DOI: 10.1001/jamanetworkopen.2026.41494 ISSN: 2574-3805

Protein Biomarkers of Lung Cancer Risk Among Individuals Who Never Smoked

Hana Zahed, Connie Bulos-Salim, Karine Alcala, Shaymaa AlWaheidi, Xiaoshuang Feng, Ying Wang, Alpa V. Patel, Loïc Le Marchand, Wei Zheng, Xiao-Ou Shu, Mattias Johansson, Hilary A. Robbins
Show PDF Cite

Importance

Approximately one-quarter of lung cancers occur in people who never smoked (PWNS). For smoking-associated lung cancer, blood-based protein biomarkers can improve risk assessment.

Objective

To identify protein biomarkers for lung cancer among PWNS.

Design, Setting, and Participants

This nested case-control study from 4 cohorts in the Lung Cancer Cohort Consortium (LC3) was conducted for discovery to identify proteins, and a cohort study was conducted using data in the UK Biobank for replication. Participants were from the US, China, and the United Kingdom and donated blood samples between 1996 and 2010. Lung cancer case-control pairs of PWNS from 4 cohorts in the LC3 were matched by cohort, age, sex, and blood collection date. In addition, 49 291 participants in the UK Biobank were included. Data were analyzed from March 2024 to May 2025.

Exposure

A total of 2821 circulating proteins measured in prediagnostic blood samples collected up to 5 years before lung cancer diagnosis for cases.

Main Outcomes and Measures

Primary measures were odds ratios (ORs) for discovery and hazard ratios (HRs) for replication.

Results

Among 180 case-control pairs of PWNS (360 participants) in the 4 LC3 cohorts, 256 participants (71.1%) were female, and the median (IQR) age at enrollment was 66 (55-70) years. In LC3, using a false discovery rate–adjusted P value threshold of .05, 2 proteins were identified as being associated with lung cancer among PWNS: carcinoembryonic antigen–related cell adhesion molecule 5 (CEACAM5) (OR, 1.65 [95% CI, 1.31-2.09]) and CEACAM6 (OR, 1.72 [95% CI, 1.33-2.22]). Associations for both proteins were successfully replicated among participants analyzed regardless of smoking history across 5 years of follow-up in the UK Biobank (CEACAM5: HR, 1.92 [95% CI, 1.66-2.22]; CEACAM6: HR, 1.86 [95% CI,1.59-2.16]). The correlation between CEACAM5 and CEACAM6 was weak (Pearson ρ = 0.25 in the LC3); therefore, both proteins maintained associations with high relative risks after mutual adjustment. For 36 proteins previously identified as robust markers of smoking-related lung cancer, most showed no evidence of an association among PWNS (median OR, 1.47 [IQR, 1.38-1.61] vs 1.01 [IQR, 0.94-1.13] in smoking-related vs nonsmoking lung cancer).

Conclusions and Relevance

In this nested case-control study, 2 proteins were identified, CEACAM5 and CEACAM6, that were associated with lung cancer in PWNS. Larger studies are needed to evaluate whether there are additional biomarkers associated with lung cancer in PWNS and whether they can identify PWNS with high lung cancer risk.

More from our Archive