Aims: Triple negative breast cancer (TNBC) and high-grade serous ovarian carcinoma (HGSOC) share many molecular and morphological similarities. The aim of this project was to analyse both tumour entities according to molecular and proteomic heterogeneity focusing on primary and recurrent tumor samples and the clinical impact.
Methods: In order to analyse this question 128 breast cancer samples were analysed by large multi-gene NGS panels detecting different types of genetic alterations such as single nucleotide variants (SNVs), insertion/deletions (InDels), copy number variants (CNVs), and fusion genes. HGSOC samples were analyzed by HTG Edge Seq platform with the Oncology Biomarker Panel that captures the expression of 2500 transcripts with Ion Torrent S5 semiconductor sequencing. Furthermore HGSOC were evaluated according to genomic instability by SNParray (N=80) or CGH (N=60) on paired samples focusing on DNA repair genes. PDX models of ovarian and breast cancer patients were established and RAD51 was evaluated both on patient samples and PDX models to discover mechanisms of PARPi resistance.
Results: The molecular analysis of the breast cancer samples showed molecular subtype changes between primary tumors and relapses in 10 of 128 (7.8%) cases. Most driver genomic alterations (55.8%) were shared between primary tumors and matched recurrences. However, in 39 of 61 cases (63.9%), additional private alterations were detected in the relapse samples only, including 12 patients with potentially actionable aberrations. The HTG analysis of the ovarian tumors resulted in 233 DEGs (adj.p ≤0,05, log2FC>0,8) after evaluation of the RNA. After further analysis of publicly available datasets 41 genes remained significant and out of these eight markers (HMGCS2, FABP4, AHRR, GREM1, ITGA5, COL5A2, WNT9B and SFRP2) were chosen for IHC validation. For three markers, differential expression within the tumor compartment could be validated on the protein level. The expression of AHRR was higher in the primary tumors compared to the recurrences (p<0,001, n=49). ITGA5 (p=0,048, n=51) and GREM1 (Z = -4,124, p<0,001, n=51) were found to be expressed higher in the recurrent tumors.
The genomic instability analysis showed that at baseline a significant proportion of tumors lacked expression in DDR biomarkers such as RAD51, PAR or FANCD2, however marked changes were noted after platinum NACT. Post-NACT re-expression of DDR markers was associated with poor outcome especially for tumors co-expressing RAD51 and PARP or FANCD2.
The search for PARPi resistance via PDX models showed a frame-restoring secondary mutation in BRCA2 in PDX405OR, an olaparib-resistant counterpart from PDX405. None of the BRCA1/2 mutant models harbored alterations in 53BP1. All PARPi-resistant models exhibited high RAD51 score, and most of them produced BRCA1 hypomorphs. Overall, 14 out of 18 BRCA1-mutant PDX models (78%) harbored BRCA1 and RAD51 foci, and were resistant to PARPi. Out of the 4 BRCA1 foci-deficient PDXs, three were PARPi resistant and one was PARPi sensitive.
Conclusion: A rather stable molecular situation between primary and recurrent breast cancer might be stated, but private actionable alterations in recurrence exist. The results of the ovarian cancers point to a relevance of proteins engaged in the processes of extracellular matrix organization (GREM1; ITGA5) and metabolic pathways (AHRR). The combination of only few but specific newly discovered signaling and survival pathways for recurrence in high-grade serous ovarian carcinoma and breast cancer make these purposeful targets for therapy and further research about these pathways very fascinating. Furthermore homolougos recombination proved to be essentially involved in ovarian tumor relapse and resistance mechanisms to PARPi therapy could be detected with PDX models and RAD51 analysis. In passing 12 papers have been published from this consortium in the TH4Respons project and an ERA PerMed grant has been obtained for the continuation of the RAD51 clinical validations (ERAPERMED2019-215).