Abstract

PURPOSE: To quantify lithium in the human brain.
METHODS: A 7Li MR spectroscopy method was developed with special features for high precision including: a) sampling a large cerebral volume to maximize the signal-to-noise ratio; b) adiabatic excitation pulses to ensure uniform spin mutation; c) morphometric analysis of the MR images of the sampled cerebrum; d) a mathematical model derived from empirical data to correct for receiver inhomogeneity effects; and e) a long interpulse delay, to eliminate errors arising from uncertain T1 values.
RESULTS: A theoretical precision of 5.2% and an accuracy of better than 7.2% in someone with a brain lithium level of 1.0 mEq per liter of cerebral volume and precision and accuracy of 6.8 and 8.6%, respectively, in someone with 0.5 mEq/L brain lithium was calculated. This level of precision was surpassed in phantoms and patients. Brain lithium in 10 patients treated with lithium carbonate varied from 0.52 to 0.87 mEq/L (mean = 0.58 mEq/L; SD = 0.17 mEq/L). Brain-to-serum lithium ratios varied from 0.50 to 0.97 mEq/L (mean = 0.77 mEq/L; SD = 0.14 mEq/L). Substantial variation in brain lithium was observed in patients with similar serum lithium.
CONCLUSIONS: A highly reliable method to quantify human brain lithium by 7Li MR spectroscopy has been implemented. Unexpected variability in brain versus serum levels of lithium was detected in patients with bipolar disease.