Choline elevation in amygdala region at recovery indicates longer survival without depressive episode: a magnetic resonance spectroscopy study.

Rationale: Depression, with variable longitudinal patterns, recurs in one third of patients. We lack useful predictors of its course/outcome, and proton magnetic resonance spectroscopy (1H-MRS) of brain metabolites is an underused research modality in finding outcome correlates.

Objectives: To determine if brain metabolite levels/changes in the amygdala region observed early in the recovery phase indicate depression recurrence risk in patients receiving maintenance therapy.

Methods: Forty-eight patients on stable-dose antidepressant (AD) maintenance therapy were analyzed from recovery onset until (i) recurrence of depression or (ii) start of AD discontinuation. Two 1H-MRS scans (6 months apart) were performed with a focus on amygdala at the beginning of recovery. N-acetylaspartate (NAA), choline-containing metabolites (Cho), and Glx (glutamine/glutamate and GABA) were evaluated with regard to time without recurrence, and risks were assessed by Cox proportional hazard modeling.

Results: Twenty patients had depression recurrence, and 23 patients reached AD discontinuation. General linear model repeated measures analysis displayed three-way interaction of measurement time, metabolite level, and recurrence on maintenance therapy, in a multivariate test, Wilks' lambda = 0.857, F(2,40) = 3.348, p = 0.045. Cho levels at the beginning of recovery and subsequent changes convey the highest risk for earlier recurrence. Patients experiencing higher amygdala Cho after recovery are at a significantly lower risk for depression recurrence (hazard ratio = 0.32; 95% confidence interval 0.13-0.77).

Conclusion: Cho levels/changes in the amygdala early in the recovery phase correlate with clinical outcome. In the absence of major NAA fluctuations, changes in Cho and Glx may suggest a shift towards reduction in (previously increased) glutamatergic neurotransmission. Investigation of a larger sample with greater sampling frequency is needed to confirm the possible predictive role of metabolite changes in the amygdala region early in the recovery phase.