S144: btk and bcl-2 activity at baseline predicts mrd status for chronic lymphocytic leukemia patients treated with ibrutinib + venetoclax in the hovon 141/vision trial

Background: Although targeted therapies have made a breakthrough in CLL treatment, many patients develop resistance, have severe side effects, or relapse during treatment. Due to the wide heterogeneity in both clinicobiological features and clinical outcomes, CLL patients require personalized management strategies. HOVON 141/VISION is a randomized phase II clinical trial that evaluates the efficacy of minimal residual disease (MRD) guided combinatorial treatment with the Bcl-2 antagonist venetoclax and the BTK inhibitor ibrutinib in relapsed/refractory (R/R) CLL patients. Aims: To identify functional biomarkers at baseline that predict MRD status at month 15 in the patients enrolled in the HOVON 141/VISION trial. Methods: Peripheral blood mononuclear cells (PBMCs) were collected from CLL patients enrolled in the HOVON 141/VISION trial (n = 186) prior to treatment (at baseline). The patient samples were divided into a training set (n = 139; 75%) and a test set (n = 47; 25%), stratified based on IGHV and TP53 aberration status. After 15 months of ibrutinib + venetoclax treatment, 48, 86 and 5 patients in the training set had obtained an undetectable (<10-4, i.e., less than 1 CLL cell detected in 10000 leukocytes by flow cytometry), intermediate (10-4 to 10-2) and high MRD (>10-2) status, respectively. Multi-color flow cytometry with fluorescent cell barcoding was applied to detect surface markers (n=14), activation markers (n=4), and phospho-proteins (n=31) in both resting and anti-IgM stimulated samples. Statistical associations between functional readouts and MRD levels were assessed with Wilcoxon non-parametric test, and p-values were Bonferroni-corrected for multiple testing. Random Forest (RF) classifier was used to predict patients’ MRD status, and its accuracy was evaluated using 3-fold MRD-stratified cross validation (CV). Results: By applying high-resolution immune profiling we were able to classify patients’ PBMCs. Increased activation of downstream effectors such as AKT (pS473), p44/42 MAPK (pT202/Y204) and p90RSK (pS380) confirmed BCR activation in response to anti-IgM stimulation. High status of peripheral blood MRD (>10-2) after 15 months of venetoclax + ibrutinib treatment was associated with upregulation of 7 proteins at baseline in both resting and stimulated B cells in the training set of samples (adjusted p<0.05). For instance, high expression level of Bim and increased activation of AKT (pS473) was associated with high MRD level (adjusted p<0.05). Using the protein profiles collected from unstimulated cells as input data, we next constructed an RF classifier to predict the MRD status. When classifying between the high and undetectable (<10-4) MRD patients in the training set, we obtained a weighted F1 score of 0.89 (SD 0.06 across the 3 CV folds). When classifying between high and undetectable/intermediate (10-4 to 10-2) MRD patients, the accuracy was somewhat decreased (weighted F1 0.83, SD 0.03). Based on the Gini impurity index, we found that Bcl-2 and BTK phosphorylation levels were among the 3 best predictors of the MRD outcome, in addition to NF-κB (Gini>0.1). Summary/Conclusion: We have developed protocols for combined immunophenotyping and protein profiling, which were successfully applied to baseline samples from patients enrolled in the HOVON 141/VISION trial. Baseline protein profiles were predictive of MRD status after 15 months of venetoclax + ibrutinib treatment and may serve to guide treatment decisions in future studies. Subsequent validation of the prediction model on the independent test set will be presented. Keywords: Targeted therapy, Chronic lymphocytic leukemia, Prediction, Clinical outcome