Breast MRI audits should calculate performance measures separately
Comprehensive audits of breast MRI exams should include separate performance benchmarks for screening and diagnostic breast MRI exams, according to a study published online April 29 by the Journal of the American College of Radiology.
Although the Mammography Quality Standards Act requires an audit of mammography practices, no formal requirements exist for screening or diagnostic MRI. Some performance measures like abnormal interpretation rate, cancer detection rate and positive predictive values have been reported, but they are often done so individually, only for screening exams, or for an aggregate cohort.
“Because performance measures are affected by the prevalence of cancer, calculations from an aggregate screening and diagnostic cohort should be interpreted with caution,” wrote the study’s lead author, Bethany L. Niell, MD, PhD, of Massachusetts General Hospital in Boston, and colleagues. The researchers evaluated breast MRI performance measures that were stratified by screening and diagnostic indications from a single academic institution.
After retrospectively reviewing an institutional database that identified all breast MRI exams performed between 2007 and 2008, Niell et al calculated the following performance measures for screening and diagnostic indications: cancer detection rate, positive predictive values, and abnormal interpretation rates.
The study included 2,444 examinations, 1,313 that were for screening and 1,131 for diagnostic indications. The study’s authors found that the cancer detection rate was 14 per 1,000 for screening breast MRI exams and 47 per 1,000 diagnostic examinations. The abnormal interpretation rate was 12 percent for screening and 17 percent for diagnostic indications.
The positive predictive values of MRI were lower for screening in comparison with diagnostic indications. Positive predictive value 1 (cancers detected after positive results on MRI) was 12 percent for screening, 28 percent for diagnostic indications and 21 percent overall. Positive predictive value 2 (cancers detected after biopsy recommended) was 24 percent for screening, 36 percent for diagnostic indications and 32 percent. Positive predictive value 3 (cancers detected after biopsy performed) was 27 percent for screening, 38 percent for diagnostic indications and 35 percent overall.
“Our results of significantly different MRI performance for clinical and diagnostic indications suggests that stratified, rather than aggregate, analyses of performance measures will provide more accurate performance estimates to guide the application of MRI in clinical practice,” Niell and colleagues concluded.