By Our Powers Combined…

Inhibitors that target PD-1 and PD-L1 immune checkpoint proteins are an emerging force in oncology thanks to their ability to prevent cancer cells from evading T cell attack. It’s not surprising, then, that a number of PD-(L)1 inhibitors have been approved for cancer treatment since 2014. But though their positive effects can turn the tide for some cancer patients, one thing prevents their wider use – treatment resistance. Could a multimodal approach – one that blends pathology, radiology, and genomics – be the solution to this problem?

Researchers sought to answer this question through a study involving data from 247 advanced non-small cell lung cancer patients with known outcomes between 2014 and 2019 (1). The data were collected from clinical visits and standard-of-care tests including biomedical imaging, histopathology, genomic assays, and, most importantly, PD-L1 expression patterns captured in diagnostic tumor biopsies. Of the patients investigated, 25 percent responded to PD-(L)1 inhibitor treatment – consistent with real-world proportions.

Next, the researchers fed the data to machine learning and analysis tools to compute risk scores for individual patients. The outcome? The predictive ability of the multimodal approach was significantly greater than that of current unimodal standards. The team hopes that these results will encourage others to explore the multimodal method and eventually even apply it to cancers beyond NSCLC. In light of this, the study’s data materials, workflows, and software have been made publicly available to hopefully springboard future research.

_{^{Credit: Images sourced from Unsplash.com}}