Hippocampal volume is reduced in schizophrenia and schizoaffective disorder but not in psychotic bipolar I disorder demonstrated by both manual tracing and automated parcellation (FreeSurfer)

Abstract

This study examined hippocampal volume as a putative biomarker for psychotic illness in the Bipolar-Schizophrenia Network on Intermediate Phenotypes (B-SNIP) psychosis sample, contrasting manual tracing and semiautomated (FreeSurfer) region-of-interest outcomes. The study sample (n = 596) included probands with schizophrenia (SZ, n = 71), schizoaffective disorder (SAD, n = 70), and psychotic bipolar I disorder (BDP, n = 86); their first-degree relatives (SZ-Rel, n = 74; SAD-Rel, n = 62; BDP-Rel, n = 88); and healthy controls (HC, n = 145). Hippocampal volumes were derived from 3Tesla T1-weighted MPRAGE images using manual tracing/3DSlicer3.6.3 and semiautomated parcellation/FreeSurfer5.1,64bit. Volumetric outcomes from both methodologies were contrasted in HC and probands and relatives across the 3 diagnoses, using mixed-effect regression models (SAS9.3 Proc MIXED); Pearson correlations between manual tracing and FreeSurfer outcomes were computed. SZ (P = .0007-.02) and SAD (P = .003-.14) had lower hippocampal volumes compared with HC, whereas BDP showed normal volumes bilaterally (P = .18-.55). All relative groups had hippocampal volumes not different from controls (P = .12-.97) and higher than those observed in probands (P = .003-.09), except for FreeSurfer measures in bipolar probands vs relatives (P = .64-.99). Outcomes from manual tracing and FreeSurfer showed direct, moderate to strong, correlations (r = .51-.73, P < .05). These findings from a large psychosis sample support decreased hippocampal volume as a putative biomarker for schizophrenia and schizoaffective disorder, but not for psychotic bipolar I disorder, and may reflect a cumulative effect of divergent primary disease processes and/or lifetime medication use. Manual tracing and semiautomated parcellation regional volumetric approaches may provide useful outcomes for defining measurable biomarkers underlying severe mental illness.

Keywords: FreeSurfer; hippocampus; manual tracing; psychotic bipolar disorder; schizophrenia.

Fig. 1.

Hippocampal volume outcomes from manual tracing and FreeSurfer in probands, relatives, and healthy controls. The scatter plots show individual hippocampal volumes derived from manual tracing and FreeSurfer in each proband, relative, or HC subject. The horizontal bars indicate group means and standard deviations. SZP, probands with schizophrenia; SADP, probands with schizoaffective disorder; BDP, probands with psychotic bipolar I disorder; SZR, relatives of probands with schizophrenia; SADR, relatives of probands with schizoaffective disorder; BDR, relatives of probands with psychotic bipolar I disorder; HC, healthy controls; L Hipp, the left hippocampus; R Hipp, the right hippocampus; MT, manual tracing; FS, FreeSurfer.

Fig. 2.

Pearson correlations between the hippocampal volume outcomes from the 2 ROI methodologies–manual tracing and FreeSurfer–in all study groups combined. Manual tracing yielded lower bilateral hippocampal volumes than FreeSurfer: left, manual tracing volume = 2721.38mm³, FreeSurfer = 3970.78mm³, 31.46% difference; right, manual tracing volume = 2645.51mm³, FreeSurfer = 4043.89mm³, 34.58% difference (all volumes are averaged across all study groups). Correlation coefficients (r) in: (1) All groups combined: left, r = .66; right, r = .63; (2) probands: SZ, left, r = .73; right, r = .63; SAD, left, r = .66; right, r = .74; BDP, left, r = .65; right, r = .64; (3) relatives: SZ-Rel, left, r = .68; right, r = .69; SZD-Rel, left, r = .6; right, r = .55; BDP-Rel, left, r = .72; right, r = .71; and (4) HC: left, r = .57; right, r = .51. All correlations are statistically significant at P < .05. SZ, probands with schizophrenia; SAD, probands with schizoaffective disorder; BDP, probands with psychotic bipolar I disorder; SZ-Rel, relatives of probands with schizophrenia; SAD-Rel, relatives of probands with schizoaffective disorder; BDP-Rel, relatives of probands with psychotic bipolar I disorder.