Real-world data and biomarkers: moving from one-size-fits-all to personalized treatment of non-metastatic colorectal cancer

Abstract

Colorectal cancer is worldwide the third most prevalent cancer and the second leading cause of cancer-related mortality. Non-metastatic disease is treated by surgical resection, with additional treatment recommendations based on tumor extent (T stage), lymph node involvement (N stage) and mismatch repair status (MMR). Patients with high-risk stage II (pT4, proficient MMR) and stage III (N+) colon cancer (CC) are recommended adjuvant chemotherapy (ACT). This one-size-fits-all ACT is overtreating the ~50% of patients who are cured by surgery, while ~30% of patients experience recurrence of disease and are undertreated. We need to improve upfront identification of over- and undertreated patients, to inform respective de-escalation and escalation strategies. To reduce overtreatment, the recommended ACT duration for CC changed from six to three months based on international trials. Because efficacy in trials with selective inclusion criteria does not always translate to effectiveness in the general population, we analyzed the impact on all patients in the Netherlands. We confirmed that shorter ACT duration was associated with superior patient-reported outcomes and no inferior survival, also not in the subgroup with high-risk stage III (T4 and/or N2) CC. Large real-world datasets provide a good opportunity to analyze subgroups, as we also show for rectal cancer patients of the rare subtype with deficient MMR (2.3%). Compared to proficient MMR, deficient MMR shows better overall survival and event-free survival. These findings help provide context to upcoming single-arm trials on immunotherapy for deficient MMR. In stage III CC patients treated with standard ACT, the strongest prognostic biomarker to date is post-surgery circulating tumor DNA (ctDNA). ctDNA detection after surgery is an indicator of minimal residual disease and is associated with a high recurrence risk (64% after 3 years). Post-surgery ctDNA-negative patients have a lower recurrence risk of 17%. Current ctDNA assays still fail to detect part of recurrences, especially to the peritoneum and lungs. To detect these false-negative recurrences, tumor stage (T4/N2) and the tumor-stroma ratio (>50% stroma) are of added value. A third of patients are ctDNA-negative and pT1-3N1 and stroma-low, they are at very low RR (3%) after ACT and may be eligible for de-escalation of treatment. Next, we molecularly characterized the patients at high risk of recurrence despite ACT, with a subgroup analysis on the patients with minimal residual disease post-surgery to isolate (lack of) response to ACT. Recurrence and chemoresistance were associated with RNA-sequencing signatures indicative of cancer-associated fibroblasts (CAFs) and DNA-mutation in SMAD4 (mediator of TGF-β signaling between CAFs and tumor cells). Lastly, we showed that organoids derived from colon tumors of patients with recurrence after ACT were more resistant to chemotherapy screens in vitro. Organoids thus provide a model to capture and study tumor-intrinsic chemoresistance, which was associated with TGF-β signaling and epithelial-mesenchymal transition. In addition, organoids may be used to study alternative treatment avenues, potentially targeted at the identified high-risk mesenchymal features. Together, real-world data and biomarkers have potential to identify which patients do not benefit from current one-size-fits-all treatment, in order to move to personalized treatment of non-metastatic colorectal cancer.