MRI can be a useful tool for taking a second look at mammograms that are considered inconclusive in a process known as “problem-solving MRI," but radiologists should be cautious about the possibility of false negatives, according to a study published in the American Journal of Roentgenology.

The imaging community is in agreement for using breast MRI in screening women with an elevated risk of cancer, usually determined by family history or the presence of BRCA genetic mutations. However, there’s no consensus on the efficacy of MRI in “problem solving” MRI. Small studies have placed the malignancy rate between 5.2 and 26.3 percent but no large-scale review has taken place, so researchers from Brigham and Women’s Hospital and Massachusetts General Hospital in Boston combed through four years of breast MRI to determine the usefulness of the problem-solving variety.

Out of over 7,000 total MRI exams, just 294 were classified as problem-solving. While the BI-RADS classification differed from patient to patient, all were recommended for additional MRI by the radiologist who read the initial mammogram. Most of these had no ultrasound correlation, while 45 percent had MRI correlation, leading to 74 biopsies. Over half of those biopsies were malignant, for a total malignancy rate of 13.6 percent over the entire series—well within the range experienced by previous studies.

While the authors said it can be a useful tool for the evaluation and management of inconclusive mammography findings, they cautioned radiologists about the relative cost of these MRI exams.

“The possibility that incidental detection of ultimately benign MRI lesions would result in MRI surveillance or biopsy is not negligible to the patient or to the healthcare system,” they wrote. “Therefore, radiologists must be thoughtful in the use of this modality to maximize timely cancer diagnosis while minimizing the rate of false positive results.”

In addition, the rise of digital breast tomosynthesis (DBT) may reduce the amount of inconclusive mammograms, according to the authors, and several states have introduced legislation mandating insurers cover the exam.

“As DBT evolves, the frequency of non-calcified equivocal mammographic findings requiring MRI evaluation should decline,” wrote the authors. “Real one-view findings may be localizable in the DBT dataset, and use of DBT diminishes the likelihood of spurious one-view findings due to overlapping tissue.”