         26.6

Grimnes et al.            2010     62      60         26.3
  (nonsmokers) (19)

Robinson et al. (25)      2011    100      66         28.1

Anderson et al. (15)      2010     75      55          ND

This studyd               2012     56      56         25.3

                          White,   Adjustment,
Reference                   %       (0-6) (b)        Design

Fourouhi et al. (16)        99          6        Cohort

Pilz et al. (23)            ND          5        Cohort

Knekt et al. (22)          100          6        Nested
Gonzalez-Molero             ND          6        Cohort
  et al. (18)
Grimnes et al.             100          6        Cohort
  (smokers only) (19)

Hurskainen et al. (20)      ND          6        Cohort

Thorand et al. (26)         ND          6        Case-cohort

Pittas et al. (24)          98          6        Nested case-
Fourouhi et al. (16)        99          6        Case-cohort

Deleskog et al. (27)        ND          5        Nested case-

Husemoen et al. (21)       100          6        Cohort

Gagnon et al. (17)          92          5        Cohort

Grimnes et al.             100          6        Cohort
  (nonsmokers) (19)

Robinson et al. (25)        90          5        Nested case-
Anderson et al. (15)        ND          2        Cohort

This studyd                100          6        Cohort

Reference                     setting                Diagnosis

Fourouhi et al. (16)      General practice   OGTT
Pilz et al. (23)          (Middle-aged)      OGTT, fasting glucose,
                            general            glycosylated
Knekt et al. (22)         General            Medication treated,
Gonzalez-Molero           General            OGTT, glycosylated
  et al. (18)                                  hemoglobin
Grimnes et al.            General            Questionnaire, OGTT,
  (smokers only) (19)                          glycosylated
                                               glucose, registry-
Hurskainen et al. (20)    (Middle-aged)      OGTT, fasting glucose,
                            general            medication treated
Thorand et al. (26)       General            Validated
Pittas et al. (24)        US female          Validated
                            nurses             questionnaire
Fourouhi et al. (16)      General practice   Self-report with
                            population         linkage to general,
                                               hospital, and death
Deleskog et al. (27)      Population         OGTT, fasting glucose
                            enriched with
Husemoen et al. (21)      General            OGTT, fasting glucose,
Gagnon et al. (17)        General            OGTT, fasting glucose,
                                               medication treated
Grimnes et al.            General            Questionnaire, OGTT,
  (nonsmokers) (19)                            glycosylated
                                               glucose, registry-
Robinson et al. (25)      Postmenopausal     Medication treated,
                            women              self-report
Anderson et al. (15)      Health care        Physician diagnoses
This studyd               General            Self report, medication
                                               treated, nonfasting

(a) Studies are ranked as in Fig. 4 based on the fixed-effect
weight in the metaanalysis.

(b) Age, sex, season of blood draw, BMI, smoking, and physical

(c) ND, no data; OGTT, oral glucose tolerance test.

(d) Copenhagen City Heart Study.

Fig. 1. Cumulative incidence of type 2 diabetes by plasma 25
(OH)D in clinical categories and seasonally adjusted quartiles.
Cumulative incidences were plotted using Fine and Gray competing
risks regression accounting for the competing risk of death.
Based on 9841 individuals from the Danish general population, the
Copenhagen City Heart Study, followed for up to 29 years after
blood sampling for measurement of 25 (OH)D.

Plasma 25 (OH)D in clinical categories

No. at risk                            Age (years)

<5 ng/mL         21     100     201     235     137     35     1
5-9.9 ng/mL     102     441     867    1030     596    120     1
10-19.9 ng/mL   260    1038    1894    2476    1471    315     6
>20 ng/mL       353    1059    1810    2303    1453    306    10

Plasma 25 (OH)D in seasonally adjusted quartiles

No. at risk                            Age (years)

1st quartile    138     598    1162    1400     803    167     3
2nd quartile    156     672    1240    1544     903    182     3
3rd quartile    215     708    1187    1508     916    203     5
4th quartile    227     658    1155    1537     982    214     7

Fig. 2. Hazard ratios for type 2 diabetes by plasma 25(OH)D in
clinical categories and seasonally adjusted quartiles.
Multivariable models were adjusted for sex, age, smoking status
(never/ever), BMI, income, and duration and intensity of leisure
time physical activities. Furthermore, the model with clinical
categories for 25(OH)D was adjusted for month of blood sampling.
Based on 9841 individuals from the Danish general population, the
Copenhagen City Heart Study, followed for up to 29 years after
blood sampling for measurement of 25(OH)D.

(ng/mL)             Participants   Events     HR (95% CI)

[greater than or
  equal to] 20          3646        243     1.0 (reference)
10-19.9                 3932        356     1.22 (1.03-1.44)
5-9.9                   1805        174     1.30 (1.06-1.59)
<5                      458          37     1.22 (0.85-1.74)

(quartiles)         Participants   Events     HR (95% CI)

4th (highest)           2397        147     1.0 (reference)
3rd                     2408        183     1.10 (0.88-1.37)
2nd                     2505        235     1.26 (1.02-1.55)
1st (lowest)            2531        245     1.35 (1.09-1.66)

Fig. 3. Hazard ratios for type 2 diabetes by a 50% lower
concentration of plasma 25(OH)D overall and in strata. Analyses
were adjusted for sex, age, smoking status (never/ever), BMI,
income, and duration and intensity of leisure time physical
activities (except the one stratified for). Age and BMI were
categorized by use of the approximate median. Based on 9841
individuals from the Danish general population, the Copenhagen City
Heart Study, followed for up to 29 years after blood sampling for
measurement of 25(OH)D. NS = not significant (P > 1.0) after
multiplication of P value by 7 according to the Bonferroni

                                                       P for


Sex                                                    NS

Age (years)                                            [10.sup.-8]
[less than or equal to] 58

Smoking                                                NS

Body mass index (kg/[m.sup.2])                         NS
[greater than or equal to] 25

Income level                                           NS

Leisure time physical activity (h/week)                NS
[less than or equal to] 2
2-4 (light activity)
[greater than or equal to] 4 or 2-4 (heavy activity)

Season of sampling                                     NS
May-October (summer)
November-April (winter)

Fig. 4. Metaanalysis of prospective studies on plasma 25(OH)D and
risk of type 2 diabetes.

The reference category is the highest category of 25(OH)D in each
study, and risk estimates are versus the lowest category of
25(OH)D in each study. On the forest plot, black box areas are
proportional to the fixed-effect weight of the individual
studies. The white diamonds represent the summary estimate, and
CIs correspond to the width of the diamonds. Complete adjustment
included adjustment for age, sex, season of blood draw, BMI or
other obesity measures, smoking, and physical activity. The Knekt
study includes both the Finnish Mobile Clinic Health Examination
Survey and the Mini-Finland Health Survey. * The Copenhagen City
Heart Study, the present study. ND = no data.

                             No. of      No. of   Follow-up   25(OH)D
Study                     participants   events    (years)    (nmol/L)

Forouhi etal. (16)            777          37        10          59
Pilz et. al. (23)             280          45         8          57

Knekt et al. (22)             1398        412        22          43

Gonzalez-Molero               412          26         4          56
  et al. (18)
Grimnes et al.                1962         64        11          ND
  (smokers only) (19)
Hurskainen et al. (20)        1082        140         9          45

Thorand et al. (26)           1683        416        11          41

Pittas et al. (24)            1167        608        15          57
Fourouhi et al (16)           1447        621        10          65
Deieskog et al. (27)          2022        134         9          60

Husemoen et al. (21)          3759        141         5          48

Gagnon et al. (17)            5200        199         5          65

Grimnes et al.                4157        183        11          53
  (nonsmokers) (19)
Robinson et al. (25)          5140        317         7          48

Anderson et al. (15)         31877        724         1          ND
This study *                  9841        810        21          41

Overall: Fixed-effect estimate (FEE)
Overall: Random-effect estimate (REE)

Studies on the oeneral population: FEE = REE (17-22. 23. 26 *)
Studies with complete adjustment: FEE = REE (16. 18-22. 26 *)

Studies stratified by desion
Cohort (15-21, 23 *): FEE = REE
Nested case-control and case-cohort (16, 22, 24-27): FEE
Nested case-control and case-cohort (16, 22, 24-27): REE

                                                Weight (%)
                             Odds ratio
Study                         (95% CI)       Fixed   Random

Forouhi etal. (16)        1.45 (0.34-5.88)     0.35    0.37
Pilz et. al. (23)         1.88 (0.61-5.82)     0.40    0.42

Knekt et al. (22)         1.49 (0.51-4.35)     0.74    0.77

Gonzalez-Molero           2.13 (1.10-4.17)   1.15      1.20
  et al. (18)
Grimnes et al.            1.47 (0.62-3.48)     1.33    1.39
  (smokers only) (19)
Hurskainen et al. (20)    1.62 (0.85-3.10)     2 15    2 23

Thorand et al. (26)       1.60 (0.84-3.05)     2.23    2.31

Pittas et al. (24)        1.92 (1.20-3.03)     3.25    3.37
Fourouhi et al (16)       2.00 (1.32-3.13)     3.32    3.44
Deieskog et al. (27)      1.72 (1.11-2.70)     4.31    4 45

Husemoen et al. (21)      1.11 (0.60-2.08)     4.97    5.12

Gagnon et al. (17)        1.43 (0.92-2.22)     6.44    7.62

Grimnes et al.            1.37 (0.89-2.10)     7.43    7.62
  (nonsmokers) (19)
Robinson et al. (25)      0.95 (0.57-1.61)    10.06   10.24

Anderson et al. (15)      1.96 (1 61-2.38)    18.36   18.29
This study *              1.35 (1.09-1.66)    33.50   32.16

Overall: Fixed-effect esti1.50 (1.33-1.67)
Overall: Random-effect est1.50 (1.33-1.68)

Studies on the oeneral pop1.38 (1.17-1.60)
Studies with complete adju1.44 (1.23-1.65)

Studies stratified by desion
Cohort (15-21, 23 *): FEE 1.52 (1.33-1.70)
Nested case-control and ca1.50 (1.33-1.66)
Nested case-control and ca1.50 (1.33-1.67)
COPYRIGHT 2013 American Association for Clinical Chemistry, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Endocrinology and Metabolism
Author:Afzal, Shoaib; Bojesen, Stig E.; Nordestgaard, Borge G.
Publication:Clinical Chemistry
Article Type:Clinical report
Date:Feb 1, 2013
Previous Article:Isotope-dilution liquid chromatography-tandem mass spectrometry candidate reference method for total testosterone in human serum.
Next Article:Data submission and quality in microarray-based microRNA profiling.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters