The Influence of Pubertal Timing on Bone Mass Acquisition: A Predetermined Trajectory Detectable Five Years before Menarche

Background: Later menarcheal age (MENA) is a risk factor for osteoporosis. It is associated with low peak bone mass (PBM). Like PBM, MENA is under strong genetic influence. We hypothesized that MENA-related bone mass differences could be predetermined before puberty.

Methods: We tested this hypothesis in 124 healthy subjects followed from age 7.9 to 20.4 yr with dual-energy x-ray absorptiometry assessment at mean ages of 8.9, 10.0, 12.4, and 16.4 yr. Six sites were measured: radial metaphysis, radial diaphysis, femoral neck, trochanter, femoral diaphysis, and L2–L4. Mean MENA (±sd) was 13.0 ± 1.2 yr. The cohort was segregated by the median of MENA into LATER (14.0 ± 0.7 yr) and EARLIER (12.1 ± 0.7 yr) subgroups.

Results: At 20.4 ± 0.6 yr, areal bone mineral density (aBMD) was lower in the LATER than the EARLIER subgroup at all six sites, with a mean difference of −0.31 Z-score (P = 0.022). Lower Z-scores in the LATER than in the EARLIER subgroup were observed at all sites at mean ages of 10.0, 12.4, and 16.4 yr, and before pubertal maturation, i.e. at 8.9 yr with a mean Z-score difference of −0.34 (P = 0.016). From mean age 8.9 to 20.4 yr, aBMD gains of all sites were similar in LATER and EARLIER subgroups, with mean of +301 and +308 mg/cm² (P = 0.402), respectively.

Conclusions: In prepubertal girls who will experience later menarche, a deficit in aBMD can already be observed before the onset of pubertal maturation, with no further accumulated deficit until PBM compared to girls with earlier menarche. This suggests that shorter estrogen exposure from prepuberty to PBM is not the main factor for increased osteoporosis risk associated with later menarche. Rather common genetic determinants of low bone mass and later puberty could be involved.

In prepubertal girls who will experience later menarche, a deficit in areal bone mineral density is already observed before the onset of pubertal maturation, with no further accumulated deficit until peak bone mass compared to girls with earlier menarche.

The relationship during adult life between the risk of osteoporosis and pubertal timing has been mostly documented in female subjects (1–10), although a recent report strongly suggests that such an association is also present in human males (11). In postmenopausal women, later age at menarche was associated with lower areal bone mineral density (aBMD) at several skeletal sites, including forearm, spine, and proximal femur (1–5, 8). It is also associated with increased risk of forearm, vertebral, and hip fractures (6, 7, 9, 10). The inverse relationship between aBMD and age at menarche has also been documented in retrospective observations in premenopausal women (1, 5, 12). The influence of menarcheal age is not negligible because epidemiological studies indicate that for the same reduced lifetime exposure to estrogen, fracture risk at the proximal femur, spine, and forearm would be greater with late menarche than earlier menopause (9, 13, 14).

Recent studies indicate that by the end of the growth period, the influence of menarcheal age is expressed not only by difference in aBMD as measured by dual-energy x-ray absorptiometry (DXA), but also by alterations in microstructure components of the distal radius and tibia as assessed by high-resolution peripheral computerized tomography (15, 16). Thus, in the same cohort as that presented here, a previous study examining distal radius microstructure at the age of 20.4 yr documented that later menarcheal age was associated with significant lower cortical volumetric density and thickness, a finding compatible with less endocortical accrual during growth (15). In a still more recent report, menarcheal age-related microstructural alterations were also observed in the distal tibia of the 20.4-yr-old group (16). In this latter study, the negative influence of later menarcheal age was documented in a group of 120 premenopausal women aged 45.8 yr on both femoral neck aBMD and distal tibia microstructure, including lower total and trabecular volumetric density and number, as well as reduced cortical thickness (16). The significant difference between the LATER and EARLIER subgroups was not attenuated by the significant age-related decline from 20.4 to 45.8 yr in both femoral neck aBMD and distal tibia microstructural components. This finding underscores the importance of later menarche in the risk of osteoporosis. Indeed, at the onset of postmenopausal bone loss, we observed in the LATER compared with the EARLIER menarcheal age subgroup a mean deficit of −0.35 and −0.42 T-scores for femoral neck aBMD and trabecular volumetric bone density (or trabecular bone volume fraction), respectively (16).

Among the putative mechanisms that may potentially explain the relationship between menarcheal age and peak bone mass (PBM) and structure, differences in the duration and/or intensity of estrogen exposure during growth are commonly accepted to play a prominent role (17–19). However, this somewhat intuitive explanation has recently been challenged by considering that both pubertal timing and PBM are traits characterized by large variance and Gaussian distribution and are under the strong influence of heritable factors (20). Therefore, it is possible that both menarcheal age and bone mineral mass acquisition may be predetermined, and low bone mass apparently associated with later menarcheal age could already be detectable before the onset of pubertal maturation.

We tested this hypothesis in a cohort of healthy girls followed from prepuberty to young adulthood; during this time period, their menarcheal age was prospectively recorded, whereas bone mineral mass accrual was longitudinally determined before, during, and after pubertal maturation. This design allows one to test whether gain in bone mineral mass from prepuberty to the age of PBM attainment significantly differs in healthy subjects with earlier compared with later menarcheal age.

Subjects and Methods

Participants

We studied 124 healthy women with mean (±sd) age of 20.4 ± 0.6 yr. They belong to a cohort followed for 12 yr and previously examined at mean ages of 7.9 ± 0.5, 8.9 ± 0.5, 9.9 ± 0.5 (21), 12.4 ± 0.5 (22), and 16.4 ± 0.5 yr (23). During 1 yr, between mean ages of 7.9 and 8.9 yr, half the cohort received a supplementation of calcium in a randomized, double-blind, placebo-controlled design as previously reported (21). The ethics committees of the Department of Pediatrics and the Department of Rehabilitation and Geriatrics of the University Hospitals of Geneva approved the protocol, and informed consent was obtained from both parents and children (21). All subjects were recruited within the Geneva area.

Clinical assessment

Weight, standing height, and body mass index (BMI; kilograms/meter²) were measured. At mean ages (±sd) 7.9 ± 0.5 and 8.9 ± 0.5 yr, pubertal stage was assessed by direct clinical examination by a pediatrician-endocrinologist. At mean ages of 10.0, 12.4, and 16.4 yr, pubertal maturation was assessed by a self-assessment questionnaire with drawings and written descriptions of Tanner’s breast and pubic hair stages. At mean ages 7.9 and 8.9 yr, all girls were classified Tanner’s stage P1, whereas at mean age 10.0 yr, 38% of them had reached Tanner’s stage P2. At each visit starting at mean age of 10 yr, menarcheal age was assessed by direct interview.

As previously described (21), the exclusion criteria at baseline were: weight/height ratio below the 3rd or above the 97th percentile, physical signs of puberty, chronic disease, malabsorption, bone disease, and regular use of medication.

Calcium intake

At each visit, the spontaneous calcium intake, essentially assessed from dairy sources, was estimated by frequency questionnaires (24).

Measurement of bone variables

As previously reported, aBMD (milligrams/centimeter²) was measured by DXA at several skeletal sites with the Hologic QDR-2000 instrument (Hologic Inc., Bedford, MA) at the first five visits (21–23) and with a cross-calibrated QDR-4500 instrument (Hologic Inc.) at the last visit (15). The six measured sites were: distal metaphysis and diaphysis of the radius; femoral neck, trochanter, and diaphysis; and L2–L4 vertebrae in anteroposterior view. The coefficient of variation of repeated measurements as determined in healthy young adults varied between 1.0 and 1.6%.

Expression of the results and statistical analysis

The various anthropometric and osteodensitometric variables are given as mean ± sd. The subjects were segregated according to the median of menarcheal age. Menarche under and above the median age of the first menstruation occurrence was defined as “EARLIER” and “LATER,” respectively. The mean values of the EARLIER and LATER menarcheal groups were calculated and expressed in absolute and relative (Z-score) terms at the different ages, i.e. before, during, and after pubertal maturation. Mean aBMD Z-score of the six skeletal sites corresponds to the mean of the individual Z-scores of each of the six skeletal sites. BMD Z-scores were generated from this healthy cohort used to establish normal reference pediatric values at different skeletal sites.

To evaluate whether EARLIER vs. LATER subgroups differ for any variable throughout time (i.e. at all ages), repeated measures ANOVA were performed after imputation for missing data (<6% for each variable). The differences between EARLIER and LATER menarcheal groups for anthropometric and osteodensitometric variables were evaluated using a two-tailed t test for unpaired values.

Two-way ANOVA was used to test the influence of chronological age compared with menarcheal age and any interaction prevailing between these two factors on mean aBMD of the six skeletal sites at mean ages of 7.9 and 8.9 yr when all girls were still prepubertal. Two-way ANOVA was also used on aBMD of each skeletal site at mean ages of 8.9 and 20.4 yr. The gains from mean age 8.9 to 20.4 yr were calculated and further subdivided for weight, standing height, and BMI in two periods from mean age 8.9 yr (all girls being prepubertal) to 16.4 yr (all girls being postmenarcheal), and from 16.4 to 20.4 yr. The statistical significance of the differences in these aBMD gains between EARLIER and LATER menarcheal groups were evaluated before and after adjustment for the calcium intervention, taken as a dummy variable.

The significance level for two-sided P values was 0.05 for all tests. The data were analyzed using STATA software, version 7.0 (StataCorp LP, College Station, TX).

Results

Anthropometric characteristics of the cohort from mean age 7.9 to 20.4 yr are given (Table 1). At the mean age of 20.4 yr, the calculated BMI was within the normal range. Spontaneous calcium intake was quite constant from mean age 10.0 to 16.4 yr, amounting to about 900 mg/d, decreasing by 10% at 20.4 yr of age (Table 1).

Table 1.

Characteristics of the cohort at different ages

Age (yr)	7.9 ± 0.5	8.9 ± 0.5	10.0 ± 0.5	12.4 ± 0.5	16.4 ± 0.5	20.4 ± 0.6
n	124	123	114	106	113	124
Height (cm)	127.7 ± 5.9	132.7 ± 6.1	138.8 ± 6.7	153.8 ± 7.9	164.0 ± 6.2	165.0 ± 6.0
Weight (kg)	26.5 ± 4.1	29.8 ± 4.9	33.2 ± 5.7	44.5 ± 8.1	56.8 ± 7.9	60.0 ± 9.2
BMI (kg/m2)	16.2 ± 1.8	16.9 ± 2.1	17.1 ± 2.1	18.7 ± 2.5	21.1 ± 2.7	22.1 ± 3.4
Calcium intake (mg · d−1)	888 ± 372	955 ± 413	900 ± 342	890 ± 364	912 ± 444	832 ± 380
aBMD (mg/cm2)
Radial metaphysis	298 ± 28	310 ± 34	314 ± 33	333 ± 44	424 ± 58	452 ± 51
Radial diaphysis	435 ± 30	450 ± 32	469 ± 33	541 ± 50	685 ± 46	710 ± 50
Femoral neck	634 ± 74	647 ± 75	675 ± 78	751 ± 103	867 ± 111	858 ± 108
Femoral trochanter	504 ± 56	523 ± 60	542 ± 62	591 ± 94	723 ± 94	708 ± 96
Femoral diaphysis	1024 ± 77	1082 ± 87	1155 ± 99	1393 ± 151	1539 ± 118	1707 ± 116
Lumbar spine (L2–L4)	616 ± 61	641 ± 63	672 ± 69	821 ± 121	1016 ± 111	1043 ± 118
Mean of the 6 skeletal sites	585 ± 44	609 ± 48	638 ± 52	750 ± 85	875 ± 74	913 ± 71

All values are means ± sd.

Between mean ages 8.9 and 16.4 yr, i.e. from prepuberty to postmenarche, aBMD gain ranged from 34.0% at the femoral neck to 58.5% at the lumbar spine (Table 1). Between the mean ages of 16.4 and 20.4 yr, the increase was much reduced, and the largest aBMD gain was recorded at the radial metaphysis (+6.6%), whereas a slight but nonsignificant decline was observed at the two proximal femur sites: −1.0 and −2.1% at femoral neck and trochanter, respectively (Table 1).

The mean (±sd) menarcheal age was 13.0 ± 1.2 yr (n = 124), ranging from 10.2 to 16.0 yr, and was well within reference values previously recorded in similar ethnic and socioeconomic populations (25). The cohort was then segregated according to the median of menarcheal age into EARLIER and LATER subgroups. The mean menarcheal age of the EARLIER subgroup was 12.1 ± 0.7 yr, ranging from 10.2 to 12.9 yr. The mean menarcheal age of the LATER subgroup was 14.0 ± 0.8 yr, ranging from 13.0 to 16.0 yr. In concordance with menarcheal age, clinical signs of pubertal maturation by the assessment of breast and pubic hair development at the mean age of 12.4 yr indicated that 49.1 vs. 12.5% of the subjects had reached stages P4–P5 according to Tanner’s classification (26) in EARLIER vs. LATER subgroups, respectively (see Supplemental Table 1, published as supplemented data on The Endocrine Society’s Journals Online web site at http//jcem.endojournals.org).

From mean age 7.9 to 20.4 yr, absolute mean aBMD values were higher in the EARLIER compared with the LATER subgroup (data no shown). As expected, these differences in aBMD Z-scores were the largest at mean age 12.4 yr (Table 2). However, they could also be observed before pubertal maturation, i.e. at mean ages of 7.9 and 8.9 yr, the greatest difference in Z-score being recorded at both radial (Δ = 0.36) and femoral (Δ = 0.54) diaphysis (Table 2). They were maintained after pubertal maturation at mean ages 16.4 and 20.4 yr (Table 2). Figure 1 illustrates the evolution of the mean Z-score of the six skeletal sites in the EARLIER compared with the LATER subgroup from prepuberty (mean age, 7.9 and 8.9 yr) to postmenarche (mean age, 16.4 yr) and further on at mean age 20.4 yr when femoral neck PBM was attained as recently reported (16).

Fig. 1.

Mean aBMD Z-score of the six skeletal sites according to the median of menarcheal age from prepuberty to PBM attainment at 20.4 yr of age. The individual Z-scores of each of the six skeletal sites are detailed in Table 3. The statistical significance at each age is indicated under the corresponding bars. Additional P values in parentheses are given after controlling for the 1-yr calcium intervention between mean ages of 7.9 and 8.9 yr. As detailed in Supplemental Table 1, the pubertal stages were P1 at mean age of 7.9 and 8.9 yr, P1–P2 at 10.0 yr, P2–P5 and P1–P5 at 12.4 yr in EARLIER and LATER subgroups, respectively. All the cohort was postpubertal at mean age of 16.4 yr. Between mean age 7.9 and 8.9 yr, statistical analysis by two-way ANOVA indicated that the significant (P = 0.001) age-dependent aBMD increment did not interact with the influence (P = 0.0038) of future menarcheal age.

Table 2.

Anthropometric and osteodensitometric Z-scores from ages 7.9 to 20.4 yr in EARLIER and LATER menarcheal age subgroups

	Age (yr)
	7.9		8.9		10.0		12.4		16.4		20.4
	EARLIER	LATER	EARLIER	LATER	EARLIER	LATER	EARLIER	LATER	EARLIER	LATER	EARLIER	LATER
Weight	0.33 ± 1.03	−0.33 ± 0.85, P = 0.0002	0.36 ± 1.04	−0.35 ± 0.83, P = 0.0001	0.39 ± 0.98	−0.41 ± 0.85, P < 0.0001	0.72 ± 1.11	−0.75 ± 1.29, P < 0.0001	0.25 ± 0.93	−0.25 ± 1.01, P = 0.0066	0.16 ± 0.94	−0.16 ± 1.04, P = 0.084
Height	0.24 ± 0.97	−0.24 ± 0.97, P = 0.0064	0.29 ± 0.95	−0.29 ± 0.97, P = 0.0010	0.40 ± 0.95	−0.41 ± 0.89, P = 0.0000	0.58 ± 0.89	−0.61 ± 1.13 0.0000	0.03 ± 0.98	−0.03 ± 1.02, P = 0.756	−0.03 ± 0.98	0.03 ± 1.03, P = 0.778
BMI	0.26 ± 1.04	−0.26 ± 0.89, P = 0.0038	0.25 ± 1.07	−0.25 ± 0.86, P = 0.0047	0.24 ± 0.99	−0.25 ± 0.95, P = 0.0092	0.39 ± 1.03	−0.40 ± 1.19, P = 0.0004	0.26 ± 0.96	−0.26 ± 0.97, P = 0.005	0.17 ± 0.98	−0.17 ± 1.00, P = 0.051
BMD
Radial metaphysis	0.10 ± 1.00	−0.11 ± 0.99, P = 0.227	0.14 ± 0.99	−0.14 ± 1.00, P = 0.119	0.06 ± 1.02	−0.06 ± 0.99, P = 0.548	0.27 ± 1.01	−0.28 ± 0.91, P = 0.004	0.36 ± 0.92	−0.37 ± 0.95, P = 0.0001	0.19 ± 1.03	−0.17 ± 0.95, P = 0.046
Radial diaphysis	0.18 ± 1.08	−0.18 ± 0.88, P = 0.0495	0.18 ± 1.04	−0.18 ± 0.93, P = 0.043	0.14 ± 1.08	−0.15 ± 0.89, P = 0.130	0.47 ± 0.94	−0.49 ± 0.82, P = 0.000	0.24 ± 0.97	−0.25 ± 0.97, P = 0.009	0.19 ± 0.91	−0.18 ± 1.07, P = 0.042
Femoral neck	0.08 ± 0.96	−0.08 ± 1.03, P = 0.364	0.15 ± 0.96	−0.15 ± 1.02, P = 0.104	0.17 ± 0.93	−0.18 ± 1.05, P = 0.061	0.47 ± 0.81	−0.50 ± 0.94, P = 0.000	0.23 ± 0.84	−0.23 ± 1.09, P = 0.014	0.18 ± 0.90	−0.18 ± 1.07, P = 0.042
Femoral trochanter	0.04 ± 1.00	−0.04 ± 1.00, P = 0.627	0.13 ± 0.13	−0.13 ± 0.13, P = 0.152	0.19 ± 0.96	−0.20 ± 1.02, P = 0.039	0.41 ± 0.81	−0.44 ± 0.99, P = 0.000	0.15 ± 0.87	−0.15 ± 1.10, P = 0.109	0.10 ± 0.90	−0.09 ± 1.08, P = 0.282
Femoral diaphysis	0.18 ± 1.06	−0.18 ± 0.91, P = 0.040	0.27 ± 1.01	−0.27 ± 0.92, P = 0.003	0.27 ± 1.00	−0.28 ± 0.93, P = 0.003	0.53 ± 0.82	−0.55 ± 0.87, P = 0.000	0.26 ± 0.94	−0.26 ± 1.00, P = 0.005	0.16 ± 0.95	−0.15 ± 1.03, P = 0.086
Lumbar spine (L2–L4)	0.11 ± 0.92	−0.11 ± 1.07, P = 0.217	0.17 ± 0.12	−0.17 ± 0.91, P = 0.063	0.25 ± 0.94	−0.25 ± 1.00, P = 0.007	0.52 ± 0.83	−0.55 ± 0.87, P = 0.000	0.19 ± 0.89	−0.20 ± 1.09, P = 0.039	0.15 ± 0.90	−0.14 ± 1.08, P = 0.107

All values are means ± sd. P < 0.0001 between EARLIER and LATER for any variable throughout time by repeated measures ANOVA. P, Statistical significance for the difference in Z-score between EARLIER and LATER subgroups.

To assess whether chronological age would interact with the influence of menarcheal age on aBMD, a two-way ANOVA was applied to the values obtained at mean ages of 8.9 and 20.4 yr. As shown in Table 3, this analysis indicated that significant differences in aBMD between EARLIER and LATER subgroups were independent of age.

Table 3.

aBMD at different skeletal sites in girls at the ages of 8.9 and 20.4 yr according to the median of menarcheal age

Menarcheal age	Age (yr)				Pa 20.4 vs. 8.9 yr	Pb EARLIER vs. LATER	Pc Interaction
	8.9 ± 0.5		20.4 ± 0.6
	EARLIER	LATER	EARLIER	LATER
No. of subjects	62	62	62	62
Age at examination (yr)	9.0 ± 0.5	8.9 ± 0.5	20.4 ± 0.6	20.4 ± 0.6
aBMD (mg/cmb)
Radial metaphysis	315 ± 32	306 ± 32	462 ± 53	443 ± 48	0.0001	0.011	0.411
Radial diaphysis	456 ± 33	445 ± 29	719 ± 46	701 ± 53	0.0001	0.005	0.517
Femoral neck	658 ± 72	636 ± 77	878 ± 97	838 ± 116	0.0001	0.009	0.460
Femoral trochanter	530 ± 60	515 ± 60	718 ± 86	699 ± 104	0.0001	0.095	0.884
Femoral diaphysis	1106 ± 88	1060 ± 80	1725 ± 111	1689 ± 119	0.0001	0.002	0.677
Lumbar spine (L2–L4)	651 ± 58	630 ± 67	1060 ± 107	1026 ± 127	0.0001	0.022	0.587
Mean of the 6 skeletal sites	620 ± 45	599 ± 48	927 ± 62	900 ± 77	0.0001	0.002	0.675

All values are means ± sd.

^a

Differences between girls at the ages of 8.9 and 20.4 yr.

^b

Differences between EARLIER and LATER menarche.

^c

Interaction between chronological and menarcheal age.

To further analyze whether the difference in aBMD observed at PBM between EARLIER and LATER subgroups may be simply due to the 2-yr lag in pubertal timing, the gains from the prepubertal mean age of 8.9 to 20.4 yr, i.e. over the 11.5 yr of follow-up, were calculated at each skeletal site. At none of the six skeletal sites was a difference observed in aBMD gain between EARLIER and LATER subgroups (data not shown). Consequently, the total mean gain in aBMD of the six skeletal sites during the 11.5 yr of follow-up from prepuberty to 20.4 yr was not significantly different between the two groups, even after controlling for the 1-yr calcium intervention between mean ages of 7.9 and 8.9 yr (Fig. 2).

Fig. 2.

Increase in mean aBMD of the six skeletal sites from mean age 8.9 (prepuberty) to 20.4 yr in the EARLIER and LATER menarcheal age subgroups. The significant difference in aBMD between the EARLIER and LATER groups recorded before the onset of pubertal maturation (mean age, 8.9 ± 0.5 yr) was maintained by the end of growth development (mean age, 20.4 ± 0.6 yr). As indicated on the figure, the gains in aBMD of the two groups over the 11.5 yr of follow-up were very similar, not significantly different between the two groups (P = 0.402), even after controlling for the 1-yr calcium intervention between mean ages of 7.9 and 8.9 yr (P = 0.416).

Lower body weight, standing height, and BMI Z-scores in the LATER subgroup were observed at prepubertal stage P1, i.e. at mean ages of 7.9 and 8.9 yr. The standing height deficit of −3.6 cm observed at 8.9 yr in the LATER subgroup, corresponding to a difference of 0.58 Z-score (Table 2), was entirely compensated by a significantly larger gain (+3.8 cm) from mean age 8.9 to 20.4 yr (Table 4). Whereas the difference in standing height was entirely cancelled at 20.4 yr, some statistically borderline differences in body weight (P = 0.084) and BMI (P = 0.051) were still observed at the age of PBM attainment (Table 2). As for aBMD, during the period of pubertal maturation until age of PBM attainment, gains in body weight and BMI values were very similar in both the EARLIER and LATER subgroups (Table 4).

Table 4.

Weight, standing height, and BMI gains from age 8.9 to 20.4 yr according to the median of menarcheal age

	Age (yr)
	8.9 to 20.4			8.9 to 16.4			16.4 to 20.4
	EARLIER	LATER	P	EARLIER	LATER	P	EARLIER	LATER	P
n	56	57		56	57		56	57
Body weight (kg)	30.0 ± 7.4 (1.0)	30.2 ± 8.4 (1.1)	0.893	27.3 ± 5.9 (0.8)	26.6 ± 6.1 (0.8)	0.541	2.7 ± 5.2 (0.7)	3.6 ± 4.8 (0.6)	0.352
Standing height (cm)	30.2 ± 4.7 (0.6)	34.0 ± 4.7 (0.6)	0.000	29.6 ± 4.8 (0.6)	32.9 ± 4.3 (0.6)	0.0002	0.6 ± 1.4 (0.2)	1.1 ± 1.2 (0.2)	0.034
BMI (kg/m2)	5.3 ± 2.5 (0.3)	5.1 ± 2.8 (0.4)	0.584	4.5 ± 2.0 (0.3)	4.0 ± 2.0 (0.3)	0.220	0.9 ± 2.0 (0.3)	1.0 ± 1.8 (0.2)	0.592

All values are means ± sd. (sem) The total gain from 8.9 to 20.4 yr was also analyzed from 8.9 to 16.4 and from 16.4 to 20.4 yr. P, Level of significance between EARLIER and LATER subgroups.

Discussion

Shorter exposure to estrogen from the onset of pubertal maturation to the age of PBM attainment is a hypothesis often put forward for explaining that later menarcheal age is associated with lower bone mineral mass and increased risk of osteoporosis fracture in adulthood (17–19, 27). In support of this hypothesis would be the demonstration that the difference in bone mineral mass would essentially be generated during the period extending from prepuberty to PBM. The foregoing results prospectively obtained in healthy subjects do not support this notion. Thus, from mean age 7.9 to 8.9 yr, although all girls were prepubertal (Tanner stage P1) as directly assessed by medical examination, there was already a significantly lower bone mass in these girls who will experience a relatively later menarche. In our study, none of the more than 120 girls examined at both mean age 7.9 and 8.9 yr displayed any objective sign of thelarche and/or pubarche. Despite this direct assessment of pubertal status, it could be argued that prepubertal females who eventually will experience an early menarche were already exposed to a higher level of estrogen than their counterparts in whom first menstruation was apparently programmed to occur 2 yr later. A previous report indicates that in healthy girls, plasma concentration of 17b-estradiol progressively rises from a mean value of about 40 pmol/liter at Tanner stage P1 to 50, 110, 165, and 230 pmol/liter at stages P2, P3, P4, and P5, respectively (28). Thus, it is only at pubertal stages P3 and P4 that the circulating level of 17b-estradiol becomes clearly elevated, with plasma concentrations 3- to 4-fold greater than in 8- or 9-yr-old prepubertal girls (28). The same pattern of increment in relation with pubertal maturation is observed for the plasma level of estrone (28). Earlier menarche was shown to be associated with higher circulating levels of estradiol in the first phase of pubertal maturation, presumably at stage P2–P3, as well as after menarche (29). Therefore, if estrogen exposure was an important determinant accounting for the menarcheal age-related difference in PBM, its influence should be particularly prominent from Tanner stage P2 on, and not before the onset of pubertal maturation. In sharp contrast with this hypothesis, analysis of our prospective data indicates that the increase in aBMD at all skeletal sites from prepuberty at 8.9 yr of age to PBM at 20.4 yr of age was very similar in those with a relatively later menarche (mean age, 13.9 ± 0.7 yr) compared with the subjects with earlier menarche (mean age, 12.0 ± 0.7 yr) (Fig. 2).

Thus, our prospective data strongly suggest that the difference in bone mineral mass is already present before puberty with very mild increment, if any, during the whole period of pubertal maturation. Therefore, our data do not support the hypothesis that difference in estrogen exposure from the onset of pubertal maturation to 20 yr of age would represent the key factor responsible for the influence of menarcheal age on PBM.

As an alternative to the estrogen exposure hypothesis, another model implying some difference in body habitus, particularly standing height and body weight, has been proposed (27). A previous report indicated that girls who enter menarche later tend to have slightly higher standing height and lower body weight related to less fatness (30, 31). A significant relationship between body fat and pubertal timing was confirmed in some but not all subsequent studies (32).

In our study, body weight of the LATER subgroup was lower by 3.5 kg at age 8.9 yr (P = 0.0001) and remained reduced by 2.9 kg (P = 0.084) at 20.4 yr compared with the EARLIER subgroup. From prepuberty to early adulthood, the gain in body weight was very similar in the two subgroups (Table 4). Thus, as for bone mineral density, the deficit in body weight recorded in the LATER group by the end of the growth period was already present before the onset of pubertal maturation. In contrast, the lower standing height by −3.6 cm observed at the mean age of 8.9 yr in the LATER group was fully compensated by a greater longitudinal growth during pubertal maturation (Table 4). In agreement with the former cross-sectional examination (31), our prospective study confirms that lower BMI in the LATER subgroup (21.5 vs. 22.7 kg/m²; P = 0.051) as observed by the end of the growth period is mainly due to a reduced body weight. Thus, our study does not sustain the notion that in healthy girls menarcheal age substantially affects the adult maximal standing height.

Several hypotheses have been put forward to explain the repeatedly observed association between body weight or fat mass and menarcheal age (31, 33). Among putative transmitters, leptin, which is positively associated with body fat whereas inversely related to age at menarche, may physiologically link fat mass to pubertal timing (32, 34, 35). As to the role of leptin in the regulation of bone mass, it is not known how its indirect effect on trabecular bone remodeling and its direct positive impact on cortical bone (36, 37) would interact and thereby may influence human bone acquisition from prepuberty to adulthood. Studies on the association between body weight, fat mass, and bone mineral mass have yielded conflicting results (38, 39). Furthermore, a recent study provided convincing evidence that in adolescents and adults, despite increased mechanical loading, fat mass had a negative correlation or no correlation with vertebral or appendicular bone mineral mass and structure, as assessed by both DXA and computed tomography (40). There is evidence that the rate of weight gain during very early childhood might be a critical determinant of menarche timing (41). Rapid infant weight gain associated with earlier pubertal maturation is a repeating intergenerational pattern that can be recorded in mothers and their children (42). Genetics influences the variability of both body weight (or BMI) and onset of menarche (43). In twin models, heredity accounts for about 75% of the variance in menarcheal age (43). Quantitatively, heredity plays a similar role in PBM variance (44).

Therefore, the relationships between menarcheal age and body weight, or more specifically body fat, can be largely genetically determined, and likewise the relationship between menarcheal age and bone mineral mass. The inverse correlation found between menarcheal age and aBMD does not mean that the former is the causal determinant of the latter. Similarly, the positive correlation found between body weight and aBMD does not mean that the former is the causal determinant of the latter, e.g. by producing hormonal factors or merely by exerting mechanical impact on bone structure.

Nevertheless, analysis of our data in relation to available genetic and environmental information on pubertal timing, body weight, and bone mass suggests that these three variables might be parts of a common programming, the nature of which remains to be elucidated.

In conclusion, the association between future menarcheal age and bone mineral density is detectable before the onset of any sign of pubertal maturation. Thus, this prospective study with follow-up of healthy subjects from mean age 7.9 to 20.4 yr suggests that the difference in PBM between earlier compared with later menarche is generated before and not during pubertal maturation.

Acknowledgements

The Swiss National Science Foundation supported this study (Grant 3247BO-109799).

Disclosure Summary: All authors have no conflict to declare.

We thank Giulio Conicella and the team of the Service of Nuclear Medicine for DXA and high-resolution peripheral computerized tomography measurements; Fanny Merminod, certified dietician, for the assessment of food intakes and having carried out this study; Pierre Casez, M.D., for the elaboration of the database; François Herrmann, M.D., MPH, for help with statistical analysis; Tara Brennan, Ph.D., for reading the manuscript; and Marianne Perez for secretarial assistance. We are indebted to Prof. D. Belli, M.D., and Prof. S. Suter, M.D., chairpersons of the Department of Pediatrics, for their constant and invaluable support in this research project.

Abbreviations

 
• aBMD

  Areal bone mineral density

 
• BMI

  body mass index

 
• DXA

  dual-energy x-ray absorptiometry

 
• PBM

  peak bone mass.