Building diagnostic neuroimaging biomarkers for psychiatric disorders using reverse inference approaches: A viable route?

The advent of structural magnetic resonance imaging (sMRI) at the end of the 20th century opened the way toward a deeper understanding of the neurophysiology of psychiatric disorders, substantiating regional structural abnormalities underlying this group of clinical conditions. However, despite abundant and flourishing scientific research, sMRI methodologies are not currently integrated into daily diagnostic practice. One reason behind this failed translation may be the prevailing approach to logical reasoning in neuroimaging: The forward inference via frequentist-based statistics. This reasoning prevents clinicians from obtaining information about the selectivity of results, which are therefore of limited use regarding the definition of biomarkers and refinement of diagnostic processes. Recently, another type of inferential approach has started to emerge in the neuroimaging field: The reverse inference via Bayesian statistics. Here, we introduce the key concepts of this approach, with a particular emphasis on the clinical sMRI environment. We survey recent findings showing significant potential for clinical translation. Clinical opportunities and challenges for developing reverse inference-based neural markers for psychiatry are also discussed. We propose that a systematic sharing of imaging data across the human brain mapping community is an essential first step toward a paradigmatic clinical shift. We conclude that a defined synergy between forward-based and reverse-based sMRI research can illuminate current discussions on diagnostic brain markers, offering clarity on key issues and fostering new tailored diagnostic avenues.

The effects of genetic and modifiable risk factors on brain regions vulnerable to ageing and disease.

We have previously identified a network of higher-order brain regions particularly vulnerable to the ageing process, schizophrenia and Alzheimer's disease. However, it remains unknown what the genetic influences on this fragile brain network are, and whether it can be altered by the most common modifiable risk factors for dementia. Here, in ~40,000 UK Biobank participants, we first show significant genome-wide associations between this brain network and seven genetic clusters implicated in cardiovascular deaths, schizophrenia, Alzheimer's and Parkinson's disease, and with the two antigens of the XG blood group located in the pseudoautosomal region of the sex chromosomes. We further reveal that the most deleterious modifiable risk factors for this vulnerable brain network are diabetes, nitrogen dioxide - a proxy for traffic-related air pollution - and alcohol intake frequency. The extent of these associations was uncovered by examining these modifiable risk factors in a single model to assess the unique contribution of each on the vulnerable brain network, above and beyond the dominating effects of age and sex. These results provide a comprehensive picture of the role played by genetic and modifiable risk factors on these fragile parts of the brain.

Atypical local brain connectivity in pediatric autism spectrum disorder? A coordinate-based meta-analysis of regional homogeneity studies.

Despite decades of massive neuroimaging research, the comprehensive characterization of short-range functional connectivity in autism spectrum disorder (ASD) remains a major challenge for scientific advances and clinical translation. From the theoretical point of view, it has been suggested a generalized local over-connectivity that would characterize ASD. This stance is known as the general local over-connectivity theory. However, there is little empirical evidence supporting such hypothesis, especially with regard to pediatric individuals with ASD (age [Formula: see text] 18 years old). To explore this issue, we performed a coordinate-based meta-analysis of regional homogeneity studies to identify significant changes of local connectivity. Our analyses revealed local functional under-connectivity patterns in the bilateral posterior cingulate cortex and superior frontal gyrus (key components of the default mode network) and in the bilateral paracentral lobule (a part of the sensorimotor network). We also performed a functional association analysis of the identified areas, whose dysfunction is clinically consistent with the well-known deficits affecting individuals with ASD. Importantly, we did not find relevant clusters of local hyper-connectivity, which is contrary to the hypothesis that ASD may be characterized by generalized local over-connectivity. If confirmed, our result will provide a valuable insight into the understanding of the complex ASD pathophysiology.

CBMAT: a MATLAB toolbox for data preparation and post hoc analyses in neuroimaging meta-analyses.

Coordinate-based meta-analysis (CBMA) is a powerful technique in the field of human brain imaging research. Due to its intense usage, several procedures for data preparation and post hoc analyses have been proposed so far. However, these steps are often performed manually by the researcher, and are therefore potentially prone to error and time-consuming. We hence developed the Coordinate-Based Meta-Analyses Toolbox (CBMAT) to provide a suite of user-friendly and automated MATLAB® functions allowing one to perform all these procedures in a fast, reproducible and reliable way. Besides the description of the code, in the present paper we also provide an annotated example of using CBMAT on a dataset including 34 experiments. CBMAT can therefore substantially improve the way data are handled when performing CBMAs. The code can be downloaded from https://github.com/Jordi-Manuello/CBMAT.git .

Unleashing the Power of Bayesian Re-Analysis: Enhancing Insights into Lecanemab (Clarity AD) Phase III Trial Through Informed t-Test.

BACKGROUND: Clinical trials targeting Alzheimer's disease (AD) aim to alleviate clinical symptoms and alter the course of this complex neurodegenerative disorder. However, the conventional approach of null hypothesis significance testing (NHST) commonly employed in such trials has inherent limitations in assessing clinical significance and capturing nuanced evidence of effectiveness on a continuous scale. OBJECTIVE: In this study, we conducted a re-analysis of the phase III trial of lecanemab, a recently proposed humanized IgG1 monoclonal antibody with high affinity for Aß soluble protofibrils, using a Bayesian approach with informed t-test priors. METHODS: To achieve this, we carefully selected trial data and derived effect size estimates for the primary endpoint, the Clinical Dementia Rating Scale-Sum of Boxes (CDR-SB). Subsequently, a series of Bayes Factor analyses were performed to compare evidence supporting the null hypothesis (no treatment effect) versus the alternative hypothesis (presence of an effect). Drawing on relevant literature and the lecanemab phase III trial, we incorporated different minimal clinically important difference (MCID) values for the primary endpoint CDR-SB as prior information. RESULTS: Our findings, based on a standard prior, revealed anecdotal evidence favoring the null hypothesis. Additional robustness checks yielded consistent results. However, when employing informed priors, we observed varying evidence across different MCID values, ultimately indicating no support for the effectiveness of lecanemab over placebo. CONCLUSION: Our study underscores the value of Bayesian analysis in clinical trials while emphasizing the importance of incorporating MCID and effect size granularity to accurately assess treatment efficacy.

A Minimum Bayes Factor Based Threshold for Activation Likelihood Estimation.

Activation likelihood estimation (ALE) is among the most used algorithms to perform neuroimaging meta-analysis. Since its first implementation, several thresholding procedures had been proposed, all referred to the frequentist framework, returning a rejection criterion for the null hypothesis according to the critical p-value selected. However, this is not informative in terms of probabilities of the validity of the hypotheses. Here, we describe an innovative thresholding procedure based on the concept of minimum Bayes factor (mBF). The use of the Bayesian framework allows to consider different levels of probability, each of these being equally significant. In order to simplify the translation between the common ALE practice and the proposed approach, we analised six task-fMRI/VBM datasets and determined the mBF values equivalent to the currently recommended frequentist thresholds based on Family Wise Error (FWE). Sensitivity and robustness toward spurious findings were also analyzed. Results showed that the cutoff log10(mBF) = 5 is equivalent to the FWE threshold, often referred as voxel-level threshold, while the cutoff log10(mBF) = 2 is equivalent to the cluster-level FWE (c-FWE) threshold. However, only in the latter case voxels spatially far from the blobs of effect in the c-FWE ALE map survived. Therefore, when using the Bayesian thresholding the cutoff log10(mBF) = 5 should be preferred. However, being in the Bayesian framework, lower values are all equally significant, while suggesting weaker level of force for that hypothesis. Hence, results obtained through less conservative thresholds can be legitimately discussed without losing statistical rigor. The proposed technique adds therefore a powerful tool to the human-brain-mapping field.

Retrospective Bayesian Evidence of Null Effect in Two Decades of Alzheimer's Disease Clinical Trials.

Despite intense research on Alzheimer's disease, no validated treatment able to reverse symptomatology or stop disease progression exists. A recent systematic review by Kim and colleagues evaluated possible reasons behind the failure of the majority of the clinical trials. As the focus was on methodological factors, no statistical trends were examined in detail. Here, we aim to complete this picture leveraging on Bayesian analysis. In particular, we tested whether the failure of those clinical trials was essentially due to insufficient statistical power or to lack of a true effect. The strong Bayes' Factor obtained supported the latter hypothesis.

Co-alteration Network Architecture of Major Depressive Disorder: A Multi-modal Neuroimaging Assessment of Large-scale Disease Effects.

Major depressive disorder (MDD) exhibits diverse symptomology and neuroimaging studies report widespread disruption of key brain areas. Numerous theories underpinning the network degeneration hypothesis (NDH) posit that neuropsychiatric diseases selectively target brain areas via meaningful network mechanisms rather than as indistinct disease effects. The present study tests the hypothesis that MDD is a network-based disorder, both structurally and functionally. Coordinate-based meta-analysis and Activation Likelihood Estimation (CBMA-ALE) were used to assess the convergence of findings from 92 previously published studies in depression. An extension of CBMA-ALE was then used to generate a node-and-edge network model representing the co-alteration of brain areas impacted by MDD. Standardized measures of graph theoretical network architecture were assessed. Co-alteration patterns among the meta-analytic MDD nodes were then tested in independent, clinical T1-weighted structural magnetic resonance imaging (MRI) and resting-state functional (rs-fMRI) data. Differences in co-alteration profiles between MDD patients and healthy controls, as well as between controls and clinical subgroups of MDD patients, were assessed. A 65-node 144-edge co-alteration network model was derived for MDD. Testing of co-alteration profiles in replication data using the MDD nodes provided distinction between MDD and healthy controls in structural data. However, co-alteration profiles were not distinguished between patients and controls in rs-fMRI data. Improved distinction between patients and healthy controls was observed in clinically homogenous MDD subgroups in T1 data. MDD abnormalities demonstrated both structural and functional network architecture, though only structural networks exhibited between-groups differences. Our findings suggest improved utility of structural co-alteration networks for ongoing biomarker development.

Revealing the Selectivity of Neuroanatomical Alteration in Autism Spectrum Disorder via Reverse Inference.

BACKGROUND: Although neuroimaging research has identified atypical neuroanatomical substrates in individuals with autism spectrum disorder (ASD), it is at present unclear whether and to what extent disorder-selective gray matter alterations occur in this spectrum of conditions. In fact, a growing body of evidence shows a substantial overlap between the pathomorphological changes across different brain diseases, which may complicate identification of reliable neural markers and differentiation of the anatomical substrates of distinct psychopathologies. METHODS: Using a novel data-driven and Bayesian methodology with published voxel-based morphometry data (849 peer-reviewed experiments and 22,304 clinical subjects), this study performs the first reverse inference investigation to explore the selective structural brain alteration profile of ASD. RESULTS: We found that specific brain areas exhibit a >90% probability of gray matter alteration selectivity for ASD: the bilateral precuneus (Brodmann area 7), right inferior occipital gyrus (Brodmann area 18), left cerebellar lobule IX and Crus II, right cerebellar lobule VIIIA, and right Crus I. Of note, many brain voxels that are selective for ASD include areas that are posterior components of the default mode network. CONCLUSIONS: The identification of these spatial gray matter alteration patterns offers new insights into understanding the complex neurobiological underpinnings of ASD and opens attractive prospects for future neuroimaging-based interventions.

Linking neuroanatomical abnormalities in autism spectrum disorder with gene expression of candidate ASD genes: A meta-analytic and network-oriented approach.

BACKGROUND: Autism spectrum disorder (ASD) is a set of developmental conditions with widespread neuroanatomical abnormalities and a strong genetic basis. Although neuroimaging studies have indicated anatomical changes in grey matter (GM) morphometry, their associations with gene expression remain elusive. METHODS: Here, we aim to understand how gene expression correlates with neuroanatomical atypicalities in ASD. To do so, we performed a coordinate-based meta-analysis to determine the common GM variation pattern in the autistic brain. From the Allen Human Brain Atlas, we selected eight genes from the SHANK, NRXN, NLGN family and MECP2, which have been implicated with ASD, particularly in regards to altered synaptic transmission and plasticity. The gene expression maps for each gene were built. We then assessed the correlation between the gene expression maps and the GM alteration maps. Lastly, we projected the obtained clusters of GM alteration-gene correlations on top of the canonical resting state networks, in order to provide a functional characterization of the structural evidence. RESULTS: We found that gene expression of most genes correlated with GM alteration (both increase and decrease) in regions located in the default mode network. Decreased GM was also correlated with gene expression of some ASD genes in areas associated with the dorsal attention and cerebellar network. Lastly, single genes were found to be significantly correlated with increased GM in areas located in the somatomotor, limbic and ganglia/thalamus networks. CONCLUSIONS: This approach allowed us to combine the well beaten path of genetic and brain imaging in a novel way, to specifically investigate the relation between gene expression and brain with structural damage, and individuate genes of potential interest for further investigation in the functional domain.

Seeking Overlapping Neuroanatomical Alterations between Dyslexia and Attention-Deficit/Hyperactivity Disorder: A Meta-Analytic Replication Study.

The present work is a replication article based on the paper "Are there shared neural correlates between dyslexia and ADHD? A meta-analysis of voxel-based morphometry studies" by McGrath and Stoodley (2019). In the original research, the authors used activation likelihood estimation (ALE), a technique to perform coordinate-based meta-analysis (CBMA), to investigate the existence of brain regions undergoing gray matter alteration in association with both attention-deficit/hyper-activity disorder (ADHD) and dyslexia. Here, the same voxel-based morphometry dataset was analyzed, while using the permutation-subject images version of signed differential mapping (PSI-SDM) in place of ALE. Overall, the replication converged with the original paper in showing a limited overlap between the two conditions. In particular, no significant effect was found for dyslexia, therefore precluding any form of comparison between the two disorders. The possible influences of biological sex, age, and medication status were also ruled out. Our findings are in line with literature about gray matter alteration associated with ADHD and dyslexia, often showing conflicting results. Therefore, although neuropsychological and clinical evidence suggest some convergence between ADHD and dyslexia, more future research is sorely needed to reach a consensus on the neuroimaging domain in terms of patterns of gray matter alteration.

Sex differences in brain homotopic co-activations: a meta-analytic study.

An element of great interest in functional connectivity is 'homotopic connectivity' (HC), namely the connectivity between two mirrored areas of the two hemispheres, mainly mediated by the fibers of the corpus callosum. Despite a long tradition of studying sexual dimorphism in the human brain, to our knowledge only one study has addressed the influence of sex on HC.We investigated the issue of homotopic co-activations in women and men using a coordinate-based meta-analytic method and data from the BrainMap database. A first unexpected observation was that the database was affected by a sex bias: women-only groups are investigated less often than men-only ones, and they are more often studied in certain domains such as emotion compared to men, and less in cognition. Implementing a series of sampling procedures to equalize the size and proportion of the datasets, our results indicated that females exhibit stronger interhemispheric co-activation than males, suggesting that the female brain is less lateralized and more integrated than that of males. In addition, males appear to show less intense but more extensive co-activation than females. Some local differences also appeared. In particular, it appears that primary motor and perceptual areas are more co-activated in males, in contrast to the opposite trend in the rest of the brain. This argues for a multidimensional view of sex brain differences and suggests that the issue should be approached with more complex models than previously thought.

Influence of Gestational Diabetes and Pregestational Maternal BMI on the Brain of Six-Year-Old Offspring.

BACKGROUND: Gestational diabetes (GD) and maternal excess weight are common pregnancy conditions that increase the risk of future complications for both the mother and her offspring. Their consequences on neurodevelopment are widely described in the literature, but less is known concerning the potential transgenerational influence on the brain structure. METHODS: We used a combination of support vectors machine and hierarchical clustering to investigate the potential presence of anatomical brain differences in a sample of 109 children aged six years, born to mothers with overweight or obesity, or to mothers diagnosed with GD during pregnancy. RESULTS: Significant effects are visible in the brain of children born to mothers with GD associated with pregestational excess weight, especially overweight instead of obesity. No differences in children's brain were observed when considering those born to normal-weight mothers. CONCLUSIONS: Our study highlights the need for clinical attention of pregnant women at risk to develop GD, and especially those with pregestational excess weight, since this status was found to be associated with detectable transgenerational brain changes. These effects may be due to the absence of specific and individualized intervention in these mothers during pregnancy.

Six actions to improve detection of critical features for neuroimaging coordinate-based meta-analysis preparation.

Coordinate-based meta-analysis (CBMA) is a research strategy widely used in the field of human brain imaging. Although dedicated tools as BrainMap or Neurosynth had been developed in past years, some of the crucial steps necessary to identify and compose the dataset are still user-based, resulting in a not standardized approach to literature search, as well as in time-consuming and prone to errors procedures. In particular, this concern involves the assessment of voxel-wise whole brain analyses in contrast to ROI-based ones, and the identification of available lists of peaks of effect (i.e., x,y,z coordinates of the foci). Here, we propose six simple actions that can be undertaken by any researcher and by the publishing system, allowing to limit the risk of erroneous decisions on the inclusion of experimental data in the meta-analytic dataset. This straightforward and useful strategy would reduce possible bias in CBMA, therefore allowing to obtain more reliable results.

A co-alteration parceling of the cingulate cortex.

The cingulate cortex is known to be a complex structure, involved in several cognitive and emotional functions, as well as being altered by a variety of brain disorders. This heterogeneity is reflected in the multiple parceling models proposed in the literature. At the present, sub-regions of the cingulate cortex had been identified taking into account functional and structural connectivity, as well as cytological and electrochemical properties. In the present work, we propose an innovative node-wise parceling approach based on meta-analytic Bayesian co-alteration. To this aim, 193 case-control voxel-based morphometry experiments were analyzed, and the Patel's κ index was used to assess probability of morphometric co-alteration between nodes placed in the cingulate cortex and in the rest of the brain. Hierarchical clustering was then applied to identify nodes in the cingulate cortex exhibiting a similar pattern of whole-brain co-alteration. The obtained dendrogram highlighted a robust fronto-parietal cluster compatible with the default mode network, and being supported by the interplay between the retrosplenial cortex and the anterior and posterior cingulate cortex, rarely described in the literature. This ensemble was further confirmed by the analysis of functional patterns. Leveraging on co-alteration to investigate cortical organization could, therefore, allow to combine multimodal information, resolving conflicting results sometimes coming from the separate use of singular modalities. Crucially, this provides a valuable way to understand the pathological brain using data driven, whole-brain informed and context-specific evidence in a way not yet explored in the field.

Tasks activating the default mode network map multiple functional systems.

Recent developments in network neuroscience suggest reconsidering what we thought we knew about the default mode network (DMN). Although this network has always been seen as unitary and associated with the resting state, a new deconstructive line of research is pointing out that the DMN could be divided into multiple subsystems supporting different functions. By now, it is well known that the DMN is not only deactivated by tasks, but also involved in affective, mnestic, and social paradigms, among others. Nonetheless, it is starting to become clear that the array of activities in which it is involved, might also be extended to more extrinsic functions. The present meta-analytic study is meant to push this boundary a bit further. The BrainMap database was searched for all experimental paradigms activating the DMN, and their activation likelihood estimation maps were then computed. An additional map of task-induced deactivations was also created. A multidimensional scaling indicated that such maps could be arranged along an anatomo-psychological gradient, which goes from midline core activations, associated with the most internal functions, to that of lateral cortices, involved in more external tasks. Further multivariate investigations suggested that such extrinsic mode is especially related to reward, semantic, and emotional functions. However, an important finding was that the various activation maps were often different from the canonical representation of the resting-state DMN, sometimes overlapping with it only in some peripheral nodes, and including external regions such as the insula. Altogether, our findings suggest that the intrinsic-extrinsic opposition may be better understood in the form of a continuous scale, rather than a dichotomy.

Disentangling predictive processing in the brain: a meta-analytic study in favour of a predictive network.

According to the predictive coding (PC) theory, the brain is constantly engaged in predicting its upcoming states and refining these predictions through error signals. Despite extensive research investigating the neural bases of this theory, to date no previous study has systematically attempted to define the neural mechanisms of predictive coding across studies and sensory channels, focussing on functional connectivity. In this study, we employ a coordinate-based meta-analytical approach to address this issue. We first use the Activation Likelihood Estimation (ALE) algorithm to detect spatial convergence across studies, related to prediction error and encoding. Overall, our ALE results suggest the ultimate role of the left inferior frontal gyrus and left insula in both processes. Moreover, we employ a meta-analytic connectivity method (Seed-Voxel Correlations Consensus). This technique reveals a large, bilateral predictive network, which resembles large-scale networks involved in task-driven attention and execution. In sum, we find that: (i) predictive processing seems to occur more in certain brain regions than others, when considering different sensory modalities at a time; (ii) there is no evidence, at the network level, for a distinction between error and prediction processing.

Interhemispheric co-alteration of brain homotopic regions.

Asymmetries in gray matter alterations raise important issues regarding the pathological co-alteration between hemispheres. Since homotopic areas are the most functionally connected sites between hemispheres and gray matter co-alterations depend on connectivity patterns, it is likely that this relationship might be mirrored in homologous interhemispheric co-altered areas. To explore this issue, we analyzed data of patients with Alzheimer's disease, schizophrenia, bipolar disorder and depressive disorder from the BrainMap voxel-based morphometry database. We calculated a map showing the pathological homotopic anatomical co-alteration between homologous brain areas. This map was compared with the meta-analytic homotopic connectivity map obtained from the BrainMap functional database, so as to have a meta-analytic connectivity modeling map between homologous areas. We applied an empirical Bayesian technique so as to determine a directional pathological co-alteration on the basis of the possible tendencies in the conditional probability of being co-altered of homologous brain areas. Our analysis provides evidence that: the hemispheric homologous areas appear to be anatomically co-altered; this pathological co-alteration is similar to the pattern of connectivity exhibited by the couples of homologues; the probability to find alterations in the areas of the left hemisphere seems to be greater when their right homologues are also altered than vice versa, an intriguing asymmetry that deserves to be further investigated and explained.

BACON: A tool for reverse inference in brain activation and alteration.

Over the past decades, powerful MRI-based methods have been developed, which yield both voxel-based maps of the brain activity and anatomical variation related to different conditions. With regard to functional or structural MRI data, forward inferences try to determine which areas are involved given a mental function or a brain disorder. A major drawback of forward inference is its lack of specificity, as it suggests the involvement of brain areas that are not specific for the process/condition under investigation. Therefore, a different approach is needed to determine to what extent a given pattern of cerebral activation or alteration is specifically associated with a mental function or brain pathology. In this study, we present a new tool called BACON (Bayes fACtor mOdeliNg) for performing reverse inference both with functional and structural neuroimaging data. BACON implements the Bayes' factor and uses the activation likelihood estimation derived-maps to obtain posterior probability distributions on the evidence of specificity with regard to a particular mental function or brain pathology.

Gray matter abnormalities follow non-random patterns of co-alteration in autism: Meta-connectomic evidence.

BACKGROUND: Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by atypical brain anatomy and connectivity. Graph-theoretical methods have mainly been applied to detect altered patterns of white matter tracts and functional brain activation in individuals with ASD. The network topology of gray matter (GM) abnormalities in ASD remains relatively unexplored. METHODS: An innovative meta-connectomic analysis on voxel-based morphometry data (45 experiments, 1,786 subjects with ASD) was performed in order to investigate whether GM variations can develop in a distinct pattern of co-alteration across the brain. This pattern was then compared with normative profiles of structural and genetic co-expression maps. Graph measures of centrality and clustering were also applied to identify brain areas with the highest topological hierarchy and core sub-graph components within the co-alteration network observed in ASD. RESULTS: Individuals with ASD exhibit a distinctive and topologically defined pattern of GM co-alteration that moderately follows the structural connectivity constraints. This was not observed with respect to the pattern of genetic co-expression. Hub regions of the co-alteration network were mainly left-lateralized, encompassing the precuneus, ventral anterior cingulate, and middle occipital gyrus. Regions of the default mode network appear to be central in the topology of co-alterations. CONCLUSION: These findings shed new light on the pathobiology of ASD, suggesting a network-level dysfunction among spatially distributed GM regions. At the same time, this study supports pathoconnectomics as an insightful approach to better understand neuropsychiatric disorders.