In the latest Lungcast, a pair of experts considered the marriage of low-dose CT scanning and machine learning to optimize lung disease interception.
Utility of low-dose computed tomography (CT) scanning for chronic lung disease has been historically low, despite recent upticks in messaging and clinical guidance informing symptom- and biomarker-based opportunities to screen.
But clinicians are beginning to uncover a greater utility of the practice for when it is applied—with goals that mirror the ambition of cardiovascular disease biomarker discovery and proceeding clinical research.
In the latest episode of Lungcast, host Albert Rizzo, MD, chief medical officer of the American Lung Association (ALA), discussed the varied benefit of CT scanning with Ravi Kalhan, MD, MS, deputy division chief of pulmonary and critical care medicine at the Northwestern Feinberg School of Medicine.
“We know that many of the low-dose CT scans done for lung cancer screening are finding evidence of emphysema in individuals without any respiratory symptoms, but certainly at risk in view of their smoking history,” Rizzo noted.
When asked whether it could help reference clinicians toward more intervention opportunities, Kalhan said he believes low-dose CT scanning is “part of the solution here.”
“We can detect early pulmonary fibrosis on those same CT scans,” Kalhan said. “Smoking is a risk factor for pulmonary fibrosis, which is shared with lung cancer—and, early application of antifibrotic treatment in patients with idiopathic pulmonary fibrosis at least is an important therapy. That’s an under-effected disease where early therapy probably helps.”
Beyond the untapped opportunity for more timely disease diagnosis and treatment, there is a “novel” benefit to improved CT scanning observed by Kalhan and Rizzo’s colleagues. The utility of novel machine learning methods on CT scans can help clinicians detect the earliest forms of lung injury.
“So, someone gets a screening CT and if they’re fortunate, they don’t have any emphysema or fibrosis nodules,” Kalhan said. “But if we could apply a novel algorithm that assessed regionally whether they have lung that is injured or is susceptible to future problems, then that is perhaps a group we could target for interception.”
Kalhan likened this strategy to longitudinal cohort cardiovascular screening research that has helped identify risk factors and biomarkers for future heart disease; the hope is a combined utility of new machine learning capability and improved screening could provide the same for respiratory disease.
Kalhan and Rizzo additionally discussed the community-based Lung Health Cohort, initiated about 18 months ago. The cohort study is assessing patients between ages 25 – 35 years old—in their perceived peak period of respiratory health—and assessing both pulmonary function status as well as behavioral status including exercise, smoking and vaping use, and their environmental factors.
Kalhan and colleagues are hoping to observe the progression of a person’s pulmonary function status toward outcomes on CT scans and how their lifelong respiratory health plays out.
The ultimate goal is to optimize lung disease interception—“determine what could be the cholesterol for the lung…figure that out, then create a platform where we could actually test therapies that could actually stop the progression toward chronic lung disease,” Kalhan said.
Lungcast is a monthly respiratory health podcast series from the ALA produced by HCPLive.
Subscribe or listen to Lungcast on your favorite platforms: