Radiomic Alterations In Tumor Microenvironment By Induction Chemotherapy Predict Treatment Outcome Of Subsequent Concurrent Chemoradiotherapy For Locally Advanced Hypopharyngeal Cancer

Abstract

Induction chemotherapy (IC) followed by concurrent chemoradiotherapy (CCRT) is recommended for larynx-preserving treatment of locally advanced hypopharyngeal cancer (LAHC). However, biomarkers to predict CCRT response for selecting larynx-preserving candidates remain lacking. This study aimed to develop an imaging biomarker using changes in radiomic features in tumor microenvironment (TME) by IC to predict treatment outcome of subsequent CCRT in LAHC. From 2006 to 2018, computed tomography (CT) scan images obtained by the same protocol before and after IC in LAHC patients were used to contour the gross tumor volumes (GTVs). By using LIFEx software (V.5.10), 47 delta-radiomics features were acquired from the absolute spatial difference of GTVs (delta-GTV) before and after IC, conceptually representing a consistent portion of TME. The data was divided to training/testing groups in the ratio of 8:2. Least absolute shrinkage and selection operator regression (LASSO) in R language was used to select features for establishing the model generating radiomic score (R score). The cutoff of R score for progression was determined by Youden index method. Multivariate regression model was used to identify prognostic factors for progression free survival (PFS). Hematoxylin and eosin stain of biopsied specimens were reviewed to identify histological phenotype of cancer-associated fibroblast (CAF). Mature (thin, wavy, and small spindle-shaped fibroblasts) and immature (large, plump spindle-shaped with prominent nucleoli) groups were categorized by using cutoff 50%. Total 59 LAHC patients with qualified paired CT images were subjected to delta-radiomic analysis. The median follow-up time was 33.5 (IQR 17.9-74.0) months. A model including 5 radiomic features from delta-GTV to predict better PFS of patients receiving subsequent CCRT was established (likelihood ratio test p < 0.05; Table). R score was validated with all dataset (area under the curve = 0.74). Low R score (< -0.34), but not volume changes of GTVs, was associated with better PFS in both univariate (p < 0.05) and multivariate (p < 0.05) analysis. In TME assessment, the histological phenotype of mature CAF, but not tumor infiltrating lymphocyte, correlated with better PFS. Low R score correlated with mature CAF category (positive and negative predictive values: 100% and 75%, respectively), implying CAF as a putative biological rationale behind delta-radiomics. The established model for TME from radiomic features of delta-GTV after IC might be a potential imaging biomarker to predict clinical outcome of subsequent CCRT in LAHC. Phenotype of CAF in TME may correlate with radiomic alterations and clinical outcome.Abstract 3864; TableThe radiomic features obtained from regression modelCoefficientHazard ratiop valueCONVENTIONAL_HUmin0.00021.00020.83HISTO_Skewness0.24721.28040.02SHAPE_Sphericity13.86051045981.78570.04GLRLM_GLNU-0.00030.99970.25GLZLM_LZLGE-0.00650.99360.64 Open table in a new tab