The Lack of Alterations in Metabolites in the Medial Prefrontal Cortex and Amygdala, but Their Associations with Autistic Traits, Empathy, and Personality Traits in Adults with Autism Spectrum Disorder: A Preliminary Study

Autism spectrum disorder (ASD) is defined by deficits in social communication and social interaction (social-emotional reciprocity, nonverbal communicative behaviors, or developing and maintaining relationships), and restricted, repetitive patterns of behavior, interests or activities. It is suggested that those with autism have difficulties with emotional relationships due to deficits in empathy, which is hypothesized to have both cognitive and affective components. Cognitive empathy is the capacity to understand other people’s feelings, intentions, and beliefs on an intellectual level, while affective empathy is the emotional response to other people’s affective states or feelings (Davis, 1983; Singer, 2006). Adults with ASD and no intellectual disability showed impairments of cognitive empathy in scores for perspective taking on the Interpersonal Reactivity Index (IRI) (Lombardo et al., 2007; Rogers et al., 2007). In our recent study, adults with ASD and no intellectual disability showed impairments of cognitive empathy rather than affective empathy on the IRI and Questionnaire of Cognitive and Affective Empathy (QCAE), and personality traits on the NEO Personality Inventory-Revised (NEO), and multiple regression analysis demonstrated that perspective taking on the QCAE and extraversion on the NEO were good predictor variables to autistic traits on the Autism-spectrum quotient (AQ) (Shirayama et al., 2022).

The neural basis of empathy is reported to involve the anterior cingulate and insula (Bernhardt & Singer, 2012). Cognitive empathy is associated with the medial prefrontal cortex and cingulate cortex, whereas affective empathy is associated with the insula (Eres et al., 2015; Fan et al., 2011). Cognitive empathy may partially involve the mechanisms underlying the theory of mind, which is shown to be associated with the medial prefrontal cortex and temporoparietal junction (Schurz et al., 2014). The medial prefrontal cortex is activated in the context of mentalizing or theory of mind tasks when people are attending to certain states of the self or others (Frith and Frith,  1999; Jackson et al., 2006), and the anterior cingulate cortex plays an important role in the executive control of attention (Fan et al., 2005).

Cognitive empathy deficits in ASD could involve impairments in social cognition, in which the amygdala might be implicated (Blair et al., 2008). The amygdala is required to recognize emotions, social interactions, and evaluating the social value of a stimulus (Adolphs, 2001). The social brain's neural network includes the amygdala, suggesting the amygdala theory of autism (Baron-Cohen et al., 2000; Nacewicz et al., 2006; Schulkin, 2007). The developmental dysfunction of the orbitofrontal-amygdala circuit is critical for the deficits in self-regulation of social-emotional behavior in autism (Bachevalier & Loveland, 2006). When emotional face processing was needed, the amygdala showed reduced activation in autistic subjects (Pierce et al., 2001). In another study, adult individuals with high-functioning ASD, including Asperger syndrome, showed less activation of the amygdala and frontal cortex when processing facial expressions (Critchley et al., 2000).

A study of the default mode network of ASD subjects found that the strength of the resting-state functional connectivity came from the core region of the medial prefrontal cortex in ASD participants (Jung et al., 2014). Social cognitive dysfunction such as self-referential processing and theory of mind involved the medial prefrontal cortex functional connectivity in an adult group of ASD (reviewed by Padmanabhan et al., 2017). In a PET study, subjects with ASD showed glucose hypometabolism in the anterior cingulate cortex, but not in the amygdala (Haznedar et al., 2000). Functional MRI studies reported enhanced amygdala activation in ASD subjects (Dalton et al., 2005; Kleinhans et al., 2009a), and that amygdala was involved in interpersonal space permeability and flexibility in ASD adults (Massaccesi et al., 2021). Whereas, individuals with ASD showed diminished activity in the medial prefrontal cortex and insula for conflicting nonverbal information (Watanabe et al., 2012), and activation in the medial prefrontal cortex during thresholded unsigned prediction errors (Mosner et al., 2019). Interestingly, the medal prefrontal cortex was activated in individuals with ASD in response to autistic characters and in typically developing individuals in response to non-autistic characters (Komeda et al., 2015). Furthermore, oxytocin increased the originally diminished brain activity in the medial prefrontal cortex in ASD (Watanabe et al.,  2014). These areas predicted autistic communication deficits in ASD. An MRI study showed a decreased-volume medial prefrontal cortex was shown in high-functioning ASD subjects compared to normal controls (Haznedar et al., 2000). However, studies of amygdala volumes in adults with ASD have shown conflicting results: reduced amygdala volume in some (Nacewicz et al., 2006; Pierce et al., 2001) and increased volume in others (Abell et al., 1999; Howard et al., 2000), with one study showing no difference (Haznedar et al., 2000). It is of note that the gray matter volume of the amygdala was correlated positively with cognitive and affective empathy, but negatively with cognitive alexithymia (Goerlich-Dobre et al., 2015).

Glutamatergic signaling, particularly as it relates to synaptic plasticity, is believed to be involved in the development of ASD (Veenstra-VanderWeele & Blakely, 2012). Previous studies utilizing 3 T Proton magnetic resonance spectroscopy (¹H-MRS) demonstrated that adults with ASD had a decreased glutamate in the anterior cingulate cortex (Tebartz van Elst et al., 2014) or no changes in glutamate in the medial prefrontal cortex (Horder et al., 2018). Meanwhile, three studies have reported no differences in glutamate plus glutamine (Glx) in the medial prefrontal cortex of ASD patients compared to controls (Aoki et al., 2012; Endres et al., 2017; Horder et al., 2018), whereas two studies showed decreased levels of Glx in the medial prefrontal cortex of ASD subjects (Bernardi et al., 2011; Tebartz van Elst et al., 2014). Recently, cannabidiol increased Glx in the medial prefrontal cortex, but decreased GABA in the adults with ASD using MRS (Pretzsch et al., 2019). Neuronal network hyperexcitability in cerebral cortex autism was supposed glutamate and GABA through maternal metabolic morbidity (Rivell & Mattson, 2019).

Adults with ASD showed increased N-acetylaspartate (NAA) levels in the medial prefrontal cortex (Aoki et al., 2012; Murphy et al., 2002), or increased NAA/choline ratio in the anterior cingulate cortex (Oner et al., 2007), whereas other studies showed significantly decreased NAA signals in the pregenual anterior cingulate of ASD adults (Tebartz van Elst et al., 2014), or a reduction in the NAA/creatine ratio in the anterior cingulate cortex of ASD adults (Libero et al., 2015), and no changes in NAA levels in the prefrontal cortex of ASD (Endres et al., 2017). NAA levels are interpreted as a marker of neuronal density and/or mitochondrial function (Clark, 1998). On the other hand, there was one report of no abnormalities in NAA levels in the amygdala of high-functioning ASD and Asperger syndrome (Kleinhans et al., 2009b).

Choline-containing compounds, glycerophosphorylcholine plus phosphorylcholine (GPC + PC), represent membrane turnover. GPC is one of the breakdown products of phospholipids, whereas PC is a precursor of the phospholipid membrane (Shirayama et al., 2004). High levels of GPC + PC were also reported in the medial prefrontal cortex of Asperger syndrome (Murphy et al., 2002). However, there were no significant changes in Cho (GPC + PC) in the amygdala of high functioning ASD and Asperger syndrome (Kleinhans et al., 2009b).

High concentrations of creatine plus phosphocreatine (Cr + PCr) were reported in the medial prefrontal cortex of Asperger syndrome (Murphy et al., 2002). On the contrary, no significant difference in Cr + PCr was shown in the amygdala of high functioning ASD and Asperger syndrome compared to controls (Kleinhans et al., 2009b).

Myo-inositol is a putative marker of glial cells because myo-inositol is transiently but actively transported into astrocytes (Barres, 2008). Reduced levels of myo-inositol were shown in the anterior cingulate of adults with high-functioning ASD compared to controls (Endres et al., 2017), whereas three studies showed no significant differences in myo-inositol levels in the medial frontal cortex of ASD (Aoki et al., 2012; Bernardi et al., 2011; Tebartz van Elst et al., 2014). On the other hand, one study showed no significant difference in myo-inositol in the amygdala of ASD adults (Kleinhans et al., 2009b).

Increases in magnetic field strengths make it possible to acquire proton (¹H) spectra from a smaller volume of interest (VOI) with fewer scan numbers than before (Shirayama et al., 2010). Recently, reliable results from the amygdala can now be obtained using new techniques with saturation bands for shimming (Nacewicz et al., 2012). Utilizing 3 T ¹H-MRS and saturation bands, we obtained acceptable measures of glutamate from the amygdala and the medial prefrontal cortex using a volume of 8 cm³ (Shirayama et al., 2017).

This study aimed to investigate the levels of NAA, glutamate, Glx, Cr + PCr, GPC + PC, myo-inositol, and glutamine in the medial prefrontal cortex and amygdala of adults with ASD and no intellectual disability using 3 T ¹H-MRS and saturation bands. Next, we examined the relationship between brain metabolites, and autistic traits on the AQ, cognitive and affective empathy scores on the QCAE and IRI, and personality traits on the NEO.

The main purpose of this study was to examine the differences in brain metabolite in the brain between ASD adults and non-ASD controls. As shown in Table 7, we found no comprehensive differences in brain metabolites concentrations in the medial prefrontal cortex and amygdala of ASD adults compared with non-ASD controls. No difference in glutamate and Glx levels in the medial prefrontal cortex supports past studies (Murphy et al., 2009; Bernardi et al., 2011; Aoki et al., 2012; Endres et al., 2017; Horder et al., 2018). However, previous studies of ASD adults showed reduced Glx in the medial prefrontal cortex (Bernardi et al., 2011; Tebartz van Elst et al., 2014), and a decrease in glutamate in the anterior cingulate cortex (Tebartz van Elst et al., 2014).

¹H-MRS studies failed to demonstrate consistent results regarding levels of brain metabolites in patients with high functioning ASD or Asperger’s syndrome. The reasons underlying the discrepancies for glutamate and Glx of ASD in the MRS study are currently unclear. This might be due to the subtle distinction in the location of interest: dorsal or ventral medial prefrontal cortex or anterior cingulate cortex. Also, magnetic field strength, scanning methods such as PRESS and STEAM, and particular sequences such as J-resolved point-resolved spectroscopy or J-editing acquisition may be factors for differentiation. Future studies will be needed to address these results.

Next end of this study was to examine the relationships between brain metabolites and psychological features detected by QCAE, IRI and NEO. We found significant correlations of brain metabolites with scores for autistic traits (AQ), empathy (QCAE and IRI) and personality (NEO) among ASD subjects, detected by Pearson correlations. The present study of ASD adults used the same data with our previous study (Shirayama et al., 2022). Significant findings by Pearson parametric correlations were checked for additional nonparametric Spearman's correlation coefficients. Further, Bonferroni correction was done for multiple comparisons when appropriate. Additionally, comparing correlations by cocor were examined to check differences in correlations between the two groups. We think that even though data did not pass Bonferroni correction, statistical differences by cocor will be helpful for future studies. The findings should be taken as preliminary or exploratory. Thus, these data interpretations may be based on effect sizes rather than statistical significance.

Glutamate and Glx in the medial prefrontal cortex in the ASD adults showed significant correlations with empathic concern on the IRI with showed significance, passing Bonferroni corrections thereafter. Furthermore, comparison of correlations by cocor showed that the correlation of glutamate in the medial prefrontal cortex with empathic concern on the IRI in ASD adults was significantly different from that in non-ASD controls. Notably, glutamate in the anterior cingulate was increased in children with ASD (Bejjani et al., 2012) and adolescents with ASD (Joshi et al., 2013). Therefore, it could be that there exists a glutamatergic dysfunction in the medial prefrontal cortex of adults with ASD.

One perspective on the pathophysiology of ASD is the excitation/inhibition imbalance theory, which proposes a relatively high ratio of excitatory to inhibitory neuronal processes (Rubenstein & Merzenich, 2003). In support of this excitation/inhibition imbalance theory, some findings for GABAergic inhibitory deficits have been found in adults with ASD (Coghlan et al., 2012). However, so far, there have been no supporting data for elevated glutamatergic signals in ASD adults. Therefore, the significant correlations of empathic concern on the IRI with glutamate and Glx in the medial prefrontal cortex of ASD adults might support glutamatergic dysfunction in adults with ASD.

Autistic traits on the AQ scores in the ASD adults showed substantially significant relationships with glutamate and Glx in the amygdala. Further, comparing by cocor showed the correlations of correlation of AQ scores with Glx in the amygdala had a trend for significance between the ASD and non-ASD groups. On the contrary, autistic traits on the AQ in the non-ASD showed substantially significant relationships with Glx in the medial prefrontal cortex. Further, comparing by cocor showed that the correlation of AQ scores with Glx in the medial prefrontal cortex had a trend for significance between the ASD and non-ASD groups. The new appearance and lack of correlations in the two groups could play a role in autistic traits on the AQ. Future studies will be needed to address these issues from a specific developing point of view.

Also, Glx in the medial prefrontal cortex of non-ASD exhibited significant correlations with neuroticism and agreeableness on the NEO. However, these correlations of non-ASD did not pass Bonferroni correction, and comparing correlations by cocor showed no differences between two groups. Here, it is noteworthy that neuroticism on the NEO was significantly altered in adults with ASD compared with non-ASD controls. A previous study showed that neuroticism was associated with resting-functional connectivity between the amygdala and medial prefrontal cortex following acute stress (Wang et al., 2018). If this mechanism between stress and neuroticism was conducted via Glx in the medial prefrontal cortex, the lack of correlation in the ASD adult might be related to alterations in neuroticism of ASD adults. Similar things could be said, such as the lack of correlation of agreeableness and openness on the NEO with glutamate and Glx in the in the medial prefrontal cortex of ASD adults, to a lesser extent. Future studies will be needed to elucidate this speculation.

NAA in the amygdala of non-ASD was correlated with agreeableness on the NEO. Although this correlation did not survive Bonferroni corrections, comparing by cocor showed that two correlations of NAA in the amygdala with agreeableness on the NEO were statistically significant between the ASD and non-ASD groups. The lack of correlations in the ASD might make sense to understand the pathophysiology of ASD. NAA is synthesized in the mitochondria of neuron, and catabolized in glia (Rae, 2014). These results could be a key for future studies.

Myo-inositol in the amygdala of ASD adults showed significant relationships with agreeableness on the NEO. Although this correlation did not survived Bonferroni corrections, comparing by cocor showed that the correlation of ASD adults was significantly different from that of non-ASD control. Myo-inositol is involved in phospholipid metabolism as a second messenger in the phophatidylinositol cycle (Kim et al., 2005), and is actively transported into astrocytes, highlighting its role as a biomarker for astrocyte (Barres, 2008). Astrocyte in the amygdala could play a role in the agreeableness of NEO in adults with ASD. via myo-inositol. Future studies will be needed to elucidate this issue.

GPC + PC levels in the medial prefrontal cortex of ASD adults showed significant relationships with online simulation and peripheral responsivity on the QCAE, and with extraversion and openness on the NEO. It is of note that online simulation on the QCAE and extraversion on the NEO were significantly altered in adults with ASD. It is likely that GPC + PC levels in the medial prefrontal cortex of ASD adults are involved in online simulation on the QCAE, and in extraversion on the NEO, However, these correlations did not pass Bonferroni correction, and were not significantly different between the two groups by comparing cocor. GPC + PC reflects membrane phospholipid turnover, and comes from phophatidylcholine and lysophosphatidylcholine via phosphodiesterase including phospholipase C, A2 and lysophospholipase. Recently, phosphodiesterase inhibitors have been proposed to treat ASD (Delhaye & Bardoni, 2021). Therefore, GPC + PC in the medial prefrontal cortex might be a good marker for ASD in future studies.

Cr + PCr in the medial prefrontal cortex of non-ASD showed significant relationships with autistic traits on the AQ, passing the Bonferroni correction thereafter. Further, comparing correlations by cocor showed that the correlation of Cr + PCr in the medial prefrontal cortex with AQ scores in the non-ASD control was statistically different from that of ASD adults. In a typical development, autistic traits on the AQ could be manipulated by Cr + PCr in the medial prefrontal cortex. Conversely, the lack of the correlation of Cr + PCr in the medial prefrontal cortex with AQ scores in ASD adults might be contributed to autistic traits in ASD. PCr is phosphorylated Cr, serving as a reserve of high energy phosphates and converting ADP into ATP where ADP is a major regulator of mitochondrial respiration (Shirayama et al., 2004).

On the other hand, Cr + PCr in the amygdala of non-ASD showed significant correlations with online simulation, peripheral responsivity and affective empathy on the QCAE. Although this correlation did not survive Bonferroni corrections, comparing by cocor showed that the correlations of Cr + PCr in the amygdala with online simulation, peripheral responsivity and affective empathy on the QCAE were statistically significant between the ASD and non-ASD groups. Since online simulation, peripheral responsivity and agreeableness on the NEO were significantly altered in ASD adults, the lack of these correlations might be contributed to alterations in online simulation, peripheral responsivity and agreeableness on the NEO in ASD adults. Future studies will be needed to address these issues from a specific developing point of view.

On the contrary, Cr + PCr in the medial prefrontal cortex of ASD adult showed significant relationships with empathic concern on the IRI. Also, Cr + PCr in the medial prefrontal cortex showed a weak correlation, although not significantly, with emotion contagion on the QCAE. Interestingly, empathic concern on the IRI showed a significant relationship with emotion contagion on the QCAE among ASD adults (Shirayama et al., 2022). However, these correlations did not survive Bonferroni corrections, and the comparison of correlations by cocor did not show statistical significance. Therefore, Cr + PCr in the medial prefrontal cortex might play a role in empathic concern on the IRI and emotion contagion on the QCAE in ASD adults to a lesser extent. Future studies will be needed.

From a psychological point of view, it should be noted that in this study, perspective taking on the QCAE and IRI failed to exhibit any correlations with brain metabolites in the ASD adults and non-ASD controls. It is noteworthy that perspective taking on the QCAE and IRI were significantly altered between the two groups, and perspective taking on the QCAE was a good predictor of autistic traits on the AQ in our recent study (Shirayama et al., 2022). Previous studies reported that perspective taking is associated with the medial prefrontal cortex (D'Argembeau et al., 2007) and that glutamate in the dorsolateral prefrontal cortex, but not the anterior cingulate, correlated perspective taking on the IRI (Montag et al., 2008). It might be that perspective taking of empathy factor is not manipulated by glutamatergic system in the medial prefrontal cortex and amygdala. Future studies will be needed.

This study addressed a gap in the literature about the investigation of brain metabolites by MRS in the medial prefrontal cortex and amygdala of ASD adults and showed substantially or statistically significant associations with autistic traits, empathy, and personality traits. The reasons are as follows: the number of MRS studies of adult ASD was small, the results of past studies were inconsistent, and the number of studies about associations of brain metabolites with autistic traits, empathy, and personality traits in adult ASD was small. A great deal of instability might be acquired using macromolecule removal or additional methods including multiecho techniques.

This study has some limitations. First, sample sizes are small. Second, participants of the two groups showed small differences in IQ between adults with ASD subjects and non-ASD controls despite recruiting participants without intellectual disability (full IQ > 80) (Supplementary Results and Discussion). Third, the ASD group showed a depressive state despite not suffering from depression (Supplementary Results and Discussion). Forth, the large number of comparisons made with clinical phenotype might underestimate the amount of spurious signal that might be present. The findings should be taken as preliminary or exploratory. Finally, there remains a possibility that technical challenges may significantly impact interpretation, which is not clearly articulated nor expressed in sufficient depth because there is difficulty in distinguishing a potentially real relationship from shared methodological variance.

Adults with ASD showed no differences from non-ASD controls in glutamate, Glx, NAA, GPC + PC, Cr + PCr, or myo-inositol levels in the amygdala or medial prefrontal cortex. However, ASD subjects did show significant correlations of localized brain metabolite levels with autistic traits, empathy components, and personality traits by standard psychological assessments. When Bonferroni corrections were done for these results, the correlations of empathic concern on the IRI with glutamate and Glx in the medial prefrontal cortex of ASD were statistically significant, and the correlation of AQ scores with Cr + PCr in the medial prefrontal cortex of non-ASD was statistically significant.