Technical development and validation of a clinically applicable microenvironment classifier as a biomarker of tumour hypoxia for soft tissue sarcoma

Background Soft tissue sarcomas (STS) are rare, heterogeneous tumours and biomarkers are needed to inform management. We previously derived a prognostic tumour microenvironment classifier (24-gene hypoxia signature). Here, we developed/validated an assay for clinical application. Methods Technical performance of targeted assays (Taqman low-density array, nanoString) was compared in 28 prospectively collected formalin-fixed, paraffin-embedded (FFPE) biopsies. The nanoString assay was biologically validated by comparing to HIF-1α/CAIX immunohistochemistry (IHC) in clinical samples. The Manchester ( n = 165) and VORTEX Phase III trial ( n = 203) cohorts were used for clinical validation. The primary outcome was overall survival (OS). Results Both assays demonstrated excellent reproducibility. The nanoString assay detected upregulation of the 24-gene signature under hypoxia in vitro, and 16/24 hypoxia genes were upregulated in tumours with high CAIX expression in vivo. Patients with hypoxia-high tumours had worse OS in the Manchester (HR 3.05, 95% CI 1.54–5.19, P = 0.0005) and VORTEX (HR 2.13, 95% CI 1.19–3.77, P = 0.009) cohorts. In the combined cohort, it was independently prognostic for OS (HR 2.24, 95% CI 1.42–3.53, P = 0.00096) and associated with worse local recurrence-free survival (HR 2.17, 95% CI 1.01–4.68, P = 0.04). Conclusions This study comprehensively validates a microenvironment classifier befitting FFPE STS biopsies. Future uses include: (1) selecting high-risk patients for perioperative chemotherapy; and (2) biomarker-driven trials of hypoxia-targeted therapies.