Patient-Specific Drug Response Prediction in Liver Cancer
21 patients, 7 drugs tested. Our models achieved 78.95% clinical correlation—predicting which patients responded and which did not.
Challenge
A clinical partner needed to validate whether organoid-based drug sensitivity testing could accurately predict patient outcomes in hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC)—two of the most challenging and heterogeneous liver cancers.
Approach
We established PDO models from 21 patients with HCC or ICC. Each model was tested against 7 drugs, including chemotherapy agents and ICB combinations. Drug sensitivity results were then compared against actual patient clinical outcomes in a blinded retrospective analysis.
Results:
Overall clinical correlation: 78.95%. Sensitivity: 80%, Specificity: 77.78%. The models accurately predicted both responders and non-responders across all drug classes tested.
Implication
This study validates the use of our ALI-PDO platform as a co-clinical tool for personalized oncology, enabling patient stratification before treatment initiation.