Author：Zhihao Lu, Huan Chen, Shuang Li, Xi Jiao, Lihong Wu, Jianing Yu, Lin Shen, Henghui Zhang

Background: Cancer therapy has been greatly revolutionized in recent years by the conceptual developments in the field of cancer immunology. Growing evidence support the utility of immune checkpoint inhibition (ICB) in gastrointestinal (GI) cancer. However, a central question lies in understanding how therapeutic responsiveness is predicted. 

Methods: To address this, we evaluated tumor FFPE specimens from 97 patients who received ICB treatment. All patients were randomly assigned into discovery (60%) and validation (40%) cohorts. Tumor RNA before ICB treatment was analyzed on a multiplex RNA immune oncology (RNA IO) profiling sequencing panel. 

Results: We show that four immune-related gene expression signatures were upregulated in responders versus non-responders in the discovery cohort. Three of the four signatures showed significant correlation with clinic response and disease control rates. However, two previously reported RNA signatures, PD-L1 expression and MMR status revealed less predictive values in GI cancers. More importantly, we identified that higher levels of a 19-gene signature were remarkably associated with favorable overall survival (OS) and progression-free survival (PFS) when compared to patients with lower levels of signature in both the discovery and validation cohorts. Of note, a joint biomarker of tumor mutation burden (TMB) and the 19-gene signature may better stratify responders from non-responders in GI cancer patients. 

Conclusions: Our data provide evidence that a responsive feature, defined by a multi-gene expression pattern across different GI cancer types, can be obtained via a RNA quantitative strategy and may be explored as a future pan-cancer biomarker.