The landscape of biomarker testing in non–small cell lung cancer (NSCLC) is undergoing a significant transformation, moving beyond foundational genomic drivers to embrace a new wave of protein-based and computationally derived markers. This expansion is driven by the approval of novel therapies, particularly antibody-drug conjugates (ADCs), and an evolving understanding of therapeutic resistance and vulnerability, as explained by Soo-Ryum (Stewart) Yang, MD, during his presentation at the
In his presentation, Yang, an assistant attending pathologist and co-director of Clinical Biomarker Development in the Department of Pathology and Laboratory Medicine at Memorial Sloan Kettering Cancer Center in New York, New York, underscored 4 key trends: the rise of protein-based immunohistochemistry (IHC) biomarkers for ADCs, the actionability of tumor suppressor genes, the therapeutic application of synthetic lethality, and the advent of computational pathology. The persistent challenge remains tissue scarcity, highlighting a critical need to develop and implement multiplex IHC and integrate broad panel next-generation sequencing (NGS, IHC, and artificial intelligence (AI) to deliver the next generation of personalized therapies to larger segments of the NSCLC patient population.
What factors have propelled the rise of protein biomarkers in NSCLC?
Beyond a cancer cell’s genetic blueprint, the expression levels of certain proteins on its surface is emerging as a collection of critical, actionable biomarkers. Instead of looking for a mutated gene, pathologists are now measuring the intensity of protein expression, which can open entirely new treatment options for patients.
Although PD-L1 IHC testing has been established to guide checkpoint inhibitor therapy, IHC testing is now being used to guide ADC usage. Yang highlighted 2 “must-test” protein biomarkers in NSCLC: HER2 and c-MET overexpression. He emphasized that these protein-level biomarkers are distinct from their genetic counterparts. This distinction is crucial, with HER2 overexpression seen in up to 20% of patients and the highest level (IHC 3+) seen in up to 3% of patients; however, there is no correlation between HER2 mutation status and overexpression, Yang noted. Most NSCLC cases with high-level gene amplification will show IHC 3+ staining, but the reverse is not true: not all 3+ cases are driven by amplification, he explained.
The approval of fam-trastuzumab deruxtecan-nxki (T-DXd; Enhertu) for HER2-positive (IHC 3+) solid tumors, including patients with NSCLC who have received prior treatment was supported by the phase 2 DESTINY-Lung01 study (NCT03505710), which utilized the HER2 scoring guidelines used in gastric cancer.2 These gastric cancer guidelines should now be applied to NSCLC testing, said Yang.
c-MET overexpression is common in NSCLC, with an actionable c-MET–high status, defined as over 50% of tumor cells with 3+ staining, is found in up to 17% of EGFR wild-type cases, Yang reported. Like HER2 overexpression, c-MET overexpression can coexist with other driver mutations but is a distinct biomarker from MET exon 14 skipping mutations and MET amplification.
In May 2025, the FDA granted accelerated approval to telisotuzumab vedotin-tllv (teliso-V; Emrelis) in this patient population, supported by data from the phase 2 LUMINOSITY trial (NCT03539536).3
Yang noted that the integration of HER2 and c-MET IHC screening presents a significant challenge to current diagnostic workflows and proposed 2 primary strategies:
- Upfront Reflex Testing: Automatically order HER2 and c-MET IHC testing on the initial diagnostic sample as part of a comprehensive reflex biomarker panel alongside NGS and other IHC tests.
- Testing at Progression: Order the tests on a per-request basis upon disease progression, either on a new biopsy or an archived sample. This aligns with their current approval in the second-line setting.
There is no single solution applicable to all practice settings, according to Yang. He recommended a flexible approach with standardized options, allowing institutions to develop optimized workflows based on multidisciplinary input and their specific resources.
What emerging biomarkers are on the horizon in NSCLC?
Several promising biomarkers are under investigation and have the potential to become part of the standard of care, further refining personalized treatment for patients with NSCLC, Yang emphasized.
KRAS Mutations
KRAS mutations occur in up to 40% of lung adenocarcinomas, with mutations in codons G12, G13, and Q61, Yang stated. The KRAS G12C mutation is the most common, followed by the KRAS G12V and KRAS G12D mutations. Yang explained that KRAS G12D mutations are associated with never or light smoking history, a lower tumor mutational burden, and lower PD-L1 expression and correlate with poorer response to chemoimmunotherapy, making them a challenge of note in the field.
KRAS G12C-specific targeted therapies, including sotorasib (Lumakras) and adagrasib (Krazati), are established, approved therapies, and the development of targeted therapies beyond KRAS G12C–directed agents, such as multi-RAS and RAS(ON) inhibitors, are in clinical trials. Yang noted zoldonrasib (RMC-9805), a KRAS G12D inhibitor, which yielded an overall response rate of 61% (n = 11) and a disease control rate of 89% (n = 16) in a phase 1 study (NCT06040541).4 The multi-RAS inhibitor daraxonrasib (RMC-6236) has also shown promise in KRAS G12V–mutant NSCLC, as well as in pancreatic cancer.
As KRAS mutations are easily detected by existing NGS and PCR technologies, they don’t present the same workflow challenges as IHC testing for HER2 and c-MET, Yang explained.
STK11 and KEAP1 Mutations
STK11 and KEAP1 mutations are tumor suppressor genes mutated in up to 20% of lung cancers, often co-mutated with KRAS, Yang continued.Mutations in STK11/KEAP1 promote an immunosuppressive tumor microenvironment, leading to primary resistance to immunotherapy. They are considered biomarkers of poor response to single-agent PD-1/PD-L1 inhibitors.
Analysis of the phase 3 POSEIDON trial (NCT03164616), which investigated first-line durvalumab (Imfinzi) with or without tremelimumab (Imjudo) plus chemotherapy in metastatic NSCLC, suggests a path to overcome this resistance, according to Yang.5 The addition of a CTLA-4 inhibitor to a PD-L1 inhibitor and chemotherapy improved progression-free survival (PFS) and overall survival (OS) in these patients. This positions STK11/KEAP1 mutations as potential biomarkers for escalating checkpoint therapy and could become a biomarker for identifying patients who need a more aggressive, multi-pronged immunotherapy approach. These mutations require broad-panel NGS to detect the full range of inactivating mutations across these tumor suppressor genes; PCR is not a feasible approach.
MTAP Deletions and Synthetic Lethality
As Yang explained, MTAP plays a key role in the purine salvage pathway. Its deletion in cancer cells impairs the activity of the enzyme PRMT5, creating a metabolic vulnerability. This “first hit” can be exploited by a “second hit”—the therapeutic inhibition of PRMT5 or MAT2a—to induce selective cancer cell death, an approach known as synthetic lethality.
MTAP deletions occur in up to 18% of lung cancers and are associated with poor outcomes, particularly with immunotherapy, Yang said. They are an emerging therapeutic target, with promising clinical trial data for PRMT5 inhibitors in MTAP-deleted lung cancers.
Detection methods include NGS, which detects homozygous deletion of the MTAP gene and requires no additional tissue if MTAP is included on the panel, but is highly sensitive to tumor purity and requires deletion on over 20% tumor cells for reliable results, according to Yang. IHC can also be used and detects loss of the MTAP protein expression in the cytoplasm. This is more sensitive, especially for low-purity samples, but requires an additional tissue slide. Yang proposed a diagnostic workflow involving using NGS for initial screening, followed by confirmatory IHC in cases where the MTAP status is retained or borderline, especially in low-purity samples.
“Despite this progress, I think tissue will still be the issue,” Yang said. “We have the same small biopsy, and we’re required to [test for] an expanding list of biomarkers that may require additional separate tissues.”
TROP2 and Computational Pathology
TROP2 is a cell surface protein widely expressed in NSCLC, making it an attractive target for ADC development. Datopotamab deruxtecan-dlnk (Dato-DXd; Datroway), an anti-TROP2 ADC, is being explored as a second-line agent. The phase 3 TROPION-Lung01 study (NCT04656652) showed a PFS benefit with Dato-DXd over docetaxel but no statistically significant OS benefit.6 The trial did not enrich for a biomarker, as previous studies found no correlation between TROP2 expression and response.
Yang explained that, to improve predictive power, investigators developed an AI-driven method using computational pathology. Here, an IHC slide is scanned, and an AI algorithm measures the optical density of TROP2 staining in the membrane and cytoplasmic components of tumor cells. This generates a TROP2 quantitative continuous score (QCS) or normalized membrane ratio (NMR), and this continuous score is converted into a binary positive/negative result.
When applied retrospectively to the TROPION-Lung01 study, TROP2 QCS/NMR positivity was predictive of higher response rates and longer PFS with Dato-DXd. Although compelling, Yang noted that this proof of concept requires prospective validation in independent cohorts. Additionally, Yang raised concerns that the biomarker is currently tied to a specific, proprietary digital pathology ecosystem, raising questions about broader accessibility and implementation on other platforms.
What does the future look like for personalized medicine in NSCLC management?
The toolbox for fighting lung cancer is expanding dramatically, moving beyond an exclusive focus on genomics and into a more holistic approach that incorporates protein analysis, AI-driven insights, and novel therapeutic strategies like synthetic lethality. These advancements are making personalized medicine a reality for broader segments of the lung cancer population.
“We’re at a point where we should be starting to explore the feasibility of multiplex IHC similar to what we did with molecular markers and NGS,” Yang concluded. “In the next few years, broad-panel NGS and IHC, along with AI, are going to be the cornerstones of comprehensive biomarker testing in lung cancer.”
References
- Yang S. Where are we heading? Expanding horizons in NSCLC biomarker testing. Presented at: 20th Annual New York Lung Cancers Symposium; November 15, 2025; New York, NY.
- FDA grants accelerated approval to fam-trastuzumab deruxtecan-nxki for unresectable or metastatic HER2-positive solid tumors. News release. US FDA. April 5, 2024. Accessed November 15, 2025. https://www.fda.gov/drugs/resources-information-approved-drugs/fda-grants-accelerated-approval-fam-trastuzumab-deruxtecan-nxki-unresectable-or-metastatic-her2
- FDA grants accelerated approval to telisotuzumab vedotin-tllv for NSCLC with high c-Met protein overexpression. FDA. April 5, 2024. Accessed November 15, 2025. https://www.fda.gov/drugs/resources-information-approved-drugs/fda-grants-accelerated-approval-telisotuzumab-vedotin-tllv-nsclc-high-c-met-protein-overexpression
- Arbour KC, Tawee T, Yaeger R, et al. Abstract CT019: preliminary safety and antitumor activity of zoldonrasib (RMC-9805), an oral, RAS(ON) G12D-selective, tri-complex inhibitor in patients with KRAS G12D non-small cell lung cancer (NSCLC) from a phase 1 study in advanced solid tumors. Cancer Res. 2025;85(suppl 2):CT019. doi:10.1158/1538-7445.AM2025-CT019
- Johnson ML, Cho BC, Luft A, et al. Durvalumab with or without tremelimumab in combination with chemotherapy as first-line therapy for metastatic non-small-cell lung cancer: the phase III POSEIDON study. J Clin Oncol. 2023;41(6):1213-1227. doi:10.1200/JCO.22.00975
- Ahn MJ, Tanaka K, Paz-Ares L, et al. Datopotamab deruxtecan versus docetaxel for previously treated advanced or metastatic non-small cell lung cancer: the randomized, open-label phase III TROPION-Lung01 study. J Clin Oncol. 2025;43(3):260-272. doi:10.1200/JCO-24-01544