Abstract
Patient-derived tumoroids have emerged as promising models for evaluating patient-specific responses to anticancer therapies, yet their clinical adoption remains limited. Although the reasons for this limited implementation are not fully elucidated, several are known and can be addressed: (i) lack of standardized protocols for tumoroid cultivation and drug exposure complicating cross-laboratory comparisons, (ii) labor- and time-intensive cultivation procedures conflicting with clinical guidelines for timely therapy initiation, (iii) prevalent use of destructive endpoint assays restricting subsequent analyses and proper growth rate correction, and (iv) poorly defined criteria for classifying tumoroid drug sensitivity that are not linked to clinical outcomes, leading to suboptimal treatment allocation in prospective studies. In this study, we developed two classifiers for assessing colorectal tumoroid sensitivity to oxaliplatin and SN-38 based on historic response rates for patients with colorectal cancer. Utilizing longitudinal, label-free confocal imaging, these classifiers offer a nondestructive method that corrects for growth rate variations and preserves patient-derived material for further analysis. Currently, these classifiers are undergoing evaluation in a prospective clinical trial to determine the feasibility of incorporating tumoroid-based drug screening into clinical decision-making. This approach lays the groundwork for next-generation molecular tumor boards, enabling anticancer treatment decisions informed by integrated functional assays and biomarkers. SIGNIFICANCE: Patient-derived colorectal tumoroids can reveal which drugs are likely to induce a tumor response, but current protocols are slow and inconsistent. We developed rapid, nondestructive imaging-based classifiers for oxaliplatin and SN-38 that account for growth rate differences across patients, enabling reliable selection of oxaliplatin- versus irinotecan-based chemotherapy regimens in colorectal cancer.