In addition to being used as a diagnostic tool, neuroimaging is also regularly employed for research and commercial purposes. When a research volunteer undergoes a brain MRI, their scan might reveal an “incidental finding.” This term describes unknown abnormalities of potential clinical importance that are discovered incidentally during a research examination.
A recent paper in the American Journal of Neuroradiology, “Incidental Findings from 16,400 Brain MRI Examinations of Research Volunteers” aims to determine how frequently this happens – and examine whether scan review processes should be amended in the interest of ethics.
Typically, a “flag and refer” scan review process is used in MRI research studies. In this workflow, a nonradiologist (usually an MRI technician or member of the research team) will review each subject’s scans and “flag” those that seem unusual. These cases are then uploaded to a reading queue of neuroradiologists.
“The driving question, for us, was ‘should we be looking at every scan?’,” says Howard Rowley, MD. The existing system tasks nonradiologists with screening and referring for further review, leaving most scans without formal interpretation.
The project began in 2002, when a cluster of suspected abnormalities were identified in research subjects. This sparked discussion about the existing workflow and prompted research into how the “flag and refer” approach would compare with a “read every scan” approach.
To Dr. Rowley, the project wasn’t merely academic.
“Over the years, I was anecdotally aware of subjects that had had volunteer MRIs and later developed serious brain conditions, including brain tumors, who should’ve been identified at the time of their initial scan and weren’t,” he says. “So, for me, the project was personal, because I knew people who could have benefited from having that early notice.”
The researchers used 16,400 MRI scans, all collected at UW-Madison between 2002 and 2020. Neuroradiologists read each of these scans, and found that 4 percent had an incidental finding that was concerning enough to need follow-up. Nonradiologists flagged less than 1 percent of the MRIs that warranted further evaluation.
Routine neuroradiologist review yielded a much higher rate of detection than selective referral. This result suggests that the typical scan review process may need to be amended – or at least, carefully considered – when designing research protocols.
“In my opinion, every research volunteer should have their scan read by a neuroradiologist,” Dr. Rowley adds. “It allows the detection of serious but potentially treatable conditions, which we know is occurring in around 4 percent of participants.”
At UW, the general approach now is to have neuroradiologists read every scan.
The article is authored by a number of past and present UW-Radiology members, including Laura Eisenmenger, MD, Aaron Field, MD, PhD, Vivek Prabhakaran, MD, PhD, and Paul Rowley, MD, Dr. Rowley’s son, who worked on the project while attending medical school at the University of Wisconsin-Madison.
By confirming the baseline rate of incidental findings in brain MRIs, the article will inform future decisions – both at UW and beyond – about how to design ethical workflows in research studies.