192P - Use of biosimulation to predict homologous recombination deficiency and PARPi benefit in patients with ovarian, pancreatic, prostate and triple negative breast cancers

Background

The use of PARP-inhibitors (PARPi) has become standard treatment for cancers with homologous recombination deficiency (HRD). Although current tests for HRD identify a subset of HRD patients, they do not identify all patients who will benefit and often rely primarily on BRCA status. Biosimulation incorporates multiple levels of gene and protein regulation, capturing the molecular fates of key HR components. A mechanistically-based biosimulation model that integrates a patient’s tumor-based genomic profile was used to identify HR signaling pathway dysregulation and differential response to PARPi.

Methods

Computational biosimulation (Cellworks) was performed on 4 real-world retrospective cohorts from TCGA (ovarian, pancreatic, prostate, TNBC). Model output, representing key HR pathways, was used to develop a classifier distinguishing patients with HRD, by comparing ovarian cancer BRCA wild-type (WT) patients(n=32) to BRCA-mutated patients (n=187). The locked classifier was prospectively validated in independent sets of ovarian, pancreatic, and prostate cancer patients (n= 336, 428, 189 respectively). Efficacy scores (ES), based on biosimulated composite cell growth in response to olaparib were evaluated in relationship to predicted HRD status in patients with WT BRCA.

Results

The HRD classifier was significantly associated with the BRCA status in all 4 validation sets (logistic regression, p < 0.001) and showed predictiveness for BRCA status in ovarian (AUC = 0.863, p < 0.001), pancreatic (AUC = 0.759, p=0.002) prostate cancer (AUC = 0.717, p < 0.001) and TNBC (AUC = 0.88, p < 0.001) patients. In all four cancer types, predicted PARPi efficacy was significantly higher in in wild-type BRCA patients predicted to be HRD (ovarian p = 0.026, prostate p < 0.001, pancreatic p < 0.001, TNBC p < 0.001).

Conclusions

In this study, an HRD classifier produced through biosimulation was predictive of PARPi benefit in real-world cohorts of ovarian, prostate, pancreatic and TNBC patients with WT BRCA. Future studies will be performed to confirm the hypothesis that biosimulation has utility in identifying WT BRCA patients who may benefit from PARP therapy.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Cellworks Group, Inc.

Funding

Cellworks Group, Inc.

Disclosure

D. Palmer: Financial Interests, Personal, Research Funding: BMS, Nucana, Medannex, Bayer, Sirtex; Financial Interests, Personal, Funding: AstraZeneca, Sirtex, Boston, Guerbet, Servier, Boehringer Ingelheim, MSD. S. Kapoor, S. Khandelwal, Y. Ullal, Y. Narvekar, A. Ghosh, A. Dey, A. Kumar, R. Ps, A. Tyagi, A. Agrawal, M.P. Castro, J. Wingrove: Financial Interests, Personal, Full or part-time Employment: Cellworks. All other authors have declared no conflicts of interest.