Specificity of choline metabolites for in vivo diagnosis of breast cancer using ¹H MRS at 1.5 T

The purpose was to determine if in vivo proton magnetic resonance spectroscopy (¹H MRS) at 1.5 T can accurately provide the correct pathology of breast disease. Forty-three asymptomatic volunteers including three lactating mothers were examined and compared with 21 breast cancer patients. Examinations were undertaken at 1.5 T using a purpose-built transmit-receive single breast coil. Single voxel spectroscopy was undertaken using echo times of 135 and 350 ms. The broad composite resonance at 3.2 ppm, which includes contributions from choline, phosphocholine (PC), glycerophosphocholine (GPC), myo-inositol and taurine, was found not to be a unique marker for malignancy providing a diagnostic sensitivity and specificity of 80.0 and 86.0%, respectively. This was due to three of the asymptomatic volunteers and all of the lactating mothers also generating the broad composite resonance at 3.2 ppm. Optimised post-acquisitional processing of the spectra resolved a resonance at 3.22 ppm, consistent with PC, in patients with cancer. In contrast the spectra recorded for three false-positive volunteers, and the three lactating mothers had a resonance centred at 3.28 ppm (possibly taurine, myo-inositol or GPC). This improved the specificity of the test to 100%. Careful referencing of the spectra and post-acquisitional processing intended to optimise spectral resolution of in vivo MR proton spectra from human breast tissue resolves the composite choline resonance. This allows the distinction of patients with malignant disease from volunteers with a sensitivity of 80% and specificity of 100%. Therefore, resolution of the composite choline resonance into its constituent components improves the specificity of the in vivo ¹H MRS method, but does not overcome the problem of 20% false-negatives.