medwireNews: Two novel signatures, one based on radiomics and the other on epigenetics, could help predict response to treatment with agents directed against the programmed cell death protein 1 (PD-1) immune checkpoint pathway.
Eric Deutsch (Gustave Roussy Cancer Campus, Villejuif, France) and colleagues who developed the radiomic signature note that although “[i]mmunotherapy has substantially changed the therapeutic strategies for cancers such as melanomas, lymphomas, and lung tumours,” not all patients have a response.
“There is therefore a need for the development of methods to identify patients who are most likely to respond to immunotherapy,” they write in The Lancet Oncology.
The investigators applied radiomic analysis – which involves translating medical images into quantitative data to generate noninvasive, image-driven biomarkers – to computed tomography images of biopsy specimens and RNA sequencing data from 135 patients with incurable or metastatic solid tumors who were involved in the MOSCATO trial.
From this they devised an eight-feature signature that enabled the estimation of tumor-infiltrating CD8 cell levels; five of the variables were radiomic features, two were tumor volumes at two locations of interest, while the final variable was the peak kilovoltage.
In the MOSCATO dataset, the radiomics-based signature could differentiate between specimens with high versus low levels of CD8 cells on 74% of occasions, and in 67% of instances in a validation set of samples from 119 patients included in The Cancer Genome Atlas.
The signature also predicted clinical outcomes in a cohort of 137 patients who had received treatment with agents targeting PD-1 or its ligand (PD-L1) in phase I studies. Specifically, patients who attained an objective response at 6 months had a significantly higher baseline radiomics score of CD8 cells than those who progressed or had stable disease. The baseline score was likewise significantly higher among patients with disease control than progression.
Furthermore, a high baseline radiomics score (above the median of 1.91 cells) was a significant and independent predictor of improved overall survival (OS) after adjusting for the number of lines of therapy, tumor volume, and the Royal Marsden Hospital prognostic score, with a hazard ratio (HR) for death of 0.52.
Median OS was 24.3 months for individuals with a high radiomic score and 11.5 months for those with a low score.
The authors of a linked commentary observe that “radiomics has the potential to be a digital biopsy technique that is both spatially guided and non-invasive and that can quantify T cell infiltration of tumours, support personalised design of immunotherapy interventions, and longitudinally monitor and assess immune checkpoint blockade response.
This study “provides an important step forward in realising that potential,” they add.
But Issam El Naqa and Randall Ten Haken (both from University of Michigan, Ann Arbor, USA) conclude: “Of course, the results of such promising retrospective radiomics analyses must still be further evaluated in prospective clinical trials.”
The other signature – termed EPIMMUNE – was derived from the epigenomic assessment of tumor samples from a discovery set of 34 patients with stage IV non-small-cell lung cancer (NSCLC) that had been treated with anti-PD-1 therapy. The epigenetic signature consisted of 301 CpG sites, the methylation levels of which were significantly linked with clinical response.
A positive EPIMMUNE signature predicted significantly improved progression-free survival (PFS) in the discovery set and in a validation cohort of 47 participants, also with stage IV NSCLC, with corresponding HRs for progression or death of 0.010 and 0.330.
However, the association of the signature with OS was significant in the discovery (HR=0.080), but not the validation set, report Manel Esteller (Bellvitge Biomedical Research Institute, Barcelona, Spain) and co-researchers.
They also identified a single DNA methylation marker – FOXP1 – that could predict clinical outcomes. In an independent group of 61 NSCLC patients given a PD-1 inhibitor, those with unmethylated FOXP1 had significantly better PFS and OS than individuals in whom the gene was methylated, with HRs of 0.415 and 0.409, respectively, in favor of unmethylated FOXP1 status.
The researchers conclude in The Lancet Respiratory Medicine: “Although further prospective clinical studies are needed to establish the true value of the EPIMMUNE signature, the epigenetic biomarkers identified herein could be helpful for selecting those patients for whom immunotherapy or strategies acting on specific intratumoral cell subpopulations could be assessed in cancer-type-specific studies and basket clinical trials.”
medwireNews is an independent medical news service provided by Springer Healthcare. © 2018 Springer Healthcare part of the Springer Nature group