1. Introduction
Prediabetes is a key and critical period for the prevention and control of diabetes mellitus (DM). It is referred to as a health condition of intermediate hyperglycemia, in which the blood glucose levels do not meet the diagnosis criteria for DM but are higher than normal thresholds. It includes two phenotypes based on the definition of intermediate hyperglycemia from the WHO and Chinese Diabetes Society, impaired fasting glucose (IFG) and impaired glucose tolerance (IGT) [
1,
2]. Prediabetes increases the average risk for the future development of type 2 DM (T2DM) by threefold to tenfold, and approximately 70% of prediabetic persons will progress to T2DM in the future [
3,
4]. Besides T2DM, persons with prediabetes are at increased risk of heart attacks and strokes [
4,
5,
6,
7]. Furthermore, the prevalence of prediabetes is increasing and imposes significant public health burdens [
4,
7].
Vitamin D is an essential fat-soluble vitamin and plays a central role in calcium-phosphate homeostasis and skeletal health. More and more studies have reported that vitamin D has various extraskeletal functions, one of which is that vitamin D may regulate glucose homeostasis through improving glucose tolerance and insulin secretion and sensitivity [
8,
9,
10,
11,
12]. Some observational and intervention studies have also indicated a significantly inverse association between vitamin D status and blood glucose level, suggesting that VDD increases the risk of DM and prediabetes, but a few studies did not find such an association [
13,
14,
15,
16,
17]. The overall evidence remains inconsistent and no conclusive evidence is available, especially for prediabetes [
12]. The evidence from China is limited and inconsistent as well. Three studies, including two in west China and one in east China, have been reported [
18,
19,
20]. Of them, one study suggested that low 25(OH)D levels might independently contribute to the incidence of prediabetes or T2DM in Chinese individuals, but the other two, which both had a large sample size, did not find that lower serum 25(OH)D concentrations were associated with a higher risk of prediabetes or T2DM.
Sex differences have been observed in the association between vitamin D status and immunomodulatory and anti-inflammatory effects in some diseases [
21,
22], insulin resistance in a large Caucasian population [
23], serum ferritin levels in Korean adults [
24], and prevalence of dyslipidaemia in the middle-aged and elderly Chinese population [
25]. A study in Americans over 50 years old demonstrated there was an association between VDD and prediabetes, but did not find any sex differences in the association between vitamin D status and prediabetes [
26]. We were interested in whether sex differences in the association between vitamin D status and prediabetes could be distorted by the heterogeneity of prediabetes phenotypes, and whether this could be observed in the Chinese population. To our knowledge, there are no previous studies that examined sex differences in the association between VDD and prediabetes phenotypes, IFG and IGT, separately. IFG and IGT have different underlying pathophysiologic mechanisms, although both have a higher risk of developing T2DM, reflecting the impairment of blood glucose regulation function under basal state and after glucose load, respectively [
3,
27]. IFG is primarily caused by hepatic insulin resistance and disturbed first-phase insulin secretion, whereas IGT primarily results from muscle insulin resistance, impaired first- and second-phase insulin secretion, and reduced β-cell sensitivity to glucose. It was observed that IFG occurs more frequently in men than in women, whereas IGT is common particularly in women, which suggests that the pathophysiologic mechanisms between men and women are not completely the same [
3]. It was reported that endogenous estrogens might stimulate insulin synthesis and secretion, and might be relevant the sensitivity of insulin, so female sex hormones may play an important role in the pathogenesis of IFG and IGT. Therefore, it will be more reasonable to consider the possible heterogeneity of prediabetes phenotypes and explore the association between VDD and IFG/IGT, instead of prediabetes, in men and women, respectively.
This present study aimed to examine whether there is sex-specific association between VDD and IFG/IGT after adjustment for other potential confounding factors including biological factors, lifestyles, etc., based on sampling data representative of general adult residents in Henan, in the central part of China.
4. Discussion
This present study found that the prevalence rate of IGT in men with VDD was 20.1%, significantly higher than that in men with non-VDD (10.5%), and that there were effect modifications by sex on the association between VDD and IGT in adults and by the family history of DM in men, suggesting VDD was independently associated with IGT in men, especially in men having a family history of DM. The OR (95%CI) of VDD adjusted for all studied variables was estimated to be 1.99 (1.24–3.19) for IGT in men and 14.84 (4.14–53.20) for IGT in men having a family history of DM, implying that after excluding the impact of other factors, having VDD doubly increased the risk of having IGT in men compared with men with non-VDD and increased by nearly 15 times the risk of having IGT in men with a family history of DM. It seems men having a family history of DM living with VDD are probably far more likely to have IGT than those who are not living with VDD, and that the association between VDD and IGT mainly exists in men with a positive family history of DM. Meantime, no independent associations between vitamin D status and prevalence of IFG were found either in men or women.
We did not find any previous sex-specific studies that examined the association between VDD and IGT/IFG, only prediabetes. For comparison, a previous study based on the data from the cross-sectional surveys of the NHANES in the USA reported that VDD was associated with risk of prediabetes in Americans over 50 years old, but there was no effect modification by sex and it did not report an association between vitamin D status and the phenotypes of prediabetes [
26]. However, studies based on the Korean National Health and Nutrition Examination Survey (KNHANES) data reported that there was a sex-based modification effect of vitamin D on blood glucose regulation and prevalence of metabolic syndrome, and also suggested that the association between vitamin D status and diabetes or metabolic syndrome was only found in men [
30,
31]. One Chinese study reported that vitamin D had a negative impact on insulin resistance only in male patients with newly diagnosed type 2 diabetes mellitus [
32]. Although these studies in Korea and China did not directly explore the association between vitamin D and IFG/IGT, they did demonstrate that the blood glucose regulation function of vitamin D perhaps only existed in men. It has been reported there is a potential interaction between sex hormones and level of vitamin D, and estrogen can enhance some functions of vitamin D [
21,
22]. To a certain extent, these findings support that the association between VDD and BGS might be sex-differentiated, if the association exists. This is in line with our findings that the association between VDD and IGT was only found in men, but not in women. However, this seems inconsistent with the study in USA mentioned above. Perhaps racial differences and dietary structure disparities lead to this inconsistency.
Additionally, a further examination on the association between VDD and IGT in men was conducted in the present study by adding the interaction term between VDD and other factors to the logistic regression model, finding that there might be an interaction between vitamin D status and family history of DM, showing a far stronger association between VDD and IGT in men having a family history of DM, suggesting that VDD greatly increased the risk of IGT in men having a family history of DM. The population having a family history of DM is a high-risk group for IGT since a positive family history of DM increases the risk of IGT by 25–64% [
33]. It is reasonable that the association between VDD and IGT is more likely to be demonstrated in the high-risk groups of IGT, such as men having a family history of DM. On the other hand, this finding suggests that the occurrence of IGT may be effectively reduced by preventing or treating VDD in men having a family history of DM, but this needs confirmation using clinical experiments.
This present study did not find any independent association between VDD and IFG either in men or women. IFG and IGT are two different phenotypes of prediabetes and both increase the risks of progressing to DM, but they have different underlying pathophysiological mechanisms, reflecting the impairment of blood glucose regulation function under basal state and after glucose load, respectively [
3,
27]. Owing to such heterogeneity in prediabetes phenotypes, the risk factors for them should not be completely the same. Thus, when exploring the association between vitamin D status and prediabetes, it was more reasonable to take IFG and IGT into account separately in this present study. Additionally, this drops a hint that it may be more reasonable to use IFG and IGT instead of prediabetes for the purposes of the diagnosis, clinical intervention, prevention, and control of DM in men and women.
Currently, there is no international consensus for the definitions of prediabetes, DM, and VDD [
1,
2,
4,
7,
11,
17,
28]. The prevalent definitions or diagnosis criteria of DM, prediabetes, and its phenotypes issued from the WHO, IDF, and ADA are used with different biomarkers, including glycated hemoglobin (HbA1c) as well as FPG and 2 h PG during OGTT, or different cut-off values for same biomarkers, all of which provide different estimates on the prevalences of DM, prediabetes, and its phenotypes in the population [
4,
7]. Considering that HbA1c being used for the definition of prediabetes has not been universally adopted globally, and that the current clinical practice in Henan for HbA1c testing has not been put into use for the diagnosis of prediabetes and diabetes in all healthcare institutions, this present study adopted the definition of prediabetes issued from the WHO and Chinese Diabetes Society, which does not include the biomarker of HbA1c [
1,
2,
6,
7]. The definition of VDD is under debate as well. The cut-off values of serum 25(OH)D concentration for the definition of VDD currently used include 10 ng/mL, 12 ng/mL, 20 ng/mL, and 30 ng/mL [
11,
17,
28]. The working group convened by the Sackler Institute for Nutrition Science at the New York Academy of Sciences and the Bill & Melinda Gates Foundation in coordination with a scientific organizing committee conducted a study and recommended the cut-off value 12 ng/mL (30 nmol/L) was for low- and middle-income countries [
28]. The cut-off value of 12 ng/mL was also recommended for the public health approach. Considering the income of Henan residents, this present study took 12 ng/mL as the cut-off value for the definition of VDD. The difference in the definitions or criteria would bring about the bias on the association between VDD and prediabetes or its phenotypes, which may be a factor leading to the inconsistent study results reported on the association between VDD and prediabetes besides the sex modification effect and heterogeneity of prediabetes phenotypes.
This present study considered the heterogeneity of prediabetes phenotypes as well as the effect modification by sex to examine the association between VDD and IFG/IGT based on a population-representative cross-sectional survey with a multi-stage sampling method, and statistical analysis methods for complex sample survey data were used incorporating both the sampling weights and the sample design instead of the estimate methods based on a simple random sample, which produced unbiased estimates and the correct standard error estimates. Therefore, the estimates in this present study are reliable. However, this present study has some limits as well. HbA1c was not included as a phenotype of prediabetes, which would result in the underestimated prevalence of prediabetes, but had no impact on the conclusion of this present study since this present study focused on the two phenotypes of prediabetes, IFG and IGT. This is a cross-sectional study design which cannot draw causal relationships, owing to the nature of cross-sectional studies, which lack temporality.