Search Continues for New Biomarkers in NSCLC

Recent research is expanding the potential biomarker landscape in non–small cell lung cancer (NSCLC) and may soon enable more precise matching of patients to their optimal therapy.

Study authors outlined the latest research on predictive biomarkers in a new review article in the journal ImmunoTargets and Therapy,¹ noting that artificial intelligence (AI) may be an essential component of the new biomarker paradigm.

The authors explained that novel therapeutic agents have led to significant breakthroughs in the treatment of NSCLC. Patients with actionable genomic alterations can now receive targeted therapies, and those without actionable alterations can be treated with immune checkpoint inhibitors (ICIs) with or without chemotherapy.

Yet, virtually all patients eventually develop resistance to immunotherapy, the authors noted. While some investigators are working to develop new types of therapies, others have been working to find improved predictive biomarkers that could help clinicians better predict responses to ICIs and, thereby, better optimize treatment plans.

The authors first reviewed the most established predictors of immunotherapy response, PD-L1 and tumor mutational burden (TMB). They noted that patients with PD-L1 tumor proportion scores (TPSs) of at least 50% are seen to be among the most likely to benefit from anti–PD-(L)1 monotherapy. Yet, the authors said reliance solely upon the 50% threshold amounts to a rather blunt measurement device and may overlook subgroups of patients with different sensitivities to ICI therapies. Complicating matters, they added that PD-L1 expression can vary over time and location and can also be impacted by inter-assay and interobserver variability.

TMB, on the other hand, is based on the premise that a higher mutational load translates to a higher likelihood of generating neoantigens that could make tumors more susceptible to immune attack. Yet, they said there is currently a lack of universally accepted assays or cutoff values for TMB, and the technique that provides the most comprehensive evaluation—whole-exome sequencing—is not feasible for widespread clinical use, they said.

Given those limitations, a number of other potential biomarkers have been researched, and many have shown promise. Genomic alterations such as KEAP1, STK11, and SMARCA4, as well as markers of metabolic pathway dysregulation, tumor-infiltrating lymphocytes, and blood-based biomarkers like circulating tumor DNA, have all been proposed as new biomarkers.

Yet, the authors said most of these biomarkers have not been shown to be effective as stand-alone biomarkers, and some would be difficult to integrate into daily clinical practice.

The authors said patient characteristics might also provide meaningful clues about therapeutic efficacy, including tobacco exposure, sex, and body mass index (BMI). The balance of the research, they said, suggests that BMI and sex are interconnected factors shaping immunotherapy outcomes. They noted that obesity promotes systemic inflammation and immune dysfunction, and yet also appears to enhance responses to PD-1/PD-L1 blockade.

“The magnitude and direction of this effect appear to differ by sex, likely reflecting hormonal regulation and differences in adipose distribution,” they wrote.

The multifactorial nature of tumor immunogenicity has led some investigators to turn toward AI and machine learning (ML) to better integrate a wide array of data to generate prognostic predictions. The investigators noted that 1 ML-based test, which used pretreatment plasma proteomic profiles to inform personalized treatment decisions, outperformed PD-L1 in identifying patients likely to benefit from treatment escalation through chemotherapy.²

The various potential biomarkers under investigation each capture distinct aspects of tumor-immune interactions, the authors explained.¹

“While several of these biomarkers have shown predictive potential, none has proven sufficient as a standalone tool, highlighting the need for multidimensional strategies to refine patient stratification,” they wrote.

The ultimate answer, though, may turn out to be not 1 single biomarker, but an integrative approach like those enabled by AI. They said AI-based models “may facilitate the integration of heterogeneous biomarker signals and support the translation of complex biological information into clinically actionable frameworks.”

References

1. Gariazzo E, Colamartini F, Ubaldi M, et al. Emerging predictive biomarkers of immunotherapy sensitivity in patients with non-small cell lung cancer. Immunotargets Ther. 2026;15:567238. doi:10.2147/ITT.S567238

2. Christopoulos P, Harel M, McGregor K, et al. Plasma proteome-based test for first-line treatment selection in metastatic non-small cell lung cancer. JCO Precis Oncol. 2024;8:e2300555. doi:10.1200/PO.23.00555