Schizophrenia (SCZ) and bipolar disorder (BPD) are both chronic psychiatric illnesses, with some overlapping clinical features. SCZ consists primarily of psychotic symptoms, including delusions, hallucinations, and disorganized behaviors, and often involves cognitive impairment. BPD often involves psychotic features (1), but its core clinical symptoms involve mood abnormalities typically involving periods of mania as well as depression. Many studies have reported genetic overlap between these two disorders (2–4), with associated familial overlap (5). As a result of such findings, increasingly patients with those disorders are studied together.

Numerous studies have investigated white matter integrity in SCZ and BPD. Diffusion imaging studies in SCZ typically report lower fractional anisotropy (FA) in one or more white matter tracts, often regionally, however, the specificity of significant abnormalities has been heterogenous across studies (6, 7). The large, multi-site ENIGMA study involving 4,322 individuals found widespread FA reduction in SCZ, with the anterior corona radiata (d=0.40) and corpus callosum (d=0.39) showing the greatest effects (8). Abnormal diffusion imaging findings in BPD have been more variable and have shown less robust findings than in SCZ. One meta-analysis reported two clusters of decreased FA, both in the right hemisphere: one close to the parahippocampus and the other close to the anterior cingulate (9). Another voxel-based meta-analytic study reported three significant clusters of decreased FA in BPD: a right posterior temporoparietal cluster and two left cingulate clusters (10). More recently, a mega-analysis across 3,033 individuals collected through the ENIGMA network reported lower FA in 29 regions, most notably within the corpus callosum and cingulum (11). Fewer studies have compared white matter anisotropy across both disorders (12, 13). A recent meta-analysis of diffusion and structural imaging white matter abnormalities in SCZ and BPD found a shared decrease in corpus callosum volume and FA across disorders, as well as SCZ-specific white matter abnormalities involving the left cingulum and the right anterior limb of the internal capsule (ALIC) (13).

Cognitive impairment occurs less commonly in BPD than in SCZ (14), where deficits in a range of domains including attention, working memory, verbal memory, and executive functioning are core features of the disorder and are generally chronologically progressive (15–17). SCZ has also been associated with an increased risk of dementia later in life, including showing precocious onset among younger individuals with the disorder (18–20). SCZ has therefore been hypothesized to be a disorder of accelerated biological aging (18, 21, 22). Accelerated aging is supported by studies reporting shared genetics between SCZ and a variety of age-related diseases (23, 24). Measures of biological aging could prove valuable for assessing SCZ patients’ risk for physical and cognitive decline and for evaluating intervention effectiveness. Direct comparisons of the brains of SCZ patients with those with age-related brain changes using identical imaging methods have however rarely been done.

TRACULA (TRActs Constrained by UnderLying Anatomy) is a software package that uses global probabilistic tractography combined with anatomical priors informed by the individual’s own anatomy to accurately map known fiber tracts in the brain and can evaluate multiple diffusion tensor imaging (DTI) parameter values within these tracts (25). TRACULA has been used in multiple research studies to evaluate different neuropathologies such as temporal lobe epilepsy (26), multiple sclerosis (27), amyotrophic lateral sclerosis (28), and obsessive-compulsive disorder (29). Only one prior study has conducted tract-based analysis in SCZ with TRACULA and used diffusion imaging data collected on the same customized ‘Connectom’ scanner used for the Human Connectome Project’s (HCP) “Young Adult” study. This study found a trend towards lower tract FA in patients, most significantly in the left anterior thalamic radiation, but no significant FA abnormalities in bipolar disorder (12). Two studies have used TRACULA in BPD. One of these found reduced FA particularly in the parietal part of the superior longitudinal fasciculus (SLF) (30a), and most notably in the SLF, cingulum-cingulate gyrus bundles, and corticospinal tracts in another study (31).

Aging has been associated with a generalized reduction in white matter FA and increased radial diffusivity (RD), particularly in later age, attributed primarily to a degeneration of myelin sheaths and loss of myelinated fibers (32–34). In Alzheimer’s dementia, diffusion imaging results have been variable (35), but further decreases in white matter integrity (relative to age-matched controls) are generally reported (36–38). Using TRACULA, reduced tract FA and increased mean diffusivity (MD) has been previously been found with aging non-uniformly across most tracts (39). Furthermore, elderly individuals with Alzheimer’s dementia or mild cognitive impairment had increased MD in the cingulum bundles compared to age-related control subjects (40). While the tract abnormalities with dementia share similarities to that reported for SCZ, these populations have not been directly compared using tract-based methods.

In the current pilot study, we use TRACULA to investigate white matter tract integrity in young healthy controls and individuals with SCZ and BPD, as well as in healthy elderly and a dementia cohort, all collected using the same scanner and acquisition protocol. Based on the most notable abnormalities reported in prior TRACULA studies in psychiatric populations (12, 30, 31), we focus our investigations on seven white matter tracts: the anterior thalamic radiation; three superior longitudinal fasciculi, the dorsal and ventral cingulum bundles, and the corticospinal tract. We hypothesized that previous TRACULA findings in SCZ and BPD will be reproducible in these major fiber tract pathways. Secondly, we hypothesized that white matter abnormalities in aging and dementia will have similar abnormalities as those seen with SCZ, but greater in severity.

Our study provides preliminary data comparing diffusion imaging metrics across younger psychiatric populations and older cohorts using an automated tract-based analysis. We found that the volumes of the white matter tracts did not differ significantly across groups, whereas there were significant differences in tract fractional anisotropy across the various tracts studied. This contrasts with several other studies which have reported decreased white matter volumes, though results have been inconsistent and highly variable. SCZ has been associated with a reduction of prefrontal or frontal white matter (62–65), temporal white matter (62), or posterior brain white matter (66) volumes. Others have reported reduced white matter volume in regions corresponding to the inferior longitudinal fasciculus in first-episode SCZ (67) or to the left fronto-occipital fasciculus in prodromal individuals who converted to psychosis (68). Studies of aging and dementia have also been associated with reductions in white matter volume (34, 69–71). Notably, white matter volumes have been reported to gradually increase in the first 40 years of life and then rapidly decrease after 60 years (70–72). The absence of substantial tract volume differences between our younger and older cohorts in our study was therefore unexpected and suggests that tract-wide volume loss may not be observed until after age 70. As our study involved measures of entire tracts, minor white structural changes in disease are likely diluted by the inclusion of less affected parts of the tracts in the analyses, as has been shown by comparing tract-based analyses results to that of voxel-based analyses (12).

Our study found a substantially reduced FA in most white matter tracts in elderly and dementia participants compared to the younger cohorts, with findings in dementia not significantly lower than in the healthy elderly population. Similarly, we found an increase in RD in the older cohorts, indicating an increase in the water diffusion in the direction perpendicular to the direction of the white matter fiber. These findings are consistent with the generalized reduced white matter integrity reported with old age (32, 34), attributed primarily to the degeneration of myelin sheaths and loss of myelinated fibers (33).

We did not find significant diffusion abnormalities in SCZ or BPD participants. FA in SCZ trended towards lower values, and RD towards higher values, suggesting potential myelin degeneration in this population, albeit to a relatively mild degree. Here again, finding statistically significant diffusion metrics in a whole white matter tract would be less likely if there are regions across the tract that are unaffected or only minimally affected. The white matter regions affected in SCZ have been highly variable across studies, and rarely involve universal white matter abnormalities (6, 7). In general, findings have implicated prefrontal and temporal lobes and the fiber tracts connecting these regions. Analysis of data from the ENIGMA schizophrenia DTI work group of 2,359 healthy controls and 1,963 schizophrenia patients from 29 independent international studies, reported that FA reductions in SCZ are widespread, and involved all major WM fasciculi, with the anterior corona radiata and corpus callosum showing the greatest effects (8). An earlier meta-analytic study found significant FA reductions primarily in two regions: the left frontal deep white matter and the left temporal deep white matter in SCZ (6). Similarly, a meta-analytic study of first-episode psychosis reported FA reduction in the right limbic white matter and left temporal white matter (73). Regarding BPD, diffusion imaging abnormalities have generally been found to be less severe than that found in SCZ (9–11), consistent with the results of our current study.

Tract-based white matter assessments in psychiatric disorders, as described in this study, are relatively infrequent. Another study using TRACULA in twenty-four SCZ patients similarly did not find significant tract effects compared to thirty healthy controls. That study however found a trend toward lower FA in the ATR (12), a tract that was the least affected in our current study. This variability is consistent with the hypothesis that the specific white matter regions most affected in schizophrenia tend to vary widely across patients. A failure to achieve statistically significant findings in schizophrenia using TRACULA may thus be related to heterogeneity across individuals combined with the relatively low sample size. Two substantially larger TRACULA studies of bipolar disorder subjects (with sample sizes of 96 and 72) have shown significant findings across several tracts (30, 31), which was not observed in a smaller TRACULA study of thirty-three bipolar disorder subjects (12). This suggests that tract-based diffusion imaging studies require a larger sample size, compared to that required for voxel-based studies, to detect groupwise FA differences (12).

We however found a significant moderate correlation between chronic psychotic symptom severity and FA in each of the three SLF tracts, which was not observed with more recent symptoms. The WERCAP Screen estimates chronic symptoms over the last year and may therefore more accurately capture cumulative psychopathology compared to the SAPS which estimates recent symptoms (i.e. over the last two weeks), which may be more indicative of state-related changes. The SLF is the largest associative fiber bundle system in the brain, and connects the frontal, temporal, and parietal lobes within the same hemisphere (74, 75). The main functions supported by the regions connected by the SLF are visual and spatial cognition, attention processes, control of motor processes and executive functions, and language functions (76). SLF-I represents the dorsal division, connecting the superior parietal and superior frontal lobes, and appears to be involved with regulating motor behavior. The SLF II originates in the caudal-inferior parietal cortex and terminates in the dorsolateral prefrontal cortex and appears to be involved in visuospatial attention. The SLF-III is the most ventral, extending from the supramarginal gyrus, anterior to the angular gyrus, to the ventral premotor and prefrontal areas. Frontoparietal dysconnectivity has been attributed to SCZ etiopathogenesis (76), and impaired SLF has been found in SCZ in several studies (7, 8, 77–79) and associated with psychotic symptom severity (79, 80). Our results are consistent with these findings and suggest that decreased integrity of SLF tracts may indicate a more advanced symptom profile.

Our study and its interpretation have some limitations. Firstly, the sample size used in our study was only modest for each group (n=28 or fewer) and thus may not have had sufficient power to detect significant group differences, particularly since the white matter regions affected are highly variable across individuals. Nevertheless, our study showed strong trends towards lower FA values in schizophrenia, and FA in the SLF showed a moderate inverse correlation with psychotic symptom severity. In the future, larger tract-based studies of diffusion data acquired similarly are needed to identify significant group effects, as well as subtypes of patients with unique patterns of white matter impairment. Secondly, results from tract-based methods may miss white matter damage if it does not involve a substantial portion of the tract since the healthier parts of the tract could dilute the effect. Thus, combining tract-based methods with voxel-based methods would overcome the disadvantages of each method, facilitating the estimation of both general and regional tract integrity (12). Thirdly, our study did not account for potential confounders in our analyses, including substance and medication use which may have influenced the findings. Antipsychotic use, for example, has been reported to increase tract FA (81, 82), while others have reported a subtle loss of white matter integrity (83). Diffusion studies have also found poorer integrity of white matter in cannabis users compared to non-users (84, 85). In addition, potential protective factors for white matter abnormalities including psychotherapy (86, 87) or cognitive training (88) may have been a confounder in our study, and such data were unavailable from our participants for analyses.

In conclusion, using the automated tractography tool TRACULA, our study showed significantly impaired white matter integrity with aging, suggesting demyelination. The pattern of white matter abnormalities in schizophrenia was similar to that in aging, but was much lesser in severity and did not meet statistical significance. In the future, larger sample sizes and aggregation with other data sets are recommended to increase the power to detect group differences.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by Washington University Institutional Review Board. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

DM wrote the original draft and conducted most of the statistical analyses. SC conducted neuroimaging processing and analyses. JS guided imaging processing and analyses. MH oversaw neuroimaging acquisition, processing, and analyses. All authors contributed to the article and approved the submitted version.