Radiomic and clinical predictors of cachexia in non-small cell lung cancer patients treated with immunotherapy

BackgroundCachexia is present in up to 50% of patients with cancer and may contribute to primary resistance to immunotherapy. Biomarkers to predict cachexia are urgently required for early intervention. Herein, we test the hypothesis that pre-treatment 18F-FDG-PET/CT-based radiomics can be used to predict cachexia and subsequently associated with clinical outcomes among patients with advanced non-small cell lung cancer (NSCLC) who are treated with immunotherapy. MethodsThis retrospective multi-institution study included 210 patients with histologically confirmed stage IIIB-IV NSCLC who were treated with immune checkpoint blockade between June 2011 and August 2019. Baseline (pre-immunotherapy) PET/CT images of 175 patients from Moffitt Cancer Center were used to train (N=123) and test (N=52) a radiomics signature to predict cachexia, which was also used to predict durable clinical benefit (DCB), progression-free survival (PFS) and overall survival (OS) subsequently. An external cohort that enrolled 35 patients from James A. Haley Veterans Hospital (VA) was used to further validate the predictive and prognostic value of this signature. ResultsA radiomics signature demonstrated cachexia prediction ability with areas under receiver operating characteristics curves (AUC) of 0.77 (95%CI:0.68-0.85), 0.75 (95%CI:0.60-0.86) and of 0.73 (95%CI:0.53-0.92) in the training, test and external VA cohorts, respectively. For the further investigation of prognostic value, this signature could identify the patients with DCB with AUC of 0.67 (95%CI:0.57-0.77), 0.66 (95%CI:0.51-0.81), and 0.72 (95%CI:0.54-0.89) in these three cohorts. Additionally, the PFS and OS were significantly shorter among patients with higher radiomics signature in all the three cohorts (p<0.05). ConclusionUsing PET/CT radiomics analysis, cachexia could be predicted before the start of the immunotherapy, making it possible to monitor the patients with a higher risk of cachexia and identify patients most likely to benefit from immunotherapy.