Plasma biomarkers predicting development of lung cancer 5 years before diagnosis Authors: Kostas A Papavassiliou, Athanasios G Papavassiliou Published: 2026 Lung cancer remains the leading cause of cancer-related mortality worldwide despite advances in tobacco control, molecular therapeutics, and low-dose computed tomography (LDCT) screening. Although LDCT reduces lung cancer mortality [1, 2], current screening strategies still rely largely on age and smoking history and remain imperfect for identifying individuals most likely to develop disease [3, 4]. Detecting biological processes that precede malignant transformation, rather than the cancer itself, therefore represents a major unmet need in cancer prevention [5]. Recently, Pandya and colleagues reported a 14-protein plasma signature that predicts lung cancer more than five years before diagnosis and provided mechanistic evidence linking this signature to environmental exposure, interleukin-1β (IL-1β) signaling, and epithelial cell-state transitions involved in early lung tumorigenesis [6]. The study integrates population-scale proteomics, machine learning, single-cell transcriptomics, experimental models of epidermal growth factor receptor (EGFR)-driven lung adenocarcinoma, and analyses of the Canakinumab Anti-inflammatory Thrombosis Outcomes Study (CANTOS), offering an unusually comprehensive investigation of a circulating cancer-risk signal. The most important contribution of this work is conceptual. Most blood-based biomarker platforms seek evidence of cancer that already exists. Circulating tumor DNA (ctDNA), DNA fragmentation profiles, methylation signatures, and multi-cancer early detection assays all aim to identify occult malignancies before they become clinically apparent (Table 1) [7–10]. These technologies represent major advances and will likely become increasingly integrated into cancer screening paradigms. However, they are fundamentally designed to detect the consequences of malignant transformation after the process of carcinogenesis has already been initiated. Table 1Evolution of blood-based biomarker strategies for lung cancer: from early detection to precision cancer interception#ApproachRepresentative landmark studiesBiological signal interrogatedClinical objectiveMajor strengthsCurrent limitationsHost-response proteomic biomarkersINTEGRAL program; Davies et al. [14, 15]Circulating proteins reflecting inflammation, tissue remodeling and host responseImprove individualized lung cancer risk predictionBlood-based; improves discrimination beyond clinical models; compatible with screening programs; demonstrates long-term risk prediction years before diagnosisBiological specificity remains limited; influenced by chronic inflammatory conditions; limited mechanistic insightTumor-derived cfDNA biomarkersDELFI [7]Cell-free DNA fragmentation patternsDetect early lung cancerNon-invasive; tumor-derived molecular signalLimited sensitivity in very early-stage disease; requires sufficient tumor DNA sheddingMCED methylation-based assaysCCGA; PATHFINDER [8, 10]Genome-wide cfDNA methylation signaturesMulti-cancer early detection and tissue-of-origin predictionHigh specificity; detects multiple cancer types simultaneously; prospective implementation data availableCost; uncertain role in population screening; lower sensitivity for some stage I cancersMulti-analyte liquid biopsyCancerSEEK [9]Combined ctDNA mutations and circulating proteinsEarly cancer detectionIntegrates complementary molecular information; broad applicabilityVariable sensitivity across tumor types and stages; requires further prospective validationTumor-promotion proteomic signaturePandya et al. [6]Circulating proteins reflecting inflammatory signaling, epithelial plasticity and a tumor-promoting microenvironmentPrecision cancer interception and biomarker-guided preventionMechanistically linked to lung carcinogenesis; validated across 8 independent cohorts; predicts risk > 5 years before diagnosis; may identify individuals most likely to benefit from preventive anti-inflammatory interventionsRequires prospective validation; specificity versus chronic inflammatory lung diseases remains to be established; clinical utility beyond current screening paradigms yet to be demonstrated#Representative blood-based biomarker approaches for lung cancer risk assessment, early detection, and cancer interception are compared according to the biological signal interrogated, principal clinical objective, major strengths, and current limitations. The study by Pandya et al. introduces a distinct paradigm by identifying a circulating proteomic signature that reflects a tumor-promoting microenvironment, thereby extending the role of blood-based biomarkers beyond early cancer detection toward biomarker-guided cancer preventionCCGA, Circulating Cell-free Genome Atlas; cfDNA, cell-free DNA; ctDNA, circulating tumor DNA; DELFI, DNA Evaluation of Fragments for Early Interception; MCED, multi-cancer early detection Evolution of blood-based biomarker strategies for lung cancer: from early detection to precision cancer interception# #Representative blood-based biomarker approaches for lung cancer risk assessment, early detection, and cancer interception are compared according to the biological signal interrogated, principal clinical objective, major strengths, and current limitations. The study by Pandya et al. introduces a distinct paradigm by identifying a circulating proteomic signature that reflects a tumor-promoting microenvironment, thereby extending the role of blood-based biomarkers beyond early cancer detection toward biomarker-guided cancer prevention CCGA, Circulating Cell-free Genome Atlas; cfDNA, cell-free DNA; ctDNA, circulating tumor DNA; DELFI, DNA Evaluation of Fragments for Early Interception; MCED, multi-cancer early detection By contrast, the 14-protein signature described by Pandya et al. appears to capture the biological conditions that permit cancer to emerge. Rather than functioning primarily as a marker of tumor burden, it may represent a systemic readout of a tumor-promoting microenvironment that precedes malignant transformation by several years. This distinction is subtle but important. Early detection aims to identify cancer at its earliest clinically actionable stage, whereas cancer interception seeks to identify and target the biological processes that drive malignant evolution before invasive disease develops. The present study therefore extends beyond the traditional framework of early detection and aligns with an emerging vision of precision cancer interception. This concept is consistent with accumulating evidence that oncogenic mutations can be present in histologically normal tissues and that additional inflammatory or environmental cues are required for malignant progression [11]. Mechanistically, Pandya and colleagues show that diverse epithelial lineages converge upon a keratin-8-positive, claudin-4-positive (Krt8+/Cldn4+) alveolar transitional state during early EGFR-driven tumorigenesis [6]. These findings build upon emerging work implicating transitional alveolar epithelial states as critical intermediates linking tissue injury, regeneration, fibrosis, and lung adenocarcinoma evolution [12, 13]. Equally compelling is the demonstration that particulate matter exposure, oncogenic EGFR signaling, and IL-1β-mediated inflammation independently and cooperatively induce components of the circulating signature [6]. These observations provide a biologically coherent framework linking environmental exposure, inflammatory signaling, epithelial plasticity, and future cancer risk. The findings further extend prior evidence that air pollution can promote EGFR-mutant lung adenocarcinoma through macrophage-derived IL-1β signaling [11]. Together, these data suggest that the circulating proteomic signal reflects a dynamic ecosystem of tumor promotion rather than the presence of an established cancer. Large international efforts, including the Integrative Analysis of Lung Cancer Etiology and Risk (INTEGRAL) program, have previously demonstrated that circulating proteins measured years before diagnosis can substantially improve risk stratification beyond traditional clinical variables [14]. Another important milestone in this field was recently provided by Davies and colleagues, who applied large-scale plasma proteomics and machine-learning approaches to identify protein signatures capable of predicting lung cancer risk up to 12 years before diagnosis [15]. By identifying 150 and 247 plasma proteins associated with lung cancer development 1–3 years and 1–5 years before diagnosis, respectively, this study established that circulating proteomic alterations precede the clinical manifestation of lung cancer by many years and demonstrated the potential of blood-based proteomic profiling for long-term risk prediction. These findings provided an important foundation for subsequent efforts aimed at developing clinically useful plasma biomarker signatures for lung cancer prevention and early intervention. Pandya and colleagues build upon this foundation but move the field in an important new direction. Whereas prior proteomic studies were primarily developed and evaluated as predictive tools, the present work attempts to explain the biological origins of the circulating signal itself. By integrating proteomic observations with experimental models, single-cell analyses, and environmental exposure studies, the authors provide evidence that the measured proteins are not merely statistical correlates of future cancer risk but may reflect active processes involved in tumor promotion. This mechanistic dimension substantially strengthens the biological plausibility of the signature and distinguishes the study from many previous biomarker-discovery efforts. Taken together, the simplicity of the 14-protein signature, its validation across eight independent cohorts, its mechanistic dissection in genetically engineered mouse models, and its retrospective clinical validation in the CANTOS trial collectively distinguish the study by Pandya et al. as a major advance toward biologically informed precision cancer interception. The translational implications are particularly intriguing. The CANTOS trial unexpectedly demonstrated that IL-1β inhibition with canakinumab reduced lung cancer incidence and mortality in exploratory analyses, providing one of the strongest clinical proofs-of-concept for pharmacologic cancer prevention [16, 17]. However, the relatively low absolute incidence of lung cancer in the overall study population resulted in a large number needed to treat, limiting enthusiasm for widespread preventive use. As is frequently the case in prevention research, identifying the appropriate population for intervention proved as important as identifying the intervention itself. Pandya et al. now suggest that individuals with elevated baseline protein signatures may derive greater benefit from IL-1β blockade, substantially reducing the number needed to treat compared with unselected populations [6]. Although these findings should be interpreted cautiously and remain hypothesis-generating, they introduce a potentially transformative concept, that is, biomarker-guided cancer prevention. Just as predictive biomarkers have enabled precision oncology in advanced disease, biomarkers of tumor-promoting biology may eventually enable precision prevention. Such an approach could enrich prevention trials for biologically susceptible individuals, improve risk-benefit ratios, reduce unnecessary exposure to preventive therapies, and accelerate development of targeted interception strategies. Importantly, however, prospective validation in dedicated prevention studies will be essential before any clinical implementation can be contemplated. The broader significance of this work becomes even more apparent when viewed alongside other biomarker approaches for lung cancer risk assessment and early detection (Table 1). Recent advances in proteomics, cfDNA fragmentation analyses, methylation-based assays, and multi-analyte liquid biopsies have demonstrated considerable promise for detecting early-stage cancers [7–10]. Yet these approaches largely seek evidence of cancers that have already emerged. The study by Pandya and colleagues instead seeks to identify the biological conditions that permit malignant evolution, thereby aligning more closely with the emerging paradigm of cancer interception. Several important questions nevertheless remain. First, the biological specificity of the signature is uncertain. The authors demonstrate enrichment in chronic obstructive pulmonary disease and idiopathic pulmonary fibrosis, raising the possibility that the panel reflects broader processes of pulmonary injury and remodeling rather than lung cancer-specific biology [6]. Whether this represents a limitation or, alternatively, a biologically meaningful measure of carcinogenic susceptibility remains to be determined. Second, the identification of Krt8+/Cldn4 + alveolar transitional cells (KACs) as a convergent cellular state raises important biological questions. Future studies should establish whether analogous KAC-like states are present within human pre-invasive lesions, including atypical adenomatous hyperplasia, adenocarcinoma in situ, and minimally invasive adenocarcinoma, using spatial transcriptomics and multiplex tissue imaging. Furthermore, although IL-1β signaling clearly promotes expansion of these transitional cells, the downstream molecular programs governing their emergence remain incompletely understood. Defining these pathways may uncover additional opportunities for cancer interception beyond IL-1β blockade alone. Third, the retrospective nature of the CANTOS analyses precludes definitive conclusions regarding biomarker-guided prevention. Fourth, implementation will require demonstration of incremental clinical utility beyond established screening approaches and validation across diverse populations, environmental exposures, smoking histories, and molecular subtypes of lung cancer. In particular, although the TALENT cohort (predominantly never-smoking Taiwanese individuals) provides encouraging proof-of-concept that selected components of the signature are elevated before diagnosis in predominantly never-smoking Asian individuals, the relatively short interval between blood sampling and diagnosis and the limited sample size preclude definitive conclusions. Larger prospective studies with longer follow-up will therefore be required to establish whether the complete 14-protein signature, or a modified version, retains robust predictive performance in never-smokers, Asian populations, and EGFR-driven lung adenocarcinoma. An additional translational challenge relates to assay standardization. The current study combines measurements obtained using Olink and SomaScan platforms, both of which provide relative rather than absolute protein quantification. Consequently, cross-platform harmonization and establishment of universally applicable clinical thresholds remain challenging. Translation into routine clinical practice will ultimately require standardized assays with absolute quantification, robust reference standards, and analytical validation suitable for regulatory approval and widespread clinical implementation. Ultimately, the importance of the 14-protein signature will depend not only on its predictive performance but also on its ability to guide interventions that improve outcomes. Prospective studies should determine whether biomarker-informed prevention strategies can reduce lung cancer incidence, whether serial measurements provide dynamic assessment of carcinogenic risk, and whether intervention based on biomarker status can meaningfully alter the natural history of disease. Nonetheless, by shifting attention from the detection of occult malignancy to the identification of a biologically actionable state of tumor promotion, Pandya and colleagues provide a compelling vision for the future of precision cancer interception and prevention. Acknowledgements Not applicable. Author contributions KAP: Conceptualization; Writing original draft and table; Writing review & editing. AGP: Conceptualization; Writing review & editing; Supervision. All authors read and approved the final manuscript. Funding The authors declare that no funding was received for this work. Data availability No datasets were generated or analysed during the current study. Declarations Ethics approval and consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement No datasets were generated or analysed during the current study.