14P - expHRD: An algorithm for the transcriptome-based estimation of homologous recombination deficiency score

Background

Homologous recombination deficiency (HRD) is a clinically-proven indicator of sensitivity against platinum-based chemotherapy and PARP inhibitors in various malignancies, including ovarian cancers. Until now, HRD prediction has been carried out by detecting the deleterious mutations of the BRCA1 or BRCA2 gene or measuring genomic scar, i.e., scarHRD. However, the scarHRD method has limitations in applying tumors without matched germline data. Investigating HRD-specific transcriptomic signatures is one of the tangible approaches to predicting HRD in tumor-only samples. Although several RNA-seq-based HRD prediction algorithms have been developed, they support mainly cohort-wise classification regarding HRD status yet do not provide an equivalent value to scarHRD. The transformation into scarHRD from transcriptomic data would be highly valuable for clinical application and research use.

Methods

Here, we developed an algorithm of transcriptome-based HRD analysis, i.e., expHRD. The prediction model was established using the elastic net in TCGA-pan cancer training set. HRD geneset for applying expHRD calculation using single-sample geneset enrichment analysis was derived by bootstrap technique.

Results

expHRD is a clinically applicable RNA-seq-based HRD prediction platform in tumor-only samples. In clinic, expHRD can provide supportive information for predicting platinum or PARP inhibitor responses by providing predicted scarHRD score. We could extract the expHRD to analyze the inter-cohort evaluation, HRD-associated gene expression profile, mutations, or tumor types in samples without matched germline data, such as cryopreserved or formalin-fixed tissues, cell lines, or organoids.

Conclusions

Developing a targeted transcriptomic panel for calculating expHRD would be a cost-effective approach to linking our platform with the clinic.

Editorial acknowledgement

Clinical trial identification

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.