Delta-radiomics and response to neoadjuvant treatment in locally advanced gastric cancer-a multicenter study of GIRCG (Italian Research Group for Gastric Cancer).

BACKGROUND

To predict response to neoadjuvant chemotherapy (NAC) of gastric cancer (GC), prior to surgery, would be pivotal to customize patient treatment. The aim of this study is to investigate the reliability of computed tomography (CT) texture analysis (TA) in predicting the histo-pathological response to NAC in patients with resectable locally advanced gastric cancer (AGC).

METHODS

Seventy (40 male, mean age 63.3 years) patients with resectable locally AGC, treated with NAC and radical surgery, were included in this retrospective study from 5 centers of the Italian Research Group for Gastric Cancer (GIRCG). Population was divided into two groups: 29 patients from one center (internal cohort for model development and internal validation) and 41 from other four centers (external cohort for independent external validation). Gross tumor volume (GTV) was segmented on each pre- and post-NAC multidetector CT (MDCT) image by using a dedicated software (RayStation), and 14 TA parameters were then extrapolated. Correlation between TA parameters and complete pathological response (tumor regression grade, TRG1), was initially investigated for the internal cohort. The univariate significant variables were tested on the external cohort and multivariate logistic analysis was performed.

RESULTS

In multivariate logistic regression the only significant TA variable was delta gray-level co-occurrence matrix (GLCM) contrast (P=0.001, Nagelkerke R: 0.546 for the internal cohort and P=0.014, Nagelkerke R: 0.435 for the external cohort). Receiver operating characteristic (ROC) curves, generated from the logistic regression of all the patients, showed an area under the curve (AUC) of 0.763.

CONCLUSIONS

Post-NAC GLCM contrast and dissimilarity and delta GLCM contrast TA parameters seem to be reliable for identifying patients with locally AGC responder to NAC.