Direct measurement is required for accurate assessment of body composition during therapy for acute lymphoblastic leukemia: A report from the T2020-003 IDEAL2 trial

Abstract Background: Obesity adversely impacts disease response and survival for children, adolescents, and adults with acute lymphoblastic leukemia (ALL). We previously demonstrated that patients undergoing induction chemotherapy experience profound changes in body composition within these first 28 days, gaining significant fat mass and losing muscle mass (“sarcopenic obesity”). Changes in weight, and related anthropometrics such as body mass index (BMI), thus may not accurately reflect changes in body composition. This discrepancy has important implications for research and clinical assessments of sarcopenia, cachexia, malnutrition, and metabolic health during therapy. Dual-energy X-ray absorptiometry (DXA) is the ‘gold standard’ imaging modality for body composition assessment. However, serial imaging requires repeated radiation exposure and is logistically challenging to integrate into clinical practice. Therefore, we investigated whether anthropomorphic measures could be used as accurate surrogates for DXA to facilitate conduct of future research trials and integration of body composition assessment into clinical practice. Patients & Methods: The T2020-003 IDEAL2 randomized phase 2 trial (NCT05082519) conducted via the Therapeutic Advances in Childhood Leukemia/Lymphoma consortium is assessing a diet, exercise, and sedentary behavior intervention during induction chemotherapy in youths 7-25 years old to reduce gain in fat mass (FM) and obesity-induced chemoresistance versus standard of care. DXA scans are performed within the first 4 days of therapy when possible and again at end of induction (EOI). Patients are concurrently measured for height, weight, and waist circumference. A planned analysis tested for correlations between FM, body fat percentage (BF%), and lean mass (LM) by DXA with available anthropometric calculations validated in general populations, including: BMI, BMI z-score, relative fat mass (RFM), body roundness index (BRI), a body shape index (ABSI), waist to height ratio (WHtR), and the conicity index (C-Index). Pearson correlation coefficients and associated 95% confidence intervals (CI) were calculated to assess linear correlation. Measurements where the lower bound of the 95% CI was ≥0.9, 0.70-0.89, or 0.4-0.69 were considered to be very strongly (“directly”), strongly, or moderately correlated, respectively. Results: As of July 1, 2025, 45 of 82 (55%) enrolled subjects had DXA scans performed at diagnosis, and 30 (37%) had DXA scans performed at both diagnosis and EOI. Amongst the 45 patients with DXA scans at diagnosis, median age was 15.8 years (range 8.3-21.1), majority were male (80%), and most were of Hispanic or Latino ethnicity by self report (80%). At diagnosis, FM was most directly correlated with BMI (r=0.98 [95% CI 0.963-0.989]) and strongly correlated with BMI Z-score (r=0.85 [0.737-0.916]), BRI (r=0.92 [0.857-0.956]), and WHtR (r=0.91 [0.837-0.949]). BF% was strongly correlated with BMI Z-score, RFM, BRI, and WHtR, but not with BMI. In contrast, no measures strongly or directly correlated with change in FM or BF% during induction. Change in FM was only moderately correlated with BMI (r=0.70 [0.453-0.846]), RFM (r=0.74 [0.491-0.875]), BRI (r=0.74 [0.497-0.877]), WHtR (r=0.76 [0.523, 0.885]). No measure was directly or strongly correlated with LM at diagnosis or change in LM during induction. Conclusion: DXA scans are challenging to obtain urgently at the start of ALL therapy, and <50% of our IDEAL2 study cohort was able to be scanned at both diagnosis and EOI. Multiple anthropometric calculations show excellent correlation to FM and BF% as surrogates of adiposity at diagnosis, but none were sufficiently correlated to estimate change in body composition by EOI. Similarly, anthropometric measurements were not adequate for assessment of LM at diagnosis or change over induction. Future research trials and clinical assessments may rely on anthropometrics as a measure of body fat at diagnosis, but clinical or research assessments focused on treatment-induced changes in body composition, or on assessments of lean mass at any timepoint, will continue to require direct radiographic assessments.