Clinical utility of proton magnetic resonance spectroscopy in characterizing breast lesions.

Proton magnetic resonance spectroscopy ((1)H MRS) of the breast has been proposed as an adjunct to the magnetic resonance imaging (MRI) examination to improve the specificity of distinguishing malignant breast tumors from benign breast tumors. In this review, we carry out a pooled analysis of the clinical breast (1)H MRS studies undertaken to date to determine the factors that influence the diagnostic performance of this method. In total, five studies of breast (1)H MRS from four independent centers around the world have been published to date. Altogether, 153 tumors were examined, 100 of which were confirmed histologically to be malignant and 53 of which were benign. The lesions presenting a detectable composite choline signal in their corresponding (1)H MR spectra were diagnosed as malignant, whereas the lesions with no choline signal were diagnosed as benign. The sensitivity and specificity of breast (1)H MRS for detecting breast cancer were 83% (95% confidence interval [CI] = 73% to 89%) and 85% (95% CI = 71% to 93%), respectively, and both values could be as high as 92% after technical exclusions. In a subgroup of 20 young women, the sensitivity and the specificity of the method approached 100%. The factors limiting the sensitivity of the examination were mainly technical. The use of the composite choline signal as a marker for malignancy in breast (1)H MRS is a robust method with highly reliable interpretation, because it is based on the appearance of a single peak. The method is likely to provide even better results with technologic advances in breast MRS that lead to the improved detection of the composite choline signal.