Search

Nach R0-Resektion eines Kolonkarzinoms im Stadium II und III gemas UICC (Union for International Cancer Control) soll die Indikation zur adjuvanten Therapie gepruft werden. Die Empfehlungen richten sich bislang nach klinischen und histopathologischen Kriterien. Im Stadium II soll die hochgradige Mikrosatelliteninstabilitat (MSI-H) bei der Therapieentscheidung einbezogen werden. Auserdem soll das Risiko fur eine erhohte Toxizitat, wie das Vorliegen von Begleiterkrankungen, eine eingeschrankte Dihydropyrimidindehydrogenase(DPD)-Funktion oder das Alter des Patienten beachtet werden. Immunonkologische Ansatze fur MSI-H-Tumoren oder molekular zielgerichtete Ansatze bei BRAF-mutierten Tumoren werden in lokalisierten Stadien perioperativ in Studien gepruft. Weitere molekulare Parameter, u. a. die zirkulierende Tumor-DNA (ctDNA), werden auf ihre prognostische und pradiktive Wertigkeit hin getestet. Die Gruppe der alteren Patienten wird perspektivisch das groste Kollektiv darstellen und sollte zunehmend beachtet werden.

In recent years, deep learning has been the key driver of breakthrough developments in computational pathology and other image based approaches that support medical diagnosis and treatment. The underlying neural networks as inherent black boxes lack transparency, and are often accompanied by approaches to explain their output. However, formally defining explainability has been a notorious unsolved riddle. Here, we introduce a hypothesis-based framework for falsifiable explanations of machine learning models. A falsifiable explanation is a hypothesis that connects an intermediate space induced by the model with the sample from which the data originate. We instantiate this framework in a computational pathology setting using label-free infrared microscopy. The intermediate space is an activation map, which is trained with an inductive bias to localize tumor. An explanation is constituted by hypothesizing that activation corresponds to tumor and associated structures, which we validate by histological staining as an independent secondary experiment.

Background : Guidance regarding adjuvant treatment decisions in stage II colorectal cancer (CRC) remains uncertain due to lack of predictive clinical or molecular markers. Recently, postoperative circulating tumour (ct)DNA has been demonstrated to be a strong prognostic marker in early colon cancer. Patients and Methods : CIRCULATE enrols patients with stage II microsatellite stable CRC in Germany (AIO) and Austria (ABCSG). Within the AIO, screening is supported by ColoPredict Plus 2.0, a molecular registry and screening platform for interventional trials. Patient-specific mutations are centrally analysed by next generation sequencing in the resected primary tumour. A postoperative plasma sample subsequently screened for the specific mutation(s). ctDNA positive (ctDNApos) patients are randomised (1:2) to control (follow up) or chemotherapy (capecitabine, oxaliplatin added a investigator's choice). ctDNA negative (ctDNAneg) patients are randomised (1:4) to be followed-up within CIRCULATE (control group) or outside the trial. Patients in the control group remain blinded to the ctDNA results. The primary objective is to compare disease free survival (DFS) of ctDNApos patients with chemotherapy or control. To demonstrate a treatment effect with a hazard ratio of 0.617 (3 year DFS rates 42.5 % vs. 25 %), 231 ctDNApos and estimated 2079 ctDNAneg patients are randomised. Secondary aims include to compare overall survival and DFS in the ctDNApos and ctDNAneg patient cohorts and ctDNA kinetics. Conclusion : The CIRCULATE trial may establish ctDNA for adjuvant treatment decision in stage II colon cancer – and with the secondary objectives – support a ctDNA guided follow up in colon cancer stage II and beyond. Micro Abstract Circulating tumour DNA (ctDNA) is a prognostic biomarker of recurrence in patients with early stage and resected colorectal cancer. The CIRCULATE study evaluates the predictive value of ctDNA with respect to adjuvant treatment in colon cancer stage II. In total, 231 patients with detectable ctDNA (ctDNApos) post resection are randomised 2:1 to chemotherapy or to control (follow-up). Patients without detectable ctDNA (approx. 2079 patients) are randomized 1:4 into control or off-study. The primary endpoint is the disease free survival.