The diagnosis of SPD, characterized by asocial tendencies, difficulties with language and paranoid, odd behavior and magical thinking, was first introduced in the DSM-III and based on the clinical profiles of patients with “borderline schizophrenia” in the landmark Danish adoption schizophrenia studies (Rosenthal et al., 1971). These studies and earlier formulations of the disorder suggested that relatives of patients with schizophrenia may display deviant psychological functioning but not all signs of schizophrenia, providing evidence for a spectrum of schizophrenia-related disorders. Many of these earlier studies of “borderline schizophrenia” included symptoms related to conceptions of BPD, but diagnostic overlap decreased when borderline criterion for paranoid ideation under stress was introduced in DSM-III-R (Spitzer et al., 1979). BPD, characterized by affective instability and impulsive behavior, was first included as a personality disorder diagnosis in DSM-III (Gunderson and Singer, 1975;Kernberg, 1977). Nonetheless, the co-morbidity of BPD and SPD is not uncommon (e.g., Zanarini et al., 1998).

Morphometric studies examining the temporal lobe in SPD report reduced gray matter volume in STG (Buchsbaum et al., 1997; Dickey et al., 1999; Hazlett et al., 2008; Kawasaki et al., 2004; Takahashi et al., 2006), middle and inferior temporal gyrus (Downhill et al., 2001), and Heschl’s gyrus (Dickey et al., 2002a). Of note, Dickey and colleagues found decreased STG volume in male (2002a) but not female SPD patients (2003). While volume reductions in medial temporal regions, including amygdala and/or hippocampal complex, are reported in schizophrenia (reviewed by Shenton et al., 2001), these findings have not been consistently observed in SPD (Dickey et al., 2002b). Only one study (Suzuki et al., 2005) reported bilateral reductions in both of these structures in SPD, while Dickey et al. (2007) recently reported reduced hippocampal volume in SPD. Volume reductions in frontal cortex regions have also been reported in studies with schizophrenia patients, but several studies suggest these areas may be relatively preserved in SPD (Hazlett et al., 2008; Kawasaki et al., 2004; Suzuki et al., 2005). Our group and others (Haznedar et al., 2004; Takahashi et al., 2004) previously reported no morphologic differences in cingulate gyrus in SPD but, using a larger sample size, we later found smaller cingulate volume compared with healthy controls, particularly in anterior cingulate cortex (ACC) gray matter (Hazlett et al., 2008).

On the other hand, BPD studies examining the frontal lobe have reported reductions in overall frontal lobe volume (Lyoo et al., 1998), left orbitofrontal cortex and right anterior cingulate volume (Tebartz van Elst et al., 2003), and anterior cingulate (BA24) gray matter volume (Hazlett et al., 2005). It should be noted that Tebartz van Elst and colleagues later found no significant differences in these regions using voxel-based morphometric MRI (Rusch et al., 2003). Some studies also report hippocampal (Irle et al., 2005; Zetzsche et al., 2007) and amygdala volume loss in BPD (Driessen et al., 2000; Schmahl et al., 2003;Tebartz van Elst et al., 2003), while others have shown no significant structural differences between BPD patients and controls in these regions (Chanen et al., 2008; New et al., 2007). Another study found smaller hippocampal volume but no differences in amygdala volume (Brambilla et al., 2004). To date, the above-mentioned studies have looked at these patient groups separately and findings suggest SPD patients have smaller STG volume, both SPD and BPD patients have abnormalities in frontal regions, particularly smaller anterior cingulate volume, and amygdala findings are less clear.

The aim of this study was to examine volume of frontal and temporal lobe regions in three age-gender- and education-matched groups: patients with BPD but no SPD traits, patients with SPD but no BPD traits, and healthy controls. This is the first study to directly compare these groups and test whether morphometric abnormalities in these brain regions are specific to one personality disorder and not the other, or whether the two patient groups differ from healthy controls in opposite directions. While both SPD and BPD patients exhibit symptoms that have been linked to temporal and frontal lobe abnormalities, these symptoms differ phenomenologically from one another. Specifically, emotion dysregulation and impulsivity are hallmark features of BPD (Skodol et al., 2002), while blunted affect and executive dysfunction are frequently associated with SPD (Kirrane and Siever, 2000). In this sense, abnormal volume in prefrontal and temporal regions may underlie affect-related symptoms in BPD (Soloff et al., 2008); in SPD, abnormal frontal and temporal lobe volume may be associated with cognitive and/or executive functioning deficits (McCloskey et al., 2005). Due to these different patterns of symptom presentation in our clinical groups, we expected to find different patterns of frontal/temporal structural abnormalities in these groups as well. Of note, since we want to be able to draw conclusions about regional specificity, we looked at a number of different frontal and temporal regions. Specifically, in the frontal lobe, our analyses focused on prefrontal and cingulate cortex Brodmann areas (BAs)— important in attention and the modulation of executive function and emotion. In the temporal lobe, we focused on regions that contribute to executive and emotion processing and have been shown to have dense interconnections: temporal gyrus regions and amygdala (Barbas, 2007; Pandya, 1995; Price, 2007). The specific temporal gyrus regions included primary auditory cortex corresponding to BA41 and BA42 and auditory association areas corresponding to BA22/STG. Also, since fMRI studies (Kim et al., 2003;Whalen et al., 2001) have begun identifying differential roles of amygdaloid subnuclei and their respective connections with the frontal lobe, we parcellated amygdala into dorsal and ventral regions.

Previous work shows that, in SPD, functional impairment such as odd speech is associated with smaller STG volume (Dickey et al., 2003). This finding is consistent with basic research (Binder et al., 2000; Joanisse and Gati, 2003) and schizophrenia work (e.g., Wible et al., 2001) suggesting the importance of temporal lobe function in language processing and production. In order to investigate individual differences in SPD, we examine correlations between symptom severity, including odd speech/thinking symptomatology, and volume of our hypothesized regions of interest.

From a functional perspective, prior work indicates that STG/BA22 has wide connections to the hippocampus-amygdala complex, cortical association areas in the frontal and parietal lobe and the thalamus, and it has been proposed that this part of the temporal cortex is involved in higher-order auditory perception (Binder et al., 1994; Engelien et al., 1995; Rajarethinam et al., 2000; Pandya, 1995). This includes phonemic discrimination and processing, the fusing of multiple sound features and analysis of temporal acoustic features of speech. Additionally, lesioning of this region produces receptive aphasia and deficits in comprehension of verbal language (Demonet et al., 2005). Considered together with these anatomical findings, the posterior STG localization of our results in both patient groups fits well with work showing that both SPD and BPD individuals exhibit auditory information processing deficits that are similar to one another, as measured by P300 auditory evoked potentials (Kutcher et al., 1989).

As such, our finding demonstrates that volumetric abnormalities in BA22 are unique to SPD and may be an important tool for differential diagnosis between these two disorders. Although it is still unclear how abnormalities in this region can affect auditory processing, it would be worthwhile to investigate the relationship between alterations in BA22 volume and socially-related functions since auditory processing plays a crucial role in social function and may be one way to distinguish between the groups clinically. For example, BA22 is located on the caudal two-thirds of the STG, an area that Gallagher and Frith (2003) found to be activated during “mentalizing” tasks involving interpretation of mental states of others. Thus, our finding may help differentiate between “mentalizing” deficits experienced by both patient groups.

The finding of reduced gray matter volume in BA22 is the most widely-replicated finding in schizophrenia (reviewed by Shenton et al., 2001). Therefore, our finding of reduced STG volume in SPD provides additional evidence for the similarity between SPD and schizophrenia patients. Our prior work showed a schizophrenia spectrum pattern for reduced STG volume with schizophrenia patients exhibiting the largest reduction compared with healthy controls, and SPD patients intermediate (Hazlett et al., 2008). This observation is consistent with prior work indicating that individuals with SPD share a broad range of similarities with schizophrenia patients in terms of genetics and neurobiology (Siever and Davis, 2004).

Our other interesting finding is that, among the SPD patients, smaller relative BA22 gray matter volume was associated with greater overall clinical (DSM) symptom severity, including greater impairment due to odd thinking and speech. These results are consistent with clinical correlates previously reported by Dickey et al. (2003) and provide further evidence that STG abnormalities may underlie the pathophysiology of SPD. A better understanding of the relationship between structural and functional abnormalities in this key language processing area and psychopathology should be a top priority for future work.

We also found larger-than-normal BA42 gray matter volume in BPD patients. This area makes up part of the primary auditory cortex and lies in Heschl’s gyrus, a region shown to be involved in sensory auditory processing. As of April 2009, there were no publications reporting morphology of BA42 in BPD patients but it is possible that increased volume or alterations in this region reflect some type of compensatory mechanism for other dysfunctional regions involved in auditory-processing demands. In previous work, BA42 volume was found to be smaller in males with SPD compared with controls (Dickey et al., 2002a). Even though BA42 makes up only a portion of Heschl’s gyrus and other work has shown preserved volume of this region in SPD (Takahashi et al., 2006), it is important to note possible reasons for discrepancies between our failure to find healthy control-SPD differences in BA42 volume and findings of reduced left Heschl’s gyrus gray matter volume by Dickey et al. (2002a). Differences may be due to heterogeneity of sample characteristics and symptom severity, methodological techniques, normal variations in the distribution of auditory cortex (Dickey et al., 2008) or medication differences. Understanding both structural and functional abnormalities in this region in BPD and SPD patients should remain a goal for future work such that measures of temporal lobe integrity may ultimately help us predict which patients will respond to different treatments.

For the amygdala, a temporal lobe region shown to have dense interconnections with the STG, no significant group differences were detected. Although some studies have shown smaller amygdala volume in BPD compared with healthy controls (Driessen et al., 2000; Rusch et al., 2003; Schmahl et al., 2003; Tebartz et al., 2003), we (New et al., 2007) and others (Brambilla et al., 2004) have found no groups differences. Perhaps amygdala abnormalities in BPD may be more functional than structural. The issue of amygdala volume requires further investigation.

Our frontal lobe analyses were not statistically significant, indicating the three groups did not differ in anterior-orbital-dorsolateral or cingulate gyrus gray matter volume. These prefrontal cortex results are consistent with our previous findings and other observations showing prefrontal volumes largely preserved in SPD in contrast to widespread prefrontal reductions in schizophrenia (Buchsbaum et al., 2002; Suzuki et al., 2005). Contrary to our hypothesis, neither patient group showed significantly smaller ACC volume compared with healthy controls. We previously reported significantly reduced cingulate (BA24, 31, 23) gray matter volume in a sample of 79 patients meeting DSM criteria for SPD compared with healthy controls (Hazlett et al., 2008). However, it should be noted that in our prior study, the SPD group may have had comorbid BPD; this was not examined because the SPD group was compared with schizophrenia patients. We have also previously shown that, compared with healthy controls, individuals with a diagnosis of both SPD and BPD exhibit greater BA24 reduction than those with only BPD (Hazlett et al., 2005). In contrast, our current sample included SPD patients without BPD. Thus, patient sample characteristics such as SPD-BPD comorbidity appear to play an important role in anterior cingulate volume abnormalities and may explain discrepancies across studies.

In the prefrontal cortex, the BPD patients showed no significant gray matter volume difference which is consistent with previous work by our group and others (Hazlett et al., 2005; Rusch et al., 2003). Nonetheless, as mentioned above, the current study did not replicate our previous findings of normal-BPD differences in anterior cingulate gray volume (Hazlett et al., 2005) which has also been reported by Tebartz van Elst et al (2003). This inconsistency may be due to sample differences in clinical symptom characteristics and severity (e.g., 50% of current sample had past history of MDD versus 76% of Hazlett et al 2005 sample).

To the best of our knowledge, this is the first volumetric study to compare SPD patients with BPD patients and healthy controls. Although all patients were unmedicated at the time of their MRI, a unique feature of the study since many MRI studies to date examine medicated patients, a primary limitation is that some had previously received psychoactive medications. However, it is important to note that when we controlled for prior medication history, our finding of smaller STG volume in SPD compared with healthy controls and BPD patients remained significant. A second limitation is that, due to power considerations, we did not include participants with a comorbid diagnosis of both BPD and SPD. This is an important group to include in future studies. Additionally, given the common comorbidity between BPD and MDD, half of the BPD patients in our study met criteria for past MDD, although it was required to be in full remission for at least two months prior to their scan. It is imperative that future BPD studies recruit large enough sample sizes so that individuals with and without past MDD can be compared. As such, it is important to note that our finding of smaller-than-normal STG volume in SPD remained significant when we controlled for past MDD (only 18% of the SPD patients had past MDD). Also, although our diagnostic groups were matched on age and gender, it is possible that gender differences exist between groups in morphometry. Unfortunately, given our current sample size, we were unable to examine whether gender effects play a role in our STG findings. Future MRI studies with larger sample sizes will allow a much-needed examination of gender effects in SPD and BPD. Lastly, we do not have family history records for our patients. It would be worthwhile to know how many SPD and BPD patients have a first-degree relative in the schizophrenia spectrum and/or bipolar spectrum and whether there is overlap in family history. Nevertheless, our STG findings suggest that future work addressing the genetic link between reduced STG volume and schizophrenia-spectrum disorders may be fruitful.