Vitamin D and assisted reproduction technologies: current concepts

A role for vitamin D in ovarian steroidogenesis is well established [4], and recent evidence has shown an effect of vitamin D on uterine receptivity [4, 6]. Importantly, vitamin D has also been shown to be involved in the pathophysiology of some disorders of women of childbearing age that are most commonly encountered among women undergoing In Vitro Fertilization (IVF) procedures:

Vitamin D may play a role in human spermatogenesis. The favourable effect of 25(OH) on human spermatozoa has been shown in vitro[26‐28] and three cross-sectional association studies on serum 25(OH)D levels and semen quality have been conducted in young healthy men without infertility problems [27, 29, 30]. The results, however, were quite discordant (Table 1). A single cross-sectional study [31] has also investigated the relationship between vitamin D status and semen parameters in a population of n = 364 infertile men as compared to n = 195 age-matched fertile controls. Interestingly, after adjustment for some relevant confounders, statistically significant associations were observed between 25(OH)D levels and sperm motility (Spearman’s coefficient = 0.12, p = 0.03) and morphology (Spearman’s coefficient = 0.12, p = 0.03) in infertile men (Table 1).

Based on these observations, the purpose of this systematic review and meta-analysis was to perform an in-depth evaluation of clinical studies assessing whether vitamin D status of patients undergoing ART could be related to cycle outcome variables. This issue is of particular interest considering that vitamin D status may be easily adjustable.

Six cohort studies have characterized serum 25(OH)D status in women undergoing ART procedures and addressed the prevalence of vitamin D deficiency among these patients [34, 38‐42] (Table 2). In five studies [38‐42], serum 25(OH)D was categorised according to clinically accepted ranges for vitamin D deficiency (<20 ng/ml), insufficiency (20–30 ng/ml) and replete (>30 ng/ml) [43]. The observed prevalence of vitamin D deficiency as defined by serum 25(OH)D levels <20 ng/m was 21% [41], 26% [42], 27% [38], 31% [39] and 99% [40] respectively in the five studies. In the study by Firouzabadi et al., vitamin D deficiency was instead exclusively defined as 25(OH)D levels < 10 ng/ml [34], and the prevalence of vitamin D deficiency was 75% with only 7% of patients having replete vitamin D levels (≥30 ng/ml) [34]. Interestingly, the studies reporting the highest frequency in vitamin D deficiency (75% for Firouzabadi et al., [34] and 99% for Aleyasin et al., [40]) were both conducted on Iranian women, confirming a strong influence of the socio-economic factors on vitamin D status [44]. After excluding these two studies, the overall mean prevalence of vitamin D deficiency among the population of women who are candidates for IVF procedures was 25% (21-31%), which is slightly lower than that reported for the general population of childbearing age women in the USA (36%) [45]. This observation seems consistent with the demographics of the IVF population, which tends to have higher socio-economic status and education level, which are both factors related to higher vitamin D status [46]. Conversely, the mean observed prevalence of women with replete vitamin D status [serum 25(OH)D levels > 30 ng/mL] was 34% (range 21-42% across studies) after excluding the studies by Firouzabadi et al. [34] and Aleyasin et al. [40].

Four studies have evaluated baseline characteristics of patients found to be significantly associated with vitamin D status, but these studies reported conflicting results [38‐41]. According to Anifandis et al., populations were homogenous with regards to various patients’ characteristics after grouping for vitamin D status [39]. Conversely, Rudick et al. reported significant differences in age and BMI between women with and without vitamin D deficiency, with patients affected being younger and with higher BMI [41]. In addition, Ozkan et al. also noted an inverse correlation between FF 25(OH)D and BMI (r = −0.25, p = 0.03) [38], while Aleyasin et al. described a linear correlation of serum 25(OH)D levels with age (r = 0.28, p = 0.01) but not with BMI [40]. Three studies have also investigated the influence of ethnicity on 25(OH)D levels [38, 41, 42]. Of the two studies conducted on conventional IVF populations, one reported Hispanic whites to have significantly lower serum 25(OH)D levels compared to Asians and non-Hispanic whites (p = 0.01) [41] while the other observed, as expected, significantly lower serum 25(OH)D levels among black versus non-black ethnicities (p = 0.001) [38]. The study conducted on recipient women undergoing an oocyte donation program reported lower 25(OH)D levels among Asians and African-Americans compared to Hispanic or non-Hispanic white (p = 0.02) [42].

Only one small study conducted sequential measurements of serum vitamin D throughout COH for IVF in order to identify cycle-related variations [47], and showed that 1,25-dihydroxyvitamin D concentrations - but not 25(OH)D and 24,25-dihydroxyvitamin D - progressively increased throughout COH [47]. Five studies instead analysed the relationship between serum and follicular fluid (FF) levels of 25(OH)D [34, 38‐40, 47], all confirming that a strong positive correlation exists between peripheral and follicular vitamin D stores (r = 0.74, p < 0.01 according to Potashnik et al., 1992; r =0.94; p < 0.001 according to Ozkan et al.; r = 0.79, p < 0.001 according to Anifandis et al.; r = 0.83, p = 0.001 according to Firouzabadi et al.; r = 0.77, p < 0.001 according to Aleyasin et al.). Of note, this observation seems to suggest that peripheral vitamin D status is a reliable indicator for 25(OH)D availability within the ovary.

The main characteristics and findings of the six studies that have addressed the role of vitamin D status on IVF/ICSI outcomes [34, 38‐42] are shown in Table 2. Two studies exclusively investigated the association between FF 25(OH)D levels and ART outcomes. Ozkan et al. were the first reporting a positive association between FF 25(OH)D levels and IVF outcomes, with patients in the highest tertile of FF 25(OH)D distribution being almost four fold more likely to achieve a clinical pregnancy (CP) compared with patients in the lowest tertile after controlling for potentially confounding factors [odds ratio (OR) for CP 3.83; 95% CI 1.20 – 12.28, p = 0.02]. In their study, FF 25(OH)D levels as an independent predictor of CP were also confirmed by means of multivariate logistic regression (confounders-adjusted OR 1.07; 95% CI 1.01 – 1.13, p = 0.01). In marked contrast with these results, Anifandis et al. observed a negative effect of increasing FF vitamin D levels on IVF outcomes. Women with higher (>30 ng/ml) FF 25(OH)D levels showed decreased embryo quality (mean score of embryo quality 5.6 ± 3.6 vs 7.02 ± 2.5 and 7.96 ± 2.6 respectively, p < 0.05) and reduced CPR (14.3% vs 32.7% and 32.3% respectively, p < 0.05) compared to patients with intermediate (20.1-30 ng/ml) or low (≤20 ng/ml) FF 25(OH)D levels. Conflicting observations have also been derived from the study by Rudick et al. (2012), in which opposite trends of association were observed according to patients’ ethnicity. In fact, after dividing the population into vitamin D deficient (<20 ng/ml), insufficient (20–30 ng/ml) and replete (>30 ng/ml) depending on serum 25(OH)D levels, vitamin D status was positively associated with increasing CPR in non-Hispanic and Hispanic whites (CPR 21%, 36% and 55% in the deficient, insufficient and replete groups respectively, p = 0.01, in non-Hispanic whites; CPR increasing from 15%, 38% and 68% in the deficient, insufficient and replete groups respectively, p = 0.03, in Hispanic whites). In contrast, an opposite inverse relationship between vitamin D status and IVF success was observed in the Asian ethnicity (CPR 64%, 34% and 14% in the deficient, insufficient and replete groups respectively, p = 0.01). Interestingly, the opposite influence of vitamin D status on IVF outcome was also confirmed on LBR among non-Hispanic whites (increasing from 14% in the deficient group to 27% and 47% in the insufficient and replete groups, respectively, p = 0.01) and Asians (with LBR decreasing from 53% to 25% and 9% across the vitamin D deficient, insufficient and replete groups respectively, p = 0.02), but not among Hispanic whites (p = 0.19) [41]. In a second study, the same authors examined serum vitamin D levels among n = 99 recipients of oocyte donation. After dividing patients into vitamin D deficient (<20 ng/ml), insufficient (20–30 ng/ml) and replete (>30 ng/ml) depending on serum 25(OH)D levels, the authors observed a significant increase in both CPR (37%, 37% and 78% respectively, p = 0.004) and LBR (31%, 30% and 59% respectively, p = 0.04) across the three groups, suggesting a specific effect of 25(OH)D levels on IVF outcomes to be mediated by endometrial receptivity [42]. However, a recent study by Firouzabadi et al., showed increased fertilization rates across groups based on serum vitamin D status. In detail, the fertilization rates associated with serum vitamin D deficiency (<10 ng/ml), insufficiency (10–29 ng/ml) and sufficiency (30–100 ng/ml) were 43%, 53%, and 59%, respectively (p = 0.05). Of note, no statistical significance was reached and vitamin D status was in general low in the population studied - with only 7% of women being vitamin D sufficient [34]. Similarly, results from the study by Aleyasin et al., which showed that CPR did not change significantly across groups based on increasing tertiles of FF vitamin D levels (29%, 30% and 30% respectively, p = 0.9) are difficult to interpret because all women in the population except for one (n = 81/82) were vitamin D deficient in the serum [40].

When pooling data for meta-analysis, data on the effect of vitamin D serum deficiency (<20 ng/ml) on CPR after IVF/ICSI procedures were extracted [34, 40‐42]. Categorisation of patients was not possible in the study by Aleyasin et al. due to the fact that all patients except for one were deficient [40]. Overall, the remaining three studies account for n = 353 women undergoing IVF/ICSI procedures and fresh embryo transfer, of which n = 115 were vitamin D deficient [as defined by serum 25(OH)D levels <20 ng/ml] and n = 238 were vitamin D sufficient [serum 25(OH)D levels ≥20 ng/ml]. Results are shown in Figure 2. A clinical pregnancy was achieved in 46/115 (40%) women with vitamin D deficiency, and in 109/238 (46%) in those with sufficient serum 25(OH)D levels. The RR of clinical pregnancy ranged from 0.60 to 1.84 [48, 49] across studies, and pooling of the results yielded a common RR of 0.89 (95% CI 0.53 to 1.49), showing a lower but not statistically significant likelihood of CP for vitamin D deficient women undergoing IVF/ICSI procedures compared with vitamin D sufficient patients (Figure 2).

The issue of whether vitamin D levels are reliable predictors of ART outcomes is still controversial. Evidence is still poor, because no RCTs are currently available and results of existing small cohort studies are very heterogeneous. Consequently, the meta-analysis performed herein fails to identify a significant association between vitamin D status and CPR after ART. Marked differences in exposure categorisation, analytic approaches, outcomes considered and general methodological design characterize the identified studies. More specifically:

Due to the presence of several sources of controversy in the interpretation of the relationship between vitamin D deficiency and poor IVF outcome, we have employed the nine criteria proposed by Austin Bradford Hill, which still stand as foundation milestones for addressing causation conditions [50]. Five (“specificity“, “temporality“, “biological plausibility“, “experimental evidence“, “analogy“) out of nine Hill’s criteria for causality are indeed fulfilled, two (“consistency” and “biological gradient”) are fulfilled by all currently available studies except for one, while the “specificity” criterion cannot be determined due to insufficient data and the “strength of association” criterion is not fulfilled likely due to the small sample size of the studies and the low magnitude of the effect. Hence, according to the Hill’s criteria, a causal relationship between vitamin D deficiency and negative outcomes should be recognised, but further research is needed to determine the strength of the association and the specific underlying mechanisms [50].

In summary, consistent heterogeneity exists among available studies and important factors need to be considered in this regard such as insufficient sample size, ethnicity, BMI, countries of origin, socio-economic status and the presumably mild magnitude of the effects. This latter aspect should however not disregard the potential relevance of vitamin D status in an IVF setting for two main reasons. Firstly, there is growing evidence that maternal vitamin D status may influence pregnancy outcome, with consistent evidence demonstrating a role of vitamin D insufficiency in pre-eclampsia and preterm birth [5]. Albeit less robust, there is also some data suggesting a relationship with gestational diabetes mellitus and late growth restriction [5]. Hence, the relevance of an advantage of vitamin D sufficiency in increasing the chances of pregnancy in IVF cycles would be strengthened by the potential, subsequent advantage in terms of improvements in obstetrics complications and neonatal health even if the magnitude of the effect is still to be measured [51]. Secondly, vitamin D deficiency or insufficiency is easily amenable to correction. Oral vitamin D supplementation at therapeutic doses is indeed a simple and cheap intervention without significant side effects. Importantly, recent studies have shown that maternal supplementation with vitamin D during second and third-trimester pregnancy at doses up to 4000 IU is safe and can effectively improve maternal vitamin D status [52]. However, results of the ongoing RCTs on vitamin D supplementation prior to IVF procedures are not yet available.