Bone turnover in pregnancy, measured by urinary C-terminal telopeptide of type I collagen (CTX), is influenced by vitamin D supplementation and is associated with maternal bone health: Findings from the MAVIDOS trial.

BACKGROUND
The pattern of change in maternal bone turnover throughout pregnancy is poorly characterized.


OBJECTIVE
We investigated changes across pregnancy in a marker of maternal bone resorption, urinary C-terminal telopeptide of type I collagen (CTX), the influence of gestational vitamin D supplementation, and associations between CTX and maternal postnatal bone indices.


DESIGN
The Maternal Vitamin D Osteoporosis Study (MAVIDOS) is a randomized, double-blind, placebo-controlled trial of 1000 IU/day cholecalciferol vs placebo from 14 weeks' gestation to birth. Maternal second void urinary α- and β-CTX were measured (Enzyme Linked Immunosorbent Assay, ELISA) at 14 and 34 weeks' gestation; DXA was performed within 2 weeks post-partum. Mann-Whitney rank sum, Spearman's rank correlation and linear regression were used to compare median CTX values within and between groups from early to late pregnancy, and associations with maternal bone outcomes.


RESULTS
372 women had CTX and 25(OH)-vitamin D measured in early and late pregnancy. CTX at 14 and 34 weeks' gestation were correlated in both placebo (r = 0.31) and cholecalciferol (r = 0.45) groups (P < 0.0001). Median CTX(μg/mmol creatinine) increased from 14 to 34 weeks' gestation in both groups (n = 372 total) [placebo (n = 188) 223.6 to 449.7; cholecalciferol (n = 184) 222.3 to 419.3; P = 0.03 for placebo vs cholecalciferol difference in CTX at 34 weeks' gestation]. The conditional increase in CTX(SD) from early to late pregnancy was greater in the placebo group (n = 188) than in the cholecalciferol group (n = 184) [placebo, mean(SD) 0.16(0.92); cholecalciferol, -0.16(1.06); p difference < 0.01]. Higher CTX at 34 weeks' gestation was associated, similarly in both groups, with lower maternal total hip and lumbar spine bone mineral content (BMC) and bone mineral density, e.g., lumbar spine BMD [β= -0.02 g/cm2/SD increase in CTX; 95%CI: -0.027, -0.002; P = 0.02, n = 283].


CONCLUSIONS
Maternal urinary CTX, a bone resorption marker, rises through pregnancy, though to a lesser degree with gestational cholecalciferol supplementation, and is inversely associated with maternal bone mass post-partum.Clinical Trial Registry number and website: ISRCTN:82927713; EUDRACT:2007-001716-23.


Introduction
The human fetal skeleton begins to mineralize between 8 and 12 weeks of gestation when primary ossification centers form in the vertebrae and long bones ( 1 ). Approximately 30 g calcium are required in total for its development, 80% of which is accrued in the third trimester. Maternal calcium homeostasis adapts to meet the calcium demands of the developing fetus (1)(2)(3), particularly through increased intestinal absorption and renal reabsorption of calcium. The maternal skeleton's role in supplying the fetus, and how it might respond to interventions such as vitamin D supplementation, are not well defined, because of the difficulty in using DXA, which involves ionizing radiation, in pre gnanc y. Biochemical markers of bone turno v er offer a noninv asi ve method of monitoring changes in bone resorption or formation during pre gnanc y ( 4 ), and pro vide some insight into the impact of pre gnanc y on maternal bone and its contribution to fetal development. There are 2 groups of bone turno v er mark ers: mark ers of bone formation [such as N-terminal collagen extension propeptide (P1NP), bone alkaline phosphatase] and markers of resorption [such as C-terminal telopeptide of type I collagen (CTX), N-terminal telopeptide of type I collagen (NTX), deoxypyridinoline, hydroxyproline]. In the vitamin Ddeficient state, plasma calcium concentrations may be reduced with a consequent rise in parathyroid hormone (PTH), leading to increased bone resorption ( 5 ). Indeed, at the population level, there is evidence of inverse associations between 25hydroxyvitamin D [25(OH)D] concentrations and markers of bone resorption such as CTX (6)(7)(8)(9). Ho we ver, fe w studies have investigated these relations across normal pregnancies, or the influence of vitamin D supplementation (10)(11)(12)(13)(14)(15)(16)(17).
The MAVIDOS (Maternal Vitamin D Osteoporosis Study) trial of vitamin D supplementation in pre gnanc y (of which the primary aim was to investigate the effect of gestational cholecalciferol supplementation on offspring bone mass) provides an opportunity to study the impact of this supplement on a marker of maternal bone resorption (CTX) and maternal postpartum bone indices ( 18 , 19 ). As the basis for this post hoc analysis, we hypothesized that vitamin D supplementation in pre gnanc y would result in reduced osteoclastic bone resorption, and thus that we would observe 1 ) an inverse association between vitamin D supplementation [or 25(OH)D concentration] and CTX concentrations, and 2 ) an inverse association between maternal CTX concentrations and measures of maternal postpartum bone mass.

MAVIDOS (ISRCTN 82927713;
EudraCT 2007-001716- 23) was a multicenter, double-blind, randomized, placebo-controlled trial of vitamin D supplementation in pre gnanc y. The primary outcome was neonatal bone mass. A detailed description of the study methods ( 18 ) and primary findings have been published previously ( 19 ).
Briefly, women attending 1 of 3 UK hospitals [University Hospital Southampton National Health Service (NHS) Foundation Trust, Southampton, United Kingdom; Oxford University Hospitals NHS Foundation Trust, Oxford, United Kingdom; Sheffield Teaching Hospitals NHS Foundation Trust (University of Shef field), Shef field, United Kingdom] for early pre gnanc y ultrasound screening (11)(12)(13)(14) weeks of gestation) between 6 October, 2008 and 11 February, 2014 were invited to participate in the study ( 19 , 20 ). Inclusion criteria were age > 18 y, singleton pre gnanc y, and gestation < 17 wk based on last menstrual period and ultrasound measurements. Women with known metabolic bone disease, renal stones, h yperparath yroidism, or hypercalciuria; those taking medication known to interfere with fetal growth; those with fetal anomalies on ultrasonography; and those already using > 400 IU vitamin D/d supplementation were excluded. A screening blood sample was obtained and analyzed on the local NHS platform [all 3 laboratories (Southampton, Oxford, and Sheffield) participate in the Vitamin D External Quality Assessment Scheme (DEQAS) vitamin D quality assurance system ( http:// www.deqas.org/ )]. Women with 25(OH)D between 25 and 100 nmol/L and serum calcium < 2.75 mmol/L were eligible to enroll fully in the study. Participants were randomly assigned in a doubleblind design to either 1000 IU cholecalciferol/d or matched placebo (Merck KGaA/Sharp Clinical Services; previously DHP-Bilcare), which was commenced before 17 weeks of gestation. All participants received standard antenatal care and could continue self-administration of dietary supplements containing ≤400 IU vitamin D/d. Data from Southampton participants were used in the current analysis because CTX was not measured in Oxford and Sheffield participants.

Maternal assessments during pregnancy
Before commencing the study medication, and again at 34 weeks of gestation, the women attended their hospital for a detailed assessment of diet (including supplement use), lifestyle (smoking, physical activity, employment), and health (past medical history, current medication use) using interviewer-led questionnaires. Ethnic group was self-reported by the participant as "White, Black Caribbean, Black African, Black Other, Indian, Pakistani, Bangladeshi, Chinese, Other Asian group or Other (specify)." This was then categorized as white or nonwhite. Anthropometric measurements included height, measured to the nearest 0.1 cm using a stadiometer (Seca), and weight, assessed to the nearest 0.1 kg using calibrated electronic scales (Seca).

Assessment of 25(OH)D and CTX
On the day that the study medication was dispensed and at 34 weeks of gestation, a nonfasted venous blood sample was obtained, and serum stored at −80 • C. 25(OH)D was assessed by chemiluminescent immunoassay (Liaison automated platform, Diasorin). All samples were analyzed in a single batch at the end of the study at Medical Research Council (MRC) Human Nutrition Research, Cambridge, United Kingdom. Details of assay performance and quality control through participation in DEQAS, National Institute of Standards and Technology (NIST), and the UK National External Quality Assessment Service (NEQAS) are given elsewhere ( 21 , 22 ).
Participants were asked to provide a second-void earlymorning urine sample to measure urine αand β-CTX on the day the study medication was started ( ∼14 weeks of gestation) and again at 34 weeks of gestation. Owing to the diurnal variation in CTX, in addition to the practical difficulties of obtaining a fasted blood sample from pregnant women, second-void early-morning urine collection by the participant at home was selected as the most appropriate measure of bone resorption. All samples were stored at −70 • C and those from Southampton participants were sent in a batch for measurement by ELISA (Immunodiagnostic Systems) at the Academic Unit of Bone Metabolism, University of Shef field, Shef field, United Kingdom. Total CTX (sum of the α and β forms) is expressed as a ratio to urine creatinine ( μg/mmol) ( 23 ). The assay had an interassay CV of 6% ( 24 ).

Maternal and neonatal DXA
Mothers and neonates underwent DXA assessment; mothers underwent DXA at the lumbar spine and hip sites, neonates at the whole-body and lumbar spine sites (instrument at the Southampton center: Hologic Disco v ery, Hologic Inc.) within 2 wk of delivery. The instrument underwent daily quality control. To maximize the scan quality, infants were undressed, clothed in a standard towel, fed, and pacified before the assessment. The total radiation dose was estimated to be 0.04 mSv, equi v alent to ∼7 d exposure to background radiation in the United Kingdom. All DXA images were re vie wed for mo v ement artefacts and quality by 2 operators who were blinded to treatment allocation.

Ethics
The trial was appro v ed by the Southampton and South West Hampshire Research Ethics Committee (07/H0502/113). MAVIDOS was re gistered prospectiv ely (ISRCTN: 82927713; EUDRACT: 2007-001716-23); full approval from the UK Medicines & Healthcare products Re gulatory Agenc y (MHRA) was granted, and written informed consent was obtained from all participants.

Statistics
Women who had urinary CTX and 25(OH)D measured at either 14 or 34 weeks of gestation, had not taken cholecalciferol supplements of > 400 IU daily in addition to the study intervention, and who delivered a live-born infant were included in the analysis. All outcomes were assessed for normality using visual inspection. Maternal characteristics in the placebo and vitamin D-supplemented groups were documented. Comparisons between median CTX concentrations in the placebo and cholecalciferol-supplemented groups in early pre gnanc y ( ∼14 weeks of gestation) and late pre gnanc y (34 weeks of gestation) were made using the Mann-Whitney Rank Sum test. In order to a v oid problems from regression to the mean and correlation between baseline value and change, early-and latepre gnanc y CTX were summarized into a change in CTX variable, calculated as the residual from the regression of the later value on the earlier; differences between groups were tested using a t test ( 25 ). This change in CTX variable was then used as an outcome variable in linear regression to explore maternal factors associated with change. Change in CTX was also used as a predictor in regression analyses examining associations between maternal CTX and neonatal bone indices. Spearman's correlation was used to test associations between CTX in early pre gnanc y and late pre gnanc y, and between CTX (in early or late pre gnanc y) and maternal serum 25(OH)D measurements (in early or late pre gnanc y).
The distribution of CTX in early and late pre gnanc y was positi vely ske wed and was therefore transformed and standardized using a Fisher-Yates transformation for inclusion in regression analyses. Fisher-Yates transformed CTX values were used as a predictor in regression analysis with DXA-assessed bone outcomes; β coefficients represent the change in bone outcome per 1-SD increase in CTX. In order to remo v e the additional influence of season on 25(OH)D concentrations, further analyses included adjustment for season of birth, using UK Meteorological Office classifications: winter (December-February), spring (March-May), summer (June-August), and autumn (September-No v ember). Co variates identified as associated with maternal CTX concentrations and known to be associated with bone mass were included in CTX-bone regression models. All analyses were performed in Stata version 14 (Statacorp).

Characteristics of the participants
In total, 965 women recruited to the MAVIDOS trial remained in the study until delivery, and of these 630 had CTX measured in either early ( n = 493) or late pre gnanc y ( n = 498). A total of 372 women had a CTX measurement at both time points (placebo group, n = 188; cholecalciferol-supplemented group, n = 184), as shown in Supplemental Figure 1 . Table 1 summarizes the characteristics of the mothers with a measurement of urine CTX in early or late pre gnanc y, demonstrating a mean age of 30.4 y, 95.0% being of self-reported white ethnicity, and 44.5% being nulliparous. There was no evidence of selective dropout (as shown in Supplemental Table 1 , comparing the participants of this substudy with the o v erall MAVIDOS trial) and thus in this subgroup analysis baseline characteristics appeared comparable between the mothers randomly assigned to placebo or cholecalciferol. Baseline maternal 25(OH)D concentration (measured at recruitment, 14 weeks of gestation) was similar in both groups. At 34 weeks of gestation mean ± SD maternal 25(OH)D concentration was significantly higher in the cholecalciferol-supplemented group (66.0 ± 20.4 nmol/L) than in those who received placebo (42.5 ± 20.6 nmol/L; P < 0.0001).

Maternal urinary CTX in early and late pregnancy
As demonstrated graphically in Figure 1 , both early-and latepre gnanc y urinary CTX were in a positively skewed distribution: a few women had very high CTX measures ( > 1500 μg/mmol creatinine). In both the placebo and 1000 IU cholecalciferol/dsupplemented mothers, urinary CTX increased markedly from 14 to 34 weeks of gestation (median 14-and 34-wk CTX: placebo group, 223.8 and 445.3 μg/mmol creatinine, respectively; cholecalciferol-supplemented group, 217.5 and 420.0 μg/mmol creatinine, respecti vely; both P -dif ference < 0.0001), as shown in Table 2 . Median CTX at 34 weeks of gestation tended to be greater in the placebo group than in the cholecalciferolsupplemented group (mean difference: 25.3 μg/mmol; P = 0.06).
Restricting the analysis to women with CTX measures in early and late pre gnanc y ( n = 372) led to greater observed differences in median CTX at 34 weeks of gestation (30.4 μg/mmol greater in the placebo group than in the cholecalciferolsupplemented group, P = 0.03), as also shown in Table 2 . The conditional increase in mean ± SD [z-score (SD)] from early to late pre gnanc y was greater in the placebo group than in the cholecalciferol group (placebo: 0.16 ± 0.92; cholecalciferol: −0.16 ± 1.06; P -difference < 0.01). Note that the ne gativ e conditional change in CTX in the cholecalciferol group is relative to the o v erall distribution of standardized conditional change in CTX so still represents a rise: both groups demonstrated a large, positive absolute increase in CTX (median percentage change from early to late pre gnanc y, placebo: 111.1%; IQR: 47.1%-210.9%; cholecalciferol-supplemented: 88.8%; IQR: 25.9%-183.1%). When stratified by baseline 25(OH)D status in early pre gnanc y (sufficient, ≥50 nmol/L; or insufficient, < 50 nmol/L), classified according to the 2011 report on dietary requirements for calcium and vitamin D from the Institute of Medicine ( 26 ), a threshold effect was observed. Women with vitamin D insufficiency at baseline in the placebo group had a greater conditional increase in CTX (0.28 SD, n = 126) than their counterparts in the cholecalciferol-supplemented group ( −0.15 SD, n = 123; P -difference < 0.01). In women with vitamin D sufficiency, similar conditional increases were observed in the placebo and cholecalciferol-supplemented groups, as shown in Table 2 .   Table 3 documents the uni v ariate associations between maternal characteristics and CTX concentrations in early or late pre gnanc y and with conditional change in CTX. Greater maternal age, parity, and height were associated with lower CTX in early and late pre gnanc y. Cholecalciferol supplementation (1000 IU/d) from 14 weeks of gestation was associated with −0.18 SD lower late pre gnanc y CTX than in those who received placebo (95% CI: −0.35, −0.001 SD; P = 0.05). There was evidence of associations between lower conditional change in CTX from early to late pre gnanc y and lower BMI, greater baseline 25(OH)D, and cholecalciferol supplementation ( −0.32 SD conditional change in CTX compared with those who received placebo; 95% CI: −0.52, −0.11 SD; P = 0.002).

Associations between early-and late-pregnancy urinary CTX and maternal serum 25(OH)D
There were modest correlations between early-and latepre gnanc y CTX in both the placebo and treatment groups (placebo group, r = 0.31; treatment group, r = 0.45, both P < 0.0001); see Supplemental Table 2 . Indeed, both earlyand late-pre gnanc y 25(OH)D were also correlated, with evidence of a greater magnitude of correlation in the placebo group ( r = 0.36, P < 0.0001) than in the cholecalciferol-supplemented group ( r = 0.22, P = 0.0003). No major correlations were demonstrated between early-or late-pre gnanc y serum 25(OH)D measurements and CTX measures. In the United Kingdom, season is strongly associated with maternal 25(OH)D in pre gnanc y ( 19 ). Supplemental Table 3 presents CTX concentrations by season, demonstrating little evidence for any seasonal effect.

Maternal CTX in late pregnancy, conditional change in CTX, and maternal postpartum bone health
In the 283 mothers with a late-pre gnanc y measure of urine CTX and postpartum DXA, greater maternal urinary CTX at 34 weeks of gestation was associated with lower maternal postpartum lumbar spine and total hip bone mineral content (BMC) and areal bone mineral density (aBMD), after adjustment for maternal age, parity, smoking, and BMI in early pre gnanc y (in the groups combined). These associations appeared more robust in the cholecalciferol than in the placebo group and CTX-bone associations are summarized in Table 5 and Figure 2 (groups combined). The direction of associations with conditional change in CTX ( Table 5 ) followed a similar but nonsignificant pattern 2 SD change in urinary CTX (Fisher-Yates score) per unit exposure. 3 Adjusted for treatment group, except when associations between the vitamin D-supplemented group and CTX were being tested. 4 Conditional change in CTX (LP CTX adjusted for EP CTX), as z score. Downloaded from https://academic.oup.com/ajcn/article/114/5/1600/6326613 by guest on 17 January 2022 and were substantially attenuated compared with those for latepre gnanc y CTX. Additional adjustment for maternal height attenuated the observed relations ( Supplemental Table 4 A, B).

Conditional change in CTX, and neonatal bone indices
A total of 321 mother-neonate pairs had a measure of early-or late-pre gnanc y maternal CTX and neonatal wholebody or lumbar spine DXA. Supplemental Table 5 presents the characteristics of the neonates. No differences in neonatal anthropometry or DXA characteristics were observed between the placebo and cholecalciferol-supplemented groups. On regression analyses ( Table 6 ) greater maternal conditional change in CTX was associated with greater neonatal total-body and total spine area, BMC, and aBMD (adjusted for neonatal sex, age at DXA scan, and gestational age at delivery). The magnitude and direction of associations were similar between the placebo and cholecalciferol-supplemented groups; no interaction was observed between conditional change in CTX and randomization group.

Discussion
In this post hoc analysis of the MAVIDOS randomized controlled trial (RCT), we have demonstrated that maternal gestational cholecalciferol supplementation is associated with a smaller gestational increase in bone resorption markers than placebo is. Although maternal urinary CTX almost doubled from 14 weeks to 34 weeks of gestation in both vitamin D-supplemented and placebo groups, the conditional increase in CTX from early to late pre gnanc y was lower in the cholecalciferol-supplemented than in the placebo group, with greater differences in conditional increase in CTX between the groups observed in vitamin D-insufficient women. Furthermore, late-pre gnanc y CTX was inversely associated with postpartum measures of maternal bone from DXA.
Our observation of increasing markers of bone resorption throughout pre gnanc y is replicated in other smaller studies, in which bone resorption markers have been shown to rise markedly through gestation, whereas markers of bone formation tend to be low in early pre gnanc y, reaching normal values by term, such that there is a net resorptive state in late pre gnanc y ( 10-17 , 27 ). Studies of maternal bone mass throughout pre gnanc y and the postpartum period support this, showing that bone mineral density (BMD) declines by a small amount [between 2% and 5% at both trabecular ( 28 ) and cortical sites in the axial and peripheral skeleton ( 29 )], but importantly, pre gnanc y does not hav e a ne gativ e impact on bone health in later life ( 30 ). Both circadian (peaking in the early morning) and seasonal (peaking during winter months) variations in bone resorption markers have been observed ( 7 , 31 ). Vitamin D has been proposed to play a role in this seasonal variation. Again, in keeping with our findings, sev eral studies hav e demonstrated a ne gativ e association between 25(OH)D status and markers of bone resorption, in both nonpregnant and pregnant women (6)(7)(8)(9). Our observation that greater conditional increases in CTX were associated with greater neonatal bone mass is in keeping with a previous study from our group using the Southampton Women's Surv e y cohort. Greater maternal bone loss in pre gnanc y (as measured by calcaneal ultrasound) was shown to be associated with increased BMD in their infants ( 32 ), consistent with associations between maternal bone resorption and transplacental calcium transfer to the developing fetus.
The link, ho we ver, between maternal vitamin D and its modulation of calcium transfer to the fetus is less clear. Circulating 1,25-dihydroxycholecalciferol [1,25(OH) 2 D (calcitriol)] increases during pre gnanc y with (most importantly) the maternal kidney and also possibly the placenta providing the necessary 1-α-hydroxylase activity ( 11 , 16 , 33 ). Increased 1,25(OH) 2 D is thought to contribute to greater maternal intestinal calcium absorption ( 34 ). Several factors are known to contribute to the regulation of the balance between maternal and fetal bone formation and resorption (osteoblast and osteoclast activity), including PTH and parathyroid hormone-related peptide (PTHrp) (35)(36)(37), receptor acti v ator of NF-κB ligand (RANKL) and osteoprotegerin (OPG), Sclerostin and fibroblast growth factor-23 (FGF-23), the Wnt pathway, oestrogens, and prolactin ( 38 ). It is, therefore, unlikely that the study of 1 potential mediating factor [maternal serum 25(OH)D concentrations] and 1 measure of bone resorption (urinary CTX) is likely to provide conclusive information about the system, although it does provide an important step toward our understanding of this complex process. Indeed, the European Food Safety Authority has identified the associations between vitamin D, bone turno v er markers, and bone   health as a current knowledge gap, as outlined in their 2016 scientific opinion article ( 39 ).
To our knowledge, MAVIDOS represents the first opportunity to study vitamin D and pre gnanc y bone markers in an interventional setting. The mechanism for the observed inverse relationship between maternal cholecalciferol supplementation vs. placebo and CTX concentrations could be related to greater calcium availability from maternal intestinal absorption for the developing fetus in mothers with impro v ed 25(OH)D status, reducing the requirement for the release of calcium by maternal bone resorption. Our findings that higher maternal serum 25(OH)D concentration in late pre gnanc y is associated with lower late-pre gnanc y CTX, after adjustment for season, and that women with vitamin D insufficiency in early pregnancy had greater conditional increases in CTX throughout pre gnanc y, support this hypothesis. Studies of calcium and vitamin D in pregnant women provide supporti ve e vidence for the resulting notion that maternal 25(OH)D status in late pre gnanc y is important in regulating maternal bone turnover to ensure calcium availability for the developing fetus. One previous RCT of calcium supplementation in pre gnanc y showed that daily consumption of a 1100-mg Ca supplement led to a 15% reduction in urinary NTX during the second and third trimesters ( 40 ). A prospective cohort showed that higher calcium intake in late pre gnanc y was associated with reduced bone resorption, with greater effects in winter, when vitamin  2 D] tended to be associated with lower bone resorption markers (serum NTX and CTX, respectively) in the second and third trimesters and postpartum period, whereas no correlation between 25(OH)D and bone resorption markers was seen in the nonpregnant controls ( 6 , 8 ). In nonpregnant women, a high bone turno v er in those who are vitamin Ddeficient has been observed, with vitamin D supplementation attenuating this effect in some studies ( 41 ) but not others ( 42 , 43 ). CTX, released by osteoclastic resorption of bone, is significantly correlated with lower total hip BMD, distal radius BMD, lumbar spine BMD, and poorer bone indices assessed by peripheral quantitative computed tomography and histology (44)(45)(46). Our finding that greater urinary CTX in late pre gnanc y was associated with lower lumbar spine and hip BMC and BMD is in keeping with this observation. Interestingly, the associations were of consistent direction but attenuated with conditional change in CTX, which may reflect lower statistical power given the smaller cohort with measures at both time points or potentially that the absolute value in late pre gnanc y rather than change in concentrations is the more important factor. Notably, although the magnitude of the associations between CTX and bone measures was attenuated after adjustment for maternal height, this was a modest effect on aBMD, suggesting CTX-aBMD relations at least partly independent of linear skeletal size.

Limitations
Our study had some limitations, the first being that, as a result of ethical stipulations, participants with 25(OH)D concentrations < 25 mmol/L at screening at 12 weeks of gestation could not be included. Ho we v er, vitamin D deficienc y had dev eloped in a proportion of women by gestational week 34, particularly in the placebo group and those with winter deliveries. Our findings may also not be rele v ant to nonwhite populations, because our study did not include many women who were not of self-reported white ethnicity. In addition, we could not exclude the possibility that some participants were taking vitamin D in addition to the study drug, although reported supplement use did not differ between groups at baseline interview. Furthermore, as a result of the primary trial design, we were unable to characterize dietary calcium intake in detail, and assays of vitamin D metabolites [1,25(OH) 2 D, 24,25(OH) 2 D], PTH, and PTHrp were not available, meaning a comprehensive assessment of maternal factors determining bone turno v er could not be made. Limitations regarding the use of CTX as a marker of bone resorption should also be recognized including its circadian rhythm and relation with food intake (although early-morning, second-void urine was used to minimize this variation). We were not able to adjust fully for hemodilution in pre gnanc y [which may also affect serum 25(OH)D concentrations] ( 47 ) and individual changes in maternal glomerular filtration rate, but did use the CTX v alue relati ve to creatinine to account for renal function ( 13 ). The assay used in this study does not distinguish between the different isoforms of CTX. α-CTX in particular is predominantly released from growing fetal bone and may cross the placenta. One study estimated that ∼9% of α-CTX and 2% of β-CTX in the mother's urine in late pre gnanc y may be coming from the fetal skeleton ( 48 ); therefore, it is not possible to quantify the fetal contribution to the mother's urinary CTX. Because the participants were recruited in early pre gnanc y, prepre gnanc y DXA scans could not be performed. Therefore, we did not have the opportunity to explore changes in BMC and BMD, and their relation with CTX, from prepre gnanc y to postpartum. Maternal breastfeeding is known to play a large role in determining amounts of bone resorption and postpartum BMD, particularly at trabecular sites ( 49 ); lumbar spine BMD has been shown to decline by a mean of 5%-10% during e xclusiv e breastfeeding ( 49 ). Although mothers underwent DXA within 2 wk postpartum, the number of days spent breastfeeding before DXA assessment may be important, but these data were not available. Finally, it should be acknowledged that this is a subanalysis of an RCT, so although the differences in CTX between groups and associations with bone indices are biologically plausible and consistent with e xisting medical literature, the y should be recognized as post hoc and require replication.

Conclusion
In conclusion, we have shown that maternal urinary CTX rises from early to late pre gnanc y, with the magnitude of this rise appearing to be reduced by gestational cholecalciferol supplementation, and is inversely associated with maternal bone mass postpartum. These findings are consistent with a protective effect of gestational vitamin D supplementation on maternal bone health and inform our understanding of bone resorption in pre gnanc y and the potential relation with vitamin D metabolism. Long-term follow-up of both mothers and offspring, with repeat assessments of bone indices, will enable us to determine better the lasting impact of cholecalciferol supplementation in pre gnanc y.
The MAVIDOS Trial Group comprises NK Arden, A Carr, M Clynes, EM Dennison, MZ Mughal, and SJ Woolford.
The authors' responsibilities were as follows-EMC and NCH: designed the research; EMC, KM, RJM, and NCH: wrote the paper; FG and RE, alongside IS and AP: undertook or advised upon the laboratory analyses; CP, SD, SRC, and HMI: performed the statistical analysis; NJB, SHK, A TP , RF, SVG, KMG, and MKJ: o v ersa w the MAVIDOS trial at the Oxford, Sheffield, and Southampton sites; NCH and CC: supervised the study and the preparation of the manuscript; CC: had primary responsibility for the final content; and all authors: contributed to manuscript preparation and read and appro v ed the final manuscript. EMC reports honoraria/travel support from UCB, Eli Lilly, Pfizer, and Amgen outside the submitted work. NJB reports remuneration from Internis Pharmaceuticals Ltd. outside the submitted work. ATP reports grants from the Arthritis Research Council during the conduct of the study. HMI reports grants from Medical Research Council, Arthritis Research UK (ARUK), and the European Union's Seventh Framework Programme during the conduct of the study; and although not directly receiving funding from other bodies, members of her team have received funding from the following companies from other work: Danone, Nestec, Abbott Nutrition. KMG reports reimbursement for speaking at Nestlé Nutrition Institute conferences, and grants from Abbott Nutrition & Nestec, outside the submitted work; in addition, KMG has a patent Phenotype Prediction pending, a patent Predictive Use of CpG Methylation pending, and a patent Maternal Nutrition Composition pending, not directly related to this work. MKJ reports personal fees from Stirling Anglia, Consilient Health, and Internis outside the submitted work. RE reports grants and personal fees