Associations between vitamin D status and diabetic complications in Chinese population with type 2 diabetes mellitus: a cross-sectional study

Background Vitamin D status has been linked to diabetes-related complications due to multiple extraskeletal effects. We aimed to investigate the association between vitamin D deficiency (VDD) and diabetic vascular complications, including diabetic retinopathy (DR), diabetic kidney disease (DKD), and diabetic foot ulcers (DFU). Methods A total of 4284 Chinese patients with type 2 diabetic mellitus (T2DM) were enrolled into the cross-sectional study. VDD was defined as serum 25-hydroxyvitamin D < 50 nmol/L. Demographic data, physical measurements, laboratory measurements, comorbidities, and related medications were collected and analyzed by VDD status. Poisson regression with robust variance estimation and binary logistic regression were performed to explore the relationship between VDD and diabetic complications. Results The prevalence of VDD, DR, DKD, DFU accounted to 71.7% (95% confidence intervals [CI]: 70.3%-73.0%), 28.5% (95% CI: 27.2%-29.9%), 28.2% (95% CI: 26.8%-29.5%) and 5.7% (95% CI: 5.1%-6.5%), respectively. The prevalence ratios (95% CI) for DR and DKD by VDD status, adjusted for demographics, physical measurements, laboratory measurements, related complications and comorbidities, and medications, were 1.093 (0.983–1.215) and 1.041 (0.937–1.156), respectively. The odds ratio (95% CI) for DFU by VDD status was 1.656 (1.159–2.367) in the final adjusted model. Meanwhile, the prevalence of VDD was significantly higher in patients with DFU compared with patients without DFU. of among VDD status DR or DKD not significant when SD: standard deviation; PR: prevalence ratios; OR: odds ratio; CI: confidence intervals.

T2DM [18,19], DR [20], DKD [21], or DFU [22] among diverse populations has been reported, uncertainties still exist due to the discordant results [23,24]. More importantly, large-scale epidemiological studies on the association of VDD and diabetes-related complications among the Chinese population are scarce, especially studies evaluating the relationship between VDD and DFU.
In this cross-sectional study, we aimed to explore the prevalence of VDD, and address the associations between VDD and three severe vascular diabetic complications (i.e. DR, DKD, DFU) in a Chinese T2DM population retrospectively.

Study Population
Participants we enrolled were admitted to the Department of Metabolism and Endocrinology and Diabetes Center of the Second Xiangya Hospital of Central South University from January 2014 to July 2018. Only inpatients aged ≥ 18 years with a definite diagnosis of type 2 diabetes were included in the study. The exclusion criteria were as follows: (1) with missing serum 25-hydroxyvitamin D (25[OH]D) data; (2) pregnant or lactating females; (3) with a known diagnosis of nephrolithiasis, glomerular or lupus nephritis, primary nephrotic syndrome, or other identified kidney diseases; (4) serum parathormone > 6.9 pmol/L or < 1.6 pmol/L; (5) serum calcium < 2.1 mmol/L or serum phosphorus < 2.1 mmol/L; (6) estimated glomerular filtration rate (eGFR) < 15 mL/min/1.73 m 2 or missing, which was calculated using the abbreviated Modification of Diet in Renal Disease (MDRD) equation: 186×(serum creatinine) −1.154 ×(age) −0.203 ×(0.742 if female) [25]. A total of 4284 participants were included in this cross-sectional study and their clinical data were extracted from the electronic medical record system. This study was complied with the Declaration of Helsinki and was approved by the ethics committee of the Second Xiangya Hospital of Central South University.

Data collection
General demographic information, including age, gender, smoking and drinking status, duration of diabetes, and family history of diabetes were collected. Physical examination (including body weight and height, blood pressure) was performed by professional caregivers. Body mass index (BMI) was calculated as weight divided by height squared and waist-hip ratio (WHR) was computed as the waist circumference divided by the hip circumference.

Definition
VDD was defined as serum 25(OH)D < 50 nmol/L (20 ng/mL). Conversely, the 25(OH)D level of the no VDD group was ≥ 50nmmol/L. The presence of DR was confirmed by a professional ophthalmologist using dilated fundoscopy according to the definition of the Global Diabetic Retinopathy Project Group [26]. DKD was defined mainly based on albuminuria and a decline of eGFR (< 60 mL/min/1.73 m 2 ), which was not caused by other causes than diabetes. DFU was mainly defined according to diabetic foot problems, such as ulceration, infection, ischemia, gangrene, or even amputation. DPN was diagnosed by analyzing clinical symptoms, neurologic examinations and the results of nerve conduction tests. HTN was defined when blood pressure was ≥ 140/90 mmHg on three separate occasions after hospital admission by physicians, a prior diagnosis of hypertension or taking antihypertensive drugs. CHD and CVD were defined as self-reported history of CHD or CVD, respectively, regardless of disease severity. The definition of dyslipidemia was as follows: TC ≥ 6.22 mmol/L, TG ≥ 2.26 mmol/L, LDL-C ≥ 4.14 mmol/L, HDL-C < 1.04 mmol/L. Additionally, subjects were divided into three groups by age (i.e., aged 18-44, young adults; aged 45-64, middle age adults; elderly, aged ≥ 65 years). Participants were also categorized into four groups based on the levels of BMI according to BMI criteria established by the Working Group on Obesity in China (WGOC) [27]: underweight (< 18.5 kg/m 2 ), normal weight (18.5-23.9 kg/m 2 ), overweight (24.0-27.9 kg/m 2 ), and obese (≥ 28.0 kg/m 2 ). Glycemic control was classified based on HbA1c levels as either good (< 7%) or poor (≥ 7%). The level of SUA was defined as normal (< 420 µmol/L) and high (≥ 420 µmol/L).

Statistical analysis
Normally distributed continuous variables were presented as the mean ± standard deviation (SD) and compared by Student's t test. The Mann-Whitney test was used for non-normally distributed continuous variables, which were reported as median and interquartile range (25-75%). Categorical variables were summarized by frequency counts with percentages, and the chi-square test was performed to evaluate differences between groups. For continuous variables with missing values < 5%, the missing values were replaced by the mean value of the corresponding variable. Two-tailed P-values < 0.05 were considered statistically significant.
In regression analyses, a total of 4176 participants without missing value in smoking status, drinking status and family history of diabetes were included. HOMA2-IR (19.7% missing) and 24HUALB (9.0% missing) were analyzed by creating a dummy variable corresponding to missing values, respectively. As prevalence of DR and DKD in T2DM patients were not rare, Poisson regression with robust variance estimation were conducted instead of logistic regression to directly estimate the prevalence ratios (PR), along with 95% confidence intervals (CI), and to avoid the overestimation of risk ratios by odds ratio [28]. The association between VDD status and DFU was still analyzed using binary logistic regression. Potential confounders (age, gender, duration of diabetes, smoking status, drinking status, BMI and WHR) and the candidate variables with a P value < 0.1 on univariate analysis (data not shown) were all included in the multivariable model to analyze the relationship between VDD status and diabetic complications of T2DM (i.e., DR, DKD and DFU).
The SPSS software (version 25.0; IBM Corp., Armonk, NY) and Stata software (version 14.0; Stata Corp., College Station, TX) were used for statistical analysis. Graphing were performed using Graphpad Prism 7 software (Graphpad Prism Software Inc., La Jolla, CA). Table 1 displays the descriptive characteristics of this T2DM study population, both overall and stratified by VDD status. A total of 4284 participants were analyzed in this study. Slightly more than half (52.6%) were male and middle age adults (aged 45-64 years) made up 57.0% of the population. The proportion of participants was similar between groups with different duration of diabetes (33.5%, 34.4% and 32.1%). By our primary definition, poor glycemic control, dyslipidemia and HTN was observed in 82.1%, 64.7% and 54.0% of subjects. The proportion of patients undergoing BPLT was 41.8% and patients received LLT accounted for 73.9%. Besides, the vast majority (97.3%) were on GLT, including insulin and/or oral hypoglycemic drugs.  17.0%, P = 0.02), relative to those without VDD. Significant differences were also found with respect to medication between the two groups, including BPLT, LLT and GLT (all P < 0.05). Besides, Additional file 1 (Table  S1) displays the prevalence of VDD and three diabetic vascular complications (i.e., DR, DKD and DFU). Overall, the prevalence of VDD accounted to 71.7% (95% confidence intervals [CI]: 70.3%-73.0%), which was defined as 25(OH)D levels less than 50 nmol/L. The prevalence of DR and DKD were very similar at 28.5% (95% CI: 27.2%-29.9%) and 28.2% (95% CI: 26.8%-29.5%), respectively. Besides, a total of 5.7% (95% CI: 5.1%-6.5%) of patients had a diagnosis of DFU in this study.  (Table S2).

The association between prevalence of DKD and VDD status
Models 1-5 in Fig. 2 present the Poisson regression with robust variance models for the assessment of the correlation between VDD status and the prevalence of DKD. The prevalence of DKD was significantly higher in the VDD group in comparison to no-VDD persons in the crude analysis (model 1)  (Table S3).

The association between DFU and VDD status
As the prevalence of DFU was not common (5.7%) in the study population, we next performed logistic regression analyses to assess the relationship between VDD status and the prevalence of DFU (Fig. 3). In the crude model (model 1), the presence of VDD was associated with an increased prevalence of DFU (odds ratio [  (Table S4) displays all variables included in the final adjusted model.

Vitamin D metrics by DFU status, overall and by gender
We also evaluated the proportions of VDD and 25(OH) levels between the DFU group and the no-DFU group (Table 2). Overall, compared with subjects without DFU, persons with DFU had higher prevalence of VDD (79.59% vs. 71.21%, P = 0.005) and lower serum 25(OH)D levels (36.96 ± 18.03 nmol/L vs. 40.97 ± 17.82 nmol/L, P = 0.001). When stratified by gender, similar results were observed in men (all P < 0.05), while there were no significant differences in the prevalence of VDD between these two groups in women (83.02% vs. 75.55%, P = 0.08). Moreover, in the DFU group, it seemed that the 25(OH)D levels in men were slightly greater than that in women, although it did not reach statistical significance (38.86 ± 19.26 nmol/L vs. 34.46 ± 16.03 nmol/L, P = 0.058).

Discussion
In this study, approximately 71.7% of Chinese hospitalized patients with T2DM developed VDD. Patients with VDD had higher prevalence of DFU after adjustment for demographics, physical measurements, laboratory indices, related treatment factors and comorbidities compared with patients without VDD, whereas the associations between VDD status and another two microvascular complications (i.e., DR and DKD) were not statistically significant.
VDD is a growing epidemic condition around the world [29], the prevalence of which varies by race, latitudes, and seasons [20]. An estimated 50%-80% of the general population is affected by vitamin D insufficiency or VDD globally [30]. In northwest and north China, the prevalence of VDD was about 75.2% and 87.1%, respectively [31,32]. VDD is also quite common among Chinese patients with T2DM. A Chinese cross-sectional survey among diabetic inpatients reported that the proportions of persons with VDD were 83.5%, the recruitment center of which was located in north China (latitude 34°-37° N) [33]. Besides, approximately 62.7% of T2DM subjects were affected in two epidemiological studies conducted in Nanjing, which is located in eastern coastal China (latitude 31°-33° N) [21]. In present study, we found that the prevalence of VDD was about 71.7% among this study population with T2DM, who were recruited in Changsha, a city located in central China (latitude 27°-29° N). Although the discordance in prevalence of VDD could partially explained by latitude, other factors such as diet and lifestyle must be considered.
Furthermore, we found that the prevalence of DR and DKD in this study population was 28.5% and 28.2%, respectively. The results were roughly in line with previous studies [34,35]. However, the proportion of DFU (5.7%, 95% CI: 5.1%-6.5%) was much lower in comparison to a previous study conducted in Wuhan, China (11.4%) [36], although it was much higher than the prevalence reported in 2010 (0.8%) by a cross-sectional study conducted in Shanghai, China [37]. We believed that the actual prevalence of DFU may be underestimated. One plausible interpretation was the presence of missed diagnosis of DFU during admission. Some patients without acute symptoms (e.g., ulceration, infection, swollen foot with pain) may not receive further examinations due to socioeconomic concerns. Besides, DFU is generally considered as the consequences of diabetic neuropathy and/or peripheral arterial disease (PAD) [7]. Sometimes patients with PAD may remain undiagnosed until severe tissue loss appears, which also add to the difficulty of the correct diagnosis of DFU [38]. We also addressed that in rural areas of China, a higher proportion of DFU remains undiagnosed because of the less medical access and limited knowledge on this severe diabetic complication. Therefore, future efforts should be directed at early diagnosis of DFU in both urban and rural areas.
DR is the leading cause of visual impairments among working-aged adults suffered from diabetes [39]. The risk factors contributed to the progression of DR include poor glycemic control, long duration of diabetes, inflammation, obesity, and hypertension [20]. Vitamin D status is hypothesized to prevent DR mainly due to its inhibitory effects on inflammation and angiogenesis [40,41]. Lu et al. [42] also demonstrated that vitamin D could decrease diabetes-induced reactive oxygen species and exert protective effects against retinal vascular damage and cell apoptosis. Many clinical studies have recognized VDD as a risk factor for DR [16,[43][44][45], whereas other epidemiological researches showed an opposite result [19,23,46,47]. In this study, we took full advantage of the data already gathered and we reported that VDD status was associated with a higher prevalence of DR when adjusting demographics, physical measurements and biochemical indices (PR: 1.132; 95% CI: 1.014-1.264). However, further adjustments of comorbidities, diabetic complications and medication use attenuated the association and removed the statistical significance. These discrepancies may be result from variations in population sampling, the diagnosis of DR and the covariates included in the regression analyses. Considering all the published data, the connection between vitamin D levels and risk of DR remains inconclusive in Chinese population and further studies are required.
DKD is one of the most common microvascular complications of diabetes, the histology of which is characterized by glomerular basement membrane (GBM) thickening, mesangial matrix expansion, nodular glomerulosclerosis, and arteriolar hyalinosis [48]. Many researches have explored the mechanisms underlying the renoprotective effects of vitamin D. The effects of vitamin D are mediated by VDR, which is widely expressed in various renal cells, such as glomerular mesangial cells, podocytes and tubular cells [49][50][51]. VDR down-regulation is associated with the severity of albuminuria in T2DM patients [52]. Accumulating evidence suggested that vitamin D may protect the kidney through alleviating oxidative stress [53], reducing inflammation response by blunting nuclear factor-kappa B (NF-κB) activation [54], preventing epithelial-to-mesenchymal transformation, and suppressing the expression of transforming growth factor-β [49,55]. Moreover, vitamin D also can inhibit the renin-angiotensin-aldosterone system (RAAS) through down-regulating renin expression, which plays a crucial role in the pathogenesis of DKD [56,57]. However, clinical studies pertaining to the association between vitamin D status and DKD have reported inconsistent results. A recent cross-sectional study of 351 Chinese inpatients with T2DM demonstrated that VDD was independently associated with DKD [21]. This was consistent with many previous studies which had linked albuminuria to low vitamin D levels [58][59][60][61][62]. Conversely, several studies reported discordant results [24,63,64]. In this study, we revealed that the correlation between VDD status and prevalence of DKD was not statistically significant after adjusting laboratory measurements, diabetic complications, related comorbidities and medications besides the adjustment for demographics and physical measurements. These conflict results may be mainly due to differences in study population and adjusted covariates. Besides, the evidence for vitamin D supplementation to prevent DKD remains weak due to the discordances in published data [65][66][67][68]. Taken together, the impacts of VDD on DKD need to be further studied in the future.
DFU, a complex and costly complication of diabetes, is associated with other severe conditions such as peripheral neuropathy, peripheral vascular disease, secondary infections and it can lead to lower extremity amputation [69]. Until now, there are only two studies from India revealed that low vitamin D may play a critical role in the pathogenesis of DFU [70,71]. We investigated the association between VDD and risk of DFU in Chinese T2DM patients for the first time. In accord with these studies [70,71], we demonstrated that there were significant 65.6% higher odds of DFU for those participants with VDD than those without VDD when adjusting all potential confounders collected in this study, including demographics, physical measurements, biochemical indices, related complications and comorbidities, as well as medications. However, Afarideh et al. reported no difference in vitamin D levels in Iranian patients with DFU compared with diabetic patients without DFU.
Regardless of this contradiction, many lines of evidence supported the favorable effects of vitamin D on DFU, especially on wound healing, which is impaired in diabetic patients due to persistent inflammation [72]. Vitamin D is essential in maintaining the normal immune system [73,74]. Vitamin D could suppress T cell proliferation and inhibit the secretion of T helper type 1 cytokines (e.g., interferon-γ and interleukin-2), while augmenting the production of T helper type 2 cytokines [75], thereby accelerating wound healing. Another study found that calcitriol, the most active vitamin D metabolite, not only augmented proangiogenic factors in keratinocytes but also induced antimicrobial peptides expression in a DFU model [76]. Besides, vitamin D may improve wound healing by suppressing endoplasmic reticulum stress [77], oxidative stress [78], and the NF-κB-mediated inflammatory gene expression [72]. More importantly, vitamin D signaling may be involved in the proliferation, migration, and differentiation of epidermal stem cells and progeny during cutaneous wound repair [79]. Despite the abundance of preclinical data regarding vitamin D and wound healing, only one randomized controlled trial that evaluated the effects of vitamin D supplementation on DFU patients has been reported [78]. Razzaghi et al. [78] demonstrated that vitamin D supplementation for 12 weeks resulted in a significant improvement on wound evolution, including ulcer length, width, depth, and erythema rate. In the light of current evidence, we proposed that VDD may be associated with a higher prevalence of DFU and vitamin D supplementation may be a potential therapeutic option for DFU patients with low vitamin D levels, although whether VDD is the cause or the result of DFU remains unknown.
Additionally, previous studies have revealed that patients with DFU had higher prevalence of VDD and lower vitamin D levels in comparison to diabetic patients without DFU [17,22,70,[80][81][82]. We reported that the prevalence of VDD in participants with DFU was about 80%. Meanwhile, the prevalence of VDD in male and female patients with DFU was approximately 77% and 83%, respectively. One explanation for these results could be their immobilization caused by DFU, thus leading to less outdoor activities and sunlight exposure [80].
The results of our study have several clinical implications for healthcare delivery. We demonstrated for the first time that low vitamin D was associated with higher prevalence of DFU in Chinese T2DM population. We highlighted that vitamin D level may be a modifiable factor in the prevention of diabetic foot complications. Besides, our findings extend the knowledge about the correlation between VDD and DR/DKD, although the statistical significance was removed in the final adjusted model. Finally, considering the high prevalence of VDD, we proposed that screening 25(OH)D levels may be beneficial for patients with diabetes. It not only incites patients to change lifestyles and dietary timely so as to increase the levels of vitamin D, but also help minimize the occurrence of complications, and improve the quality of life.
Our study has some important strengths, including the relatively large sample size, a well-defined study population, the availability of multiple covariates, and strong quality control. Notwithstanding these strengths, some limitations should be considered. First, the study design was cross-sectional, thus the temporality of this association between VDD status and DFU cannot be confirmed. Second, the current study was a single-center study, thus our results may not be generalizable to the entire Chinese T2DM population. Third, other potential confounders, including diet, sunlight exposure, physical activity, economic status, season of vitamin D detection, supplementation of vitamin D (by food or drugs) were not available in the analysis. To clarify the associations between vitamin D levels and diabetes-related complications and to assess the benefits of vitamin D supplementation, multi-center randomized controlled trials and larger-scale prospective studies are required.

Conclusion
VDD is a very common condition among Chinese T2DM patients. Decreased vitamin D levels were associated with a higher prevalence of diabetes-related complications, especially DFU. The association between VDD status and DFU was significant and independent of numerous potential confounders, including demographics, physical measurements, biochemical indices, related comorbidities and complications, as well as medication use. Vitamin D supplementation by dietary or other intervention strategies to correct VDD in Chinese diabetic population may help prevent the development of diabetic complications.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download. Additionalfile1.docx