Abstract 582P
Background
High-grade serous carcinoma (HGSC) is an aggressive subtype of ovarian cancer. It is commonly diagnosed at late-stage with intra-abdominal spread that is characterized by extensive heterogeneity. We studied how clinical responses are affected by multi-omics variations and metastatic heterogeneity.
Methods
We have established a prospective, longitudinal HGSC cohort (N=292) treated at Turku University Hospital in Finland (DECIDER, NCT04846933). Whole-genome and transcriptome sequence data were obtained from >750 fresh tumour samples collected from tubo-ovarian and metastatic sites at diagnosis, interval surgery and relapse. All patients have a thorough, curated clinical information. They were treated with debulking surgery followed by chemotherapy (PDS, N=150), neoadjuvant chemotherapy (NACT) followed by debulking surgery and chemotherapy (N=140), or chemotherapy only (N=2).
Results
DECIDER cohort allows us to characterize the impact of treatment to tumors’ genomes and transcriptomics and intrinsic heterogeneity, as well as their associations to clinical parameters, such as treatment failure. For example, only patients with homologous BRCA1/2 mutations or homologous recombination (HR) signature had longer platinum free intervals (PFI), but patients with heterogenic mutations did not differ from patients without HR signature. Largest hazard ratios for PFI were detected for residual disease after surgery, HR signature and accumulation of ascites. We detected patients with highly variable genomic profiles between tumour sites, and even heterogenic whole-genome duplications (WGD) were detected. Most of the detected genomic variation comes from the differences between tumour sites and neoadjuvant chemotherapy had only limited effect on genomic scale whereas impact on transcriptome was clear.
Conclusions
Known clinical and genomic parameters explain a large part of primary treatment responses in HGSC. However, heterogeneity complicates identification of novel clinically relevant variations and our study approach facilitatesovercoming this issue. Single biopsy samples have unique signals not detected in other samples, and in order to avoid misguided conclusions, multi-sample collection is suggested.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
European Union’s Horizon 2020 research and innovation programme, grant agreement No 965193 for DECIDER.
Disclosure
All authors have declared no conflicts of interest.