In this study, we combined structural covariance and graph-theory analyses to investigate the associations between mild behavioral impairment (MBI) as identified by the MBI-C and the topological properties of the structural brain networks in patients with a biologically-defined Alzheimer’s Disease (AD). Our investigation unveiled significant correlations between MBI sub-scores and network properties in specific brain regions associated with behavioral regulation. In particular, MBI-affective dysregulation scores were inversely correlated with the degree and the nodal efficiency of the orbital part of right the inferior frontal gyrus (IFG). Positive correlations, instead, were found between the regional network properties of the right posterior cingulum cortex (PCC) and MBI-impulse dyscontrol subscores (degree) and MBI-social inappropriateness scores (degree and nodal efficiency).
Our findings provide new insights into our understanding of the associations between brain network properties and MBI across the AD spectrum. In particular, a preserved small-world organization of structural brain networks mirrors previous findings observed in cognitively normal subjects with MBI (25). This provides additional evidence in favor of the theory positing that the small-world topology serves as a fundamental principle governing the organization of networks in the human system (65–69).
At the local level, the presence of affective dysregulation was associated with a decreased segregation and integration ability of the IFG in the information processing across brain networks. The IFG has been shown to be involved in depression by means of functional (70–76) and structural connectivity studies (76–79). In particular, alterations of the right IFG structural connectivity were revealed by DTI in late-life depression in subjects both cognitively normal (76) and suffering from memory deficits (78, 79). Concerning previous studies on MBI, Shu et al. found that betweenness centrality, a local connectivity indicator that measure the influence of a node on the information flow, was lower in frontal brain regions encompassing the right IFG when patients with MBI were compared to controls (25). In line with this study, our findings highlight the key role of the IFG in processing and transmitting affective information, providing further evidence that a reduced centrality in this brain region may be associated with affective dysregulation.
Concerning the impulse-dyscontrol and social inappropriateness domains, positive associations were found between the MBI subscores and the centrality and integration ability of the right PCC. The PCC is considered a crucial hub for the posterior Default Mode Network (DMN) regulation (80, 81) and is known to be heavily involved in AD progression (82–85). Most importantly, both structural and functional alterations of the PCC have been specifically attributed to the presence of NPS in AD. In particular, lower measures of microstructural integrity of the cingulum bundle have been associated with irritability and agitation in clinically-defined MCI and early AD (86). In a cohort of preclinical (amyloid and tau positive) AD patients the presence of irritability and lability at baseline, among other NPS, predicted the subsequent development of hypometabolism in PCC (87). Although our findings might appear in contrast with these investigations showing a reduced structural and functional integrity of the PCC in AD patients, however, recent studies also reported an increased functional connectivity between the PCC and the lateral orbitofrontal cortex in depression (74). Given the role of PCC in the DMN, it has been postulated that enhanced orbitofrontal-PCC connectivity with PCC hyperactivity (25) might be linked to sad ruminating thoughts in depression (25, 74). Local measures of increased metabolism in the PCC at baseline were also documented in the presence of irritability and lability(87). Moreover, a higher efficiency in the information processing of the precuneus, a region closely linked with the PCC from both an anatomical and functional perspective (88, 89), have been found by Shu et al. in clinically-defined normal cognition MBI subjects as opposed to healthy controls (25). Overall, our findings highlight the involvement of the PCC in impulse dyscontrol and social inappropriateness. However, due to the conflicting results reported in the literature, further studies are needed to understand the neurophysiological mechanisms underlying the positive association between network properties and MBI scores observed in our study.
Of note, while widespread alterations in fronto-parietal regions were found in previous studies investigating connectivity changes related to MBI correlates (12, 13, 25, 90), we found that the MBI subscores of affectivity, impulse dyscontrol and social inappropriateness were associated with local network properties of specific brain regions such as the IFG and PCC. This difference could be partially attributed to the methodological differences between our graph analysis approach and the seed-based connectivity analyses performed in these studies (12, 13). On the other hand, a correlation between domain-specific MBI-C subscores and connectivity measures was not performed in the graph-theory study of Shu et al (25). This impedes the accurate localization of their reported connectivity changes. In fact, MBI-C total score accounts for a global effect caused by NPS that are not necessarily related to each other from the pathogenetical viewpoint. In this way, MBI-C total score could fail to effectively correlate with neuroimaging patterns specific to distinct neuropsychiatric sub-domains. Finally, the fact that these studies did not use a biological definition of AD in their inclusion criteria could have played a role.
This is the first study that investigated the topological abnormalities in structural networks driven by impairment in the MBI subdomains in a sample of biologically-defined AD patients. The unique strength of our study lies in the utilization of biologically-defined AD patients, setting it apart from previous research that primarily relied on clinically defined subjects with normal cognition(25) and MCI (11–13). In contrast to these previous studies, our research incorporates information on the underlying biological aetiology, which enhances our ability to establish the specificity of our findings to AD as opposed to other potential pathologies. Moreover, our study distinguishes itself from prior research endeavors that employed metrics other than MBI-C for evaluating the burden of neuropsychiatric symptoms (NPS). MBI-C scores have been proved to be a reliable tool not only to predict the risk of a further cognitive impairment, but also to assess the current psychological and behavioural changes of MCI and AD patients (91). NPI scores have been observed to overestimate the prevalence of NPS (10). Therefore, MBI-C scores might reflect the neuroimaging alterations underlying NPS in early stage AD patients more accurately than NPI scores.
It is important to consider some limitations when interpreting our findings. Firstly, we need to acknowledge that our sample had a relatively limited size. Secondly, our study lacked a control sample. This would have been instrumental to confirm that the correlations we describe between clinical phenotype and structural connectivity measures are specific to MBI-involvement rather than to AD-progression. Future studies should explore if these correlations are exclusive to MBI-due-to AD (14). Finally, despite MBI-C is tailored at the detection of NPS in the pre-dementia stage, we included some patients who were found to be affected by early cognitive impairment in our sample. This might allow some claims on the reliability of their MBI-C score reports. However, the caregivers were carefully given instructions to assist the subjects in the compilation of the questionnaire or, when necessary, to compile it in their stead. The fact that the burden of the NPS was seen to increase with disease progression, as revelead by the significant difference in disease duration (p = 0.002) supports the reliability of the compilation process, since NPS have been previously reported to increase with disease duration in AD (92). Notwithstanding, further studies should consider a procedure of compilation supervised by qualified personnel.