Mutational profiling is increasingly performed to identify oncogenic alterations that can be matched to particular molecularly targeted drugs. This 'precision medicine' paradigm has resulted in improved outcomes for patients with specific molecular cancer subtypes, but the efficacy of many targeted therapies has been underwhelming. New data from a genomic study of cancer evolution emphasize the importance of considering 'oncogenic dependencies' in precision-medicine strategies.

Credit: Macmillan Publishers Limited

Previous genome-profiling studies have revealed that certain oncogenic alterations tend to occur together, whereas others rarely co-occur and, thus, seem to be mutually exclusive. These findings suggest that mutations are conditionally selected during cancer evolution, depending on whether certain other alterations are present or absent, based on their overall, rather than individual, effect on cell fitness. “To identify such dependencies, we designed an algorithmic approach that accounts for low-frequency alterations, multiple confounding factors introduced by mixing multiple tumour types and subtypes, and spurious correlations owing to the large number of alterations considered,” explains lead author Giovanni Ciriello.

Using this algorithm, the investigators studied the co-occurrence or mutual exclusivity of 505 candidate 'selected functional events' (SFEs), comprising copy-number alterations and somatic mutations, in the genomes of 6,456 human tumours; 224 pairs of SFEs were found to co-occur, whereas 407 pairs were found to be mutually exclusive. These 631 pairwise relationships involved only 365 SFEs, indicating that many alterations have co-dependencies with multiple aberrations. Interestingly, most relationships were shared by more than one of the 23 tumour types represented in the dataset. KRAS alterations had the most co-dependencies, including mutual exclusivity with alterations involving receptor-tyrosine kinases (EGFR and FGFR1) and other MAPK regulators (NRAS, BRAF, and NF1), as well as co-occurrence with RBM10 and STK11 mutations. Co-dependencies involving TP53 mutations were most frequent and ubiquitous. RNF43, which encodes an E3 ubiquitin ligase, was identified as a less-well-characterized gene with many co-dependencies, mostly converging on activition of the WNT pathway. Co-occurring mutations highlighted the existence of crosstalk between functional pathways, particularly involving MAPK and WNT. Importantly, in vitro studies established the relevance of co-occurring alterations to either drug sensitivity or resistance.

“We have provided a conceptual and methodological framework to study each tumour as a system of co-operating events, generated a pan-cancer map of oncogenic dependencies, and studied their implications in terms of activation of cellular pathways and response to therapy,” summarizes Ciriello. “Our work could serve as a reference for functional and therapeutic studies of co-dependent oncogenic events.”

identifying these dependencies is critical to anticipating drug resistance and designing context-aware therapeutic strategies

Importantly, despite mounting evidence that specific oncogenic alterations have different effects when they occur together or individually, functional studies and clinical trials of molecularly targeted agents are often focused on single genomic alterations, independent of the different cellular and genomic contexts in which they occur. Indeed, this practice is becoming more common, as exemplified by the current proclivity of the oncology community for basket trials. For many cancers, however, therapeutic strategies in which selected targetable alterations are considered in isolation will probably be futile. Ciriello clarifies: “if the effect of one alteration depends on the presence or absence of others, so will its sensitivity to specific treatments; therefore, identifying these dependencies is critical to anticipating drug resistance and designing context-aware therapeutic strategies.” He concludes, “towards this goal, we are currently working with data collected from tens of thousands of patients to determine the most promising drug combination for each of them.” Thus, considering networks of alterations as interdependent factors that drive cancer progression will improve precision medicine.