Genomic loss of heterozygosity and survival in the REAL3 trial

Background: Homologous recombination deficiency (HRD) measured using a genomic signature for loss of heterozygosity (LOH) predicts benefit from rucaparib in ovarian cancer. We hypothesized that some oesophagogastric cancers will have high-LOH which would be prognostic in patients treated with platinum chemotherapy.

Methods: Diagnostic biopsy DNA from patients treated in the REAL3 trial was sequenced using the Foundation Medicine T5 next-generation sequencing (NGS) assay. An algorithm quantified the percentage of interrogable genome with LOH. Multidimensional optimization was performed to identify a cut-off dichotomizing the population into LOH-high and low groups associated with differential survival outcomes.

Results: Of 158 available samples, 117 were successfully sequenced; LOH was derived for 74 of these. A cut-off of 21% genomic LOH defined an LOH-high subgroup (n=10, 14% of population) who had median overall survival (OS) of 18.3 months (m) versus 11m for the LOH-low group (HR 0.55 95% CI 0.19-0.97, p= 0.10). Progression free survival (PFS) for LOH-high and LOH-low groups was 10.7m and 7.3m (HR 0.61 (95% CI 0.21 - 1.09, p=0.09). Sensitivity analysis censoring operated patients (n=4), demonstrated OS of 18.3m vs. 10.2m (HR 0.43, 95% CI (0.20-0.92), p=0.02; PFS was 10.5m vs. 7.2m (HR 0.55, (95% CI 0.26-1.17), p=0.09 for LOH-high and LOH-low.

Conclusion: HRD assessment using an algorithmically derived LOH signature on a standard NGS panel identifies oesophagogastric cancer patients with high LOH who have prolonged survival when treated with platinum chemotherapy. Validation work will determine the signature's predictive value in patients treated with a PARP inhibitor and with platinum chemotherapy.