Abstract
BACKGROUND: Resection of colorectal liver metastases (CLM) can cure disease, but many patients with extensive disease cannot be fully resected and others recur following surgery. Hepatic arterial infusion (HAI) chemotherapy can convert extensive liver disease to a resectable state or decrease recurrence risk, but response varies and no biomarkers currently exist to identify patients most likely to benefit. METHODS: We performed a retrospective cohort study of CLM patients receiving HAI chemotherapy whose tumors underwent MSK-IMPACT sequencing. The frequency of oncogenic alterations and their association with overall survival (OS) and objective response rate were analyzed at the individual gene and signaling pathway levels. RESULTS: Three hundred and seventy patients met inclusion criteria: 189 (51.1%) who underwent colorectal liver metastasectomy followed by HAI + systemic therapy (Adjuvant cohort), and 181 (48.9%) with unresectable CLM (Metastatic cohort) who received HAI + systemic therapy, consisting of 63 (34.8%) with extrahepatic disease and 118 (65.2%) with liver-restricted disease. Genomic alterations were similar in each cohort, and no individual gene or pathway was significantly associated with objective response. Patients in the adjuvant cohort with concurrent Ras/B-Raf alteration and SMAD4 inactivation had worse prognosis while in the metastatic cohort patients with co-alteration of Ras/B-Raf and TP53 had worse OS. Similar findings were observed in a validation cohort. CONCLUSIONS: Concurrently altered Ras/B-Raf and SMAD4 mutations were associated with worse survival in resectable patients, while concurrent Ras/B-Raf and TP53 alterations were associated with worse survival in unresectable patients. The mutual exclusivity of Ras/B-Raf, SMAD4, and TP53 may have prognostic value for CLM patients receiving HAI.