On the relation of white matter brain abnormalities and the asociality symptoms in schizophrenia outpatients – a DTI study

1 Institute of Psychology, Jagiellonian University, Krakow, Poland, 2 Psychosis Research and Psychotherapy Unit, Association for the Development of Community Psychiatry and Care, Krakow, Poland, 3 Department of Psychiatry, Faculty of Medicine, Masaryk University and University Hospital Brno, Brno, Czech Republic, 4 Community Psychiatry and Psychosis Research Center, Chair of Psychiatry, Medical College, Jagiellonian University, Krakow, Poland, 5 Lentis Research, Lentis Psychiatric Institute, Groningen, The Netherlands, 6 Faculty of Behavioural and Social Sciences, University of Groningen, Groningen, The Netherlands, * Email: przemyslaw.adamczyk@uj.edu.pl The study was conducted by the Krakow Schizophrenia Research Group, Krakow, Poland


INTRODUCTION
The brain origins of psychotic symptoms, known as Kraepelin's 'dementia praecox', were further indicated by Bleuer as an essential disintegration of mental processes in schizophrenia (Jablensky, 2010). Nowadays, it resonates with the disconnection theory (Friston, 2002;Friston et al., 2016), where neuropathology of schizophrenia points at abnormal functional connections associated with subtle changes in the structure of axons located across the brain.
Recent evidence implicates that altered white matter (WM) structure in schizophrenia can be regarded as a long-term effect of disturbed myelinogenesis, e.g. lower levels of oligodendrocyte-and myelin-related proteins (Cassoli et al., 2015;Schoonover et al., 2019), but also as a consequence of excitotoxicity, cytoskeletal abnormalities (Uranova et al., 2004) or disturbances in neurogenesis, e.g. weakened synaptic pruning (Alba-Ferrara and de Erausquin, 2013;Klauser et al., 2016). Finally, the abnormal structural connectivity and microstructural changes on synapses lead to the functional disturbances and manifestation of psychopathological symptoms related to schizophrenia (Friston 2002;Friston et al., 2016).
Apart from positive symptoms (e.g. hallucinations, delusions, paranoid thoughts) and various socio-cognitive deficiencies, negative symptoms (i.e., alogia, apathy, avolition, anhedonia, asociality, blunted affect, and poverty of speech) remain as an extremely important hallmark of the diagnosis and considered a core feature of schizophrenia. Indeed, negative symptoms are believed to be the most stable characteristics for schizophrenia psychopathology (Andreasen, 1982;Harrow and Jobe, 2018), and social anhedonia is recognized as a predictor of later schizophrenia spectrum disorder development in young adults (Kwapil, 1998).
Hence, the investigation on biomarkers of this phenomenon, including structural changes in brain connectivity, that affect various aspects of brain functioning is of great importance, as the occurrence of negative symptoms has a destructive impact on individual functioning in schizophrenia (e.g. occupational impairment, financial dependence, poor social relationships and worse quality of life) and remains one of the most important scientific and societal problems related to schizophrenic psychosis remediation (Kaiser et al., 2011;Bijanki et al., 2015;Dollfus and Lyne 2017;Correll and Schooler, 2020).
Recently, analyses of WM alterations in the schizophrenic brain are performed using diffusion tensor imaging (DTI). DTI is based on the estimation of the diffusion of water molecules in tissues as this allows a reconstruction of images of brain fibers' cytoarchitecture (Emsell et al., 2016). One of the most widely used DTI measurements is fractional anisotropy (FA), which is generally interpreted as a biomarker of the integrity of WM bundles (Assaf and Pasternak, 2008). Other commonly used parameters are axial diffusivity (AD), regarded as the index of axonal damage; radial diffusivity (RD), a measure of the level of myelinization (Karlsgodt, 2016); mean diffusivity (MD), considered a complex measure of the surrounding cytoarchitecture (Emsell et al., 2016).
Recent data undoubtedly indicate the existence of abnormal WM structure in schizophrenia (Kelly et al., 2018;Koshiyama et al., 2020). In particular, a recent meta-analysis of 29 DTI studies in schizophrenia, including 1963 patients, showed differences in the diffusion parameters between clinical and healthy subjects in 20 tracts in total, indicating the WM alterations in schizophrenia are widespread and affecting the majority of fiber bundles in the brain, with the greatest effect size in the corpus callosum and the anterior corona radiate (Kelly et al., 2018). Other research consistently revealed that WM abnormalities were detected in different structures, in various groups of patients (e.g. medication naïve, after the first episode, chronic) and WM alteration has a potential to be a biomarker used as a diagnostic criterion, even in a prodromal stage of the illness (Pettersson-Yeo et al., 2011;Alba-Ferrara and de Erausquin, 2013).
Yet, regarding the existing DTI studies on the structural biomarkers of the negative symptoms in schizophrenia, the observed inconsistency of the results could be due to the high heterogeneity of clinical images between schizophrenic individuals. The lack of application of detailed clinical assessment of specific negative symptoms is also one of the most important issues.
First of all, two clinical tools were commonly used in a majority of DTI research in schizophrenia the Scale for the Assessment of Negative Symptoms (SANS, Andreasen 1982) or the negative subscale of the Positive and Negative Symptoms Scale (PANSS, Kay et al., 1987;van der Gaag et al., 2006;Liemburg et al., 2013;Stiekema et al., 2016). In brief, most of the studies involving these scales find no association between their scores and reduced WM integrity, i.e., lower FA values (Fujiwara et al., 2007;Skelly et al., 2008;Abdul-Rahman et al., 2011;Choi et al., 2011;Yan et al., 2012;Kelly et al., 2018), whereas some find negative Balevich et al., 2015;Ochi et al., 2020) or positive associations (Camchong et al., 2011;Bijanki et al., 2015).
Furthermore, in some studies (Balevich et al., 2015;Sun et al., 2015) clinical assessment did not seem to be prioritized, as the calculated associations were provided only between a total score of each scale, instead of correlations between particular subscales. The specific investigations on the relationships between diffusion parameters and clinical subscales or even single items concerning negative symptoms are still scarce.
Pieces of evidence indicate that lower FA in the left frontal lobe were related with SANS anhedonia-asociality domain (Asami et al., 2014;Ohtani et al., 2014); SANS avolition were negatively correlated with FA in the corpus callosum (Nakamura et al., 2012); PANSS (N1)-blunted affect was related with lower FA in uncinate fasciculus (Luck et al., 2011); avolition-apathy domain (The Schedule for the Deficit Syndrome (SDS); Kirkpatrick et al., 1989) negatively correlate with FA abnormalities in the reward system, i.e., amygdala-insular connections (Amodio et al., 2017); deficit patients (SDS) revealed a greater reduction in WM integrity than non-deficit patients in the corpus callosum and right posterior thalamic radiation (Tan et al., 2020); patients with persistent negative symptoms exhibit a different pattern of WM abnormalities as compared to patients without negative symptoms (Hovington et al., 2015); WM abnormalities related to the treatment resistance were also associated with the severity of the negative symptoms (Kochunov et al., 2019).
Summarizing, although consistent results revealed the existence of diverse WM abnormalities in schizophrenia in comparison to healthy controls, the association of those changes with the occurrence of the negative symptoms seems still to be not yet fully detected and understood. The above discrepancy of DTI results on negative symptomatology may be more specifically considered as an effect of the inconsistency of the clinical assessments of the same symptoms by different scales. For example, anhedonia is not included in the PANSS, while in the SANS it is rated together with asociality, but with no discrimination on anticipatory and consummatory anhedonia, of which the first one is most characteristic for schizophrenia (Marder and Galderisi, 2017;Yan et al., 2019).
Interestingly, on the contrary to the SANS and PANSS, the Brief Negative Symptom Scale (BNSS) (Kirkpatrick et al., 2011) rates avolition more precisely as based on both internal subject's feeling and observed behavior, and evaluates both consummatory and anticipatory anhedonia (Marder and Galderisi, 2017). BNSS belongs to the second generation of the negative symptoms scales and consists of 5 specific subdomains of negative symptoms: blunted affect, alogia, asociality, anhedonia, and avolition (Garcia-Portilla et al., 2015;Kumari et al., 2017;Ahmed et al., 2019;Strauss et al., 2019), which may have the potential to separate neurobiological substrates and to become new therapeutic targets (Kirkpatrick et al., 2011). The concurrent validity of both BNSS and PANSS is high and a recent study (Kaliuzhna et al., 2020) revealed that both amotivation factors reach a relatively high negative association with diminished left ventral striatal activation. However, the effect sizes for BNSS were much higher and the authors emphasize that the use of specialized scales like BNSS is crucial in MRI studies directly addressing negative symptoms (Kaliuzhna et al., 2020).
Hence, the primal aim of this study is to investigate the association between WM abnormalities and the severity of specific subdomains of psychopathological symptoms measured by PANSS and BNSS in a sample of schizophrenia outpatients, with a special focus on the specific aspects of the negative symptomatology.
First, we determined the differences in WM diffusivity parameters in the main tracts in the examined clinical group in comparison to healthy controls. Replication of the alternations in WM in a group of patients consistent with those reported in the literature (Kelly et al., 2018;Koshiyama et al., 2020) serves as a rationale for the secondary correlation analysis of the altered DTI parameters with psychopathology.
More specifically, in the present study, we investigate the severity of specific negative symptoms using the five-factor BNSS (Ahmed et al., 2019) and five-factor PANSS (van der Gaag et al., 2006) models with additional application of a 2-factor PANSS negative symptoms structure, i.e., social amotivation and diminished expression (Liemburg et al., 2013). However, BNSS may be regarded as a more specific clinical tool than PANSS, as relies on both clinical and subject internal experience (Marder and Galderisi, 2017) and the 5 subdomains construct model of negative symptoms is clinically well settled (Ahmed et al., 2019, Strauss et al., 2018. On the other hand, since PANSS is one of the most widely used clinical tools in neuroimaging research the referential value of such measurement is indispensable. Therefore, we expect more pronounced BNSS subdomains associations compared to PANSS factors. Next, based on the reported widespread alternations of WM in schizophrenia (Kelly et al., 2018;Koshiyama et al., 2020) we postulate that WM regions related to the severity of the negative symptoms may be subtle, more diffuse and not restricted only to the main tracts, but expressed beneath the cortical regions with abnormal functional activation. This assumption is supported by studies on grey matter structure and its relation to negative symptoms, eg. blunted affect (Guessoum et al., 2020) or apathy (Bègue et al., 2020). Thus, in the present study, besides commonly used tract-based spatial statistics (TBSS) analysis, we used a novel DTI approach more focused on smaller parts of the WM tracts localized beneath cortical regions of interest (ROI), which ROI-FA values will serve as the basis for secondary analyses of the relationship between WM integrity and specific psychopathology subdomains.

Subjects
The study included 30 schizophrenia outpatients (SCH) and 30 sex-, age-, and education-matched healthy controls (CON). The clinical group consisted of people with schizophrenia (27 paranoid subtype, 2 undifferentiated, 1 schizoaffective) as diagnosed with the ICD-10 by an experienced psychiatrist based on clinical interviews and medical documentation; recruited through the local network of outpatient clinics and rehabilitation centers in Krakow, Poland.
The inclusion of patients with undifferentiated schizophrenia and schizoaffective disorder was based on a clinical premise, that diagnostic subtyping has no predictive validity and will be abandoned in future classifications, e.g. ICD-11 (Reed et al., 2019) and that schizoaffective disorder shares similar WM pathologies with schizophrenia (Kaluser et al., 2016).
All participants provided written informed consent for participation in the study. Procedures were designed following the ethical standards of the World Medical Association Declaration of Helsinki (2013) and approved by the Research Ethics Committee of the Institute of Psychology, Jagiellonian University in Krakow, Poland.
All clinical subjects were in a stable psychopathological condition for at least 8-12 weeks before the assessment. The exclusion criterion was a history of head injuries, seizures, substance dependence, or any serious current somatic illnesses. Before MRI data acquisition, the PANSS, Kay et al., 1987;Van der Gaag et al., 2006;Liemburg et al., 2013 andthe BNSS, Kirkpatrick et al., 2011;Ahmed et al., 2019 were assessed by experienced psychiatrists. The mean dose of antipsychotics for each subject from the clinical group was calculated as chlorpromazine equivalents (Atkins et al., 1997;Woods, 2003;Gardner et al., 2010).
The Polish adaptation of the Montreal Cognitive Assessment (MoCA, available at www.mocatest.org; Nasreddine et al., 2005) was used as a general measure of basic cognitive skills for all of the subjects. The groups did not differ in terms of sex and age, but they did differ in years of education (CON > SCH), although this difference was not found at the educational level (Chi²=4.592; P=0.204). As expected, a difference in cognitive performance (MoCA result; CON > SCH) was found, with a lower total score in the SCH group, and revealed the cognitive impairments prominent and characteristic for schizophrenia (Adamczyk et al., 2016).
Demographic and clinical data are presented in Table I. The data were normally distributed.

Preprocessing of diffusion data
For DTI data preprocessing and analysis, the FSL package (FMRIB Software Library v5.11) was used. Anatomical images were skull-stripped with BET (Smith et al., 2006). Motion correction was performed using eddy (Andersson and Sotiropoulos, 2016) and distortion correction was performed with FLIRT (Jenkinson et al., 2002). After every step, the quality of data was checked manually by experienced researchers. Voxelwise statistical analysis of the FA data was carried out using TBSS (Smith et al., 2006). First, the tensor model was fitted to diffusion data using FDT, thus resulting in the creation of brain FA images. All subjects' FA data were then aligned into a common space with FNIRT. Next, based on the mean FA image from each subject, the mean FA skeleton was created, which represents the centers of all tracts common to the group. Each subject's aligned FA data were then projected onto this skeleton and used for voxelwise cross-subject statistics. RD (radial diffusivity), AD (axial diffusivity), and MD (mean diffusivity) data were also Subjects demographics and clinical data were presented as mean (±SD) for quantitative data. The significance level in all statistical analyses equaled P<0.05.
warped, merged, and projected onto the original mean FA skeleton. Additionally, since TBSS analysis is based on values extracted from voxels contained in the FA skeleton, which does not include the subject's entire WM, it represents only all tracts that are common to the whole group. Thus, besides standard TBSS between-group analysis, we focused on selected regions of interest (ROI) containing cortex and adjacent WM, and extracted values from WM skeletons voxels within those areas (individually for each subject). We assume that correlations between these ROI-FA values and psychopathology can provide more reliable results.
Therefore, to obtain cortical projections of WM results, we decided to test between-group differences in FA values within 17 selected cortical ROI-FA. Regions were selected according to previously published data on brain abnormalities in schizophrenia and revealed as relevant attribution to negative symptom (Shaffer et al., 2015;Abram et al., 2017;Walton et al., 2018;Li et al., 2018;Brady et al., 2019;Bègue et al., 2020;Guessoum et al., 2020).
The masks for each ROI were created separately and fitted to the common space. The shared voxels of each ROI mask and the respective mean FA skeleton were extracted individually for each subject. Namely, the FA values from each mask in each subject's FA skeleton image were extracted, averaged, and used for statistical ROI-FA analysis. We decided to obtain only the FA measure from masks as this parameter is widely used in other publications concerning WM and psychopathology in schizophrenia, but not MD, RD, or AD.

Statistical analysis
Voxelwise DTI analyses were performed using nonparametric permutation-based testing with the Randomise command (Winkler et al., 2014) controlling for sex, age and illness duration. The Threshold-Free Cluster Enhancement (TFCE) method was used with Familywise Error (FWE) correction; 10.000 permutations were calculated. P<0.001 were considered significant.
Between-group comparison of ROI-FA values (t-tests) was performed using IBM SPSS Statistics for Windows (version 23). The effect size was calculated with Cohen's d with small sample size correction. False discovery rate (FDR) (Benjamini and Hochberg, 1995) was used for multiple testing corrections for all extracted ROIs threshold at α level P<0.05.
The relationships between PANSS, BNSS and ROI-FA values were computed using partial Spearman rank-order correlation, controlling for sex, age, illness duration and medication (chlorpromazine equivalent). Besides the calculation with standard five factors PANSS (Kay et al., 1987;Van der Gaag et al., 2006) i.e., positive symptoms, negative symptoms, disorganization symptoms, excitement and emotional distress, the additional analysis was conducted, where the negative symptoms were divided into 2 subdomains -expressive deficits and social amotivation (Liemburg et al., 2013). For BNSS 5-factor model was applied, which includes blunted affect, alogy, anhedonia, avolition and asociality subdomains (Strauss et al., 2018;Mucci et al., 2019). The FDR correction for multiple tests with α level P<0.05 was used.

RESULTS
The DTI TBSS results of the between-group comparisons of FA, MD, RD and AD parameters showed widespread differences across most of the WM bundles in the brain (Fig. 1 left-panel). In particular, the FA values were lower and MD, AD, RD were higher for the SCH group as compared to the CON group (P<0.001, TFCE, FWR corrected). The structures in which all the parameters changed (FA, MD, RD, AD) were identified bilaterally (i.e., the body and splenium of the corpus callosum, the medial lemniscus, the anterior, posterior and retrolenticular limb of internal capsule, the superior corona radiata, the sagittal stratum, the external capsule, the fornix); they were also identified selectively in the left (i.e., the anterior corona radiata) and right hemisphere (i.e., the superior longitudinal fasciculus, the corticospinal tract, the posterior corona radiata).
Next  Table II and Fig. 1 (right-panel).
Finally, the correlation analysis of the ROI-FA values with psychopathology has been provided in the above ROIs in which significant between-group differ-ences were found. The negative association between asociality BNSS subscale and several regions after FDR correction for multiple comparisons was revealed. In particular, increased asociality was associated with decreased FA in the left oPFC (r=-0.46, P=0.043), pCC (r=-0.56, P=0.016), Prec (r=-0.49, P=0.035), and the right pCC (r=-0.54, P=0.016), HG (r=-0.54, P=0.016), and Prec (r=-0.70, P<0.001). The detailed correlation results are presented in Table III

DISCUSSION
In the present study, we aimed at investigating the relationship between the WM integrity abnormalities and psychopathological symptoms in schizophrenia using BNSS and PANSS scales.
First of all, results of between-group comparisons on DTI parameters and ROI-FA values confirmed the existence of essential bilateral and widespread disruptions of WM cytoarchitecture in schizophrenia in line with previous studies (Parnanzone et al., 2017;Kelly et al., 2018;Koshiyama et al., 2020). Thus, the clinical group of outpatients investigated in our study may be regarded as representative, as sharing the abnormal patterns of WM changes similar to those revealed by big-data study on a cohort of people with schizophrenia, e.g. these found in the major tracts of the corpus callosum and the corona radiata (Kelly et al., 2018). Moreover, the presented results on differences in all diffusivity parameters (FA, AD, MD, RD) with the most widespread RD WM alterations are in line with the recent study on 696 schizophrenia patients (Koshiyama et al., 2020). In particular, we found widespread bilateral differences, which indicates a breached microstructural integrity in schizophrenia, i.e., FA values were lower and RD, AD and MD values were higher. Consistently, results from ROI-FA values revealed various bilateral differences in the frontal, temporal and parietal lobes. Moreover, we found a higher general level of diffusivity (MD) in the schizophrenia group in the same regions as the FA-related findings described above, with additional changes in the cingulum and the uncinate fasciculus. Noteworthy, consistent with big-sample studies (Kelly et al., 2018;Koshiyama et al., 2020) the RD changes were the most distributed DTI parameter, as compared to FA, AD, and MD measures. These findings support the evidence that the diffusivity changes in schizophrenia are most possibly a result of myelin disruptions (Cassoli et al., 2015;Mighdoll et al., 2015). The RD is considered a more sensitive measure to detect abnormalities related to myelin disruptions than FA (Joo et al., 2020). This is especially evident in regions that demonstrate a significant number of coherently oriented axons, e.g. corpus callosum (Karlsgodt, 2016). Moreover, molecular studies showed that the dysfunction of oligodendrocytes has an impact on disrupted myelination processes in schizophrenia (Cassoli et al., 2015;Mighdoll et al., 2015). Although, further studies are needed to better determine this phenomenon. At last, the differences that appeared along the axis (AD) indicate the impact of other factors that contribute to WM abnormalities, e.g. an inappropriate neurodevelopmental environment that results in an abnormal process of establishing synapses and pruning axonal connections (Al-ba-Ferrara and de Erausquin, 2013; Klauser et al., 2016) seemingly provided to the brain disconnection (Friston et al., 2016).
Overall, it should be pointed that presented differentiated findings on the WM abnormalities distribution indicated by various measures (i.e., FA, MD, RD, AD) is seemingly related to the fact, that each of those DTI parameters were concerned specifically sensitive to different biological phenomena (e.g. FA -WM integrity, RD -WM myelination level, AD -WM axonal damage; MD -WM surrounding cytoarchitecture). Although, such an interpretation of the tissue properties based on the DTI parameters is still under debate (Wheeler-Kingshott and Cercignani, 2009). To investigate the exact mechanism underlying different alternations of WM in schizophrenia, further research is needed to better determine the biological characteristics of the brain tissues and their relation to the specific DTI measure (Kelly et al., 2018).
Second, the significant effects on the specific associations of the BNSS asociality with a lower WM integrity was found in the left oPFC, right HG, and bilateral pCC and Prec. This indicates that the abnormal structure of these regions may be concerned as a neurobiological substrate of functional disturbances crucial for the manifestation of specific psychopathological symptoms, i.e., asociality. This is consistent with the previous DTI study (Viher et al., 2016) which revealed that the negative symptoms (DSM-V) were related to the disrupted WM structure in bilateral prefrontal cortices and the right temporal lobe.
However, to the best of our knowledge, apart from the presented findings, there is a lack of other research investigating the relationship between schizophrenia symptoms and abnormalities of WM beneath the specified cortical regions (e.g. ROI-FA). Although, our results are supported by recent research that indicates that alterations of WM tracts may be considered to represent neural underpinning of changes in structural properties of grey matter regions, which are inextricably interconnected at a cellular level. In particular, reductions in cortical thinning were shown to be related to the increased FA in the intracortical WM (Di Biase et al., 2019), elevated RD in adjacent WM was associated with increased cortical folding in the dorsolateral PFC (Schultz et al., 2017) and reduced volume of WM was shown to be related to the volume reduction in the neighboring regions of grey matter (Colibazzi et al., 2013).
Considering abnormal brain structure in schizophrenia, some studies revealed pronounced fronto-temporo-parietal cortical thinning (right STG/ TPJ, parahippocampal gyrus, and cingulate cortex) (Bodnar et al., 2014), or smaller volume of the right parahippocampal gyrus and STG (Benoit et al., 2012) in schizophrenia patients with persistent negative symptoms. At last, the left medial-oPFC cortex thinning was found to be selectively related to negative symptoms severity . Interestingly, in a longitudinal study involving at-risk for psychosis adolescents, negative symptoms were found to be related to the increased grey matter loss in the left hemisphere, e.g, STG/SupG, pCC, cerebellum and limbic lobe (McKechanie et al., 2016).
Furthermore, in the case of the functional MRI studies, hypoactivation of the Prec and the pCC was found to be related to the higher scores on the SANS anhedonia/asociality subscale (Shaffer et al., 2015;Guessoum et al., 2020) and avolition/apathy domain (Shaffer et al., 2015), and reduced resting-state functional connectivity in the precuneus was found to be positively correlated with avolition-apathy domains in schizophrenia patients (Forlim et al., 2020). Finally, the above-mentioned disconnectivity between large scale networks, such as frontal-cingulate-parietal connections within the default mode network (DMN) may be considered crucial for negative symptoms manifestation in schizophrenia (Lefort-Besnard et al., 2017).
Noteworthy, the medial PFC, temporal lobe, pCC and precuneus are considered as an essential part of DMN and a social network crucial for social cognition, which deficits partially overlap the asociality concept (Millan et al., 2014;Pelletier-Baldelli and Holt, 2020). Interestingly, the neurobiological basis of the structural and functional associations between precuneus and asociality is recognized to be dependent on the oxytocin level (Churchland and Winkielman, 2012;Kumar et al., 2015;Strauss et al., 2015). In particular, previous studies showed that the decreased plasma oxytocin level is associated with increased asociality (Strauss et al., 2015) and administration of intranasal oxytocin can reduce asociality symptoms in patients (Churchland and Winkielman, 2012). Complementary, other studies showed that enhancement of the oxytocin transmission can affect functional connectivity between precuneus -amygdala (Kumar et al., 2015) or precuneus -left dorsolateral prefrontal cortex (Kumar et al., 2019) connections. The above findings support the theory on the contribution of the precuneus to the social cognition in schizophrenia, mediated by oxytocin level decrease, which is strongly related to one related deficit of social cognition -the severity of asociality (Guessoum et al., 2020).
Yet, our results on altered WM structure beneath the left oPFC, right HG, bilateral precuneus and pCC seem to be complementary with findings on disturbed oxytocin-related functional connectivity in the DMN, which may be regarded as a potential mechanism of the neu-ral substrate of social cognition deficits in schizophrenia and/or manifestation of asociality-related symptomatology. The pCC and precuneus are midline areas involved in higher-order and social processing (Cavanna and Trimble, 2006;Leech et al., 2012). Therefore, diminished WM connections in these regions might be responsible for disruption in social-motivation processes and turn resulting in increased asociality. This is supported by previous findings on the specific relationship of the anhedonia-asociality domain (assessed with SANS) with disrupted WM integrity in the left superior fronto-occipital fasciculus (Asami et al., 2014), the left posterior oPFC -ACC connections (Ohtani et al., 2014) or right cingulum bundle . Finally, the results of the presented BNSS correlations indicate that alterations in the smaller parts of tracts located beneath grey matter (ROI-FA) may be an effective approach for the measurement of the specific associations of brain regions with negative symptoms.
On the other hand, no significant correlation with the negative symptoms subscale of PANSS was found, neither within the five-factor model of PANSS subscales (Van der Gaag et al., 2006), nor the analysis of the additional two specific subdomains of the PANSS negative symptoms, i.e., social amotivation and expressive deficits (Liemburg et al., 2013). The presented BNSS and PANSS results may indicate that the relation of WM disturbances with negative symptoms was indeed indirect and narrow down to specific deficits, e.g. BNSS asociality subscale; supporting the previously mentioned discrepancies between clinical scales (Kumari et al., 2017;Marder and Galderisi, 2017). Importantly, regarding the concurrent validity for the BNSS, SANS or PANSS, it should be pointed out that these clinical scales are highly inconsistent and measure slightly different aspects of the negative symptomatology (Kirkpatrick et al., 2011); e.g. asociality is often understood as social amotivation and therefore is treated by many researchers as one of the elements of the avolition/apathy domain (Kaiser et al., 2017). Interestingly to this issue, an exceptional DTI study on schizophrenia which applied BNSS (Stämpfli et al., 2019), despite revealed subtle between-group fiber density (FD) differences, reported no correlation between negative symptoms, neither for FA nor for FD parameters. Although, in contrast to the present study, in the study of Stämpfli and colleagues (2019), no commonly observed FA WM alterations were found (Kelly et al., 2018) and the BNSS analysis was limited to the two-factor model, i.e., apathy and diminished expression, beyond current five-factor model of BNSS subdomains analysis (Ahmed et al., 2019).
Summarizing, due to some limitations of our study (e.g. limited sample size), the presented results should be interpreted with caution, especially the correlation analysis. Nevertheless, the presented results may serve as a valuable input for a deeper understanding of WM structure changes and its relation to the specific psychopathology symptoms in schizophrenia. Noteworthy, in the present study on schizophrenia outpatients, we detected characteristic differences in WM parameters in schizophrenia, which served as a rationale for provided correlation analysis of specific PANSS and BNSS symptoms subscales, as of great importance. The presented results of the ROI-FA calculation gave a deeper insight into the associations between WM structure abnormality and the manifestation of psychopathology symptoms. However, the used ROI-FA approach, similar to other methods (e.g. TBSS method focusing on the major WM tracts) is still affected by the problems with precise abnormal brain region alignment and localization accuracy. For a more profound understanding of the nature of WM abnormalities in schizophrenia and its relation with psychopathology, the more sophisticated methods need to be incorporated in further studies, e.g. ROI specified tractography and structural connectivity analysis. The more precise analysis of FA values on particular WM tracts may provide more precise measures of WM abnormalities in schizophrenia. Although, the replication of presented results on a bigger sample in the future is required.
Furthermore, even in our analyses, we controlled the parameters that potentially affected WM integrity in schizophrenia, i.e., sex, age, illness duration, and dosage of antipsychotic medication, these issues require special attention, as still under scientific debate. On the one hand, some studies indicated that WM integrity may be affected by the type of medication or illness duration (Ozcelik-Eroglu et al., 2014;Samartzis et al., 2014). On the other hand, the recent studies on big-sample sizes revealed that the dose of medication (calculated as the chlorpromazine equivalent) did not affect the diffusivity parameters of WM (Kelly et al., 2018;Koshiyama, 2020;Joo et al., 2020;Gurholt et al., 2020).
Nevertheless, another problematic issue concerns nowadays the open-source, big databases where usually lack the full clinical information of patients, e.g. PANSS, BNSS results. Thus, our research strongly suggests the necessity of implementing consistent and more specific clinical scales, e.g. five-factor BNSS, as standard for psychopathology measurement in clinical DTI investigations on schizophrenia.
To summarize, further study on this topic is highly required. In particular, multidisciplinary studies combining functional and structural approaches in one investigation may be an important prospect that allows analysis of the relations between WM abnormalities, altered synaptic transmission, genes, and associated my-elination disturbances and psychopathology in schizophrenia. The better organization of the recent knowledge on the relation between the negative symptoms and their biological underpinnings may contribute to the development of more effective neurotherapeutic interventions, i.e., state-of-the-art neurostimulation methods aimed at psychopathological symptoms reduction, e.g. TMS or tDCS, which methods potentially may effects in an improvement of social functioning.

CONCLUSIONS
In the present study, we revealed the association between disrupted WM integrity in the fronto-temporo-parietal regions (the left oPFC, right HG, bilateral Prec and pCC) and specific negative symptoms in schizophrenia, i.e., asociality. Further studies on this topic require big neuroimaging data set and implementation of more specific, consistent and more homogenous clinical scales used in DTI studies on schizophrenia, e.g. five-factor BNSS symptoms scale.