Colorectal cancer liver metastases: Improving selection of patients for different treatment strategies

Potential under- and overtreatment of patients with colorectal cancer liver metastases (CRLM) is considered a clinical problem. Proper patient selection for local and/or systemic treatment strategies may improve outcome in these patients. So, this thesis aimed to tailor personalized treatment plans by making multidisciplinary liver expertise more accessible and to provide new insights in treatment of early recurrent CRLM. Additionally, the performance of radiomics in predicting response to treatment was evaluated. The added value of a dedicated multidisciplinary panel of liver specialists to assess feasibility of local treatment strategies in patients suffering from CRLM is studied in Chapter 2. Diagnostic imaging were reviewed by an expert panel consisting of four hepatobiliary surgeons and two interventional radiologists. In 20.0% of patients initially assigned to systemic treatment of CRLM, upfront local treatment was deemed feasible after re-evaluation by this expert panel. Assessment by such panels may reduce undertreatment of patients with CRLM. The implementation of an online expert panel in clinical practice is reported in Chapter 3. This quality improvement initiative was introduced to provide non-liver centers with online multidisciplinary liver expertise and select patients for referral. Chapter 4 highlights the rules and regulations, which apply for e-consultation services, under current legislation in the Netherlands. There are no legal obstacles or practical objections to the introduction of expert panels in clinical practice, provided that these requirements are met. Chapter 5 describes the perspectives of patients on the use of online expert panels in transmural care. Based on two focus groups, three major themes were identified: ‘data management’, ‘expertise’, and ‘information and coordination’. The potential hazard of privacy violation associated with digital data exchange was accepted, provided that the use of digital data supported a higher purpose, like improving patient’s health, education or research. The potential benefit and timing of repeat local treatment of recurrent CRLM in patients with a short disease-free interval is often debated during MDT meetings. Chapter 6 shows the results of a literature review and meta-analysis of individual patient data on overall survival after repeat hepatectomy for recurrent CRLM, stratified for disease-free interval. Disease-free interval between hepatic resections was considered a prognostic factor for overall survival after repeat hepatectomy, but should not be used as selection criterion for repeat hepatectomy in patients suffering from recurrent CRLM. Chapter 7 shows the findings of an observational cohort study to determine the potential benefit of repeat local treatment, as compared to systemic therapy, for patients with early recurrence of CRLM. Patients who received repeat local treatment, consisting of upfront or repeat local treatment after neoadjuvant systemic therapy, showed improved survival compared to patients subjected to systemic therapy alone. In a sub-analysis of patients with recurrent CRLM within four months, no significant difference in overall survival was observed between treatment strategies. The prognostic value of disease-free interval and the effect of perioperative systemic therapy in patients receiving repeat local treatment for recurrent CRLM are described in Chapter 8. Although perioperative systemic therapy at initial treatment of CRLM may prolong disease-free interval after initial treatment with curative intent, no survival benefit was observed in patients receiving perioperative systemic therapy at repeat local treatment of CRLM, stratified for disease-free interval, previous exposure to chemotherapy, or RAS mutation status. Radiomics is increasingly used to tailor patient selection by predicting response to treatment. Chapter 9 gives an overview on the predictive performance of radiomics in patients with gastrointestinal cancer treated with (neoadjuvant) systemic or radiation therapy. High discriminatory power and accuracy of individual radiomic features and models were reported. However, methodology varied considerably among studies. Direct effects of validated prediction models need to be further clarified.