Referee 2's Review:

The authors examine hippocampal shape in late-life depression using 3 T MRI. This is an interesting topic that indeed requires further investigation. The MRI technique including image acquisition and data analysis methods are state of the art. The technical MRI part of the paper is up to date and well done.

However I have major concerns with the patient selection and assessment of patient-dependant covariates. In my opinion the clinical assessment has been carried out in a sloppy and negligent way, limiting the psychiatric and neuroscientific statement dramatically. The authors should elaborate on their neuroscientific limitations and submit to a more technically oriented neuroimaging journal.

Major concerns:

The cognitive assessment using only MMSE is very poor. Considering the age range of subjects included it is very likely that this sample contains patients with mild cognitive impairment, or even incipient dementias. The effects of these diseases on hippocampal shape and volume are much more pronounced than those effects reported in major depression.

The authors do not mention how the excluded relevant comorbid diseases. They mention that the NIMH Diagnostic Interview Schedule assesses major depression, sleep problems, melancholia, psychosis, mania, substance abuse and dependence. However it is unclear if they specifically screened for post traumatic stress disorder (PTSD) which has high comorbidity with major depression and huge influences on hippocampal shape and volume.

Further, life events that significantly influence hippocampal shape and volume have not been assessed. Disease characteristics such as number of depressive episodes and any treatment details including treatment time, medication type and dosage that have relevant impact on hippocampal volume are not considered.

The depressed and nondepressed groups are not well matched, containing different numbers of individuals with a significant age difference. Even if age is introduced as covariate, nonlinear age effects on total brain volume may introduce a systematic error.

The authors elaborate on the resemblance of hippocampal atrophy in this sample with findings in dementia of Alzheimer's type and infer that there might be an association of these diseases. However their finding might also suggest a contamination of the study group by MCI or incipient DAT patients due to insufficient screening methods.

Minor concerns:

Please clarify if psychiatric diagnosis meeting exclusion criteria refers to lifetime diagnosis or only to current diagnosis.

Not correcting for multiple comparisons weakens the results, although this is commonly used in neuroimaging with these sample sizes.

Concerning the method of defining the hippocampal perimeter, reference 33 points towards an article on caudate volume measurement.

**********
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.