Session Item

The biomarker development pipeline is fraught with failure. The plethora of published predictors and prognosticators for radiation oncology provide biological insight but lack practical application. The development of tumour genomic signatures addresses some of the limitations associated with other approaches for biomarker development. Accessible and expanding online resources allow for validation in multiple cohorts. There are multiple methods for signature implementation that have proven accuracy and have entered the clinical arena. The use of a signature versus a single marker can reduce intra-tumour heterogeneity and improve performance. The most widely developed signatures in radiation oncology are those that assess tumour hypoxia.

Work in Manchester focused on developing tumour type specific signatures for assessing hypoxia. Our head and neck signature developed with Oxford was validated in four cohorts, taken forward for application using customised Taqman arrays and shown to predict benefit from the addition of hypoxia-targeted treatment to radiotherapy in a fourth cohort (ARCON trial). The signature was assessed prospectively in the UK NIMRAD trial, which closed to recruitment in May 2019. Hypoxia signature scores were generated for most of the 338 patients recruited into the NIMRAD trial.  A bladder signature was validated in four of six cohorts and shown to predict benefit from the addition of hypoxia-targeted treatment to radiotherapy in a seventh (BCON trial). The signature is progressing to final testing in a biomarker driven trial. A sarcoma signature has been validated in four independent cohorts including a randomised trial (VorteX) and is being taken forward using the NanoString platform. A prostate signature was validated in nine cohorts and is also progressing using NanoString.

A number of key elements in developing genomic signatures are highlighted. (1) Multiple cohorts are obtained for validation in silico, via collaboration and/or by obtaining funding to build unique cohorts involving patients treated with radiotherapy and generating transcriptomic data. (2) Signatures are tested to show independence from existing and emerging prognostic factors, e.g. CINSARC and sarculator (sarcoma); Decipher and a 31 risk loci classifier reflecting genomic instability (prostate). (3) Considerable effort is put into identifying platforms for clinical application that involves testing gene probes, choosing suitable endogenous reference genes, and assessing assay performance.

Within the rapidly changing field of biomarkers for oncology the radiation oncology community need to catalyse competitiveness and accelerate progress. A change in culture is needed to improve biomarker development, evaluation, translation and implementation. There is a need to identify “what we do best and finding more ways of doing less of it better”.