Objective
We sought to characterize the relationship between serum 25-hydroxy vitamin D (25-OH D) levels and implantation and clinical pregnancy rates in women who undergo a euploid blastocyst embryo transfer.
Study Design
This retrospective cohort study, conducted in an academic setting, included 529 cycles in which comprehensive chromosome screening was performed as part of routine infertility care with an autologous transfer of 1 or 2 euploid blastocysts. After excluding repeat cycles there were 517 unique cycles representing 517 women for evaluation. Vitamin D levels from serum samples obtained on the day of ovulation trigger in the fresh in vitro fertilization cycle were analyzed. The primary outcome was ongoing pregnancy rate as defined by sonographic presence of fetal heart rate at >8 weeks’ gestation.
Results
For the population as a whole, serum vitamin D ranges and pregnancy outcomes did not correlate. Furthermore, pregnancy rates did not differ when comparing women in different strata of vitamin D levels (<20, 20-29.9, and ≥30 ng/mL). No meaningful breakpoint for vitamin D levels and ongoing pregnancy rate was identified using receiver operating characteristic analysis with the resultant line possessing an area under the curve of 0.502. Multivariate logistic regression controlling for age, transfer order, race, season, and body mass index did not yield a different result. The study was powered to detect an 18% difference in ongoing pregnancy rates between patients grouped by the 3 vitamin D ranges.
Conclusion
In women undergoing euploid embryo transfer, vitamin D status was unrelated to pregnancy outcomes. Measuring serum 25-OH vitamin D levels does not predict the likelihood that euploid blastocysts will implant. These results may not apply to women who do not undergo extended embryo culture, blastocyst biopsy for comprehensive chromosome screening, and euploid embryo transfer.
There has been much interest in what has been deemed an epidemic of 25-hydroxy vitamin D (25-OH D) (the precursor to the hormone 1,25 [OH 2 ] vitamin D) deficiency in the United States. This deficiency has been linked to a wide range of chronic disease states including malignancy, cardiovascular, metabolic, and immune diseases. Vitamin D has also been implicated in reproductive health with observational data showing vitamin D deficiency is associated with adverse pregnancy outcomes including preeclampsia, small for gestational age in singleton live births, and increased risk of cesarean delivery. However, intervention studies normalizing serum vitamin D concentrations in pregnancy fail to show improved outcomes in terms of birthweight. Given its potential for impact in very early gestation, vitamin D status in patients undergoing infertility treatment has garnered considerable interest.
To date, there are limited data on the impact of vitamin D status and outcomes from infertility treatment. One observational study examined vitamin D in the serum and follicular fluid (FF) of 84 patients undergoing in vitro fertilization (IVF). The data showed that serum and FF vitamin D levels were highly correlated and that FF vitamin D levels were associated with clinical pregnancy, defined as a visible gestational sac on transvaginal ultrasound. Overall, FF vitamin D levels in the clinical pregnancy group were 34.4 ng/mL vs 25.6 ng/mL in the nonpregnant group. It is important to note that although FF levels were different, this does not determine whether the effect was at the level of the ovary/oocyte or the endometrium, given that serum levels correlated well with FF levels.
A follow-up validation study with 88 patients evaluated the relationship between serum vitamin D levels and pregnancy rates within various race categories. Lower vitamin D levels correlated with diminished pregnancy rates in non-Hispanic whites but no differences were noted among Hispanic whites. Interestingly, vitamin D levels and pregnancy rates were inversely related in Asians with higher levels prognosticating lower outcomes. These types of studies are further complicated by the intrinsic difficulty of studying parameters in relation to self-reported race vs an individual’s genetic ancestry.
Although these studies show a difference in pregnancy rates between groups with disparate vitamin D levels, they do not evaluate the mechanism by which vitamin D impacts outcomes by attempting to control for the endometrium and/or embryonic reproductive competence. This was done in a subsequent study evaluating vitamin D levels in 99 donor oocyte recipients who had an embryo transfer in a synthetic cycle. Recipient patients who had nonreplete vitamin D (<30 ng/mL) had a live birth rate of 31% vs 59% in those patients who were vitamin D replete. There was no difference in adjusted live-birth rates among recipients who were vitamin D deficient (<20 ng/mL) vs vitamin D insufficient (<30 mg/mL). Given that this was an oocyte donor-recipient model with presumably high-quality embryos, the diminution in pregnancy rates was attributed to impaired endometrial receptivity.
The data on vitamin D in health and disease, particularly in obstetric and infertility health and disease, are still evolving. As such, we sought to further study its predictive value in infertile patients undergoing stimulated IVF cycles by controlling as best possible for extrinsic factors such as blastulation, aneuploidy, and embryoendometrial synchrony. In doing so, the impact of vitamin D on pregnancy rates and role in reproductive biology might be further elucidated.
Materials and Methods
Patient population
In this retrospective cohort study, we identified 529 autologous first fresh or cryopreserved euploid embryo transfers at a single institution from December 2012 through December 2013. We subsequently excluded from analysis 12 transfers that represented repeat cycles, leaving an analytic sample of 517 unique cycles representing 517 women. The patients in this study had undergone comprehensive chromosome screening as part of their routine infertility care as a way of improving pregnancy rates and decreasing miscarriage rates. This is offered to all patients at our center as part of routine infertility care when undergoing IVF. All patient characteristics and cycle parameters were identified from the electronic medical record.
All patients underwent IVF care per clinic standard. Standard regimens for controlled ovarian hyperstimulation using purified urinary follicle stimulating hormone or recombinant follicle stimulating hormone and leutinizing hormone activity in the form of low-dose human chorionic gonadotropin (hCG) or human menopausal gonadotropins along with gonadotropin releasing hormone (GnRH) agonist (long down-regulation or microdose flare) or GnRH antagonist to prevent a premature leutinizing hormone surge were utilized. Monitoring of IVF cycles were per practice routine. Oocyte maturation was induced with recombinant hCG (typically 500 μg) or purified urinary hCG (typically 10,000 IU) or with GnRH agonist (leuprolide acetate 2 mg in 2 doses 12 hours apart) ± 1500 IU of hCG when 2 or 3 follicles reached or exceeded 17-18 mm or when the follicular cohort was deemed to be mature by the patient’s primary physician.
Transvaginal oocyte aspiration was performed approximately 36 hours later. Cumulus stripping was performed and all mature oocytes underwent intracytoplasmic sperm injection owing to the need to perform preimplantation genetic testing. Once embryos reached the blastocyst stage, a trophectoderm biopsy was performed and processed for real-time, quantitative polymerase chain reaction–based comprehensive chromosome screening as previously described.
Ultrasound-guided fresh embryo transfer was performed on day 6 if the embryo and endometrium were synchronous. Synchrony was determined by ensuring there was not a premature progesterone elevation in the late follicular phase and that embryos had fully blastulated to allow for trophectoderm biopsy late in the afternoon of day 5. Luteal support was provided with vaginal progesterone in fresh cycles. Embryos that blastulated late were biopsied on day 6 and vitrified for potential transfer in a subsequent synthetic cycle. Synthetic cycles were similarly conducted per clinic routine with estradiol for endometrial proliferation and intramuscular progesterone in oil to induce secretory transformation. In synthetic cycles, intramuscular progesterone was continued for luteal support as is the clinic routine.
Pregnancy outcomes were defined as not pregnant (no appearance of serum hCG 8-9 days after blastocyst transfer), chemical pregnancy loss (detection of serum hCG without progression to a sonographically visible gestational sac), clinical pregnancy loss (pregnancy loss after ultrasound confirmation of a gestational sac), and ongoing/delivered (sonographic presence of fetal heart rate beyond the first trimester and ongoing or delivered at the end of the study period).
Patient cycle parameters were identified from the electronic medical record including age at time of transfer and self-reported ethnicity. Race was categorized as American Indian/Alaska Native, Native Hawaiian/Pacific Islander, Asian, non-Hispanic white, African American, Hispanic/Latina, or other.
Vitamin D status
Patient vitamin D status was determined by measuring serum levels of 25-hydroxy vitamin D (25-OH D) in stored, frozen, never previously thawed samples drawn on day of ovulation trigger using the Siemens Advia Centaur XP immunoassay system (Siemens, Munich, Germany). The intraassay and interassay variation coefficients were 8.4 and 9.2, respectively. Serum 25-OH D levels were categorized as deficient (<20 ng/mL), insufficient (20-29.9 ng/mL), and replete (≥30 ng/mL).
Data analysis
Pregnancy outcomes in relation to mean serum vitamin D level were assessed. The use of vitamin D supplementation was not assessed and serum levels are reflective of all vitamin D sources including dietary, exogenous supplements, and endogenous conversion of sunlight. This was done by first comparing mean vitamin D levels in patients who became pregnant (chemical, clinical, ongoing/delivered) to those who did not become pregnant. Groups were analyzed utilizing a 2-tailed Student t test.
To identify an optimal breakpoint for vitamin D level for analysis, a receiver operator characteristic (ROC) curve was constructed utilizing continuous vitamin D serum levels and ongoing pregnancy (evidence of fetal heart beat at the time of discharge at 8 weeks’ gestation) or not ongoing (which includes not pregnant, chemical pregnancy loss, clinical pregnancy loss). Next, serum vitamin D ranges were categorized according to clinically accepted ranges of deficiency (<20 ng/mL), insufficiency (20-29.9 ng/mL), and replete (>30 ng/mL). Pregnancy outcomes were compared between groups using a Pearson χ 2 statistic.
Finally, a multivariate logistic regression was employed to control for patient and cycle characteristics. Specifically, the factors included in the regression were patient age at the time of oocyte retrieval, the number of euploid embryos transferred (either 1 or 2), number of prior IVF cycles patients underwent, body mass index, self-reported ethnicity, the season during which the oocyte retrieval took place (accounting for changes in sun exposure leading into retrieval), and season in which the embryo transfer took place (accounting for changes in sun exposure during implantation and early pregnancy). Each of these were controlled for individually and then all factors together.
In all cases, an alpha error <0.05 was considered significant. The retrospective analysis of data was institutional review board approved by Western IRB, protocol 20021333.
Results
Demographics
Of the 517 women who underwent a euploid fresh or cryopreserved transfer during the study time frame, 40.2% (208/517) had serum 25-OH D levels that were categorized as deficient (<20 ng/mL), 41.4% (214/517) had levels categorized as insufficient (20-30 ng/mL), and 18.4% (95/517) were considered replete (>30 ng/mL).
The mean patient age was 35.1 (range, 21.7–45.8) years and the mean number of embryos transferred was 1.25 (range, 1–2). Of the 517 patients, 196 had undergone prior failed retrieval cycles. The mean number of prior retrieval IVF cycles the patients had undergone outside of the retrieval cycle studied here was 0.72, median was 0, and mean was 0. Patient’s self-reported race distribution was American Indian/Alaska Native (n = 1, 0.2%), Native Hawaiian/Pacific Islander (n = 0, 0.0%), Asian (n = 96, 18.7%), non-Hispanic white (n = 362, 70.0%), African American (n = 22, 4.3%), Hispanic/Latina (n = 22, 4.3%), or other (n = 14, 2.7%).
Patient demographics are summarized in Table 1 .
| Demographic | Vitamin D range | P value | ||
|---|---|---|---|---|
| Deficient, <20 ng/mL | Insufficient, 20-29.9 ng/mL | Replete, ≥30 ng/mL | ||
| Age, y | 34.5 (4.5) | 35.1 (4.0) | 36.4 (4.2) | < .01 a |
| BMI | 25.7 (5.8) | 25.1 (5.5) | 24.1 (4.8) | .06 |
| No. of patients who had undergone IVF previously | 62 (32.5%) | 90 (41.9%) | 44 (45.8%) | .04 b |
| Average no. of euploid embryos transferred | 1.3 (0.46) | 1.2 (0.42) | 1.2 (0.37) | .03 c |
| Race | < .001 b | |||
| White | 113 (54.9) | 165 (76.8) | 84 (87.5) | |
| African American | 13 (6.3) | 8 (3.7) | 1 (1.0) | |
| Asian | 61 (29.6) | 28 (13.0) | 7 (7.3) | |
| Other | 19 (9.2) | 14 (6.5) | 4 (4.2) | |
a P < .01 for comparisons between groups 1 and 3; 2 and 3
b P < .001 for comparisons between groups 1 and 2; groups 1 and 3
ROC curve
An optimal breakpoint for the relationship between vitamin D level and ongoing pregnancy was not identified. The ROC curve is shown in the Figure . The resultant line had an area under the curve of 0.502, very nearly approximating the line of no discrimination.
Pregnancy outcomes
Overall mean serum vitamin D ranges and pregnancy outcomes (ongoing pregnancy vs no ongoing pregnancy, which includes no pregnancy and chemical or clinical pregnancy losses) were compared. The mean serum vitamin D level was 24.5 ng/mL for patients with no ongoing pregnancy and 23.9 ng/mL for patients with an ongoing pregnancy ( P = .88).
A subsequent analysis involved utilizing vitamin D ranges of <20, 20-29.9, and ≥30 ng/mL compared to pregnancy outcome. Table 2 shows ranges compared to not pregnant, chemical loss, clinical loss, and ongoing pregnancies. Ongoing pregnancy was the most common outcome in each group and when comparing the outcomes, but there was no difference in vitamin D level and pregnancy outcome ( P = .43).
