AI-powered blood test detects lung cancer with DNA analysis
Stanford researchers have developed a way to screen for lung cancer by combining next-gen molecular DNA quantification with machine learning. The experimental technique requires only a blood test and could, with refinement, replace low-dose CT scanning as an initial exam for longtime smokers.
While development is early, the new way—called LungCLiP for Lung Cancer Likelihood in Plasma—identified stage-1 lung cancer in some 63% of patients with known disease, in 69% of those with stage-2 disease and in 75% of those in stage 3.
Specificity was a respectable 80%, although that figure would give 20% of patients false positives, which can lead to undue anxiety and costly downstream procedures.
Reporting their work in Nature, senior study author Maximilian Diehn, MD, PhD, and colleagues note they trained Lung-CLiP using samples from a discovery cohort of 104 patients with early-stage non-small-cell lung cancer and 56 risk-matched controls undergoing annual radiologic screening for lung cancer at four cancer centers.
They used a multi-tiered machine-learning approach, training a model to estimate the probability that a DNA mutation is tumor-derived.
On the upside of the cost considerations, LungCLiP’s performance was similar to that of tumor-informed DNA analysis and can be achieved without the need for tissue genotyping.
The latter entails tests to find any changes in specific genes or chromosomes, along with tumor-panel tests looking at multiple genes simultaneously. These tests can cost from $300 to more than $10,000, often with a high out-of-pocket burden for patients, according to the American Cancer Society.
LungCLiP’s greatest promise lies in its potential to serve as an initial screen for “some of the approximately 95% of high-risk patients in the U.S. who, despite being candidates for low-dose CT, are not being screened for various reasons, including limited access and concerns with false positives,” write senior study author Maximilian Diehn, MD, PhD, and colleagues.
Patients with positive Lung-CLiP tests would then be referred for low-dose CT, the authors add.
“Although Lung-CLiP is less sensitive than low-dose CT, this hybrid approach could potentially increase the total number of patients screened, and therefore the number of lives saved annually in the U.S., from the current annual value of around 600 to closer to the projected maximum of around 12,000” lives.
The study is available in full for free.
Dave P. has worked in journalism, marketing and public relations for more than 30 years, frequently concentrating on hospitals, healthcare technology and Catholic communications. He has also specialized in fundraising communications, ghostwriting for CEOs of local, national and global charities, nonprofits and foundations.