Scientists identify blood proteins that predict lung cancer five years early

Lung cancer kills more people worldwide than any other malignancy. It is also, paradoxically, one we have learned to detect and treat with increasing sophistication over the past two decades—yet fewer than one in three patients survives five years after diagnosis. Now a team of more than eighty researchers working across four continents believes they may have found a way to stop the disease before it starts.

The scientists identified fourteen proteins in the blood that predict lung cancer development with striking accuracy—more than five years before a tumor appears on any scan. The discovery emerged from analysis of 48,000 blood samples drawn from the UK Biobank, processed through machine learning algorithms to isolate the protein signature. When researchers combined this signature with patient age, smoking history, and prior lung disease, their predictive model outperformed every existing risk assessment tool. The team then validated the finding across eight additional datasets worldwide, including a cohort from Taiwan composed largely of people who had never smoked. The work was published Thursday in the journal Cell.

What makes the finding potentially transformative is not just the prediction itself, but what the researchers discovered about what drives it. Using mouse models and cell cultures, they showed that these fourteen proteins surge when a specific inflammatory pathway activates—a pathway that smoking and air pollution can trigger. This suggests that lung cancer does not arise from genetic mutations alone. Rather, as Charles Swanton, the study's lead author and clinical director of the Francis Crick Institute in London, explained it: smoke causes both mutations and inflammation, and the two together cause cancer. The same inflammatory signature appeared elevated in people who later developed chronic obstructive pulmonary disease and pulmonary fibrosis, pointing to a common inflammatory environment that precedes all three diseases.

Inflammation, unlike genetic damage, is something medicine might be able to attack before cancer takes hold. To test this possibility, the researchers examined data from a randomized controlled trial of canakinumab, an anti-inflammatory drug that targets the same pathway linked to the protein signature. The trial had been designed to test the drug's effect on heart attacks, where it showed only modest benefit. But among the 2,300 trial participants who had above-average levels of the fourteen proteins, canakinumab reduced lung cancer risk by nearly half. Swanton compared the potential to statins: doctors identify patients with high LDL cholesterol and treat them with drugs that significantly lower their risk of heart attack and stroke. "This is a kind of equivalent to an LDL for cancer," he said.

Yet the path from discovery to clinical use remains uncertain and fraught. The protein signature needs validation through additional studies, and a practical blood test must be developed for patient use. Most critically, a dedicated clinical trial must prove whether canakinumab actually prevents lung cancer—not merely correlates with lower risk in retrospective data. Roy Herbst, chief of medical oncology at Yale School of Medicine, framed the central question: "Will it be clinically significant? Will we be able to block this enough at the right stage to prevent cancer?" Peter Mazzone, a pulmonologist at Cleveland Clinic, raised another concern: canakinumab carries significant side effects, including increased risk of infection and sepsis. Even in a narrowly defined high-risk population, the drug might prove too toxic for the benefit to justify the harm. A different medication targeting the same pathway might work better and safer.

Beyond drug treatment, the protein signature could reshape how doctors approach screening. Currently, American physicians recommend annual low-dose CT scans for people aged fifty to eighty with a twenty pack-year smoking history who currently smoke or quit within the past fifteen years. But many eligible people never get screened. A blood test identifying those most likely to benefit could increase uptake. It might also flag people outside current screening criteria who carry genuine risk—particularly never-smokers, a population where lung cancer detection remains inadequate. Mazzone noted there is "a great need" for better screening in this group, though the new research suggests only that some components of the protein signature might help identify at-risk never-smokers; more data is required.

Douglas Arenberg, a professor of medicine at the University of Michigan who was not involved in the study, called the findings a promising starting point toward a long-elusive public health goal. "Preventing lung cancer was an elusive holy grail for a very long time," he said. The researchers may have identified not just a risk marker, but one that also predicts who would benefit from a specific preventive drug. Herbst, reflecting on three decades in oncology, noted that lung cancer has transformed from an incurable disease to a treatable condition in some patients. "But the greatest benefit will still be detecting it at the earliest stages, or even preventing it," he said. "This is a step forward in that direction."