Impaired Development of Bone Mineral Density During Chemotherapy: A Prospective Analysis of 46 Children Newly Diagnosed with Cancer



Osteopenia and osteoporosis are becoming increasingly recognized in children with cancer, though reasons for these changes are poorly understood. The purpose of the present study was to evaluate longitudinal changes in bone mineral density (BMD) and bone turnover in newly diagnosed children with a malignancy. Lumbar spine (L2–L4) and femoral neck bone mineral density (BMDareal, g/cm2) was measured by dual-energy X-ray absorptiometry in 46 children (age 2.9–16.0, median 8.0 years; 15 leukemias, 12 lymphomas, 19 solid tumors) at diagnosis, and after 6 months from the baseline. The apparent volumetric bone mineral density (BMDvol) was calculated to minimize the effect of bone size on BMD. Serum levels of osteocalcin (OC), type I collagen carboxy-terminal propeptide (PICP), and type I collagen carboxy-terminal telopeptide (ICTP) were analyzed at diagnosis, and during a 6-month follow-up. A significant decrease in lumbar BMDvol (–2.1%, p < 0.05), and in femoral BMDareal (–9.9%, p = 0.0001) and BMDvol (–8.5%, p = 0.0001) was observed after 6 months when compared with baseline measurements. The markers of bone formation (PICP, OC) were significantly decreased, and the marker of bone resorption (ICTP) was significantly increased at diagnosis as compared with normal values. By the end the follow-up, the levels of PICP and OC were normalized, whereas the level of ICTP continued to increase indicating that there was a negative balance in bone turnover. A deficient accumulation of bone mass might predispose children with a malignancy to impaired development of peak bone mass. A controlled study determining the benefits of an early intervention on bone turnover should be considered in these patients.


Improvements in diagnostic and therapeutic procedures have led to increased survival rates in childhood malignancies. Approximately two-thirds of all patients are cured and reach adulthood. There is increasing concern over the quality of life in these patients. Skeletal problems are being ever more commonly recognized in children with a malignancy at diagnosis, during therapy, and after completion of treatment.(1–8) We have previously demonstrated reduced bone mineral density (BMD) in long-term survivors of childhood acute lymphoblastic leukemia (ALL), as well as in children with a malignancy at the cessation of their chemotherapy.(9,10) However, little is known about the longitudinal changes in bone metabolism and mineralization in children undergoing treatment for malignancies. Low bone turnover and diminished BMD have been reported in children during chemotherapy.(2,11–13)

The causes for these changes are most likely multifactorial. Some antineoplastic treatments, such as corticosteroids, methotrexate, and radiotherapy, have been described to be harmful to the development of bone mass and density.(1–10,14,15) Alterations in vitamin D metabolism, and treatment-induced hypogonadism and growth hormone (GH) deficiency are also factors that might influence BMD in these children.(2,7,12–13,16,17) Impaired accumulation of skeletal mass during childhood and adolescence may lead to low peak bone mass and predispose these children to suffer osteoporosis and fractures later in adulthood.

The purpose of the present study was to evaluate longitudinal changes in bone turnover and BMD in newly diagnosed children with a malignancy.


The study series comprised 46 (70%) of the 66 Caucasian patients (22 males, 24 females) who had been diagnosed with a childhood malignancy in the Kuopio University Hospital between January 1995 and December 1997 and treated with chemotherapy (Table 1). Patients were studied at diagnosis and during a 6-month follow-up from the diagnosis. None of the patients had a condition known or suspected to affect bone metabolism prior to diagnosis (a growth-affecting chronic disease, bone disease, systemic corticosteroid treatment during the preceeding 6 months, history of malignancy or radiotherapy, mental retardation, physical disability). No patient was diagnosed to have GH deficiency or was treated with GH. Twenty (30%) of the 66 patients were not included in the study: 17 (4 ALL, 13 solid tumors) due to inadequate study compliance to participate, 2 ALL patients due to Down's syndrome, and 1 patient with solid tumor due to a refusal. The study protocol was approved by the Research Ethics Committee of Kuopio University Hospital, and written informed consent was obtained from every parent and age-appropriate patient.

Table Table 1. Clinical Data of the Study Population
original image

Due to the heterogeneity of the diagnoses, patients were divided into three groups: 1) those with leukemias (ALL, n = 11, acute myeloblastic leukemia, n = 4), 2) those with lymphomas (n = 12), and 3) those with solid tumors (n = 19). The patients with leukemias were treated according to the protocols of the Nordic Society of Pediatric Hematology and Oncology (NOPHO).(18,19) The patients with lymphomas and solid tumors were treated according to the international cancer protocols consisting of various multiagent chemotherapy regimens.(20–26) Twelve patients (8 with lymphoma, 3 with acute myeloblastic leukemia, 1 with Burkitt leukemia) completed their chemotherapy at 5.5 (4.0–6.0) (median [range]) months from diagnosis, and had the last follow-up measurement performed at that stage. All the other patients continued their chemotherapy beyond the 6-month follow-up point. Table 2 presents data on the cumulative doses of those antineoplastic agents that were used in all three disease groups. The duration of hospitalization during the study period was determined as an estimate of immobilization (Table 2).

Table Table 2. Treatment Data (Median [Range]) of the Study Population During the Follow-up Period
original image

BMD and evaluation of bone age

Areal bone mineral density (BMDareal, g/cm2) of the lumbar spine (L2–L4) and left femoral neck were measured with dual-energy X-ray absorptiometry (DXA) (Lunar DPX; Lunar Radiation Corp., Madison, WI, U.S.A.) at diagnosis and at median 6 months from the baseline. The BMD measurements were performed between February 1995 and May 1998. Due to device-based soft tissue requirements, femoral BMD was measured only for children above 6 years of age (n = 28). Due to a measurement artifact, lumbar BMD values were not obtained for two patients at diagnosis and three patients at the follow-up, and femoral BMD for one patient at diagnosis and two patients at the follow-up. The short-term reproducibility expressed as the coefficient of variation (CV, %) for the spine is 0.8% and for the femoral neck 2.3%.(27) The long-term reproducibility (CV) of our DXA instrument based on repeated phantom measurements during the study was 0.5%. To minimize the effect of bone size on BMD values, bone apparent volumetric mineral density (BMDvol, g/cm3) was calculated from the areal BMD values: lumbar BMDvol = BMDareal (g/cm2) × [4/(π × width of measurement area in lumbar spine)]; femoral BMDvol = BMDareal (g/cm2) × (4/π) × (height of measurement area/measurement area of femoral neck).(28) The BMD results were compared with those of 121 healthy age- and gender-matched Finnish controls (60 males, 61 females, age 3.5–18.9 years).(27,28) Bone age at diagnosis was determined for 31 (67%) patients by one of the authors (J.K.) using the Tanner-Whitehouse (RUS) method.(29)

Bone biochemical and hormonal status

Bone biochemical and hormonal parameters were determined at diagnosis, and at medians of 2 weeks, 1 month, 3 months, and 6 months from the baseline. To evaluate bone formation, serum human intact osteocalcin (OC) and type I collagen carboxy-terminal propeptide (PICP) were measured by radioimmunoassays (Brahms Diagnostica, Berlin, Germany, and Orion Diagnostica, Espoo, Finland, respectively). The sensitivity of the intact OC assay was 0.5 μg/l; the intra- and interassay CVs were < 7.0% and < 11.6%, respectively. The sensitivity of the PICP assay was 0.2 μg/l; the intra- and interassay CVs were < 5.0% and < 7.6%, respectively. To evaluate bone resorption, serum type I collagen carboxy-terminal telopeptide (ICTP) was measured by radioimmunoassay (Orion Diagnostica). The sensitivity of the ICTP assay was 0.5 μg/l; the intra- and interassay CVs were < 7.0% and < 10.0%, respectively. A two-site immunoradiometric assay was used for the evaluation of intact parathyroid hormone (PTHint) (Nichols Institute Diagnostics, San Juan Capistrano, CA, U.S.A.). Serum levels of calcium, phosphate, magnesium, alkaline phosphatase, albumin, alanine aminotransferase (ALT), and creatinine were determined by standard methods, and laboratory-specific age- and gender-matched reference data were used (Kuopio University Hospital, Kuopio, Finland).


Statistical analyses were carried out with the SPSS for Windows (version 6.0.1; SPSS, Inc., Chicago, IL, U.S.A.) statistical program. To facilitate the comparison of data, Z scores were calculated for the BMD values and for the follow-up measurements of the bone biochemical markers from the mean and SD values of healthy Caucasian children of the same age.(27–28,30,31) A percentage change in the absolute BMD values was calculated between the follow-up and the baseline measurements {[(BMD follow-up – BMD diagnosis)/BMD diagnosis] × 100}. Kruskal–Wallis nonparametric test was used to compare the BMD values between the three disease groups (leukemias, lymphomas, solid tumors) with p values of < 0.05 were considered significant. Wilcoxon nonparametric test was used to compare the Z scores of BMD and bone biochemical markers against a constant for the controls. Similarly, the follow-up measurements of BMD and of the biochemical parameters were compared with the baseline measurements using the Wilcoxon test. A Bonferroni correction was performed for the multiple comparisons. Age, body mass index (BMI), relative height, duration of hospitalization, and the cumulative doses of antineoplastic agents between the disease groups were analyzed using Kruskal–Wallis test. Spearman correlation coefficients were calculated for the correlations of the cumulative doses of antineoplastic agents with BMD and bone biochemical markers. Stepwise regression analysis was used to study the effects of gender, disease group (leukemias, lymphomas, solid tumors; converted into two dummy variables), puberty, hospitalization days, and the changes in BMI and relative height values during the follow-up on the percentage changes in BMD.


Table 3 presents data on the age, bone age, BMI, and relative height at the time of diagnosis of the childhood malignancy. No significant difference was observed between the chronological age and bone age at diagnosis. BMI and relative height values were comparable with the age- and gender-matched healthy children.(27) After the follow-up, BMI was significantly increased (17.0 [12.6–27.5] kg/m2 vs. 16.2 [11.9–25.8] kg/m2, p < 0.01) (median [range]), and relative height significantly decreased (–0.3 [–2.5 to 2.0]) SDS vs. 0.0 [–2.3 to 2.4] SDS, p < 0.0001) as compared with baseline. No significant difference was observed in age, BMI, and relative height values between the patients with leukemias, lymphomas, and solid tumors at diagnosis or after the follow-up (Table 3). The data on the pubertal development were available for 45 (98%) patients of whom 30 (15 males, 15 females) were prepubertal and 15 (7 males, 8 females) pubertal at the end of the follow-up.

Table Table 3. The Clinical Characteristics (median [range]) of the Study Population at Diagnosis
original image

Figure 1 presents data on the individual absolute lumbar and femoral BMDvol values at diagnosis and after the 6-month follow-up. The mean (SD) absolute BMD values, the mean (SD) percentage changes in the absolute BMD values, and the BMD Z scores (mean [95% confidence intervals]) during the study are presented in Table 4.

Table Table 4. The Bone Mineral Density Data of the Study Population at Diagnosis and After the Follow-up, and the Percentage Change in the Absolute BMD Values During the Follow-up
original image
Figure FIG. 1.

(A and B) The individual lumbar (L2–L4) (A) and femoral neck (B) apparent volumetric BMD (g/cm3) values at diagnosis and at 6 months from diagnosis.

The mean lumbar and femoral BMDareal and BMDvol values presented as Z scores did not differ significantly from control values at diagnosis. After the follow-up, a significant decrease in the absolute lumbar BMDvol and femoral BMDareal and BMDvol values was observed as compared with the absolute BMD values at diagnosis (Table 4). In prepubertal patients, the loss in the absolute femoral BMDareal and BMDvol was significantly higher than in pubertal patients during the 6 months (–15.8% [7.6%] vs. –8.0% [6.0%], p < 0.01; –12.5 [9.5%] vs. –5.2 [5.5%], p < 0.05, respectively) (mean [SD]). No difference in the BMD Z scores was observed between the patients with leukemias, lymphomas, and solid tumors at diagnosis or at the end of the follow-up (data not shown).

The absolute values of biochemical markers of bone turnover are presented in Table 5. The markers of bone formation (PICP and OC) were significantly lower (–3.0 [–3.6 to –2.4]; –1.5 [–1.7 to –1.3], p < 0.0001, respectively) (mean Z scores [95% confidence intervals]) and the marker of bone resorption (ICTP) significantly higher (0.5 [0.05–1.0], p < 0.05) at diagnosis as compared with normal values (Fig. 2). The changes in bone markers over the follow-up are presented in Table 5 and Fig. 2. A significant increase in PICP and OC was observed at 3 and 6 months, and in ICTP at 6 months in comparison with baseline measurement (Table 5). At the end of the follow-up, the mean Z score of ICTP was significantly increased (1.5 [1.0–2.0], p < 0.0001) as compared with normal values, whereas the levels of PICP and OC were normal (–0.2 [–0.7 to 0.4]; 0.2 [–0.2–0.7], respectively; NS) (Fig. 2).

Table Table 5. Serum Levels (Mean [SD]) of the Markers of Bone Formation (PICP, OC) and Bone Resorption (ICTP) at Diagnosis and During the Follow-up
original image
Figure FIG. 2.

The levels of the serum markers of bone formation (PICP, OC) and bone resorption (ICTP) as Z scores (mean) at diagnosis and during the 6-month follow-up. *p < 0.05, **p < 0.0001 versus normal values (comparisons were performed for values at diagnosis, and for values at 6 months) PICP, type I collagen carboxy-terminal propeptide; OC, human intact osteocalcin; ICTP, type I collagen carboxy-terminal telopeptide; dg, diagnosis; wk, week; mo, month.

Table 6 presents data on the absolute values of serum calcium, phosphate, magnesium, PTHint, albumin, ALT, and creatinine. Serum levels of calcium (–1.5 [–1.9 to –1.0], p < 0.0001), phosphate (–1.4 [–1.8 to –0.9], p < 0.0001), magnesium (–1.3 [–1.7 to –0.9], p < 0.0001) and PTHint (–0.4 [–0.8–0.05], p < 0.05) (mean Z scores [95% confidence intervals]) were significantly decreased at diagnosis compared with normal values. Serum albumin at diagnosis was decreased in 18 (40%) patients. Serum creatinine was found to be elevated in eight (17%), and serum ALT in five (11%) patients at diagnosis. The changes in the biochemical parameters over the follow-up are presented in Table 6. A significant increase in ALT was observed at 2 weeks, and in phosphate and PTHint at 6 months in comparison to the baseline measurement. Serum calcium, magnesium, albumin, and creatinine levels did not change significantly over the follow-up time as compared with the measurements at baseline (Table 6).

Table Table 6. Serum Biochemical Parameters (Mean [SD]) at Diagnosis and During the Follow-up
original image

PICP, alkaline phosphatase, and OC at diagnosis or at the 6-month follow-up did not correlate with the percentage change in BMD. No correlations were observed between the reduction in BMD and the cumulative doses of the antineoplastic agents by the end of the follow-up period. The cumulative dose of methotrexate correlated positively with ICTP at 6 months (r = 0.36, p < 0.05). No correlations were observed between the other antineoplastic agents and any of the bone markers at the end of the follow-up.

In a stepwise regression analysis, an increase in relative height was independently associated with a decrease in lumbar BMDareal and BMDvol (r2 = 0.26, p = 0.02; r2 = 0.30, p = 0.01, respectively), and diagnosis of cancer before the onset of puberty was independently associated with a decrease in femoral BMDareal (r2 = 0.21, p = 0.03). The duration of hospitalization and the type of disease were not independently associated with BMD.


As far as we are aware, this is the first prospective study including different childhood malignancies where both lumbar and femoral neck BMD have been analyzed together with markers of bone metabolism, and where a correction for bone size was performed by calculating the BMDvol. Impaired development of BMDareal in children may be due to disturbed bone growth which would thus not represent a real impairment in bone density.(28) The correction for bone size is particularly important when evaluating longitudinal changes in childhood malignancies where growth failure during treatment commonly occurs.(16,17)

In the present study, no decrease in BMD was observed at the time of diagnosis in comparison with healthy age- and gender-matched controls. After 6 months from diagnosis, significant reductions in both lumbar volumetric and femoral areal and volumetric BMD were found as compared with the baseline measurement. The decline in BMDvol was evidence that this represented a real deficit in bone density in these patients. Also, the association between the increase in relative height and the decrease in lumbar volumetric BMD indicated a depressed increment of bone density in the growing spine during treatment for a malignancy.

The bone density values did not differ among the children with leukemias, lymphomas, and solid tumors, suggesting that it is the treatment components rather than the type of disease that are contributing factors to the decrease in BMD. The spine, which is predominantly trabecular bone, undergoes a more rapid bone turnover than the femoral neck, which contains more cortical bone. These two anatomical sites can thus respond in different ways to specific treatments.(32) In previous studies, the bone loss induced by corticosteroids has been shown to be most rapid in the lumbar spine.(32) Thus, the detected reduction in lumbar BMD could be partly explained by the higher susceptibility of trabecular bone to corticosteroid therapy which is commonly used in the treatment of childhood malignancies. Other anticancer agents, such as methotrexate, are also known to impair the development of bone mass and density.(14) However, we found no independent correlation between the reduction of BMD and any of the antineoplastic agents. The role of individual drugs in inducing changes in bone density in chronic diseases is difficult to assess clinically because other treatment components and the diseases themselves may also affect bone development.(1,33)

Prolonged periods of disease and treatment-related hospitalization are common in childhood malignancies. In the present study, the patients were hospitalized for considerable periods of time leading to decreased physical activity. The loss of bone during immobilization has been found to be higher in younger patients and to be greater in weight-bearing bones.(34) In cortical bone, immobilization-induced bone loss has been shown to be prominent after 3–6 months of inactivation.(34) Pathological fractures predominantly in the femur and osteoporosis following immobilization have been observed in children with cerebral palsy.(35) The bone loss caused by immobilization could thus partly explain our finding of a greater reduction in femoral than in lumbar BMD at 6 months from diagnosis, as well as the observation that the decrease in femoral BMD was higher in prepubertal than in the older, pubertal patients.

Serum OC has been shown to be a sensitive and specific marker of osteoblastic activity, reflecting the rate of bone mineralization.(36) Serum PICP is released into the circulation during type I collagen synthesis reflecting bone formation.(36) ICTP is a product of type I collagen degradation, which can be measured in serum and compared directly with the serum markers of bone formation measured from the same sample. A significant correlation has been shown between the serum levels of ICTP and the rate of bone resorption in histomorphometry and calcium kinetic studies.(37) In adults, serum ICTP has been found to be less sensitive than the urinary markers in monitoring the effects of antiresorptive agents,(36) but in children, the sensitivity of the response of ICTP to therapeutic interventions has been equal to that of the urinary pyridinium cross-links.(38)

We found that the markers of bone formation were significantly decreased, and the marker of bone resorption significantly increased already at diagnosis, reflecting a negative balance in bone turnover even though no reduction in BMD was observed at that stage. The markers of bone metabolism are known to respond relatively rapidly as compared with changes in BMD. Secretion of PTH-related peptide, tumor-related cytokines, disturbances in vitamin D metabolism, and GH secretion, leukemic infiltration of the bone marrow, and disease-related poor health and physical inactivity are factors that can impair bone metabolism in children with a malignancy already at an early stage of the disease.(2,12–13,39) In our study, the level of ICTP was increased, whereas the levels of OC and PICP were normalized, indicating increased bone resorption during the follow-up which is in accordance with the reduction observed in BMD at 6 months after diagnosis. The positive correlation between the cumulative dose of methotrexate and ICTP at the end of the follow-up pointed to methotrexate as being a factor contributing to the increased bone resorption.

OC and ICTP are largely cleared by the kidneys.(36) The elevated serum ICTP and OC levels could thus be partly due to impaired renal function. However, we observed no increase in the level of serum creatinine during the study period. Serum PICP is largely cleared via the hepatic endothelial cells, and impaired hepatic function would lead to increased serum PICP levels.(36) We found a significant increase in serum ALT at 2 weeks from diagnosis during the intensive induction stage of chemotherapy, with the level returning to normal in the follow-up measurements. Thus, the increased levels of PICP by 6 months from diagnosis in our study do not seem to be due to impaired liver function.

Increased urinary excretion has been reported as a cause for low serum magnesium in children with ALL.(40) Magnesium depletion may lead to reduced PTH secretion through inefficient stimulation of adenyl cyclase.(41) This could partly explain the low levels of serum calcium observed in our patients since PTH participates in calcium mobilization from bone and in reabsorption of ionic calcium from the renal tubules. However, the increase in the level of PTH while the serum magnesium remained low was evidence that the secretion of PTH in childhood malignancies is influenced also by other mechanisms which need to be further evaluated. The low levels of serum total calcium during the follow-up could partly be attributed to the observed hypoalbuminemia which has been considered to be a metabolic response to fever and infection in children with a malignancy.(42) The concentrations of serum ionic and urinary calcium and the intake of calcium were not analyzed in our study.

In conclusion, our data revealed increased bone resorption and decreased bone formation already at diagnosis of a childhood malignancy. The negative balance in bone turnover persisted throughout the follow-up, and a reduction in BMD was observed at 6 months from diagnosis. In addition to the optimal treatment of the possible late endocrinopathies, such as growth hormone deficiency and hypogonadism which might disturb bone metabolism in survivors of a childhood malignancy, a controlled study to assess the benefits of an early intervention program on bone turnover in newly diagnosed children with cancer should be undertaken. Ensuring adequate calcium intake, physical therapy, GH therapy for children with GH deficiency, and vitamin D and calcitonin treatments have been described as potential methods for improving BMD in children.(5,7,43–45)


The authors wish to thank Pirjo Halonen, M.Sc., for her statistical assistance. P.A. is grateful to Finnish Pediatric Research Foundation for financial assistance.