Supplementary Table from Preclinical <i>In Vivo</i> Validation of the RAD51 Test for Identification of Homologous Recombination-Deficient Tumors and Patient Stratification
1073 Background: Next generation sequencing (NGS) of tBx is the basis for precision medicine. Most tBx used for NGS are archival primary tumors, often acquired several years before starting matched treatment. Analysis of ctDNA may better capture the GA landscape of MBC. We aimed to compare the concordance of GA detection by NGS in synchronously acquired tBx and ctDNA in MBC pts. Methods: MiSeq Amplicon-based NGS (panel of 59 cancer-related genes) was performed in both tBx and ctDNA at disease progression. The concordance of GA in tBx vs ctDNA was determined at patient level and at mutation (mut) level in clinically actionable genes (PIK3CA, AKT1, ERBB2, ESR1, PTEN). False negative result in ctDNA (FN-ctDNA) defined as mut detected in tBx but not in ctDNA. Results: 28 pts identified (luminal [lum] 21, HER2+ 5, triple negative 2), median prior lines of therapy 4.5 (0-15). Most pts had visceral metastasis (71%); most biopsies were from non-visceral sites (67%), mainly breast/nodes/skin (59%). In 16 pts (57%), tBx and ctDNA had complete concordant results (4 were wild-type). Concordance was 100% in non-lum and 43% in lum pts (P=0.01). Clinically actionable GA were found in 20/28 (71%) pts. FN-ctDNA rate was 25%; these pts had a trend towards having non-visceral metastasis (OR 3.1, P=0.32). Focusing in clinically actionable genes, concordance was 52% at mut level, being lower for ESR1 mut (Table). Interestingly, ctDNA analysis identified 6 ESR1 mut that were not detected in tissue, including 3 pts with a double ESR1 mut. Conclusions: NGS in ctDNA is feasible and the results may be informative for pts management. Our results suggest that ctDNA may be useful for assessing GA in non-lum MBC and the emergence of ESR1 mut in lum MBC. Double ESR1 mut in ctDNA suggests a mechanism of convergent evolution in acquired resistance to endocrine treatment. Causes of discordant tBx/ctDNA (e.g. tumor burden / heterogeneity, technical issues related to ctDNA isolation / processing), warrant further study in larger datasets. N mut detected Concordant mut tBx ctDNA Overall N (%) ERBB2 1 1 1 1 (100) PIK3CA 9 7 9 7 (78) ESR1 10 13 16 7 (44) AKT1 2 0 2 0 PTEN 0 1 1 0 Overall 22 22 29 15 (52)
<div>Abstract<p>PARP inhibitors (PARPi) are approved drugs for platinum-sensitive, high-grade serous ovarian cancer (HGSOC) and for breast, prostate, and pancreatic cancers (PaC) harboring genetic alterations impairing homologous recombination repair (HRR). Detection of nuclear RAD51 foci in tumor cells is a marker of HRR functionality, and we previously established a test to detect RAD51 nuclear foci. Here, we aimed to validate the RAD51 score cut off and compare the performance of this test to other HRR deficiency (HRD) detection methods. Laboratory models from <i>BRCA1</i>/<i>BRCA2</i>-associated breast cancer, HGSOC, and PaC were developed and evaluated for their response to PARPi and cisplatin. HRD in these models and patient samples was evaluated by DNA sequencing of HRR genes, genomic HRD tests, and RAD51 foci detection. We established patient-derived xenograft models from breast cancer (<i>n</i> = 103), HGSOC (<i>n</i> = 4), and PaC (<i>n</i> = 2) that recapitulated patient HRD status and treatment response. The RAD51 test showed higher accuracy than HRR gene mutations and genomic HRD analysis for predicting PARPi response (95%, 67%, and 71%, respectively). RAD51 detection captured dynamic changes in HRR status upon acquisition of PARPi resistance. The accuracy of the RAD51 test was similar to HRR gene mutations for predicting platinum response. The predefined RAD51 score cut off was validated, and the high predictive value of the RAD51 test in preclinical models was confirmed. These results collectively support pursuing clinical assessment of the RAD51 test in patient samples from randomized trials testing PARPi or platinum-based therapies.</p>Significance:<p>This work demonstrates the high accuracy of a histopathology-based test based on the detection of RAD51 nuclear foci in predicting response to PARPi and cisplatin.</p></div>
Supplementary Figure from Preclinical <i>In Vivo</i> Validation of the RAD51 Test for Identification of Homologous Recombination-Deficient Tumors and Patient Stratification