Information about immune cell proportions and tumor stage improves the prediction of recurrence in patients with colorectal cancer

Summary: Predicting cancer recurrence is essential to improving the clinical outcomes of patients with colorectal cancer (CRC). Although tumor stage information has been used as a guideline to predict CRC recurrence, patients with the same stage show different clinical outcomes. Therefore, there is a need to develop a method to identify additional features for CRC recurrence prediction. Here, we developed a network-integrated multiomics (NIMO) approach to select appropriate transcriptome signatures for better CRC recurrence prediction by comparing the methylation signatures of immune cells. We validated the performance of the CRC recurrence prediction based on two independent retrospective cohorts of 114 and 110 patients. Moreover, to confirm that the prediction was improved, we used both NIMO-based immune cell proportions and TNM (tumor, node, metastasis) stage data. This work demonstrates the importance of (1) using both immune cell composition and TNM stage data and (2) identifying robust immune cell marker genes to improve CRC recurrence prediction. The bigger picture: Colorectal cancer is a significant global public health issue, and accurately predicting its recurrence remains a challenge despite advances in screening and treatment. Accurate recurrence prediction is crucial for clinicians to make informed decisions about treatment and follow-up care, leading to timely interventions that may improve outcomes and potentially prolong survival. In this study, we present a method that combines immune cell information with clinical data to enhance the accuracy of recurrence risk prediction. The predictive performance of the method was validated using two independent cohorts of patients with colorectal cancer of different races. Our approach has implications for both clinical practice and research, as it helps to suggest treatment strategies by predicting recurrence risk and identifying potential therapeutic targets based on immune cell information.