Examining Bone Surfaces Across Puberty: A 20-Month pQCT Trial


  • Saija A Kontulainen,

    1. Department of Orthopaedics, Faculty of Medicine, University of British Columbia, Vancouver, Canada
    Search for more papers by this author
  • Heather M Macdonald,

    1. School of Human Kinetics, University of British Columbia, Vancouver, Canada
    Search for more papers by this author
  • Karim M Khan,

    1. Department of Family Practice, Faculty of Medicine, University of British Columbia, Vancouver, Canada
    Search for more papers by this author
  • Heather A McKay PhD

    Corresponding author
    1. Department of Orthopaedics, Faculty of Medicine, University of British Columbia, Vancouver, Canada
    2. Department of Family Practice, Faculty of Medicine, University of British Columbia, Vancouver, Canada
    • Division of Orthopedic Engineering Research, Faculty of Medicine, University of British Columbia, 828 West 10th Avenue, Vancouver, BC V5Z 18L, Canada
    Search for more papers by this author

  • The authors have no conflict of interest.


This follow-up study assessed sex differences in cortical bone growth at the tibial midshaft across puberty. In both sexes, periosteal apposition dominated over endosteal resorption. Boys had a greater magnitude of change at both surfaces, and thus, a greater increase in bone size across puberty. Relative increase in cortical bone area was similar between sexes.

Introduction: Generally, sex differences in bone size become most evident as puberty progresses. This was thought to be caused, in part, by greater periosteal apposition in boys, whereas endosteal apposition prevailed in girls. However, this premise is based on evidence from cross-sectional studies and planar measurement techniques. Thus, our aim was to prospectively evaluate sex-specific changes in cortical bone area across puberty.

Materials and Methods: We used pQCT to assess the tibial midshaft (50% site) at baseline and final (20 months) in girls (N = 68) and boys (N = 60) across early-, peri-, and postpuberty. We report total bone cross-sectional area (ToA, mm2), cortical area (CoA, mm2), marrow cavity area (CavA, mm2), and CoA/ToA ratio.

Results: Children were a mean age of 11.9 ± 0.6 (SD) years at baseline. At the tibia, CoA ranged from 230 ± 44, 261 ± 50, and 258 ± 46 in early-, peri-, and postpubertal girls. In boys, comparable values were 223 ± 36 (early), 264 ± 38 (peri), and 281 ± 77 (postpubertal). There was no sex difference for ToA or CoA at baseline. Increase in ToA and CoA was, on average, 10% greater for boys than girls across maturity groups. The area of the marrow cavity increased in all groups, but with considerable variability. The increase in CavA was significantly less for girls than boys in the early- and postpubertal groups. Change in CoA/ToA was similar between sexes across puberty.

Conclusion: Both sexes showed a similar pattern of change in CoA at the tibial midshaft, where periosteal apposition dominated over endosteal resorption. Boys showed a greater magnitude of change at both surfaces, and thus, showed a greater increase in bone size across puberty. The relative increase in cortical area was similar between sexes. These pQCT findings provide no evidence for endosteal apposition in postmenarchal girls.


THE STRUCTURAL basis of bone fragility is firmly rooted in the growing years and becomes fully expressed during aging in both sexes.(1–4) Higher vertebral fracture rates in women compared with men are associated with sex differences in bone size and structure.(3) Structural parameters that may underpin similar sex differences in fracture rates at the proximal femur have yet to be clearly elucidated. However, until the recent advancement in measurement technologies, we have been unable to directly evaluate whether there is sexual dimorphism in long bone size and structure in the growing skeleton. The early literature describes sex differences in bone length and size that are reportedly established during the peripubertal growth period when endochondral ossification increases the length of long bone and periosteal apposition increases cross-sectional bone area (bone size) to a greater extent in boys than girls.(5–8) In his landmark studies, Garn undertook radiographic cross-sectional comparisons of the second metacarpal and showed endosteal (i.e., endocortical) apposition of bone in both sexes. Garn and colleagues observed that this event began earlier and was of greater magnitude in girls than boys.(6–8) These classic studies proposed that endosteal apposition in adolescent girls resulted from the pubertal estrogen surge to supply calcium for reproduction. Garn and colleagues further postulated that endosteal bone lost in menopause as a result of diminished estrogen levels had been accrued during puberty for that eventuality.(6–8) Later studies provided evidence that cortical wall thickness (estimated from DXA images) increased in the femoral shaft as a result of endosteal apposition in girls(9) but not boys.(10)

In contrast, recent cross-sectional comparisons of cortical bone structure by QCT(11) and MRI(12) gave evidence that the medullary cavity increased with advanced maturation and age in both sexes. This did not support the earlier observation at the metacarpal from plain radiographs(6–8) or from the DXA-derived estimations.(9) This discrepancy may reflect either site-specificity or differences in measurement techniques. Because of their planar nature, neither radiographs(5–8,13) nor single-(14) or dual-energy X-ray absorptiometry(5,15–19) are able to measure bone size and structure. Thus, it is only recently that technologies such as pQCT have advanced to study growth at both endosteal and periosteal surfaces.(20–22) To our knowledge, there are no longitudinal studies that have characterized surface-specific adaptations in cortical bone across puberty in both sexes.

Our overall aim in conducting this study was to substantiate previous theory by using repeated pQCT measurements at the tibial midshaft to examine bone structure across three maturational time-points in boys and girls. Based on the findings of Garn, we anticipated that bone structural adaptations would be specific to both stage of maturity and the sex of the child. Specifically, we hypothesized that as girls achieved menarche they would deposit bone on the endosteal surface, whereas this would not be evident in boys.



Subjects were participants in the University of British Columbia Healthy Bones II Study. This exercise intervention trial has been detailed elsewhere.(23) Baseline measurements for this study of 128 children (68 girls) were obtained 4 months after cessation of the Healthy Bones exercise intervention. We found no evidence of an effect of the earlier exercise intervention on pQCT outcomes at the tibia at study baseline.

Furthermore, baseline physical activity scores did not differ between former participants and controls. Thus, all subjects were analyzed together regardless of the previous group assignment. We assessed children at two time-points, on average 20 months apart. The Ethical Review Board at the University of British Columbia approved the protocol, and all parents and participating children provided written informed consent.


Maturity was assessed by self-report of breast (girls) and pubic hair (boys and girls) stage.(24) Menarcheal status was assessed by questionnaire at each measurement time. Girls were grouped into one of three maturity groups according to their menarcheal status over the 20-month period (Table 1). Girls in group 1 (EARLY) were premenarcheal at baseline and remained premenarcheal. Group 2 (PERI) girls were premenarcheal at baseline but reached menarche during the follow-up period (Table 1). Group 3 (POST) girls were postmenarchal at baseline and final.

Table Table 1.. Baseline and Follow-up Characteristics of the Subjects
original image

For boys, maturity groups were based on Tanner pubic hair stage, and we classified boys as EARLY if they were in Tanner stages 1-3, PERI if Tanner stage 4, and POST if Tanner stage 5 at follow-up (Table 1).

Anthropometry and questionnaires

The following procedures completed at both baseline and 20-month have been documented in detail elsewhere.(23) Briefly, standing height was measured to the nearest 0.1 cm using a wall-mounted digital stadiometer (model 242; Seca, Hanover, MD, USA), and body weight was assessed to the nearest 0.1 kg using an electronic scale (model 840; Seca). Tibial length was measured twice as the distance from the tibial plateau to the tip of the medial malleolus (to the nearest millimeter) using an anthropometric tape. We used the mean of two measures to represent the midpoint of the tibia. Parent(s) completed a health history questionnaire for each child. Daily dietary intake of calcium (mg/day) was estimated with a validated food frequency questionnaire.(25) Moderate to vigorous physical activity during the previous 7 days was determined by a modified version of the Physical Activity Questionnaire for Children (PAQ-C).(26,27)

Bone measures

We used pQCT to acquire one 2.5-mm slice (voxel size, 0.5 mm) of the left tibia at the 50% site, measured proximally from the distal endplate of the tibia. A 30-mm planar scout scan provided a view of the joint line that we used to locate a standard anatomical reference. pQCT scans were analyzed using Bonalyse software (Bonalyse 2.1; BonAlyse Oy, Jyväskylä, Finland). A threshold algorithm without a contour (D-mode) was used to separate cortical and trabecular bone. Thresholds were determined from histogram profiles of the bone images by line analysis. Based on these evaluations, the thresholds we used were 540 mg/cm3 to separate cortical and trabecular bone and 171 mg/cm3 to separate bone from soft tissue. Outcome variables were total cross-sectional area of bone (ToA, mm2), cortical bone cross-sectional area (CoA, mm2), the ratio of these variables (CoA/ToA), and the area of marrow cavity (CavA, mm2; ToA-CoA). In our laboratory, the short-term precision (CV%) at the midtibia for ToA and CoA were 0.7%, and 0.9%, respectively. A phantom was scanned daily to maintain quality assurance.

Statistical analysis

We compared baseline and follow-up values within each group with paired-samples t-tests. Between-maturity and between-sex differences in bone variables were evaluated using a general linear model (GLM) multivariate analysis. Dependent variables were baseline and change in bone outcome variables (ToA, CoA, CoA/ToA, CavA), and fixed factors were sex (M, F) and maturity category (EARLY, PERI, POST). If interactions between fixed factors and baseline or change were statistically significant, we performed a separate analysis to assess differences between maturity categories within sex and between sexes within each maturity category. Posthoc range tests and multiple comparisons (between maturity categories) were performed by Tukey's honestly significant difference adjustment. In all tests, an α-level <5% was considered statistically significant.


Comparison across maturity groups and between sexes at baseline

A description of subjects at baseline and final and percent change for height, weight, maturity, intake of dietary calcium, and physical activity participation is provided in Table 1. For ToA at baseline, there were differences across maturity groups in boys, but not girls. ToA was 21% (81 mm2; 95% CI, 32-73) greater in POST compared with EARLY boys. CoA was 13% (32 mm2; 95% CI, 0.6-63) greater in PERI and 21% (49 mm2; 95% CI, 5-93) greater in POST compared with EARLY boys. At baseline, there were no differences in ToA, CoA, and CavA between sexes within each maturity category.

Comparison of 20-month changes within maturity groups

EARLY, PERI, and POST boys and girls all increased ToA and CoA significantly over the 20-month measurement period (Table 2; Fig. 1). We observed a significant increase from baseline in CoA/ToA for EARLY girls only and EARLY and PERI (but not POST) boys. We also noted an increase in CavA for all maturity groups in both sexes (Fig. 1), but these changes were statistically significant in EARLY and PERI groups only (Table 2).

Table Table 2.. Bone Characteristics at Baseline and at 20-Month Follow-up
original image
Figure FIG. 1..

Schematic diagram depicting cortical bone growth over 20 months. Average increases in ToA and CavA are drawn to scale.

Comparison of 20-month change across maturity groups

For girls, the increase in ToA was 5% (20 mm2; 95% CI, 4-35) greater for the PERI compared with POST girls. CoA increased 8% (14 mm2; 95% CI, 4-23) more for EARLY and 5% (14 mm2; 95% CI, 5-22) more for PERI compared with POST girls.

For boys, ToA increased similarly across maturity groups. PERI boys increased CoA 3% (14 mm2; 95% CI, 1-28) more than EARLY and 9% (24 mm2; 95% CI, 2-45) more than the POST boys. Changes in CavA and CoA/ToA ratio did not differ across maturity groups in either sex.

Comparison of 20-month changes between girls and boys

Boys showed a significantly greater increase in ToA and CoA compared with girls at EARLY, PERI, and POST (Tables 2 and 3). In addition, the increase in CavA was greater for boys than girls in EARLY and POST groups (Table 3). This was apparently caused by boys' greater overall growth rate, because we observed no difference for changes in CoA/ToA between girls and boys.

Table Table 3.. Comparison of the 20-Month Changes in Bone Size (ToA) and Cortical Area (CoA), Their Ratio (CoA/ToA), and the Area of Marrow Cavity (CavA) Between Girls and Boys at Three Maturational Time-Points
original image


Our aim was to test the widely accepted models of bone accrual regarding sex differences in surface-based bone apposition across three maturational time-points using a novel measurement technique. We hypothesized that increased endosteal apposition in menarcheal girls would be reflected by a decrease in the area of the medullary cavity and an increase in cortical wall thickness as represented by an increase in the CoA/ToA ratio. Also, if existing paradigms held true, boys and early pubertal girls would show increases in both ToA and CavA as maturity advanced.

Our findings did not support these hypotheses. Cortical bone growth in both peri- and postpubertal girls was dominated by periosteal apposition, and there was no evidence of endosteal apposition in either sex across pubertal groups. In contrast, CavA increased for both girls and boys in early and peripuberty. This greater increase in CavA for boys compared with girls was explained by boys' greater overall increase in bone dimensions (i.e., greater periosteal apposition and endosteal resorption). Diminished growth rate as girls reached postpuberty was reflected in the small increases in ToA and CoA and the maintenance of the CavA. Although boys increased ToA on average 10% more than girls across pubertal groups, the change in CoA/ToA ratio was similar between sexes.

We find support for these findings in two cross-sectional comparisons that used QCT and MRI to assess the femoral shaft.(11,12) These studies showed that cross-sectional area of the medullary cavity was greater in more mature children and young adults in both sexes.(11,12) Together, these findings suggest a re-evaluation of the commonly accepted paradigm that after sexual maturity (menarche) bone is apposed at the endosteal surface in girls. This concept was originally established based on the landmark findings of Garn at the second metacarpal using radiograph techniques. These extensive studies showed endosteal apposition occurred in both sexes, but it was reported to begin earlier and was of greater magnitude in females than males.(6–8)

Early studies of cortical bone were based on extrapolations from breadth measures obtained from radiographs. Significant errors have been reported in the estimation of these two-dimensional cross-sectional parameters using this technique.(28) Van Gerven et al.(28) obtained cut cross-sections of femora and compared estimates of cortical area measured from anterior-posterior X-rays with direct morphometric measures of CoA. They reported differences as high as 25% between methods. More recent evidence from a DXA study suggested that increased cortical wall thickness was a result of endocortical apposition in postmenarchal girls.(5) This finding has been later questioned because of poor accuracy of DXA-estimated bone geometry at the femoral shaft.(12,29)

There is considerable variability in the relatively sparse pediatric literature studying changes in long bone geometry. This, in part, reflects the heterogeneity that exists among children for bone growth as well as the vast array of bone sites measured and technologies used to measure them. Our study could not discern whether endosteal apposition occurred in other skeletal sites or differs between bone regions. A recent 12-month MRI follow-up of the humeral shaft reported that endosteal apposition was detected at the midhumerus but not the distal humerus in postpubertal girls.(30) The authors suggested that this was because of the differential timing of regional growth, whereby the distal skeleton matures in advance of the proximal skeleton.

This study had a number of strengths. pQCT provides a relatively novel and reliable means to study cortical bone area.(31) We used prospective pQCT data to describe changes in total and cortical bone area across maturational time-points in both sexes. According to our knowledge, there are no other in vivo studies that have examined growth in cortical bone area relative to overall increase in bone size. The increase in CoA/ToA in early puberty for both boys and girls and in peripuberty for boys showed that the cortex contributed relatively more to overall bone size than at later stages of maturity. In adulthood CoA/ToA was reportedly greater than in childhood.(12) Direct measures, in vitro, of bone cross-sections obtained from an archaeological sample of children showed that the average CoA/ToA at the tibial shaft decreased from 0.72 to 0.68 from age 12-15 years.(32) This was explained by possible dietary insufficiency in this population.(32) The lower CoA/ToA ratios we observed in this study reflect the gross differences in the study subjects and the differences in measurement techniques.

This study also had a number of limitations. We acknowledge that the comparison across pubertal groups was cross-sectional, and we were thus unable to capture the nuances of a longitudinal growth trajectory. Also, Tanner staging is a broad and thus imperfect means to define maturity, because there may be maturational differences even within a maturity category. Furthermore, the timing and temporal sequence of pubertal events is different in boys and girls. Thus, although Tanner stages provides an estimate of where children are along the sexual maturation continuum, these time-points will not be exactly equivalent across sexes. This was highlighted by the considerable growth (12.3 cm) that occurred in postpubertal boys compared with girls (3.6 cm). However, because we assessed a relatively small sample of postpubertal boys, results from this group should be accepted with caution. We strove to minimize the variability of Tanner stage assessments by using repeated measurements to classify boys and menarcheal status to classify girls. Ideally, if sufficient longitudinal data were collected, children would be aligned on a common maturational landmark such as peak height velocity. This would reduce some of the known variability associated with growth and with the self-assessment of maturity.(24) Differences in lifestyle factors between groups of children, such as dietary intake of calcium and physical activity habits, may also confound study outcomes. To address this, we evaluated both factors by questionnaire and found no statistically significant between group differences at either measurement time-point. Finally, there are two methodological issues with pQCT technique for assessing pediatric bone that warrant discussion. First, it is not possible to ascertain the same exact location along the length of growing bone over time. Thus, we used fixed anatomical reference points to locate the same relative region along the long bone length at each measurement. Second, we used the default voxel size of 0.5 mm to limit the scan time and thus reduce the likelihood of movement. This may have limited our ability to accurately assess the small changes in cortical bone area. In closing, long-term prospective studies designed to evaluate surface-specific bone apposition in different regions of the same bone and between the upper and lower limbs are needed to further characterize the sexual dimorphism in bone size and structure, and changes in material properties (i.e., apparent volumetric cortical and trabecular density).

In summary, both sexes showed a similar pattern of cortical bone growth at the tibial midshaft where periosteal apposition predominated over endosteal resorption. Boys' changes were of greater magnitude at both periosteal and endosteal surfaces, and thus, boys showed a greater increase in bone size across puberty. The relative increase in cortical area was similar between sexes. In conclusion, these findings suggest re-evaluation of the commonly accepted paradigm that bone is apposed at the endosteal surface in postmenarchal girls. Longer-term descriptive studies and mechanistic studies of factors that regulate these growth processes are essential.


Financial support to conduct this study was provided by the Canadian Institutes of Health Research and the Michael Smith Foundation for Health Research. We also acknowledge the Ministry of Education in Finland, and the Academy of Finland (203689) for postdoctoral support for SK. The authors thank Sarah Manske, BSc, for help with data analysis and Kam Sandhu, BSc, for the pQCT precision results. Finally, many thanks to Teppo Järvinen, MD, PhD for valuable comments on the manuscript.