Machine learning finds rate of change—not value of ovarian cancer biomarker—indicates recurrence
Researchers utilized a machine learning algorithm to determine that a higher rate of change—rather than actual value of cancer antigen 125 (CA125)—is associated with abdominal recurrence of ovarian cancer. Findings may help identify patients most likely to benefit from imaging surveillance of the disease.
“The rate of change of CA125 is more predictive of abdominal recurrence than actual CA125 value or absolute or relative change in CA125 level,” authors wrote in the study published online May 19 in the Journal of the American College of Radiology. “This small hypothesis-generating study demonstrates that the rate of change of CA125 may potentially help in optimizing the use of abdominopelvic CT in surveillance of patients with ovarian cancer.”
Current National Comprehensive Care Network (NCCN) guidelines recommend using the CA125 biomarker with imaging to detect cancer recurrence. But the relationship between CA125 levels and detection of disease via imaging is unknown, according to first author Atul B. Shinagare, MD, with Brigham and Women’s Hospital’s Department of Radiology in Boston, and colleagues.
The study included 57 patients who had advanced high-grade ovarian cancer, underwent cytoreductive surgery and received surveillance via abdominopelvic CT and corresponding CA125 levels.
Radiologists assessed four measures of CA125: actual CA125 levels at the time of CT, absolute change since prior CT, relative change since prior CT and rate of change since prior CT.
A machine learning algorithm found the rate of change in CA125 to be the most predictive measure of abdominal recurrence. It was “significantly” higher preceding CT imaging showing abdominal recurrence.
The group also found that CA125 rises more rapidly before CT studies that showed abdominal recurrent disease. A faster rate of CA125 increase was greatly associated with recurrence.
Shinagare et al. noted their study had many limitations. The two most impactful being its small sample size and retrospective nature.
Authors noted their results would require larger, prospective studies for validation, but their impact on making imaging-related decisions on the basis of CA125 levels may prove to be invaluable for patients.
Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.