MRI breast ca staging provided benefits, shortcomings

A meta-analysis by Australian and European researchers indicates that MR staging identifies additional disease in nearly one of five women previously diagnosed with breast cancer. It also suggests that women may undergo more extensive surgeries than originally planned because of false-positive MR findings.

Evidence of MRI’s value as an adjunct to mammography and ultrasound for the management of patients with breast cancer continues to accumulate.

The medical literature also shows, however, that many women with localized disease undergo mastectomy or even more extensive surgical resection than initially planned based on MRI staging.

Adequate informed consent for this special population of women is essential because preop MRI may mean further studies and delays to definitive treatment in some cases, said principal investigator Nehmat Houssami, MD, a medical epidemiologist at the University of Sydney.

The trade-off between true-positive and false-positive detection, however, remains the most significant issue. “All women should be told of this before having MR,” Dr. Houssami said.

Methods

Dr. Houssami and colleagues in Australia, Italy, and the UK reviewed data from 19 studies including 2,610 women previously diagnosed with breast cancer.

The group assessed MRI’s accuracy in the detection of multifocal and multicentric disease missed by conventional imaging and correlated study results with the quality of the diagnostic standard of reference used. They also estimated the proportion of women whose surgery plan changed due to staging results.

The investigators found that MRI staging detected additional disease in 16% of patients, leading to changes in surgical treatment ranging from local excision to mastectomy to more extensive surgeries. Nearly half of these procedures, however, were based on false-positive findings. Researchers published their findings in the Journal of Clinical Oncology (26: 3248-3258, 2008).

The incremental accuracy of MRI staging fell from 99% to 86% as the quality of the reference standard increased. The correlation was statistically significant (P = .016). The true-positive to true-negative ratio was 1.9 to 1. Slightly more than 8% of patients converted from wide local excision to mastectomy, and 11.3% went from wide local excision to more extensive surgery.

In women with MRI-detected lesions who did not have additional malignancy on histology, conversion from local excision to mastectomy and from local excision to more extensive surgery was 1.1% and 5.5%, respectively.

All evidence in this context comes from observational studies only, according to Dr. Houssami. More research is needed to determine whether the MR-based management plan is truly beneficial or unnecessary.

Prospective multicenter randomized trials in women with a new breast cancer diagnosis could provide answers.

“We need randomized trials to truly assess the implications of preop MR on longer-term outcomes in women with breast cancer,” Dr. Houssami said.

He emphasized that “this is an area where imaging research must look at clinical endpoints beyond detection capability of a new imaging test.”

Future studies

MR improvements or the incorporation of complementary technologies such as MR spectroscopy can help false-positive results.

Dr. Houssami said refinement and standardization of interpretation criteria for breast MRI might also help radiologists distinguish between benign lesions and malignant findings.

False positives for surgical conversion are a serious issue, and they should be avoided by ensuring that all significant MR-detected findings get confirmed by needle biopsy before any surgical intervention is planned or carried out, Dr. Houssami said.

The meta-analysis also identified methodological issues, namely the effect of the reference standard.

This is a finding that radiologists and researchers need to consider in the design of future studies, Dr. Houssami said.

This article is adapted from ONI’s sister publication Diagnostic Imaging online (July 2008).