A biomarker combination indicating resistance to FOLFOX plus bevacizumab in metastatic colorectal cancer: Results of phase I of the PERMAD trial

Abstract

Background: Antiangiogenic agents such as bevacizumab (bev) are widely used in combination with chemotherapy to treat metastatic colorectal cancer (mCRC). Predictive markers indicating resistance to antiangiogenic agents are elusive. Cytokines and angiogenic factors (CAF) may enable such a prediction. The PERMAD trial has two phases: Phase I, reported here, aims to establish a CAF marker combination (CAFmC) that enables early prediction of treatment resistance in treatment naive patients with mCRC receiving bev plus mFOLFOX6. Phase II will prospectively evaluate this CAFmC and randomize between an early replacement of bev by aflibercept or continuation of bev when the CAFmC indicates imminent progress. Method(s): In phase I 41 out of 50 patients recruited in 15 centers in Germany and Austria with treatment naive mCRC under FOLFOX plus bev treatment were evaluable for CAF analysis. 102 different, preselected CAFs were prospectively collected and centrally analyzed in plasma samples (n=647) obtained prior to treatment and biweekly until radiologic progress determined by CT scan every 2 months. The values of various CAFs were affected by both, chemotherapeutic treatment itself as well as progress. These CAF were excluded from the bioinformatic analysis. Using the remaining CAF we employed a machine learning approach to define a combination of 5 CAF whose change in values/pattern correlated with later progress at least 2 months prior to radiologic progress as determined by CT. Out of various classifiers examined, a random forest classifier provided a CAF set with the highest accuracy. Result(s): Using the samples described above and a random forest algorithm we established a CAFmC comprising 5 CAF whose specific change in value/pattern over time indicated treatment resistance 3months prior to radiologic progress with an accuracy of 83%. The CAFmC was established and cross-validated in two cohorts of 26 and 15 patients, respectively. Conclusion(s): Using advanced bioinformatics we identified a CAFmC that points out treatment resistance to FOLFOX plus Bev in patients with mCRC 3 months prior to radiological progress. A decision process using this marker combination will be evaluated in the randomized phase II of the trial.