The role of delayed bone age in the evaluation of stature and bone health in glucocorticoid treated patients with Duchenne muscular dystrophy

The physiological process of bone modelling and remodelling in childhood involves complex and sensitive endocrine and immunological signalling systems, which may be affected by factors such as physical conditioning, poor linear growth, obesity, pubertal delay, vitamin D deficiency, and alterations in calcium homeostasis [1, 2]. Low Bone Mineral Density (BMD) and an increased risk of fractures is a well-established consequence of many chronic diseases in childhood, including Duchenne muscular dystrophy (DMD) [3, 4].

Several studies have described a higher incidence of low-trauma long-bone (appendicular) fractures in steroid-naïve boys with DMD compared to the normal population [5, 6], and a significantly increased risk of appendicular and vertebral fractures in glucocorticoid (GC) treated DMD [7].

In untreated DMD, it is suggested that reduced bone strength is related to decreased muscle tension on bone and muscle disuse as well as nutritional factors, disturbances in the calcium homeostasis and increased activity of inflammatory cytokines [1, 8]. Histomorphometric analyses of bone biopsies in DMD showed clear differences in all compartments between patients with DMD compared to healthy age-matched reference data. Thus, the total length of the bone biopsy was reduced, as was the cortical shell (thickness). Moreover, the total trabecular bone volume per total volume was reduced, due to a reduction in trabecular thickness. These findings indicate significant reduction in bone formation and a relatively increased bone resorption (increased endosteal resorption) [9]. GC treatment exacerbates the situation by disrupting the osteoblast-osteoclast balance and coupling [10], compromising the biomechanical properties of bone [11], and inhibiting the absorption of calcium from the gastrointestinal tract and the reabsorption of calcium in the renal tubuli [10]. GC treatment is also associated with 25-hydroxy-vitamin D (25(OH)D) deficiency and by itself increased fracture risk [12]. Further, steroid induced obesity, poor linear growth [13], hypogonadism and pubertal delay [14] add to the complexity of bone health evaluation in GC treated DMD [15] .

Fractures in the long-bones are painful and can be detrimental to the ambulatory function of a patient with DMD, especially if they occur in the late-ambulatory phase [5]. Vertebral fractures may cause severe pain, but may also be asymptomatic and thereby underdiagnosed [16]. The long-term consequences of permanent deformity on pain and function following repeated childhood vertebral compression remain unstudied [4]. Patients with mild or even asymptomatic fractures may have a high risk of future vertebral fractures [16, 17]. Bone health monitoring, identification of at-risk patients and timely intervention is thus an important part of the management of DMD.

Dual-energy X-ray absorptiometry (DXA) is the standard measure of bone mass and density for clinical use in children [18, 19]. However, DXA has important limitations in the evaluation of the fracture risk in a growing child. Bone strength is also dependent on bone size, geometry, architecture, bone matrix quality, and anthropometric variables, which may be independent of BMD [18, 19]. Children can have microarchitectural changes and an increased fracture risk even with a normal BMD [4, 20].

In children with short stature or growth delay, BMD is systematically underestimated due to the two-dimensional presentation of bone area (g/cm2) in which bone thickness or depth is not factored into DXA estimates of BMD [21]. The International Society for Clinical Densitometry (ISCD) recommend that bone mineral content (BMC) and BMD in this situation should be adjusted using either bone mineral apparent density (BMAD) [22] or height-for-age Z-score (HAZ) [23] for the spine, and HAZ for total body less head measurements (TBLH) [19]. However, no prediction equation for bone mass based on height, weight and bone area has been published [20].

As a part of the Norwegian Duchenne muscular dystrophy cohort study [24], we have previously reported preliminary findings of considerably retarded bone maturation in a group of GC treated DMD patients [25]. We have also reported that the delay in skeletal maturation increases with age [25]. Bone age in adolescence is closely related to pubertal development and growth spurt, which are stunted in GC treated DMD [14, 26]. Therefore, our aim was to examine the extent of delayed bone age in boys with GC treated DMD, and to explore the implications of such delay for the evaluation of short stature and low BMD Z-scores. We hypothesized that a delay in bone age is closely related to the stunted growth and low BMD Z-scores in GC treated DMD. Furthermore, that taking bone age into account in the evaluation of BMD in GC treated boys with DMD is imperative, as the delay may influence evaluations of bone health in this patient group.

We examined the extent of delay in skeletal maturation and the relationship between bone age delay, growth and DXA measured BMD Z-scores in GC treated boys with DMD. Our results revealed a significant delay in skeletal maturation in this patient group. The delay increased with age and duration of GC treatment, and for some boys the delay was up to several years. Our study did not reveal any significant difference in bone age delay between groups treated with prednisolone and deflazacort. The bone age of the GC naïve participants was within normal variation. Interestingly, the bone age of previously GC treated participants was comparable to GC naïve participants, indicating that the delay in skeletal maturation may be reversible upon termination of GC treatment.

The use of the G&P atlas of normal skeletal maturation of the hand and wrist [39] has several limitations [42]. The reference population of the atlas were healthy, well-nourished, middle-class, Caucasian American children and adolescents, who underwent normal pubertal development and growth spurt in the 1930s. Recent studies suggest that skeletal maturation may be reached earlier and that maturation may differ considerably between individuals and ethnicities, and with nutrition, disease and medication [42‐45]. However, given its limitations, the G&P atlas still supplies valuable information about the skeletal maturation of individuals over time.

In contrast to our results, a previously published study found that bone age was compatible with chronological age in 33 GC treated boys with DMD [46]. The boys were 8.4–11.2 years old and had been treated with GCs for 32–66 months. Our participants were older and had been GC treated for a longer period, which possibly explains why bone age delay was not revealed in the previous study.

In our cohort the GC treated group was lower in height than the GC naïve group. A limitation of our study is the use of different measurement methods (standing height, lay flat or arm span), as well as possible measurement error due to lower limb contractures and lumbar lordosis. Familial short stature was not investigated, and historic growth data collected from medical records may have been inaccurate. However, both the low actual height and the increasing arrest of linear growth in GC treated DMD over time is in accordance with previous studies [13]. We found that much of the difference in height between the groups was eliminated when the height of the GC treated boys was corrected for bone age delay, and was no longer significant. Importantly, the number of boys with a height below the 5th percentile was reduced from 31 to 13 when age was corrected for skeletal delay.

In contrast to previous studies [13], the difference in weight and iso-BMI between groups was not significant. The lack of significant differences in weight in our study may however be due to insufficient power.

Previous studies have examined the effect of bone age delay on BMD Z-scores in other conditions, presenting results that correspond well with our findings [47, 48]. A study of paediatric brain tumour patients found a lack of correlation between chronological age and bone age BMD Z-scores in a subgroup of peripubertal males aged 11–15 years [47]. The authors suggested that even a small difference between bone age and chronological age greatly impacted the BMD Z-score due to the increased rate of bone mineralisation following the rapid expansion of bone volume related to normal pubertal growth spurt in boys. Thus, bone age adjustment should be added to BMD evaluations for children exhibiting abnormal growth velocities [47].

In a study of 151 healthy children and 61 children with metabolic and bone disorders, significant differences were found between BMD Z-scores based on chronological age and bone age respectively, with increased Z-scores after recalculation of BMD according to bone age in pubertal boys and children with pituitary deficiency [48]. It was suggested that considering a delayed bone age in the interpretation of BMD in the setting of pituitary hormone deficiency prevents an overestimation of skeletal deficits [19].

The clinical significance of our study with respect to the evaluation of fracture risk is yet unclear. Lateral spine radiographs of asymptomatic boys were not available, and the number of VFs may be considerably underestimated [49]. All five participants with radiologically confirmed VFs were treated with deflazacort, and the prevalence of LB fractures was higher among the deflazacort treated boys than the prednisolone treated group. However, the deflazacort treated boys were older than the prednisolone treated boys at the time of our study. Bone age adjusted BMD Z-scores at the time of VFs are not available, so that we do not have the data to explore the “true” risk of VFs in relation to corrected Z-scores. A previous study of otherwise healthy fracture-prone children found a bone-age adjusted BMD Z-score ≤ − 2.0 in only 8% of a cohort who had sustained fractures, although BMD Z-scores were significantly lower in fracture-prone children when compared to controls [50]. Interestingly, the same study found asymptomatic VFs in 15% of the fracture-prone children with a history of only appendicular fractures when screened by spinal radiography. Spinal radiographs of the controls were not available [50]. We argue that there is a need for larger, prospective studies in which bone age adjusted BMD Z-scores are reviewed in connection with lateral spine radiographs of boys with DMD on different GC regimens.

VFs can occur in children even with BMD Z-scores within normal range [4, 20, 51], and Z-scores may vary substantially depending on the normative database used [52]. The role of DXA in the evaluation of fracture risk has been altered accordingly in the 2018 DMD care considerations [51]. A low BMD Z-score is not an independent indication for bisphosphonate treatment, but rather serves as an adjuvant tool in the overall assessment of the bone health of the individual patient over time. Our results indicate that the significance of a fall in BMD Z-scores over time can only be evaluated in light of advancement or arrest in skeletal maturation over the same period of time. We suggest that if skeletal maturation is arrested, an apparent fall in BMD Z-scores may be erroneously weighted in the decision-making process regarding GC adjustments or indication for bone sparing treatment of the patient.

The results from our study correspond to a recent report of bone age delay among 12 GC treated DMD patients, as well as higher bone age adjusted BMD Z-scores for all patients [53]. Our results support the authors’ argument that the failure of DXA BMD Z-scores to predict fractures in GC treated DMD may be due to the lack of appropriate reference databases and failure to correct age-based BMD Z-scores for delay in skeletal maturation.

GC induced hypogonadotropic hypogonadism is a well-established consequence of long-term GC treatment [14]. We found delayed or absent signs of puberty among the adolescents in the GC treated group, while the GC naïve boys progressed through normal puberty. There was a relation between pubertal delay and low BMD Z-scores. As the rate of mineral accrual is linked more closely to skeletal maturation and pubertal development than to chronological age [18, 54], we suggest that testosterone replacement therapy should be considered in the presence of confirmed hypogonadism in steroid treated boys with DMD of pubertal age, not only in order to induce male secondary gender characteristics [35], but also to improve skeletal maturation and bone mineralisation as part of the pubertal development.

We found that the delay of skeletal maturation was more pronounced in the carpals than in the phalanges (Fig. 1) [25]. The rate of bone turnover in the trabecular compartment, such as the carpals and vertebrae is more rapid than that in the cortical one, and there is growing evidence that the two anatomical sites can react differently in disease states and as a result of specific treatments [9, 54, 55]. GCs have an affinity for trabecular bone [54], and failure of bone formation rather than increased bone resorption seems to be the main mechanism underlying glucocorticoid-associated bone loss [9, 10, 56]. A possible connection between marked delay of the maturation of the carpal bones and low vertebral BMD remains to be studied. If such a connection is found, the delayed carpal maturation might be used as an indication for vertebral fracture risk in GC treated DMD in the future.

In children with short stature, the two-dimensional nature of DXA will systematically underestimate bone density. Several mathematical methods of calculation have been demonstrated to improve the clinical accuracy of DXA in this situation, of which BMAD and/or HAZ are endorsed by the ISCD [19]. In a recent study, spine BMAD reference ranges are provided based on the Bone Mineral Density in Childhood Study (BMDCS) [57]. The study includes 2014 children aged 5–19 years at baseline. However, children of height, weight or BMI below the third percentile or above the 97th percentile, delayed or advanced pubertal development, use of medications known to influence bone metabolism or medical conditions that threatened normal bone accretion were excluded. The BMDCS cohort thus has important limitations in assessing children with an increased risk of fractures, such as GC treated boys with DMD [20, 58].

Further, we have demonstrated that the GC treated boys with DMD are not only short in stature, but suffer from significant delay of skeletal maturation. Calculation of BMAD or HAZ does not correct for this delay. Correcting BMD for bone age can be done directly and requires no assumptions about bone depth. We suggest that this method, which is relatively easy to perform, provides a useful impression of BMD based on the actual maturation of the bones.

Moreover, there is a need for further and larger studies of bone health and endocrine care in DMD, as well as for disease-specific reference databases for the evaluation of DXA results [19, 53, 59]. We suggest that such studies also include assessment of skeletal maturation.

Our study reveals significant delay in the skeletal maturation of boys with GC treated DMD. Our results indicate that considering skeletal delay in the evaluation of growth and BMD Z-scores is imperative in the evaluation of bone health and decision-making concerning short stature and fracture risk of GC treated boys with DMD.