MR Spectroscopy After MRI Cuts Down on Benign Biopsies

September 25, 2007

NEW YORK -- Proton magnetic resonance (MR) spectroscopy may eliminate 68% of biopsies for benign non-mass breast lesions found on standard MR imaging (MRI) without missing any cancers.

NEW YORK, Sept. 25 -- Proton magnetic resonance (MR) spectroscopy may eliminate 68% of biopsies for benign nonmass breast lesions found on standard MR imaging (MRI) without missing any cancers.

These findings of a small study suggest the "quick and well tolerated" imaging technique may be a clinically useful supplement to MRI, said Lia Bartella, M.D., of Memorial Sloan-Kettering Cancer Center here, and colleagues in the October issue of Radiology.

MRI has become more common in breast cancer screening for high-risk women because it finds more abnormalities than mammography. But about 80% of breast lesions biopsied because of MRI findings are found to be benign.

MR spectroscopy has been suggested as an adjunct to breast MRI because it can be used to detect elevated levels of choline compounds, which indicate an active tumor, in a suspicious lesion.

Lesions that are not masses or lumps but enhance on MRI often require biopsy because they can reflect either benign hormonal changes or cancer.

However, non-mass lesions typically include more fat tissue than mass lesions, which could cause problems for interpretation of proton MR spectroscopy.

So, the researchers prospectively evaluated MR spectroscopy performance compared with histology. The study included 32 consecutive women (mean age 48.5) with a suspicious (78%) or biopsy-proven malignant (22%) enhancing non-mass lesion of 1 cm or larger at MRI.

The initial diagnostic MRI was done to screen women at high risk for breast cancer in 41% of cases, to evaluate disease extent in 50%, and for problem solving in 9%.

MR spectroscopy took 10 to 20 minutes, "important if proton MR spectroscopy is to be included in a clinical protocol for the diagnosis of breast cancer." Findings were considered positive if the choline resonance peak signal-to-noise ratio was at least two.

Histology after MR spectroscopy showed 12 of the 32 non-mass lesions (37%) to be cancerous.

Four lesions were considered high risk. These were lobular carcinoma in situ, atypical lobular hyperplasia, atypical ductal hyperplasia, and chronic inflammation with atypia. The remaining lesions were benign on histology and included fibrocystic changes, fat necrosis, and ductal hyperplasia.

MR spectroscopy was positive for 15 (47%) non-mass lesions. This included all 12 malignant lesions, of which two were ductal carcinoma in situ, for a sensitivity of 100% (95% confidence interval 74% to 100%).

MR spectroscopy findings correctly identified as negative 17 of 20 noncancerous non-mass lesions for a specificity of 85% (95% CI 62% to 97%).

Overall, MR spectroscopy had a positive predictive value of 80% (95% CI

Excluding lesions that were biopsy-proven to be cancer at the initial MRI, 25 non-mass lesions were recommended for biopsy. Among these, MRI had a positive predictive value of 20%.

Adding MR spectroscopy significantly increased the positive predictive value to 63% for selecting women for biopsy (P