Figure 1. Regions of reduced gray matter volume in the 54 men with schizotypal personality disorder compared with the 54 male healthy control subjects. The gray matter volume reductions (red) in the schizotypal personality disorder compared with healthy control subjects are shown using the brain template in SPM software. These regions include widespread gray matter volume reductions in the bilateral frontal, temporal, and parietal regions (see Table 3 for specific coordinates). An uncorrected threshold of P < .001 with an extent threshold of 30 voxels was used for graphical reporting. All the regions shown with volume reductions also satisfied the false discovery rate criterion with a corrected P < .05 (Table 3). A indicates anterior; L, left; P, posterior; R, right.
Figure 2. Relationships between the relative volumes of gray matter in the cortex (A) and left superior temporal gyrus (STG) (B) and age. The 54 men with schizotypal personality disorder (SPD) and 54 male healthy control (HC) subjects showed significant age-related gray matter volume reductions in the cortex (SPD: rho = −0.70, P < .001; HC: rho = −0.70, P < .001) and left STG (SPD: rho = −0.43, P = .001; HC: rho = −0.48, P < .001). However, there were no age by group interactions for these regions (cortex: F106 = 0.74, P = .77; left STG: F106 = 0.65, P = .86) (Table 4).
Figure 3. Correlation between regional gray matter volumes and Structured Interview for Schizotypy negative symptom scores (Table 5). ACG indicates anterior cingulate gyrus; OFC, orbitofrontal cortex; SFG, superior frontal gyrus; STG, superior temporal gyrus. The eFigure graphs this relationship for 21 additional regions.
Asami T, Whitford TJ, Bouix S, et al. Globally and locally reduced MRI gray matter volumes in neuroleptic-naïve men with schizotypal personality disorder: association with negative symptoms [published online February 6, 2013]. JAMA Psychiatry. doi:10.1001/jamapsychiatry.2013.665.
eFigure. Relationships between negative symptoms derived from the Structured Interview for Schizotypy and regional gray matter volumes
Asami T, Whitford TJ, Bouix S, Dickey CC, Niznikiewicz M, Shenton ME, Voglmaier MM, McCarley RW. Globally and Locally Reduced MRI Gray Matter Volumes in Neuroleptic-Naive Men With Schizotypal Personality DisorderAssociation With Negative Symptoms. JAMA Psychiatry. 2013;70(4):361–372. doi:10.1001/jamapsychiatry.2013.665
Author Affiliations: Laboratory of Neuroscience, Clinical Neuroscience Division, Department of Psychiatry, Boston Veterans Affairs Healthcare System, Brockton Division, Brockton (Drs Asami, Bouix, Dickey, Niznikiewicz, Shenton, Voglmaier, and McCarley); Psychiatry Neuroimaging Laboratory, Department of Psychiatry, Brigham and Women's Hospital, Boston (Drs Asami, Whitford, Bouix, Dickey, Niznikiewicz, and Shenton), Harvard Medical School, Massachusetts; and School of Psychology, University of New South Wales, Sydney, New South Wales, Australia (Dr Whitford).
Importance Some, but not all, previous magnetic resonance imaging studies have indicated smaller cortical and local gray matter volumes (GMVs) in men with schizotypal personality disorder (SPD) compared with healthy control (HC) subjects. Thus, there is need for a whole-brain comparison to resolve inconsistencies and provide hitherto generally absent data on the association between GMV and symptoms.
Objective To use voxel-based morphometry to evaluate a large sample of neuroleptic-naive men with SPD compared with group-matched HC subjects on local and global GMV and to identify associations with symptoms, especially negative symptoms. Also, to determine whether age-related GMV reductions are greater in men with SPD than HC subjects, providing presumptive evidence on possible progression.
Design, Setting, and Participants This naturalistic study involved 54 neuroleptic-naive men with SPD and 54 male HC subjects aged 18 to 55 years recruited from the community and scanned on the same 1.5-T GE magnetic resonance imaging scanner. Participants were group matched on age, socioeconomic status, handedness, and IQ.
Main Outcome Measures Cross-sectional voxel-based morphometry, GMV in subjects with SPD and HC participants, and the relationship to clinical symptoms.
Results A voxelwise analysis showed participants with SPD had significantly smaller GMV in the left superior temporal gyrus and widespread frontal, frontolimbic, and parietal regions compared with HC subjects. Most of these regional volumes were strikingly and significantly correlated with negative symptoms: the more the volume reduction, the more negative symptoms. Global cortical GMV and most regional GMV showed significant negative relationships with age in both those with SPD and HC subjects, without any group by age interactions.
Conclusions and Relevance Men with SPD showed global and widespread smaller regional GMV. The regional structural abnormalities were correlated with the severity of a participant's negative symptoms. While the pattern of GMV loss is similar to that in schizophrenia, the similar patterns of HC-SPD age-related GMV reduction suggest that SPD showed no progressive GMV loss, pointing to an important difference in the biological mechanisms of SPD and schizophrenia.
Schizotypal personality disorder (SPD) is a schizophrenia spectrum disorder with genetic, symptomatic, cognitive, and social characteristics similar to schizophrenia.1- 3 It is relatively common, with a lifetime prevalence estimated at 3.9%.4 Recent magnetic resonance imaging (MRI) studies have provided evidence of MRI brain similarities between SPD and schizophrenia (see articles on schizophrenia by McCarley et al5 and Shenton et al6). Schizotypal personality disorder MRI abnormalities compared with healthy control (HC) subjects include both global7,8 and regional volume reductions in gray matter of the neocortex (Table 1). Reports of regional gray matter volume (GMV) reductions in SPD include (1) temporal lobe superior temporal gyrus (STG),9,17,21,26,30,31 Heschl gyrus within the STG,10 middle temporal gyrus,8,29 fusiform gyrus,27 and hippocampus16,24; (2) frontal lobe superior and inferior frontal gyri,8,21 insula,20,21 and cingulate gyrus29,30; (3) parietal lobe8,28; and (4) basal ganglia.11,15,18 However, relative to schizophrenia, the number of MRI volumetric studies of SPD is smaller. Moreover, many of the studies looked at small regions of interest (ROIs), had findings that were sometimes inconsistent, and generally had a small number of participants, with variability in demographic factors that are known to influence brain volume such as age, sex, handedness, IQ, socioeconomic status (SES), and neuroleptic medication use.32- 42
Neuroleptic medications are of particular concern in light of nonhuman primate evidence suggesting that these medications can influence brain structure in and of themselves.43 Thus, a study of neuroleptic-naive subjects with SPD may provide a clearer understanding of brain abnormalities in the schizophrenia spectrum.
Another important variable in SPD is sex because previous studies have demonstrated important sex differences in SPD, including a greater rate of lifetime prevalence of SPD in men (4.2%) compared with women (3.7%).4 In terms of symptom clusters, males with SPDs are most different from females with SPDs in interpersonal relationships, having significantly fewer close relationships and also more prevalent negative symptoms associated with social deficits.44 Males with SPD also show a greater disturbance of verbal learning and abstraction than females with SPD.45 Not surprisingly, structural MRI studies have also shown sex differences on regional brain volumes in subjects with SPD,9,14,19,22,42 as well as in schizophrenia46- 48 and in HC subjects.32,34,35 These sex differences indicate that we must separately investigate structural brain abnormalities and their symptom associations in men and women with SPD to avoid a sex confound. In our previous study, a voxel-based morphometry (VBM) analysis demonstrated volumetric gray matter (GM) reductions in several brain regions, including the left STG in the neuroleptic-naive women with SPD compared with matched HC subjects8 (Table 1).
In terms of men with SPD, our previous ROI analyses demonstrated GMV reductions in the cortex (6.9% reduction)7 and left STG (9.2% reduction)9 compared with matched HC subjects. Although limited, these findings suggest that cortical brain abnormalities in men with SPD are widespread but are more pronounced in some brain regions.
For the purpose of identifying local GMV abnormalities in SPD throughout the brain, a whole-brain VBM method is useful. However, there are only a few VBM studies of SPD.20,21 Furthermore, to our knowledge, there have been no VBM studies that have investigated GMV reductions in neuroleptic-naive men with SPD and none providing clues as to reductions associated with negative symptoms, especially prominent in men.
The current whole-brain cross-sectional VBM analysis investigated volume differences in global and local GM in a large sample of 54 neuroleptic-naive men with SPD compared with 54 male HC subjects. To avoid potential confounds that could affect brain structure, the SPD and HC groups were carefully matched for sex, age, handedness, IQ, and the subject's own SES and parental SES (Table 2).Based on the previous studies (Table 1), we speculated that men with SPD would show GMV reductions in multiple cortical regions—in similar regions but to a smaller extent than the reductions previously reported in patients with established schizophrenia. Once volume differences were confirmed, differences in the patterns of age-related GMV changes were evaluated between the groups to investigate whether these 2 groups would show similar patterns of age-related volume changes. Finally, correlation analyses were conducted between GMVs and clinical measures to investigate the biological implications of structural abnormalities in the men with SPD. Given the predominance of negative symptoms in males,44 we hypothesized that negative symptoms would be closely associated with GMV reductions, particularly in frontal regions known to play a role in emotion regulation and regions associated with social cognition. We have found these to be reduced in schizophrenia and associated with negative symptoms.49
A total of 54 neuroleptic-naive men diagnosed as having DSM-III or DSM-IV SPD and 54 male control subjects were recruited from the community through advertisements. Of those, 6 male subjects with SPD and 14 male control subjects were used in our previous ROI study.7 Subjects with SPD were recruited via the following advertisement:
Sixth sense/very shy: A study at Harvard Medical School seeks right-handed people who believe they have ESP, telepathy, or a “sixth sense”; often mistake noises for voices; sense the presence of others when alone; have extreme social anxiety (or discomfort) in social situations involving unfamiliar people; and have few friends.
Sixth sense/very shy: A study at Harvard Medical School seeks right-handed people who believe they have ESP, telepathy, or a “sixth sense”; often mistake noises for voices; sense the presence of others when alone; have extreme social anxiety (or discomfort) in social situations involving unfamiliar people; and have few friends.
The SPD advertisement tapped the DSM-IV diagnostic criteria for SPD, which include (1) ideas of reference, (2) odd beliefs and superstitions or sixth sense, (3) abnormal perceptual experiences, (4) odd thinking and speech, (5) suspiciousness, (6) constricted affect, (7) odd and peculiar appearance or behavior, (8) no close friends, and (9) extreme social anxiety.
Of the 3001 individuals who responded to the SPD advertisement, 1536 male participants underwent an extensive telephone screening process that used the following inclusion criteria: (1) age between 18 and 55 years; (2) right handedness50; (3) English as the primary language; (4) no history of neurologic disorder or loss of consciousness longer than 5 minutes; (5) no history of electroconvulsive therapy, drug or alcohol dependence in the past 5 years, or abuse in the past year; and (6) no history of using neuroleptics at any time or psychotropic medications in the past year.
Of the 1536 subjects, 242 met the telephone inclusion criteria, including positive responses to at least 3 of the previously mentioned SPD criteria on screening questions. The Structured Clinical Interview for DSM-IV –Patient Edition (SCID)51 and its personality disorder version (SCID-II)52 were then used to make DSM-IV diagnoses, exclude Axis I psychotic and bipolar disorders from both groups, and exclude Axis I and Axis II diagnoses from control subjects. A total of 118 male subjects were diagnosed as having SPD and 65 men with SPD underwent MRI. The fact that fewer subjects had MRIs than were eligible was primarily owing to unavailability of the scanner—subject diagnostic characteristics did not play a role in selection. Finally, 54 men with SPD were chosen for this study, matched on demographic data with male HC participants.
We have previously described in detail the clinical and demographic characteristics of the subjects with SPD.8,44 Interviews were conducted by either a licensed psychiatrist (C.C.D. or R.W.M) or a licensed psychologist (M.M.V. or M.E.S.). Interrater reliability for the diagnosis of SPD was high (κ = 0.89; n = 25).9
Interviewers were trained to detect nuances of behavior and history and to ask follow-up questions to establish the correct diagnosis. In the rare instances in which the first interviewer was uncertain about the diagnosis, a second licensed psychiatrist or psychologist interviewed the subject, and a consensus was obtained. All 54 subjects with SPD who underwent MRI and other parts of the protocol met the full DSM-IV SPD diagnostic criteria (having ≥5 of the 9 characteristics). Subjects with SPD had a mean of 1.5 additional DSM-IV personality diagnoses, with the most common being paranoid (n = 19), avoidant (n = 16), obsessive-compulsive (n = 12), borderline (n = 10), and schizoid (n = 9) personality disorders. Fifteen men with SPD also met the criteria for Axis I disorders, including depression (n = 1); dysthymia (n = 9); generalized anxiety disorder (n = 4); and body dysmorphic disorder (n = 1).
Male control subjects were recruited from the community through a different advertisement and similarly underwent the SCID and SCID-II. Control participants had the additional inclusion requirement of no family history of psychotic or bipolar illness and no personal history of an Axis I or personality disorder diagnosis. Control subjects were group matched with subjects with SPD for demographic variables including age, Wechsler Adult Intelligence Scale–III IQ based on Vocabulary and Block Design subscales, handedness, and both the subject's own and parental SES.53 The project was approved by the institutional review boards at Harvard Medical School and Boston Veterans Affairs Healthcare System; after a full description of the study was provided to the participants, written informed consent was obtained.
Clinical symptoms were measured using the Structured Interview for Schizotypy (SIS).54,55 Use of the SIS began in the middle of subject recruitment; therefore, only 21 of the 54 men with SPD completed the SIS. Based on our previous results from SPD on the most prominent positive and negative SIS items in our population and the Kendler criteria, we calculated the positive symptom score by summing scores of (1) ideas of reference, (2) magical thinking, (3) illusions, and (4) psychoticlike symptoms, while the negative symptom score summed scores of (1) social isolation, (2) introversion, (3) restricted emotion, and (4) sensitivity.15,55 There were no significant demographic differences between the male subjects with SPD with and without SIS scores (eFigure).
The MRI protocol used 2 pulse sequences on a 1.5-T MRI system (GE Medical Systems), as described.8 A 3-dimensional Fourier transformed spoiled-gradient-recalled acquisition sequence yielded a coronal series of contiguous 1.5-mm images (echo time, 5 milliseconds; repetition time, 35 milliseconds; repetition, 1; nutation angle, 45°; field of view, 24 cm; acquisition matrix, 256 × 256 × 124; voxel dimension, 0.9375 × 0.9375 × 1.5 mm).
The theory and algorithms of VBM using Statistical Parametric Mapping version 5 (SPM5) software (Wellcome Department of Cognitive Neurology) are well documented.56,57 In the present cross-sectional study, VBM was performed using the Diffeomorphic Anatomical Registration Through Exponentiated Lie (DARTEL) algebra tool in SPM5.58 After realigning the T1-weighted images so that the anterior commissure–posterior commissure line was horizontal and the midsagittal plane was vertical, the images were segmented by the unified segmentation approach in SPM557 into probability maps of gray and white matter and cerebrospinal fluid. These gray and white probability maps were then rigid-body aligned (3 rotations and 3 translations) according to the Montreal Neurological Institute (MNI) template and resampled into 1.5-mm isotropic voxels. These aligned images were still in their native space (www.fil.ion.ucl.ac.uk/spm/doc/manual.pdf). Next, a population GM was created by nonlinear registration of the resampled gray and white matter probability maps using DARTEL, which provided a higher dimensional (thus more accurate) warping procedure than that typically used in standard VBM.58 The white matter maps were used to achieve this improved registration, but they were not incorporated into the statistical analysis. The GM maps of each subject were then spatially nonlinearly normalized to the population template, and Jacobian modulated. To bring the final analysis into standard MNI space, the population GM template was registered to the MNI space. All the individual GM maps residing in the population template space were then coregistered to MNI using the same affine transformation. Finally, these images were smoothed with an 8-mm full-width at half maximum Gaussian kernel.
To estimate group difference in global cortical GMV, individual cortical GMVs were calculated as per the procedure of Asami et al49 and Whitford et al.59 Consistent with our previous ROI study for male SPD,7 cortical GM was defined as all the cerebral GM, excluding the basal ganglia and thalamus. First, a binary image of cortical GM was created using the Wake Forest University PickAtlas toolkit.60 This binary image was then convolved with all subjects' preprocessed GM images, and the volumes of cortical GM were calculated by summing the constituent smoothed modulated values. Finally, relative volumes were calculated using intracranial content (ICC) volumes measured by 3-dimensional slicer (www.slicer.org).
To investigate group differences in regional GMVs, the framework of the general linear model61 was implemented using an analysis of covariance model. A 2-sample t test was applied, with ICC volume as a confounding covariate to correct for global anatomical variations.62 The resulting set of voxel values for each contrast constituted a statistical parametric map of the t statistic, SPM(t). The SPM(t) values were displayed at an uncorrected threshold of P < .001, with an extent threshold of 30 voxels for graphical reporting. The purpose of the current whole-brain VBM analysis was to identify local regions contributing to the global (cortical) GMV reduction in the male subjects with SPD observed in our previous study.7 Therefore, consistent with previous VBM analyses,63- 65 false discovery rate–corrected P < .0566 was applied to maintain a balance between appropriate correction for multiple comparisons, while also minimizing the potential for type 2 errors. In the text and tables, we discuss only those results surviving correction at this level.
As was done with the global cortical GMVs, the regional volumes were calculated as per the procedure of Asami et al49 and Whitford et al.59 First, the regions for which the VBM analysis showed significant GMV differences between the 2 groups were extracted as a gray-scale image using the save option on the SPM5 graphic user interface. Then the gray-scale image was transformed into a binary image, and this binary image was parcellated into gyri using the Wake Forest University PickAtlas toolkit60 and SPM image calculator. These binary images were then convolved with all subjects' preprocessed GM images, and the volumes of each ROI-defined region were calculated by summing the constituent modulated values. Finally, relative volumes of each region were calculated using ICC volumes.
To assess relationships between age and relative volumes of cortical and each ROI-defined GM in the SPD and HC groups, Spearman correlations were conducted, with P < .05 as significant (2 tailed). In addition, age by group interaction analyses were conducted for the cortical and each ROI-defined GM regions using analysis of variance, with cutoff P values of .05, to investigate whether the SPD and HC groups had similar patterns of age-related GMV reductions.
To evaluate contributions of structural changes to behavioral abnormalities in SPD, we evaluated associations between the positive and negative symptom scores derived from SIS and each ROI-defined GMV with significant reductions in those with SPD vs HC subjects. Based on our previous work, we hypothesized correlations with negative symptom scales. Results are reported as P < .05 (2 tailed).
There were no significant group differences in age, handedness, subject and parental SES, or IQ (Table 2).
There was a significant group difference in the global GMV; namely, the 54 neuroleptic-naive men with SPD showed smaller relative volume of the cortical GM compared with 54 matched male HC subjects (mean [SD], SPD: 42.0 [3.9], HC: 43.5 [3.1] [units are percentage of total ICC]; t106 = 2.23; P = .03).
Reflecting the result of group difference in the global GMV, the voxelwise whole-brain analysis showed significant widespread GMV reductions in the 54 men with SPD compared with the 54 HC subjects (false discovery rate–corrected P < .05). These regions included (1) temporal regions of the bilateral superior and middle temporal gyri, fusiform gyrus, and left inferior temporal gyrus; (2) frontal regions of the bilateral superior and middle frontal gyri, orbitofrontal cortex, insula, cingulate gyrus (both anterior [subgenual and affective subregions] and posterior), and right precentral gyrus; and (3) parieto-occipital regions of bilateral postcentral gyrus, right supramarginal gyrus, precuneus, and inferior occipital gyrus. Although significant reductions were observed in the multiple cerebral regions, each reduction was small (typically <300 voxels), as shown in Table 3, Figure 1, and the eFigure. We found no region where the men with SPD had significantly larger volumes compared with the HC subjects.
There were no differences in regional GMVs between men with SPD with and without comorbid Axis I or Axis II disorders. Among the 11 subjects with SPD who also had first-degree relatives with psychosis, there was no significant GMV difference when compared with the participants with SPD without a family history of psychosis (eFigure).
The results are summarized in Table 4 and Figure 2. The Spearman correlation analyses showed statistically significant negative relationships between age and relative volumes of the cortical GM, as well as most of the ROI-defined GM regions, in both those with SPD and HC participants. However, analysis of variance showed no age by group interactions.
There were no significant correlations between patients' positive symptom scores and their GMV in any of the 26 ROIs. In contrast, as hypothesized, significant negative correlations were observed between the negative symptoms and the volume of 22 of the 26 ROIs in patients with SPD (Table 5, Figure 3, and the eFigure). This was unlikely to be a chance finding, given that the probability of obtaining 22 of 26 significant correlations with an alpha level of 0.05 was P < 10−11 by the binomial theorem. Consequently, we then investigated the relationship between each of the 4 items on the SIS negative symptom scale and the regional volume of the 26 ROIs. As detailed in Table 5 and summarized here, significant negative correlations were observed between social isolation and 18 GM regions, restricted emotions and 11 regions, introversion and 13 regions, and sensitivity and 19 GM regions. The binomial distribution probability of 11 or more successes out of 26 trials with a probability of success of 0.05 was P < 10−7, which remains significant for the 4 items after Bonferroni correction (eFigure).
This study demonstrated statistically significant GMV reductions in global and widespread brain regions in the 54 neuroleptic-naive men with SPD compared with the 54 matched male HC subjects. To our knowledge, this is the first study to demonstrate such widespread GMV reductions in what is, we believe, the largest sample of neuroleptic-naive men with SPD compared with HC subjects. Our extensive results, not observed in other VBM studies,20,21 may be owing to the improved sensitivity of the DARTEL method67 and the careful group matching by age, sex, handedness, IQ, and the subject's own and parental SES. The global analysis finding of a significant GMV reduction in the cortex in the men with SPD was in line with our previous result, arrived at using a ROI method.7 In the voxelwise analysis, significant widespread GMV reductions were discovered in multiple cortical regions in the men with SPD compared with HC subjects. The size of each reduced region was comparatively small, in contrast to previous studies that have identified widespread volumetric GM reductions in patients with schizophrenia.49,68 Relative to women with SPD, the men with SPD showed GMV reductions in widespread local brain regions.
In the frontal lobe, volume reductions were confirmed in several frontal surface gyri: bilateral superior (mainly the medial region), middle, orbitofrontal, and right precentral. Previous studies have suggested that the volume of frontal regions, particularly Brodmann area 10, may be preserved in subjects with SPD.1,29 However, we found that men with SPD had GMV reductions at least in small regions in the medial frontal regions that were associated with the default mode network. Within frontolimbic regions, SPD evinced smaller GMVs in the bilateral insula, affective and subgenual subregions of the anterior cingulate gyrus, and posterior cingulate gyrus, similar to previous studies.8,20,21,29
In the temporal lobe, the subjects with SPD showed volume reductions in the bilateral superior and middle temporal gyri, fusiform gyrus, and left inferior temporal gyrus compared with the HC subjects. The volume reduction in the STG is the most consistent finding in both the ROI9,26,31 and VBM8,21 analyses in SPD; volume reductions in the other temporal regions have also been reported previously.17,27 The fusiform gyrus is thought to be important in facial recognition,69,70 and our previous MRI–evoked-potential studies have demonstrated these structural-cognitive relationships in schizophrenia.71,72 Subjects with SPD are also thought to have deficits of facial recognition73,74; the finding of fusiform gyrus volume reductions might be an anatomical substrate of poor facial recognition in SPD. Although our previous ROI did not demonstrate a difference in fusiform GMV,12 the voxel by voxel approach, as well as larger subject sample, could detect small abnormalities in fusiform gyrus in SPD. In parietal and occipital lobes, this study provided new evidence that men with SPD showed GMV reductions in the bilateral postcentral gyrus, precuneus, right supramarginal gyrus, and inferior occipital gyrus compared with HC subjects. Lateral parietal regions are also believed to be related to the default mode network,75 while the inferior occipital gyrus and fusiform gyrus are believed to be involved in the ventral stream of visual processing. Thus, it is feasible that abnormalities in this region might also be related with poor facial recognition in individuals with SPD.
While the neurobiological mechanism underlying the smaller GMVs in SPD is unknown, it is consistent with the finding of GMV reductions observed in schizophrenia, but without the postonset progression. Neuropathologic investigations indicate the basis of GMV reductions in schizophrenia is neuropil reduction.76,77 We have speculated that, in schizophrenia, there is a cortical circuit abnormality of deficient gamma-aminobutyric acidergic recurrent inhibition that may result in excitotoxic GMV loss. While this mechanism could promote progressive GMV reductions in schizophrenia, we speculate that SPD has a less severe alteration in circuitry, therefore shows no progressive GMV changes.
The absence of progression in SPD is supported by a recent longitudinal MRI study showing no progressive GMV changes in the STG after a 2.9-year interval in the patients with SPD compared with HC subjects.31 In our data, the similar patterns of the age-related volume decline between the 2 groups may provide provisional evidence that, in contrast to patients with schizophrenia, male subjects with SPD do not show progressive loss on neocortical and almost all regional volumes,36,78,79 although it is necessary to confirm our findings in a longitudinal analysis. This caveat notwithstanding, our results point to an intriguing difference between schizophrenia and the spectrum disorder of SPD, a difference that invites further neurobiological study.
Finally, the correlation analyses, although tentative because of the number of correlations performed, demonstrated relationships between the local GMV reductions and behavioral abnormalities in the men with SPD. This confirmed our prediction of the association between negative symptoms and GMVs, a prediction generated because severe negative symptoms and fewer close friends are the most salient and prominent features of male SPD,44 thus most likely to be associated with volumetric reductions. The association of reduced STG and fusiform gyrus volumes with negative symptoms is of interest in suggesting that deficits in early sensory processing may be related to negative symptoms. We noted also that previous studies have reported that negative symptoms are among the risk factors for social dysfunction, lower quality of life, and suicide in subjects with schizophrenia spectrum disorders.80- 82 Because studies have shown that individuals with SPD have worse social function and higher frequency of attempted suicide,83,84 this association of negative symptoms and volumetric reduction provides insight into the pathophysiology of SPD.
To our knowledge, there are no previous MRI studies reporting widespread structure-negative symptom relationships in men with SPD. In previous studies—which either included only female subjects or combined male and female subjects with SPD—only a small number of regions have been demonstrated to have a relationship with negative symptoms. Examples of these regions include Brodmann area 22 (STG), showing the negative association with the interpersonal impairment score (which involved no close friends29) and the caudate, showing the relationship with restricted emotions.15 In contrast, our current study revealed that the widespread GM regions—not limited local regions—were associated with the negative symptoms. These GM regions included frontal (mainly medial regions), frontolimbic, temporal, and parietal regions. These regions have been reported to have associations with negative symptoms in schizophrenia.85- 88 We note the negative symptom—GM association did not definitively indicate that the GMV changes caused the negative symptoms because various extraneous factors (eg, social factors) could be causative of, or could potentiate, GMV changes. Finally, with respect to the GMV changes, because this never-medicated subject population excluded medication as a confound, we believe the observed GMV-negative symptom association is important in providing information on the pathophysiology of schizophrenia spectrum disorders.
Of particular interest to the negative-symptom profile is its relationship to the default network and the location of SPD GM deficits. Recent functional MRI studies and fractionation analyses75 suggest the core of the default network is the posterior cingulate and anterior medial prefrontal cortex, regions prominently represented in GM deficits in SPD. The fractionation analysis also identified default network components related to self in the present; the SPD regions with GMV deficits had a strong overlap with this aspect of the self, including the dorsomedial (anterior cingulate) prefrontal cortex, temporopartietal junction regions, and lateral temporal cortex, although the temporal pole did not show a deficit in SPD. These SPD regions also showed correlations with negative symptoms. Interestingly, functional MRI studies point to abnormalities in the resting-state default network in schizophrenia and, of particular note, in first-degree relatives of persons with schizophrenia.89,90 While there are many ways of conceptualizing the default network activity, one prominent theory is that it is related to social cognition because the brain regions in the default network and those active in social cognition overlap.91 Thus, we hypothesized that the prominent social deficits/negative symptoms shown in SPD may be, at least in part, related to GM abnormalities in the default network, regions that showed association with negative symptoms.
The present data indicate that contributions to negative symptoms likely also stem from a defective anatomic substrate in other, widespread brain regions in addition to the default network. Abnormalities in GMV in auditory, somatosensory, and fusiform regions were associated with negative symptoms, regions involved in sensory and language processing, and possibly interfering with social interaction through defective processing of social cues. Moreover, as reviewed by us in our article on orbitofrontal cortex in schizophrenia,92 abnormalities in this region have consistently been associated with social deficits, as we found in the SPD population.
The correlation analyses between negative symptoms and regional volumes, although hypothesis driven, should be considered exploratory owing to the number of correlations, therefore needing confirmation in future studies. A longitudinal study would offer stronger evidence that men with SPD do not (or do) have progressive GMV loss.
This study revealed volumetric GM reductions and symptom correlations in a large and neuroleptic-naive sample of single-sex (male) subjects with SPD. A correlation analysis indicated that the widespread GMV reductions were strongly associated with negative symptoms in men with SPD. These reductions also overlapped with the default network and the regions active in social cognition and perception of social cues, suggesting a possible anatomical substrate of negative symptoms. The absence of an age-related difference in GM change between those with SPD and HC subjects suggested that, unlike schizophrenia, there is no progressive volume reduction in SPD. Future longitudinal studies would be useful in confirming this important point of difference between the 2 disorders.
Correspondence: Robert W. McCarley, MD, Department of Psychiatry, 116A, Boston Veterans Affairs Healthcare System, Brockton Division, Harvard Medical School, 940 Belmont St, Brockton, MA 02301 (firstname.lastname@example.org).
Submitted for Publication: June 6, 2012; final revision received June 6, 2012; accepted August 21, 2012.
Published Online: February 6, 2013. doi:10.1001/jamapsychiatry.2013.665
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by Merit Awards from the Department of Veterans Affairs (Drs Shenton and McCarley); a Research Enhancement Award (Drs Shenton and McCarley); a Middleton Award (Dr McCarley); a Veterans Affairs Advanced Career Development Award (Dr Dickey); and grants K05MH070047 and R01MH50740 (Dr Shenton) and R01 MH 40799 and R01 MH 052807 (Dr McCarley) from the National Institute of Mental Health. Dr Whitford's work is supported by an Overseas-Based Biomedical Training Fellowship from the National Health and Medical Research Council of Australia (NHMRC 520627), administered through the Melbourne Neuropsychiatry Centre at the University of Melbourne, Australia.
Additional Contributions: We acknowledge Marie Fairbanks for her administrative support on this study.