Subjects and methods

SUBJECTS

Our study is based on a one year follow up of a previously published cross sectional study with school children from five schools in Copenhagen.10 Of the 343 (201 girls, 142 boys) healthy individuals participating in the cross sectional study, 97% (192 girls, 140 boys) were re-examined one year later (mean (SD), 1.02 years (0.07)). Table 1 gives basic information on the study population, but more detailed information has been published previously.10

PUBERTY AND ANTHROPOMETRY

One hundred and seventy six girls and 137 boys participating in the follow up study had their pubertal stages assessed according to Tanner17 18 at the first examination. Breast development in girls and pubic hair development in boys were used to assess pubertal stage. Height and weight were measured before DXA scanning. Height was determined to the nearest 1 mm using a stadiometer. Weight was measured to the nearest 0.1 kg using a digital electronic instrument. Subjects wore only pants and a cotton T-shirt when weighed.

BONE MINERAL ASSESSMENT

Whole body BMC (measured in grams of hydroxyapatite) and bone size expressed as anterior–posterior projected BA (measured in cm²) were determined by DXA scanning using a Hologic 1000/W (Hologic Inc, Waltham, Massachusetts, USA). For analysis, software version 5.61 was used. Subjects wore only pants and a cotton T-shirt during the scan. For quality control, spine phantoms were scanned daily. The coefficients of variation for these BMC and BA measurements on a spine phantom over a period of two years (n = 358) were 0.37% and 0.28%, respectively. For each scanning the entrance radiation dose level was 15 μSv, with an effective dose not more than 10 μSv, equal to about one day’s background radiation in Denmark.

STATISTICAL METHODS

The weight for height and the height for age distributions of the individuals were compared with a Danish sex specific reference.19 The result of this analysis has been reported earlier10; it showed that the children were taller for age and heavier for height compared with the 20 year old Danish reference data.19 This is in accordance with the general pattern that both the height of the children and the prevalence of overweight children has increased in Scandinavia.20-22

Annual gain in height (ΔH (cm/year)), weight (ΔW (kg/year)), BMC (ΔBMC (g/year)), and BA (ΔBA (cm²/year)) were calculated from the two measurements corrected to exactly one year by dividing the difference in measurement by the difference in age (years). Annual bone calcium accretion rates expressed as mg calcium/day (ΔCa) according to pubertal stages were calculated assuming that 32.2% of the BMC measured by Hologic instruments is calcium23 (ΔCa (mg/day) = (ΔBMC (g/year)*1000*0.322)/365).

We derived smooth centile curves for ΔBA and ΔBMC versus mean age over the period using the LMS method.24 This summarises the distribution of the dependent variable (for example, ΔBMC) given the covariate by a Box-Cox power (L), depending on the skewness of the distribution, the median (M), and the coefficient of variation (S), giving smooth curves for the dependence of L, M, and S on the covariate. In our study a modified version adding 300 to the BA and BMC accretion was used, because negative accretion occurred; thus, the centiles are based on the assumption that the variable Z, defined as:

is standard normally distributed. Therefore, these formulae may be used to calculate the Z score of an observed value of ΔBMC or ΔBA given the value of the covariate. The range of the covariate age was reduced to eliminate edge effects caused by single extreme values. Thus, we considered the age interval 6.5–19.5 years (both sexes).

The influence of pubertal stage on ΔBA and ΔBMC was estimated by maximum likelihood in a linear mixed model allowing the variance to depend on pubertal stage. The pubertal stages were entered as four binary covariates, each measuring the change from one pubertal stage to the next. All tests were based on the likelihood ratio test statistic.

Results

Table 2 shows annual accretion rates in BMC (g/year), BA (cm²/year), and calcium (mg/day) according to sex and pubertal stage at first examination.

GAIN IN BMC AND GAIN IN BA RELATED TO AGE AND PUBERTAL STAGES

The sex specific, age dependent centile curves for ΔBA (fig 1) and ΔBMC (fig 2) showed a large variation around the median, especially for the boys between 11 and 16 years. The Box-Cox power (L), the median (M), and the coefficient of variation (S) for ΔBA (cm²/year) and ΔBMC (g/year) by sex and age are shown in tables 3 and 4, respectively. Using these values, it is possible to calculate Z scores for individual children. The girls’ curves peaked earlier than those of the boys’ for both ΔBA (fig 1) and ΔBMC (fig2). The peak ages for ΔBA were 12.3 and 13.4 years for girls and boys, respectively, whereas the peak ages for ΔBMC were 12.5 and 14.2 years, respectively. The ΔBA and ΔBMC curves were of similar shape except that ΔBA peaked earlier than ΔBMC for both girls and boys, most distinctly in boys (figs 1 and 2). Eight boys between the ages of 11 and 13.5 years had a ΔBMC above the 90th centile. Six of these boys (75%) were in pubertal stage III, while only 8% of the boys below the 90th centile were in pubertal stage III. The youngest boys in pubertal stage III (< 13.5 years; n = 8) had a particularly high ΔBMC, with a median Z score of 1.72 (range, −0.54 to 2.29).

ΔBA and ΔBMC were significantly associated with pubertal stage for girls (χ²(4) = 88.54 and χ²(4) = 64.20, respectively; both p < 0.0001) and boys (χ²(4) = 67.12 and χ²(4) = 55.29, respectively; both p < 0.0001). ΔBA was highest in pubertal stage III for both girls and boys, but not significantly different from pubertal stage II in girls and boys (girls: χ²(1) = 0.187, p = 0.67; boys: χ²(1) = 2.07, p = 0.15), and borderline significantly different from pubertal stage IV in boys (χ²(1) = 3.52, p = 0.06). ΔBMC was highest in pubertal stage III for both girls and boys, but only borderline significantly different from pubertal stage II in girls (χ²(1) = 2.83, p = 0.09), and pubertal stage IV in boys (χ²(1) = 2.975, p = 0.08).

Discussion

Our longitudinal study on whole body bone mineralisation confirmed the close relation between pubertal stages and gain in BMC and bone size described previously for regional bone mineralisation. Our study showed a dissociation in time between peak growth in bone size and the peak growth in BMC. To our knowledge, this is the first study reporting longitudinal data on whole body BA in a wide age range for both girls and boys.

We have only presented centile curves for whole body BA accretion and for whole body BMC accretion. We chose not to use bone mineral density (BMD) calculated as BMC/BA because BMD measured with anterior–posterior osteodensitometry represents a mixture of true density and skeletal size, as described previously.10 25-27

We suggest that our data may be used for assessing growth of whole body BMC and BA of children for clinical or research purposes. However, for young boys in pubertal stage III, extremely high Z scores for BMC accretion may be anticipated. Attained BA and BMC can be evaluated using the data based on the first examination of these children.10 However, there are some limitations to the use of our dataset as a reference because of the limited number of children in each age and sex group, and because only 25% of the children approached agreed to participate, as described in more detail previously.10 Nevertheless, it is one of the most comprehensive sets of data on whole body measurements available at present. For daily clinical use, an individual may be plotted on centile charts like those shown in figs 1 and 2. (Single copies of large centile charts can be obtained from the authors.) If a more accurate centile or Z score is required, the Z score can be calculated using the equations presented above by inserting the measured ΔBA or ΔBMC and L, M, and S values corresponding to the mean age over the period presented in tables 3 and 4. When using these data for calculation of Z scores for ΔBA and ΔBMC, the calculation of ΔBA and ΔBMC should be based on measurements determined with an approximate one year interval corrected exactly to one year, as described above.

Our data indicate that the median accretion rates for both ΔBA and ΔBMC are similar for the two sexes before puberty (tables 3 and 4). This is in agreement with Theintz et al,5 who found no sex differences between the ages of 9 and 11 years for bone mass accretion in lumbar spine, femoral neck, and femoral shaft. Recently, Martin et al have published data on whole body BMC accretion also showing that the accretion rates were similar for girls and boys aged 9–10 years (130 g/year).6

We observed an earlier age for peak accretion in both ΔBA and ΔBMC for girls (12.3 and 12.5 years, respectively) compared with boys (13.4 and 14.2 years, respectively), which is in agreement with the earlier age of puberty in girls. An earlier age for the regional bone mass accretion peak for girls compared with boys was also found by Theintzet al.5 The age at peak accretion in both ΔBA and ΔBMC in our study was similar to that seen by Martin and colleagues6 in Canadian children, where the difference in age between peak in ΔBA and ΔBMC was also most pronounced in boys. The earlier peak in BA accretion rate in relation to the peak in BMC accretion rate in both girls and boys indicated a dissociation between growth in bone size and growth in BMC during puberty. The dissociation between accretion rate in BA and BMC is also supported by a study showing that the fracture rate in children is highest around peak height velocity, and this could not be explained by increase in physical activity.28 Blimkieet al found that a lag in cortical bone thickness and mineralisation relative to linear skeletal growth appeared to be the predominant factors associated with the increased fracture incidence in Belgian boys during the growth spurt.29 This dissociation between growth in bone size and growth in BMC is in agreement with our longitudinal data. In summary, studies of both regional and whole body mineralisation corroborate the dissociation in time between growth in bone size and growth in BMC during puberty that was pointed out many years ago by Krabbeet al.30

Few longitudinal whole body DXA datasets have been published. Lloyd and colleagues31 published data from an intervention study of 12 year old girls in mid-puberty using a QDR Hologic 1000/W, as used in our study, although the software version was not mentioned. From their data, a BMC accretion rate of 257 g/year can be calculated, which is similar to the median ΔBMC for 12 year old girls in our study (255 g/year). In the study of Martin et al,6 in which bone measurements were obtained by a QDR Hologic 2000 (software version 5.56A), the peak accretion for ΔBA in boys and girls was 212 and 174 cm²/year, respectively, which is below our values for girls and a little above our values for boys. In the same study, the peak accretion for ΔBMC in boys and girls was 320 and 240 g/year, respectively, which is similar to our values for girls but above our values for boys. When our data were expressed according to pubertal stage, the values in pubertal stage III were similar to the peak values in the study by Martinet al,6 so the difference in accretion values according to age might be caused by differences in the timing of puberty between the two studies.

Our data on bone calcium accretion (table 2) showed that in normal children in mid-puberty, a relatively high calcium retention is needed. The bone calcium accretion values calculated from our DXA measurements are similar to the values obtained by others using balance studies. Jackman et al examined the relation between calcium retention and calcium intake in girls aged 12–14 years (mean, 12.7) and found an increase in retention with increase in calcium intake.4 The retention was measured by balance studies over two weeks. For a calcium intake of ∼ 1100 mg/day in that study, the calcium retention was ∼ 300 mg/day for girls at menarche. In our study, girls in Tanner stage III had a median calcium intake of 1146 mg/day32 and a median calcium retention of 220 mg/day over the following year, which means that the retention in shorter periods was probably considerably higher. Thus, in both balance and DXA studies, it has been shown that calcium retention is high in mid-puberty. Therefore, during puberty, children with a low calcium intake need a high calcium absorption rate to obtain an average bone mineralisation. Most children in Western countries who do not drink milk will have a very low calcium intake, typically 300–400 mg/day.33 If their calcium intake is below 500 mg/day, more than 50% of the ingested calcium should then be retained in the bones to obtain an average mineral accretion. It is not known to what extent children can adapt to a low calcium intake by increasing the absorption capacity, but the high calcium retention in normal puberty indicates that a low calcium intake during puberty may limit bone mineralisation. Whether a negative influence on bone mineralisation during puberty will also affect peak bone mass negatively is unknown because it has not been shown to what extent children can catch up later in adolescence. To examine this further, longitudinal studies are needed that follow children who have a varying calcium intake from before puberty to adulthood.

In conclusion, we have presented centile curves for annual bone area and bone mineral content accretion according to sex and age in school age children and adolescents. We found a very strong relation between pubertal stages and both bone size and bone mineral accretion. In puberty we observed a dissociation in time between the peak in bone size accretion and the peak in bone mineral accretion.