Translating cancer genome insights into practical clinical tools

Just ten years ago, the first tentative steps were being made using next-generation sequencing (NGS) profiling to identify gene mutations across cancer types. Today, it is increasingly routine in some parts of the world to profile tumours for biomarker genes; however, harnessing the wealth of information for clinical benefit is still a challenge.

Most cancer patients have multiple mutations, not just a single mutation, and many tumours harbour infrequent mutations. Because of this, interrogation of thousands of patient samples is required to achieve the statistical modelling power necessary to understand the interactions between many common and rare mutation combinations. In collaboration with the International Working Group for Prognosis in Myelodysplastic Syndromes (MDS), more than 3,300 samples were analysed – with subsequent confirmation using 1,120 samples from an independent validation cohort – linking uniform profiling with diagnostic annotations, treatment details and outcome data (Nat Med. 2020;26:1549–1556). Regular discussions with the interdisciplinary team over more than four years have been central to turning complicated data into simple, practical, diagnostic and risk-classification tools for use by practising oncologists. As evidenced by our work on TP53 mutations, the way single biomarkers are used for disease classification and risk stratification was redefined. Using detailed molecular annotation, it was demonstrated that one in three patients with TP53 mutation had a mutation in one allele, while the remaining patients had biallelic mutations. Underpinned by the statistical power of our database, only patients with biallelic TP53 mutations were shown to have a high-risk phenotype and poor outcomes, whereas those with monoallelic mutation had clinical presentation and outcomes similar to patients with wild-type TP53 (Nat Med. 2020;26:1549–1556). Within two years of the publication of the findings, allelic state differentiation was incorporated into the 2022 WHO diagnostic guidelines (Leukemia. 2022;36:1703–1719).

Risk stratification in MDS is generally based on chromosomal surveillance; however, half of patients have no molecular findings. By combining clinical parameters with molecular findings, including chromosomal abnormalities and gene mutations, the collaborative group devised a risk-classification schema capable of providing a patient-specific score. This complex formula was used to develop a simple web-based risk-classification tool, with corresponding smartphone apps. Since the International Prognostic Scoring System for MDS publication in 2022 (NEJM Evid. 2022;1:EVIDoa2200008), the web portal, which has been translated into a number of different languages, has been used for over 260,000 patients in more than 65 countries around the world with approximately 200 users daily.

This is a unique timepoint in genome profiling research – the tools available are becoming increasingly sophisticated, enabling both bulk and single-cell technologies to delineate specific gene mutations across distinct cell lineages in patient samples, and describe effects at the level of initiating cells and cells that define disease progression or treatment resistance. There is a lot to celebrate, but there is also so much more to do. At the moment, researchers’ efforts focus on leveraging data from a handful of biomarkers that can practically and realistically be generated through panel-based assays. In the future, a key challenge is to be able to combine information from a patient’s germline genome with genome sequencing and multimodal profiling of their tumour to further improve diagnosis and treatment.