Associations between body composition measurements (BCM) and phase 1 (P1) oncology clinical trial outcomes.

Abstract

e18648 Background: Cancer-related malnutrition and cachexia can lead to body composition changes. BCM can be assessed at the third lumbar (L3) vertebra by CT, which is available as part of pre-trial evaluation. We previously found that malnutrition and low psoas muscle area (PMA) are associated with adverse P1 outcomes including higher rates of ≥ Grade 3 toxicity (G3T). Here we evaluate the relationships between comprehensive cross-sectional muscle and adipose tissue BCM at L3 on P1 outcomes. Methods: Baseline CT scans for 82 patients (pts) were reviewed and images at the level of L3 were identified by 3 independent reviewers. A CT L3 image selected by at least 2 reviewers underwent analysis by Slice-O-Matic software (Tomovision, Canada) to generate BCM including: skeletal muscle area (SMA), skeletal muscle radiodensity (SMD), and adipose tissue area [intermuscular (IMAT), visceral (VAT), subcutaneous (SAT), and total adipose (TAT)] in cm2. SMA was normalized by height (m2), yielding cross-sectional skeletal muscle index (SMI). We stratified pts by having a SMI, SMD, IMAT, VAT, SAT, and TAT above or below the median value. We evaluated for associations between BCM and the following outcomes: rates of ≥ G3T, frequency of dose reductions/interruptions, hospitalizations, tumor response, disease control, duration on study (DOS), and overall survival (OS). Chi-square analysis was used to determine statistical significance between groups. Kaplan-Meier curves were used to compare DOS and OS. A multivariable analysis (MVA) was conducted via logistic regression to evaluate the association between SMI, VAT, PMA and ≥ G3T controlling for age and gender. Results: 82 P1 pts were included (38 M, 45 F), with a median age of 60 (range 28-85). The most common disease site was gastrointestinal (33%). Mean SMI was 44.78 cm2/m2 (range 25.70-79.89). Higher SMI was associated with a reduced risk of ≥ G3T (36.6% vs 58.5%; p = 0.047) and a trend towards improved OS (p = 0.07). There was no association between SMD, IMAT, SAT, or TAT and toxicity, however, higher VAT was associated with reduced risk of ≥ G3T (31.7% vs 63.4%, p = 0.004), and improved response to therapy (p = 0.001). A MVA controlling for age and gender showed that reduced SMI (AUC 0.7072), increased VAT (AUC 0.7597), and reduced PMA (AUC 0.757) were similar in their ability to predict ≥ G3T. Conclusions: P1 trials are designed to determine the safety and tolerability of investigational agents. In this population of P1 pts, BCM including higher baseline CT L3 SMI and VAT were associated with a reduced risk of ≥ G3T. BCM were also tied to efficacy as high VAT was associated with improved tumor response while a trend towards improved OS was noted for pts with higher baseline SMI. Future research should examine the value of integrating CT-based BCM into dose-selection algorithms when evaluating safety in P1 trials to minimize treatment-related toxicity and optimize therapeutic benefit.