Two blood proteins illuminate a neglected lung-cancer question

One of the hardest problems in cancer prevention is knowing whom to screen when the usual risk factors are absent. For people who have never smoked, standard smoking-history-driven lung screening is often irrelevant. Researchers analyzed 2,821 circulating proteins in prediagnostic blood from 180 lung-cancer cases and 180 matched controls across four cohorts, then examined candidate associations in the UK Biobank.

Two markers stood out: CEACAM5 and CEACAM6. In the discovery analysis, the odds ratios per standard-deviation increase were 1.65 (95% CI 1.31–2.09) and 1.72 (1.33–2.22). Associations were reproduced in an independent cohort, with hazard ratios of 1.92 and 1.86 across the replication population. Some markers associated with smoking-related lung cancer performed differently among those who had never smoked. This is evidence that the biological signature of risk may vary meaningfully across populations.

That is not yet evidence that measuring either protein should trigger CT scanning. A useful screen must estimate absolute risk, have acceptable calibration and specificity, change an actionable decision, and ultimately improve important outcomes while avoiding excess overdiagnosis and unnecessary investigations. The discovery population was small, heavily Asian in composition, and the replication set had a different smoking-status mix.