The Degree of Prepregnancy Vitamin D Deficiency Is Not Associated With Gestational Diabetes in Women Undergoing ART.

This study is a post hoc analysis of a prospective single-center clinical study. The primary study was concerned with the effect of vitamin D status on IVF/ICSI outcome ( 11 ). Data were collected between January 2017 and December 2018 at the Reproductive and Genetic Hospital of CITIC-Xiangya in Changsha, China. The study was approved by the ethics committee of the hospital (approval number: LL-SC-2018-014) and was carried out with the written, informed consent of all participants. From 3600 initial patients considered eligible for the study, the final cohort of the primary study included 2569 women, who underwent fresh embryo transfer at the hospital during the study period. However, among women who got pregnant, data about the incidence of GDM were only available for 1593 patients, who comprised the final cohort for this study. These women received IVF/ICSI treatment, followed by fresh embryo transfer, as already described in the study protocol ( 11 , 12 ). Inclusion criteria were: (1) age from 18 to 40 years; (2) first IVF/ICSI treatment; (3) received IVF/ICSI treatment, as well as fresh embryo transfer; (4) informed consent was given by the patient for participation in the study; and (5) clinical pregnancy diagnosed. The exclusion criteria were (#1 to #11 of the following criteria were screened for every patient before treatment): (1) received oocyte donation; (2) uterine malformation; (3) endometriosis; (4) uterine adhesions; (5) untreated hydrosalpinx; (6) uterine myoma; (7) Cushing syndrome; (8) adult-onset adrenogenital syndrome; (9) hypothalamic or pituitary disease causing infertility; (10) diabetes mellitus type 1 or 2 prior to pregnancy; (11) hypertension prior to pregnancy (blood pressure values above 140 mmHg systolic or 90 mmHg diastolic; or taking antihypertensives); and (12) incomplete data regarding the incidence of GDM. Total and free vitamin D were measured before pregnancy, 1 day before embryo transfer, but after hormonal stimulation, both using a commercially available enzyme-linked immunosorbent assay (ELISA) kit (DIAsource ImmunoAssays: Total, RRID: AB_3068001; Free, RRID: AB_2890998). Free vitamin D levels can be obtained directly or indirectly through calculation. In this study, it was measured directly, which is more accurate ( 8 ). According to the Endocrine Society, total vitamin D levels are classified as follows: (1) vitamin D sufficiency (≥30 ng/mL); (2) vitamin D insufficiency (20 to <30 ng/mL); and (3) vitamin D deficiency (<20 ng/mL) ( 13 , 14 ). However, there is no set range for total 25(OH)D during exceptional circumstances, such as pregnancy, as well as for specific ethnic groups. Likewise, there are no reference levels for free 25(OH)D to date. Suggested values for free 25(OH)D by Zeng et al are 5.67 pg/mL [equivalent to 20 ng/mL total 25[OH]D] and 8.50 pg/mL [equivalent to 30 ng/mL total 25[OH]D] ( 15 ). Concerning the outcomes, this paper focuses on a secondary endpoint of the primary study: the incidence of GDM. During pregnancy, all study patients were screened for gestational diabetes using a 75-g oral glucose tolerance test, as previously described ( 12 ). Gestational diabetes was diagnosed if it was not present prior to pregnancy, and blood glucose levels were in one of the following categories: (1) ≥92 mg/dL (5.1 mmol/L) before glucose intake; (2) ≥180 mg/dL (10.0 mmol/L) 1 hour after glucose intake; or (3) ≥153 mg/dL (8.5 mmol/L) 2 hours after glucose intake, following the International Association of Diabetes in Pregnancy Study Groups (IADPSG) 2010 guidelines ( 12 , 16 ). Patients were contacted by phone during the pregnancy and after delivery to follow up on further complications and outcomes. Primary data was analyzed using SPSS version 29.0 (SPSS Inc., Chicago, IL, USA). Diagrams were created with GraphPad Prism 6 (GraphPad Software, San Diego, CA, USA). All study participants from the original study were included in this analysis if data regarding GDM was available. Therefore, we did not perform a sample size calculation. In the descriptive statistics, values are expressed either as frequency (%) or median (interquartile range [IQR]). The Chi-square (χ 2 ) test for categorical variables and the Mann-Whitney U test for continuous variables were carried out to compare GDM and non-GDM patient groups. One-way analysis of variance (ANOVA) was used to compare GDM occurrence in the different vitamin D quintiles. We also created a multivariate logistic regression model, considering factors that had a significant P value in the descriptive statistics. For this analysis, we created 2 models: Model A, considering significant factors in literature: maternal age, obesity (via body mass index [BMI]), and prediabetes (via blood sugar measurement prior to fertility treatments). Model B included significant factors from Table 1 (maternal age, BMI, blood sugar, anti-Müllerian hormone [AMH], luteinizing hormone [LH], and estradiol). All patients were of Chinese descent, while other risk factors, such as a family history of diabetes, were not recorded. P < .05 was considered statistically significant. Characterization of the study population—binary analysis Data in the table are shown as frequency, n (%) for categorical variables, or median (IQR) for continuous variables. Comparisons were made using the Chi-square (χ 2 ) test for categorical variables and Mann-Whitney U test for continuous variables. Abbreviations: 25(OH)D, 25-hydroxyvitamin D; AFC, antral follicle count; AMH, anti-Müllerian hormone; BMI, body mass index; FSH, follicle-stimulating hormone; GDM, gestational diabetes mellitus; ICSI, intracytoplasmic sperm injection; IVF, in vitro fertilization; LH, luteinizing hormone; RR, blood pressure.

In accordance with the criteria mentioned in “Methods,” a total of 1593 patients undergoing embryo transfer after IVF/ICSI were included in this study. The median age was 29 years (IQR, 27-31). These patients were grouped into GDM and non-GDM patients for binary analysis ( Table 1 ). A total of 256 (16.1%) of the 1593 patients were diagnosed with GDM during their pregnancy. Age, BMI, and blood sugar measured prior to embryo transfer showed a statistically significant difference between the 2 groups ( P < .05), as did the hormones AMH, LH, and estradiol. Contrastingly, neither total nor free 25(OH)D was statistically significantly different ( P = .340 and .849 respectively) when comparing GDM and non-GDM patients. Following the clinical practice guidelines of the Endocrine Society for total vitamin D ( 13 ), only 47 (2.9%) patients had a sufficient total 25(OH)D supply, according to the guidelines; 696 (43.7%) patients had an insufficient and 850 (53.4%) had a deficient vitamin D status ( Fig. 1 ). Distribution of study population in total and free 25-hydroxyvitamin D [25(OH)D] categories. Data are shown as frequency, n (%). Comparison was made with the Chi-square (χ 2 ) test. Total 25(OH)D was classified into 3 groups as follows: deficiency (<20 ng/mL), insufficiency (20 to <30 ng/mL), and sufficiency (≥30 ng/mL). Free 25(OH)D was classified into 3 groups using the values suggested by Zeng et al ( 15 ): deficiency (<5.67 pg/mL) insufficiency (5.67 to <8.50 pg/mL), and sufficiency (≥8.50 pg/mL). For free vitamin D, there are no internationally established reference values. Using the suggested values from Zeng et al ( 15 ), only 7 (0.4%) study participants had a sufficient supply of free 25(OH)D of more than 8.50 pg/mL. Another 237 (14.9%) were insufficient and the large majority, 1349 (84.7%), were deficient ( Fig. 1 ). The study group was divided into quintiles, separately according to total and free vitamin D, to analyze the incidence of GDM in each of the quintiles. One-way ANOVA showed no significant difference in GDM incidence in the different quintiles for both total and free 25(OH)D ( P = .831 and .799 respectively) ( Figs. 2 and 3 ). Comparison of GDM rates by quintiles of total 25(OH)D. Comparisons were made using one-way ANOVA. Quintiles of total 25(OH)D were defined in ng/mL as follows: Q1 (<15.94), Q2 (15.94 to <18.50), Q3 (18.50 to <20.67), Q4 (20.67 to <23.80), and Q5 (≥23.80). Abbreviations: 25(OH)D, 25-hydroxyvitamin D; GDM, gestational diabetes mellitus. Comparison of GDM rates by quintiles of free 25(OH)D. Comparisons were made using one-way ANOVA. Quintiles of free 25(OH)D were defined by concentration/pg/mL as follows: Q1 (<3.97), Q2 (3.97 to <4.48), Q3 (4.48 to <4.96), Q4 (4.96 to <5.48), and Q5 (≥5.48). Abbreviations: 25(OH)D, 25-hydroxyvitamin D; GDM, gestational diabetes mellitus. To further explore the association between vitamin D and GDM in this relatively vitamin D insufficient/deficient population, we used multivariate logistic regression, taking into account confounding factors. Factors in Model A are established risk factors for GDM from the available literature, namely, maternal age, BMI, and blood sugar prior to fertility treatments. Considering these factors, there was no statistical association between the degree of either total or free 25(OH)D deficiency with the incidence of GDM, with a P value of .266 and .123, respectively ( Tables 2 and 3 ). For Model B, factors that were significantly different between GDM and non-GDM patients (from Table 1 ) were included in the analysis: maternal age, BMI, blood sugar, AMH, LH, and estradiol ( Tables 4 and 5 ). Similarly, this multivariate logistic regression also showed no significant association ( P = .297 and .094 respectively) ( Tables 4 and 5 ). Multivariate logistic regression for total 25(OH)D and GDM incidence (model A) Comparisons were made using multivariate logistic regression. Abbreviations: 25(OH)D, 25-hydroxyvitamin D; B, regression coefficient; BMI, body mass index; GDM, gestational diabetes mellitus; OR, odds ratio. Multivariate logistic regression for free 25(OH)D and GDM incidence (model A) Comparisons were made using multivariate logistic regression. Abbreviations: 25(OH)D, 25-hydroxyvitamin D; B, regression coefficient; BMI, body mass index; GDM, gestational diabetes mellitus; OR, odds ratio. Multivariate logistic regression for total 25(OH)D and GDM incidence (model B) Comparisons were made using multivariate logistic regression. Abbreviations: 25(OH)D, 25-hydroxyvitamin D; AMH, anti-Müllerian hormone; B, regression coefficient; BMI, body mass index; GDM, gestational diabetes mellitus; LH, luteinizing hormone; OR, odds ratio. Multivariate logistic regression for free 25(OH)D and GDM incidence (model B) Comparisons were made using multivariate logistic regression. Abbreviations: 25(OH)D, 25-hydroxyvitamin D; AMH, anti-Müllerian hormone; B, regression coefficient; BMI, body mass index; GDM, gestational diabetes mellitus; LH, luteinizing hormone; OR, odds ratio.

Most of the 1593 women taking part in this study did not have a sufficient vitamin D supply. In this relatively vitamin D insufficient/deficient cohort of women undergoing IVF/ICSI and embryo transfer, the degree of both total and free vitamin D deficiency before pregnancy was not associated with the incidence of GDM. A total of 256 (16.1%) patients developed GDM, which reflects the high incidence of this pregnancy complication. This is also in accordance with the meta-analysis by Gao et al, which found that the GDM prevalence in China was fairly high at 14.8% in 2019 ( 3 ). The present study also confirms the high prevalence of vitamin D deficiency. Prior to their pregnancy, most patients were either vitamin D insufficient (43.7%) or even deficient (53.4%), totaling 97.1% of the patients. Therefore, only 47 (2.9%) had a sufficient total 25(OH)D supply (≥30 ng/mL), according to current international general guidelines for vitamin D status ( 13 ). Almost all patients in the study population had a lack of vitamin D, which is very common in China, although the percentages vary across the country. Studies in Beijing ( 17 ) and Guizhou ( 18 ) found a very high percentage of vitamin D deficiency among pregnant women (total 25(OH)D < 20 ng/mL), of 96.8% and 92.8%, respectively. Other studies also found a fairly high prevalence of vitamin D insufficiency/deficiency (total 25(OH)D < 30 ng/mL) of 65.9% in Liuzhou ( 19 ) and 97.4% (of whom 57.5% were vitamin D deficient) in Chengdu ( 20 ), which is similar to the findings of our study. Important factors influencing vitamin D status in China include vitamin D supplementation, a lack of sunlight exposure due to a high level of indoor activities, and dietary habits ( 17 , 21 ). Furthermore, only 7 (0.4%) patients had sufficient free 25(OH)D levels according to the suggested cutoff of 8.50 pg/mL suggested by Zeng et al ( 15 ). This shows that there were even fewer patients with sufficient free 25(OH)D levels, compared to the percentage of patients with a sufficient total 25(OH)D supply (as seen in Fig. 1 ). Considering that the measurements were made before embryo transfer, but after hormonal stimulation, a possible explanation could be that the increase in estrogen due to hormonal stimulation led to an increase in DBP synthesis ( 8 ). Because more vitamin D would then be bound to its transport protein DBP, the proportion of free and bioavailable vitamin D would decrease. This is the first study that includes measurements of total and free vitamin D to evaluate its relationship with GDM. Therefore, we cannot compare the measurements of free 25(OH)D to other similar studies. With such a small proportion of the study cohort having a sufficient supply of total and free vitamin D, the findings of this study focus on a vitamin D insufficient/deficient population. In this population, statistical analyses showed that the degree of vitamin D deficiency is not associated with GDM incidence—for both total and free vitamin D. Taking into account confounders, multivariate logistic regression also indicates that there is no association in this cohort between total or free 25(OH)D and GDM ( P = .266 and .123 in Model A and P = .297 and .094 in Model B) ( Tables 2 to 5 ). In other words, the degree of vitamin D deficiency does not affect the incidence of GDM. However, given the low number of women with sufficient vitamin D levels, clear conclusions on whether higher levels of vitamin D would have decreased the risk of GDM cannot be made. Comparing our results to previous studies on the association between total 25(OH)D and GDM incidence in non-ART pregnancies, these studies present heterogeneous results. Meta-analyses by Rizzo et al ( 10 ) and Wei et al ( 22 ) suggested that there is an uncertain relationship between vitamin D and GDM, whereby some studies present a significant association ( 23 , 24 ) and others show similar results to our study ( 25-27 ). We also reviewed similar studies in the last 10 years in a Chinese/Taiwanese population (Supplementary Table S1 ( 28 )). While all 9 studies agree that there is a high prevalence of vitamin D deficiency, 4 of them found a statistically significant relationship between vitamin D and GDM ( 22 , 29-31 ). One of the studies found that vitamin D was neither associated with GDM nor fasting glucose in Chinese women ( 32 ), while the other 4 studies show associations only under certain conditions. Three of them suggest that there is a J-shaped relationship ( 33 ) rather than a linear relationship, that is, that there is only a significant association with GDM above a certain level of vitamin D (>20 ng/mL ( 33 , 34 ) or >30 ng/mL ( 35 )). Another study found that vitamin D was only associated with GDM in the second but not the first trimester of pregnancy. Because most women in our study cohort were vitamin deficient/insufficient, the findings fit to the latter studies. In any case, it is noteworthy to mention that all above-mentioned studies were not conducted in women undergoing IVF/ICSI treatment, a procedure that increases GDM risk. The strengths of this study include that it is a post hoc analysis of an originally prospective, clinical study with a large cohort of the same ethnic background. This reduced an important confounding factor, which is often a problem in large-scale meta-analyses. Furthermore, it is the first study on the association between prepregnancy vitamin D and GDM in women who undergo ART, which additionally includes free vitamin D. A limitation that remains to be established is that there are no widespread guidelines for vitamin D levels for different sexes, ethnicities, or pregnant women, which is very important, as this study solely includes Chinese women. Because it is such a specific cohort, the results are not directly comparable to other study populations. Additionally, vitamin D measurements were made prior to pregnancy, 1 day before embryo transfer. For further analysis, one could measure vitamin D levels throughout the pregnancy, thereby also comparing how total and free vitamin D levels progress during pregnancy, and possibly influence the incidence of GDM. Additional steps for future studies on this topic are to take into account more quantitative data, such as the results of the oral glucose tolerance test. Furthermore, data on the family history of diabetes, which is an important risk factor for diabetes, was not available. It would also be useful to measure more factors, such as insulin resistance, which play a role in the genesis of (gestational) diabetes. In our study, vitamin D measurements were made before pregnancy, which is a unique opportunity in IVF/ICSI patients, as it is possible to study the effects of a certain factor before pregnancy even occurs. This is the first study to investigate the effects of prepregnancy vitamin D concentrations on GDM, which would allow for the prevention of diseases. However, the measurements were made after hormonal stimulation and although free 25(OH)D is known to not be influenced as much as total 25(OH)D, more studies are needed at this point in time to evaluate the reliability. It is a strength of this study that we analyzed both free and total vitamin D, especially given that total vitamin D concentrations might be affected by estrogen after ovarian stimulation for egg retrieval. However, the methods used are based on ELISA technology. Measurements based on liquid chromatography–mass spectrometry (LC-MS) would be more accurate but are very expensive and were not available for the study. This represents a clear study limitation. In the studies that Rizzo et al ( 10 ) reviewed, a correlation between total 25(OH)D and GDM was only found in studies that used LC-MS assays. However, from the 9 studies in Supplementary Table S1 ( 28 ), 2 used LC-MS and one of them also found that there was no statistical association ( 32 ); similarly, studies using immunoassays also found an association ( 29 , 30 ). This study has used a variety of statistical methods, 2 different types of vitamin D determinations, and 2 different regression models, which agree that there is no statistical association between vitamin D and GDM incidence in this cohort. This shows that despite different approaches and independent analysis methods, the result is the same in this cohort. In conclusion, in this prospective study in Changsha, China, with women undergoing IVF/ICSI, the incidence of GDM was not associated with the degree of vitamin D deficiency/insufficiency before pregnancy—for both total and free vitamin D.