Treatment Course With Antidepressant Therapy in Late-Life Depression

To the Editor: We read with great interest the article by Sheline et al. (1) in the November 2012 issue of the Journal on the prediction of antidepressant response based on the volume of the hippocampus on MRI. This is a very exciting field that has attracted significant attention in the medical community and from patients. The idea that MRI could finally provide anatomical evidence for some mental disorders could have a significant impact on the general perception and acceptance of some of these diseases. Much progress has been made in this field, but some technological limitations must be considered before getting overexcited about the results of research using MRI quantification.

The resolution of the state-of-the-art MRI scanners is in the range of 0.8 mm, which is excellent. MR imaging data are mainly segmented based on the similarity of signal intensity of neighboring voxels or by selecting regions of interest.

We would like to highlight an important technical limitation that should be considered before accepting these results as conclusive. Sheline et al. observed an average hippocampal volume of 8,298.2 mm³ in the group of patients who achieved remission from depression and 7,942.3 mm³ in those who did not achieve remission after treatment. The values are provided in volume (mm³), and although the volumetric difference seems large, the difference represents less than 0.3 mm on each dimension of the measured volume. The cubic root of 8,298.2 mm³ is 20.24 mm and the cubic root of 7,942.3 mm³ is 19.95 mm, so the approximate difference in each dimension measured of the volume of hippocampus is in the range of 0.3 mm, which is below the resolution of MRI.

We would kindly suggest that the authors check the precision and accuracy of the segmentation process provided by MRI volumetric quantification on the scanner and workstation chosen for the study before we could accept their results as conclusive.