Total body bone measurements: A cross-sectional study in children with acute lymphoblastic leukemia during and following completion of therapy

Abstract

Background

Abnormalities in bone mineral density (BMD) occur in children treated for acute lymphoblastic leukemia (ALL). However, BMD estimates have been performed using varied instruments, reference data, and interpretations. This exploratory cross sectional study to evaluate bone mass in children with ALL, uses an algorithm that serially adjusts for variables known to affect pediatric bone measures by dual energy X-ray absorptiometry (DXA), based on models developed in 1,218 healthy children and adolescents.

Procedure

Anthropometry, DXA scans, and factors with possible influence on bone mass were evaluated in 21 ALL patients receiving chemotherapy and 20 in the follow-up phase. Main outcome was treatment group differences in Z-scores for total body bone mineral content (BMC), bone area (Area), and areal BMD (aBMD).

Results

Mean Z-scores for the entire study population for BMC, Area, and aBMD were significantly less than zero. Among possible contributing factors, only calcium intake was a significant co-variate. Comparison between treatment groups showed that least-square mean Z-scores for patients on-therapy for at least 12 months were significantly lower than those off therapy for at least 12 months (P: 0.0008–0.044), except for BMC at last step of the algorithm (adjusted for sex, age, ethnicity, height, weight, and bone area).

Conclusions

Evaluation of total body DXA by this algorithm is consistent with better general bone status in those off-therapy. However, in this small exploratory study, the lack of significant difference between Z-scores for fully adjusted BMC in on- versus off-therapy groups suggests possible risk of low peak bone mass. Additional longitudinal evaluation is warranted. Pediatr Blood Cancer 2009;52:33–38. © 2008 Wiley-Liss, Inc.

INTRODUCTION

Children diagnosed with acute lymphoblastic leukemia (ALL) have been reported to be at increased risk of decreased bone mineral density (BMD), compared to appropriately matched healthy peers 1, 2. Potential etiologies for developing low bone mass in children with ALL include the disease itself 3, long-term corticosteroid use 4, chemotherapy agents including methotrexate 5, and cranial radiation 6. There have, however, been inconsistencies in studies evaluating the effect childhood ALL and its treatment have on BMD. Reported frequencies of BMD more than 1–2 standard deviations below age matched controls range from 8 to 77% 7–10. Some studies have shown no difference in BMD among children with ALL compared to healthy controls after correcting for bone size 11, whereas others report decreased BMD in children with ALL at diagnosis compared to controls, with the difference disappearing as treatment progressed. Given these inconsistencies, there is a need to accurately determine the risk factors for bone-related problems, including low bone mass, in children with ALL. This may help guide the institution of appropriate interventions to reduce morbidity from later osteoporosis.

Estimates of BMD in children, with or without ALL, have been performed using a wide variety of instruments, reference data sets, and interpretation methods, making it difficult to compare results. Indeed, the classification of low bone density or mass in children varies depending on characteristics of the reference population, including the sex and ethnic distribution, as well as the specific scanner model used 12. Use of the same instrument, appropriate pediatric bone mineral reference data, and consistent interpretation approach are necessary to better describe the clinical variation in childhood bone mass and to monitor its changes over time and with treatment. Dual energy X-ray absorptiometry (DXA) has been widely used in the pediatric population 13, 14. However, the clinical meaning of DXA bone results in children has been unclear 13–15.

An algorithm for evaluation of pediatric bone mass was developed from a cross-sectional study of 1,218 healthy ethnically diverse volunteers 13. This sex-specific algorithm adjusts for age, ethnicity, height, weight, and bone area—all variables that affect bone measurements by DXA. This is used to overcome the limitation of the two-dimensional measurement of bone by DXA, as bone is a structure with three dimensions, all of which change with growth and maturation. DXA measures bone area (length and width) only, missing bone thickness. In this exploratory study we evaluated bone mass by DXA in 41 children with ALL, using this algorithm to interpret results from the same scanner used to develop the algorithm.

METHODS

Subjects

The study cohort was a convenience sample of children and adolescents with ALL treated at the Herbert Irving Child and Adolescent Oncology Center at Columbia University Medical Center and its affiliates. Eligibility included diagnosis of ALL between ages 3 and 18 years and status in the active or follow-up phases of chemotherapy. Race and ethnicity were self reported. Informed consent was obtained from all patients, parents or legal guardians as appropriate as per the standards of the Columbia University Medical Center and St. Luke's-Roosevelt Hospital Institutional Review Boards. Assent was obtained from children 7 years and older.

Treatment History

All patients were treated with multi-agent chemotherapy protocols, as per the Children's Oncology Group or Dana Farber Cancer Institute ALL Consortium protocols. The medical record was abstracted for information on leukemia risk group 16(standard vs. high risk), duration of chemotherapy, time since completion of chemotherapy if applicable, history of cranial radiation and primary corticosteroid (prednisone vs. dexamethasone). Patients, who received more than one type of corticosteroid as part of their treatment regimen, were classified by the type of corticosteroid administered for the longer duration. Information on history of bone fractures was collected.

Physical Examination

Body weight was measured to the nearest 0.1 kg on a digital scale (O'Haus CD-11, Pinebrook, New Jersey) and height to the nearest 0.1 cm using a wall-mounted stadiometer (Holtain, Crosswell, Wales). Body mass index (BMI) was calculated as weight in kg divided by the square of the height in m (m²). Z-scores for weight, height, and BMI were derived from the CDC 2000 growth curves 17.

Laboratory Measures

Blood samples were obtained by venipuncture for analysis of serum 25 hydroxy vitamin D (25-OHD) and 1,25 dihydroxy vitamin D (1.25-OHD) and intact parathyroid hormone (PTH) levels. Sera for these assays were stored and the assays were run as a group. These tests are not obtained as part of routine clinical management. Details of the assays are provided in the supplemental information section.

Bone Mineral Measurements by DXA and Interpretation

Measures of total body bone mineral content (BMC in g), total body bone area (Area in cm²), and total body aBMD (g/cm²) were obtained by DXA (Lunar DPX-L, Lunar Corporation, Madison, WI, pediatric software version 3.8G) between May 7, 2003 and February 18, 2005. Details of our quality control procedures are provided in the supplemental section 18.

The total body bone measurements were interpreted using a three level algorithm developed from results in 1,218 healthy volunteers ages 6–18 years using the same scanner, representing Asian, African-American, Hispanic, White, and other backgrounds. Details of the algorithm have been reported previously 13. Briefly, it serially adjusts for factors known to affect DXA measurements, including age and ethnicity (Level 1), height and weight (Level 2), and bone area (Level 3) based on sex-specific models (Table I). Serial application of Levels 1–3 evaluates the contributions of height and weight and then bone size (estimated as two-dimensional bone area with DXA) to total body bone mass for an individual, compared with the expected effects of these variables on bone mass among healthy subjects of the same sex, age, and race. If stepwise adjustment for these variables results in appropriate values for the patient's reference group, then intrinsic bone mass compromise is unlikely. However, if adjustment yields lower-than-expected values then a deficit in bone mineral mass is suggested. The clinical approach for results that are consistent with small body size (improved Z-scores at Level 2) or small bones (improved Z-scores at Level 3) would differ from that for low bone mass that is independent of body and bone size (persistently low Z-scores at Level 3). This interpretation in steps prevents over-adjustment that may mask information. The outcomes are sex-specific Z-scores for bone measurements at each level of adjustment that may suggest why an individual or group of patients has Z-scores for bone measurements that differ from expectation.

Table I. Variables Included in Sex-Specific Adjustments of Total Body Bone Measurement Results by DXA 13

Level	Bone measurement	Variables adjusted for at each Level
		Age	Race/ethnicity	Height	Weight	TBBA
a Total body bone mineral content by DXA (g); b Total body bone area by DXA (cm^2); c Total body bone mineral density (g/cm^2, “areal density”).
1	TBBMCa	x	x
	TBBAb	x	x
	TBBMDc	x	x
2	TBBMC	x	x	x	x
	TBBA	x	x	x	x
	TBBMD	x	x	x	x
3	TBBMC	x	x	x	x	x
	TBBMD	x	x	x	x	x

Behavioral Measures

Study participants completed questionnaires for assessment of dietary intake of calcium and vitamin D, and of physical activity. Dietary intake was assessed with the Harvard Youth Adolescent food frequency questionnaire, which is available in both English and Spanish 19. This method captures intake of total calories, calcium, fat-soluble vitamins, and vitamin D over the preceding year. Recommendations were that parents complete the survey for subject ages 3–11 years, and parents assist in completion of the survey as needed for subjects 12 years and older.

Physical activity was assessed with a modified Slemenda questionnaire which asks the subject to recall activities for the week prior to the study visit by checking off from a list of sports and activities and choosing defined time intervals for each 20. The subject is also asked if the last week is representative of his or her usual pattern of physical activity. The questionnaire is completed by the subject with assistance from a parent as needed. The variable used for this analysis was the screen time (television/computer time) per week.

Statistical Methods

Descriptive statistics, mean, standard deviation, and range, were calculated for the Z-scores of the bone measurements, weight, height, and BMI. Analysis of covariance was used to test the hypothesis that the mean Z-scores for the different subsets of the ALL group were equal. Calcium intake estimated from the food frequency questionnaire was used as a covariate in all analyses. Multiple comparisons were performed using Fisher's protected least significant difference procedure. The adjusted mean scores are presented in the tables. Separate analyses were presented for the Z-scores for each bone measurement, BMC, aBMD, and Area. The level of significance for all statistical tests was 0.05. All statistical calculations were performed using the SAS statistical software package for personal computers.

RESULTS

Participant Characteristics

The characteristics of the study population are presented in Table II. Forty-five subjects were enrolled; 41 subjects were evaluable. Two subjects did not complete the DXA scans because of multiple missed appointments and were therefore excluded from further analyses. Two additional subjects, ages 21 and 26 years at the time of the DXA scans, were diagnosed with ALL between ages 3 and 18 years; their data was excluded from the analyses as they were considerably above the upper age limit of the algorithm such that comparisons could not be made to the reference population. Z-scores for weight and BMI were significantly greater than zero (Table III). There were no sex or race differences in Z-scores.

Table II. Participant Characteristics

	All patients (N = 41)
Age: median (range)	10 (range 3–19)
Gender (male:female)	25:16
Race/ethnicity
Asian	1 (2.5%)
Hispanic	21 (51%)
African-American	2 (5%)
White	16 (39%)
Other	1 (2.5%)
Treatment status
Treatment group 1
On-therapy < 1-year	7 (17%)
Median age (years)	5
Treatment group 2	19 (46%)
On-therapy > 1-year and	14 (34%)
Off-therapy < 1-year	5 (12%)
Median age (years)	8
Treatment group 3
Off-therapy > 1-year	15 (37%)
Median age (years)	12
Primary corticosteroid
Prednisone	26 (63%)
Dexamethasone	15 (37%)
Cranial radiation	9 (22%)
History of fracture	4 (10%)
25 OH vitamin D: mean (range)	23.1 ng/ml (6.0–36.9)
1,25 Di-OH vitamin D: mean (range)	54.6 pg/ml (14.3–104.5)
Intact PTH: mean (range)	34.6 pg/ml (6.6–122.7)

Table III. Study Population Z-Scores for DXA Total Body Bone Measurements and for Height, Weight, and BMI*

	Level 1a			Level 2a			Level 3a		CDC 2000b
	BMC	Area	aBMD	BMC	Area	aBMD	BMC	aBMD	Ht	Wt	BMI
Mean Z-score is significantly different from zero are denoted by bold values. Total Body Bone and Anthropometric Measurements: BMC, bone mineral content (g); Area, bone area (cm^2); aBMD, areal bone mineral density (g/cm^2); BMI, body mass index [wt in kg/(ht in m^2)]; a Level (1) Z-score for sex, age, and ethnicity; Level (2) Z-score for sex, age, ethnicity, height, and weight; Level (3) Z-score for sex, age, ethnicity, height, weight, and bone area 13; b Z-scores from CDC 2000 Growth curves 17.
Mean	−0.4	−0.4	−0.8	−0.9	−0.5	−1.1	−0.8	−1.0	−0.2	0.6	0.8
SD	1.1	1.1	1.2	1.4	1.3	1.5	1.3	1.3	0.9	1.1	1.3
Range	−3.4, 1.5	−3.1, 2.1	−3.7, 1.6	−4.4, 2.7	−4.3, 2.2	−3.8, 2.3	−3.6, 1.9	−3.9, 1.5	−2.0, 1.8	−2.3, 2.4	−3.3, 2.5

Twenty-one patients were receiving chemotherapy, and 20 patients had completed all planned chemotherapy. Prednisone and dexamethasone were the primary corticosteroids used in 26 subjects and 15 subjects, respectively. Nine subjects had received cranial radiation as part of their therapy for ALL and were evenly divided between the on therapy and off therapy groups. Four subjects had a history of bone fracture in the interval since diagnosis of ALL. All fractures occurred while patients were receiving chemotherapy, with one fracture (radius) during induction, and three fractures (heel, radius, and ankle) sustained during the consolidation/intensification phase.

The mean 25-OHD, 1,25-OHD, and PTH levels were as follows: 24.6 ng/ml (6.0–103.5), 54.6 pg/ml (14.3–104.5), and 34.6 pg/ml (6.6–122.7). One 25-OHD level (6.0 ng/ml) was below the reference range, but the PTH value for this patient was in the normal range. Three individuals had PTH levels above the reference range (122.7, 82.6, and 69.2 pg/ml), with 25-OH levels of 28.9, 16.9, and 17.0 ng/ml respectively. There were no clinical features shared by these patients. All patients with low vitamin D levels were male, but there were no associations with race, leukemia risk group, treatment protocol, use of cranial radiation or history of fractures in this subgroup.

Calcium intake was inadequate (as defined by intake of calcium below age adjusted adequate intake values 21) in 19 (50%) of 38 subjects with available data. Only one (7%) of 15 subjects below age 8 years reported low intake of calcium, whereas 18 (78%) of 23 subjects ages 9 years and above had inadequate intake. The percentage of subjects with inadequate calcium intake did not differ when looking at calcium intake from the diet alone, or diet intake with calcium supplements combined.

Total Body Bone Measurements by DXA

Z-scores for the entire study group for total body BMC and aBMD from Levels 1 to 3 and for Area from Levels 1 to 2 of the algorithm 13 are presented in Table III. For the entire study group, all Z-scores were significantly less than zero. There were no sex or race differences in bone Z-scores.

We considered factors that could be possible predictors of the observed Z-scores for DXA bone measurements (Supplemental Table I). Cranial irradiation, primary steroid use, history of fractures, serum vitamin D and PTH, nutritional measures, and physical activity were not significant predictors of bone measurement Z-scores. Two factors were significant predictors in the final models; (1) the distinction between actively receiving chemotherapy versus completing planned chemotherapy and (2) calcium intake. The effect of calcium intake was not large, but it remained significant with therapy status in the model.

Based on this finding, the patients were further divided into three treatment groups: Group 1: subjects on chemotherapy for 12 months or less (n = 7); Group 2: subjects on chemotherapy for more than 12 months or off chemotherapy for 12 months or less (n = 17); and Group 3: subjects off chemotherapy for more than 12 months (n = 17). These time periods were consistent with reports of fracture risk 22, with subjects in group 2 being at highest risk.

After adjustment for calcium intake, the three therapy groups' least squares mean values for Z-scores for total body DXA bone measurements at Levels 1–3 were compared to each other and to zero (Table IV). The Group 3 Z-scores were not different from zero, and were significantly greater than those for Group 2 (P < 0.001–0.017), except for BMC at Level 3 (final adjustment for bone area). All Group 2 Z-scores were significantly different from zero (less), as well as significantly less than either Group 1 or 3 or both.

Table IV. Pair Wise Comparison Between Treatment Groups for Total Body Bone Measurement Z-Scores [Least Square Mean (SE), Adjusted for Calcium Intake]

Group	Level 1a			Level 2a			Level 3a
	BMCb	Areab	aBMDb	BMC	Area	aBMD	BMC	aBMD
Least square mean Z-score is significantly different from zero are denoted by bold values. a Levels of sex-specific algorithms for assessment of total body bone measurements by DXA with serial adjustments for (1) age, and ethnicity; (2) age, ethnicity, height, and weight; (3) age, ethnicity, height, weight, and bone area 13; b BMC, bone mineral content (g); Area, bone area (cm^2); aBMD, areal bone mineral density (g/cm^2); * Significantly different from Group 3; ** Significantly different from Group 1.
1. On therapy < 12 months	−0.15 (0.29)	−0.88 (0.36)^1	−0.42 (0.39)	−0.85 (0.52)	−0.59 (0.50)	−0.71 (0.51)	0.18 (0.48)	−0.46 (0.48)
2. On therapy > 12 mos OR Off therapy < 12 months	−0.99 (0.20)^1,2	−0.86 (0.25)^1	−1.56 (0.27)^1,2	−1.76 (0.36)^1	−1.17 (0.34)^1	−2.09 (0.36)^1,2	−1.38 (0.33)^2	−1.72 (0.33)^1,2
3. Off therapy > 12 months	−0.06 (0.19)	0.02 (0.23)	−0.21 (0.25)	−0.27 (0.34)	0.02 (0.32)	−0.29 (0.33)	−0.51 (0.31)	−0.41 (0.31)
P-values	0.0020*	0.0440* (2 vs. 1)	0.0008*	0.0046*	0.0171*	0.0008*	0.0119**	0.0072*
	0.0255**	0.0139* (3 vs. 1)	0.0226**			0.0359**		0.0409**

DISCUSSION

The children and adolescents with ALL evaluated in this study had lower than expected total body BMC, Area, and aBMD as adjusted for sex, age, and race/ethnicity. These findings persisted after adjustment for height and weight, and after further adjustment for bone area (an estimate of bone size). When divided into three groups based on treatment status, the decrease in bone mass was most striking in patients on chemotherapy for more than 12 months or off for 12 months or less compared with those off chemotherapy for more than 12 months. The total body bone results were less than expected at all levels of the algorithm, and significantly less except for BMC at Level 3 (final adjustment for bone area). The lack of between-treatment group difference for this value at the last step of the algorithm suggests that there may be a persistent effect of therapy on total body bone mass. The findings in this small pilot cross sectional study suggest that there is an effect of therapy on total body bone mass in pediatric patients with ALL, that this improves when therapy ceases, but that there may be a persistent decrease in BMC for bone area, an estimate of bone size.

These results are consistent with a report on DXA measurements of total body and lumbar spine and peripheral quantitative CT (pQCT) of the radius in 53 ALL survivors 23. The patients had completed therapy at least 1-year before the study. Borderline greater total body and L1–L4 spine bone area in the ALL survivors was observed compared to controls. Decreased midradial cortical thickness but increased periosteal and endosteal circumference by pQCT was observed. This is consistent with decreased BMC within the bone envelope measured by DXA. Although we cannot conclude that what they observed in a cross section of the radius explains our total body DXA findings, the outcome from our algorithm is consistent with the geometry they measured by pQCT.

Our findings are consistent with other reports 23–25, in that that we noted the most significant impact on bone status in subjects at least 1-year on chemotherapy and within 1-year following completion of therapy. In a longitudinal study of total body aBMD 25, a significantly higher increase in aBMD was observed in survivors of ALL as compared with gender, age, and pubertal stage matched controls, suggesting that there is an improvement in total bone mass with time.

The relationship of total bone mass to fracture risk during treatment cannot be adequately assessed in our small study. History of fracture was not a predictor of low Z-scores in this diverse patient group, but decrement with therapy may be as important as a low absolute Z-score 24. Findings from our algorithm for DXA measures may contribute to understanding the mechanism. In the off-therapy group the least squares mean Z-scores were not significantly below zero and significantly greater than the on-therapy group at Levels 1–2. It was only for BMC after adjustment for bone area at Level 3 that these groups did not differ, suggesting that bone area for age, height, and weight were fine, but BMC could be decreased. This could be explained by decreased bone thickness (the dimension not measured by DXA), decreased cortical thickness (consistent with the pQCT of the radius results) or deficit in volumetric bone density. Longitudinal evaluation of ALL patients with both DXA and techniques that assess regional volumetric density and bone geometry could address these mechanisms.

Of all the factors considered as predictors for Z-scores for bone measures, only therapy status and calcium intake by food frequency questionnaire were significant. The effect of calcium intake was not large, but it remained significant with therapy status in the model. Although the comparison of treatment group mean values was similar before and after adjustment for calcium intake, the addition of calcium information reduced the variability. This may have clinical implications as low calcium intake in childhood has been associated with reductions in bone mass and increased fracture risk in adults, although its impact on BMD in children is less clear 26, 27. The lack of effect of the primary steroid (prednisone or dexamethasone) on bone outcomes is consistent with recent reports 28, 29.

A major strength of our study is the interpretation of the bone status of the subjects with ALL using an algorithm that serially adjusts for variables that affect bone measurements by DXA, but are not part of the manufacturers' current reference standards 13. The recently reported decrease in adult stature compared to siblings in ALL survivors, including those treated with chemotherapy alone, emphasizes the importance of taking height (a major contributor to bone mass) into account in evaluation of bone status 30.

The main limitation of our cross-sectional pilot study is that we cannot assess change in bone mass in individual subjects, or evaluate the effects of specific treatment or other contributing factors. Longitudinal evaluation of patients at diagnosis and during therapy with attention to patients with fracture and specific bone measurement patterns is warranted.

DXA is widely available, safe, and easy to perform in children and adolescents. Consistent interpretation of longitudinal findings in patients with ALL could further evaluate the findings of this exploratory study suggesting that deficit in bone mass may persist 1-year or more after therapy. Interpretation with an approach like our algorithm that adjusts for factors known to affect DXA bone measures may suggest mechanisms and identify specific risk factors.

Acknowledgements

This work was supported by the Joseph LeRoy and Ann C. Warner Fund, Inc.