Clinical Endocrinology

Skeletal morbidity in childhood acute lymphoblastic leukaemia


Dr J. H. Davies, Lecturer in Paediatric Endocrinology, Department of Child Health, Cardiff University, Heath Park, Cardiff CF14 4XN, UK. Tel: +44 2920 742274; Fax: +44 2920 745157; E-mail:


Acute lymphoblastic leukaemia (ALL) is the commonest childhood malignancy, with a typical age of presentation between 3 and 5 years. This equates to a 65 in 100 000 risk for a boy throughout childhood up to 14 years of age, and a 46 in 100 000 risk for a girl.1 With current treatment regimens, children with ALL can now expect a better than 80% chance of 5-year survival, with the majority now surviving into adulthood.2 Skeletal morbidity is increasingly being recognized in these individuals and may occur at diagnosis, during and following treatment. This is an important problem, which may result in fractures, pain, loss of mobility and deformity, with resultant adverse consequences on quality of life. The recognition of these complications is essential to enable the institution of rational strategies for monitoring and optimizing bone health.

This review will consider the many influences that adversely affect bone health in children treated for ALL and will focus on the commonest abnormality, osteopenia (Fig. 1). Osteopenia is characterized by a low bone mineral density (BMD). A low BMD during childhood may lead to a failure to achieve peak bone mass, which is thought to be a major determinant for the subsequent risk of osteoporotic fractures in later life.3 Recent clinical, animal and in vitro studies have elucidated some of the factors that lead to osteopenia in childhood ALL and these are discussed below.

Figure 1.

Influences on bone in children treated for childhood acute lymphoblastic leukaemia. +, positive influence; –, negative influence.

Skeletal morbidity at diagnosis of childhood ALL

The leukaemic process itself has an adverse effect on bone as at diagnosis up to 69% of children have a radiographic abnormality such as juxtametaphyseal lucent bands and lytic lesions.4 Up to 25% of children have radiographic evidence of osteopenia, and fractures are observed in 10%5,6 (Table 1). Histomorphometric evaluation of iliac crest biopsies at this time have confirmed that the bone is undermineralized.6 There are conflicting reports as to whether skeletal involvement at diagnosis has an influence on prognosis.7–9

Table 1.  Fractures in childhood acute lymphoblastic leukaemia
  • *

    Thirty-one per cent of subjects were still receiving maintenance therapy.

At diagnosis
Rogalsky et al.5107Fractures in 50% (n = 6) of those presenting with bone pain
Halton et al.6 40Fractures in 10%
During treatment
Rogalsky et al.5107Pathological fractures in 12%, nonpathological fractures in 13%
Atkinson et al.50 16Fractures in 19%
Halton et al.14 40Fractures in 39%
van der Sluis et al.51 61Fracture incidence ↑ × 6 compared to healthy controls
Following treatment
Nysom et al.58 95Fracture incidence at age 6–9 years, 33% (n = 6/18), at 15–18 years, 40% (n = 8/20), at > 29 years, 100% (n = 4/4)
Kadan-Lottick et al.64 75*Fractures in 28%, fracture incidence twice that in those receiving therapy or had discontinued for < 6 months
Strauss et al.281765-year cumulative fracture incidence of 28%

Bone metabolism at diagnosis

At diagnosis, serum markers of bone formation, procollagen type I C-terminal propeptide (PICP) and bone alkaline phosphatase (AP), are significantly lower than expected.10,11 Low concentrations of the bone collagen degradation peptide, carboxyterminal telopeptide of type I collagen (ICTP), have also been demonstrated.11 These findings indicate that the leukaemic disease process is associated with a low bone turnover state. Acquired GH insensitivity may also occur at diagnosis, as evidenced by a decrease in serum levels of IGF-1 and IGFBP-3 with increased urinary GH excretion, and reduced bone metabolism.12 Bone metabolism may be impaired by factors secreted by leukaemic cells, such as osteoblast-inhibiting factor and parathyroid hormone-related peptide.13 Direct infiltration of leukaemic cells into bone and expansion of the marrow spaces may result in destruction of the spongiosa.13 However, no correlation between leucocyte count at diagnosis and markers of bone formation has been demonstrated.11

Mineral homeostasis at diagnosis

In a series of 40 newly diagnosed children with ALL, 95% had low circulating 1,25-dihydroxyvitamin D3 concentrations and 64% had hypercalciuria, which together may negatively affect bone mineral accrual.6 Renal dysfunction could have accounted for these findings but was not demonstrated. Although receptor-mediated binding of 1,25-dihydroxyvitamin D3 to leukaemic lymphoblasts may have contributed to the depleted serum levels, this was not the case.6

Osteopenia during treatment of childhood ALL

Osteopenia is the commonest skeletal abnormality that occurs during treatment (Table 2). In one study, the incidence of radiographic osteopenia increased from 13% at diagnosis to 83% by 24 months of therapy, with 65% of patients demonstrating densitometric evidence of osteopenia at the lumbar spine.14 These latter observations were made in children administered a treatment regimen using cranial irradiation and high doses of both chemotherapy and glucocorticoids. More recently, during treatment of children with a contemporary UK treatment protocol which did not include administration of cranial irradiation, significant reductions in bone mineral content at the hip and lumbar spine were still demonstrated.15

Table 2.  Prospective studies of BMD during treatment for childhood acute lymphoblastic leukaemia
ReferenceBone densitometrynCRTControlsComments
  1. BMAD, bone mineral apparent density; % BMC, % of control bone mineral content; BMD vol, volumetric BMD; CRT, some of the cohort received cranial irradiation; n, sample size; N, no; SD, standard deviation; SXA, single X-ray absorptiometry; Y, yes; **other childhood malignancies included; ↑, increase; ↓, decrease; ↔, no change.

Atkinson et al.50SXA16NManufacturer radial BMD −2·3 SD (< –1 SD in 70%)
Halton et al.14DEXA40YLocal lumbar spine BMD ( from baseline in 47% of children), lumbar spine BMC ( from baseline in 64% of children)
Henderson et al.84DEXA37**YLocal↓ of ≥ 0·5 SD femoral BMD in 23% of patients, lumbar spine BMD in 27% of patients
Arikoski et al.35DEXA46**YLocal lumbar spine BMAD from baseline (−2·1%), femoral BMAD from baseline (−8·5%), femoral BMD from baseline (−9·9%), femoral BMD (−0·7 SD), femoral BMAD (−0·7 SD), BMD lumbar spine
Arikoski et al.36DEXA28**YLocal femoral BMAD from baseline (−10·1%), femoral BMD from baseline (−11·3%)
Boot et al.21DEXA32NLocal whole body BMD (−1·1 SD), lumbar spine BMD
van der Sluis et al.51DEXA61NLocal whole body BMD (−0·68 SD), lumbar spine BMD (∼−0·65 SD), lumbar spine BMAD
Davies et al.15DEXA14NLocal femoral neck % BMC (82%), femoral trochanter % BMC (72%), lumbar spine % BMC (89%), whole body % BMC

In one study, 39% of children sustained fractures on treatment, with impaction fractures at the distal tibia and distal femur occurring most frequently14 (Table 1), although vertebral compression fractures have also been reported.16 However, the clinical evaluation of fracture incidence during treatment is problematic as many fractures are subclinical and concomitant glucocorticoid administration may have ameliorated the symptoms. As children become osteopenic and fracture even when in remission, factors other than the leukaemic disease process may have adverse effects on bone metabolism. These factors are discussed below.

Bone metabolism during treatment

Serum markers of bone formation, bone AP and PICP, are suppressed immediately after the administration of combinations of chemotherapeutic agents and glucocorticoids.12 Recovery of these markers to near normal levels coincided with the administration of fewer agents within the same treatment regimen.17,18 Hyperleptinaemia also occurs during treatment15,19 and recent in vitro studies with primary human osteoblast-like (HOB) cells have shown that leptin has direct beneficial effects on bone formation.20 It has been speculated that resistance to leptin may contribute to a reduction in bone metabolism in these individuals.15 Unlike at diagnosis, GH resistance does not occur during treatment.12,17,21

Mineral homeostasis during treatment

Bone mineral accrual may be adversely affected by low serum 1,25-dihydroxyvitamin D3 concentrations and hypercalciuria which occur during treatment.14 Reduced 1,25-dihydroxyvitamin D3 concentrations may result from disordered synthesis of 1,25-dihydroxyvitamin D3 secondary to the glucocorticoid administration.22 Furthermore, an inappropriately low PTH response in those with mild hypocalcaemia contributes to the hypercalciuria.6 Hypomagnesaemia and hypermagnesuria also occur in these patients, and have been associated with the use of aminoglycosides and glucocorticoids.14,23 As magnesium is an important co-factor in hydroxylation reactions for the production of 1,25-dihydroxyvitamin D3,24 these latter biochemical abnormalities could further contribute to the low circulating 1,25-dihydroxyvitamin D3 concentrations.


Glucocorticoid administration is an important therapeutic intervention in childhood ALL as glucocorticoids induce leukaemic cell lysis.25 Treatment with glucocorticoids also predisposes to osteopenia and fractures.26 Dexamethasone is now the preferred glucocorticoid used for the treatment of childhood ALL because of its better CNS penetrance and antileukaemic effect. When given at the same anti-inflammatory potency, those treated with dexamethasone have a greater fracture-risk and incidence of osteonecrosis compared with those given prednisolone.27,28

The administration of pharmacological doses of glucocorticoids results in supraphysiological plasma levels of glucocorticoids and direct adverse effects on bone.29 The rate of bone loss is correlated with the glucocorticoid dose and there is greater loss of trabecular than cortical bone. Bone histomorphometric studies in adult humans have shown that exogenous glucocorticoids decrease bone formation by causing a marked reduction in the number of osteoid seams, a low mineral apposition rate and decreased mean wall thickness.30 In addition, the amount of bone replaced in each remodelling cycle was reduced by 30%. Similar studies have not been undertaken in children to determine glucocorticoid effects on bone modelling. In vitro, glucocorticoids inhibit osteoblast proliferation, type I collagen synthesis and local regulators of osteoblast function such as IGF-1 and IGFBP-3, effects which are more marked with dexamethasone than prednisolone.31,32

It is possible that in children, as has been shown in adults, glucocorticoids have indirect adverse effects on bone by causing reduced gonadal function, decreased intestinal calcium absorption, hypercalciuria, secondary hyperparathyroidism and steroid-induced myopathy.26,33 Moreover, glucocorticoid administration to children has been associated with vitamin D deficiency and low serum 1,25-dihydroxyvitamin D3 concentrations.34

Chemotherapeutic agents

During treatment of children with chemotherapy alone there is suppression of serum markers of bone formation coupled with a reduction of BMD.12,35,36 Those children administered high-dose combination chemotherapy are more likely to have osteopenia14 and have greater suppression of their serum markers of bone formation,12 when compared to those given fewer agents at a reduced dosage. These findings indicate that chemotherapeutic agents have direct adverse effects on bone. As chemotherapeutic agents are given in combination, it is impossible to determine from clinical studies the relative contribution of each agent to the resulting osteopenia. Thus animal and in vitro models have been employed to evaluate the direct effects of chemotherapeutic agents on bone.

Of the chemotherapeutic agents used to treat childhood ALL, methotrexate is known to predispose to bone pain, osteopenia and fractures even in low doses.37,38 In animal studies, methotrexate and doxorubicin inhibit new bone formation and impair fracture healing.39,40 Short-term administration of therapeutic doses of methotrexate or doxorubicin to rats caused a significant reduction in trabecular bone volume.41 Both drugs diminished the bone formation rate by 60%, and the cytotoxic effects on osteoblasts were reflected in the reduced volume and thickness of osteoid. Osteoclasts, however, were less affected, as the numbers of osteoclasts and the extent of their activity were no different from that observed in untreated rats.41,42In vivo data support this latter observation, as markers of bone resorption were suppressed to a much lesser extent than markers of bone formation during treatment of childhood ALL with chemotherapy.12,18

Age-related bone loss may be due to an inadequate supply of osteoblast precursor cells and a resultant decline in osteoblast numbers.43 It is possible that the mechanism of chemotherapy-induced osteopenia is similar. Recent evidence suggests that more chemotherapeutic agents than previously appreciated may have direct deleterious effects on bone in the clinical setting. In vitro, clinically relevant concentrations of all the chemotherapeutic agents used to treat childhood ALL caused a dose-dependent reduction of HOB cell numbers.44,45 Combinations of these agents diminished HOB cell numbers to a greater degree than when agents were given individually, and reductions in cell numbers preferentially occurred in less differentiated osteoblast phenotypes.45,46 In other studies, a reduction in osteoprogenitor cell numbers resulted in reduced skeletal mass.47 Thus during treatment with chemotherapy, although more differentiated osteoblast phenotypes may retain their ability to maintain a mineralized matrix, new bone formation may diminish from a depleted osteoprogenitor cell pool.

Some chemotherapeutic agents impair osteoblast function in vitro as evidenced by a reduction in type I collagen and alkaline phosphatase synthesis, a reduction in osteoblast responsiveness to 1,25-dihydroxyvitamin D3 and inhibition of mineralization.48 These latter observations, together with the reduction in osteoblast numbers, may account for the suppression of serum markers of bone formation in children receiving chemotherapy and contribute to the resultant osteopenia during and after treatment.

Chemotherapeutic agents also have indirect effects on bone metabolism as systemic factors that maintain bone quality, such as IGF-1 and IGF-BP3, are reduced during treatment for childhood ALL.12 Furthermore, vincristine-induced neuropathy can lead to dyspraxia which may increase the predisposition to a fall and fracture. Vincristine administration has also been associated with avascular necrosis of the femoral head in individuals treated for other malignancies.49

Dietary factors during treatment

During treatment, a sufficient dietary intake of energy, calcium, phosphorous, iron, vitamin C, folate and protein has been demonstrated.50 Calcium intake during therapy has not been shown to influence BMD.51 Although unproven, it is possible that intestinal mucositis from intensive chemotherapy may predispose to malabsorption of factors such as vitamin D and magnesium, and contribute to the abnormal mineral homeostasis.

Osteopenia following treatment of childhood ALL

Early studies of BMD in survivors of childhood ALL who have completed therapy need to be interpreted in the context of the change in treatment protocols over the last 15 years. In the UK prior to 1990, most children received cranial irradiation as CNS prophylaxis. More recently, children were randomized to receive either cranial irradiation or intrathecal methotrexate, whereas currently cranial irradiation is the first-line therapy for those with CNS disease at presentation. In addition, changes in chemotherapeutic regimens have also occurred which may in turn influence bone mass acquisition. Therefore, as treatment for childhood ALL changes, so outcome in terms of bone morbidity may change.

Another confounding factor is that initial studies using dual energy X-ray absorptiometry (DEXA) to evaluate BMD in survivors of ALL did not adjust for the influence of bone size on BMD. DEXA tends to underestimate BMD in small subjects and overestimate it in larger individuals. In this review therefore we have focused on studies in which DEXA measurements have taken into account the influence of body size (Table 3).52,53 By contrast, the use of quantitative computerized tomography (CT) has the advantage of measuring volumetric BMD and is less influenced by changes in bone size.

Table 3.  Cross-sectional studies of BMD after completion of treatment for childhood acute lymphoblastic leukaemia
ReferenceBone densitometrynCRTControlsMean/median age (years) at assessmentComments
  1. GHD, growth hormone deficiency; n, sample size; N, no; Y, yes; *31% subjects were still receiving maintenance therapy; **other childhood malignancies included; †qCT measurement; ↑, increase; ↓, decrease; ↔, no change.

Gilsanz et al.54qCT 42YLocal11·7 lumbar spine BMD vol (−0·93 SD), received CRT (n = 29); lumbar spine BMD vol, no CRT (n = 13)
Nussey et al.55DEXA 64**YManufacturer20·6GHD not treated (n = 14), femoral neck BMD (90% of control), lumbar spine BMD (88% of control); GHD treated with GH (n = 10), femoral neck BMD, lumbar spine BMD; no GHD, no GH treatment (n = 15), femoral neck BMD, lumbar spine BMD
Henderson et al.56DEXA 60**YLocal12·4 lumbar spine BMD, 8% had BMD ≤ −2 SD
Aisenberg et al.85DEXA 40**YLocal25·8 femoral neck BMD (−0·68 SD), distal radius BMD (−1·57 SD), whole body BMD (−0·33 SD), lumbar spine BMD
Arikoski et al.69DEXA 29YLocal17·1 femoral neck BMD (96% of control), lumbar spine BMD (96% of control)
Nysom et al.58DEXA 95YLocal16·2 lumbar spine BMD (−0·4 SD), whole body BMD (−0·4 SD)
Arikoski et al.86DEXA 48**YLocal 9·3 lumbar spine BMAD (−0·77 SD), femoral neck BMAD (−0·98 SD), femoral neck BMD (−1·02 SD), lumbar spine BMD
Brennan et al.60DEXA/SXA/qCT 31YLocal23·0 lumbar spine BMD vol (−1·25 SD), lumbar spine BMD (−0·74 SD), femoral neck BMD (−0·43 SD), forearm BMC (−1·3 SD)
H-Nijmann et al.65DEXA 24YManufacturer25·1 femoral neck BMD (−1·1 SD), lumbar spine BMD (−1·1 SD), radial BMD (−1·8 SD), femoral trochanter BMD
Warner et al.63DEXA 35YLocal12·0 hip % BMC (89%), lumbar spine % BMC (92%)
van der Sluis et al.61DEXA 23NLocal17·2 lumbar spine BMD, whole body BMD, lumbar spine BMAD
Kadan-Lottick et al.64DEXA 75*YManufacturer 9·3 whole body BMD
Kaste et al.57qCT141YManufacturer15·9 lumbar spine BMD vol (−0·78 SD)
Lequin et al.87DEXA/ultrasonometry 21NLocal16·5 lumbar spine BMD, tibial ultrasonometry
Tillmann et al.62DEXA 28NLocal10·7 lumbar spine BMAD (0·303 g/cm3vs. 0·323 g/cm3 control), whole body BMD, whole body % BMC, lumbar spine BMD
Mandel et al.72DEXA106YManufacturer15·9 lumbar spine BMD, femoral neck BMD
Brennan et al.73DEXA/qCT 53NLocal11·2 distal radial trabecular BMD vol, lumbar spine BMD, whole body qCT BMD, total radial BMD vol

Cranial irradiation

In historical cohorts of ALL survivors, severe abnormalities in BMD were seen exclusively in those treated previously with cranial irradiation.54–56 Furthermore, greater bone mineral decrements were found in those children treated with cranial irradiation using 2400 centigray (cGy) compared to those given 1800 cGy.57 Radiation-induced damage to the hypothalamo–pituitary axis may result in GH deficiency which is known to predispose to decreased bone mass,58 and GH deficiency was observed in up to 19% of ALL survivors at this time.59 Consistent with these data was the finding in ALL survivors that BMD was normal in both GH sufficient individuals and in those treated for their GH insufficiency.55

More recently, these latter findings have been challenged by a study by Brennan et al.,60 who demonstrated that following cranial irradiation there was no difference in the reduction of vertebral BMD between GH sufficient, insufficient and deficient adult survivors of childhood ALL, as measured by both quantitative CT and DEXA. This latter study was undertaken in survivors who had ceased therapy for a longer period than those studied by Nussey and colleagues,55 and the authors speculated that the reduction in BMD that they reported was the result of the direct effects of chemotherapeutic agents and glucocorticoids on bone. Moreover, reduced BMD has been demonstrated following cessation of treatment in ALL survivors given chemotherapy but without cranial irradiation administration.21,61,62 Therefore, apart from cranial irradiation, other factors appear to contribute to the adverse effects on BMD following treatment.

Chemotherapy and glucocorticoids

Following treatment of ALL, some have demonstrated that osteopenia was associated with the previous administration of glucocorticoids28 and certain chemotherapeutic agents, such as methotrexate and mercaptopurine,62,63 although others have found no such associations.58,64,65In vitro studies show there is poor recovery of osteoprogenitor cells following prior exposure to chemotherapy,46 and in rats the adverse histomorphometric effects of chemotherapy on bone remain months after cessation of treatment with no evidence of recovery.66

Physical inactivity following treatment

Following cessation of treatment, Warner et al.,63 demonstrated a significant positive correlation between a reduction in exercise capacity and physical fitness, measured by peak oxygen consumption, and the decrement in %BMC at the hip and spine. In the same subjects, a reduction in the level of habitual daily activity was also associated with osteopenia at these sites. In other studies, using a questionnaire and an accelerometer, the association between reduced physical activity and reduced BMD at the hip and spine has been confirmed,61,62 although no reduction in whole-body BMD has been demonstrated.64 The number of days admitted to hospital, as a proxy for immobilization, has not been shown to influence BMD outcome in survivors.35 Physical inactivity and reduction in exercise capacity may result from the side-effects of prior chemotherapeutic agent administration67 or be a consequence of the psychosocial sequelae of ALL.68

Dietary factors following treatment

There are conflicting data as to whether dietary calcium intake has beneficial effects on BMD in ALL survivors.56,61,62,64 Reduced serum 1,25-dihydroxyvitamin D3 concentrations have been found 6 months after completion of therapy in those with reduced lumbar BMD,50,69 whereas in long-term survivors calcium homeostasis appears normal.64

Other factors influencing BMD following treatment of childhood ALL

Male gender and treatment during adolescence are both independent risk factors that increase susceptibility to osteopenia and fractures in ALL survivors.28,57,62,69,70 It is unclear why this should occur although adult males may have increased susceptibility to glucocorticoid effects on bone when compared to females.71 Furthermore, as bone mineral accretion accelerates threefold during puberty, treatment during this time might be expected to have a greater adverse effect on BMD than if treatment was given at a younger age.

Catch-up of previously reduced BMD may occur with increasing length of time off therapy,63,64,72,73 and data from long-term ALL survivors suggest that bone metabolism is normal.60,64,65 Whether osteopenia in childhood translates into an increased fracture risk in late adulthood is uncertain, and further long-term prospective studies of volumetric BMD are required to address this important question. Many genetic polymorphisms are thought to influence BMD in healthy children,74 but these have not been shown to influence BMD in children treated for ALL.75 In children, changes of bone geometry also influence bone strength, even in the presence of normal BMD. A study of childhood ALL survivors treated without cranial irradiation showed that at the radius there was endosteal bone loss but also periosteal apposition at the same site.73 Such findings indicate that following an insult to bone formation from chemotherapy administration, there is a compensatory adaptive change to bone shape so that strength may be preserved.

Other contributing factors to the osteopenia may be over-treatment of irradiation-induced hypopituitarism with hydrocortisone or thyroxine,76 or the use of GnRH analogues for gonadotrophin-dependent precocious puberty.77

Reducing the skeletal morbidity associated with the treatment of childhood ALL

Given that physical activity is reduced in some ALL survivors, promotion of mobility and exercise during and after treatment could be beneficial to bone mass acquisition. A simple jumping programme, given for example during the prepubertal years, improves bone mass at the hip and lumbar spine.78 Nutritional supplementation during treatment may also be advantageous as a high intake of milk during childhood and adolescence is associated with increased bone mass at maturity,79 and calcium supplementation augments the bone response to physical activity.80 Vitamin D supplementation, even when given during infancy is associated with an increase in BMD at the femoral neck in prepubertal girls.81 By contrast, a reduction in BMD is associated with a high consumption of carbonated beverages during adolescence,82 an effect likely to be due to milk displacement from the diet. There is anecdotal evidence that the use of pamidronate may ameliorate the development of osteopenia during treatment of childhood ALL.83 However, there must be caution in the wider applicability of this study as the subject numbers were small, and more importantly, the effect of pamidronate is unknown in terms of its effect on the leukaemic disease process and its treatment.


Until long-term prospective studies show otherwise, individuals treated for childhood ALL represent an at risk group for the subsequent development of osteoporosis and fractures. Thus continued surveillance for skeletal morbidity into late adulthood is necessary. Attention to bone health should occur at each follow-up evaluation, including enquiry about musculoskeletal pain and also advice given regarding the optimization of nutrition, mobility and exercise. During childhood and adolescence, longitudinal evaluation of linear growth, including sitting height, is necessary, as poor height velocity may indicate the presence of vertebral compression fractures or GH deficiency. In those with symptoms of bone pain, radiography is indicated to establish the presence of a fracture and DEXA scanning to estimate BMD. Following cessation of ALL treatment, if there are recurrent fractures coupled with a low BMD, there should be consideration of bisphosphonate therapy. The impact on bone health in those treated for childhood ALL, of interventions such as exercise and nutritional supplementation, both during and following treatment, deserves further study.


The authors would like to thank LATCH Llandough Hospital aims to treat children with cancer with hope, Llandough Hospital, Penarth, Cardiff, UK and Novo Nordisk for financial support.