Pre-treatment MDCT-based texture analysis for therapy response prediction in gastric cancer: Comparison with tumour regression grade at final histology

Purpose An accurate prediction of tumour response to therapy is fundamental in oncology, so as to prompt personalised treatment options if needed. The aim of this study was to investigate the ability of preoperative texture analysis from multi-detector computed tomography (MDCT) in the prediction of the response rate to neo-adjuvant therapy in patients with gastric cancer. Material and methods Thirty-four patients with biopsy-proven gastric cancer were examined by MDCT before neo-adjuvant therapy, and treated with radical surgery after treatment completion. Tumour regression grade (TRG) at final histology was also assessed. Image features from texture analysis were quantified, with and without filters for fine to coarse textures. Patients with TRG 1–3 were considered responders while TRG 4–5 as non- responders. The response rate to neo-adjuvant therapy was assessed both at univariate and multivariate analysis. Results Fourteen parameters were significantly different between the two subgroups at univariate analysis; in particular, entropy and compactness (higher in responders) and uniformity (lower in responders). According to our model, the following parameters could identify non-responders at multivariate analysis: entropy (≤6.86 with a logarithm of Odds Ratio − Log OR −: 4.11; p = 0.003); range (>158.72; Log OR: 3.67; p = 0.010) and root mean square (≤3.71; Log OR: 4.57; p = 0.005). Entropy and three-dimensional volume were not significantly correlated (r = 0.06; p = 0.735). Conclusion Pre-treatment texture analysis can potentially provide important information regarding the response rate to neo-adjuvant therapy for gastric cancer, improving risk stratification.