Osteoporos Int (2002) 13:379–387 ß 2002 International Osteoporosis Foundation and National Osteoporosis Foundation
Osteoporosis International
Original Article Bone Mass, Bone Metabolism, Gonadal Status and Body Mass Index H. Rico1, I. Arribas2, F. J. Casanova3, A. M. Duce5, E. R. Herna´ndez1 and J. Cortes-Prieto4 Departments of 1Medicine, 4Medical Specialties and 5Surgery, and 2Central Laboratory, ‘Prı´ncipe de Asturias’ University Hospital, University of Alcala´, Madrid; and 3Department of Preventive Medicine and Public Health, Autonomous University of Madrid, Spain
Abstract. Weight and gonadal status are the main determinants of bone mass in women. Because of this it is important to study which influences it more. The effect of weight (expressed as body mass index, BMI) and gonadal status of women on total-body bone mineral content (TBBMC) and regional bone mineral content (BMC) was investigated. A total of 373 normal women (mean age 48.9 13.4 years) were studied: 171 postmenopausal women (mean age 59.3 9.5 years; years since menopause 11.3 6.7 years); 76 perimenopausal women (mean age 48.9 2.2 years); and 126 premenopausal women (mean age 34.7 7.4 years). In all the women, TBBMC and regional BMC were determined by dual-energy X-ray absorptiometry. Also biochemical markers of bone metabolism (total alkaline phosphatase and tartrate-resistant acid phosphatase) and serum estrone and estradiol were determined. When the women were stratified by gonadal status and BMI, thin women (BMI 520 kg/m2) had significantly lower TBBMC and regional BMC, lower gonadal steroid concentration and higher levels of biochemical markers than overweight (BMI 25–30 kg/m2) and obese (BMI 430 kg/m2) women, regardless of gonadal status. Overweight and obese women had findings suggestive of increased parathyroid activity, but greater bone mass. Weight rather than gonadal steroid concentration is the main determinant of bone mass in women regardless of gonadal status.
Correspondence and offprint requests to: Prof. Dr. H. Rico, Departamento de Medicina, Universidad de Alcala´, E-28801 Madrid, Spain. Tel. +34 91 885 45 52. Fax: +34 91 885 45 26. e-mail:
[email protected]
Keywords: Body mass index; Bone mass; Bone metabolism; Gonadal status; Non-weight-bearing bones; Total and regional bone mass; Weight-bearing bones
Introduction The high economic and social costs of osteoporosis [1] have encouraged research in recent years, leading to advances in bone mass measurement and knowledge of the biologic markers of bone remodeling, which are extremely useful for understanding osteoporosis and abnormal bone metabolism [2,3]. Bone mass measurement techniques and markers of bone remodeling show important variations in relation to the gonadal status of women [2–6]. However, aside from the mechanical influence of weight on overall bone mass and on weight-bearing and non-weight-bearing bones [7,8], weight also influences gonadal status. Women who are heavier or have a greater body mass index have been reported to have higher levels of female gonadal steroids [9,10], due to the aromatization of gonadal steroid precursors in fat and muscle tissue in both premenopausal and postmenopausal women [11,12]. If weight influences gonadal steroid levels [9–12] in a large group of women of the same gonadal status, it may also affect bone mass and bone metabolism. Such an effect should be evident in the levels of biologic markers of bone remodeling. Therefore, this study was undertaken to examine how weight and gonadal status in women affect total-body bone mineral content (TBBMC), regional bone mineral content (head, arm, trunk and leg BMC), and bone metabolism in a broad sample of normal women stratified by gonadal status
380
(premenopausal, perimenopausal and postmenopausal) and body mass index (thin, normal, overweight and obese). We consider both of these factors very important for understanding bone metabolism.
Material and Methods
H. Rico et al.
the floor. The individual’s weight was that given by the densitometer as the value of total body mass. Both measurements were made with the subject in pyjamas and barefoot. The body mass index was calculated as the weight in kilograms divided by the square of height in meters (BMI, kg/m2). The characteristics of the groups (number, age, years since menopause, anthropometric data, etc.) are summarized in Tables 1–4.
Subjects A total of 373 normal women (mean age 48.9 13.4 years) were studied. Of these, 171 were postmenopausal, all with a natural onset of menopause, defined as the absence of a menstrual period for at least 12 months and serum levels of FSH 430 U/l (mean age 59.3 9.5 years; years since menopause (YSM): 11.3 6.7 years). Another 76 women were classified as perimenopausal, defined as having missed three to six menstrual periods with widely varying menstrual cycles and with episodes of menstrual bleeding occurring at intervals of more than 35, and less than 90 days (mean age 48.9 2.2 years). A third group contained 126 premenopausal subjects (mean age 34.7 7.4 years). In all premenopausal women, menstrual histories indicated current and prior menstrual regularity (11–13 cycles/year). All the subjects were from the health district of the Alcala´ University Hospital (Alcala´ de Henares, Madrid, Spain). The premenopausal and perimenopausal women had visited the clinic of the Rheumatology Department for nonspecific pain for which no organic cause was found. The postmenopausal women were referred by a cohort of general physicians for routine evaluation of the risk of suffering osteoporosis, following a protocol used in this public health district. All the subjects gave written informed consent. The Office for Protection from Research Risks of the University of Alcala´ approved the study. Prior to enrolling each candidate, a complete medical history was compiled and a physical examination was made. In all the women, normality was established by an interview and biochemical measurements. Also, radiologic imaging of the thoracic and lumbar spine excluded vertebral deformities, defined as the loss of more than 25% of the height of the anterior, middle or posterior vertebral body compared with normal reference values in subjects matched for age and sex [13]. The women participating in the study had not taken any medication that could interfere with calcium metabolism (e.g., corticosteroids, vitamin A and D, anticonvulsants, thyroid hormones, lithium, fluoride, zinc, calcium, heparin and/or contraceptives) for at least 4 years before entering the study, or for more than 3 months before that time. All led active lives, but none practiced any sports. Only 5% of the women smoked, but none more than 10 cigarettes/day. Alcohol intake among the participants was sporadic, and coffee intake did not exceed 100 ml/day in any case. None of the women followed a special diet. In all the groups, height was measured using a Harpenden stadiometer with a mandible plane parallel to
Bone Mass Study As in previous studies [14], the first full-body bone densitometry was carried out in all the women at baseline and the second 1 year or more later. Studies lasted an average of 21 min and were done in the decubitus position and at times not near to food intake. Dual-energy X-ray absorptiometry (DXA) was the system used (Norland XR-26, Norland, Fort Atkinson, WI). In our laboratory this system has a coefficient of variation (CV) of = 1.3% in the assessment in vivo of total (TBBMC) and regional bone mass (head, arm, trunk and leg BMC). These values were calculated from three or four measurements made in 6 women over 10 weeks. The CV for weight was 0.6%. The XR-26 was calibrated daily with a specific calibrator supplied by Norland.
Laboratory Studies The participants were not permitted to smoke, to drink coffee, tea or alcohol, or to exercise for 24 h before the day of investigation. Urine samples were collected in the morning after an overnight fast. Hematologic and biochemical studies were done on venous blood samples obtained in a fasting state at 0800 hours. Blood samples were centrifuged and the serum was stored at –208C until analyzed. Biochemical measurements were done of blood glucose, transaminases, g-glutamyltransferase (GGT), serum creatinine (SCr), calcium (SCa), phosphate (SP), total proteins (STPr), bilirubin, total alkaline phosphatase (TAlPh) and tartrate-resistant acid phosphatase (TRAP). A coagulation study was done. In all cases serum calcium was corrected for proteins (CSCa) according Parfitt [15]. Biochemical parameters were measured in serum using a BM/Hitachi 717 automated analyzer system (Boehringer, Mannheim, Germany). TRAP was quantitated in serum in the Hitachi automated analyzer as the substrate a-naphthyl phosphate, using a reagent from Boehringer Laboratories (Boehringer, Mannheim, Germany) that reacts specifically with isoenzyme 5b synthesized by the osteoclasts [16]. A biochemical study was done on 24-h urine to confirm the normality of calcium excretion. Twenty-four hour urinary calcium excretion (UCa) was determined by atomic absorption spectroscopy using a Perkin Elmer model 5000 spectrophotometer (Perkin Elmer, Norfolk, CT); renal tubular reabsorption of phosphate was quantitated in 24-h urine (%TRPh). Serum estrone
Bone, Gonadal Status and BMI
(SE1) and serum estradiol (SE2) were determined around day 8–10 of the menstrual cycle in premenopausal women and 8–10 days after the onset of menstruation in perimenopausal women. FSH was determined with SE1 and SE2 in perimenopausal and postmenopausal women. In all the women, these determinations were done with the other biochemical determinations. SE1 and SE2 were measured in samples using a 1230 Arcus fluorimeter (LKB, Turku, Finland). In our laboratory these biochemical determinations have intra-assay and interassay coefficients of variation of less than 6% for a concentration of 25 pg/ml. All samples from each woman were analyzed in the same assay to eliminate interassay variation. Assay reproducibility was determined by testing four samples five times in five different runs. For TRAP, the intraassay CV was 4.5% for a concentration of 66 nkat/l–1.
Statistical Studies All results were expressed as mean SD. The normal distribution of data was confirmed by calculating skew and kurtosis before applying standard tests. The continuous variables studied in each group (defined by the nominal variables of gonadal status (premenopausal, perimenopausal and postmenopausal) and BMI (thin BMI 520 kg/m2; normal, BMI 20–25 kg/m2; overweight, BMI 25–30 kg/m2; and obese, BMI 430 kg/ m2)) were compared by analysis of variance to determine the effects of the nominal variables with a post-hoc test of differences between groups using the Bonferroni/ Dunn test. A minimum p value of 50.05 was the necessary condition for statistical significance. Simple and multiple regression analysis was used to examine the relationships between the continuous variables. Correlation, and partial correlation, adjusted for confounding variables, were used to examine the relationships between continuous variables. In the correlation and regression tests, a p value of 50.05 was the necessary condition for statistical significance. Data were processed on a Macintosh computer using the StatView 4.02 statistical package (Abacus Concepts, Berkeley, CA).
381 Table 1. Number and characteristics of the overall group of women stratified by gonadal status Premenopausal Perimenopausal Postmenopausal (n = 126) (n = 76) (n = 171) Age (years) Height (m) Weight (kg) %IBWeight BMI (kg/m2 SCr (mmol/l) SP (mmol/l) CSCa (mmol/l) STPr (g/l) TRAP (nkat/l) TAlPh (mkat/l) SE1 (pmol/l)d SE2 (pmol/l)d UCa (mmol/24 h) %TRPh TBBMC (g) Head BMC (g) Arms BMC (g) Trunk BMC (g) Legs BMC (g)
34 1.60 60.9 107 23.7 80 1.17 2.34 72.3 55.7 2.3 331 507 3.9 86 2515 472 352 896 821
+
7 0.05ab 15.7 28 6.2 8a 0.13 0.10c 2.8 6.7ab 0.4ab 99 123 0.8 3 401ab 83a 68ab 167ab 186ab
48 1.57 61.3 112 24.8 81 1.18 2.33 71.7 58.4 2.4 258 351 4.0 85 2364 441 326 827 769
2 0.04 10.4 21 4.5 7c 0.11 0.08c 2.1 6.1c 0.5c 49 76 0.8 3 268a 125 46 106c 92c
59 1.56 60.6 113 24.9 85 1.15 2.37 72.2 64.5 2.7 88 53 4.1 86 2170 405 307 752 700
9 0.06 10.5 22 4.5 10 0.12 0.10 2.9 7.0 0.4 15 15 1.1 4 257 106 52 125 90
Percent of ideal body weight (%IBWeight), body mass index (BMI), serum creatinine (SCr), serum phosphate (SP), corrected serum calcium (CSCa), serum total proteins (STPr), tartrate-resistant acid phosphatase (TRAP), total alkaline phosphatase (TAlPh), serum estrone (SE1), serum estradiol (SE2), urinary calcium (Uca), percent tubular phosphate reabsorption (%TRPh) and total-body (TBBMC) and regional bone mineral content (BMC). a p50.0001 versus postmenopausal women; bp50.005 versus perimenopausal women; cp50.005 versus postmenopausal women; d p50.0001 between all; according to ANOVA with post-hoc Bonferroni/Dunn test.
BMC and leg BMC differed in all three groups (p 50.005 to 50.0001). Of the biochemical variables, SCr and ScaC were higher in postmenopausal women. TRAP and TAlPh behaved similarly, being higher in postmenopausal women than in premenopausal and perimenopausal women (p50.0001 in both), and higher in perimenopausal women than in premenopausal women (p50.005).
Results
(ii) Women Stratified by Gonadal Status and BMI
Differences among the Variables Studied
Premenopausal (Table 2). SP was significantly higher in thin women than in the other women. SP was higher in women of normal weight than in overweight and obese women (p50.0005 in both). The only significant difference in CSCa concentration was between thin and obese women (p50.005). TRAP did not differ between thin and normal-weight women, and in both groups it was greater than in overweight and obese women (all p50.0001). TAlPh was higher in thin women than in the other groups (p50.0001 to 50.005). SE2 showed the same differences as TRAP. UCa and %TRPh differed only between normal-weight and obese women (p50.0005).
(i) Women Stratified by Gonadal Status Alone (Table 1) ANOVA with post-hoc Bonferroni/Dunn confirmed that TBBMC was greater in premenopausal women than in perimenopausal (p50.0005) and postmenopausal women (p 50.0001); TBBMC was also greater in perimenopausal women than in postmenopausal women (p50.0001). Head BMC differed significantly only between premenopausal and postmenopausal women (p50.0001). Arm BMC was lower in perimenopausal and postmenopausal women than in premenopausal women (p50.0005 and 50.0001, respectively). Trunk
382
H. Rico et al.
Table 2. Number and characteristics of the premenopausal women stratified by BMI Premenopausal women (n = 126)
Age (years) Height (m) Weight (kg) %IBWeight2 BMI (kg/m ) SCr (mmol/l) SP (mmol/l) CSCa (mmol/l) STPr (g/l) TRAP (nkat/l) TAlPh (mkat/l) SE1 (pmol/l) SE2 (pmol/l) UCa (mmol/24 h) %TRPh TBBMC (g) Head BMC (g) Arms BMC (g) Trunk BMC (g) Legs BMC (g)
Thin (n = 31)
Normal (n = 64)
Overweight (n = 15)
Obese (n = 16)
30 1.62 48.2 82 18.3 78 1.26 2.36 72.2 58.6 2.5 254 479 4.2 86 2239 462 306 797 711
35 1.60 56.6 99 22.1 80 1.17 2.34 72.8 57.1 2.3 311 467 4.6 87 2432 465 341 862 769
38 1.59 70.0 125 27.8 82 1.07 2.33 72.6 49.4 2.1 401 615 3.2 85 2776 488 367 995 925
36 1.59 94.0 166 36.9 78 1.08 2.28 73.4 50.1 1.9 484 619 3.2 84 3140 498 458 1090 1093
7 0.05 4.0 6 1.4 8 ab 0.11 0.11d 3.1b 5.4ab 0.3ab 56 b 107b 0.3 2 b 295 72ab 41 b 121b 103
7 0.05 4.4 6 1.3 9 c 0.12 0.10 3.1b 5.8b 0.3b 82 b 103b 0.9 d 3 b 247 90e 44 e 129c 105
4 0.04 5.6 3 1.0 7 0.09 0.10 2.0 5.8 0.2d 68 106 0.3 3 d 107 87e 66a 76d 73
8 0.05 16.0 27 6.1 7 0.03 0.06 2.9 6.9 0.2 35 124 0.8 4 479 69 68 223 284
See Table 1 for abbreviations. p50.0005 versus normal-weight women; bp50.0001 versus overweight and obese women; cp50.0005 versus overweight and obese women; d p50.005 versus obese women; ep50.0001 versus obese women; according to ANOVA with post-hoc Bonferroni/Dunn test. a
TBBMC was lower in thin and normal-weight women than in overweight and obese women (p50.0001 in both). TBBMC also differed significantly between overweight and obese women (p50.0005). There were no differences in head BMC between groups. Arm BMC differed in all women versus obese women (all p50.0001). Trunk BMC and leg BMC showed differences similar to TRAP.
Perimenopausal Women (Table 3). SP was higher in thin and normal-weight women than in overweight and obese women (all p50.0001). TRAP and TAlPh showed the same differences. SE1 was lower in thin and normalweight women than in overweight and obese women (p50.0001) and higher in obese women than in overweight women (p50.005). SE2 was lower in thin women than in overweight women (p50.0001), and higher in overweight women than in normal-weight and obese women (p50.0005). UCa was lower in obese women than in the other groups (p50.0001); %TRPh was lower in obese women than in thin (p50.0005) and normal-weight women (p50.0001). The differences between groups in TBBMC were significant but, paradoxically, perimenopausal obese women had a lower TBBMC than overweight women (p50.0005). The differences between groups in arm BMC were similar to those observed for TBBMC. The trunk BMC and leg BMC of the thin women were lower than in the other groups (p50.0005 to
50.0001), but did not differ between overweight and obese women. Postmenopausal women (Table 4). SP was higher in thin and normal-weight women than in overweight and obese women (p50.0001 in both). CSCa was higher only in thin women than in overweight or obese women (p50.0005 in both). TRAP, TAlPh, UCa and %TRPh were higher in thin and normal-weight women than in overweight and obese women (p50.0001 in both). %TRPh was higher in overweight women than in obese women (p50.005). SE1 and SE2 were lower in thin and normal-weight women (p50.0001), and SE1 was lower in overweight women than in obese women (p50.005). Thin and normal-weight women had a smaller TBBMC than overweight and obese women (p50.0005 in both). Paradoxically, head BMC was greater in thin women than in overweight and obese women (p50.0005). Arm BMC showed differences similar to TBBMC. Trunk BMC and legs BMC was lower in thin and normal-weight women versus overweight and obese women (p50.0005).
Correlation Studies In the overall group of women, correlation studies revealed a significant positive correlation between age and BMI (r = 0.14, p50.005, 95%CI (CI) 0.04 to 0.24), SCr (r = 0.20, p50.0005, CI 0.10 to 0.30), CSCa (r =
Bone, Gonadal Status and BMI
383
Table 3. Number and characteristics of the perimenopausal women stratified by BMI Perimenopausal women (n = 76)
Age (years) Height (m) Weight (kg) %IBWeight2 BMI (kg/m ) SCr (mmol/l) SP (mmol/l) CSCa (mmol/l) STPr (g/l) TRAP (nkat/l) TAlPh (mkat/l) SE1 (pmol/l) SE2 (pmol/l) UCa (mmol/24 h) %TRPh TBBMC (g) Head BMC (g) Arms BMC (g) Trunk BMC (g) Legs BMC (g)
Thin (n = 15)
Normal (n = 31)
Overweight (n = 15)
Obese (n = 15)
47 1.57 47.8 87 19.2 82 1.24 2.36 71.1 62.4 2.9 209 309 4.4 86 2061 405 274 738 643
49 1.59 58.7 105 23.3 83 1.23 2.31 71.4 61.8 2.5 246 353 4.2 87 2350 461 320 820 756
49 1.56 65.3 121 26.7 79 1.10 2.33 73.1 52.7 2.1 278 425 4.2 85 2681 537 383 928 838
49 1.56 77.1 146 32.0 76 1.07 2.36 71.7 51.8 2.1 316 329 3.2 82 2443 369 348 857 868
1 0.02 1.4 2 0.4 5 a 0.08 0.08 1.6a 4.2a 0.4a 24d 44 0.4 3 f 150 59f 27f 64f 35
2 0.04 4.1 6 1.3 8 a 0.06 0.07 2.1a 3.6bc 0.4a 34 70 0.7 2 243 129 29 i 114a 58
2 0.03 1.5 8 1.3 10 0.13 0.08 1.5 4.7 0.3 23e 75 0.3 4 161eh 130a 46 98 69
2 0.05 5.0 9 1.6 5 0.08 0.10 2.7 4.0 0.5 i 44 75 f 0.8 g 4 169 116 21 85 71
See Table 1 for abbreviations. p50.0001 versus overweight and obese women; bp50.0005 versus overweight and obese women; cp 5 0.005 versus thin women; dp50.0001 versus overweight women; e p50.0005 versus normal and obese women; fp50.0001 versus others; gp50.0001 versus thin and normal-weight women; hp50.0005 versus thin and obese women; i p50.005 versus overweight women; according to ANOVA with post-hoc Bonferroni/Dunn test.
a
Table 4. Number and characteristics of the postmenopausal women stratified by BMI Postmenopausal women (n = 171)
Age (years) YSM Height (m) Weight (k)g %IBWeight2 BMI (kg/m ) SCr (mmol/l) SP (mmol/l) CSCa (mmol/l) STPr (g/l) TRAP (nkat/l) TAlPh (mkat/l) SE1 (pmol/l) SE2 (pmol/l) UCa (mmol/24 h) %TRPh TBBMC (g) Head BMC (g) Arm BMC (g) Trunk BMC (g) Leg BMC (g)
Thin (n = 19)
Normal (n = 81)
Overweight (n = 56)
Obesity (n = 15)
55 6 1.58 46.2 83 18.4 87 1.25 2.44 71.5 68.7 3.0 74 45 4.5 88 2065 460 276 704 623
60 12 1.57 56.9 104 22.9 85 1.18 2.37 72.4 66.5 2.8 84 49 4.5 87 2099 416 291 713 676
59 10 1.55 65.2 124 27.0 83 1.10 2.35 72.1 61.7 2.5 93 61 4.3 85 2260 381 328 800 744
58 12 1.53 82.3 163 35.2 87 1.04 2.33 72.1 59.5 2.3 107 61 4.3 81 2349 358 368 869 774
5 5 0.06 3.3 6 1.1 8 a 0.08 0.12b 1.8a 5.9a 0.2a 10a 11 a 0.6 a 3 b 136 b 83a 54b 61a 45
10 9 0.06 4.4 7 1.3 8 a 0.12 0.09 3.1a 6.8a 0.3a 11a 14 a 1.2 a 4 b 279 92a 43 a 118 a 91
8 7 0.06 5.4 7 1.4 11 0.09 0.10 3.1 6.0 0.3c 15 15 1.0 c 3 216 116 c 41 124 64
7 8 0.05 12.6 21 4.6 12 0.12 0.12 2.4 5.9 0.4 15 8 1.0 3 196 111 81 108 83
See Table 1 for abbreviations. p50.0001 versus overweight and obese women; bp50.0005 versus overweight and obese women; cp50.005 versuss obese women; according to ANOVA with post-hoc Bonferroni/Dunn test.
a
384
0.14, p50.05, CI 0.03 to 0.24), TRAP (r = 0.42, p50.0001, CI 0.33 to 0.51) and TAlPh (r = 0.33, p50.0001, CI 0.23 to 0.42). There was a significant negative correlation between age and height (r = –0.37, p50.0001, CI –0.45 to –0.27), TBBMC (r = –0.44, p50.0001, CI –0.52 to –0.35), regional BMC (r = –0.27 to –.40, p50.0001 in all, CI –0.48 to –0.22), SP (r = 70.14, p50.05, CI –0.24 to –0.04) and, naturally, SE1 (r = –0.73, p50.0001, CI –0.78 to –0.68) and SE2 (r = 70.66, p50.0001, CI –0.71 to –0.60). TBBMC and regional BMC (except head BMC) (r = 0.67 to 0.51, p50.0001 in all, CI 0.61 to 0.59), SE1 (r = 0.29, p50.0001, CI 0.19 to 0.38) and SE2 (r = 0.11, p50.05, CI 0.01 to 0.21) were weightdependent. There was a significant negative correlation between weight and SP (r = –0.40, p50.0001, CI –0.49 to –0.31), CSCa (r = –0.19, p50.005, CI –0.29 to –0.08), TRAP (r = –0.39, p50.0001, CI –0.48 to –0.30), TAlPh (r = –0.40, p50.0001, CI –0.48 to –0.31), UCa (r = 70.41, p50.0001, CI –0.49 to –0.32) and %TRPh (r = –0.33, p50.0001, CI –0.42 to –0.23). TBBMC and regional BMC correlated significantly with SE1 (r = 0.61 to 0.25, p50.0001, CI 0.17 to 0.67) and SE2 (r = 0.55 to 0.29, p50.0001, CI 0.19 to 0.62). The correlation remained positive, but lower, for SE1 (r = 0.26, p50.0001) and SE2 (r = 0.28, p50.0001) when a partial correlation was done using age and weight as confounding variables. There was a significant positive correlation between TRAP and TAlPh (r = 0.53, p50.0001, CI 0.45 to 0.60) and significant negative correlations between TRAP and SE1 (r = –0.62, p50.0001, CI –0.68 to –0.55) and SE2 (r = –0.59, p50.0001, CI –0.65 to –0.62), and between TAlPh and SE1 (r = –0.51, p50.0001, CI –0.58 to 70.43) and SE2 (r = –0.46, p50.0001, CI –0.54 to 70.38). In the overall group of the women, there was a significant positive correlation between BMI and SE1 (r = 0.20, p5 0.0001, CI 0.10 to 0.30), and between SE1 and SE2 (r = 0.75, p50.0001, CI 0.66 to 0.82). According to gonadal status, there was a positive correlation BMI and SE1 (r = 0.39, p50.0001, CI 0.23 to 0.53) and SE2 (r = 0.65, p 50.0001, CI 0.54 to 0.74) in premenopausal women. In perimenopausal women there was a significant positive correlation only with SE1 (r = 0.67, p50.0001, CI 0.51 to 0.78), and in postmenopausdal women there was a significant positive correlation between BMI and SE2 (r = 0.31, p50.0001, CI 0.17 to 0.44) and with SE1 (r = 0.49, p50.0001, CI 0.37 to 0.60).
Regression Studies (Table 5, 6) The simple regression coefficients of the variables that showed significant correlations are listed in Table 5. In the overall group of the women bone loss was two-fold greater in non-weight-bearing bone (ß = –0.06 in arms) than in weight-bearing bone (ß = –0.03 in trunk and legs). The influence of BMI was greater in non-weight-
H. Rico et al. Table 5. Regression coefficients of age and BMI in relation to variables with which they are significantly correlated in the overall group of women studied
Age vs
BMI vs
Height TBBMC Head BMC Arm BMC Trunk BMC Leg BMC SCr SP SCa TRAP TAlPh SE1 SE2 TBBMC Arm BMC Trunk BMC Leg BMC SP SCa TRAP TAlPh SE1 SE2 UCa %TRPh
ß
p
95% CI
–89.0 –0.02 –0.04 –0.06 –0.03 –0.03 0.32 –15.5 19.9 0.73 10.1 –0.07 –0.04 0.01 0.07 0.03 0.03 20.6 10.4 0.4 9.3 0.09 0.05 0.56 0.28
50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.05 50.005 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001 50.0001
–112 to –65 –0.02 to –0.01 –0.05 to –0.03 –0.08 to –0.04 –0.04 to –0.03 –.04 to –0.02 0.18 to 0.46 –26.7 to –4.3 6.2 to 33.7 0.57 to 0.88 7.2 to 13.0 –0.08 to –0.06 –0.05 to –0.04 0.01 to 0.01 0.07 to 0.08 0.03 to 0.03 0.03 to 0.03 20.1 to 21.2 10.1 to 10.6 0.38 to 0.41 8.9 to 9.6 0.08 to 0.09 0.05 to 0.06 0.54 to 0.58 0.28 to 0.29
See Table 1 for abbreviations.
bearing (ß = 0.07 in arms) than in weight-bearing bone (ß = 0.03 in trunk and legs). The influence of age was greater for SE1 than for SE2 (ß = –0.07 vs –0.04); the influence of BMI also was greater for SE1 than for SE2 (ß = 0.09 vs 0.05). Age and BMI had inverse effects on TBBMC and regional bone mass. According to gonadal status, only in postmenopausal women was the bone loss greater in non-weight-bearing bone (ß = –0.05 in arms) than in weight-bearing bone (ß = –0.01 in legs). In multiple regression analysis of TBBMC representing bone mass as the dependent variable and age, BMI, SE1, SE2, TRAP and TalPh as independent variables (Table 6), in the overall group of women, TBBMC was inversely dependent on age (ß = –6.8, p 5 0.0001, CI – 9.6 to –3.9) and positively dependent on BMI (ß = 31.9, p 50.0001, CI 25.4 to 38.3) and SE2 (ß = 0.37, p 5 0.05, CI 0.02 to 0.76). The measures of the regional BMC (arms, trunk and legs) have similar results to TBBMC. In premenopausal women TBBMC was only influenced by BMI (ß = 48.9, P 5 0.0001, CI 38.9 to 59.0). In perimenopausal women, BMI was the most significant factor (ß = 31.0, p 5 0.0005, CI 24.7 to 32.3) and SE2 and TalPh less significant (ß = 1.0, p 5 0.05, CI 0.2 to 1.9 and ß = –156.2, p 5 0.05, CI –288 to –24 respectively). In postmenopausal women, TBBMC was influenced only by BMI (ß = 14.1, p 5 0.005, CI 5.0 to 23.3, respectively).
Bone, Gonadal Status and BMI
385
Table 6. Multiple regression analysis between TBBMC as dependent variable versus age, BMI, serum estrone (SE1, pmol/l) and serum estradiol (SE2, pmol/l) in the overall group of women studied, and according to gonadal status Coefficient
SE
t value
p value
95% CI
1929.3 –6.8 31.8 0.37
206.4 1.4 3.2 0.19
9.3 –4.6 9.6 1.8
50.0001 50.0001 50.0001 50.05
1523 to 2335 –9.6 to –3.9 25.4 to 38.3 –0.02 to 0.76
Premenopausal women Intercept BMI (kg/m2)
1763.7 48.9
347.7 5.1
5.0 9.6
50.0001 50.0001
1075 to 2452 38.9 to 59.0
Perimenopausal women Intercept BMI (kg/m2) TalPh (mkat/l) SE2 (pmol/l)
1773.2 31.0 –156.2 1.0
730.2 2.9 66.1 0.4
2.4 8.7 –2.3 2.4
50.05 50.005 50.05 50.05
315 to 3231 24.7 to 32.3 –288 to –24 0.2 to 1.9
Postmenopausal women Intercept Age (years) BMI (kg/m2)
2457.7 –10.8 14.4
314.5 1.8 4.6
7.8 –5.9 3.1
50.0001 50.0001 50.005
1836 to 3078 –14.3 to –7.2 5.2 to 23.6
All women Intercept Age (years) BMI (kg/m2) SE2 (pmol/l)
Only significant regressions are shown.
Discussion Our results for the overall group of women stratified by gonadal status concur with reports that bone mass differs significantly in premenopausal, perimenopausal and postmenopausal women [5,6], and that postmenopausal women have increased levels of bone remodeling markers [2,3]. On the other hand, with regard to the influence of BMI and/or estrogens on bone mass, multiple regression analysis indicated that in premenopausal and perimenopausal women, BMI was more influential than estrogens, and that in postmenopausal women BMI continued to be an important positive factor for bone mass, and age and years since menopause was a negative factor. The influence of BMI or weight as a major determinant of bone mass has been well documented [14,17,18]. Stratification of postmenopausal women by weight 115% or more above ideal body weight [17] has demonstrated that heavier women have significantly greater BMC in the spine, hip and radius than normal-weight women. In a study [18] of obese perimenopausal and postmenopausal women and agematched women of normal weight, obesity was defined as body weight more than 10% over normal weight. Obese postmenopausal women had a significantly higher vertebral BMD than non-obese women. The importance of weight for bone mass is indicated by the fact that the effect of weight is greater in weight-bearing bones than in non-weight-bearing bones [7,8]. This was confirmed by our study, in which regression analysis showed that bone loss was two-fold greater in non-weight-bearing bones than in weight-bearing bones and that by multiple regression analysis the effect of BMI was positive in all women and according to gonadal status.
The correlation between gonadal steroids and bone mass is debated. In contrast with the well-documented protective effect exogenous estrogens have against bone loss and fracture, less is known about the effect on bone of endogenous estrogens [19]. One study reported a positive association between radial BMC and estrogen excretion [20]. Two studies that failed to report an association between serum estrogens and bone mass had a small sample size and insufficient power [21,22]. In our study, we observed a significant correlation between TBBMC and regional BMC and serum gonadal steroid concentrations, a correlation that persisted, although it was marginal, when age and BMI were taken into consideration as confounding variables. It is more interesting to examine findings in women stratified by gonadal status and BMI. In women stratified by gonadal status, thin women showed more significant differences compared with obese women; in thin women, TBBMC and regional BMC were lower, gonadal steroid levels were lower, and bone remodeling markers were higher. It is possible that calcium and vitamin D homeostasis is disturbed in obese women, resulting in reduced calcium, increased parathyroid activity, reduced 25hydroxyvitamin D and increased 1,25-dihydroxyvitamin D [19]. Bell et al. [23] assert than the vitamin D– endocrine system is altered in obese subjects, with secondary hyperparathyroidism, enhanced renal tubular reabsorption of calcium and increased circulating 1,25dihydroxyvitamin D. On the other hand, histomorphometric studies of obese subjects indicate the possible existence of secondary hyperparathyroidism [24] with increased tubular calcium reabsorption and decreased %TRPh [25]. These observations were confirmed by our study. In thin and normal-weight premenopausal women,
386
SP levels are greater than in overweight and obese women, and renal calcium elimination is enhanced, both findings indicative of parathyroid hyperactivity [25]. These disturbances were also observed in perimenopausal women, in whom SP and UCa elimination were lower in the obese group. Identical results were reported in postmenopausal women, suggesting parathyroid hormone (PTH)-dependent hyperactivity in obese women regardless of gonadal status. It is paradoxical that the obese and overweight women had findings suggestive of enhanced parathyroid activity with increased TBBMC and regional BMC. PTH affects cortical bone preferentially but our findings indicate that TBBMC, constituted by 80% cortical bone, and total legs, also largely cortical bone, have increased bone mass. This suggests that mechanical stimuli and/or higher estrogen levels have more effect on bone mass than PTH. In fact, obese women have less postmenopausal loss of bone mass than thin women [17,18,26]. Our results also indicate a significant correlation between TRAP and TAlPh, which is logical given the coupling of these enzymes in bone remodeling [27]. We also found evidence, confirming the findings of earlier studies [28], of an inverse correlation between TRAP and SE1 and SE2 and between these parameters and TAlPh. This is also logical and explains why there is less bone remodeling at higher estrogen levels. On the other hand, BMI showed a negative correlation with TRAP and TAlPh, and a positive and more significant correlation with SE1 (p50.0001) than with SE2, which had an almost marginal correlation (p50.05). This indicates more peripheral conversion of estrone than estradiol. In conclusion, our findings indicate that BMI, rather than gonadal steroid concentrations, is the main determinant of bone mass in women, regardless of gonadal status, and that age and years since menopause in postmenopausal women have a negative effect on bone mass. They also indicate that overweight and obese women have biochemical findings indicative of enhanced parathyroid activity, although they have more bone mass. Given the influence of BMI on bone mass, it can be deduced that the mechanical effect of weight on the skeleton and adiposity, which favors the conversion of androstenedione to the metabolically active estrogen, estrone, is more important than any parathyroid hyperactivity that may exist. This also explains why the loss of bone mass with age in weightbearing bones is only half that observed in non-weightbearing bones.
References 1. Lindsay R. The burden of osteoporosis: cost. Am J Med 1995;98 (Suppl 2A):S9–11. 2. Hassager C, Christiansen C. Measurement of bone mineral density. Calcif Tissue Int 1995;57:1–5. 3. Bikle DD. Biochemical markers in the assessment of bone disease. Am J Med 1997;103:427–36. 4. Soda MY, Mizunuma H, Honjo SI, Okano H, Ibuki Y, Igarashi M.
H. Rico et al.
5. 6.
7.
8. 9.
10. 11. 12. 13. 14.
15. 16. 17.
18.
19.
20. 21.
22. 23. 24. 25.
Premenopausal and postmenopausal bone mineral density of the spine and proximal femur in Japanese women assessed by dualenergy X-ray absorptiometry: a cross-sectional study. J Bone Miner Res 1993;8:183–9. Rico H, Revilla M, Herna´ndez ER, Villa LF, Alvarez de Buergo M. Total and regional bone mineral content in relation to menopause. Maturitas 1992;15:233–40. Rico H, Herna´ndez ER, Seco C, Villa LF, Ferna´ndez S. Quantitative peripheral computed tomodensitometric study of cortical and trabecular bone mass in relation with menopause. Maturitas 1994;18:183–9. Felson DT, Zhang YQ, Hannan MT, Anderson JJ. Effects of weight and body mass index on bone mineral density in men and women: the Framingham study. J Bone Miner Res 1993;8:567– 73. Edelstein SL, Barrett Connor E. Relation between body size and bone mineral density in elderly men and women. Am J Epidemiol 1993;138:160–9. MacDonald PC, Edman CD, Hemsell DC, Porter JC, Siiteri PR. Effect of obesity on conversion of plasma androstenedione to estrone in postmenopausal women with and without endometrial cancer. Am J Obstet Gynecol 1978;130:448–50. MacDonald PC, Siiteri PK. Relationship between extraglandular production of estrone and the occurrence of endometrial neoplasm. Gynecol Oncol 1974;2:2159–63. Longcope CL. Methods and results of aromatization studies in vivo. Cancer Res 1982;42:3307–11. Grodin JM, Siiteri OK, McDonald PC. Source of oestrogen production in postmenopausal women. J Clin Endocrinol Metab 1973;36:207–14. Rico H, Fraile E, Revilla M, Aguado F, Villa LF, Martin FJ. Vertebral and metacarpal morphometry as indicators of generational improvement. Clin Rheumatol 1997;16:279–83. Rico H, Revilla M, Villa LF, Alvarez de Buergo M, RuizContreras D. Determinants of total and regional bone mineral content and density in postpubertal normal women. Metabolism 1994;43:263–6. Parfitt AM. Chlorothiazide induced hypercalcemia in juvenile osteoporosis and hyperparathyroidism. N Engl J Med 1969;281:55–9. Rico H, Villa LF. Serum tartrate-resistant acid phosphatase as a biochemical marker of bone remodeling. Calcif Tissue Int 1993;52:149–50. Dawson-Hughes B, Dhipp C, Sadowski L, Dallal G. Bone density of the radius, spine, and hip in relation to percent of ideal body weight in postmenopausal women. Calcif Tissue Int 1987;40:310–4. Ribot C, Tremollieres F, Pouilles JM, Bonneu M, Germain F, Louvet JP. Obesity and postmenopausal bone loss: the influence of obesity on vertebral density and bone turnover in postmenopausal women. Bone 1987;8:327–31. Cauley JA, Salamone LM, Lucas FL. Postmenopausal endogenous and exogenous hormones, degree of obesity, thiazide diuretics, and risk of osteoporosis. In: Marcus R, Feldman D, Kelsey J, editors. Osteoporosis. San Diego: Academic Press, 1996:551–76. Rae MH, Mole PA, Paterson CR. Endogenous factors affecting bone mineral content in post-menopausal women. Maturitas 1991;13:319–24. Deutsch S, Benjamin F, Seltzer V, Tafreshi M, Kocheril G, Frank A. The correlation of serum estrogens and androgens with bone density in the late postmenopause. Int J Gynaecol Obstet 1987;25:217–22. Nelson DA, Kleerekoper M, Parfitt AM. Bone mass, skin color and body size among black and white women. Bone Miner 1988;4:257–64. Bell NH, Epstein S, Greene A, Shary J, Oexmann MJ, Shaw S. Evidence for alteration of the vitamin D–endocrine system in obese subjects. J Clin Invest 1985;76:370–3. Compston JE, Vedi S, Ledger JE, Webb A, Gazet JC, Pilkington TR. Vitamin D status and bone histomorphometry in gross obesity. Am J Clin Nutr 1981;34:2359–63. Fitzpatrick LA, Bilezikian JP. Actions of parathyroid hormone.
Bone, Gonadal Status and BMI In. Bilezikian JP, Raisz LG, Rodan GA, editors. Principles of bone biology. San Diego: Academic Press, 1996:339–46. 26. Pocock N, Eisman J, Gwinn T, Sambrook P, Kelly P, Freund J, et al. Muscle strength, physical fitness, and weight but not age predict femoral neck bone mass. J Bone Miner Res 1989;4:441–8.
387 27. Risteli L, Risteli J. Biochemical markers of bone metabolism. Ann Med 1993;25:385–93. 28. Rico H, Arnanz F, Revilla M, Perera S, Iritia M, Villa LF, et al. Total and regional bone mineral content in women treated with GnRH agonists. Calcif Tissue Int 1993;52:354–7.
Received for publication 6 July 2001 Accepted in revised form 15 November 2001