Differential prognostic impact of liver resection by the site of concurrent extrahepatic diseases in patients with colorectal cancer liver metastases: a nationwide multicenter study (correspondence)

Dear Editor, We read with great interest the results of a nationwide multicenter study by Kato et al1 suggesting that hepatectomy improves overall survival in patients with colorectal liver metastases (CRLM) and concurrent extrahepatic disease (EHD). While we commend the effort, we feel that certain pitfalls substantially limit the causal inferences drawn from the results in this study. We confirm that this correspondence adheres to the requirements outlined in the TITAN Guideline Checklist 20252. The study is highly susceptible to immortal-time bias, which is not addressed in the analysis. Overall survival is measured from the date of CRLM diagnosis, but hepatectomy occurs later, often months later, creating an inherent survival advantage for the surgical group because patients must survive long enough, remain resectable, and maintain adequate performance status to reach surgery. This bias cannot be corrected using inverse probability of treatment weighting alone. Contemporary methodology in oncology clearly demonstrates that any treatment delivered after cohort entry must be modeled as a time-dependent exposure or analyzed with a landmark approach to avoid falsely attributing survival time to the intervention itself. Recent work in advanced colorectal cancer illustrates how standard survival analyses can exaggerate the benefit of surgery when immortal time is not explicitly accounted for, even in large population-based datasets3. For precisely this reason, methodological studies broadly emphasize that time-dependent covariates are essential in comparative surgical research4. Without addressing the issue, the hazard ratios reported by Kato et al. likely overstate the true effect of hepatectomy. The study underestimates the extent to which biological selection, rather than surgery itself, may account for the observed survival differences. Patients who undergo hepatectomy represent a highly selected subgroup with more favorable molecular and chemotherapy-response characteristics. Yet, key biological determinants such as RAS and BRAF mutation status, microsatellite instability, and radiological tumor response were not included in the propensity model. Recent prospective studies demonstrate that molecular profiles directly shape resectability and long-term outcomes in metastatic colorectal cancer. RAS or BRAF mutations, in particular, strongly predict reduced likelihood of resection and inferior survival5. In-depth analyses have also shown that BRAF mutation status and molecular subtype substantially modify recurrence risk after CRLM surgery6. Lack of these variables likely leaves substantial residual confounding by tumor biology. The EHD classification used in the study oversimplifies a complex and heterogeneous prognostic landscape. A patient with both lung and peritoneal metastases was categorized into each site-specific subgroup, complicating interpretation. This is concerning as prognosis does not merely depend on the presence of a given site but on metastatic burden, combinations of sites, and disease biology7. For instance, peritoneal metastases are not a uniform entity, as extensive heterogeneity exists in response to systemic therapy, patterns of spread, and suitability for cytotoxic interventions8. Therefore, we believe that the study risks overgeneralizing its conclusion that hepatectomy is beneficial regardless of EHD site. Moreover, recent large-scale analyses reaffirm that resection in the setting of EHD can be beneficial but only in carefully selected patients with minimal, controllable extrahepatic burden9. Hence, selection bias, not purely surgical efficacy, remains central to these outcomes. Kato et al’s conclusions therefore warrant more caution than stated. In summary, while the study contributes important multicenter data, the unaddressed immortal-time bias, omission of biological and response-based variables, and oversimplified treatment of EHD heterogeneity limit its causal interpretations. Refining analytical techniques and incorporating molecular and disease-response parameters will be essential for future investigations aiming to define which patients with CRLM and EHD are truly candidates for hepatic resection.