Sexual Dimorphism and the Origins of Human Spinal Health

Abstract

Recent observations indicate that the cross-sectional area (CSA) of vertebral bodies is on average 10% smaller in healthy newborn girls than in newborn boys, a striking difference that increases during infancy and puberty and is greatest by the time of sexual and skeletal maturity. The smaller CSA of female vertebrae is associated with greater spinal flexibility and could represent the human adaptation to fetal load in bipedal posture. Unfortunately, it also imparts a mechanical disadvantage that increases stress within the vertebrae for all physical activities. This review summarizes the potential endocrine, genetic, and environmental determinants of vertebral cross-sectional growth and current knowledge of the association between the small female vertebrae and greater risk for a broad array of spinal conditions across the lifespan.

The smaller female vertebra is associated with greater spinal flexibility and lesser vertebral strength, probably contributing to a broad array of spinal conditions across the lifespan.

Essential Points

• Factors related to sex-based differences in fetal skeletal development

• Newborn girls are born with smaller vertebral cross-sectional dimensions than newborn boys, which could represent the human adaptation to fetal load in bipedal posture

• The greater spinal flexibility and lesser vertebral strength associated with smaller female vertebrae are likely determinants of the risk for spinal deformities in girls and vertebral fractures in older women

Among the main areas of progress in osteoporosis research is the recognition that this condition may have its antecedents in intrauterine development (1, 2). Support for the notion that osteoporosis may result from perturbations in the fetal programming of skeletal growth is based primarily on epidemiological data showing a relationship between low birth size and lower bone mass later in life (1–11). Interestingly, the association between birth weight and adult bone mass in a cohort of female twins appeared to be mediated by skeletal size (3). The importance of the fetal environment to skeletal development is further substantiated by reports that maternal smoking, nutrition, vitamin supplementation, and physical activity are linked to offspring’s bone mass (1, 12–19). However, whether birth weight predicts future fracture risk is the subject of interest and controversy (1, 8, 20–24).

Despite a broad agreement that the fetal environment shapes one’s future health, the newborn phenotypes linked to bone disease risk are unknown. As the field of imaging progresses, our understanding of fetal development is making important strides. Advances in magnetic resonance imaging (MRI) allow detailed measures of skeletal structure in healthy infants without the need for sedation, contrast, or radiation exposure (25–27). Using MRI, we recently found that newborn girls have smaller vertebrae than boys of the same length and weight (27), a striking sexual dimorphism that could represent the female adaptation to fetal load in bipedal posture and probably has major implications for women’s spinal health (27).

This review summarizes current knowledge on the differential effects of sex on human vertebral growth, spinal flexibility, and strength and their possible endocrine, genetic, and environmental determinants. It also highlights data supporting the notion that the greater prevalence of spinal deformities in girls and vertebral fractures in older women are related to small vertebrae and the downside of walking upright. Notably, although current research supports the presence of sex differences in vertebral cross-sectional area (CSA), knowledge of their clinical implications is limited, and the mechanisms for this sexual dimorphism are yet to be defined.

Bipedal Locomotion, Pregnancy, and the Evolution of the Female Spine

Erect posture and bipedal locomotion are defining features of humans associated with unique skeletal morphogenesis originating about 2 million years ago (28–36). Unlike in other primates, the bones of the lower extremities in humans evolved to be stronger and larger, with greater joint surface areas than those in the upper extremities (29). Longer legs allowed early hunter-gatherers greater stride length and increased speed and endurance for long-distance walking or running (29). The high percentage of slow-twitch muscle fibers necessary for endurance running may have derived in humans from a genetic mutation of the ACTN3 gene (37).

In humans, the axial skeleton also developed distinct morphological traits that allow the greater spinal mobility needed for orthograde bipedality (38–42), adaptations that are most relevant to the woman’s need to compromise between bipedal locomotion and pregnancy. During gestation, women must maintain bipedal posture and balance despite body weight increases of 15% to 25% (43–46). After 12 weeks of gestation, the uterus can no longer be contained within the pelvis and moves superiorly and anteriorly into the abdomen. To compensate for this bipedal obstetric load and counteract the shift in the center of mass, pregnant mothers habitually extend their spine and increase the degree of lumbar lordosis (LL) (Fig. 1) (42, 47, 48).

Figure 1.

(a) Schematic representation of increasing lumbar lordosis during pregnancy to compensate for fetal load and maintain a stable position of the center of mass (black and white circle with crosshairs). Adapted from (28). (b) MRI image of a 30-year-old woman at 25 weeks’ gestation showing marked lordosis. Figure 1a adapted with permission from Macmillan Publishers Ltd: Whitcome KA, Shapiro LJ, Lieberman DE. Fetal load and evolution of lumbar lordosis in bipedal hominins. Nature. 2007; 450:7172.

Relaxin, a peptide of the insulin-like growth factor (IGF) family secreted by the corpus luteum and placenta, exerts a regulatory effect on the musculoskeletal and other systems through binding to its receptor in various tissues (49–51). Serum levels of this hormone increase 10-fold during the first trimester, remaining high until late pregnancy, and then become serologically undetectable in the first few days postpartum (52). Relaxin alters the properties of cartilage and tendon by activating collagenase and facilitates the spine and pelvic flexibility necessary to accommodate the enlarging uterus. In the lumbar spine, joint laxity is most notable in the anterior and posterior longitudinal ligaments (35, 53–55). Both joint laxity and the degree of LL are less pronounced in nulliparous than multiparous women and are associated with the number of pregnancies (56, 57). The degree to which increased LL during pregnancy is the consequence of ligamentous laxity or an adaptive trait evolved by natural selection is unknown.

Although variations in lumbar vertebral morphogenesis, such as vertebral wedging and orientation of lumbar facets, have been suggested to facilitate LL (42, 45, 58), the structural basis for the greater LL in women compared with men has been poorly understood. Our limited understanding of the structural phenotypes linked to upright posture stems from a lack of animal models that can replicate human bipedalism. The mammalian vertebral column is highly variable, reflecting a wide range of adaptations to differences in lifestyles, from natural burrowers to flying bats. Even the axial skeleton of nonhuman apes lacks the lordosis and range of motion needed for the fully erect posture observed in modern humans (29, 42, 59). Although mice and rat models are useful in bone metabolism research, the growth plates of their long bones remain open throughout their lifespan, allowing continuous skeletal growth and modeling (60–62).

Sexual Dimorphism in Axial Skeletal Development

Animal investigations indicate that skeletal development begins in the embryo and is mediated by multiple signaling pathways, including Hox genes, Wnts, Hedgehogs, bone morphogenetic proteins, fibroblast growth factors, Notch/Delta, and others (63–66). Mesenchymal cells in chick and mice embryos are laid down in spatiotemporal patterns where bones of the axial and appendicular skeletons will form (67). In humans, most osteochondral progenitors become chondroblasts that initiate endochondral bone formation between the eighth and 12th week of gestation, but it is not until the third trimester that the majority of bone mineralization occurs (68, 69). Growth of the appendicular and axial skeletons results from two different processes, probably regulated by different means (70–73). In the extremities, growth in bone length occurs by endochondral bone formation at the growth plates, whereas increases in bone width occur by apposition of subperiosteal bone, a process that begins before birth and continues throughout life (73). In the vertebrae, growth occurs by endochondral ossification, which commences in the central area of the cartilage anlage and expands toward the periphery in all directions (71, 72).

The human skeleton exhibits noticeable sexual dimorphisms during intrauterine development. In the appendicular skeleton, these differences are related predominantly to the tempo and degree of skeletal maturation. Epiphyseal ossification is more advanced in girls than in boys during the third trimester of gestation, and a greater number of ossification centers are depicted in radiographs of preterm and full-term newborn girls than boys (74). Throughout childhood, the rate of skeletal maturation remains more advanced in girls, who on average achieve skeletal maturity and cessation of longitudinal growth 2 years earlier than boys (72, 75, 76).

Whereas the differences in appendicular skeletal maturation between girls and boys have been known for more than a century (77), we only recently learned that factors related to sex also program fetal vertebral development (27, 77). Previous studies examining sex differences in axial skeletal mass at birth used projection techniques that did not account for the confounding effect of vertebral size and yielded discrepant results. Although most found that sex had no effect on lumbar spine bone mass (78–81), one showed that boys had higher bone mass than girls but did not adjust for differences in body size or examine bone morphology (82). Based on three-dimensional assessments of skeletal structure, recent observations indicate that the CSAs of vertebral bodies are on average 10% smaller in healthy newborn girls than in newborn boys, a difference that was independent of gestational age, birth weight, body length, and vertebral height (Fig. 2) (27). In contrast, sex did not influence intrauterine cross-sectional growth of the appendicular skeleton, and neither the CSA nor the cortical bone area at the midshaft of the humerus differs between boys and girls (27). Because smaller vertebral cross-sectional dimensions render the female spine more flexible, a proposed explanation for the sexual dimorphism is that it improves maternal performance in posture and locomotion (27). Studies are needed to establish the generalizability of these sex differences in the cross-sectional dimensions of the axial but not appendicular skeletons in other newborn populations.

Figure 2.

Representation of plasma testosterone levels in the male fetus, infant, and adolescent (83), with overlying values for vertebral CSA (mean ± standard deviation) displaying increasing sex differences from birth to prepubertal and postpubertal stages. Female values are ~10% smaller at birth, 15% smaller during childhood, and 25% smaller at sexual maturity and young adulthood (27, 93, 134).

Although the mechanisms responsible for the smaller female vertebral body are unknown, it probably results from complex interactions involving sex steroids, growth hormone (GH), and IGF (84). Both insulin and IGF-2 are known regulators of fetal growth (85), but no investigations to date have revealed sexual dimorphism in these values. Estrogens and androgens also promote bone accretion; estrogens are particularly key for the regulation of epiphyseal function, whereas androgens promote periosteal new bone formation, which has a dramatic effect on the width of the bone (86–89). Although the absence of the SRY gene in girls may dictate the sexual dimorphism in vertebral development, the important downstream events warrant investigation. Careful evaluations of newborns with disorders of sex development, such as girls with congenital adrenal hyperplasia, could aid in deciphering the role of androgens in mediating the fetal development of the axial skeleton. Conversely, testicular feminization syndromes may demonstrate the opposite effect (90).

Sexual Dimorphism in Postnatal Axial Skeletal Development

Infancy

Infancy is a developmental stage associated with marked gains in bone and muscle. Remarkably, the sexual dimorphism in vertebral size further increases during infancy. On average, vertebral CSA in girls is 10% smaller than in boys at birth and 15% smaller at 6 months of age (27, 91), a discrepancy similar to that previously observed in prepubertal children (Fig. 2) (92, 93). This differential effect of sex on vertebral cross-sectional growth has been reported in children of European, Mexican, and African descent (92–94).

In the first few months of life, infants experience a transient activation of the hypothalamo-pituitary-gonadal axis and go through a “mini-puberty” (Fig. 2) (95–104). In male infants, testosterone concentrations can reach values that are similar to those found in men (95–97). After this postnatal surge, androgen production decreases to remain at prepubertal levels from 6 months of age until adolescence (96). In female infants, an increase in both estradiol and testosterone also occurs; however, the rise in testosterone is lower than that in boys (103, 104). The mini puberty affects the reproductive organs, causing penile and prostate growth in boys and breast enlargement in girls (90). Ultrasound measures of testicular, ovarian, and uterine volumes are also larger in the first 6 months of life than at 1 year of age (105–107). The effects of this infant gonadal activation are not confined to the reproductive organs. The greater rise in testosterone in the infant boy has been suggested to be responsible for his increased skeletal growth (108). Boys grow faster than girls during the first 6 months of life, and the greatest sex difference in growth velocity is observed concomitantly with the peak of postnatal gonadal activation (108). Old investigations in newborn male monkeys showed that blockage of the neonatal secretion of gonadotropins and testosterone results in diminished muscle mass and bone size in the axial and appendicular skeleton (109).

Infants experience remarkable skeletal development despite being largely motionless and spending most of the day asleep; this development is particularly notable in the axial skeleton (110–112). The maintenance of skeletal integrity in the presence of relative inactivity provides a strong case for an unidentified mechanism that regulates bone metabolism. Emerging data from animal investigations support a relationship between brown adipose tissue (BAT) and bone mass (113–121). Impairment in BAT function or ablation of beige fat in mice leads to development of low bone mass (118, 119). It has also been suggested that BAT influences skeletal metabolism by modulating the activity of the sympathetic nervous system (SNS) (113, 117, 118). In mice, uncoupling protein 1 (UCP1) activity exerts a protective effect on bone mass, possibly through alterations in central neuropeptide Y pathways known to regulate SNS activity (119, 122).

Additionally, a recent study found that a neutralizing antibody against follicle-stimulating hormone (FSH) increases cellular mitochondrial density, promotes brown/beige fat thermogenesis, and prevents muscle and bone loss in mice (121). Notably, human anatomical studies indicate that BAT is established in fetuses within the fifth month of gestation (123, 124), simultaneously with a decrease in fetal FSH levels (96). At the time of birth, BAT abundance peaks, as reflected by levels of UCP1, before declining concomitantly with a period of marked increases in FSH during the first 3 months of postnatal life (96). Because the greatest amounts of BAT are present at birth, followed by a period of marked bone accretion (91, 125, 126), the notion that BAT function would either directly or indirectly influence infant skeletal metabolism is a hypothesis in need of testing.

Adolescence

Variations in the accumulation and distribution of fat and muscle are important contributors to the different sexual characteristics of humans, which are already present before puberty. Compared with prepubertal boys, prepubertal girls have greater subcutaneous fat but less musculature, shorter legs, and smaller vertebral cross-sectional dimensions (93, 127). Increases in cortical thickness and CSA of the long bones in the lower extremities are driven primarily by mechanical stresses associated with increasing weight (128–132), consistent with analytic models suggesting that muscle activity related to the maintenance of upright position is a major determinant of cross-sectional bone growth. Although there are also strong correlations between weight and vertebral CSA in boys and girls, differences in weight and musculature account for a small part of the sex-related variations in vertebral CSA (93, 133). Therefore, factors other than mechanical stresses and related to sex have a role in the regulation of vertebral cross-sectional growth.

Throughout childhood, the CSA of the vertebral bodies in prepubertal girls is ~15% smaller than that in boys matched for age, height, and weight (92), a sexual dimorphism that further increases during puberty and is greatest at sexual and skeletal maturity, when the vertebral cross-sectional dimensions are ~25% smaller in women than in men, even after differences in body size are considered (Fig. 2) (27, 92, 93, 129, 134).

Puberty is the time of life when the greatest skeletal growth and bone accrual occur (135–141). The high activity of osteoblasts and osteoclasts during puberty is documented by the rise of bone formation and bone resorption markers in blood and urine (136, 138, 140, 142–146). Increases in both activities occur in the early stages of pubertal development in girls and reach their zenith in midpuberty. The concentration of bone metabolism markers decreases rapidly during the later stages of puberty. Conversely, the rise observed in boys is progressive and reaches its peak at later stages of pubertal development, showing a slow decrease thereafter (136, 142, 143, 146). These observations are in line with the different tempo of growth in boys and girls. Vertebral cancellous bone density does not differ in male and female subjects before the onset of puberty or by the end of the second decade of life (Fig. 3) (147). However, increases during sexual development begin and reach peak values earlier in girls than in boys (147–149). Interestingly, sex differences in bone accrual parallel differences in the rate of peak height velocities (147, 150). Peak growth velocity in boys typically is reached 2 to 3 years later, and boys continue growing for ~2 to 3 years longer than do girls (140, 150–153).

Figure 3.

Computed tomography measures of vertebral bone density in male (blue) and female (red) subjects, aged 7 to 20 years, showing no sex differences in bone density during the first decade of life and in young adulthood. Although gains in bone density during sexual development are similar for both sexes, the increase begins and reaches peak values at an earlier age in female subjects, consistent with differences in pubertal timing. Data from (147).

Several pieces of evidence suggest that differences in sex steroid actions on bone are associated with sex-specific effects of GH and its downstream effector IGF-1 on bone cells. GH directly stimulates osteoblastogenesis and bone formation (154–156). GH is secreted by the pituitary gland in a sex-dependent fashion. The female pituitary secretes GH continuously, with frequent peaks and short GH-free intervals. Conversely, male GH peaks are higher and less frequent, and the GH-free intervals last longer. These sex differences of GH secretion lead to sex dimorphism in liver gene expression that depends on STAT5b, the GH receptor downstream mediator (157–159). IGF-1 influences both linear and cross-sectional growth and exerts a direct action on growth plate chondrocytes and osteoblasts responsible for building cortical and cancellous bone (160–167). IGF-1 also promotes the synthesis of RANK-L, which is responsible for osteoclast differentiation and activation (168–170). Moreover, osteoblasts express IGF-1 under parathyroid stimulation, and several other hormones, including thyroid hormone and estrogen, induce paracrine IGF-1 secretion by osteoblasts, whereas glucocorticoids inhibit it (171–179). However, the complex interactions between sex steroids and the GH–IGF-1 system remain to be fully delineated in humans (136, 140, 180–185). In animal studies, low concentrations of estrogens stimulate the hepatic production of IGF-1, whereas large concentrations exert an inhibitory effect (180, 186). Androgens act primarily at the pituitary level but only after being converted into estrogens by the enzymatic activity of aromatase (180).

BAT was thought to disappear after infancy. However, recent data suggest that BAT is often present in children and that both the amount and the activity of BAT increase during puberty (91, 126, 140, 141, 187–191). Notably, pubertal gains in BAT during adolescence, like those for muscle, are generally greater in boys than in girls (189). Several independent groups have also reported a positive association between BAT and the amount of bone in adolescents and adults (183, 192, 193); this association may be due in part to temporal changes in endocrine secretory factors (such as IGFBP2 and Wnt10b) or confounding from the strong positive relationship of brown fat to muscle mass (183, 184, 192). BAT volume has also been associated with long bone structure (183, 184, 192). In adolescence, BAT volume is related to femoral CSA and cortical bone area (192). Additionally, young women with BAT also have thicker femoral cortices and lower Pref-1 compared with women without BAT, suggesting that BAT may be involved in the regulation of stem cell differentiation into the bone lineage at the expense of adipogenesis (183). Whether BAT is also related to vertebral cross-sectional growth is unknown.

Adulthood

The amount of bone that is gained during adolescence is the main contributor to peak bone mass (PBM) in young adulthood, which in turn is a major determinant of the risk for osteoporosis and fragility fractures in older adults (141, 185, 194–200). The time of life in which peak bone mass in the axial skeleton is attained has been the subject of controversy, with estimates based on projection techniques ranging from the third to the fifth decade of life (137). However, spinal PBM as measured by computed tomography (CT) is achieved by the time of sexual and skeletal maturity, when the amount of bone in the vertebral body and vertebral size reach peak values (Fig. 3) (147, 195, 196).

According to a recent longitudinal study, measures of vertebral volume, bone mineral density (BMD), and bone mineral content by CT scan in 50 girls at skeletal maturity remained unchanged 3 years later (137). Whether sex differences in vertebral size further increase during adulthood is controversial. Although most data indicate that little or no bone is gained from the periosteal surface of vertebral bone in adult women and the overall CSA of their vertebrae remains stable (137, 201), some studies suggest male vertebrae increase in size over life (201–205). Unlike vertebral size, cancellous bone density, the other major indicator of vertebral strength, starts to decrease in both sexes by early adulthood (193, 205–211). Previous histological examinations of vertebrae and iliac crest showed evidence that the loss of cancellous BMD may occur as early as the third decade (206–208, 210).

The basis for the lower PBM in the axial skeleton of women lies in the smaller female vertebrae because differences in vertebral bone density are less salient or nonexistent (134, 212–214). Even after differences in body size are accounted for, the CSA of vertebral bodies is 25% smaller in women than in men (134).

Measures of vertebral bone density and CSA track through adolescence until the time of PBM (215). Values for vertebral CSA at the beginning of puberty in both girls and boys accounted for ~90% of the variations seen at sexual maturity 3 years later (Fig. 4) (215). Tracking for this phenotype is as strong as that observed for standing height (215), knowledge that highlights our potential ability to determine which children are prone to develop low values for vertebral CSA at skeletal maturity. Dual-energy X-ray absorptiometry (DXA) measures of bone density and mineral content also track through childhood until skeletal maturity; this is true in the axial and appendicular skeleton and for both male and female subjects (216, 217).

Figure 4.

Longitudinal measurements of vertebral CSA in 20 girls from the commencement of puberty (Tanner stage 2) to its completion (Tanner stage 5). Yellow lines represent changes in values for the children in the lowest baseline quartiles; purple and green lines depict changes in values for the girls in the middle quartiles at baseline; blue lines represent changes in values for the girls in the highest baseline quartiles. Values are shown (a) for each girl and (b) for each quartile. The regression lines across Tanner stages for this phenotype differed between quartiles and did not overlap. Adapted with permission from Loro ML, Sayre J, Roe TF, Goran MI, Kaufman FR, Gilsanz V. Early identification of children predisposed to low peak bone mass and osteoporosis later in life. J Clin Endocrinol Metab. 2000;85:3908–3918.

Accumulating evidence suggests that marrow mesenchymal stem cells differentiate into bone or fat through alternative activation of mutually exclusive transcriptional programs (218–222) and that bone marrow adiposity plays a critical role in affecting bone quantity (223). In healthy teenagers and young adults, the amount of bone in the axial and appendicular skeleton is negatively related to marrow adiposity measured by MRI, a reciprocal relation independent of body size and present in both sexes (224, 225). Moreover, a prospective longitudinal study in young women found that increases in femoral cortical bone thickness as measured by CT scans, were inversely related to decreases in marrow adiposity (226). To date, no study has examined the potential link between marrow adiposity and vertebral CSA.

Other Potential Determinants of Vertebral CSA

Genetic factors

Heredity is a well-known determinant of skeletal bone mass, as measured with dual photon or X-ray absorptiometry. Convergent data from sibling pairs and twin studies have estimated the heritability of spinal bone mass to be 80% of its variance (227–233). Additional support for this genetic influence comes from studies showing low lumbar bone mass in daughters of women with osteoporosis and in first-degree relatives of patients with osteoporosis (229, 234). Because measures of spinal bone mass, based on projection techniques, are greatly influenced by bone size, the reported associations in bone mass could reflect the confounding effect of vertebral geometry (235–238). Earlier candidate gene and family studies identified the vitamin D receptor, type I collagen alpha 1, and low-density lipoprotein receptor–related protein 5 as determinants of DXA measurements of BMD (239–243). In the last decade, genomewide association studies have identified >60 loci associated with BMD (244); many are also associated with bone values in childhood (245–247). However, these loci have small effect sizes and collectively explain only ~6% of the variance associated with BMD, and many more loci remain to be discovered (244). Recently, the rare variant EN1 and a common variant near SOX6 have been associated with bone mass in both childhood (246) and adulthood (245); the smaller effect sizes observed in the adult study (245) suggest that the genetic influence on bone properties is lower in adulthood than in childhood.

The obvious application of genetic studies on vertebral structure is the discovery of markers that would allow early identification of subjects at risk for spinal disease. Spondylolysis (248, 249), adolescent idiopathic scoliosis (AIS) (250, 251), and vertebral fragility fractures (244) are polygenic diseases affecting millions of people. Though widely investigated by linkage analysis and candidate gene association studies, the genetic basis of these complex diseases is still undefined. A better characterization of the phenotypes and structural basis for these deformities would greatly aid in the identification of disease predisposition genes. Greater understanding of the interactions between genetic predisposition and modifiable factors, such as physical activity and diet, in determining axial skeletal modeling would also provide new ways to prevent spinal disorders.

Behavioral and environmental factors

Bone accrual begins before birth with genetic programming and is altered by epigenetic modifications (252, 253). Dietary alterations in humans can affect that genetic program, leading to subsequent changes in skeletal mass (230, 232, 254–257). Although nutritional variables are important modifiable factors that influence bone health, the effect they have on vertebral morphogenesis is unknown. A 6-year longitudinal assessment of calcium intake on bone in a diverse cohort of 1743 children found that dietary calcium had a positive effect, after adjustment for age, height velocity, and physical activity, on bone accrual at the lumbar spine in girls (258). Because most growth-related increases in bone mass observed in DXA pediatric values are caused by increases in vertebral size, the reported gains could, in part, reflect the confounding effect of skeletal geometry on DXA measures (237).

Although the beneficial effects of nutrition on vertebral cross-sectional growth are yet to be defined, the influence of exercise on axial skeletal development is better documented. Multiple observational and interventional studies suggest that weight-bearing activities are associated with high or increased bone mass in the axial skeleton (257, 259–264). Emerging evidence suggests that exercise may increase vertebral dimensions (265–267). Using physical activity records from the Northern Finland Birth Cohort 1966, Oura et al. (266) reported that high exercise levels from 14 to 46 years of age were associated with larger MRI values for vertebral CSA at 47 years of age. Additionally, young women who participated in physical activity four or more times a week had larger vertebral CSAs in the perimenopausal years (266). Another study using anterior and lateral DXA scans found that gymnastic exposure in girls was associated with larger vertebral cross-sectional geometry (265). This benefit was observed even in girls of low gymnastic experience and years after they discontinued gymnastics. Thus, the effect of physical activity on the growing axial skeleton implicates vertebral geometric adaptation to mechanical loading that may persist well beyond activity cessation into adulthood (265, 266, 268).

Archaeological data indicate that the strength and CSA of the axial and appendicular skeleton have decreased over time alongside technological development (269, 270). Using MRI and specimens from Swedish and British bone collections, Junno et al. (270) found the CSA of the lumbar vertebra to be smaller and the vertebral height taller today than in medieval times. Higher levels of physical activity leading to bone adaptation and potential genetic selection pressures favoring more robust bone structure probably account for differences in vertebral CSA (270). A physically less demanding lifestyle could have led to selection for more fragile vertebrae.

Clinical Implications of the Female Vertebral CSA for the Growing Skeleton

The spine provides flexibility, support, and protection of the spinal cord and nerves. Phenotypic variations that have evolved to be beneficial in one set of circumstances can be detrimental in other conditions. The smaller cross-sectional dimensions of the female vertebral body, though probably facilitating the lordosis needed during pregnancy (42, 47, 48, 271), could also increase the risk for a broad array of spinal conditions across the lifespan, such as exaggerated lordosis (272), spondylolysis (273), scoliosis (274, 275), vertebral wedging (276, 277), and vertebral fractures (134, 214, 278, 279) (Table 1).

Table 1.

Clinical Studies Assessing Vertebral CSA in Relation to LL, Spondylolysis, Scoliosis, and Vertebral Wedging

Study (Ref.)	n	Age (y)	Imaging Modality	Outcomes	Findings
Wren et al., 2017 (272)	40 girls and 40 boys	9–13	MRI	Vertebral CSA	Significant negative correlation between vertebral CSA and LL in girls and boys.
				LL	Girls have significantly smaller vertebral CSA and greater degree of LL.
Wren et al., 2017 (273)	Spondylolysis: 16 girls and 19 boys	9–14	MRI	Vertebral CSA	Girls and boys with spondylolysis have smaller vertebral CSA and greater LL.
	Controls: 36 girls and 50 boys			LL	Significant negative correlation between vertebral CSA and LL.
				Spondylolysis
Ponrartana et al., 2016 (274)	AIS: 35 girls and 11 boys	10–15	MRI	Vertebral CSA	Girls and boys with AIS have significantly smaller vertebral CSA and taller IVD.
	Controls: 35 girls and 11 boys		CT	IVD height	Girls have significantly taller IVD
				AIS
Wren et al., 2017 (276)	AIS: 25 girls	9–15	MRI	Vertebral CSA	Significant negative correlation between vertebral CSA and lateral thoracic vertebral wedging.
				Lateral thoracic wedging
	50 girls and 50 boys	9–15	MRI	Vertebral CSA	Significant negative correlation between vertebral CSA and anterior lumbar vertebral wedging.
				Sagittal lumbar wedging
Poorghasamians et al., 2017 (277)	27 girls	9–13	MRI	Changes in vertebral CSA and sagittal lumbar wedging	Progression of lumbar vertebral wedging is associated with lesser vertebral cross-sectional growth, whereas regression is the consequence of greater cross-sectional growth.

The height of the intervertebral disc (IVD), the compliance of its fibrous cartilage, and the dimensions of adjacent vertebrae are major determinants of spinal mobility (133, 275, 280). A greater range of motion occurs when the IVD is tall or the vertebral CSA is small. For comparable disc thickness and stiffness, the smaller female vertebral CSA results in greater flexion and extension and lateral flexion (133), a notion consistent with knowledge that the lumbar spine of young girls has greater flexibility than that of boys (281–283). Previous radiographic studies suggest that children with slender vertebrae have greater range of motion of the spine than those with larger vertebral bodies (284, 285). More recently, bending flexibility was found to be associated with smaller vertebral CSA and taller IVDs in healthy girls (275).

Spinal curvatures

LL is the anterior curvature of the lumbar spine, expressed in humans as a response to upright posture, which develops during childhood, increases throughout adolescence (286), and is more prominent in women than in men (133, 287–290). Support for the notion that smaller female vertebrae represent the human adaptation to bipedal obstetric load comes from a recent study showing that when compared with adolescent boys, adolescent girls had significantly smaller vertebral CSA and a greater degree of lumbar curvature (272). Notably, differences in LL between boys and girls were closely related to sex differences in vertebral CSA (272).

Spondylolysis is a stress fracture of the pars interarticularis of the vertebra, which occurs most commonly in the lower lumbar spine of adolescents with prominent LL (291–293). This condition affects ~6% of children and is thought to be the result of mechanical stresses from hyperextension of the spine (291, 294–297). Recent data show that in addition to a greater degree of LL, boys and girls with spondylolysis have significantly smaller vertebral CSA when compared with controls (Fig. 5), a difference that persisted even after the data were adjusted for height and weight (273). Whether the small vertebral body size found in patients with spondylolysis also translates to a small pars interarticularis and neural arch is unknown. Facet height, width, and orientation have all been proposed in the pathogenesis of spondylolysis (298, 299). Studies are needed to determine the degree to which exaggerated LL (mediated by small vertebral CSA) and small pars interarticularis predict spondylolysis.

Figure 5.

MRI images of the lumbar spine showing a greater LL angle (67° vs 17°) and smaller fourth lumbar vertebral CSA (7.2 cm² vs 9.1 cm²) in (a) a 12-year-old girl with spondylolysis and (b) a 12-year-old girl without spondylolysis.

Scoliosis affects ~7 million people in the United States, mostly adolescents, and 85% of cases are considered idiopathic (300–303). Beyond knowledge that female sex and a family history are major risk factors for its development (304), the specific genes and phenotypes that confer susceptibility or resistance to the deformity are unknown. Recent evidence suggests that girls with AIS have taller IVDs relative to vertebral width and greater spinal flexibility when compared with peers without spinal deformity (275). These findings are in agreement with data showing that girls who participate in sports, such as gymnastics and dancing, have both greater joint flexibility and a higher risk of developing AIS (305–310). Additional support for a role of vertebral morphology in the development of spinal deformity comes from two previous radiographic studies showing that girls with slender vertebrae have a greater tendency to develop progressive scoliosis than those with larger vertebral body widths (284, 285). Results from a preliminary retrospective study also show that MRI measures of lumbar vertebral CSA in children with AIS were significantly smaller and the IVDs taller than those obtained in children without spinal deformities, as measured with CT scans (274). Prospective studies using the same imaging modality in patients and controls are needed to validate these findings.

Multiple reports based on dual photon or X-ray absorptiometry showed lower spinal mass in children with AIS (311–317). Measurements of spinal bone mass at the time of diagnosis have even been suggested as a prognostic factor of curve progression (318). Because patients with AIS have substantially thicker IVDs (274), which have lower radiographic attenuation than bone, low spinal bone measures could, in part, reflect the confounding effect of a greater proportion of intervertebral fibrocartilage in the spine (235, 237, 238).

Vertebral wedging

Just as the term plasticity in engineering refers to the ability of solid material to undergo deformation in response to load, skeletal plasticity refers to the ability of growing bone to alter modeling as a consequence of mechanical stresses (319). Wedging of the vertebral body, a key structural characteristic of spinal curvatures (45, 276, 320, 321), is the result of a simultaneous increase in longitudinal growth on the convex side of a curve and inhibition on the concave side (322–324). In LL, greater axial loading in the posterior portions of the vertebral body leads to anterior-posterior wedging of the lumbar vertebrae (320). In scoliosis, compressive loading on the concave side of the curve inhibits longitudinal growth, and tensile loading on the convex side accelerates it, resulting in lateral wedging of the thoracic vertebrae (322–327). Previous studies have shown that the degrees of anterior wedging in LL and lateral wedging in scoliosis are both inversely correlated to vertebral CSA (276, 277).

Unlike the permanent deformations that adult vertebrae sustain under load, asymmetrical vertebral growth in children has the capacity to change shape in response to mechanical stresses (328, 329). In the immature skeleton, vertebral cross-sectional growth is an important determinant of the plasticity of the vertebral body; regression of vertebral wedging is associated with greater vertebral cross-sectional growth, whereas progression is the consequence of lesser cross-sectional growth (277). Makino et al. (327) recently showed the degree of wedging in thoracic vertebrae to diminish a year after posterior corrective surgery for AIS. A similar phenomenon occurs in pediatric patients with vertebral fractures secondary to leukemia or hypercortisolism (330–336). In contrast to the mature skeleton, these pathological fractures often regain normal dimensions after treatment (330, 337–339). Interestingly, older teenagers, like adults, are less likely to regain vertebral body height (339, 340), supporting the notion that vertebral body plasticity is a property of the immature axial skeleton.

Clinical Implications of the Female Vertebral CSA for the Aging Skeleton

Vertebral fragility fractures

The smaller female vertebra confers a biomechanical disadvantage that increases the stress within vertebrae for all physical activities and limits the loading capacity of the spine for life (134, 203, 214, 341). During compression along the longitudinal axis of a vertebral body, its loading capacity is directly proportional to its CSA and to the material, or compressive, strength of the bone (Fig. 6). Conversely, mechanical stresses are inversely proportional to CSA (stress = load/CSA); therefore, vertebrae with a smaller CSA have higher stresses than vertebrae with a larger CSA for a similar load. It has been estimated that sex differences in vertebral size result in 30% to 40% greater mechanical stress within vertebral bodies in young women than in men for equivalent loads (134). Not surprisingly, vertebral fractures, the most common clinical manifestation of osteoporosis, have a higher prevalence in women than men (342–345).

Figure 6.

The strength of the vertebral body is determined by vertebral CSA, bone density, microarchitecture, and bone tissue properties. Fractures (arrow) occur when the loads applied to a vertebral body exceed its strength.

Approximately 25% of all postmenopausal women in the United States are diagnosed with vertebral fractures, caused by the inability of the vertebral body to withstand the loads associated with normal daily activities (340–342, 346–348). However, most vertebral fractures are clinically silent (349, 350), highlighting the difficulties in assessing their true prevalence and justifying the varying results between studies of sex differences in vertebral fractures. In the Framingham cohort, a greater percentage of men had vertebral fractures than women in the younger age groups, but more women had vertebral fractures in the oldest age group, consistent with the notion that young men have more frequent exposure to high-load activities and injuries than young women (345). Available data indicate that men at all ages have greater vertebral strength, largely because of their larger CSA (214, 351).

Cancellous bone density alone is not a sufficient explanation for vertebral fractures. Many women with low bone density values, regardless of technique, do not experience fractures, and there is substantial overlap in bone density between women with and without radiographic evidence of vertebral compression (278, 352, 353). In an attempt to understand why some patients with low vertebral bone density do not have fractures, other properties of bone that contribute to its strength, such as the vertebral dimensions and material properties and microarchitecture of cancellous bone, have been considered in the pathogenesis of osteoporotic fractures (202–204, 278, 279, 352, 354–361) (Fig. 6).

Several clinical studies have also examined the role of vertebral cross-sectional dimensions as a possible determinant of vertebral fractures (Table 2) (201, 202, 214, 278, 279, 361). Overall, when compared with men and women without fragility fractures, patients with fractures had, on average, 7.7% smaller vertebral CSA (279). Notably, in a case-control study of matched pairs of older women with reduced bone density and spinal osteoporosis, whose main difference was the absence or presence of vertebral fractures, the CSAs of the unaffected vertebrae in the fracture group were ~8% smaller than in the nonfracture group (278).

Table 2.

Comparisons of Vertebral CSA, Width, and Volume in Subjects With and Without Vertebral Body Fracture

Study (Ref.)	n	Age (y)	Imaging Modality	Outcomes	Findings
Gilsanz et al., 1995 (278)	Vertebral fracture: 32 women	60–89	CT	Vertebral CSA	Women with vertebral fractures have significantly smaller vertebral CSA.
	Controls: 32 women			Vertebral fractures
Duan et al., 2001 (343)	Vertebral fracture: 76 (50 women)	18–92	DXA	Vertebral CSAa	Women and men with vertebral fractures have significantly smaller CSA.
	Controls: 1013 (686 women)			Vertebral fractures
Nielsen et al., 1993 (355)	Vertebral fracture: 62 women	66.8 ± 8.1b	DXA	Vertebral width	Women with vertebral fractures have smaller vertebral width, but the difference did not reach statistical significance.
	Controls: 415 women	53.0 ± 6.5		Vertebral fractures
Vega et al., 1998 (358)	Vertebral fracture: 30 men	47–83	DXA	Vertebral width	Men with vertebral fractures have significantly smaller vertebral width.
	Controls: 26 men			Vertebral fractures
Seeman et al., 2001 (204)	Vertebral fracture: 95 men	17–91	DXA	Vertebral width	Men with vertebral fractures have significantly smaller vertebral width.
	Controls: 395 men			Vertebral fractures
Ito et al., 1997 (211)	Vertebral fracture: 58 women	55–79	CT	Vertebral volumec	Women with vertebral fracture have smaller vertebral volume, but the difference did not reach statistical significance.
	Controls: 65 women			Vertebral fractures
Duan et al., 1999 (359)	Vertebral fracture: 163 women	18–87	DXA	Vertebral volumed	Women with vertebral fractures have significantly smaller vertebral volume.
	Controls: 479 women			Vertebral fractures

^aVertebral CSA calculated as π × width/2 × depth/2.

^bAge range not provided.

^cVertebral volume calculated as vertebral body area × mean vertebral height (vertebral CSA values were not provided).

^dVertebral volume calculated as (projected area)^3/2.

Vertebral fracture cascade

The main risk factor for sustaining a vertebral fracture is having suffered a previous one (362–367), a phenomenon often called the vertebral fracture cascade (364, 366). It has been estimated that the risk of another vertebral fracture is more than four times that of the first fracture and significantly greater than the recurrence rate associated with other fragility fractures (367). Why one vertebral fracture is an imminent risk for a subsequent fracture is unclear (366, 368). It should be noted that when compared with men, women are more likely to sustain first and subsequent fractures (367). Data also show that this elevated risk for vertebral fracture recurrence in women persists even after adjustment for age and BMD (362). Clearly, studies are needed to establish the potential role of vertebral CSA in the vertebral fracture cascade and whether patients who refracture are those with the smallest vertebral cross-sectional dimensions.

Conclusions and Future Directions

We now know that factors related to sex program fetal vertebral growth and give rise to the outmost phenotypic difference between the male and female skeletons. During evolution, the female axial skeleton probably reacted to the unique requirements of pregnancy by modeling the CSA of the vertebral body to facilitate spinal flexibility. Unfortunately, this programming not only renders the female spine more flexible but also lessens its strength. Even though this concept has only recently been formalized, a strong argument can be made that smaller vertebral CSA is a key risk factor for spinal deformities in adolescence and vertebral fractures in older adults.

Well-designed studies across multiple pediatric populations are needed to definitively establish the role of vertebral CSA as a risk for exaggerated lordosis, spondylolysis, and scoliosis. Studies are also needed to better examine vertebral cross-sectional growth as a determinant of vertebral plasticity in the immature spine and to investigate the potential for wedge deformities to reshape or resolve over time. A better understanding of the differences in vertebral growth and the reshaping potential of the growing vertebral body between sexes could help explain the 10 times greater prevalence of severe scoliosis in girls than boys. It could also aid in the design of preventive and corrective treatments for wedge deformities in children.

Additionally, sex discrepancies in vertebral CSA probably contribute to the higher incidence of fragility fractures in the axial skeleton of older women than men. Studies are needed to determine the strength of this phenotype in predicting vertebral fracture and to provide reference standards for vertebral CSA measures in adults of varying ages and body sizes; only then would we be in a position to establish a threshold value for vertebral fragility fractures. Future work should also compare the strength of vertebral size measures with bone density values in predicting fracture risk, specifically in women with a history of previous vertebral fracture because they are at substantial risk for additional fractures. This research could play a crucial role in the development of an integrated predictive index for fracture risk based on vertebral size and bone density, alongside sex, age, weight, and clinical history.

Although current data suggest that deficient vertebral cross-sectional growth in early life could be associated with an array of adverse spinal conditions, they do not preclude the possibility that vertebral growth can be optimized through mechanical or dietary interventions. Physical activity and mechanical loading in children are known to influence bone density, shape, and size, especially when performed frequently and regularly. Although much more work is needed, it is tempting to think that the clinical relevance of vertebral cross-sectional growth will soon be better delineated. Such knowledge will provide a more rational way to diagnose, prevent, and treat spinal pathologies in the young and in older adults.

Glossary

Abbreviations

AIS

adolescent idiopathic scoliosis

BAT

brown adipose tissue

BMD

bone mineral density

CSA

cross-sectional area

CT

computed tomography

DXA

dual-energy X-ray absorptiometry

FSH

follicle-stimulating hormone

GH

growth hormone

IGF

insulin-like growth factor

IVD

intervertebral disc

LL

lumbar lordosis

MRI

magnetic resonance imaging

PBM

peak bone mass

SNS

sympathetic nervous system

UCP1

uncoupling protein 1