Investigation Yields Potential New Genetic Biomarker for Anti–PD-1 Success in Melanoma

Scientists appear to have identified a series of cancer-associated fibroblast (CAF) genes that with the potential to predict whether patients with melanoma will respond to anti–programmed cell death protein 1 (anti–PD-1) therapy.

The authors say the findings could one day help clinicians better manage existing patients and give drug developers potential new targets to improve the efficacy of anti–PD-1 therapy. Their findings were published in Frontiers in Medicine.

They wrote that even though immunotherapy has led to tremendous benefits for many patients, a significant proportion of those patients—30% to 60%—will not respond. That is a major problem, given the high cost and serious adverse events that can be associated with the therapy, they said.

“In terms of the clinical response, costs, and side effects, it is urgent to find biomarkers to predict the efficacy of anti–PD-1 therapy,” they wrote.

Existing methods of predicting response to anti–PD-1 therapy include PD-L1 immunohistochemical assays, although they noted that PD-L1 expression can vary over time, making the biomarker less than certain. Other biomarkers, such as tumor mutational burden, microsatellite instability, and molecular subtypes, can be effective, but they also come with a high cost and testing can be inconvenient for patients.

In search of a better biomarker, the authors turned to CAFs, which constitute most stromal cells in the tumor microenvironment. These cells appear to have an important impact on immunotherapy response by promoting tumor immunosuppression and immune escape. However, they have not yet been investigated to the point of identifying predictors of response to anti–PD-1 therapy.

In order to identify the potential role of CAF-associated genes, the investigators used the ç to pull 3 gene expression datasets from patients with melanoma who had not been given anti–PD-1 treatment. From there, they identified CAF-related module genes using weighted co-expression network analysis. They also conducted differential gene analysis on the sets and utilized several computational biology methods to identify a 6-gene panel of CAF-related genes that appeared to be related to anti–PD-1 success: CDK14, SYNPO2, TCF4, GJA1, CPXM1, and TFPI.

The authors next used validation sets to assess their panel using receiver operating characteristic (ROC) curves.

“The multigene panel demonstrated excellent combined diagnostic performance with the area under the curve of ROC reaching 90.5 and 75.4% [to approximately] 100% in the discovery and validation sets, respectively,” they explained.

The authors cited several limitations to their study. Additional validation will be needed, and their sample size was small, with only 87 samples. They said it will be important to validate their findings in a clinical patient cohort before the genes can be relied upon to make clinical decisions.

Still, they said their study resulted in a key step forward by identifying genes of interest that may one day become the basis of an important prognostic tool.

“The multigene panel may become a potential biomarker panel to guide immunotherapy in the future,” they concluded.

Reference

Tian L, Long F, Hao Y, et al. A cancer associated fibroblasts-related six-gene panel for anti-PD-1 therapy in melanoma driven by weighted correlation network analysis and supervised machine learning. Front Med (Lausanne). Published online April 11, 2022. doi:10.3389/fmed.2022.880326