A simple blood test might prevent a disease that kills two million people yearly
Each year, lung cancer quietly claims nearly two million lives — most of them in people who had no warning until it was too late. A team of scientists at the Francis Crick Institute in London has now identified a 14-protein signature in blood plasma that can detect who will develop the disease up to five years before diagnosis, offering, for the first time, a credible biological window for prevention. The discovery also points toward an existing anti-inflammatory drug as a possible shield against the disease — a reminder that medicine sometimes finds its answers in unexpected places. The road from insight to intervention remains long, but the direction, at last, is clear.
- Lung cancer kills 1.8 million people annually, and the vast majority of those deaths occur because the disease is caught too late — making early detection not a convenience but a matter of survival.
- A machine-learning analysis of nearly 48,000 blood samples uncovered 14 plasma proteins that flag future lung cancer cases with over 75% accuracy, years before any symptom appears.
- The finding reframes lung cancer as a preventable inflammatory process — smoking mutates cells, pollution inflames them, and cancer follows — opening the door to intervention before the disease ever begins.
- Reanalysis of an old cardiac drug trial revealed that canakinumab cut lung cancer risk by 50% in people carrying the protein signature, suggesting a drug already in existence could be repurposed as a cancer preventive.
- The path to clinical use is obstructed by serious hurdles: the drug costs $73,000 per year, carries significant toxicity, is unavailable in many high-burden countries, and the signature itself must still be validated across far more diverse global populations.
Lung cancer kills nearly two million people every year, and most of those deaths come too late to prevent — because the disease announces itself only after it has already taken hold. A team led by Charles Swanton at the Francis Crick Institute in London may have found a way to change that. By analyzing blood plasma samples from roughly 48,000 volunteers in the U.K. Biobank, they used machine learning to identify 14 specific proteins that appear in people who will go on to develop lung cancer. When tested on a separate group of 12,000 volunteers, the signature successfully identified more than 75% of those who later received a diagnosis — with a median lead time of over five years. The results were published in Cell in May and held up across eight additional international datasets.
What makes the finding especially compelling is what it reveals about cancer's origins. The 14-protein signature was strongest in people whose blood showed signs of inflammation — the kind triggered by smoking and air pollution. This fits a picture the same team has been assembling: environmental exposure awakens dormant mutant cells in the lungs, and the resulting inflammation eventually tips them toward cancer. If that sequence can be interrupted, the disease might be stopped before it starts.
That logic led the researchers to canakinumab, an FDA-approved anti-inflammatory drug made by Novartis. Reanalyzing data from a large cardiac trial called CANTOS, they found that participants carrying the 14-protein signature who received the drug saw their lung cancer risk fall by 50%. A medication designed for inflammatory disorders may, in the right population, double as a cancer preventive.
But the distance between discovery and widespread use is considerable. The protein signature was developed largely from populations in the U.K., U.S., and East Asia, and must be validated across more diverse groups before it can be used clinically. Canakinumab itself is a serious obstacle — it carries significant side effects, has been flagged by European regulators for toxicity concerns, and costs $73,000 per year in the United States. It is neither registered nor available in India, where lung cancer is a major burden. Swanton's team is candid about these limitations, framing their work as proof of concept rather than a finished solution. The blood proteome, they have shown, holds legible clues to cancer risk years before disease appears. Whether those clues can be translated into something affordable and accessible for the world's most vulnerable patients remains the defining challenge ahead.
Lung cancer kills nearly two million people every year. Most of those deaths are preventable—if we could catch the disease before it takes hold. A team of scientists led by Charles Swanton at the Francis Crick Institute in London believes they may have found a way to do exactly that: a simple blood test that can identify who will develop lung cancer years before symptoms appear.
The discovery rests on an elegant insight. Blood plasma—the liquid portion of blood—carries proteins from every organ and tissue in the body. These proteins tell a story about what is happening inside us. If you could read that story before disease strikes, you might be able to intervene. Swanton's team did exactly that. They examined blood samples from roughly 48,000 volunteers in the U.K. Biobank, a vast repository of health data and biological material from half a million people. Using machine learning, they identified 14 specific plasma proteins that appear in people destined to develop lung cancer. They call it the 14-protein signature.
When the researchers tested their model on 12,000 volunteers whose data had been held back from training, the results were striking. Among 75 people who later developed lung cancer, the blood test successfully flagged more than 75 percent of them—years before diagnosis. The median time between the blood test and actual diagnosis was 5.1 years. The signature held up when tested against eight additional datasets, including samples from lifelong non-smokers. The work was published in Cell in May.
What makes the discovery particularly intriguing is what it reveals about how lung cancer actually develops. The 14-protein signature was most pronounced in people whose blood showed signs of inflammation—the kind triggered by smoking and air pollution. This aligns with earlier research from the same team showing that environmental pollution wakes up dormant mutant cells in the lungs. The emerging picture is this: smoking causes mutations in lung cells; inflammation from environmental exposure follows; and eventually, those inflamed, mutated cells become cancer. If that sequence is correct, then treating the inflammation before cancer takes root might prevent the disease entirely.
That possibility led the researchers to an unexpected place: an old drug called canakinumab, made by Novartis. The medication is already approved by the FDA to treat inflammatory disorders. Years ago, it was tested in a large clinical trial called CANTOS, where it showed modest benefits in heart disease patients. Swanton's team reanalyzed that trial data and found something remarkable. Among the 2,300 CANTOS participants who carried the 14-protein signature and received canakinumab, the risk of developing lung cancer dropped by 50 percent. A drug designed for one purpose might prevent another disease entirely.
But the path from discovery to treatment is long and uncertain. The 14-protein signature was derived from a relatively homogeneous population—the U.K., the U.S., and East Asia. Before it can be used clinically, it must be validated in more diverse populations worldwide. A diagnostic test would need to be developed that can reliably detect all 14 proteins in a blood sample. Then canakinumab itself presents serious obstacles. The drug carries significant side effects; the European Medicines Agency concluded its toxicity could outweigh its benefits. And it is staggeringly expensive—$73,000 per year in the United States, with treatment typically requiring multiple doses over long periods. The drug is neither registered nor available in India, where lung cancer is a major public health burden.
Swanton's team acknowledges all of this. They note that even if canakinumab proves effective in preventing lung cancer, cheaper alternatives with better safety profiles will be essential before the approach can help people worldwide. The work is a proof of concept, a demonstration that the blood proteome holds clues to cancer risk years before the disease manifests. Whether those clues can be translated into a practical, affordable tool for prevention remains an open question. But for the first time, there is a credible path forward.
Notable Quotes
The 14-protein signature was more pronounced when specific inflammatory pathways involving smoking and air pollution were activated— Research findings from Swanton et al.
Canakinumab could be a potential drug to prevent future lung cancer in those with the 14-protein signature— Swanton et al., based on CANTOS trial retrospective analysis
The Hearth Conversation Another angle on the story
So you're saying a blood test can predict lung cancer five years early. How is that even possible?
The blood carries proteins from every tissue in your body. If you compare the protein patterns of people before and after they get sick, you can spot what changed. These 14 proteins appear to shift in people headed toward lung cancer, sometimes years before any tumor shows up.
And the test caught 75 percent of cases? That seems almost too good.
It is good, but there's a catch. They tested it on people whose blood had already been stored. Real-world use would be different. You'd have to screen millions of people to find the ones at risk, and you'd need to make sure the signature works the same way in different populations.
The drug they mention—canakinumab—it's not new, right?
No, it was designed for heart disease. But when they looked back at old trial data, they noticed people with the 14-protein signature who took it had half the lung cancer risk. That's intriguing, but it's also a small signal from a study that wasn't designed to test that.
What's the real barrier here?
Money and safety. The drug costs $73,000 a year and has serious side effects. In most of the world, including India, it's not even available. You'd need something cheaper and safer before this could actually help people.