Abstract 1790P
Background
In patients with metastatic malignancies there is an urgent need for predictive biomarkers. Fluids containing tumor cells, like pleural effusion or ascites, are easily accessible and could potentially provide information on drug sensitivities ex vivo..
Methods
Image-based single-cell drug response testing (pharmacoscopy) on fluid samples containing malignant cells is used to assess drug response variability on an intra- and interpersonal level. A population of malignant and healthy cells is incubated with a drug panel for 24 hours. After staining with fluorescent antibodies, cells are imaged using automated microscopy. A convolutional neural network infers cell types (malignant vs normal) directly from single-cell images. Using bulk RNA sequencing and targeted next generation sequencing, transcriptomic and genomic data is correlated with the ex vivo drug responses. Eventually, data will be matched with clinical response.
Results
So far, the cohort includes 154 samples from 110 patients. 70% of patients had a sample with sufficient viability and cancer cell content permitting pharmacoscopy. In 19 patients, pharmacoscopy was repeated on specimens taken at different time points, for which we observed a high intra-individual reproducibility of the drug response profiles. Taken together, these results highlight the technical feasibility and robustness of pharmacoscopy on fluid biopsies. At this early stage of this study, we want to highlight one case of a patient with BRAF p.V600E mutated lung adenocarcinoma. Combined tyrosine kinase inhibition with dabrafenib (BRAF inhibitor) and trametinib (MEK inhibitor) was started and resulted in a partial response. In parallel, pharmacoscopy showed an ex vivo response to dabrafenib as the most effective drug.
At the meeting, updated results incorporating transcriptomic and genomic data in correlation with drug responses will be presented.
Conclusions
Pharmacoscopy on fluid samples is a feasible diagnostic tool to assess drug responses. Integration of drug response profiles and morphological profiles with molecular measurements including genomic and transcriptomic profiling will provide further insights into the molecular mechanisms underlying drug response variability.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
University Hospital Zurich.
Funding
European Research Council (ERC).
Disclosure
C. Britschgi: Financial Interests, Personal, Other: AstraZeneca; Financial Interests, Personal, Other: Pfizer; Financial Interests, Personal, Other: Roche; Financial Interests, Personal, Other: Takeda; Financial Interests, Personal, Other: Janssen-Cilag; Financial Interests, Personal, Other: Boeringer Ingelheim. All other authors have declared no conflicts of interest.