Survivors of acute lymphoblastic leukemia (ALL) are at risk of osteoporosis and obesity. We studied bone mineral density (BMD), percent of fat mass (%FM), and activity levels in survivors of ALL treated without radiotherapy. Lumbar and total areal BMD (g/cm2) and %FM were measured in 28 survivors (aged 5.7-14.7 years) of childhood ALL by dual-energy X-ray absorptiometry (DXA) scan (GE Lunar, Prodigy) an average of 5 years after completion of chemotherapy (UK Medical Research Council randomized trial protocol XI [UKALL XI]). One boy fractured his arm during treatment. Apparent volumetric lumbar BMD (BMD vol; g/cm3) was calculated and %FM was adjusted for sex and age (%FM adj). Physical activity was measured by accelerometer and questionnaire. The results were compared with 28 sex- and age-matched healthy controls. Total body and lumbar areal BMD (g/cm2) were not different between the ALL group and the control group. However, mean lumbar BMD vol in survivors of ALL was significantly lower than in controls (0.303 ± 0.036 g/cm3 vs. 0.323 ± 0.03 g/cm3; p < 0.01), which mostly was caused by the difference in boys (0.287 ± 0.032 g/cm3 vs. 0.312 ± 0.027 g/cm3; p < 0.05). Weekly activity score by questionnaire was significantly lower in the ALL group than in the control group (geometric mean 50 vs. geometric mean 74; p < 0.05). Male gender, low activity levels and an intravenous (iv) high dose of methotrexate were associated with low lumbar BMD vol. Patients who received an iv high dose of methotrexate (n = 18) had significantly higher %FM adj than those with intrathecal methotrexate only (n = 10; 141 ± 70% vs. 98 ± 37%; p < 0.05). In conclusion, male survivors of childhood ALL have reduced lumbar BMD vol, whereas no such difference was seen in girls. Overall, survivors of ALL were physically less active than their healthy controls and lower activity correlated with lower lumbar BMD vol and higher %FM adj.
ACUTE LYMPHOBLASTIC leukemia (ALL) is the commonest childhood malignancy, accounting for 26% of all childhood cancers. Dramatic improvements in treatment mean that ∼70% of children with standard risk disease can now expect to be cured.(1,2) However, as survival rates have improved, there has been an increasing recognition of adverse long-term effects of treatment such as effects on neuropsychological functioning and cardiac, growth, and endocrine function including osteoporosis and obesity.(3,4)
Possible etiological factors for osteoporosis include the leukemia per se, chemotherapy, particularly steroids and methotrexate and cranial irradiation. In addition, poor nutritional intake, reduced vitamin D levels and low levels of physical activity may have an impact on bone development in these children.
Most previous studies have looked at bone health in ALL survivors who received cranial radiation as prophylaxis against central nervous system disease.(4–11) Children in the United Kingdom treated according to the Medical Research Council protocols (UKALL), which included low-dose cranial irradiation (18-24 Gy), had significantly reduced bone mineral density (BMD) in the spine, less so the in femoral neck.(12) The effect of chemotherapy alone on bone health is much less clear. To date, the studies (outside the United Kingdom) that have looked at children who have been treated with chemotherapy alone have shown normal (spine)(13,14) or reduced BMD (femur and spine).(15,16) There have as yet been no studies examining BMD in survivors of ALL treated according to the UKALL randomized trial protocol XI (UKALL XI), which did not include cranial radiation for standard risk patients.
In addition to the effects seen on BMD, survivors of ALL have an increased incidence of obesity when defined by measuring body mass index (BMI).(17–19) Studies of body composition have confirmed an increase in body fat in survivors of ALL when compared with the normal population and survivors of other malignancies.(4,20) Cranial irradiation,(4) chemotherapy, particularly dexamethasone,(21) and reduced energy expenditure secondary to reduced habitual physical activity(22) have been proposed as contributors to the development of obesity. Decreased physical activity is a known risk factor for obesity(23) as well as for reduced BMD.(24,25)
We aimed to study BMD and body composition in survivors of childhood leukemia treated with chemotherapy alone and to assess the effect of physical activity on these changes.
MATERIALS AND METHODS
All 31 children with ALL treated on the UKALL XI protocol who were alive and still followed up at the Sheffield Children's Hospital were asked to participate in the study. The study was approved by the local ethics committee, and informed consent was obtained from parents and children of an appropriate age. They were all in their first clinical remission and none had received cranial radiation. Twenty-eight (17 boys; 11 girls) of them participated in the study (1 child refused and 2 children did not attend for the study appointment). The mean age of the patients was 10.7 years (range, 5.7-14.7 years) and they were studied an average of 4.5 years (range, 1.5-7.1 years) after discontinuing their chemotherapy. Eighteen children were prepubertal, 6 children in pubertal stages 2 or 3 and 4 children in stages 4 and 5 using Tanner standards. Auxological characteristics are given in Table 1.
Table Table 1.. Clinical Characteristics of Subjects
All patients were treated according to the previously described UKALL XI.(26) Briefly, all children had received 4 weeks of induction chemotherapy, which was followed by two 5-day intensification blocks at weeks 5 and 20. Eighteen patients received intrathecal plus intravenous (iv) high-dose methotrexate (group A), and 10 patients received intrathecal methotrexate only (group B) as central nervous system (CNS)-directed treatment. Thirteen children also received an 8-week third intensification block at week 35. Cumulative doses of drugs were 5400-10800 IU/m2 of asparaginase; 43.5-46.5 mg/m2 of vincristine; 6120-6520 mg/m2 of prednisolone; 180 mg/m2 of daunorubicin; 1000 mg/m2 of etoposide; 2000-3200 mg/m2 of cytarabine; 800-2480 mg/m2 of thioguanine; and methotrexate, 1520-1740 mg/m2 orally, 18-24 g/m2 intravenously, and 120-212.5 mg/m2 intrathecally. One boy with ALL presented with severe back pain but no signs of fracture on X-ray and 1 other boy fractured his radius and ulna during the treatment.
Twenty-eight (17 boys; 11 girls) sex and age matched healthy children of hospital staff were recruited as controls. Their mean age was 10.4 years (range, 5.2-15.1 years). Fifteen of the children were prepubertal, 8 children were in pubertal stages 2 or 3 and 5 children were in stages 4 and 5. Auxological characteristics are given in Table 1. Physical activity in controls was measured only by questionnaire.
Children over 12 years of age and parents of those under 12 years were sent a validated modified questionnaire(27) assessing exercise levels ∼1 week before their clinic appointment. Patients were asked to indicate how many times they participate in different forms of exercise for more than 15 minutes during their free time in an average week. Exercise was classified into three categories, strenuous exercise (heart beats rapidly), moderate exercise (not exhausting), and mild exercise (minimal effort), and examples were provided to further clarify the level of exercise. A total score was generated (exercise in an average week) by summing the reported frequencies for each of the three categories (strenuous, moderate, and mild) after multiplying each category with the corresponding metabolic equivalents (MET) score (9, 5, or 3, respectively). MET is a multiple of resting metabolic rate, which represents a value of energy costs of activities.(27)
A questionnaire was used to assess daily milk (ml) and weekly dairy product (milk, yogurt, cheese, and ice-cream) consumption. Patients were divided subsequently into three groups: low, none or only one dairy product once (or less) per week; moderate, at least two dairy products two to three times a weeks or daily milk consumption ⅓-½ pints; and high, two dairy products six to seven times a week or four to five times a week and daily milk consumption >½ pint, or daily milk consumption >1 pint.
Physical activity was measured also by an uniaxial accelerometer (CSA-7164; Computer Science and Applications Inc., Shalimar, FL, USA) in patients with ALL. Subjects were asked to wear the accelerometer on their right hip through the day (usually 10-12 h, not during the night) on 3 consecutive days after the day (always Wednesday) they visited the hospital for the dual-energy X-ray absorptiometry (DXA) scan. Thus, each 3-day period included 2 days of normal school and 1 weekend day. The actual time the accelerometer had been on the hip was reported also. Data were analyzed by a specific software program provided by the manufacturer and a mean activity count per hour generated over the period the accelerometer was worn.
Total body and lumbar (L2-L4) bone mineral content (BMC) and bone area (BA) were measured by DXA (GE Lunar Prodigy 10326, software version 2.26; GE Lunar, Madison, WI, USA). BMD was calculated by BMC/BA. The BMD data obtained from DXA are expressed as grams per centimeter squared, that is, not true volumetric density and, therefore, are influenced greatly by the bone size. To minimize the effect of bone size on BMD values, two methods were used. First, a predicted total body BMC was calculated from the mathematical model by Warner et al.(28) based on a cohort of children that takes into account the influence of age and height. The results were expressed as a percentage:
Second, apparent volumetric BMD (BMD vol) of the lumbar spine was calculated according to the method of Kröger et al.,(29) which is based on DXA measurements of 84 children. This model assumes the lumbar vertebra to be a cylinder and corrects its areal BMD value with the width (depth) of the vertebra. Width of vertebrae L2-L4 were obtained from the scan analysis reports. The BMD vol in the spine (L2-L4) was calculated as
In addition, total body and lumbar BA for height was calculated by dividing total or lumbar BA by height.
Fat mass (FM) was measured also by DXA at the same time as BMD measurements. Percentage of FM (%FM) was calculated: %FM = FM/tissue mass × 100. Because %FM is highly sex and age dependent, percentage of FM was adjusted for age- and sex- specific normal values: %FM adj = %FM/(predicted %FM for that height, weight, age, and sex) × 100. Predicted %FM was calculated from the published normal data.(30)
Statistical analysis was done using SPSS, Inc. (version 10.0; SPSS, Inc., Chicago, IL, USA) and Minitab (version 10.1; Minitab, Inc., Coventry, UK) software packages. Differences within the ALL group were assessed by Student's t-test and differences between the survivors of ALL and their sex- and age-matched controls were measured by paired t-test. Before the analysis, activity data were log-transformed in both groups to achieve normal distribution of data. Results of these two parameters are shown as geometric mean, +1 tolerance factor (TF) [+ 1 TF = antilog(mean + 1 SD)] and −1 TF [−1 TF = antilog(mean − 1 SD)]. Relationships between variables were evaluated by Pearson correlation and linear regression analysis. Partial correlation controlling for age and sex was used to assess the relationship between physical activity and BMD. Multiple regression analysis was used for prediction models. Those variables not reaching significance at the 5% level were discarded in a backward stepwise manner.
Mean height, weight, and BMI in children with ALL was normal with no difference between genders (Table 1). Only 1 child had BMI SD Score (SDS) > +2.0. Patients in group I (with third intensification) tended to be heavier than those in group II (without third intensification; +0.63 ± 1.2 SDS vs. −0.15 ± 0.72 SDS; p = 0.06), but the difference did not reach the 5% significance level.
Milk and dairy products consumption
One child with ALL and 1 control child had low milk consumption, whereas there were 11 patients with ALL and 7 control subjects with moderate milk consumption.
BMC, BA, BA for height, and areal BMD (g/cm2) of total body and lumbar region were not different between the ALL group and the control group (Table 2). However, mean lumbar BMD vol in survivors of ALL was significantly lower than in their sex- and age-matched controls (0.303 ± 0.036 g/cm3 vs. 0.323 ± 0.003 g/cm3; p < 0.01; Fig. 1A). The difference in lumbar BMD vol mostly was caused by the difference in boys (n = 17; 0.287 ± 0.032 g/cm3 vs. 0.312 ± 0.027 g/cm3; p < 0.05; Fig. 1B), which was seen across all ages and pubertal stages. In girls, the lumbar BMD vol was more similar between the ALL and control groups (0.326 ± 0.029 g/cm3 and 0.339 ± 0.027 g/cm3; p = 0.28). Patients in group A (intrathecal and iv methotrexate; n = 18) tended to have lower lumbar BMD vol than those in group B (only intrathecal methotrexate; n = 10), but the difference was not statistically significant (0.294 ± 0.038 vs. 0.317 ± 0.027; p = 0.07). There was no significant difference in BMDs between the dietary groups.
Table Table 2.. The Results of BMC, BA, BA for Height and BMD in Both Groups
The mean %FM in survivors of ALL was not significantly different from controls in either sex (18.8 ± 10.2% vs. 16.9 ± 9.1% in males and 26.6 ± 9.9% vs. 23.5 ± 7.7% in girls, respectively). However, when %FM was adjusted for age and sex (%FM adj), patients in group A (intrathecal and iv methotrexate) had significantly higher %FM adj than those in group B (only intrathecal methotrexate; 141 ± 70% vs. 98 ± 37%; p < 0.05; Fig. 2), whereas the difference in the BMI SDS was not significant (0.37 ± 1.1 SDS vs. −0.1 ± 1.1 SDS, respectively). Survivors of ALL who had %FM adj > 120% (n = 13; mean age, 10.8 years) had significantly lower total BMD than those with %FM adj < 120% (n = 15; mean age, 10.6 years; 0.913 ± 0.058 vs. 0.966 ± 0.062; p < 0.05). There was no such difference in the control group.
Weekly activity score by questionnaire was significantly lower in the ALL than in the control group (mean 50 ± 1 TF [26; 96] versus mean 74 ± 1 TF [40; 138]; p < 0.05) with no sex difference. Only 2 children with ALL had an activity score above 100 points, whereas in the control group there were 8 such subjects. Activity score measured either by questionnaire or accelerometer was not different between the treatment groups and did not correlate with time from end of treatment.
In the ALL group, physical activity measured by accelerometer was significantly correlated to activity levels measured by questionnaire (r = 0.57; p < 0.005). Activity score by accelerometer tended to be higher in boys with ALL compared with that in girls but did not reach significance (4.4 × 104 ± 1 TF [3.0 × 104; 6.5 × 104] vs. 3.6 × 104 ± 1 TF [2.7 × 104; 4.7 × 104]; p = 0.08). Using partial correlation corrected by age and sex, total BMD and lumbar BMD vol were both positively correlated to activity levels measured either by accelerometer (r = 0.56, p < 0.01 and r = 0.42, p = 0.05, respectively) or by questionnaire (r = 0.39, p = 0.07 and r = 0.58, p < 0.005, respectively).
In the ALL group, physical activity by accelerometer was negatively correlated to %FM adj (r = −0.39; p < 0.05; Fig. 3). Patients with %FM adj > 120% were less active (p < 0.05) measured either by accelerometer (mean 3.5 × 104 ± 1 TF [2.6 × 104; 4.7 × 104] counts/h) or by questionnaire (mean 36 ± 1 TF [36; 52]) than the group of %FM adj < 120% (mean 4.7 × 104 ± 1 TF [3.2 × 104; 6.7 × 104] counts/h and mean 67 ± 1 TF [33; 134], respectively).
We undertook multiple regression analysis with lumbar spine BMD vol as the dependent variable and gender, age, height, weight, BMI, pubertal stage, daily and weekly activity scores, %FM, and treatment groups as independent variables. The independent factors, which in combination most strongly predicted lumbar BMD vol in patients, were gender (p < 0.0001), weekly activity score (p < 0.0001), and iv high-dose methotrexate (p = 0.002) accounting for 73% of the variability overall. Thus, male sex, low weekly activity score, and high-dose methotrexate were associated with low lumbar BMD vol.
We found that children with ALL treated with chemotherapy alone had reduced lumbar BMD vol and were physically less active than their healthy controls. Three characteristics: gender, physical activity, and iv high-dose methotrexate explained 73% of the variability in lumbar BMD vol in survivors of ALL.
The difference in BMD was significant only in the spine and after correcting the areal BMD values for bone size using a model by Kröger et al.(29) This model assumes vertebrae to be cylindrical and adjusts the areal BMD by vertebral width as a proxy for depth. This allows for calculation of apparent BMD vol (g/cm3), which has been validated against magnetic resonance imaging (MRI)-derived BMD.(31) Lumbar areal BMD is highly dependent on body size (r2 = 0.71 for height; r2 = 0.75 for weight).(32) Calculation of BMD vol using Kröger's method reduces the dependency on body size (r2 = 0.32 for height; r2 = 0.37 for weight).(32) Correction for bone size is particularly important in subjects with short stature where low areal BMD values can be influenced by small bones.(33) Our patients and control subjects were auxologically well matched (Table 1) and had normal height and weight. However, when examined in more detail, the mean lumbar BA was 2% lower, but the mean width was 2% greater in the ALL group compared with controls. Thus, for the same BA, a vertebra in subjects with ALL was relatively “short” and “wide,” but with a relatively big volume [volume = height × 3.14 × (width/2)](2) for measured area. Accordingly, BMC for bone volume (g/cm3) was more reduced than BMC for the BA (g/cm2) compared with those in controls (p = 0.007 and p = 0.08, respectively). Our results are similar to findings by Arikoski et al.(34) who found that apparent volumetric but not areal lumbar BMD was reduced in 22 patients with ALL at the end of chemotherapy. Results from other studies are conflicting; reduced(16) and normal(13,14) spinal BMD have been found in survivors of childhood ALL without cranial irradiation. In a recent report by Goulding et al., apparently healthy girls with previous fracture and low spinal apparent BMD vol (g/cm3) had nine times the risk of new fractures within 4 years.(35) In a separate study, cross-sectional size of the radius was significantly decreased in girls with forearm fractures compared with those without fractures.(36) Thus, bone size is of importance when assessing bone strength, with clear implication for our population of children with ALL. However, there was no significant history of fractures in either gender; only 1 boy fractured (radius and ulna) during treatment.
Vertebrae consist mainly of trabecular bone where bone turnover is higher than in cortical bone and potentially are more susceptible to damage by factors such as chemotherapy. Glucocorticoids and methotrexate are both known to cause bone loss. Osteoporosis is a well-known complication of long-term use of glucocorticoids. Methotrexate osteopathy has been found in children with ALL and in infants with intracranial tumors treated with high doses of methotrexate.(37–39) The mechanism by which methotrexate affects bone has been suggested to be related to methotrexate's antifolic acid effects(38) and to the inhibition of osteoblast proliferation shown in cultured human osteoblasts.(40) In our study, 18 patients received iv high-dose methotrexate with a cumulative dose of up to 24 g/m2. Although BMD was not significantly different from those treated with intrathecal methotrexate only, the high-dose iv methotrexate regimen was, in combination with gender and activity, an important factor predicting BMD vol. In patients receiving high-dose methotrexate, after adjusting for the effects of gender and activity, lumbar BMD vol was reduced 0.029 g/cm3 that is, nearly 10% of the average BMD vol value.
One of the most interesting findings was that the difference in lumbar BMD vol was seen mainly in male patients. This difference was seen across ages and independently of pubertal status. In Arikoski et al.'s study, male gender and a history of cranial irradiation were associated with lower lumbar and femoral BMD in 29 survivors of childhood ALL investigated a median of 8 years after discontinuing chemotherapy (20 of them received also cranial irradiation).(6) However, in their stepwise regression analysis male gender remained the only significant negative predictor of both lumbar and femoral neck BMD.
Male gonads are more sensitive to certain cytotoxic drugs than the female.(41,42) However, none of our patients received directly gonadotoxic agents such as alkylating agents, procarbazine, vinblastine, cisplatin etc. One boy had slightly delayed puberty (prepubertal at the age of 13.9 years). Even after excluding this boy (and his control) and the other two pairs where pubertal stage did not match exactly (stage 1 vs. stage 2), the difference in lumbar BMD vol remained significant.
Another possible explanation is a gender difference in sensitivity to glucocorticoids. In patients with Addison's disease on steroid replacement, 32% of men versus 7% of women had decreased areal BMD of < −2.0 SDs.(43) There was no difference between men and women with regard to age, duration of treatment, or glucocorticoid dose. In men, but not in women, a significant correlation was found between glucocorticoid dose per body weight and BMD of the lumbar spine, but not of the femoral neck. Thus, it may be that trabecular bone in the male gender is more sensitive to glucocorticoids than it is in the female gender.
DXA scanning provides information about body composition as well as bone density. The mean percentage of body fat in survivors of ALL was not significantly different from controls in either gender. To analyze body composition data within the ALL group, we used the normal reference data of percentage of body fat published by Braillon et al.(30) Our control children had more body fat than expected from the reference values (109% vs. 100%). This may reflect differences in machine technology between Hologic, Inc. and GE Lunar instruments. Using %FM adj adjusted for age and sex, we found that patients who received iv high-dose methotrexate had higher %FM adj than those who received only intrathecal methotrexate. There is no previous data on the effect of methotrexate on body composition in patients with leukemia. Weight gain, a reported problem associated with adjuvant chemotherapy (including methotrexate) for breast cancer, has been attributed to an increase in both total body water and in FM.(44) ALL survivors treated with cranial irradiation have been found to be leptin insensitive; they had higher leptin levels per unit of FM than controls.(45) It has been shown in the ob/ob mouse that leptin has a negative effect on BMD.(46) If leptin resistance was present also in survivors treated with chemotherapy only, this might contribute to the observed reduction in BMD. No data on leptin levels in children treated with chemotherapy only is currently available. We also found that patients with higher percentage of body fat than expected (%FM adj > 120%; n = 13) had significantly lower total BMD than those with %FM adj < 120%, whereas no such difference was seen in the control group. It is known that body fat can affect the DXA measurements of bone mass either by under- or overestimating the true BMD. The higher the proportion of fat in the extraosseus lean tissue, the more the DXA-measured BMD overestimates the true value.(47) Accordingly, in patients with an increased %FM, the true BMD may be even lower than the BMD measured by DXA. Eleven out of the 13 patients in the %FM adj > 120% group received high-dose methotrexate. Therefore, patients who have received high-dose methotrexate are more likely to have increased body fat percentage and reduced BMD 4-5 years after discontinuing chemotherapy. Increased body weight relative to bone mass has been identified as an important predictor of fracture risk in adolescent girls.(35)
Physical activity during childhood and adolescence is an important determinant of body FM and peak bone mass.(24,25) Reduced energy expenditure secondary to reduced habitual physical activity has been found in overweight children treated for ALL.(22) However, the measurement of physical activity in children has been limited by methodological problems. The doubly labeled water method, used in the Reilly study,(22) is not easily available. In this study, we used two methods: a modified validated questionnaire by Godin and Shepard(27) and an accelerometer. The latter has been validated previously as an objective monitor of children's physical activity from preschool years to adolescents in both laboratory and field settings.(48,49) We found that activity level measured by accelerometer during a 3-day period was significantly correlated with the questionnaire activity score in an average week in the ALL group. Mean activity score by accelerometer tended to be higher in boys than in girls with ALL, but the difference was not statistically significant.
Our subjects with ALL were less active than their controls in keeping with other studies.(22,50) After correcting for age and sex, BMD was positively correlated to activity levels measured either by accelerometer or by questionnaire in the ALL group. No such correlations were found in the control group. It is possible that the positive relationship between physical activity and BMD is seen only in individuals with relatively low BMD values. This would be similar to the asymptotic relationship seen between growth hormone (GH) secretion and height velocity SDs in prepubertal children,(51) where subnormal growth rates were associated with low GH secretion (as in GH deficiency), but normal growth rates occurred over a wide range of GH output. An alternative explanation may be the greater variation, that is, dynamic range, in exercise undertaken by ALL children compared with controls (weekly activity score ranged from 14 to 311 in the ALL group and from 20 to 252 in the control group).
In conclusion, we have shown significantly lower lumbar BMD vol values in survivors of childhood ALL than their healthy controls. None of the patients had received cranial irradiation. Reduced lumbar BMD vol was associated with male gender, low weekly activity levels, and iv high-dose methotrexate. Peak bone mass built up during teenage years is an important determinant of fracture risk in late adulthood. Each 1 SD (∼10%) decrement in BMD measurement in adults increases the risk of fracture ∼1.5-fold.(52) Prolonged follow-up will allow us to examine the evolution of changes in BMD in this cohort and may provide further mechanistic information. Nevertheless, the data currently suggest that physical activity is an important factor in determining bone mass in this group and patients who have completed treatment for leukemia should be encouraged to return to normal activity levels as soon as possible.
We thank the children and their parents for participating in our study. V.T. was supported by the Parents Association of Children with Tumors and Leukemia and C.E. was supported by a grant (CP1019/0104) from the Cancer Research Campaign, London, UK.