P613: analysis of ex-vivo response to single drugs and drug combinations in chronic lymphocytic leukemia patient samples

Background: Despite major advances in the development of novel treatments for Chronic Lymphocytic Leukemia (CLL), difficulties arise due to the high heterogeneity, in particular that of drug resistance in individual patients, requiring the identification of novel personalized treatments. Addressing this challenge requires experimental approaches that can mimic the tumor microenvironment while using CLL cells from patient samples, as well as powerful, reliable statistical analyses of drug efficacies or synergistic effects in drug combinations. Aims: Identify novel efficacious drugs and synergistic drug combinations using data from high throughput ex-vivo drug response assays, and look for patterns in relation to drug types and mechanisms of action. Methods: An experimental approach was used to mimic the tumor microenvironment, keeping the viability of CLL cells from patient samples under control during luminescent cell viability assays. A drug sensitivity score (DSS) was used to assess the response (viability) of cells to 516 drugs and 20 x 39 drug combinations over a range of concentrations, together with a statistical methodology that filters out noise while accounting for batch effects across plates, allowing for reliable results. For each patient, single drugs were selected for ex vivo drug-combination assays on the basis of certain criteria on their experimental DSS and IC50 values. The experimental design for combinations used one of the drugs at a predefined anchoring concentration, while varying the concentration of the second drug. The synergy or antagonism of drug combinations were quantified for two different reference models, which assumed either Bliss independence or Loewe additivity. The uncertainties in our results were quantified by taking into account both fitting errors and uncertainties in the control data. Single drugs and drug combinations were ranked for each patient according to their DSS and synergy metric, respectively. Results: We identified 3 clusters of single drugs, in which their DSS scores were either mostly high, mostly low or a mixture of high and low across patients. Our ranking of single drugs by DSS identified new candidate drugs for CLL treatment, to be further investigated in additional ex vivo assays. Using our statistical approach and comparing two different synergy models, for each patient we reliably identified specific drug combinations with significantly high synergy scores, to be investigated in future experiments. Summary/Conclusion: Analysis of our data from ex vivo high throughput drug screening assays reliably identified novel candidate drugs for CLL, as well as drug combinations with significantly high synergistic effects, which we aim to investigate further in future experiments. Keywords: Synergy, Drug sensitivity, Targeted therapy, Drug interaction