Study designs, baseline and reference data

When interpreting studies of Ca economy in pregnancy and lactation it is essential to consider the study design and the limitations this may impose. Longitudinal study designs, in which serial measurements are made prospectively on the same individual, are the most informative. This is because the likely changes in bone measurements, biochemical markers, dietary intakes and Ca absorption and excretion within an individual are relatively small compared with the range of absolute values in the population. Cross-sectional studies are less able to detect such changes unless the sample size is very large. In addition, retrospective studies are more difficult to interpret because of potential confounding by factors that may not be accurately recalled such as previous weight, factors affecting vitamin D status or use of hormone contraception. In studies of lactation retrospective studies are rarely able to adequately capture the specifics of infant feeding practice, compounded by a lack of consistent definitions for the term ‘lactation’ which can cover a range of breast-feeding behaviours that differ in duration of exclusive and partial lactation, number of feeds per d, and the timing and extent of complementary feeding⁽Reference Prentice, Pettifor, Juppner and Glorieux^3, Reference Sowers, Crutchfield and Jannausch^6, Reference Prentice, Laskey, Jarjou, Bonjour and Tsang⁷⁾.

The most stringent study design is to collect data from women before they become pregnant and to follow them until after lactation has stopped and when sufficient time has elapsed for any permanent effects to become apparent. Serial measurements are required over the same period of time in a comparable group of women who are neither pregnant nor lactating and have not been so recently (non-pregnant non-lactating; NPNL), in order to account for potential variation due to increasing age, weight change and instrument performance⁽Reference Laskey and Prentice^8–Reference Naylor, Iqbal and Fledelius¹³⁾. In addition, it is important to include a comparative group of non-breast-feeding (NBF) women measured serially after parturition in order to differentiate between the effects of lactation and those of a recent pregnancy in breast-feeding (BF) mothers⁽Reference Laskey and Prentice⁸⁾. Although useful for the interpretation of longitudinal studies, it is important to recognise that data from NPNL and NBF women should not be used to make judgements about long-term benefits or detriments of reproduction on maternal Ca economy because, depending on the population being studied, such groups may include women who are less able to conceive⁽Reference Sowers¹⁴⁾ and/or those with differing socio-economic and lifestyle backgrounds.

In practice, such an ideal study is challenging and difficult to achieve. As a result, many researchers have used more limited designs. For studies of metabolic changes in pregnancy, measurements made in the first or second trimester have commonly been used to provide reference data for each individual. Such studies, although valuable, cannot provide information about any post-conception changes that occur very early in pregnancy. Similarly, data collected in the weeks after delivery cannot be used to quantify the net response to pregnancy unless the possible effects of lactation are considered for BF subjects. For studies of lactation, measurements obtained early in the postpartum period (generally within 1 or 2 weeks) have often been used as the initial reference point with serial measurements thereafter. Such studies cannot be used to draw definitive conclusions about the net effects of pregnancy and lactation, because the influence of a recent pregnancy may still be apparent at the initial ‘baseline’ value. Additional complexities arise in the design of lactation studies because of the need to make serial measurements at specified intervals. Typically, these have been set variously relative to the day of delivery (i.e. at set times after delivery), stage of lactation (for example, at peak lactation and/or cessation of lactation) or lactational amenorrhoea (for example, time to first menses).

Measures of bone mineral content, density and size

Studies of Ca economy during pregnancy and lactation require estimates of change in skeletal Ca content. Bone consists of an organic matrix strengthened by deposits of Ca and other minerals; the skeleton contains about 99 % of the total amount of Ca in the body⁽Reference Ilich and Kerstetter¹⁵⁾. The skeleton of an adult woman contains approximately 1 kg Ca⁽Reference Widdowson, Dickerson, Comar and Bronner¹⁶⁾. There are two types of bone: cortical and trabecular⁽Reference Marks, Odgren, Bilezikian, Raisz and Rodan¹⁷⁾ and their distribution ensures that a bone can withstand forces without breaking. Cortical bone is dense and compact, found mainly in the shafts of long bones and surrounding other bones, for example, vertebrae, and mostly has mechanical and protective functions. Trabecular bone has an open spongy structure, is found in the ends of long bones and throughout other bones, and is more metabolically active. There is a greater ratio of trabecular:cortical bone in the axial skeleton (spine and hip) and in distal regions of the appendicular skeleton (wrists and ankles) than in the shafts of the long bones. The response of bone to physiological and environmental stimuli can differ between regions of the skeleton. It is important, therefore, to obtain information from several skeletal sites when considering the effects of pregnancy and lactation on Ca economy and bone health.

There are several methods for the in vivo determination of human bone mineral content (BMC), density and size. The most commonly used method is dual-energy X-ray absorptiometry (DXA)⁽Reference Prentice, Davies and Cole^18, Reference Laskey¹⁹⁾. DXA has largely replaced dual-photon absorptiometry (DPA), which used radioisotopes rather than X-rays as the source of ionising radiation. DXA and DPA measure BMC, bone area (BA) and areal bone mineral density (aBMD; equivalent to BMC/BA in g/cm²). These provide information about the total amount of mineral present, the size and areal density of bone in the scanned regions, all of which contribute to bone strength⁽Reference Orwoll²⁰⁾. Most or nearly all researchers report only aBMD because this is the variable that is measured with the greatest precision and is most widely used clinically. However, interpretation is more difficult when DXA data are not reported in full and may account for some of the inconsistencies in results and conclusions between different studies. Estimates of Ca content, and the contribution of the skeleton to Ca economy, are obtained by making assumptions about the proportion of Ca present in the mineral phase of human bone, generally considered to be about 38 %⁽Reference Olausson, Laskey and Goldberg¹¹⁾. In addition, there are systematic differences in aBMD measurements between different DXA manufacturers.

The reproducibility of DXA is good; the CV of aBMD varies from 1 to 3 %, depending on scanning system and skeletal sites⁽Reference Laskey, Prentice and Hanratty²¹⁾. This allows relatively small changes in aBMD within individuals to be detected with confidence. DXA instruments have been optimised for measurements in adults. Several systems have introduced software to enable measurements to be made in infants and children. However, although there are studies that have considered the accuracy and precision of DXA for use in infants⁽Reference Koo²²⁾, there is a lack of consistency between different systems. Estimates of neonatal and infant bone accretion and comparisons of results generated with different instruments, therefore, are problematic and must be viewed with caution⁽Reference Prentice, Laskey and Goldberg^23, Reference Jarjou, Prentice and Sawo²⁴⁾.

There are several points that should be considered when interpreting DXA data. First, values derived by absorptiometry represent an integration over the whole of the organ within the skeletal envelope in the region being scanned and cannot distinguish between cortical, trabecular and non-osseous tissue (for example, within the medullary cavity). Second, aBMD represents the X-ray attenuation within a two-dimensional projection of a three-dimensional structure and is not a measure of true density. As a consequence, aBMD is dependent on bone volume, and bones with the same volumetric density but of different size can have different aBMD⁽Reference Prentice, Parsons and Cole^25–Reference Prentice, Schoenmakers and Laskey²⁷⁾. This can be addressed to some extent by adjusting for bone and body size, for example, using a regression method⁽Reference Prentice, Parsons and Cole²⁵⁾ or algorithms⁽Reference Adams and Shaw^26, Reference Molgaard, Thomsen and Prentice^28, Reference Fewtrell²⁹⁾. Because of this volume effect, the interpretation of aBMD in longitudinal studies can be complicated, especially when there are changes in bone size, for example, during growth or, potentially, in pregnancy or lactation. In such studies, BA can be used to adjust BMC for bone size (BA-adjusted BMC) and the influence of body weight and weight change considered separately⁽Reference Olausson, Laskey and Goldberg¹¹⁾. Third, the skeleton is responsive to changes in body weight (the greater the weight gain the greater the BMC and aBMD)⁽Reference Olausson, Laskey and Goldberg¹¹⁾. Finally, because of technicalities associated with bone-edge detection, BA, and hence BMC and aBMD, is influenced by the amount of mineral present and the depth of overlying soft tissue and may change slightly when the mineral content or the tissue depth changes (the greater the mineral present or overlying tissue, the greater the BA)⁽Reference Prentice, Schoenmakers and Laskey^27, Reference Laskey, Murgatroyd and Prentice³⁰⁾. Bone edge detection effects are therefore likely to accompany bone mineral mobilisation/replenishment and changes in body weight and, although generally small, need to be considered when interpreting longitudinal data.

Another X-ray technique for measuring bone mineral and size is quantitative computed tomography (QCT). In contrast to DXA and DPA, QCT measures volumetric bone mineral density (vBMD; g/cm³) and distinguishes between regions of cortical and trabecular bone. In addition, this method measures cross-sectional BA and can provide information about bone shape and the biomechanical properties of the skeleton. Peripheral QCT (pQCT) is designed specifically to measure appendicular skeletal sites, such as the bones of the forearm (radius and ulna) and the leg (tibia, fibula and femur). The reproducibility of vBMD measurements by pQCT is about 2–5 %⁽Reference Veitch, Findlay and Ingle³¹⁾.

Quantitative ultrasound (QUS) is a third technique used for studying bone. Two QUS variables are generally reported: broadband ultrasound attenuation (BUA), which is a measure of the energy lost when ultrasound passes through bone mineral and soft tissue, and velocity or speed of sound, through bone. Although these variables are regarded as proxy measures for bone density, the validity of this assumption is uncertain, especially during pregnancy in the presence of peripheral oedema⁽Reference Johansen and Stone³²⁾. In addition, the reproducibility of QUS is relatively poor⁽Reference Laskey and Prentice^33–Reference Njeh, Hans and Li³⁵⁾ which limits its use in longitudinal studies.

DXA, DPA, QCT and pQCT are based on ionising radiation. The radiation dose received during a set of DXA, DPA or pQCT scans is, in general, similar to the daily exposure to background radiation⁽Reference Adams and Shaw^26, Reference Blake, Naeem and Boutros^36, Reference Njeh, Fuerst and Hans³⁷⁾, while that received from QCT is higher⁽Reference Kalender³⁸⁾. Although the dose of radiation is low, whole-body and axial skeletal sites of pregnant women generally are not scanned using DXA, DPA or QCT for research purposes, in order to minimise unnecessary exposure of the fetus and because the results cannot distinguish between maternal and fetal tissues. Peripheral X-ray techniques in which the fetus is not exposed to additional radiation, such as forearm absorptiometry⁽Reference Black, Topping and Durham^39–Reference More, Bettembuk and Bhattoa⁴²⁾ and pQCT⁽Reference Wisser, Florio and Neff⁴³⁾, are used for studies in pregnant women.

Measures of bone turnover, mineral metabolism and excretion

Supporting information on the contribution of bone metabolism to Ca economy during pregnancy and lactation can be obtained through studies of bone turnover markers. In addition, indices of mineral metabolism, and calciotropic and other hormones, are useful for identifying underlying mechanisms.

Bone undergoes continuous turnover through the actions of bone-resorbing osteoclasts and bone-forming osteoblasts⁽⁴⁴⁾. Within a single bone-remodelling unit, osteoclasts erode an area of the mineralised surface to produce a resorption cavity. Over a period of time, this is refilled by bone matrix secreted by osteoblasts, which is subsequently mineralised. In the young adult, the process of bone resorption and formation is usually tightly coupled. This results in overall maintenance of the skeleton with little net change in mineral content⁽Reference Prentice⁴⁵⁾. Bone mineral accretion occurs when bone formation exceeds resorption, for example, during growth. Bone mineral loss occurs when resorption exceeds formation, for example, during age-related bone loss.

Markers of bone resorption include collagen breakdown products such as crosslinks (for example, deoxypyridinoline), hydroxyproline and segments of the N-telopeptide (NTx) or C-telopeptide (CTx). Deoxypyridinoline and hydroxyproline are measured in urine; NTx and CTx may be measured in either urine or blood. Markers of bone formation measured in blood include products of osteoblastic synthesis of new bone matrix, such as N- and C-propeptides of type I collagen (P1NP and P1CP, respectively), and proteins involved in osteoblast function such as osteocalcin and bone-specific alkaline phosphatase (ALP). In studies of pregnancy, the use of assays that are specific for bone-specific ALP is essential because in addition to total ALP derived from extra-skeletal sources, the placenta produces an isoenzyme of ALP which is excreted into the circulation.

There are several well-established laboratory techniques for the analysis of markers of mineral metabolism and calciotropic hormones, including sensitive and specific immunoassays and HPLC. These topics are covered more fully elsewhere⁽Reference Bilezikian, Raisz and Martin⁴⁶⁾. Technical variations often mean there are difficulties in drawing comparisons between results generated with different assay methods or in different laboratories because of a lack of methodological standardisation for many of the indices relating to mineral and bone metabolism. In addition, there is biological variation in the measured concentration of many of these factors due to circadian rhythms, the pattern of breast-feeding, and the effects of periodic exogenous influences such as recent food intake. For example, the plasma concentration and urinary output of CTx is higher at night than in the afternoon⁽Reference Hannon and Eastell⁴⁷⁾, plasma prolactin concentration is raised after a breast-feed⁽Reference Riordan and Auerbach⁴⁸⁾ and urinary hydroxyproline excretion is increased temporarily after ingestion of animal protein⁽Reference Gasser, Celada and Courvoisier⁴⁹⁾. Interpretation of urinary markers is further complicated by the variety of urine collection methods that are used, such as a random spot sample, the first void of the day, or a timed collection over a set period, most commonly 2 or 24 h. The choice of collection method should be dictated by the specific question being addressed. For example, studies of urinary Ca output require 24 h collections with no restrictions on eating habits, whereas studies of renal phosphate reabsorption require a 2 h collection under fasting conditions. Thus the interpretation of biochemical data may depend on the time and conditions when samples were collected. Ideally, samples should be collected in a standardised way with respect to time of day, recent food intake, and time elapsed since the last meal and/or breast-feed.

A further complication is that the concentrations of blood-borne analytes in pregnancy are affected by the increase in plasma volume and resulting haemodilution. Albumin concentration has been used as a proxy measure to derive a correction factor for haemodilution⁽Reference Kaur, Godber and Lawson^50, Reference Olausson, Laskey and Smith⁵¹⁾ but such corrections are not universally applied. Pregnancy is also accompanied by an increase in glomerular filtration rate, which can affect the interpretation of urinary measures. Other factors that need to be considered when interpreting biochemical data during pregnancy and lactation are most notably the need to distinguish between the contribution of the fetus, placenta and mammary gland to bone markers and hormones in the maternal circulation⁽Reference Kovacs and Kronenberg^4, Reference Naylor, Iqbal and Fledelius^13, Reference Kaur, Godber and Lawson⁵⁰⁾. Such issues are rarely discussed in published reports, and studies need to be reviewed with this in mind.

Calcium intake, absorption and balance

Measurements of Ca intake, absorption and balance are important when considering Ca economy in pregnancy and lactation, but can be challenging for subjects and investigators, and difficult to interpret⁽Reference Weaver, Weaver and Heaney⁵²⁾.

Many techniques are used to assess dietary Ca intake⁽Reference Boushey, Weaver and Heaney⁵³⁾. All dietary assessment methods, such as weighed records, diaries, FFQ, diet histories and 24 h recalls have their advantages and disadvantages for use in different subject and population groups. No method is appropriate for all situations and the choice depends on what aspect of the diet is under scrutiny: for example, the description of habitual diet, monitoring dietary change, quantifying nutrient intakes, identifying food sources or characterising eating patterns and food groups⁽Reference Rutishauser, Black, Gibney, Vorster and Kok^54, Reference Goldberg, New and Bonjour⁵⁵⁾. There are relatively few foods that are rich sources of Ca and some are consumed infrequently. A combination of dietary assessment techniques may be needed to provide a more detailed indication of customary Ca intake, especially in populations where milk and milk products are not major components of the diet⁽Reference Prentice, Laskey and Shaw⁵⁶⁾. In addition, drinking water and some condiments, flavourings, medicines and over-the-counter preparations and supplements may contain substantial amounts of Ca. Although assessments of Ca intake should include the contribution from these sources, it is uncommon for such information to be routinely collected in dietary studies and surveys, or for the necessary compositional information to be included in food databases. It is important therefore to appreciate fully the dietary assessment methods used when comparing Ca intakes of pregnant and lactating women in different studies.

The absorption of Ca from foods depends on many factors, both endogenous, such as the efficiency of absorption in the intestine and the production of gastric acid, and exogenous, such as vitamin D status (25-hydroxyvitamin D (25OHD) via intake or sun exposure) and the consumption of dietary components that enhance or inhibit Ca absorption⁽Reference Heaney⁵⁷⁾. The retention of dietary Ca also depends on the extent to which Ca is excreted through faeces, urine and sweat. The traditional Ca balance study, which measures the difference between intake and output, requires collection of all food consumed and all excretory products over a period of several weeks. In practice, Ca excretion in sweat is rarely quantified but an estimate applied. The use of stable isotopes of Ca (⁴⁸Ca, ⁴⁴Ca, ⁴²Ca) allows for the direct quantification of intestinal Ca absorption efficiency over 1–4 d without the need for faecal collections⁽Reference DeGrazia, Ivanovich and Fellows⁵⁸⁾. Ca absorption efficiency can also be determined by more indirect methods, such as quantifying the effect of an oral Ca load on plasma Ca concentration and urinary Ca excretion⁽Reference Weaver, Rothwell and Wood⁵⁹⁾. Such methods require advanced data-modelling and assumptions about variables that are difficult to measure⁽Reference O'Brien, Donangelo and Zapata⁶⁰⁾. There is a lack of information about the validity of these models, estimates and assumptions when applied to pregnancy and lactation.

Maternal bone mineral mobilisation: bone mineral studies

Bone mineral studies have provided evidence that bone mineral mobilisation occurs during human pregnancy and lactation with replenishment of skeletal mineral in the later stages of lactation and after breast-feeding has stopped (Fig. 1). Tables 1–5 and Figs. 2–4 summarise the published results from longitudinal studies in Caucasian women, with Ca intakes close to recommendations, that have investigated whole-body and regional changes in bone mineral measured as change in aBMD or BA-adjusted BMC using DPA or DXA. It was not possible to include change in BMC without correction for BA because few authors provide this information, but where such data are available they have been incorporated in the text. Similarly, the results of the few studies that have used pQCT, QCT or ultrasound are not included in the tables but described in the text. In addition, it has not been possible to provide estimates of the variation between individuals in these studies because insufficient data were provided in most cases. It should also be noted that relatively few studies included comparisons with contemporaneous NPNL controls or, in studies of lactation, comparisons between BF and NBF mothers.

Table 1 Mean changes (%) in bone mineral at different sites, measured with dual-energy X-ray absorptiometry (DXA) or dual-photon absorptiometry (DPA), between pre-pregnancy (PRE) and up to 6 weeks postpartum (POST)†

NPNL, non-pregnant, non-lactating; NBF, non-breast-feeding; BA, bone area; BMC, bone mineral content; RS, radius shaft; RW, radius wrist; aBMD, areal bone mineral density; QCT, quantitative computed tomography. Statistically significant: * P ≤ 0·05, ** P ≤ 0·01, *** P ≤ 0·001, NS, non-significant, as indicated in original papers.

†All data are for measurements without correction for changes in body weight.

‡ Bone mineral data taken from paper.

§ Bone mineral data derived from tables or figures in paper.

‖ Implausible values? (see text).

¶ Data adjusted for weight.

†† In women of pre-pregnant BMI <  19·8 kg/m².

‡‡ In women of pre-pregnant BMI 19·8–26·0 kg/m².

§§ In women of pre-pregnant BMI > 26·0 kg/m².

‖ ‖ Note: value of − 3·5 % cited in abstract is incorrect.

¶¶ Derived from the original paper by authors of the present review from whole-body scan divided into subregions.

††† Given the small sample size, the original authors concluded that there is a tendency for a decrease in bone mineral status at the spine, but not at the femoral neck or radial shaft.

Table 2 Mean changes (%) in bone mineral during 3–6 months lactation at different sites, measured with dual-energy X-ray absorptiometry (DXA) or dual-photon absorptiometry (DPA)†

NPNL, non-pregnant, non-lactating; NBF, non-breast-feeding; aBMD, areal bone mineral density; RW, radius wrist; BMC, bone mineral content; BA, bone area; RS, radius shaft; QCT, quantitative computed tomography. Statistically significant: * P ≤ 0·05, ** P ≤ 0·01, *** P ≤ 0·001, NS, non-significant, as indicated in original papers.

† Data are for measurements without correction for changes in body weight, unless specified.

‡ Bone mineral data taken from paper.

§ Bone mineral data derived from tables or figures in paper.

‖ Derived by authors from whole-body scan divided into subregions.

¶ Data adjusted for weight.

Table 3 Mean net changes (%) in bone mineral at 12 months postpartum at different sites measured with dual-energy X-ray absorptiometry (DXA) or dual-photon absorptiometry (DPA)†

NPNL, non-pregnant, non-lactating; NBF, non-breast-feeding; aBMD, areal bone mineral density; RW, radius wrist; BA, bone area; BMC, bone mineral content; RS, radius shaft. Statistically significant: *P ≤ 0·05, **P ≤ 0·01, ***P ≤ 0·001, NS, non-significant, as indicated in original papers.

† Data are for measurements without correction for changes in body weight.

‡ Bone mineral data taken from paper.

§ Bone mineral data derived from tables or figures in paper.

‖ Bromocryptine after 6 months.

Table 4 Mean net changes (%) in bone mineral at different sites, measured with dual-energy X-ray absorptiometry (DXA), between early lactation and post-lactation†

NPNL, non-pregnant, non-lactating; NBF, non-breast-feeding; aBMD, areal bone mineral density; BA, bone area; BMC, bone mineral content; RW, radius wrist; RS, radius shaft. Statistically significant: ** P ≤ 0·01, *** P ≤ 0·001, NS, non-significant, as indicated in original papers.

† All data are for measurements without correction for changes in body weight, unless specified.

‡ Bone mineral data taken from paper.

§ Data adjusted for weight.

Table 5 Mean net changes (%) in bone mineral at different sites measured with dual-energy X-ray absorptiometry (DXA) after resumption of menses†

NPNL, non-pregnant, non-lactating; NBF, non-breast-feeding; aBMD, areal bone mineral density; RS, radius shaft. Statistically significant: * P ≤  0·05, ** P ≤  0·01, *** P ≤  0·001, NS, non-significant, as indicated in original papers.

† All data are for measurements without correction for changes in body weight.

‡ Bone mineral data taken from paper.

§ Bone mineral data derived from tables or figures in paper.

Fig. 2 Mean percentage change in bone area-adjusted bone mineral content (BA-adjusted BMC) during pregnancy (pre-pregnancy to 2 weeks postpartum; ▒; n 34) and in non-pregnant, non-lactating (NPNL) controls (□; n 84). Values are means, with standard errors represented by vertical bars. WB, whole body; LS, lumbar spine; TH, total hip; FN, femoral neck. Data taken from Olausson et al. ⁽Reference Olausson, Laskey and Goldberg¹¹⁾.

Fig. 3 Percentage changes in bone area-adjusted bone mineral content from baseline (2 weeks postpartum) to 3, 6 and 12 months postpartum and 3 months post-lactation (PL) for women lactating > 9 months (n 20). (●), Whole body; (■), lumbar spine; (▲), femoral neck; (▾), trochanter. Values are means, with standard errors represented by vertical bars. Modified from data published by Laskey & Prentice⁽Reference Laskey and Prentice⁸⁾.

Fig. 4 Percentage changes in bone area-adjusted bone mineral content at the spine from baseline (2 weeks postpartum) to 3, 6 and 12 months postpartum and post-lactation (PL) (12 months postpartum or 3 months post-lactation for mothers who breast-fed for more than 9 months). Subjects are grouped according to length of lactation: <  3 months (●; n 12); 3–6 months (■; n 13); 6–9 months (▲; n 14); > 9 months (▾; n 20); formula feeders (non-breast-feeding; ♦; n 11). A group of twenty-two non-pregnant non-lactating controls (○) was studied in parallel. Values are means, with standard errors represented by vertical bars. Modified from data published by Laskey & Prentice⁽Reference Laskey and Prentice⁸⁾.

The mean changes in bone mineral in Tables 1–5 are presented for the different skeletal sites as given in, or derived from, the original papers, before any adjustment for the weight changes associated with pregnancy or lactation, unless stated. When used to study change over time within an individual, the value unadjusted for weight change provides a measure of net or actual change in mineral content, and therefore of Ca mobilisation from or accretion into the skeleton, provided that there is no accompanying change in bone size. Adjustment for weight change allows the effects of pregnancy and lactation on the skeleton to be considered independently of weight effects⁽Reference Olausson, Laskey and Goldberg¹¹⁾.

Maternal bone mineral mobilisation: bone turnover studies

Supporting evidence for bone mineral mobilisation during human pregnancy and lactation with later replenishment of bone mineral (Fig. 1) comes from biochemical and stable-isotope studies of bone turnover.

Bone mineral mobilisation: osteoporosis and fractures

Further evidence for bone mineral mobilisation comes from rare cases of osteoporotic fragility fractures, often vertebral, that occur during late pregnancy and in lactation. The aetiology is unknown, although in one study nine of eleven subjects had at least one of the traditional risk factors for osteoporosis, including low body weight, family history of fragility fractures or osteoporosis, low vitamin D status or smoking. Data from this study suggested that women with a low aBMD before pregnancy were at increased risk of fracture in late pregnancy or postpartum⁽Reference O'Sullivan, Grey and Singh¹¹⁹⁾. However, it has also been reported that fragility fractures in pregnancy and lactation can occur in the absence of low aBMD⁽Reference Rousiere, Kahan and Job-Deslandre¹²⁰⁾. There is little evidence that osteoporosis of pregnancy and lactation is related to diet⁽Reference Gruber, Gutteridge and Baylink^121, Reference Smith and Phillips¹²²⁾.

No prospective studies have investigated if there is an increased risk of osteoporosis in later life that can be attributed to pregnancy or lactation. Findings from retrospective studies investigating relationships between parity, lactation history and bone mineral measurements in pre- and postmenopausal women are inconsistent. Studies report positive associations between parity or lactation history and greater bone mineral⁽Reference Specker and Binkley^77, Reference Aloia, Vaswani and Yeh^123–Reference Schnatz, Barker and Marakovits¹²⁷⁾, an inverse association⁽Reference Lissner, Bengtsson and Hansson^128, Reference Wardlaw and Pike¹²⁹⁾ or no significant association⁽Reference Specker and Binkley^77, Reference Henderson, Sowers and Kutzko^130, Reference Paton, Alexander and Nowson¹³¹⁾. Secondary analysis of survey data from the third National Health and Nutrition Examination Survey (NHANES III) of 819 women aged 20–25 years indicated that those who had been pregnant as adolescents had the same BMD as women pregnant as adults and as nulliparous women. Those who had breast-fed as adolescents had higher BMD than those who had not breast-fed⁽Reference Chantry, Auinger and Byrd¹³²⁾.

Regarding relationships between parity or lactation history and hip fracture incidence in later life, however, most studies suggest either no association or a protective effect. Studies have reported no association with parity⁽Reference Alderman, Weiss and Daling¹³³⁾, reduced hip fracture incidence with increasing parity⁽Reference Hoffman, Grisso and Kelsey^134–Reference Michaelsson, Baron and Farahmand¹³⁶⁾, and an association between longer duration of lactation and lower risk of hip fracture⁽Reference Alderman, Weiss and Daling^133, Reference Kreiger, Kelsey and Holford^137–Reference Cumming and Klineberg¹³⁹⁾. There are very few data in non-Caucasian populations in developing countries, but retrospective studies have found no associations between aBMD and parity or lactation history in Bangladeshi or Sri Lankan women⁽Reference Chowdhury, Sarkar and Roy^140, Reference Lenora, Lekamwasam and Karlsson¹⁴¹⁾, and no differences in bone dimensions between South African Bantu women who had had two or fewer children compared with seven or more⁽Reference Walker, Richardson and Walker¹⁴²⁾. One study found a greater aBMD and reduced prevalence of osteoporotic fracture in multiparous compared with nulliparous postmenopausal Colombian women⁽Reference Cure-Cure, Cure-Ramirez and Teran¹⁴³⁾.

Intestinal absorption and renal excretion of calcium

Studies of Ca absorption efficiency and renal Ca excretion have demonstrated that physiological contributions to maternal Ca economy are made by increased absorption in pregnancy, decreased excretion in lactation and both increased absorption and decreased excretion post-lactation (Fig. 1).

Fetal calcium accretion and breast-milk calcium secretion

There is wide variation in fetal Ca accretion and in breast-milk Ca secretion, the other components of maternal Ca economy. Relatively little is known about the Ca content of the fetal skeleton other than that derived directly from studies of stillborn fetuses⁽Reference Ziegler, O'Donnell and Nelson^155–Reference Apte and Iyengar¹⁵⁷) and indirectly from maternal Ca balance studies⁽Reference Heaney and Skillman¹⁰³⁾ and measures of skeletal size in studies of fetal growth and development among neonates of different gestational age, with assumptions made about bone composition⁽Reference Fomon and Nelson¹⁵⁶⁾. Studies of neonatal and infant bone mineral using single-photon absorptiometry (SPA) and, more recently, DXA have added to the literature⁽Reference Jarjou, Prentice and Sawo^24, Reference Koo, Bush and Walters^158–Reference Harvey, Javaid and Poole¹⁶²⁾ but many assumptions have to be made (see Introduction) and there are ongoing difficulties with the technology and associated software that can lead to problems with interpretation⁽Reference Sawyer, Bachrach and Fung¹⁶³⁾. Evaluation of methodologies against a neonatal pig model have improved confidence in the DXA technique for assessing total Ca and mineral content of small babies⁽Reference Koo, Hammami and Hockman^164, Reference Koo, Massom and Walters¹⁶⁵⁾ but these have not been conducted for all instruments. Nevertheless, variations in fetal bone accretion at different stages of gestation, between individuals and between different pregnancies in the same mother, need to be considered in studies of maternal Ca economy in pregnancy.

After the colostral phase, breast-milk Ca concentration is relatively constant during the first 3 months of lactation, averaging about 200–300 mg/l (5·0–7·5 mmol/l) depending on the population⁽Reference Prentice, Laskey, Jarjou, Bonjour and Tsang⁷⁾, but declines progressively thereafter⁽Reference Prentice, Laskey, Jarjou, Bonjour and Tsang^7, Reference Vaughan, Weber and Kemberling^166, Reference Laskey, Prentice and Shaw¹⁶⁷⁾. The concentration of Ca in breast milk is independent of the volume of milk produced⁽Reference Prentice, Laskey, Jarjou, Bonjour and Tsang^7, Reference Jarjou, Goldberg and Coward¹⁶⁸⁾, and variation in both results in wide differences in breast-milk Ca secretion between individual mothers and between populations at the same time postpartum⁽Reference Prentice, Laskey, Jarjou, Bonjour and Tsang^7, Reference Jarjou, Goldberg and Coward¹⁶⁸⁾. The reasons for these differences are not known, although genetic effects may play a role; for example, polymorphisms in the PTH/PTH-related protein (PTHrP) receptor 1 gene have been associated with differences in breast-milk Ca concentration⁽Reference Jones, Laskey, Rushworth, Goldberg, Prentice, Prentice, Filteau and Simondon¹⁶⁹⁾. PTHrP may be one determinant of breast-milk Ca concentration, because associations have been shown with the concentration of PTHrP in breast milk⁽Reference Uemura, Yasui and Yoneda^170, Reference Seki, Kato and Sekiya¹⁷¹⁾ and in plasma⁽Reference DeSantiago, Alonso and Halhali¹⁷²⁾. However, Ca is associated with the casein, phosphate and citrate fractions of human milk and it is probable that the major determinants of breast-milk Ca concentration are those that regulate the concentration of these components⁽Reference Kent, Arthur and Mitoulas¹⁷³⁾. Studies investigating the possible influence of maternal Ca intake and vitamin D status are described later.

Lactation and postpartum

After delivery, total plasma Ca concentration returns towards a value similar to that before pregnancy⁽Reference Black, Topping and Durham^39, Reference Ritchie, Fung and Halloran^67, Reference Kent, Price and Gutteridge^115, Reference Kovacs¹⁵⁰⁾, possibly in parallel with the return of plasma volume to pre-pregnancy levels. BF women tend to have higher total and ionised plasma Ca concentrations than before pregnancy, during pregnancy or in NPNL women⁽Reference Prentice, Pettifor, Juppner and Glorieux³⁾ but similar to those observed in NBF mothers at the same stage postpartum⁽Reference Prentice, Pettifor, Juppner and Glorieux^3, Reference Kalkwarf, Specker and Ho⁹⁵⁾.

The plasma concentration of PTH during the first few months postpartum is similar to⁽Reference Ritchie, Fung and Halloran^67, Reference Krebs, Reidinger and Robertson⁸⁴⁾ or slightly decreased⁽Reference Chan, Nelson and Leung^91, Reference Prentice, Jarjou and Stirling^113, Reference Specker, Tsang and Ho^190, Reference Sowers, Zhang and Hollis¹⁹¹⁾ compared with before pregnancy or shortly after delivery. The plasma concentration of 1,25(OH)₂D is also either unchanged⁽Reference Ritchie, Fung and Halloran^67, Reference Uemura, Yasui and Kiyokawa¹⁰⁷⁾ or slightly decreased⁽Reference Prentice, Jarjou and Stirling^113, Reference Sowers, Zhang and Hollis¹⁹¹⁾ compared with pre-pregnancy or those of NPNL women. Increases in 1,25(OH)₂D concentration during the first months postpartum in both BF and NBF mothers have been reported⁽Reference Krebs, Reidinger and Robertson⁸⁴⁾. In general, BF women tend to have lower plasma PTH concentrations but higher 1,25(OH)₂D concentrations than NBF women at the same time postpartum⁽Reference Krebs, Reidinger and Robertson^84, Reference Kalkwarf, Specker and Ho^95, Reference Kalkwarf, Specker and Heubi^148, Reference Sowers, Zhang and Hollis^191, Reference Hillman, Sateesha and Haussler¹⁹²⁾. However, BF mothers nursing twins have elevated plasma concentrations of both PTH and 1,25(OH)₂D compared with those nursing single infants⁽Reference Greer, Lane and Ho¹⁹³⁾. Elevated PTH and 1,25(OH)₂D have been reported in BF women relative to early lactation and to NPNL women during the later stages of lactation and after breast-feeding stops⁽Reference Kalkwarf, Specker and Bianchi^83, Reference Kalkwarf, Specker and Ho^95, Reference More, Bhattoa and Bettembuk^109, Reference Prentice, Jarjou and Stirling^113, Reference Kent, Price and Gutteridge^115, Reference Cross, Hillman and Allen^144, Reference Specker, Tsang and Ho^190, Reference Sowers, Zhang and Hollis¹⁹¹⁾, although the pattern is not consistent. The increases in PTH and 1,25(OH)₂D may play a role in the replenishment of bone mineral post-lactation through their effects on intestinal absorption and renal retention of Ca.

The plasma concentration of calcitonin decreases during the first months postpartum in both BF and NBF women compared with shortly after delivery⁽Reference Krebs, Reidinger and Robertson⁸⁴⁾. The concentration in BF women has been reported to be higher than in NPNL women in some studies⁽Reference Prentice, Jarjou and Stirling^113, Reference Dahlman, Sjoberg and Bucht¹⁹⁴⁾ but not others⁽Reference Ritchie, Fung and Halloran^67, Reference Krebs, Reidinger and Robertson^84, Reference Greer, Tsang and Searcy¹⁹⁵⁾ and to be raised in mothers nursing twins⁽Reference Greer, Tsang and Searcy¹⁹⁵⁾. No changes have been observed in later lactation⁽Reference Ritchie, Fung and Halloran^67, Reference Krebs, Reidinger and Robertson⁸⁴⁾.

In general, the early postpartum changes in PTH, 1,25(OH)₂D and calcitonin do not correlate with breast-milk Ca content or with changes in maternal bone mineral and bone turnover markers⁽Reference Ritchie, Fung and Halloran^67, Reference Krebs, Reidinger and Robertson^84, Reference Sowers, Zhang and Hollis¹⁹¹⁾. This suggests that the Ca homeorrhesis of lactation is not driven by the three classical calciotropic hormones, and that Ca loss into breast milk drives the hormonal response and other factors that direct the Ca flux out of and into bone in response to lactation. However, Ca supplementation is associated with the expected lowering effects on PTH and 1,25(OH)₂D⁽Reference Kalkwarf, Specker and Ho⁹⁵⁾ indicating that Ca homeostatic mechanisms are intact and capable of regulating plasma Ca concentrations⁽Reference Prentice, Pettifor, Juppner and Glorieux^3, Reference Kalkwarf, Specker and Ho⁹⁵⁾.

An indicator of PTH activity is nephrogenous cyclic AMP production. No differences have been found between women who have recently ceased breast-feeding and either non-lactating control women or those who have not recently been pregnant⁽Reference Kalkwarf, Specker and Ho^95, Reference Kent, Price and Gutteridge^115, Reference Kent, Price and Gutteridge¹⁴⁷⁾. It is considered probable that a key regulator of Ca and bone metabolism during the first weeks of lactation is PTHrP. This is produced by the lactating mammary gland, possibly under the influence of prolactin, and is released into the maternal bloodstream and into breast milk⁽Reference Lippuner, Zehnder and Casez^151, Reference Sowers, Hollis and Shapiro¹⁹⁶⁾. The plasma concentration is high after delivery and declines over time, possibly in association with the decrease in prolactin concentration and the return of menstruation⁽Reference Dobnig, Kainer and Stepan^90, Reference Sowers, Hollis and Shapiro¹⁹⁶⁾. PTHrP is elevated in BF women compared with NBF women⁽Reference Sowers, Hollis and Shapiro^196, Reference Grill, Hillary and Ho¹⁹⁷⁾ and in those who have weaned their infants⁽Reference Lippuner, Zehnder and Casez¹⁵¹⁾ in the first weeks after delivery. However, it is virtually undetectable at 6 months postpartum even in women who continue to breast-feed⁽Reference Sowers, Zhang and Hollis¹⁹¹⁾. Higher concentrations of PTHrP have been shown to correlate with greater reductions in maternal aBMD at the lumbar spine and femoral neck postpartum⁽Reference Sowers, Hollis and Shapiro¹⁹⁶⁾ but not in established lactation⁽Reference Dobnig, Kainer and Stepan⁹⁰⁾. The key role for PTHrP in the first months of lactation is further supported by a clinical case report of a woman with PTH deficiency whose requirement for Ca and 1,25(OH)₂D therapy decreased during breast-feeding, a circumstance that was attributed to elevated PTHrP concentrations⁽Reference Mather, Chik and Corenblum¹⁹⁸⁾. However, the biology of PTHrP is complex and the evidence for its role in human lactation is inconsistent and needs further investigation.

Other hormonal changes of lactation may be involved in regulating Ca and bone metabolism in BF women. For example, lactation is associated with increased prolactin concentrations, which suppress the hypothalamic–pituitary–ovarian axis resulting in low oestrogen concentrations and amenorrhoea⁽Reference Howie, McNeilly and Houston¹⁹⁹⁾. Both prolactin and oestrogen have recognised direct effects on Ca and bone metabolism and may be involved in Ca homeorrhesis. It is notable that, in respect to low oestrogen concentrations, lactation has parallels with the postmenopausal period, also a time when mineral is mobilised from the skeleton. The changes in bone mineral during the first few months of lactation, therefore, may be related, at least in part, to low oestrogen concentrations⁽Reference Kolthoff, Eiken and Kristensen^41, Reference Kalkwarf, Specker and Bianchi^83, Reference Affinito, Tommaselli and di Carlo^85, Reference Kalkwarf and Specker⁸⁶⁾. However, NPNL women of reproductive age who are oestrogen deficient as a result of gonadotrophin-releasing hormone (GnRH) agonist therapy have higher Ca excretion, suppressed PTH and 1,25(OH)₂D, i.e. a pattern that does not resemble the metabolic response to lactation⁽Reference Kovacs and Kronenberg^4, Reference Kovacs¹⁸⁵⁾.

Influence on the mother in pregnancy

For women in the UK and USA with a Ca intake close to recommendations, changes in maternal bone mineral using DXA during pregnancy appear to be independent of dietary Ca intake, as shown by observational studies⁽Reference Sowers, Crutchfield and Jannausch^6, Reference Olausson, Laskey and Goldberg¹¹⁾. In contrast, observational studies among populations where Ca intakes are low suggest that the skeletal response may be dependent on maternal Ca intake. A detailed longitudinal study of bone Ca turnover during pregnancy and lactation in Brazilian women (mean Ca intake 463 mg/d) found significant positive associations with dietary Ca intake in early and late pregnancy and in early lactation; a higher Ca intake was associated with improved Ca balance⁽Reference O'Brien, Donangelo and Zapata⁶⁰⁾. A study in Mexico reported smaller increments in bone turnover markers from the second to third trimester and lower NTx in pregnant women with a higher dietary intake (average intake about 500 mg/d). Two studies using ultrasound have reported that pregnant women consuming less than 1000 mg Ca/d in Spain⁽Reference Aguado, Revilla and Hernandez⁷⁶⁾ or less than 568 ml (1 pint) of milk/d in the UK⁽Reference Javaid, Crozier and Harvey⁶⁹⁾ had a greater decrease in calcaneal bone ultrasound measures during pregnancy than those women with higher intakes, although in the UK study neither an overall correlation with milk intake nor a relationship with Ca supplement use was observed.

There have been few Ca supplementation studies of pregnant women that have investigated directly the effect of maternal Ca intake in pregnancy on bone mineral and the results are inconsistent. A randomised, double-blind, placebo-controlled Ca supplementation study (1500 mg Ca/d as calcium carbonate) of Gambian women (mean intake about 350 mg Ca/d) from 20 weeks of pregnancy to parturition demonstrated, contrary to expectations, lower BA-adjusted BMC of the hip measured at 2 weeks postpartum in the Ca-supplemented group⁽Reference Jarjou, Laskey and Sawo²⁰⁶⁾. This, combined with more accentuated lactational bone changes that were observed at the lumbar spine, distal radius and whole body, and the accompanying biochemical effects, suggest that the Ca supplement in pregnancy had disrupted the processes of adaptation to the habitually low Ca intake of these women. A non-blinded randomised supplementation study among thirty-six Chinese women (mean baseline dietary intake 480 mg Ca/d) allocated to remain on their habitual diet (group I) or supplemented with milk powder (containing 350 mg Ca/d; group II) or both milk powder and 600 mg Ca/d as calcium carbonate (950 mg Ca/d in total; group III) from 18 weeks pregnancy to 6 weeks postpartum reported a higher aBMD at 45d postpartum of the whole body and spine in group III v. group I, and of the spine in group II v. group I. There was no difference between the groups at the hip⁽Reference Liu, Qiu and Chen²⁰⁷⁾. Some of the differences between the Gambian and Chinese studies may relate to the fact that the outcome measures were obtained in the Gambian study several weeks after supplementation was stopped, whereas the data in the Chinese study were collected at the end of the supplementation period. The reported increase may, therefore, have reflected a bone remodelling transient and the effect may have been temporary. Furthermore, there is no indication of whether the Chinese women were, or had been, lactating. A study of pregnant Indian women⁽Reference Raman, Rajalakshmi and Krishnamachari²⁰⁸⁾, from an area with a habitual Ca intake of about 300 mg/d⁽Reference Krishnamachari and Iyengar²⁰⁹⁾, found a tendency towards an increase in hand bone density and significant increase of the fourth metacarpal bone as assessed by radiodensitometry in those who were supplemented with 600 mg Ca/d (as calcium lactate) from 20 weeks pregnancy to term compared with women receiving 300 mg/d or placebo.

The expected effects of an increase in Ca intake on bone resorption have been noted in studies of pregnant women. Pregnant Mexican women experienced a 14 % reduction in the bone resorption marker NTx after supplementation for 12 d⁽Reference Janakiraman, Ettinger and Mercado-Garcia²¹⁰⁾. This parallels findings from the Chinese supplementation study described above in which lower hydroxyproline excretion was observed in the supplemented groups at the end of the treatment period⁽Reference Liu, Qiu and Chen²⁰⁷⁾. These studies provide further evidence that the physiological response to a higher Ca load remains effective during pregnancy. However, rare cases of life-threatening milk alkali syndrome (hypercalcaemia, metabolic alkalosis and renal insufficiency) during or after pregnancy have been reported in women consuming large quantities of Ca-containing supplements as antacids⁽Reference Gordon, McMahon and Hamblin²¹¹⁾ or combining moderate antacid consumption with a high dietary Ca intake⁽Reference Caplan, Miller and Silva²¹²⁾. Total Ca intakes of 2500 mg/d have not been shown to cause milk alkali syndrome⁽²¹³⁾, and this is reflected in the recent tolerable upper intake levels set by the Institute of Medicine in 2010 of 2500 mg/d for pregnant or lactating women aged 19–50 years⁽²¹⁴⁾.

Influence on the mother in lactation

Observational and supplementation studies have demonstrated that the skeletal response to lactation is independent of the BF mother's Ca intake⁽Reference Prentice¹⁴⁹⁾. Most observational studies of BF women have shown no significant relationship between dietary Ca intake and changes in bone mineral during lactation⁽Reference Laskey, Prentice and Hanratty^21, Reference Kolthoff, Eiken and Kristensen^41, Reference Sowers, Randolph and Shapiro^79, Reference Sowers, Corton and Shapiro^89, Reference Lopez, Gonzalez and Reyes⁹⁴⁾. Similarly, controlled supplementation studies have demonstrated little or no effect of increases in Ca intake on changes in bone mineral, intestinal Ca absorption efficiency, renal Ca handling or Ca metabolism during or after lactation⁽Reference Polatti, Capuzzo and Viazzo^82, Reference Kalkwarf, Specker and Bianchi^83, Reference Cross, Hillman and Allen^87, Reference Kalkwarf, Specker and Ho^95, Reference Kalkwarf, Specker and Heubi¹⁴⁸⁾ even among Gambian women with a very low dietary Ca intake⁽Reference Prentice, Jarjou and Stirling^113, Reference Prentice, Jarjou and Cole^152, Reference Fairweather-Tait, Prentice and Heumann²¹⁵⁾. Transient effects of Ca supplements on aBMD have been reported in BF women during and after lactation⁽Reference Polatti, Capuzzo and Viazzo^82, Reference Kalkwarf, Specker and Bianchi⁸³⁾ but these are also observed in NBF and NPNL women and are likely to be due to the expected alterations in bone remodelling, similar to those seen when Ca is used as an anti-resorptive agent in older women⁽Reference Prentice, Pettifor, Juppner and Glorieux³⁾.

Adolescent mothers may be an exception, although the evidence is inconclusive. In a US dietary intervention study from 2 to 16 weeks postpartum in which forearm BMC was measured by SPA, control BF adolescents on their normal diet of 900 mg Ca/d had a 10 % decrease in BMC. In contrast, experimental adolescent and adult BF groups who received dietary advice to increase daily Ca intake through dairy products and other Ca-rich foods and supplements (to ≥  1600 and 1200 mg Ca/d, respectively) had no significant decreases (3 and 5 %, respectively)⁽Reference Chan, McMurry and Westover²¹⁶⁾. In a Gambian study, no significant effect of age (teenage v. adult women) was observed on changes in BA-adjusted BMC of the radius measured by SPA or biochemistry during lactation among BF women randomised to receive a Ca supplement (714 mg Ca/d) for 12 months⁽Reference Prentice, Jarjou and Stirling^113, Reference Prentice, Jarjou and Cole¹⁵²⁾.

There is evidence that Ca intake during pregnancy may influence the mother's response to lactation. In the Gambian study described earlier among women with a low Ca intake in a population where breast-feeding is continued for 18–24 months, there was evidence that Ca supplementation (1500 mg/d) during the latter half of pregnancy resulted in more pronounced lactational bone mineral mobilisation from the lumbar spine and distal radius measured up to 12 months postpartum⁽Reference Jarjou, Laskey and Sawo²⁰⁶⁾. The Ca-supplemented group also had biochemical changes measured at 13 weeks of lactation consistent with greater turnover of mineral between the maternal skeleton and the extracellular pool, and greater urinary Ca excretion. These effects may represent a disruption of the processes of adaptation to a low dietary Ca intake and research is ongoing to determine whether they are temporary or remain after breast-feeding stops.

Observational studies and the wide inter-individual and geographical variations in breast-milk Ca concentration have suggested that breast-milk Ca content may be influenced by maternal Ca intake during lactation or during the previous pregnancy⁽Reference Ortega, Martinez and Quintas^217, Reference Prentice, Dibba and Jarjou²¹⁸⁾. However, Ca supplementation studies of women during lactation⁽Reference Kalkwarf, Specker and Bianchi^83, Reference Prentice, Jarjou and Cole¹⁵²⁾, and more recently during pregnancy⁽Reference Jarjou, Prentice and Sawo²⁴⁾, have demonstrated that breast-milk Ca concentration is independent of maternal Ca intake, even amongst women with very low Ca intakes. In addition, because breast-milk Ca secretion is regulated by the casein, phosphate and citrate components, it is now recognised that maternal Ca intake is unlikely to influence breast-milk Ca secretion directly⁽Reference Kent, Arthur and Mitoulas¹⁷³⁾.

Influence on the mother in later life

Few studies have investigated whether a low Ca intake during pregnancy and lactation increases the risk of postmenopausal osteoporosis⁽Reference Prentice, Pettifor, Juppner and Glorieux³⁾. In studies that have attempted to look for interactions between Ca intake and reproductive history, no associations have been identified⁽Reference Kleerekoper, Peterson and Nelson²¹⁹⁾. However, African women with low habitual dietary Ca intake, high parity and long lactation periods are not at increased risk of fragility fractures in old age compared with Western women⁽Reference Walker, Richardson and Walker^142, Reference Aspray, Prentice and Cole^220–Reference Walker²²²⁾.

Influence on the child

Early studies of body Ca content of newborn infants suggest that fetal Ca accretion is influenced by maternal nutrition⁽Reference Apte and Iyengar¹⁵⁷⁾. Infants born to mothers from a poor socio-economic community in India had lower bone density, assessed by radiodensitometry within 48 h of birth, in the arms and legs than infants born to matched controls from a more affluent group⁽Reference Krishnamachari and Iyengar²⁰⁹⁾. Infants born to mothers supplemented with Ca, either 300 or 600 mg/d, during pregnancy had higher radiographic bone density of their arms and legs than those born to controls, but there was no difference between the supplemented groups⁽Reference Raman, Rajalakshmi and Krishnamachari²⁰⁸⁾. A DXA study has suggested that infants in rural areas of The Gambia have lower whole-body BMC, and hence total body Ca content, than infants of the same age in Western populations⁽Reference Jarjou, Prentice and Sawo²⁴⁾. However, the extent to which these results in Indian and African women reflect low maternal Ca intakes as opposed to small maternal and fetal size associated with poor general nutrition is unclear. The Gambian study also showed that Ca supplementation (1500 mg/d) of the mothers during pregnancy had no significant effect on fetal bone mineral accretion, as measured by SPA and DXA at 2 weeks, or on birth weight and other anthropometry⁽Reference Jarjou, Prentice and Sawo²⁴⁾. An intervention study in the USA showed a higher whole-body BMC 2d after delivery in the offspring of women in the lowest quintile of dietary Ca intake ( <  600 mg/d) randomised to receive 2000 mg Ca/d in pregnancy compared with those given placebo and those with higher dietary Ca intakes⁽Reference Koo, Walters and Esterlitz²²³⁾. In addition, studies looking at dietary determinants of birth weight and fetal bone dimensions have suggested that there are positive associations between fetal growth and bone mineral and Ca-rich foods, such as dairy products⁽Reference Chan, McElligott and McNaught^224, Reference Chang, O'Brien and Nathanson²²⁵⁾. It is possible therefore that, in the Indian and Gambian studies, shortages of other nutrients may have prevented a response to the increased maternal Ca intake, but suggests that Ca alone does not limit fetal bone accretion in these populations⁽Reference Prentice, Laskey and Goldberg²³⁾. However, it is possible that a low maternal Ca intake may be limiting in mothers with poor vitamin D status, but to date there have been no studies directly exploring this possibility.

Based on a small number of studies, there are conflicting indications about whether maternal Ca intake during pregnancy influences the bone mineral accretion of the child in the long term. An observational study in India among women with a low Ca intake reported that women with a higher frequency of intake of Ca-rich foods during pregnancy had children with higher BMC and aBMD of the spine and whole body at 6 years of age than mothers with lower intakes of Ca-rich foods⁽Reference Ganpule, Yajnik and Fall²²⁶⁾. However, in an Australian longitudinal study, no association was found between maternal dietary intake of Ca during pregnancy and aBMD at the spine, hip or whole body of their children at 8 years of age⁽Reference Jones, Riley and Dwyer²²⁷⁾. Additionally, in the Gambian study described above⁽Reference Jarjou, Prentice and Sawo²⁴⁾, there was no evidence of a beneficial effect of Ca supplementation in pregnancy on skeletal dimensions as measured by crown–heel length and head circumference at 12 months of age⁽Reference Jarjou, Prentice and Sawo²⁴⁾ or on stature at age 5–10 years⁽Reference Hawkesworth, Sawo and Fulford²²⁸⁾.

Influence on the mother

There is no evidence that the biological requirement for vitamin D is increased by pregnancy and lactation because only small amounts of vitamin D and its metabolites cross the placenta or are transferred into breast milk⁽Reference Kovacs^{5, 213,} Reference Specker²³⁵⁾. Frank clinical vitamin D deficiency in adults causes osteomalacia, hypocalcaemia and secondary hypoparathyroidism; there is no evidence to suggest that this worsens during pregnancy⁽Reference Kovacs⁵⁾. In theory, poor vitamin D status during pregnancy and lactation, at 25OHD concentrations above those associated with clinical vitamin D deficiency, might compromise Ca homeorrhesis, such as the ability to increase intestinal Ca absorption and renal Ca retention, and might lead to a more exaggerated maternal skeletal response and compromise the mother's bone health. However, the extent to which this is the case is not known. Few studies have investigated the possible interaction between vitamin D status and maternal Ca and bone metabolism during pregnancy and lactation. One observational study reported that British women who were pregnant during the winter had greater reductions in QUS bone variables than those pregnant during the summer, suggesting an interaction with vitamin D status⁽Reference Javaid, Crozier and Harvey⁶⁹⁾.

Maternal vitamin D status during lactation also influences the concentration of breast-milk vitamin D metabolites. Vitamin D (cholecalciferol and ergocalciferol) transfers readily into breast milk from the maternal circulation, 25OHD less so and 1,25(OH)₂D hardly at all⁽Reference Kovacs⁵⁾. The concentrations of vitamin D and its metabolites in breast milk parallel those in the mother's circulation, but at lower concentrations. In US women, the breast-milk concentration of vitamin D increased 10-fold to a peak within 48 h of a single exposure to UVB radiation at 1·5 minimal erythmal dose and remained above baseline levels for at least 2 weeks⁽Reference Greer, Hollis and Cripps^236, Reference Hollis²³⁷⁾. These changes closely paralleled the concentrations of maternal serum vitamin D but were lower by approximately 10- to 15-fold. Similarly, oral supplementation with vitamin D₃ or D₂ has been shown to increase the vitamin D content of breast milk, with smaller increases in 25OHD⁽Reference Basile, Taylor and Wagner^238–Reference Wagner, Hulsey and Fanning²⁴⁰⁾. Very high concentrations were measured when vitamin D was given at therapeutic doses during pregnancy to treat an underlying clinical disorder⁽Reference Greer, Hollis and Napoli²⁴¹⁾. These data, and those from animal studies, suggest that only unmetabolised vitamin D is found in significant quantities in milk and thus is the predominant dietary form of vitamin D available to the exclusively breast-fed infant⁽Reference Greer, Hollis and Cripps²³⁶⁾.

It is also plausible that maternal vitamin D status might influence the incorporation of Ca into breast milk. However, no association between breast-milk Ca concentration and maternal vitamin D status (25OHD) was observed in a study of British and Gambian women⁽Reference Prentice, Yan and Jarjou²⁴²⁾, and no differences in breast-milk Ca were observed between US mothers who consumed 50 μg/d (2000 IU/d) or 100 μg/d (4000 IU/d) supplemental vitamin D between 1 and 4 months of lactation compared with historical controls consuming 10 μg/d (400 IU/d)⁽Reference Basile, Taylor and Wagner²³⁸⁾.

Influence on the child

Fetal 25OHD, as measured in cord blood, mirrors that in the maternal circulation, at similar or slightly lower concentrations. Therefore, maternal vitamin D status in pregnancy is the key determinant of neonatal vitamin D status⁽Reference Kovacs^5, Reference Specker^243, Reference Cockburn, Belton and Purvis²⁴⁴⁾, and, together with infant UVB skin exposure and the limited supply through breast milk, of vitamin D status in the first months of life⁽Reference Specker, Vieira and O'Brien^{154, 213, 231,} Reference Specker^243, Reference Pawley and Bishop²⁴⁵⁾. Vitamin D deficiency in the pregnant mother is associated with congenital rickets, craniotabes and hypocalcaemia in the newborn, and rickets in infancy^{(213, 214, 231,} Reference Specker²⁴³⁾. There is evidence that maternal vitamin D status during pregnancy at 25OHD concentrations above that associated with clinical deficiency may influence fetal and infant bone growth and dental development^(231, Reference Specker²⁴³⁾, although the data are conflicting. Birth weight and neonatal BMC and bone turnover have been related to season of birth in countries where maternal vitamin D status is seasonally dependent⁽Reference McGrath, Keeping and Saha^246–Reference Namgung, Tsang and Lee²⁴⁸⁾. Positive associations have been reported between birth weight and length and maternal vitamin D intake⁽Reference Mannion, Gray-Donald and Koski²⁴⁹⁾ and infant vitamin D status⁽Reference Weiler, Fitzpatrick-Wong and Veitch²⁵⁰⁾ among infants in Canada; however, these observations were confounded by maternal milk intake because Canadian milk is fortified with vitamin D. Infants of Australian mothers who were vitamin D deficient at 28–32 weeks of pregnancy (25OHD < 28 nmol/l) had shorter knee–heel length at birth than other infants, indicating a difference in long-bone growth, but other birth measures were unaffected⁽Reference Morley, Carlin and Pasco²⁵¹⁾. A study in The Gambia, in which all women had a plasma 25OHD concentration >50 nmol/l at 20 weeks of pregnancy, found no significant relationships between maternal vitamin D status and infant growth or bone mineral during the first year of life⁽Reference Prentice, Jarjou and Bennett²⁵²⁾. Maternal vitamin D status during pregnancy may have long-term effects on bone mineral accretion in childhood. A low concentration of maternal 25OHD in late pregnancy has been associated with lower whole-body and lumbar spine BMC in UK offspring at 9 years of age; maternal UVB skin exposure and vitamin D supplement use in late pregnancy were also predictors⁽Reference Javaid, Crozier and Harvey²⁵³⁾.

In pregnant women at risk of low vitamin D status, vitamin D supplementation in mid–late gestation with doses ranging from 10 to 30 μg/d (400 to 1200 IU/d) has demonstrated greater cord and plasma Ca concentrations, lower plasma ALP concentrations, smaller fontanelle size and lower incidence of growth retardation and neonatal hypocalcaemia in the newborns⁽Reference Cockburn, Belton and Purvis^244, Reference Brooke, Brown and Bone^254–Reference Mallet, Gugi and Brunelle²⁶⁰⁾ and effects on subsequent infant growth⁽Reference Brooke, Butters and Wood²⁵⁵⁾. Other studies have reported no effects on birth weight⁽Reference Mallet, Gugi and Brunelle²⁶⁰⁾ or infant forearm bone mineral⁽Reference Congdon, Horsman and Kirby²⁶¹⁾. It should be noted that many of these studies were small and did not have randomised, controlled protocols.

As described in the previous section, unmetabolised vitamin D is the predominant form of vitamin D transferred postnatally from the mother to the breast-fed infant. The concentrations of vitamin D in breast milk are influenced by the mother's UVB exposure and dietary intake. Supplementation of US women during lactation with doses of 10 μg/d (400 IU/d) vitamin D has been shown to have relatively little influence on the vitamin D status of their breast-fed child, but increases in serum 25OHD concentrations have been observed in the infants of US lactating women consuming supplemental vitamin D at doses of 50–160 μg/d (50–6400 IU/d)⁽Reference Specker^235, Reference Basile, Taylor and Wagner^238–Reference Wagner, Hulsey and Fanning²⁴⁰⁾.

There is considerable controversy over the definition of vitamin D adequacy for pregnant and lactating women that takes into account the requirements for the mother and infant, other potential health outcomes for the mother and child, and the UVB exposure and/or supplemental doses required to achieve it⁽Reference Abrams²⁶²⁾. Large supplementation trials^(263–265) are currently ongoing in the UK, USA and Canada to provide more definitive evidence.