Background: Cancer recurrence is an important predictor of survival outcomes in patients with colorectal liver metastasis (CRLM), undergoing curative resection. While cancer recurrence occurs frequently in patients with CRLM, perioperative treatment remains arbitrary and the NCCN guidelines recommended several treatment options, including colectomy, hepatectomy, or systemic chemotherapy. Therefore, identification of patients who are at highest risk of recurrence is critical for developing a precision oncology strategy which might include just frequent surveillance (in low-risk patients) or a more aggressive treatment approach with chemotherapy or other treatments (in high-risk patients). Herein, we performed a genomewide expression profiling analysis to identify and develop a gene signature for predicting recurrence in patients with CRLM.

Methods: We analyzed a total of 383 CRLM patients, which included a genomewide expression dataset of 63 patients (GSE81423) and 320 patients from independent clinical cohorts. During the biomarker discovery phase, we utilized rigorous bioinformatic approaches to analyze transcriptomic profiling data in CRLM patients with and without recurrence. The robustness and performance of this gene signature was subsequently interrogated in independent clinical cohorts (training cohort, n=169; validation cohort, n=151) using qRT-PCR assays. Finally, using Cox regression analysis, we evaluated the clinical significance of the gene signature by comparing its performance with several key clinicopathological factors.

Results: We identified a 6-gene panel, which robustly categorized patients with recurrence in the discovery cohort (AUC = 0.90, 95% CI = 0.79-0.96). The validation efforts in the clinical cohorts revealed that the panel was a significant predictor of recurrence in the training (AUC = 0.83, 95% CI = 0.76-0.88) and validation cohorts (AUC = 0.81, 95% CI = 0.74-0.87). In univariate and multivariate analysis using the Cox's proportional hazard regression model along with other clinicopathological factors, high-risk patients defined by the transcriptomic panel associated with a significantly higher risk of recurrence in both cohorts [training cohort: Hazard ratio (HR) = 3.18; 95% CI = 2.03-4.98; P < 0.01, validation cohort: HR = 2.60; 95% CI = 1.73-3.92; P < 0.01]. By combining our transcriptomic panel with key clinical features, we established a risk-stratification model which was even superior than the gene panel, and emerged as an independent predictor for recurrence (AUC = 0.85, 95% CI = 0.78-0.90).

Conclusions: We identified and developed a novel gene signature that robustly predicts recurrence in CRLM patients; highlighting its clinical impact in the a more appropriate patient selection and development of improved precision treatment strategies in patients with advanced metastatic colorectal cancer.

Citation Format: Yuma Wada, Mitsuo Shimada, Ajay Goel. A novel gene signature that predicts recurrence following hepatectomy in patients with colorectal liver metastasis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 651.