LUNG CANCER CAUSES FAR MORE DEATHS THAN ANY OTHER FORM OF CANCER, and most of those casualties result from tardy diagnoses. In 2011, a study with more than 50,000 participants showed that early screening could reduce death in people most at risk—heavy smokers—by as much as 20%. So for the past five years, physicians have been ordering an increasing number of chest CT scans for early detection.

But those scans often reveal small nodules in the lung—tiny growths that are benign most of the time. Doctors have no way of knowing whether a nodule is safe or cancerous unless they perform a biopsy, most often by snaking a fiber optic cable through the nose and mouth to reach the nodule and take a small piece of it.

That procedure can cause bleeding or complications in sensitive tissue. “And often the bronchoscope can’t get all the way out to the periphery of the lung, where many of these nodules arise,” says Avrum Spira, director of the Translational Bioinformatics Program at Boston University School of Medicine. “So about half the time, the procedure can’t give a diagnosis.”

What’s needed are better ways to screen. Two types of blood test look promising—one searches for tumor cells circulating in the blood, while the second detects other biomarkers that cancer can shed. “But these techniques may not be sensitive enough in people who have very small or early cancers, like the kind you might find with CT scan screening,” says Anil Vachani, associate professor of medicine at the University of Pennsylvania, whose lab is exploring diagnostic markers for lung cancer.

So Spira took a different tack. A couple of years ago, he began analyzing cells of the windpipe, which connects the lungs to the world outside. “The concept is called ‘field of injury,’ ” explains Spira. “The cells in the windpipe, exposed to smoke and pollution, will have genomic alterations. Those should be able to tell you whether there is a high risk that you have a cancer deep in your lung.” Of the hundreds of genes whose activity was changed by exposure to smoke, the researchers found about 20 whose activity could fairly reliably predict whether the patient would go on to develop lung cancer within one year. The research was published in the New England Journal of Medicine in 2015.

Since then, Spira’s team has also been looking at tissues of the nose to see whether exposure to cigarette smoke may cause genomic changes there. In one recent study, the researchers swabbed the noses of more than 500 patients who were already being evaluated for suspicious nodules, to see whether nasal cells could help predict cancer noninvasively.

The researchers measured gene expression in the nose cells of these patients and watched to see which patients developed lung cancer within one year. They identified 30 genes that correlated strongly with the presence of cancer.

Then they paired this information with a model developed by Michael K. Gould, a lung cancer expert, and his colleagues. The model offers a statistical way to predict lung cancer risk by looking only at the patient’s medical history and the size of nodules in the CT scan. “Adding information about gene expression to the Gould model allowed us to detect cancer noninvasively 90% of the time,” says Spira. The study was published February 27 in the Journal of the National Cancer Institute.

“The test has high negative predictive value—meaning that a negative test gives the doctor confidence that the patient does not have cancer,” says Spira. There are false positives, though—tests that point to cancer when it isn’t there—so a biopsy may be needed to follow up. But the test may be able to reduce the number of patients who undergo lung biopsies and are found not to have cancer.

And because the nose test is even less invasive than a windpipe test, it might easily be used even on those with only casual exposure to smoking or pollution. “It could serve as a molecular pap smear for lung cancer,” says Spira.

The test is also useful for those who have the gene changes but turn out not to have lung cancer. These patients are at the highest risk of developing the disease and may benefit from closer observation. Physicians might also give these patients one of several experimental drugs that could reduce their risk of developing lung cancer. Now in clinical trials, such drugs include zileuton, a medication often used to treat asthma, iloprost, now used to treat pulmonary hypertension, and common aspirin.

“Once we know who to watch,” Spira says, “these drugs and other interventions can start saving lives.”