Familial resemblance and diversity in bone mass and strength in the population are established during the first year of postnatal life

Abstract

Familial resemblance and diversity in bone structure and strength in adulthood are determined in part during growth. Whether these characteristics are established during gestation or shortly after birth is not known. Total-body, lumbar spine, and femoral neck size and mass and indices of tibial bending strength and distal radial compressive strength were measured using bone densitometry and quantitative computed tomography in 236 girls at 18.5 years of age. Among them, 219, 141, and 105 girls had crown-heel length (CHL) and weight recorded at birth and at 6 and 12 months of age, and then height and weight were recorded at 3, 5, 10, 13, and 15 years of age in 181, 176, 127, 111, and 228 girls, respectively. Of these girls, 101 and 93 girls also had bone structure assessed at 11 and 13 years of age, respectively. Similar bone measurements were made once in 78 mother-father pairs. CHL and weight at birth did not correlate or did so weakly with bone traits in girls at 18 years of age. By contrast, CHL at 6 months correlated with the height, bone traits, and strength at puberty and at 18 years of age (r = 0.24–0.56, p < .001) in girls and with their parents' height and bone traits (r = 0.15–0.37, p < .05). When the girls' CHL at 6 months was stratified into quartiles, the absolute and relative differences in bone traits observed at puberty (∼11.5 years) were maintained as these traits tracked during the ensuing 7 years. Similarly, weight at 6 months correlated with the girls' bone traits at puberty and 18 years of age (r = 0.22–0.55, p < .05). During puberty and at 18 years of age, the girls' bone traits correlated with the corresponding traits in their parents (r = 0.32–0.43, p < .01). It is concluded that familial resemblance in bone structural strength and the position of an individual's bone traits relative to others in adulthood are likely to be established during the first year of life. Thus susceptibility to bone fragility late in life has its antecedents established early in life. © 2010 American Society for Bone and Mineral Research

Introduction

Research into the pathogenesis of bone fragility has concentrated on the amount of bone loss accompanying advancing age and menopause.1 However, bone fragility also depends on the underlying structure of cortical and trabecular bone assembled during growth.2 For example, bones constructed with thick cortices and thick and densely connected trabeculae are likely to better tolerate resorptive removal of bone than bones with thin cortices and thin, widely spaced trabeculae.3

These structural features—bone size and shape, cortical area and thickness, and trabecular number, thickness, and connectedness—and their diversity or variance in adulthood are largely due to genetic factors. This diversity in structure is first expressed during growth, probably before puberty.4–7 The ranking or percentile location of a person's individual bone trait within the population distribution is established at some time during growth, and this location may make an important contribution to its ranking in old age when falls and fractures occur.8

Whether a person's individual trait is assigned its ranking during intrauterine life or after birth is uncertain. If the position of a person's bone size, mass, density, or strength index is determined during intrauterine life and the trait then tracks during gestation, trait dimensions at birth should predict the trait location in adulthood. If tracking is established at some time after birth, trait dimensions at some time after birth, but not at birth, should predict trait dimensions in adulthood. The purpose of this study was to examine these alternatives and to explore the implications of the findings.

Methods

The participants included 236 girls in Jyväskylä, Finland. These girls were a subgroup of the Calcium and Exercise Study (CALEX) (n = 396) with measurements of bone size and structure at 18 years of age.9 Growth charts from birth to 16 to 20 years of age providing gestational age, crown-heel length (CHL), and weight at birth and 6 and 12 months of age and height and weight at 3, 5, 10, 13, 15, and 18 years of age were obtained from the School Health Care System after obtaining consent from the participants and their parents or guardians. There were 219 girls with CHL and weight documented at birth, 141 and 105 girls with records of CHL and weight at 6 and 12 months of age, and 181, 176, 127, 111, and 228 girls with records of height and weight measured within 5 months before or after the third, fifth, tenth, thirteenth, and fifteenth birthdays, respectively. There were 199 mothers and 118 fathers measured once and 78 families having both parents and the daughter with all the data available for analysis. Height of the girls during adolescence and height of their parents were measured using a stadiometer to the nearest 0.1 cm. Weight was measured using an electric scale to the nearest 0.1 kg.

Among the 236 girls with bone morphology assessed at 18 years of age (range 16 to 20 years), 101 and 93 had their bone assessed at 11 years of age (range 10 to 13 years) and at 13 years of age (range 12 to 15 years), respectively. Bone area and bone mineral content (BMC) of the total body (TB), left femoral neck (FN), and lumbar spine (L₂ to L₄) were measured using dual-energy X-ray absorptiometry (Prodigy, GE Lunar Corp., Madison, WI, USA, with software Version 9.3).9 Precision, expressed as a coefficient of variation (CV%), was less than 1% for TB and lumbar spine (LS) measurements and less than 1.8% for FN. The left tibial shaft was scanned using peripheral quantitative computed tomography (pQCT; XCT-2000, Stratec, Medizintechnik, GmbH, Pforzheim, Germany) at a point 60% of the lower leg length above the lateral malleolus.10 The left radius was scanned at the distal 4% of forearm length, defined as the distance between the ulna olecranon and the styloid process. Images were processed using the manufacturer's software (Version 5.40) and the Geanie 2.1 (Bonalyse Oy, Jyväskylä, Finland). Bone total cross-sectional area (CSA, ie, area within the periosteal surface) and bone strength index (BSI) were obtained. BSI, an index of bending strength of the tibial shaft, was derived as I × cBMD, where I is the total-bone polar moment of inertia (I_max + I_min) and cBMD is the cortical volumetric bone mineral density (vBMD).11, 12 the BSI of the distal radius, an index of compressive strength, was derived as CSA × (vBMD)², where CSA is radial total CSA and vBMD is total vBMD. The CV% for CSA and BSI for repeated measurements was less than 2%. The study was approved by the ethical committees of the University of Jyväskylä and the Central Hospital of Central Finland. Informed consent was provided by all subjects and their parents or guardians.

Data were all normally distributed. Pearson correlations were used to examine associations between anthropometric features in early life and bone traits in adolescence and young adulthood using all available data. Regression analysis was used to quantify the variance of traits in girls at 18 years of age explained by their parents' corresponding traits and their own CHL and weight at birth and at 6 months separately or combined. The girls' CHL and weight at birth and 6 months and height and weight at 18 years of age were ranked into percentiles at any given age, and then the changes in percentile location from birth to 18 years of age and 6 months to 18 years of age were described. Girls with complete data were categorized into quartiles according to their CHL (and weight) at birth or at 6 months of age for comparison of the group differences in height (and weight) from birth and bone traits during adolescence to demonstrate whether the four groups remain distinct from each other during growth. Results are reported as means ± SD unless otherwise stated. A p value of less than .05 was considered statistically significant.

Results

Of the 236 girls, 4.1% were born preterm (<32 to 37 weeks), and 3.0% were born small for gestational age. CHL and weight were 49.7 ± 1.8 cm and 3.5 ± 0.5 kg at birth, 66.0 ± 2.2 cm and 7.4 ± 0.8 kg at 6 months, and 165.5 ± 5.3 cm and 59.8 ± 9.8 kg at 18 years, respectively. The average height and weight were 165.3 ± 6.0 cm and 71.0 ± 13.4 kg in mothers and 177.2 ± 5.8 cm and 82.9 ± 9.6 kg in fathers, respectively. The mean values of bone parameters in the adolescent girls are given in Table 1.

Table 1. Anthropometry and Bone Traits in Adolescent Girls at 11, 13, and 18 Years (mean ± SD)

Vists:	First (n = 101)	Second (n = 93)	Third (n = 236)
Age (years)	11.3 ± 0.7	13.3 ± 0.7	18.3 ± 1.1
Height (cm)	145.8 ± 7.8	158.5 ± 6.9	165.8 ± 5.6
Weight (kg)	38.5 ± 8.4	48.8 ± 10.7	60.3 ± 9.9
Total-body BMC (g)	1400 ± 260	1891 ± 343	2458 ± 369
Total-body aBMD (g/cm^2)	0.947 ± 0.055	1.04 ± 0.08	1.16 ± 0.07
Total-body bone area (cm^2)	1471 ± 207	1803 ± 222	2108 ± 213
L2–L4 BMC (g)	23.3 ± 5.8	35.1 ± 8.5	49.1 ± 7.9
L2–L4 aBMD (g/cm^2)	0.821 ± 0.11	0.993 ± 0.14	1.20 ± 0.12
L2–L4 bone area (cm^2)	28.0 ± 3.8	34.9 ± 4.4	40.7 ± 3.9
Femoral neck BMC (g)	3.24 ± 0.54	4.04 ± 0.72	4.87 ± 0.67
Femoral neck aBMD (g/cm^2)	0.818 ± 0.10	0.93 ± 0.13	1.06 ± 0.14
Femoral neck bone area (cm^2)	3.84 ± 0.49	4.28 ± 0.41	4.59 ± 0.30
Distal radial CSA (mm^2)	232 ± 43	290 ± 49	312 ± 42
Distal radial BSI (10^3 mg^2/cm^4)	189 ± 43	240 ± 61	365 ± 73
Tibial shaft CSA (mm^2)	369 ± 53	412 ± 55	472 ± 56
Tibial shaft BSI (10^5 mg · cm)	18.4 ± 3.9	21.9 ± 4.2	27.6 ± 4.8

Trajectories of growth in body size are established at 6 months of age, not at birth

CHL at birth was weakly or not associated with height and total and regional bone size, mass, and strength at 18 years of age (r = 0.1–0.2). By contrast, CHL at 6 months and at any age thereafter significantly correlated with these traits (r = 0.24–0.57; Table 2 and Fig. 1). Only 17% of girls' height tracked within ±10 percentiles of their CHL ranking at birth, and 40% deviated from this position by more than ±30 percentiles during 18 years. By contrast, 36% of girls' height tracked within ±10 percentiles of their CHL ranking at 6 months, and only 20% deviated from this position by more than ±30 percentiles during 18 years.

Table 2. Correlation Coefficients (r) Between Bone Traits at 18 Years and Crown-Heel Length (CHL) and Weight at Birth and 6 Months of Age

Bone traits at 18 years	BMC			Bone area			BSI		Bone CSA
Early-life body size	TB	LS	FN	TB	LS	FN	Tibia	Radius	Tibia	Radius
Note: Boldface means p ≤ .05; otherwise, NS. BMC = bone mineral content; BSI = bone strength index; CSA = cross-sectional area; TB = total body; LS = lumbar spine 2–4; FN = femoral neck.
Gestational age	−0.05	−0.07	−0.07	−0.03	−0.05	0.01	−0.02	−0.05	−0.01	−0.05
CHL at birth (n = 219)	0.17	0.15	0.10	0.19	0.20	0.13	0.14	0.07	0.13	0.18
CHL at 6 months (n = 141)	0.55	0.48	0.48	0.57	0.49	0.39	0.38	0.33	0.39	0.24
Weight at birth (n = 219)	0.17	0.13	0.10	0.19	0.16	0.18	0.23	0.06	0.21	0.18
Weight at 6 months (n = 141)	0.53	0.41	0.39	0.55	0.33	0.36	0.38	0.31	0.40	0.23

Figure 1. The correlation coefficient (r) of total-body and regional BMC (left panel) and bone strength indices (BSI) of distal radius and tibial shaft (right panel) at 18 years of age versus CHL at birth and at 6 and 12 months or height thereafter (upper panel) and weight (lower panel) at various ages.

Similarly, weight at birth was weakly or not associated with weight (r = 0.13, NS) and total and regional bone traits at 18 years of age, whereas weight at 6 months of age significantly correlated with these traits (Table 2). Unlike height, correlation of weight in early life versus bone traits at 18 years of age strengthened with advancing age in childhood (Fig. 1). A similar degree of tracking as that observed for height from 6 months to 18 years of age was found for weight from 3 to 18 years of age.

There were no differences in height or bone traits at 11.5 or 18 years of age between girls ordered by CHL quartiles at birth (Figs. 2 and 3). By contrast, in girls ordered by CHL quartiles at 6 months of age, differences in height and bone traits (ie, total body, lumbar spine, and femoral neck and tibial and radial size and estimates of strength) between the lowest and highest quartiles were observed at 11.5 years, and these differences were maintained as these traits tracked during the ensuing 7 years (Figs. 2 and 4). Notably, the third and fourth quartiles' BSIs, but not other bone traits, converged at 18 years of age (Fig. 4).

Figure 2. (A) When girls were classified according to quartiles of crown-heel length (CHL) at birth, the resulting ranking of height quartiles converged to be indistinguishable as age advanced. (B) When classified according to CHL at 6 months, ranking of height remained distinct through childhood to 18 years of age. (C) Differences in height between quartiles of CHL at 6 months were maintained during puberty (··· = 1st; —— = 2nd; - - - = 3rd; === = 4th quartile; mean ± SE, n = 93).

Figure 3. Quartiles of crown-heel length (CHL) at birth did not predicted the ranking and tracking of total-body, femoral neck, and lumbar spine bone area (BA) and bone mineral content (BMC) and tibal and radial total cross-sectional area (CSA) and bone strength index (BSI) through adolescence (··· = 1st; —— = 2nd; - - - = 3rd; == = 4th quartiles. ^†Significantly different between 1st and 4th quartiles; mean ± SD, n = 93).

Figure 4. Quartiles of crown-heel length (CHL) at 6 months of age predicted the ranking and tracking of total-body, femoral neck, and lumbar spine bone area (BA) and bone mineral content (BMC) and tibal and radial total cross-sectional area (CSA) and bone strength index (BSI) through adolescence (··· = 1; —— = 2nd; - - - = 3rd; -- = 4th quartiles. ^†Significantly different between 1st and 4th quartiles; mean ± SD, N = 93).

Likewise, quartiles of girls ordered by weight at birth showed no differences in bone traits during adolescence (data not shown), but when ordered by weight at 6 months, those with weight in the highest quartile had significantly greater bone size and mass of total body, lumbar spine, and femoral neck and BSI of the radius and tibia at 18 years of age than girls with weight at 6 months in the lowest quartile (Fig. 5).

Figure 5. Quartiles of weight at 6 months of age predicted the ranking and tracking of total-body, femoral neck, and lumbar spine bone area (BA) and bone mineral content (BMC) and tibal and radial total cross-sectional area (CSA) and bone strength index (BSI) through adolescence (··· = 1st; —— = 2nd; - - - = 3rd; -- = 4th quartiles. ^†Significantly different between 1st and 4th quartiles; mean ± SD, n = 93).

The apple doesn't fall far from the tree

When the girls reached 18 years of age, their height and bone traits correlated with the corresponding traits in their parents (r = 0.32–0.43, p < .001). This familial resemblance in stature and bone traits first became detectable when the girls were 6 months of age. For example, CHL at 6 months correlated with their parents' height and bone traits (Table 3).

Table 3. Correlation Coefficients (r) of Girls' Crown-Heel Lengths (CHL) at Birth and 6 Months of Age Versus Their Parents' Height and Bone Traits

Mothers	Height	BMC			Bone area			BSI		Bone CSA
Girls		TB	LS	FN	TB	LS	FN	Tibia	Radius	Tibia	Radius
Boldface means p ≤ .05; otherwise, NS. BMC = bone mineral content; BSI = bone strength index; CSA = cross-sectional area; TB = total body; LS = lumbar spine 2–4; FN = femoral neck.
CHL at birth (n = 199)	0.21	0.18	0.17	0.12	0.19	0.09	0.08	0.05	0.13	0.05	0.14
CHL at 6 months (n = 141)	0.47	0.26	0.34	0.29	0.33	0.37	0.28	0.17	0.19	0.16	0.15
Fathers
Girls
CHL at birth (n = 118)	0.12	0.08	0.01	0.01	0.11	0.12	0.02	0.21	0.06	0.21	0.18
CHL at 6 months (n = 118)	0.27	0.26	0.29	0.29	0.29	0.44	0.24	0.32	0.00	0.36	0.31

In the regression analysis, the daughters' heights at 18 years were independently and equally predicted by their CHLs at 6 months and midparental heights (defined as half the sum of heights of father and mother) (Table 4). CHL at 6 months and midparental height each accounted for 36% of the variance in daughters' height at 18 years of age, whereas 53% of the total variance was explained when the two were combined (Table 4). Similarly, the daughters' total-body and femoral neck BMC values and tibial shaft BSI values at 18 years of age were independently predicted by their CHLs or weights at 6 months and the corresponding bone traits in each parent (Table 5).

Table 4. Regression Coefficients of Girls' Height at 18 Years Versus Their Crown-Heel Length (CHL) at 6 Months (Model 1), the Midparental Height (Half the Sum of Parents' Heights) (Model 2), Respectively, and the Two Combined (Model 3) (β ± SE)

n = 78 pairs		Girls' CHL at 6 months	Midparental height	r^2
Model 1: Only CHL at 6 months as independent variable. Model 2: Only midparental height as independent variable. Model 3: Both CHL at 6 months and midparental height as independent variables. Boldface means p ≤ 0.05; otherwise, NS. r^2 is the explained variance of the dependent variable (girls' height at 18 years) by the independent variable(s).
Girls' height at 18 years	Model 1	0.60 ± 0.04		0.36
	Model 2		0.61 ± 0.04	0.36
	Model 3	0.42 ± 0.04	0.45 ± 0.04	0.53

Table 5. Regression Coefficients of the Girls' Bone Traits at 18 Years of Age Versus Their CHL, Weight at 6 Months of Age (β ± SE), and/or the Same Traits in Their Parents

Girls' traits at 18 years (n = 78)		CHL at 6 months	Weight at 6 months	Mothers' same trait	Fathers' same trait	r^2
Models 1, 2, and 3: Only CHL, weight at 6 months, or parents' corresponding trait as independent variables. Model 4: CHL at 6 months and parents' corresponding trait as independent variable. Model 5: Weight at 6 months and parents' corresponding trait as independent variable. Boldface meansd p ≤ .05; otherwise, NS. r^2 is the explained variance of the dependent variable by independent variable(s).
Total-body BMC	Mode 1	0.55 ± 0.04				0.30
	Model 2		0.56 ± 0.10			0.31
	Model 3			0.30 ± 0.10	0.38 ± 0.10	0.24
	Model 4	0.50 ± 0.10		0.15 ± 0.10	0.36 ± 0.10	0.48
	Model 5		0.49 ± 0.09	0.21 ± 0.09	0.37 ± 0.09	0.49
Lumbar spine BMC	Mode 1	0.48 ± 0.07				0.23
	Model 2		0.39 ± 0.11			0.16
	Model 3			0.46 ± 0.10	0.36 ± 0.10	0.41
	Model 4	0.24 ± 0.10		0.35 ± 0.10	0.40 ± 0.10	0.56
	Model 5		0.18 ± 0.11	0.36 ± 0.11	0.40 ± 0.11	0.45
Femoral neck BMC	Mode 1	0.48 ± 0.08				0.23
	Model 2		0.44 ± 0.10			0.19
	Model 3			0.27 ± 0.10	0.37 ± 0.10	0.19
	Model 4	0.41 ± 0.11		0.14 ± 0.11	0.42 ± 0.11	0.44
	Model 5		0.23 ± 0.11	0.31± 0.11	0.44 ± 0.11	0.42
Tibial shaft BSI	Mode 1	0.38 ± 0.09				0.14
	Model 2		0.42 ± 0.11			0.18
	Model 3			0.28 ± 0.09	0.36 ± 0.09	0.19
	Model 4	0.38 ± 0.10		0.21 ± 0.10	0.29 ± 0.10	0.37
	Model 5		0.37 ± 0.10	0.27 ± 0.10	0.34 ± 0.10	0.38
Distal radial BSI	Mode 1	0.33 ± 0.09				0.11
	Model 2		0.25 ± 0.12			0.06
	Model 3			0.45 ± 0.09	0.21 ± 0.09	0.23
	Model 4	0.13 ± 0.11		0.50 ± 0.11	0.22 ± 0.11	0.25
	Model 5		0.04 ± 0.13	0.21 ± 0.15	0.41 ± 0.16	0.29

Discussion

We report that the ranking of an individual's CHL or height in the population distribution was achieved during the first 6 months of life, not during gestation. A similar degree of tracking in body weight was established, but later, at 3 years of age. CHL at 6 months and thereafter, but not at birth, predicted height and bone size, mass, and strength almost two decades later and predicted these traits in the parents. CHL or height then tracked in a trajectory of growth from 6 months of age through adolescence to young adulthood. This also applied to bone traits such as total and regional bone mass and size, tibial and radial cross-sectional area, and indices of bending and compressive strength first measured at 11.5 years of age and then during 7 years to maturity at 18 years of age. These observations suggest that the location of a person's individual skeletal trait relative to others and the familial resemblance of height and these bone traits at maturity appear during the first 6 months of postnatal life.

Several cross-sectional and longitudinal studies confirm that trait variances and the locations of a person's individual bone traits are established before puberty and track within their assigned trajectory.5, 7, 9, 13 For example, Loro and colleagues reported tracking in bone size and mass over 3 years in adolescents; an individual with a large vertebral or femoral shaft cross section or higher vertebral vBMD or femoral cortical area retained this position from prepuberty to maturity.7 Wang and colleagues reported that variances in tibial cross-sectional size, shape, and density in girls were established before puberty and were no less than in their mothers.5 These structures also tracked over 2 peripubertal years.5 Percentile ranking of total-body BMC was maintained during 7 years in adolescence; 77% of the 236 eighteen-year-old girls had total-body BMC values within ±10 percentiles of their locations 7 years earlier.9 Traits also track during adulthood to old age. Garn and colleagues monitored 744 women and men in whom 90% of the variance in cortical thickness in adulthood was present at maturity.14 Emaus and colleagues reported that distal and ultradistal radial BMD measured by single-energy X-ray absorptiometry tracked during 7 years of follow-up in 5366 women and men aged 45 to 84 years; the correlation coefficients of two measurements 7 years apart from each other werer greater than 0.9.15

None of the preceding studies identified whether the trajectory of growth was established during gestation or after birth. Bjørnerem and colleagues reported that femoral length deviated from tracking by one or more quartiles in 87% of fetuses during gestation.16 Smith and colleagues implied that tracking occurred during gestation because smaller-than-expected crown-rump length in the first trimester doubled the risk for a low weight at birth.17 However, only 38 of 1289 smaller-than-expected fetuses remained small throughout gestation, whereas 1251 did not, suggesting that tracking is uncommon during gestation.

The observations that CHL and weight at 6 months, but not at birth, correlated with skeletal traits at puberty and adulthood in these girls and in their parents and that the strength of the correlations persisted from 6 months of age on for height and from 3 years on for weight support the view that tracking is established by growth after birth and not during gestation. Several studies suggest that variance and tracking in CHL or height and bone traits are established in early postnatal life. Pietilainen and colleagues surmised that tracking of body length occurred from birth to young adulthood. However, birth length correlated weakly (r = 0.3) but height at 3 years of age correlated strongly with adult height (r = 0.8).18 In a twin study, Silventoinen and colleagues reported that heritability of CHL or height was little at birth (16%), whereas it was virtually entirely established at 2 years of age (91% versus 94% at maturity).19 While expression of heritability in twins may differ from that in singletons owing to difference in placental membrane and vascular structure leading to nutritional imbalances between the twins,19 this report and the data in twins are consistent. Oliver and colleagues and Javaid and colleagues reported that weight at 1 year correlated with bone strength of the radius and tibia in adulthood.20, 21 One prospective study reported that robusticity (cross-sectional area/length) of the second metacarpal at 3 months of age predicted robusticity at 8 years (r = ∼0.6).22 We confirm these observations; tracking of body size and bone traits is low at birth but is rapidly established in the first year of postnatal life.

As reported by Javaid and colleagues and Oliver and colleagues, body weight at 6 months, not at birth, predicted bone traits in young adulthood.20, 21 However, unlike height, the correlations between weight in early life and bone traits in young adulthood strengthened as age advanced, perhaps reflecting the greater contribution of environmental factors to variance in weight than it does to variance in height in industrialized countries.23

No single cohort study has monitored the progress of a trait throughout intrauterine and postnatal life to old age. Linking trait trajectories during gestation to the end of life requires that inferences are made from information of a piecemeal nature. The published literature and this prospective study suggest that deviation from tracking is common during gestation but that ranking of traits such as height, skeletal size, and mass is largely established during the first year of life, and these traits then track in their assigned trajectory of growth through childhood, adolescence, and puberty into adulthood.

The growth-related origin of bone morphology in adulthood and old age is supported by many studies. Healthy premenopausal daughters of women with spine or hip fractures have structural abnormalities at the corresponding site.24 Girls with forearm fractures during puberty have reduced distal radius vBMD before puberty, and this deficit in vBMD persisted into young adulthood.25 Exercise in children leaves structural benefits in old age.26 Because of temporal differences in upper and lower body growth, onset of anorexia nervosa before puberty produces deficits in appendicular bone mass, whereas onset during puberty produces deficits in the axial skeleton.27

This study extends the preceding observations by suggesting that expression of structural diversity in the population and of familial resemblance in bone traits appear in the first 6 months of postnatal life. Three observations support this inference: CHL at 6 months of age, but not at birth, (1) predicted the parents' height and bone traits, (2) predicted an individual's height and bone traits at puberty and at 18 years of age, and (3) predicted their height and bone traits at 18 years of age independent of familial factors such as their parents' corresponding traits.

The appearance of familial resemblance and the diversity of height and bone traits at 6 months, but not at birth, may be partly the result of the release from maternal constraint, which permits the expression of maternal, paternal, and individual genetic factors that determine postnatal growth. Finding an association between CHL at 6 months and height and bone traits at 18 years of age independent of any parental contribution supports the role of factors specific to the individual that influence height and bone traits at 18 years of age. Since growth is rapid, exposure to adverse environmental factors, for example, malnutrition, illness, or risk factors, may alter the trajectory of growth, which, even if small, can alter the ranking of an individual's trait in adulthood. For example, Cooper and colleagues report that in over 7000 men and women, the risk of hip fracture nearly doubled in those whose rate of childhood height gain was in the lowest quartile compared with those whose growth rate was in the highest quartile.28

This study has several limitations. First, we did not measure these girls' bone traits in early life. Tracking was inferred from the correlation between CHL and weight in early life and bone traits in adolescence because CHL or height predicts bone traits throughout childhood with similar strength.29–31 Indeed, CHL at 6 months and height at 18 years equally predicted bone traits at 18 years. Thus we infer that the bone traits track as strongly as stature. Second, the data derived from growth charts were retrospective. We did not have quantitative assessment of environmental factors (ie, physical activity, nutrition, and diseases history) in early life. Thus we were unable to assess whether these environmental factors in early life explain some of the variance in bone traits at 18 years. In addition, the study also included a few participants born preterm and small for gestational age. However, although born small, they were not shorter and had no deficit in the examined bone traits at 18 years of age than those who were born full term,32 suggesting that the size at birth explains a small amount, if any, of the variance in bone traits at 18 years. Finally, we have no explanation for the convergence of bone strength of the radius and tibia at 18 years of age in the third and fourth quartiles of CHL at 6 months.

We conclude that growth during the first year of life, not gestation, establishes the familial resemblance in bone mass and strength and thus their ranking relative to others in adulthood. Susceptibility to bone fragility late in life may have its antecedents early in life.

Disclosures

The study sponsors played no role in the design, methods, data management, or analysis, nor in the decision to publish. All the authors state that they have no conflicts of interest.

Acknowledgements

We thank the research staff—Drs Patrick HF Nicholson and Aiping Zheng, Mrs Shu Mei Cheng and Heli Vertamo, Ms Sirpa Mäkinen, and Mr. Erkki Helkal—for their valuable contributions. This study was supported by grants from the Academy of Finland, the Finnish Ministry of Education, the University of Jyväskylä, and the Juho Vainion Foundation. The study was made possible, in part, by the Bridge Funding Grant Award of ASBMR 2006 to SC and by a grant to QW, the recipient of ESCEO-Amgen fellowship award, in 2008.