On June 4 2026, a paper was published in the journal Cell, in which researchers from The Francis Crick Institute, University College London and elsewhere, identified a set of blood proteins that appeared to predict lung cancer five years before it became clinically apparent. In addition, the researchers proposed that a human monoclonal antibody called Canakinumab, which targets interleukin-1 beta (a component of the body’s immune response), might reduce the risk of lung cancer among individuals with this protein signature, based on a retrospective analysis of an earlier clinical trial.
The dramatic appeal of these claims attracted significant media attention, leading many people to believe that they could simply do a blood test to predict lung cancer and then take a drug to prevent it from occurring. However, medical research is far too complex to simplify into generic headlines such as these. It is important to explore the facts behind the research and understand the fallacies in making overtly optimistic predictions based on the findings.
The original trial
The drug being studied is called Canakinumab, an anti-inflammatory injection used in certain types of arthritis. Because of its ability to suppress inflammation, earlier researchers were keen to see if it could reduce future cardiovascular events in people who had previously suffered a heart attack. This formed the basis of the CANTOS trial (Canakinumab Antiinflammatory Thrombosis Outcome Study), published in 2017 in the New England Journal of Medicine. More than 10,000 patients with previous myocardial infarction and elevated levels of high-sensitivity C-reactive protein (indicating ongoing inflammation) were randomised to receive three different doses of the drug or a placebo.
The CANTOS trial was designed to assess cardiovascular outcomes, not cancer. After approximately four years of follow-up, people who received the drug had fewer cardiovascular events than those who received the placebo. However, those who received the drug also had more fatal infections, which is not surprising because interleukin-1 beta is involved in the body’s defence against infection. This risk was already known — and was further confirmed by the trial.
An additional finding
A surprising finding from a subsequent analysis — which was not part of the original CANTOS trial objective — was that fewer cases of lung cancer occurred in the Canakinumab group than in the placebo group. This attracted considerable academic attention and led to further publications proposing that inflammation within the lung, mediated by interleukin-1 beta, might play a role in cancer progression. Researchers suggested that the lower cancer numbers may be due to the drug blocking this inflammatory pathway.
However, such an observation could also arise for reasons other than a genuine preventive effect. One important limitation is that there was no systematic screening for lung cancer during the CANTOS trial. Participants did not undergo routine CT scans or other chest imaging to identify early cancers. The reported cases were simply those that came to medical attention because patients developed symptoms or underwent imaging for unrelated reasons.
This creates an important uncertainty. Some participants may already have had undiagnosed lung cancer when the trial began. Others may have developed lung cancer during the study but remained undetected. Without systematic screening at the beginning and end of the trial, it is impossible to know the true number of cancer cases in either group. Therefore, the observation of fewer diagnosed lung cancers in the treatment arm does not necessarily mean that the drug prevented lung cancer from developing.
This remains one of the major limitations of using the CANTOS findings to support a lung-cancer prevention strategy. Coupled with the lack of prior evidence that this line of treatment prevents lung cancer, the findings should be interpreted with caution.
What does the new paper say?
Essentially, a section of the recent Cell paper is a re-examination of the findings from the CANTOS trial. The authors assume that the reduction in lung cancer in the CANTOS trial was a genuine effect of the drug and then attempt to provide a biological explanation for it. However, even nine years after the publication of the original CANTOS paper, this observation remains isolated and has not been convincingly replicated in independent studies.
In the recent Cell paper, as part of their search for lung cancer predictors, researchers analysed blood samples from the U.K. Biobank, an ongoing project in which volunteers contribute health information for research purposes. Using machine-learning techniques, they identified a 14-protein plasma signature among people in the Biobank cohort who eventually developed lung cancer. The authors describe this pattern as reflecting a “perturbed lung environment”, suggesting that these markers may indicate lungs that are not in an optimal state of health.
Interestingly, individuals with this protein signature were not only more likely to develop lung cancer, but also other chronic lung diseases such as COPD and pulmonary fibrosis. The findings were subsequently validated in other cohorts, including a Taiwanese group with fewer smokers. Other forms of cancer did not share the same relationship.
Many media reports interpreted these findings as evidence that a blood test can predict lung cancer five years earlier. The reality is more nuanced. These proteins are not cancer markers in the conventional sense, unlike PSA, which is a recognised marker for prostate cancer. Rather, they may simply identify people whose lungs have already been adversely affected by smoking, pollution or both. In that sense, the test may be identifying unhealthy lungs rather than detecting cancer itself.
Nevertheless, the discovery is significant. It highlights the ability of machine-learning techniques to analyse large volumes of biological data and identify patterns that would otherwise be difficult to detect. Future studies may determine whether these markers can be refined and independently validated as tools for identifying people who would benefit from more intensive screening.
What does it signify?
The significance of this research lies in the fact that it is neither practical nor economical to screen entire populations for lung cancer. Doctors already know that people with a heavy smoking history are at greater risk and therefore recommend targeted screening. In addition, if this protein signature genuinely identifies people at particularly high risk, it could eventually help make screening programmes more efficient.
Whether treatment with Canakinumab actually reduces the incidence or progression of lung cancer remains uncertain. At present, there is no independent study that convincingly replicates the unexpected lung cancer findings of the CANTOS trial. Subsequent research studies using Canakinumab as a treatment option in lung cancer patients yielded no benefit. Importantly, the well-established risk of fatal infections is a serious concern while considering use of this drug among otherwise healthy people with an intent to prevent lung cancer.
What you should know
In summary, interpretation of medical research findings requires careful analysis, scrutiny of key references and cross-domain knowledge. Complex processes such as cancer and inflammation are multifactorial, involving numerous interacting mechanisms, many of which remain incompletely understood. Simplistic conclusions, especially when based on subgroup analyses that were not part of the original study design, should therefore be treated with caution. Unexpected findings from individual research studies should be independently replicated before they are translated into clinical practice.
Meanwhile, the most effective measures to reduce the burden of lung cancer remain quitting smoking and reducing exposure to air pollution.
(Dr. Rajeev Jayadevan is convener, research cell, Kerala State IMA and honorary senior consultant gastroenterologist, Sunrise Hospital Cochin. rajeevjayadevan@gmail.com)
Published – June 13, 2026 05:34 pm IST