Division of Bone Diseases, WHO Collaborating Center for Osteoporosis Prevention, Department of Rehabilitation and Geriatrics, Geneva University Hospitals and Faculty of Medicine, Geneva, Switzerland
Address reprint requests to: Thierry Chevalley, MD, Division of Bone Diseases, Department of Rehabilitation and Geriatrics, Geneva University Hospitals and Faculty of Medicine, Rue Micheli-du-Crest 24, CH-1211 Geneva 14, Switzerland
The authors state that they have no conflicts of interest.
Late menarche is a risk factor for fragility fractures. We hypothesized that pubertal timing–dependent alterations in bone structural components would persist from peak bone mass to menopause, independent of premenopausal bone loss. We studied the influence of menarcheal age (MENA) on femoral neck BMD (FN aBMD) by DXA and microstructure of distal tibia by HR-pQCT in healthy young adult (YAD; 20.4 ± 0.6 [SD] yr, n = 124) and premenopausal middle-aged (PREMENO; 45.8 ± 3.4 yr, n = 120) women. Median of MENA was 13.0 ± 1.2 and 13.1 ± 1.7 yr in YAD and PREMENO, respectively. In YAD and PREMENO (n = 244), FN aBMD (R = −0.29, p = 0.013), as well as total volumetric BMD (Dtot; R = −0.23, p = 0.006) and cortical thickness (Ct.Th; R = −0.18, p = 0.011) of distal tibia were inversely correlated to MENA. After segregation by the median of MENA in EARLY and LATE subgroups, the significant influences of both MENA (p = 0.004) and chronological age (p < 0.0001) were observed for FN aBMD and trabecular bone volume fraction of the distal tibia with similar differences in T-scores between LATE and EARLY subgroups in YAD (−0.36 and −0.31 T-scores) and PREMENO (−0.35 and −0.42 T-scores) women. Ct.Th was negatively influenced by MENA, whereas trabecular thickness (Tb.Th) was negatively influenced by chronological age. There was a striking inverse relationship between cross-sectional area and Ct.Th (R = −0.57, p < 0.001). In conclusion, the negative influence of late menarcheal age at weight-bearing sites as observed by the end of skeletal growth remains unattenuated a few years before menopause and is independent of premenopausal bone loss. Alterations in both bone mineral mass and microstructural components may explain the increased risk of fragility fractures associated with later menarcheal age.
Several factors related to the reproductive life of women influence the risk of fragility fractures. The age at which the production of sexual hormones ceases represents an important determinant of this risk. Epidemiological studies also indicate that variations in pubertal timing are another factor that influences the risk of osteoporosis. Late menarche is associated with low bone mass at several sites of the skeleton in both premenopausal and postmenopausal women. These findings are supported by observations that late pubertal maturation is linked to an increased risk of vertebral and nonvertebral fragility fractures.
Retrospective epidemiological studies have shown that the inverse relationship between bone mineral mass or density and age of menarche seen in premenopausal women has generally been ascribed to the duration of sex hormone exposure up to the time of peak bone mass. These findings imply that, the longer the duration of sex hormone exposure, the higher peak bone mass will reach. During pubertal maturation, early cross-sectional analysis of appendicular bone, at least in the upper limb, indicated some distinct sex dimorphisms. In female subjects, increase in bone mineral mass may be caused more by endosteal than periosteal accrual. Assuming that estrogen is responsible for this enhancement in endosteal deposition, it may be inferred that, at the time of peak bone mass, cortical thickness is inversely related to menarcheal age. In fact, we recently reported that this was indeed the case at the level of the distal radius, a non–weight-bearing site. Whether similar, menarcheal age-related alterations also occur at weight-bearing sites, such as femur and tibia, is currently unknown.
In this study, we investigated the influence of menarcheal age on bone mineral mass at the level of the proximal femur by DXA and on microstructural components of the distal tibia by high-resolution pQCT (HR-pQCT). Our studies were carried out on both 20- and 45-yr-old healthy women to examine whether this influence of menarcheal age at the time of peak bone mass attainment was also detectable in a cohort of mid-40s, premenopausal women close to the mean age of menopause onset. The design of this study allowed us to test the hypothesis that any deleterious effect of premenopausal bone loss in healthy women would be additive to the persisting negative influence of late menarche.
MATERIALS AND METHODS
The study was carried out in 124 healthy young adult women with a mean age of 20.4 ± 0.6 (SD) yr. During the same period, 120 healthy premenopausal women, 45.8 ± 3.4 yr of age, underwent identical examination as the younger cohort. The protocol was approved by the Ethics Committee of the University Hospitals of Geneva. Thirty-seven percent of the subjects were mother-daughter related. All 124 subjects of the younger group belonged to a 12-yr cohort study and had been previously examined at mean ages of 7.9, 8.9, 9.9, 12.5, and 16.4 yr. Between 7.9 and 8.9 yr of age, one half of the cohort received a calcium supplementation as previously reported in detail.
All subjects lived within the Geneva area. The 244 participants were examined over a period of 1.5 yr, from November 2005 to April 2007. Technical staff and instruments used for data acquisition were unchanged throughout the study. Weight, stadiometer standing height, and body mass index (BMI, kg/m2) were determined in all subjects. Menarcheal age, as defined by the exact age of first menstruation, was assessed prospectively and retrospectively in the 20- and 45-yr-old cohorts, respectively. The persistence of regular menstruation at the time of examination was a prerequisite for inclusion in the cohort of women 45.8 ± 3.4 yr old.
Spontaneous calcium intake, essentially assessed from intake of dairy sources, was estimated by frequency questionnaires, as previously described for the young healthy adult (YAD) cohort.
Protein intake was assessed in the last examination described in this report by a frequency questionnaire. The total animal protein intake was expressed in either grams per day or grams per kilogram body weight per day. This included dairy, meat, fish, and egg proteins.
Usual physical activity and weight-bearing physical activity were assessed by a frequency questionnaire. It included organized sports, recreational activity, and usual walking and cycling. Subsequently, the collected data were converted and expressed as physical activity energy expenditure (PAEE; kcal/d) using established conversion equations.
Measurement of bone variables
Areal BMD (aBMD, mg/cm2) was measured by DXA at the level of the femoral neck (FN) with an Hologic QDR-4500 instrument (Waltham, MA, USA) as previously reported. The CV of repeated aBMD measurements varied between 1.0% and 1.6%. Volumetric bone density and microstructure were determined at the distal tibia by HR-pQCT on an XtremeCT instrument (Scanco Medical, Basserdorf, Switzerland) as previously described. The following variables were measured: total (Dtot), cortical (Dcort), and trabecular (Dtrab) volumetric BMD expressed as milligram hydroxyapatite (HA) per cubic centimeter; trabecular bone volume fraction (BV/TV, %), trabecular number (Tb.N), thickness (Tb.Th, μm), and spacing (Tb.Sp, μm); mean cortical thickness (Ct.Th, μm), and cross-sectional area (CSA, mm2). The in vivo short-term reproducibility of HR-pQCT at the distal tibia assessed in 15 subjects with repositioning varied from 0.7% to 1.0% and from 3.0% to 4.9% for BMD and for trabecular architecture, respectively. These reproductibility ranges are similar to those recently published.
Expression of the results and statistical analysis
The various anthropometric and osteodensitometric variables are given as mean ± SD. The differences in density, microarchitecture, and clinical characteristics among healthy young adult and premenopausal middle-aged women were assessed by unpaired Student t-test or by Wilcoxon signed rank test whether the variables were normally distributed. The relationships between menarcheal age and bone variables were examined by univariate regression analysis, both without and after adjustment for age and cohort (generation) effects. Likewise, the relation between menarcheal age in a subgroup of 46 mother-daughter pairs was assessed. The two cohorts of young adult (YAD) and premenopausal (PREMENO) women were also segregated according to the median of menarcheal age (MENA). Two-way ANOVA applied to repeated measure design for the mother-daughter relationship (n = 46), as defined as the repeated factor, was used to test the influence of chronological age compared with menarcheal age, and any interaction prevailing between these two factors, on FN aBMD and microstructural variables of the distal tibia. To further confirm that the influence of menarcheal age was still present after taking into account the correlation between mothers and daughters as shown in Fig. 4, we applied an additional two-way ANOVA test after exclusion of the 46 mothers and their daughters.
T-score of FN aBMD was based on the regional reference values used in the present clinical unit, which is dedicated to the diagnosis of osteoporosis. T-scores of the microstructural components were calculated from an external cohort of healthy 34-yr-old French women who were recently measured on the same HR-pQCT model as the one used in this study. The data were analyzed using STATA software, version 7.0 (StataCorp, College Station, TX, USA).
Comparisons between 20-yr-old and middle-aged premenopausal women
Anthropometric and lifestyle variables:
The age of menarche was very similar in the two cohorts (Table 1). The standing height was also virtually the same in the young adult (YAD) and in the middle-aged premenopausal women (PREMENO), whereas body weight and consequently BMI were greater in the older group (Table 1). Calcium intake was similar, whereas protein intake, corrected for body weight and expressed as grams per kilogram body weight per day was slightly (+2.8%), although not significantly, greater in the PREMENO group (Table 1). The level of physical activity was 32% greater in the YAD compared with the PREMENO group (p = 0.014).
Table Table 1.. Characteristics of Healthy Young Adult and Middle-Aged Premenopausal Women
The mean FN aBMD was lower by 6.1% in PREMENO compared with YAD (Table 2). According to the regional reference values used in the present clinical unit, which is dedicated to the diagnosis of osteoporosis, FN aBMD T-scores were −0.22 ± 0.12 and +0.26 ± 0.11 (SD) in PREMENO and YAD, respectively. Several densitometric and microstructural elements were significantly lower in PREMENO compared with YAD (Table 2), including: Dtot (−4.6%), Dtrab (−11.6%), and its mathematically related value, BV/TV (−11.4%), as well as Tb.Th (−7.5%). The reduction in both Dtrab (or BV/TV) and Tb.Th was mathematically associated with a significant enlargement in Tb.Sp (+6.1%) in the PREMENO group (Table 2). A statistically but biologically not relevant increase in cortical density (+1.0%) and a nonsignificant increase in CSA (+3.2%) were recorded in the PREMENO compared with the YAD group (Table 2). In contrast, Ct.Th was similar in the two cohorts (Table 2).
Table Table 2.. Femoral Neck aBMD and Microstructure of Distal Tibia in Healthy Young Adult and Middle-Aged Premenopausal Women
Influence of menarcheal age
Negative relationships were observed between MENA and several bone variables (Table 3), before and after adjustments for age and cohort effect. Similar significant inverse relationships were observed for both aBMD of the femoral neck (Fig. 1), as measured by DXA, and total volumetric BMD (Fig. 2) of the distal tibia, as assessed by HR-pQCT. Statistical significance with p < 0.05 was also obtained for Ct.Th (Table 3). Inverse relationships with MENA were close to statistical significance for Dcort and Dtrab (or BV/TV; Table 3). Adjustment for the calcium intervention in YAD did not modify the correlation coefficients between menarcheal age and FN aBMD nor any of the measured microstructural elements of the distal tibia.
Table Table 3.. Correlation Coefficients (R) Between Menarcheal Age and aBMD at Femoral Neck and Microstructural Variables of Distal Tibia in 244 Healthy Young Adult and Middle-Aged Premenopausal Women
Associated with the strong effect of menarcheal age on Ct.Th was a trend for greater CSAs (Table 3; Fig. 3). Thus, according to the calculated regression line, a 2-yr difference in MENA (e.g., from 12–14 yr) correlated to a 2.7% increase in CSA. As shown in Fig. 3, there was a highly significant inverse relationship between cortical thickness and CSA. This inverse relationship was independent of menarcheal age, with R values of −0.57 and −0.55 (p < 0.001) in EARLY and LATE subgroups, respectively. The regression equation of this relationship indicates that a 20% decrease in Ct.Th (i.e., from 1.0 to 0.8 mm) was associated with an increase in CSA of 8.3% (Fig. 3). BV/TV was also inversely related to CSA. The slope of the regression line (CSA = −7.2 BV/TV + 742, R = −0.18, n = 244, p = 0.004), however, was less pronounced, because a 20% decrease of BV/TV was associated with an enlargement of CSA by only 3.1%.
Each cohort was divided into relatively early (EARLY) and late (LATE) menarcheal age groups according to the median time of the first menstruation. Age at examination did not differ between the EARLY and LATE groups (Table 4). In both cohorts, BMI were between 5% and 7% lower in LATE compared with the corresponding EARLY group (Table 4). Two-way ANOVA indicated that body weight and BMI were significantly influenced by both chronological and menarcheal age, although without any evidence for a statistical interaction between these two factors.
Table Table 4.. Characteristics of Healthy Young Adult and Middle-Aged Premenopausal Women Segregated by the Median of Menarcheal Age
Femoral neck aBMD was significantly lower in LATE compared with EARLY, as well as in PREMENO compared with YAD (Table 5). These observations were similar for both Dtot and Dtrab (or BV/TV) measurements of the distal tibia. Tb.Sp was wider in LATE compared with EARLY as well as in PREMENO compared with YAD (Table 5). As analyzed by two-way ANOVA, there was no statistical evidence for an interaction between the effect of chronological and menarcheal age on FN aBMD, Dtot, Dtrab (or BV/TV), and Tb.Sp (Table 5). A trend for such dual influences without any interaction between chronological and menarcheal age was recorded for Dcort and Tb.N (Table 5). In contrast, Tb.Th was strongly influenced by chronological age, although barely affected by menarcheal age. Conversely, although Ct.Th was strongly influenced by menarcheal age, it was not associated with chronological age.
Table Table 5.. Femoral Neck aBMD and Microstructure of Distal Tibia in Healthy Young Adult and Middle-Aged Premenopausal Women Segregated by the Median of Menarcheal Age
More than one third (46 of 124) of YAD were daughters of PREMENO women. The anthropometric and bone characteristics of this subgroup (data not shown) were not statistically different from the values of the whole cohort presented in Tables 1 and 2. Menarcheal age of PREMENO mothers was significantly correlated (R = 0.39, n = 46, p < 0.01) to that of their daughters included in the YAD cohort (Fig. 4).
To take into account this correlation between mothers and daughters, the respective influence of both chronological age and menarcheal age on FN aBMD and microstructural components of distal tibia was analyzed by two-way ANOVA after exclusion of the 46 mother-daughter pairs. The levels of significance were very similar to those described in Table 5 (data not shown).
Figure 5 shows the negative influence of menarcheal age in both YAD and PREMENO for both FN aBMD and BV/TV as expressed in T-scores. Between LATE and EARLY menarche groups, differences in T-scores on femoral neck aBMD were very similar in young adult (−0.36 T-score) and in premenopausal (−0.35 T-score) women. At the distal tibia, similar differences in T-scores of BV/TV between LATE and EARLY menarche groups were also observed in young adult (−0.31 T-score) and premenopausal (−0.42 T-score) women. From YAD EARLY to PREMENO LATE, there was a progressive decrease in mean T-score from +0.26 to –0.57 and from +0.50 to –0.48 in FN aBMD and distal tibia BV/TV, respectively. The differences in FN aBMD in relation to aging (20 versus 45 yr) were –0.48 and –0.47 T-score in EARLY and LATE, respectively (Fig. 5). The corresponding differences in BV/TV between YAD and PREMENO were –0.55 and –0.67 T-score (Fig. 5), respectively.
Based on circumstantial evidence, previous reports have suggested that, for the same underlying reason of reduced lifetime exposure to estrogen, fracture risk at the proximal femur, spine, and forearm would be greater with late menarche as with earlier-onset menopause. This study provides insight into the relationships between menarcheal age and bone macro- and microstructure that sustains the notion that pubertal timing in women is an important factor that modulates the risk of sustaining a fragility fracture during adulthood.
The measurement of aBMD at the level of the proximal femur is still considered the gold standard for establishing the diagnosis of osteoporosis. In this study, aBMD was measured during the same time period and on the same DXA machine in two generations of healthy women, defined as subjects free of any clinically detectable sexual maturation disorder. The differences in T-scores between late and early menarche groups on femoral neck aBMD were very similar in young adult (−0.36 T-score) and in premenopausal (−0.35 T-score) women. This observation is compelling and suggests that the deficit observed at the age of peak bone mass, which is attributable to late pubertal timing, persists virtually unchanged until a time close to the onset of menopause. In other words, the negative influence of relatively late menarcheal age on femoral neck aBMD remains unabated during the reproductive life span.
In our study, femoral neck aBMD was on average 6.1% lower in the 45-yr-old premenopausal group than in the 20-yr-old adult cohort. The apparent decrease in femoral neck aBMD from the beginning of the third decade to the mid-40s corresponded to a mean reduction of –0.48 T-score. The difference in femoral neck aBMD between the 20- and the 45-yr-old cohorts was very similar in the early (−0.48 T-score) and late (−0.47 T-score) menarche groups. The decrease in femoral neck aBMD T-scores in premenopausal women as estimated by the difference between the 20- and 45-yr-old cohorts is in agreement with cross-sectional and longitudinal studies.
Thus, from 20 to 45 yr, the influence of aging is additive to that of relatively late pubertal timing on femoral neck aBMD. The demonstration of these additive effects substantiates previous studies that indicate that late menarche is associated with low BMD in premenopausal women.
Furthermore, by using noninvasive high-resolution technology, our study showed that the influence of menarcheal age on femoral neck aBMD is associated with alterations in microstructural components of the distal tibia. These modifications observed in women in their early 20s, in relation to pubertal timing, were often more prominent in women in their mid-40s, that is to say a few years before the cessation of the ovarian function. This was the case for total density, trabecular density, and cortical and trabecular thickness of the distal tibia. It should be noted that, although premenopausal microarchitectural bone loss is associated with a reduction in trabecular but not in cortical thickness, the influence of menarcheal age was much more pronounced on the width of the cortical shell. To our knowledge, this is the first report indicating such a differential influence of chronological and menarcheal age on cortical and trabecular tissue analyzed at the same skeletal site.
A decrease in the amount of bony tissue within the envelope of skeletal pieces may be considered as a stimulus for a compensatory increase in periosteal apposition. An increase in CSA would reflect such a biomechanically driven phenomenon. This concept was developed for the construction of the femoral neck during growth in relation with its strength in old age. It was proposed that greater periosteal apposition leading to a wider femoral neck was offset by even greater endocortical resorption, so that the same net amount of bone would be distributed as a thinner cortex further from the neutral axis, increasing the resistance to bending despite lowering volumetric BMD. In our study, a later menarcheal age was linked to a larger cross-sectional area of the distal tibia, which was correlated with a thinner cortical shell and a lower volumetric trabecular density (or BV/TV).
The significant reduction in femoral aBMD observed in women in their mid-40s compared with those in their early 20s was associated with several alterations in distal tibia microstructure. These altered parameters that were particularly prominent included trabecular density (−11.6%), which seems more likely to be caused by reduced trabecular thickness (−7.5%) than number (−3.2%). As a mathematically derived variable, trabecular space was larger (+6.1%) in the 45-yr-old premenopausal group than in the 20-yr-old women. A trend for larger cross-sectional area of the distal tibia was observed in relation to the lower trabecular density (or BV/TV) detected in the premenopausal women.
Several recent reports have examined, by HR-pQCT, the relation of microstructural alterations at distal radius and/or tibia to osteoporosis. Differences were observed between premenopausal and postmenopausal women in several microstructural components of the distal radius and tibia. Furthermore, although spine and hip aBMD were similar in osteopenic postmenopausal women, those with a history of fracture had lower Dtrab in the ultradistal radius than those free of fracture. Likewise, in postmenopausal women, Dtot, Dtrab, Ct.Th, and Tb.Th at the distal radius were significantly lower in subjects who had sustained a fracture than in controls, even after adjustment for total hip BMD. In another study, women with Colles' fracture had inferior Dtot, Ct.Th, and Tb.N in the distal radius. In a cross-sectional study on the microstructure of the distal radius, the influence of sex and age between 20 and 90 yr was studied. With aging, the decrease in BV/TV was similar in women and in men, but whereas women had a significant reduction in Tb.N associated with an increase in TbS, in men, the most prominent decrease was in Tb.Th. Taken together, these studies provide evidence that the use of HR-pQCT can provide greater insight into the pathogenesis and evaluation of the risk of osteoporotic fracture.
Our study further supports this notion by specifying which microstructural components are affected by menarcheal age. It suggests that the influence of pubertal timing, as recorded at the time of peak bone mass attainment (i.e., around age 20 yr) remains unaltered 25 yr later, just before menopause. Interestingly, a decrease of similar magnitude in total volumetric BMD, trabecular volumetric BMD, cortical thickness, and trabecular thickness was observed by HR-pQCT at the distal tibia in women with later occurrence of menarche and at the distal radius in postmenopausal women who had sustained a fracture compared with controls, even after adjustment for total aBMD. In the latter study, alterations of cortical and trabecular architecture were associated with fractures in postmenopausal women, in part independently of decreased aBMD measured by DXA. The age-dependent decline in macro- and microstructural components that are related to bone strength, as measured at the level of both proximal femur and distal tibia, adds to the deficit triggered by late pubertal timing. This age-related difference was somewhat greater for BV/TV; however, it was of similar magnitude independent of whether women experienced relatively early (−0.55 T-score) or late (−0.67 T-score) menarche. Our data are thus compatible with the notion that the time of menarche is inversely related to the risk of vertebral and nonvertebral fragility fractures, independent of the aging process.
Both pubertal timing and peak bone mass are under the strong influence of hereditable factors and can be moderately affected by common environmental determinants. The hereditable component of pubertal timing was well identified by both twin and mother-daughter relationship studies. Our study both complements and furthers this notion by providing evidence that this familial relationship exists in premenopausal women and their daughters measured at the time of peak bone mass attainment.
Several aspects limit the interpretation of our data. Menarcheal age was prospectively assessed only in the 20-yr-old cohort. The retrospective assessment made in the premenopausal cohort reduces the accuracy of the information to within about a year, possibly half a year in some subjects, instead of within a month as when prospectively recorded. Whereas areal BMD was measured by DXA technology at the level of the proximal femur, the microstructural components were determined by HR-pQCT in the distal tibia. It certainly would have been more informative to analyze the microstructural components at the level of the proximal femur. This scientifically logical approach was not possible because of both technological constraints and the necessity to minimize X-ray irradiation at the hip level for ethical reasons. In this condition, our study design prevented us from reliably establishing which bone microstructural components are responsible for the effect of menarcheal age on proximal femur aBMD. The interpretation of the data regarding the role of menarche on structural components related to bone fragility is also limited by the fact that the two cohorts were not randomly selected among the general population. Nevertheless, each cohort was quite homogenous and representative of the general healthy population of the Geneva area, in terms of anthropometric, lifestyle, and osteodensitometric variables, as well as the mean and range of menarcheal age.
In conclusion, variations in the timing of the onset of reproductive life is associated with differences in several macro- and microstructural components of bone. In healthy women, a delay of ∼2 yr in menarcheal age is linked to significant decreases in T-scores of the femoral neck aBMD and trabecular density at the distal tibia as assessed by HR-pQCT. This influence of menarcheal age, observed at the time of peak bone mass attainment, remains evident and significant 25 yr later at an age close to the onset of menopause. These data contribute to a better understanding of how late menarche within the physiological range increases the risk of fragility fractures in postmenopausal women.
The authors thank Giulio Conicella and the team of the Service of Nuclear Medicine for DXA and HR-pQCT measurements, Fanny Merminod, certified dietician, for the assessment of food intakes and having carried out this study, Pierre Casez, MD, for the elaboration of the database; François Herrmann, MD, MPH, for help with statistical analysis; Tara Brennan, PhD, for reading the manuscript; and Marianne Perez for secretarial assistance. We are indebted to D Belli, MD, and S Suter, MD, chairpersons of the Department of Pediatrics, for constant and invaluable support in this research project. The Swiss National Science Foundation supported this study (Grant 3247BO-109799).