On growth and scoliosis

The natural growth of the healthy human spine is abundantly described in literature [25‐27]. On average, it declines from more than 12 cm/year in the first year of life to about 2.3 cm/year from age 6 until puberty. Girls have a growth spurt at age 12–13 and boys at age 14–15, with stronger spinal growth in the second year of puberty. After that, there is a decline to zero at adulthood (data from Stücker [26]). The spine mainly grows in the vertebral bodies, lumbar more than thoracic more than cervical [27]. The cervical and thoracic intervertebral discs only grow slightly in the first four years, after which disc height stabilizes. Lumbar discs grow somewhat faster until about the age of 7 and also show a small growth spurt at the start of puberty [27]. With AIS, growth looks a bit different (Fig. 2). Yim et al. [6] studied growth of 611 Asian girls with AIS and 296 age-matched controls between the age of 12 and 16. The AIS girls were divided between moderate (Cobb angle 20°–40°) and severe scoliosis (more than 40°). In a cross-sectional analysis, the girls with severe scoliosis were significantly shorter than controls at the age of 12, but significantly taller at the age of 15 (Fig. 2). The girls with mild scoliosis were as tall as controls at the age of 12, but as tall as severe scoliotic girls at the age of 15. Consequently, the growth rate was much higher in girls with severe AIS (3.6 cm/year) than in girls with moderate AIS (2.3 cm/year) or healthy controls (1.6 cm/year) (Fig. 2). The authors also provide longitudinal data of girls developing AIS, which show the same trend [6]. The later onset of growth in girls with severe AIS may be related to a significantly later menarche (5.9 months compared to controls) that was also observed [6]. That late menarche correlates with AIS has also been reported by others [28‐30], but how this leads to a mechanical deformation remains as yet unexplained.

In recent years, multiple studies showed that intervertebral discs are higher in AIS patients than in healthy age-matched controls. Chen et al. [31] found 10–20% higher intervertebral discs at all thoracic levels of the spine, and Ponrartana et al. [11] described an increase in lumbar discs of some 20% in both boys and girls with AIS. Brink and colleagues also found that the difference in spinal growth between AIS patients and controls is mainly due to the intervertebral discs [12]. In principle, there are two mechanisms that cause increased intervertebral disc height: reduced spinal compression [32] and higher intradiscal pressure [33, 34]. These mechanisms seem to exclude each other, since disc height is reported to be inversely related to disc pressure [35], i.e.: reduced axial compression relaxes the disc, increases disc height and reduces intradiscal pressure [34]. Nevertheless, despite the reported increase in disc height [11, 12, 31], Meir et al. measured an intradiscal pressure of about 0.25 MPa in anaesthesized AIS patients in supine position [13, 36]. This is 70–100% more than in age-matched, not anaesthesized controls (0.12–0.15 MPa) [36]. Increased disc pressure in AIS patients may be related to the continued presence of notochordal cells in adult intervertebral discs [37], which have large vacuoles with a high concentration of hydrophilic solutes that elevate osmotic pressure [38, 39]. Also, notochordal cells induce and stimulate matrix synthesis in the intervertebral disc [40, 41], which may further increase static pressure and disc height.

Normally, notochordal cells vanish at the onset of puberty [40, 42] and this appears to be related to dynamic loading, presumably as a result of the activities of daily life. In an ex-vivo study using porcine intervertebral discs, Li et al. [43] showed that the number and vitality of notochordal cells decrease with higher amplitude and frequency of dynamic compression. Guehring and colleagues obtained similar results in an in vivo rabbit model [44]. Thus, that notochordal cells normally disappear at the onset of puberty may be due to the increased amplitude and frequency of dynamic loading during the activities of daily life, as well as higher muscular force and body mass. The loss of notochordal cells then results in a decrease of intradiscal pressure [45]. Reversely, the continued presence of notochordal cells in adult AIS patients suggests that the intradiscal pressure is high and that the dynamic loading of the spine was insufficient to eliminate the notochordal cells.

Not only the intervertebral discs, but also the vertebral bodies are higher in AIS patients than in healthy controls. Guo et al. [14] report consistently higher vertebral body height at thoracic levels, while Ponrartana et al. [11] found the same at lumbar levels. Following the Law of Hueter–Volkmann [17], these data suggest that the vertebrae are subjected to a reduced spinal compression, thereby allowing enhanced longitudinal growth. At the same time, the cross-sectional areas of the vertebrae appear to be smaller, resulting in more slender vertebrae in AIS patients [11]. A smaller cross-sectional area of the vertebrae is also indicative for a reduced spinal compression, because bones in general grow more stout under increased loading to maintain equal stress [46]. Thus, spines in AIS patients are more slender than in age-matched controls. This theoretically creates a higher risk of Euler buckling, but Euler buckling is not considered the mechanism underlying AIS, because the spine is not an elastic rod [47] but rather a chain of vertebrae hinged by flexible intervertebral discs [48, 49]. Also, the deformation is not instantaneous, but rather slow and related to the growing spine [50]. Yet, increased vertebral growth contributes to differential growth, which underlies the scoliotic deformation of the spine [18, 19, 50, 51].

Along with vertebral morphology, vertebral bone mineral density has been assessed in AIS patients. Bone mineral density is a strong indicator of spinal loading since it highly correlates with muscle mass (r = 0.92) in a population of children, women and men [52]. This implies that bone mineral density is a strong indicator of muscular activity. Cook et al. [53] already reported a significantly lower bone mineral density in AIS patients, 15% less than in healthy controls. Lee and colleagues also reported lower bone mineral density in AIS patients as compared to age-matched controls, and more so in patients with higher Cobb angles [54]. More recently, Ramos et al. [55] performed an MRI study in which they observed a vertebral bone quality (VBQ) score of 2.5 in AIS patients, compared to 2.1 in age-matched controls, also indicating poorer bone quality. These studies all support the notion that the spines of AIS patients are subjected to less dynamic compression than the spines of age-matched controls. This fits with the observations of higher growth rates and increased intervertebral disc height discussed above. Also, the lower bone mineral density in the thoracic spine [54] may be indicative for a lower spinal compression and higher risk of scoliotic deformations than the lumbar or cervical levels.

Faster growth, higher intervertebral discs, more slender vertebral bodies and low vertebral bone density are all indications that spinal compression is reduced in AIS patients. This compression mainly results from gravity and muscular activity [50]. Barrios and colleagues report that girls with AIS have a progressive decrease of BMI and more general a change in body composition (somatotype) [56]. In their large study on Asian girls, Yim and colleagues also reported body weight and body mass index (BMI) as a function of age [6]. Interestingly, girls that develop mild AIS weigh less at age 12 than healthy controls and more than girls with severe AIS. At age 15, however, the average weight of the three groups is more or less the same. Related to body height, however, there is a considerable difference between AIS patients and healthy controls: BMI is 17 or less in AIS patients at age 12 and more than 18.5 in healthy controls; through the years the difference remains more than one unit (Fig. 3). It is further interesting to note that the absolute BMI of the AIS patients is close to 17.5, the definition of anorexia nervosa in adults [57]. Patients with anorexia and other eating disorders are indeed at much greater risk to develop scoliosis than healthy controls [58].

Miyagi et al. [59] reported the lean mass index (LMI) of AIS patients and healthy controls, essentially the amount of muscle mass in relation to body height and as such a measure of muscularity. As expected, LMI is lower for AIS patients at all ages between 10 and 17. Further, LMI, BMI and bone mass all declined with increasing Cobb angle, while fat mass did not. This implies that loss of weight is mainly due to loss of muscle mass, which also results in lower bone mass. Starcevic–Klasan and colleagues reported that also muscular power is reduced in AIS patients: they matched AIS patients (n = 93) and healthy controls (n = 90) for age, weight, height and BMI, and found that iliopsoas strength was reduced by more than 10% [60]. Shahidi et al. [61] report that muscles on either side of the apex are atrophic compared to non-scoliotic individuals, and demonstrate levels of collagen similar to severe degenerative spinal pathologies. These observations all support the notion that muscular activity is reduced in AIS patients, not only by muscular mass but also by muscular strength. It leads to less spinal compression, lower bone mass, lower BMI, and accelerated growth of vertebrae and intervertebral discs [50].

While bone and cartilage grow by osmotic pressure of cells and matrix, tendons, ligaments, muscles, nerves and other tissues need external tension in order to grow. This way, the tensegrity and mechanical functionality of the musculoskeletal system remain intact during growth [50]. Ligaments, tendons and other fasciae essentially consist of parallel, discontinuous collagen fibrils that are interconnected by a non-collagenous matrix (Fig. 4a) [62]. This non-collagenous matrix consists of glycoproteins like proteoglycans, but also lubricin that allows sliding and elastin that provides elasticity (for a detailed description of tendon and ligament composition and architecture, see Thorpe et al. [63]). The majority of proteoglycans are small, leucine-rich proteoglycans (SLRPs), of which decorin is the most abundant one (80%) [62]. Decorins are known to inhibit the lateral fusion of collagen fibrils [64] by binding non-covalently to specific amino acids of the collagen fibrils. Glycosaminoglycans are able to connect to decorins on adjacent collagen fibrils to form an interfibrillar bridge, thereby securing mechanical integrity under tension (Fig. 4b) [65].

Mechanical stiffness and strength of tendons and ligaments are related to the cross-link density of the collagen fibrils involved: the more cross-links, the stiffer and stronger the ligament. In order to allow growth, tension must apply and the interfibrillar proteins must break, so that collagen fibrils can slide along each other longitudinally, and then reconnect in order to restore mechanical integrity (Fig. 4c) [66]. The collagen fibrils are then replenished by cell-created procollagens to fill the gaps and restore mechanical integrity (Fig. 4c). The breaking of interfibrillar bridges is essentially microdamage induced by dynamic mechanical loading [67], presumably by the activities of daily life. Thus, the growth of ligaments, tendons and other fasciae is a process of continuous remodelling under dynamic longitudinal tension. In adult tendons and ligaments, the cross-linking density is so high that fibrillar sliding is no longer possible [68]. In AIS, ligamentous growth appears to be inhibited and one reason may be the lack of dynamic loading that is required to break the cross-links and allow fibrillar sliding of the collagens and ligamental growth. Another reason may be that hormones affect the physiology and mechanical properties of ligaments; this is discussed in “A role for hormones” section.

Dynamic loading thus induces micro-damage to tendons and ligaments and thereby facilitate remodelling and growth. Static tension, on the other hand is reported to suppress the turnover and degradation of connective tissues [69, 70]. Saini et al. [71] review this phenomenon in detail and ascribe this effect to the non-conformity of enzymes to cleavage sites along the collagen fibrils, that is: matrix metalloproteinases and collagenases are no longer able to reach and digest the catalytic domains along the collagen fibrils. Also, stretched tendons and ligaments contract perpendicularly due to the mechanical Poisson effect and thereby decrease the permeability of the matrix for the interstitial transport of enzymes. Similarly, collagen degradation was reported to be inhibited by collagen cross-linking, as in matured organisms [68]. Thus, the condensation of the extracellular matrix decreases the availability and efficacy of enzymes and thereby reduce the growth potential of tendons and ligaments. On the other hand, dynamic loading enhances the transport of degrading enzymes through the matrix [72] and thereby support remodelling and growth.

The conceptual framework presented here provides a mechanistic, mechanobiological explanation for the differential growth of the spinal column and the ligaments attached to it, resulting in the characteristic scoliotic deformity of the spine. It appears that reduced muscularity is a major risk factor for the development of AIS, as it has an effect on the growth of both, the spine and the ligaments, but in opposite directions, resulting in differential growth. The concept does not tell, however, what underlies a reduced muscular strength in the first place. Indeed, there may be multiple causes for that, like a neuromuscular disease, a genetic mutation, a brain disorder (e.g. of the cerebellum or the pineal gland), an eating disorder, or simply disuse, all of which have been suggested as an underlying cause of scoliosis [1‐4]. In this context, one may wonder whether the classical differentiation between neuromuscular scoliosis (e.g. from Duchenne muscular distrophy) and adolescent idiopathic scoliosis is really that different, except that in the case of muscular dystrophy the cause is obvious and in AIS it is more subtle and appears later in life. Brink and colleagues already pointed this out in a direct anatomical comparison between patients with neuromuscular and adolescent scoliosis [12]. The conceptual framework presented in Fig. 5 suggests that differential growth underlies all types of juvenile and adolescent scoliosis.

As explained in the introduction, skeletal growth is a mechanical force, but it is also subjected to a large variety of hormones [80]. Some of these directly affect growth (like growth hormone produced by the pituary gland and IGF-1 produced by the liver), and it appears that e.g. integrated growth hormone levels are substantially and significantly higher in AIS patients under 12, but not different at higher age [80, 81]. Other hormones are related to brain function (growth hormone releasing hormone (GHRH) produced by the hypothalamus) or nutrition (insulin produced by the pancreas; hunger hormones ghrelin and leptin, produced by the stomach and fat cells, respectively) [82]. Nutrition presumably determines the growth and strengthening of muscles, as suggested by the higher incidence of scoliosis in patients with eating disorders [58]. Melatonin (produced by the pineal gland) is widely investigated in the context of scoliosis, because experimental rat studies show that a depletion is related to the development of scoliosis [78] and that the volume of the pineal gland is smaller in AIS patients than in healthy controls [83]. Nevertheless, its mechanism remains obscure [84‐86]. An interesting cue may be that a high concentration of melatonin receptors is found in the cerebellum [87], the small brain that governs motor control and balance [88] and thus may underlie muscular activity. Another interesting hormone is relaxin, a hormone produced by the prostate in men and in breasts and ovary in women [89]. Relaxin appears to interfere with connective tissues, presumably as a competitor of SLRPs like decorin that function as anchor points for cross-links between collagen fibrils [79]. Considering the late menarche in many girls developing AIS [6], it is also noticeable that relaxin production rises after ovulation and declines towards menstruation. A late first menstruation delays the production of relaxin and thereby may increase collagenous cross-linking in ligaments and reduce its growth potential. Estrogen, as a last example, increases collagen turnover and decreases the stiffness of tendons and ligaments by inhibiting lysyl oxidase and decreasing cross-linking [90, 91]. A lack of estrogen, as a result of late menarche, thus enhances cross-linking and reduces growth potential of tendons and ligaments. Interestingly, levels of oestradiol are indeed reported to be lower in girls with AIS than in age-matched controls [92, 93]. Overall, a large variety of hormones interacts with tissues, functions and each other in many different ways [94]. An exploration of these from the perspective of differential growth is of great interest and a topic for future research.

It has long been recognised that a scoliotic deformity of the spine is due to differential growth between the vertebrae and intervertebral discs on the one hand, and the interconnecting ligaments and fasciae on the other hand [18, 19, 24, 50]. If large enough, such growth discrepancy inevitably leads to twisting and bending of the spine [23, 73]. The question thus is, why ligamental growth does not keep up with spinal growth and what physical or biological cues affect this discrepancy. Such cues are essentially risk factors that may be identified and modulated to allow early diagnosis and prevention.

It appears that low muscular strength may be such a risk factor: it leads to enhanced spinal growth (Hueter–Volkmann Law), conservation of notochordal cells, increased disc pressure, and reduced bone mineral density. Also, less dynamic loading results in less microdamage in ligaments and fasciae and thereby inhibits remodelling and longitudinal growth of ligaments (Fig. 5). Girls are in general less muscular than boys, and may suffer more from severe AIS for that reason, but their susceptibility may also be related to the levels of estrogen and relaxin, which affect ligamental remodelling. There are many reasons why muscular strength may be limited: lack of exercise, reduced uptake of nutrients (eating disorder), disturbed hormone levels, disturbed motor control or proprioception, and more. While all these factors have been related to AIS, it is not the whole story: there are many girls and boys with reduced muscular strength who do not develop AIS, and there are AIS patients who seem to be physically fit. Indeed, a risk factor is not a cause or explanation, but a possible contribution to the disease. The underlying cause of AIS is multifactorial [1, 2] and the causes may differ between AIS patients. If a sufficient number of risk factors apply, the disease progresses. Differential growth provides a useful conceptual framework to think about the mechanical phenotypes of genetic, hormonal, neural, and physical cues.

Early indicators of AIS include low BMI, low muscular strength and late menarche. Of these, BMI appears to have some discriminatory power [6], and it is easy to assess by measuring length and weight. An interesting observation was made by Tomaschewsky [95], who identified an inhibited forward flexion of at least one spinal level in about 16% of 689 children 9–10 year old. This points at relatively short dorsal ligaments and could be interpreted as a first observable sign of hypo-normal flexibility and differential growth. Of the children with inhibited forward flexion, 27% developed scoliosis within one year after the assessment, where the usual incidence is about 2%. Inhibited forward flexion thus appears to be a strong early indicator of AIS that is easily assessed by a simple forward bending test. Muscular strength is relevant as argued above, but its assessment in the pre-pubertal phase has as yet not been reported. Children with low BMI, inhibited forward flexion and/or low muscular strength may be subjected to hormonal or neuromuscular screening to assess the deeper causes of AIS. Such screening should be performed well before the onset of the scoliotic deformation, because once scoliotic, it is extremely difficult to normalise the alignment of the spine.

Once an increased risk to develop AIS has been established, early treatment or even prevention should be conservative or minimally invasive. One target may be to train core stability and enhance the daily activities of the young children. Such dynamic loading will remodel the spinal ligaments, thereby avoiding locked growth. Also, it will compress the intervertebral discs, overload the notochordal cells and thereby avoid excessive osmotic pressure in the nucleus pulposus. Physical training is important for all children, but particularly important for children at risk to develop AIS. In patients with muscular dystrophy, like Duchene, physical training may be difficult or impossible. Instead, one may think of targeting the intervertebral discs, which should be considered the motor of differential growth and spinal deformity [50]. A controlled release of intradiscal pressure by degrading enzymes [96] may be considered as a minimal invasive alternative to surgical stabilization. Hormone therapies also may be considered to counteract differential growth. On the one hand, spinal growth may be modulated (e.g. by reducing growth hormone or strengthening of muscles), on the other hand ligamental remodelling may be enhanced by the suppletion of e.g. estrogen [90]. This field is largely unexplored and somewhat risky, since hormone expressions vary strongly in time and hormones tend to have multiple functions and mutual interactions.