Correlation of Body Mass Index to Dual-Energy X-Ray Absorptiometry in Assessment of Body Composition during Therapy for Childhood Acute Lymphoblastic Leukemia

Introduction

Obesity is associated with increased morbidity and mortality in children diagnosed with high-risk acute lymphoblastic leukemia (HR-ALL). The mechanism of this adverse influence is thought to be due to adipose tissue and adipokines affecting immune function and/or decreasing chemotherapy efficacy. Adiposity, and not body weight, is therefore the central feature for exploring the influence of obesity on outcome in HR-ALL. To date, multiple international consortia have investigated this association using the surrogate measure of Body Mass Index (BMI). BMI estimates body composition based on height and weight and is therefore not able to distinguish between fat mass, bone density, and lean muscle. While BMI is generally a good indicator of overall adiposity, changes in weight that may occur over the course of therapy reflect not only a change in fat mass, but also changes in other components of body composition. We therefore hypothesized that BMI may not be a sensitive measure for assessing change in adiposity during therapy.

Methods

Body composition was assessed serially by BMI and the gold-standard Dual-energy X-ray Absorptiometry (DXA) as an ancillary aim in children with newly diagnosed HR-ALL enrolled on a prospective clinical trial investigating bone health. Children were treated as per Children’s Oncology Group HR-ALL regimens (CCG1961, AALL0232, AALL1131). Body composition was compared at three time-points (TP): within 24 hours of diagnosis (TP1), 28 days later (TP2, end of induction), and at the end of delayed intensification (TP3, mean interval of 8.2 months in the cohort). Age and tanner stage at diagnosis, gender, and ethnicity were collected. DXA was used to analyze body composition as defined by total mass of “body minus head” and separated into respective percentages of lean muscle, fat mass, and bone mineral content. BMI was converted to a percentage per Center for Disease Control and Prevention age- and gender- population norms. BMI percentage was compared to body fat percentage (BF%) overall and individually at each TP. Changes in body composition across TPs were evaluated by DXA. IRB approval was obtained and informed consent documented for all subjects.

Results

Of 51 subjects enrolled in the trial, a sub-cohort of 34 (66.6%) had sufficient DXA data for analysis. Of these, 85% (29/34) had DXA and BMI data at all 3 TPs, while 5/34 were too ill at diagnosis to complete the imaging. There were no significant differences in age, gender, and tanner stage between those included in the DXA sub-cohort and excluded; the cohort with DXA data consisted of a significantly higher prevalence of self-identified Hispanic subjects (31/34, 91% vs 11/17 65%). BF% by DXA was significantly correlated with BMI when observations from all TPs were combined (n=96, Spearman rho 0.6, p=0.002) and at each TP (TP1 rho=0.6, p=0.002; TP2 rho =0.7, p<0.001;TP3 rho=0.6, p=0.002). When change in BMI was compared to change in BF%, however, there was no significant correlation (TP1 vs TP2 rho=0.3, p=0.189; TP1 vs TP3 rho=0.1, p=0.632). Change in weight also did not correlate with change in BF% (TP1 vs TP2 rho=-0.2, p=0.420; TP1 vs TP3 rho=0.2, p=0.311); in fact, average BF% increased across all TPs despite a decrease or no change in weight (Table 1). On evaluation by DXA, change in weight was noted to be primarily due to a loss in lean muscle and gain in fat mass (Table 1).

Conclusion

BMI is an adequate cross-sectional assessment of obesity but was insufficient to assess longitudinal changes in body composition. Continued investigation into the association between ALL and obesity would be aided by prospective trials utilizing the gold-standard DXA or similarly sensitive measurements.