Prognostic and predictive value of a pathomics signature in gastric cancer.

The current tumour-node-metastasis (TNM) staging system alone cannot provide adequate information for prognosis and adjuvant chemotherapy benefits in patients with gastric cancer (GC). Pathomics, which is based on the development of digital pathology, is an emerging field that might improve clinical management. Herein, we propose a pathomics signature (PS) that is derived from multiple pathomics features of haematoxylin and eosin-stained slides. We find that the PS is an independent predictor of prognosis. A nomogram incorporating the PS and TNM staging system shows significantly improved accuracy in predicting the prognosis compared to the TNM staging system alone. Moreover, in stage II and III GC patients with a low PS (but not in those with a high PS), satisfactory chemotherapy benefits are observed. Therefore, the PS could serve as a prognostic predictor in patients with GC and might be a potential predictive indicator for decision-making regarding adjuvant chemotherapy.