Bone 32 (2003) 449 – 454
Longitudinal changes in bone mineral density during normal pregnancy M. Kaur,a D. Pearson,b I. Godber,c N. Lawson,c P. Baker,d,1 and D. Hoskinga,* a
Division of Mineral Metabolism, City Hospital, Hucknall Road, Nottingham NG5 1PB, UK Department of Medical Physics, City Hospital, Hucknall Road, Nottingham NG5 1PB, UK c Department of Clinical Chemistry, City Hospital, Hucknall Road, Nottingham NG5 1PB, UK d Department of Obstetrics and Gynaecology, City Hospital, Hucknall Road, Nottingham NG5 1PB, UK b
Received 13 November 2002; accepted 17 December 2002
Abstract Pregnancy is a common physiological event that could affect peak bone mass and the risk of developing osteoporosis later in life. There have been few longitudinal studies over a complete reproductive cycle of any size to show whether bone mineral density (BMD) changes. We have measured BMD by dual-energy X-ray absorptiometry in 46 normal women before conception and then again immediately after delivery and compared them with 30 control women who failed to conceive. Fifteen women were osteopenic in preconceptual BMD, but there was no difference between those who did or did not become pregnant. During pregnancy there was a small and statistically nonsignificant decline in BMD at all sites. The decrease at the trochanteric region was 4.2%, while losses at other sites were about 1%. The decline at the trochanter exceeded the least significant change between two measurements (5.04%) in 17 women (40.5%) with significant changes within individuals being much less common at the other measurement sites. The nonpregnant controls showed small increases in BMD of 0.3%–1.9% but no woman lost more than the least significant change. At the trochanter there was a significant difference (P ⫽ 0.013) between those who did and did not become pregnant. There was a good correlation between changes in BMD at all sites and no significant difference in the slope of these correlations between the pregnant and control groups. Correlations with lumbar spine were total hip, r ⫽ 0.46, P ⫽ 0.0001; femoral neck, r ⫽ 0.49, P ⫽ 0.0005; and trochanter, r ⫽ 0.66, P ⬍ 0.0001. © 2003 Elsevier Science (USA). All rights reserved. Keywords: Pregnancy; Bone mineral density; Trochanter; Osteoporosis
Introduction All women lose bone after menopause and the main determinants of whether this will lead to osteoporosis are the rate of loss and its starting point represented by the peak bone mass. Pregnancy and lactation are common physiological events that could affect peak bone mass and thus the development of osteoporosis later in life. In theory the risk to the maternal skeleton should be small since the 30 g of calcium in the full-term neonate  should be met by adaptation in maternal calcium homeostasis. Although the renal conservation of calcium is less efficient in pregnancy [2,3], active calcium absorption by the * Corresponding author. Fax: ⫹44-115-962-7900. E-mail address: [email protected]
(D. Hosking). 1 Current address: St. Mary’s Hospital, Manchester, UK.
intestine doubles by 24 weeks . Even if the whole of the neonates needs were met by the maternal skeleton, containing 1000 g of calcium, the effect on structure should be relatively slight. Currently there is no consensus as to whether bone mineral density (BMD) changes during the reproductive cycle and if so whether any losses are subsequently replaced. The technological and methodological basis for this uncertainty has been recently reviewed in detail . Most previous studies have been either too small or cross-sectional in nature while what is needed are large-scale longitudinal measurements through a complete reproductive cycle [6 – 8]. We have been able to recruit a substantial group of normal young women who were contemplating pregnancy and measure BMD preconceptually. Those who subsequently became pregnant were measured again soon after delivery and at several time points in the postpartum year.
8756-3282/03/$ – see front matter © 2003 Elsevier Science (USA). All rights reserved. doi:10.1016/S8756-3282(03)00017-6
M. Kaur et al. / Bone 32 (2003) 449 – 454
In the present article, we describe the longitudinal changes in BMD during pregnancy.
Patients and methods One-hundred twenty-two healthy women who were contemplating pregnancy were recruited by newspaper and hospital advertisements for this prospective study. The women were required to be in good general health as judged by medical history, physical examination, and routine laboratory screening tests. They were excluded if there was evidence of metabolic bone disease or they were taking any medication such as glucocorticoids that might affect the skeleton. Body mass index (BMI) was measured at baseline. BMD of the lumbar spine (L1–L4) and hip (femoral neck, trochanter, total hip) were measured by dual energy X-ray absorptiometry (DXA) on a QDR 2000 (Hologic, Waltham, MA). All women were measured at entry to the study before conception. Forty-two women became pregnant during the course of the study. In these women, BMD was measured again within 2 weeks of completion of a full-term pregnancy. Thirty women who failed to conceive had their BMD measured again at 12–24 months to serve as a control group. Positioning of patients during absorptiometry and data analysis were standardised as were machine calibration and technician training. Quality assurance of the densitometric technique followed the manufacturer’s protocol using the Hologic spine phantom. The long-term coefficient of variation (CV%) for each site was 1.8% for lumbar spine (LS), 1.8% for trochanter (TR), 1.9% for femoral neck (FN), and 1.5% for the total hip (TH) region . The least significant change (LSC) in BMD that could be detected depended on the precision error and was calculated as 2.8 * CV% based on a 95% confidence interval . The number of patients in each group who had gained or lost more than the LSC in BMD between baseline and the postdelivery or 12-month follow-up scan was calculated. The study protocol was approved by the Nottingham City Hospital Ethics Committee and all participants gave formal written consent. Groups were compared at baseline using one-way analysis of variance (ANOVA) or a Mann–Whitney test where data were not normally distributed. A repeated-measures ANOVA  was used to see if there was a significant change in BMD between visits and a significant difference between the pregnancy and control groups with time. A Pearson’s correlation coefficient was used to investigate the association between BMD at baseline and the change in BMD. An ANOVA was used to compare the associations between the changes in BMD in the pregnancy and control groups . A 2 test was used to compare categorical variables. Where multiple end points were compared, a modified Bonferonni adjustment was applied to the significance level . The modification corrects for the correla-
Table 1 Baseline characteristics of both groups of women in the study
Age (years)a Age at menarche (years)a Nulliparous women (n) Multiparae: total pregnancies (n) Years since last pregnancy (median, range) BMI (kg m⫺2)a Calcium intake (g/day)a BMD (g/cm2)a L1–L4 Total hip Femoral neck Trochanter Smokers (n) Alcohol (units/week) 0–10 11–20 ⬎21
Pregnancy (n ⫽ 46)
Controls (n ⫽ 30)
31 ⫾ 5 12.9 ⫾ 1.4 22 28
32 ⫾ 5 12.9 ⫾ 1.3 15 15
24.5 ⫾ 3.2 864 ⫾ 270b
26.2 ⫾ 5.5 949 ⫾ 208c
1.036 ⫾ 0.140 0.954 ⫾ 0.110 0.839 ⫾ 0.101 0.715 ⫾ 0.094 7
1.016 ⫾ 0.140 0.942 ⫾ 0.114 0.823 ⫾ 0.111 0.713 ⫾ 0.093 5
37 9 0
22 6 2
Mean ⫾ SD. n ⫽ 41. c n ⫽ 18. a
tion between the change in BMD at different sites within lumbar spine and hip. The most conservative value of Bonferonni adjustment was used based on the weakest correlation between BMD at the lumbar spine and the total hip region.
Results In the 42 patients who were recruited to the pregnancy group (mean age ⫾ SD 31 ⫾ 5 years, range 22 to 40 years), the mean (⫾SD) time between the preconception DXA and the postdelivery DXA was 14 ⫾ 3 months, (median 12 months, range 9 to 22 months). The mean (⫾ SD) time to follow up in the 30 controls (mean age 32 ⫾ 5 years, range 24 to 42 years) was 16 ⫾ 3 months (median 15 months, range 11 to 24 months). The baseline characteristics of the patients and controls are shown in Table 1. There was no significant difference between the groups at baseline in any measurement including years since last pregnancy (Mann– Whitney test). There was no change in the BMI of the control group during the study (preconception 26.2 ⫾ 5.4 kg/m2, follow-up 26.7 ⫾ 5.4 kg/m2) but the pregnant women did gain weight so that there was a nonsignificant gain in BMI from 24.4 ⫾ 3.2 kg/m2 prior to conception to 27.4 ⫾ 3.6 kg/m2 after delivery. In the group as a whole (n ⫽ 72) there was a good correlation between the preconceptual LS and FN BMD (r ⫽ 0.73, P ⬍ 0.001) and other hip sites. At the preconceptual measurement 15 women (21%) were osteopenic (T score below ⫺1) at both LS and FN. None were osteoporotic.
M. Kaur et al. / Bone 32 (2003) 449 – 454
Fig. 1. Absolute change in BMD (g cm⫺2) by site between preconception and postdelivery in the pregnancy group.
There was no difference between the BMD of those who did and did not become pregnant. There was no correlation between preconceptual BMD at any site and dietary calcium intake. In the patients who became pregnant there was a small, but statistically nonsignificant, decrease in BMD at all measurement sites (Fig. 1) by the postdelivery scan, with the greatest change occurring at the trochanter (LS ⫽ ⫺0.9%, TH ⫽ ⫺1.2%, FN ⫽ ⫺0.7%, TR ⫽ ⫺4.2%). The number of patients who showed a decrease in BMD that exceeded the LSC for the particular measurement site (5.04% at LS and TR, 5.32% at FN, and 4.2% at TH) was LS ⫽ 3 (7.1%), TH ⫽ 6 (14.3%), FN ⫽ 8 (16.7%), and TR ⫽ 17 (40.5%); these patients are identified in Fig. 1. The total number of women with a decrease in excess of the LSC at any site was 20 (47.6%). Only two patients showed a gain in BMD that exceeded the LSC at the spine or hip. There was no correlation between the change in BMD during pregnancy and the preconceptual value at any site. There was no change in the quality control of DXA throughout the study as measured by the Hologic spine phantom. There was no significant difference in baseline BMD or the change in BMD in those patients on oral contraceptives prior to conception (n ⫽ 23) compared to the women using other methods (barrier, n ⫽ 13; IUCD or none, n ⫽ 6). There was no correlation between the birth weight of the baby (mean 3.4 ⫾ 0.5 kg, range 2.4 – 4.5 kg) and the degree of bone loss at any site. There was a slight increase in BMD at all measurement sites (Fig. 2) in those who failed to become pregnant (LS 1.9%, TH 1.1%, FN 1.1%, TR 0.3%). There was no correlation between these changes and the baseline BMD. A small number of women showed gains greater than the LSC
(LS ⫽ 1, TH ⫽ 1, FN ⫽ 3, TR ⫽ 0). No patient showed a decrease in BMD greater than the LSC. There was a significant difference between the change in BMD of the pregnancy and control groups at the trochanter (pregnancy ⫺0.028 ⫾ 0.035 g cm⫺2; control ⫹0.003 ⫾ 0.14 g cm⫺2; 0.046 ⬎ P ⬎ 0.042). There was a good correlation between the change in BMD at individual sites indicating a generalised loss of bone. There was no significant difference in the slope of these correlations between the pregnancy and control groups, implying that changes across measurement sites in the control group appeared to show the same relationship as the pregnant group (Fig. 3). These data were, therefore, combined. In the pregnant group the correlations with LS were TH, r ⫽ 0.46, P ⫽ 0.001; FN, r ⫽ 0.49, P ⫽ 0.005; and TR, r ⫽ 0.66, P ⬍ 0.0001. The correlations between LS and TR are shown for both groups in Fig. 3.
Discussion This longitudinal study of spine and hip BMD during pregnancy is one of the largest to be completed, although a recent investigation measured spine and radius but not hip . Both are generally reassuring with respect to bone loss in terms of the magnitude and distribution of change. We had been concerned that the method of recruitment might lead to a preponderance of women with a low BMD who might be more likely to volunteer if they had a family history of osteoporosis. However, the prevalence of osteopenia at 21% was very similar to that expected in the normal population (14.5%). Those who became pregnant were evenly divided between first and subsequent pregnan-
M. Kaur et al. / Bone 32 (2003) 449 – 454
Fig. 2. Absolute change in BMD (g cm⫺2) by site between baseline and 1-year follow-up scans in the control group.
cies but this did not seem to influence the change in BMD. This is consistent with previous epidemiological evidence that failed to show an association between parity and BMD [13–15], although a recent study from Nottingham did show a positive influence of parity on femoral neck and total radius BMD but not on spine, trochanter, or total body . There has also been concern about the heterogeneity of BMD changes because of the inclusion of adolescent or late premenopausal pregnancies at a time when there may be an altered response of the skeleton to the demand for calcium
Fig. 3. Change in trochanteric BMD against change in lumbar spine BMD in the pregnancy (closed symbols) and control groups (open symbols).
. Most of the present group of women were in their late 20s or early 30s at the time of their pregnancy when dietary calcium intake has a diminishing influence on skeletal consolidation and peak bone mass ; this may explain why there was no correlation between preconceptual BMD and calcium intake. Although formal dietary records (including milk intake) were available in 40 of the 42 pregnancies, it is interesting that only 16 women exceeded the recommended threshold value of 957 mg/day that would be needed to maintain optimal peak bone mass . Very few studies in pregnancy have included a control group but this does provide some assessment of drift in the densitometric technique and controls for age-related changes in BMD as well as other clinical characteristics. There was no difference for any baseline variable, including BMD, between those who did and did not become pregnant. This might be an indirect indication that the failure to conceive was not solely due to poor ovarian function with all its implications for skeletal health. The control group showed a small gain in BMD of 1.9% at the spine and 1.1% at the hip, which was independent of age and baseline BMD. Young women are known to consolidate their skeleton once longitudinal growth is completed but the changes seen in the nonpregnant women were twice those reported in a 4-year study of 18- to 26-year-old college students . One possible explanation is that the women who were hoping to conceive made important lifestyle changes such as increasing dietary calcium intake or reducing alcohol and cigarette consumption, which might have resulted in a small augmentation of skeletal consolidation. Unfortunately none of these variables were measured in the present study al-
M. Kaur et al. / Bone 32 (2003) 449 – 454
though the generally poor calcium intake before conception may have been a stimulus for change given the widespread perception that calcium requirements are increased in pregnancy. Although the loss of bone during pregnancy was relatively small and not significantly different from the preconceptual value in the group as a whole, 17 of the 42 women (40.5%) showed a fall in BMD of more than 5.04% at the femoral trochanter, indicating that in these individuals there was a significant decrease during pregnancy. This finding is confirmed by the significant difference between controls and pregnant women in the change in BMD at the trochanter, a finding that remains significant after correcting for the use of multiple endpoints. The number of women experiencing a decrease in excess of the least significant change in BMD at other sites was much smaller and this preponderance of loss at the trochanter with maintenance of BMD at the spine is intriguing. The lumbar spine and trochanter both contain approximately equal proportions of cortical and trabecular bone and so a difference in the metabolic activity of the two types of bone is unlikely to be the explanation. One possibility is that the femoral neck and trochanter respond differently to the effects of weight bearing and the demands of calcium homeostasis during pregnancy. Weight gain during pregnancy would be expected to maintain hip BMD but a recent study of bone strength during puberty might provide a possible explanation . There are important differences in the pattern of bone growth at puberty between girls and boys [21,22]. In girls periosteal bone apposition stops while endosteal growth continues with narrowing of the medullary cavity, while in boys there is continued periosteal growth with little change in the endocortical diameter. Thus although both sexes will have the same cortical thickness, the bone diameter, which contributes to mechanical strength, is reduced in girls . It has been suggested that in girls the accumulation of endosteal bone of little mechanical value might provide a reservoir of calcium which can be mobilised during pregnancy . Differences in the loading of the trochanter and femoral neck during pregnancy might result in loss at the former with preservation at the latter site. There is some support for such an hypothesis in that during walking it is the cortical bone of the femoral neck that is the main weight-bearing region of the proximal femur rather than the trabecular bone of the trochanter [24,25]. The stresses at these two sites are also different with compression at the femoral neck owing to weight bearing and tension at the trochanter owing to the effect of the hip abductors. In circumstances such as pregnancy where there seems to be a change in bone remodeling , there will be maintenance of bone at sites that are highly stressed, such as the femoral neck, but loss in regions of lower stress such as the trochanter. This might not have major mechanical consequences if this loss was largely confined to trabecular bone since 80% of the load in this region is carried by the cortex .
Prospective studies of BMD measured by DXA before and after pregnancy are relatively scarce. A large study of 38 Hungarian women  found a 2.1% decrease at the lumbar spine while a smaller study of 10 women from the UK found a 2.0% decrease . Only the smaller study measured changes at the hip and found a 3.6% decrease at the total hip and a 4.8% decrease at the trochanter. The results of these two studies are in broad agreement with the present findings. A small study of 5 women from Finland  found that all but one lost, on average, approximately 3% at the lumbar spine but the changes at the femoral neck were variable and mostly lay within the least significant change. Longitudinal studies using single- or dual-photon absorptiometry, reviewed by Sowers , were inconsistent and either found no change or (in one study) a loss at femoral neck and midradius. The present study may also shed light on the nature of pregnancy-associated osteoporosis. It is uncertain whether this condition is due to a systematic loss of bone during pregnancy in women with preexisting osteoporosis sufficient to cause fractures or whether it is an idiosyncratic loss that is so severe that it causes architectural failure . We found that pregnancy was associated with a generalised reduction in bone mineral density but that this would not be of sufficient magnitude to cause fractures unless the woman already had a substantial decrease in bone mass. Moreover, the amount of loss was not related to the preconceptual BMD, which is reassuring in that women with osteopenia, who made up almost one-fifth of our volunteer sample, were not more at risk of excessive loss than those with a normal BMD. The pattern of change in BMD that we describe is also interesting in that pregnancy-associated osteoporosis involves the spine more commonly than the hip [28,29] and yet we found a preponderance of loss at the trochanteric region of the hip with minimal loss at the spine. This implies that pregnancy-associated osteoporosis is an idiosyncratic phenomenon rather than a generalised loss of bone in those with a preexisting low bone mass. Although the 4.2% loss at the trochanter and the 1.2% decrease at the total hip that we found is much less than the 4.42% gain in total hip BMD described on recovery of pregnancy-associated osteoporosis , this will also include a contribution from lactation that may be associated with substantial bone loss . In conclusion, this study shows that pregnancy is usually associated with small changes in BMD predominantly in the trochanteric region of the hip. These changes are too small to cause pregnancy-associated osteoporosis, which is more likely to represent an idiosyncratic response.
Acknowledgments This study was funded by the National Osteoporosis Society, Camerton, Bath, UK. We are also grateful for the
M. Kaur et al. / Bone 32 (2003) 449 – 454
assistance of Sue Cawte in the Department of Nuclear Medicine and Pat San for nursing support.
References  Givens MH, Macy IG. The clinical composition of the human fetus. J Biol Chem 1933;102:7–17.  Dahlman T, Sjoberg HE, Bucht E. Calcium homeostasis in normal pregnancy and puerperium. Acta Obstet Gynecol Scand 1994;73: 393– 8.  Pedersen EB, Johannsen P, Kristensen S, Rasmussen AB, Emmertsen K, Lauritsen JG, Wohlert M. Calcium, parathyroid hormone and calcitonin in normal pregnancy and pre-eclampsia. Gynecol Obstet Invest 1984;18:156 – 64.  Heaney RP, Skillman TG. Calcium metabolism in normal human pregnancy. J Clin Endocrinol 1971;33:661–70.  Sowers M. Pregnancy and lactation as risk factors for subsequent bone loss and osteoporosis. J Bone Miner Res 1996;11:1052– 60.  Cross NA, Hillman LS, Allen SH, Krause GF, Viera NE. Calcium homeostasis and bone metabolism during pregnancy, lactation and post weaning: a longitudinal study. Am J Clin Nutr 1995;61:514 – 523.  More C, Bettembuk P, Bhattoa HP, Balogh A. The effects of pregnancy and lactation on bone mineral density. Osteoporos Int 2001; 12:732–7.  Richie LD, Fung EB, Halloran BP, Turnlund JR, Van Loan MD, Cann CE, King JC. A longitudinal study of calcium homeostasis during human pregnancy and lactation and after resumption of menses. Am J Clin Nutr 1998;67:693–701.  Sahota O, San P, Cawte SA, Pearson D, Hosking DJ. A comparison of the longitudinal changes in quantitative ultrasound with dualenergy X-ray absorptiometry: the four-year effects of hormone replacement therapy. Osteoporos Int 2000;11:52– 8.  Blake GM, Wahner HW, Fogelman I. The evaluation of osteoporosis: dual energy X-ray absorptiometry and ultrasound in clinical practice. 2nd ed. London: Martin Dunitz, 1999, p. 389 –90.  Chow S, Liu J. Design and analysis of clinical trials. New York: Wiley, 1998, p. 291–304.  Armitage P. Statistical methods in medical research. Oxford: Blackwell, 1971, p. 285– 8.  Fox KM, Magaziner J, Sherwin R, Scott JC, Plato CC, Nevitt M, Cummings S. Reproductive correlates of bone mass in older women. J Bone Miner Res 1993;8:901– 8.
 Kritz-Silverstein D, Barrett-Connor E, Hollenbach KA. Pregnancy and lactation as determinants of bone mineral density in postmenopausal women. Am J Epidemiol 1992;136:1052–9.  Murphy S, Khaw K-T, May H, Compston JE. Parity and bone mineral density in middle-aged women. Osteoporos Int 1994;4:162– 6.  Grainge MJ, Coupland CAC, Cliffe SJ, Chilvers CED, Hosking DJ. Reproductive, menstrual and menopausal factors: which are associated with bone mineral density in early postmenopausal women? Osteoporos Int 2001;12:777– 87.  Heaney RP, Abrams S, Dawson-Hughes B, Looker A, Marcus R, Matkovic V, Weaver C. Peak bone mass. Osteoporos Int 2000;11: 985–1009.  Matkovic V, Heaney RP. Calcium balance during human growth: evidence for threshold behaviour. Am J Clin Nutr 1992;55:992–96.  Recker RR, Davies KM, Hinders SM, Heaney RP, Stegman MR, Kimmel DB. Bone gain in young adult women. JAMA 1992;268: 2403– 8.  Schoenau E, Neu CM, Rauch F, Manz F. The development of bone strength at the proximal radius during childhood and adolescence. J Clin Endocrinol Metab 2001;86:613– 8.  Bass S, Delmas PD, Pearce G, Hendrich E, Tabensky A, Seeman E. The differing tempo of growth in bone size, mass, and density in girls is region specific. J Clin Invest 1999;104:795– 804.  Katzman DK, Bachrach LK, Carter DR, Marcus R. Clinical and anthropometric correlates of bone mineral acquisition in healthy adolescent girls. J Clin Endocrinol Metab 1991;73:1332–9.  Seeman E. The structural basis of bone fragility in men. Bone 1999; 25:143–7.  Lotz JC, Cheal EJ, Hayes WC. Stress distributions within the proximal femur during gait and falls: implications for osteoporotic fracture. Osteoporos Int 1995;5:252– 61.  Oden ZM, Selvitelli DM, Bouxsein ML. Effect of local density changes on the failure load of the proximal femur. J Orthop Res 1999;17:661–7.  Black AJ, Topping J, Durham B, Farquharson RG, Fraser WD. A detailed assessment of alterations in bone turnover, calcium homeostasis, and bone density in normal pregnancy. J Bone Miner Res 2000;15:557– 63.  Holmberg-Marttila D, Sievanen H, Tuimala R. Changes in bone mineral density during pregnancy and postpartum: prospective data on five women. Osteoporos Int 1999;10:41– 6.  Smith R, Athanasou NA, Ostlere SJ, Vipond SE. Pregnancy-associated osteoporosis. Q J Med 1995;88:865–78.  Khovidhunkit W, Epstein S. Osteoporosis in pregnancy. Osteoporos Int 1996;6:345–54.  Phillips AJ, Ostlere SJ, Smith R. Pregnancy-associated osteoporosis: does the skeleton recover? Osteoporos Int 2000;11:449 –54.