Among pregnant women with relatively low average baseline vitamin D status in Dhaka, 3rd-trimester vitamin D3 supplementation (35,000 IU/week) significantly raised maternal and neonatal (cord blood) serum 25(OH)D concentrations above the IOM cut-off for sufficiency (50 nmol/L) in virtually all participants without inducing hypercalcemia or other apparent short-term clinical adverse effects. This study contributes pharmacokinetic data relevant to South Asia, as well as preliminary evidence in support of a vitamin D3 no observed adverse effect level (NOAEL) of 35,000 IU/week in the third trimester of pregnancy.
The present findings build primarily on those of Hollis et al. in South Carolina, who randomized 502 pregnant women at 12 to 16 weeks gestation to receive vitamin D3 at doses of 400 IU/day, 2000 IU/day, or 4000 IU/day; 350 participants (70%) were followed-up until delivery [13
]. From a baseline 25(OH)D of 58 nmol/L, 2000 IU/day and 4000 IU/day raised 25(OH)D to means at the time of delivery of 98 nmol/L (rise of 40 nmol/L) and 111 nmol/L (rise of 52 nmol/L), respectively [13
]. In the 4000 IU/day group, 82% attained 80 nmol/L at delivery, compared to 97% in the present study (5000 IU/day). In 2012, Dawodu et al. presented unpublished findings from a trial in United Arab Emirates in which 2000 IU/day and 4000 IU/day raised mean 25(OH)D from ~20 nmol/L to ~65 nmol/L (rise of 45 nmol/L) and ~90 nmol/L (rise of 70 nmol/L) at delivery, respectively [24
]. Our modeled mean increment in maternal 25(OH)D of 102 nmol/L corresponded to a rise of 0.82 nmol/L/mcg/day (95% CI, 0.72 to 0.91), which was similar to the value often cited for non-pregnant adults (~0.7 nmol/L/mcg/day [25
]) but smaller than the effect observed with lower vitamin D3 doses (e.g., 1.6 nmol/L/mcg/day for doses of ~800 IU/day [27
]). This is consistent with the hypothesized constraint in hepatic vitamin D–to–25(OH)D conversion that occurs at a 25(OH)D concentration of about 80–90 nmol/L [28
In a preceding pilot trial in Dhaka, we found that 3rd
-trimester regimens of 14,000 IU/week (≈2000 IU/day) and 35,000 IU/week (≈5000 IU/day) led to mean 25(OH)D concentrations of 76 nmol/L (rise of 36 nmol/L) and 98 nmol/L (rise of 57 nmol/L), respectively, after 10 weeks of supplementation [16
]. Thus, compared to the present study, the average response to 35,000 IU/week was less pronounced in the pilot study. However, direct between-study comparisons are tempered by differences in study design (e.g., shorter period of follow-up in the pilot study) and 25(OH)D assays. In comparison to the LC-MS/MS method used here, the immunoassay used in the pilot study (Diasorin Liaison Total) may have under-estimated 25(OH)D because of raised DBP concentrations in pregnancy [29
] or relative imprecision at the higher 25(OH)D range attained in the vitamin D group [30
Similar to previous trials [13
], we observed substantial inter-individual variability in the maternal Δ25(OH)D, such that the SD on the mean 25(OH)D nearly doubled from 18.4 nmol/L at baseline to 30.7 nmol/L at delivery in the vitamin D group; in contrast, in the placebo group, the SD at baseline was 20.1 nmol/L and at delivery was 18.1 nmol/L. However, we could not identify any participant factor other than intervention group that significantly explained Δ25(OH)D. The range of published mean cord blood 25(OH)D concentrations in South Asia, from 21 [32
] to 59 [33
] nmol/L, bounds the mean cord 25(OH)D of 39 nmol/L in the control group. A normal cord 25(OH)D range based on functional outcomes has not been defined; however, Zeghoud et al. (1997) reported that newborn 25(OH)D
30 nmol/L was associated with suppression of PTH concentrations [34
], and 25(OH)D
50 nmol/L (the 25(OH)D threshold for sufficiency set by the IOM) has been associated with a low risk of rickets [35
]. In our study, 95% of newborns had a cord-maternal 25(OH)D ratio of 0.5 to 1.5, suggesting that universal maternal prenatal 25(OH)D
100 nmol/L would ensure that nearly all cord concentrations are above 50 nmol/L.
We remain guarded in our interpretation of these pharmacokinetic data because the clinical implications of the observed changes in maternal-fetal 25(OH)D and resultant maternal PTH suppression in pregnancy remain to be defined. Hollis et al. concluded from their trial findings “that the current vitamin D EAR and RDA for pregnant women issued in 2010 by the IOM should be raised to 4000 IU of vitamin D per day” [13
]. However, we do not believe that above-RDA doses should yet be recommended as routine practice. Further clinical data are required before conclusive statements can be made regarding the appropriateness of high-dose antenatal vitamin D supplementation. However, like Hollis et al., we did not encounter any evidence of vitamin D toxicity. Vitamin D did not affect average urinary calcium excretion, yet we did see a small but significant increase in the mean adj-Ca in the vitamin D group compared to the placebo group. Several factors suggested this was not likely to be evidence of harm. First, this trend was attenuated and non-significant when the analysis involved uncorrected serum calcium, a less conservative endpoint used in previous trials. Second, we did not observe a linear escalation in adj-Ca over the entire duration of supplementation, but rather found an initial increase in the rate of the natural rise in adj-Ca that occurs in late pregnancy [36
] followed by an attenuation of the between-group difference after about 2 months of supplementation. This suggested a homeostatic adjustment to the altered vitamin D status, with no further increase in serum calcium after a steady-state 25(OH)D was reached. Third, a positive association between adj-Ca and 25(OH)D was apparent in the range of 25(OH)D below 100 nmol/L but not above 100 nmol/L; evidence of possible harm would be more consistent with the opposite finding of a serum calcium that was tightly controlled (i.e., unassociated with 25(OH)D) within the lower range of 25(OH)D, but deflected upward at higher 25(OH)D. All maternal uncorrected and adj-Ca values were below the threshold for hypercalcemia used by the IOM in the 1997 vitamin D DRIs (2.75 mmol/L) [10
] and the 2010 revision (2.63 mmol/L) [7
]. We set a conservative upper limit of reference range for trial monitoring (2.60 mmol/L), which was crossed by one participant in the vitamin D group; however, this was not logically attributable to vitamin D toxicity based on normal repeat serum calcium, normal ca:cr, and a 25(OH)D that was well within the range considered to be non-hypercalcemic [37
]. Based on biochemical findings, and the lack of apparent clinical adverse effects, we cautiously conclude that 35,000 IU/week for an average of 10 weeks (up to 18 weeks) was tolerated by study participants, and could be considered an upper dose limit for use in future research.
This RCT had several limitations. It was not designed to draw precise inferences regarding pregnancy/birth outcomes, which limits its clinical applications. The biochemical dose–response curve would have been improved by the inclusion of other dose levels. Because of the high dose of vitamin D not previously studied in pregnancy, enrolment was limited to healthy women at low risk of pregnancy complications, and who were most likely to adhere to the protocol; as such, generalizability of the findings may be limited. In addition, enrolment was not conducted throughout the entire season, but rather occurred during a period in which vitamin D status was declining from its summer peak. Mechanistic inferences were limited because we did not measure serum concentrations of 1,25(OH)2D, vitamin D3, or DBP. As well, we did not calculate total intake of vitamin D due to the lack of adequate information regarding vitamin D content in the Bangladeshi food supply.