Quantifying brain connectivity signatures by means of polyconnectomic scoring

Abstract

A broad range of neuropsychiatric disorders are associated with alterations in macroscale brain circuitry and connectivity. Identifying consistent brain patterns underlying these disorders by means of structural and functional MRI has proven challenging, partly due to the vast number of tests required to examine the entire brain, which can lead to an increase in missed findings. In this study, we propose polyconnectomic score (PCS) as a metric designed to quantify the presence of disease-related brain connectivity signatures in connectomes. PCS summarizes evidence of brain patterns related to a phenotype across the entire landscape of brain connectivity into a subject-level score. We evaluated PCS across four brain disorders (autism spectrum disorder, schizophrenia, attention deficit hyperactivity disorder, and Alzheimer's disease) and 14 studies encompassing ~35,000 individuals. Our findings consistently show that patients exhibit significantly higher PCS compared to controls, with effect sizes that go beyond other single MRI metrics ([min, max]: Cohen's d = [0.30, 0.87], AUC = [0.58, 0.73]). We further demonstrate that PCS serves as a valuable tool for stratifying individuals, for example within the psychosis continuum, distinguishing patients with schizophrenia from their first-degree relatives (d = 0.42, p = 4 × 10^-3, FDR-corrected), and first-degree relatives from healthy controls (d = 0.34, p = 0.034, FDR-corrected). We also show that PCS is useful to uncover associations between brain connectivity patterns related to neuropsychiatric disorders and mental health, psychosocial factors, and body measurements.

Fig 1.. Computation of polyconnectomic score.

The computation of the polyconnectomic score (PCS) relies on connectome summary statistics (CSS). These statistics represent the strength and direction of the association between brain connections and the phenotype of interest (POI), measured using regression coefficients for scaled variables or Cohen’s d for binary variables. CSS can be based on either a discovery dataset or a previously conducted independent study. The PCS for an out-of-sample individual can then be computed as the weighted average of the CSS and the individual’s brain connectivity map, capturing how closely a subject’s connectome resembles the brain signature associated with the POI. The efficacy of PCS is evaluated by comparing scores between cases and controls. PCS, polyconnectomic score; POI, phenotype of interest; d, Cohen’s d.

Figure 2.. Polyconnectomic score for autism spectrum disorder, schizophrenia, attention deficit hyperactivity disorder, and Alzheimer’s disease.

Connectome summary statistics (CSS) are estimated from a previously conducted study (left column). These statistics quantify the strength of the association (x-axis) and the level of significance (y-axis) for each brain connection in relation to (A) autism spectrum disorder, (B) schizophrenia, (C) attention deficit hyperactivity disorder, and (D) Alzheimer’s disease. Blue and red dots represent connections with decreased and increased functional connectivity in patients compared to controls, respectively. These CSS are used to calculate the polyconnectomic score (PCS) in an independent study (middle column). In each study, we statistically compare PCS levels between patients and controls (y-axis; white dots indicate group means) to assess the method’s efficacy in capturing brain connectivity signatures linked to neuropsychiatric disorders. Asterisks denote studies where significant differences in PCS between groups were observed, as estimated by t-test statistics (FDR-corrected). Logistic regression analysis (right column) is used to evaluate the predictive power of PCS in classifying individual diagnoses by estimating the area under the receiver operating characteristic curve (AUC; x-axis for false positive rate, y-axis for true positive rate). A dotted line at an AUC of 0.5 corresponds to random guessing. CSS, connectome summary statistics; PCS, polyconnectomic score; ASD, autism spectrum disorder; SCZ, schizophrenia; ADHD, attention deficit hyperactivity disorder; AD, Alzheimer’s disease; AUC, area under the curve of the receiver operating characteristic curve.

Figure 3.. Computation of polyconnectomic score using meta-analytic summary statistics.

A leave-one-out meta-analysis is conducted to derive robust connectome summary statistics (CSS) for calculating the polyconnectomic score (PCS) in independent studies. Within each study (x-axis), differences in PCS levels between patients and controls are estimated using Cohen’s d (y-axis). Connections from the CSS are thresholded based on p-value significance levels, ranging from no threshold to approximately Bonferroni correction (p-value threshold < 1 × 10⁻⁵). Asterisks indicate studies where significant differences in PCS between patients and controls are observed, as estimated by t-test statistics (FDR-corrected). PCS, polyconnectomic score; ASD, autism spectrum disorder; SCZ, schizophrenia; ADHD, attention deficit hyperactivity disorder; AD, Alzheimer’s disease.

Figure 4.. Using polyconnectomic score for schizophrenia to stratify individuals across the psychosis continuum

The polyconnectomic score (PCS) for schizophrenia (SCZ) is computed across the psychosis continuum, including patients with SCZ, schizoaffective disorder (SCA), and psychotic bipolar disorder (BD), as well as first-degree relatives from each group, and healthy controls. (A) Violin plots display the distribution of PCS-SCZ (y-axis) for each group (x-axis; white dot denotes the mean). (B) A histogram shows the frequency count (y-axis) of PCS-SCZ values (x-axis) among individuals in each group. Patients with SCZ present the largest differences in PCS-SCZ compared to healthy controls (d = 0.77), followed by SCA (d = 0.66) and BD (d = 0.57). SCZ, schizophrenia; SCA, schizoaffective disorder; BD, bipolar disorder; SCZ-rel, first-degree relatives of schizophrenia patients; SCA-rel, first-degree relatives of schizoaffective disorder patients; BD-rel, first-degree relatives of bipolar disorder patients; PCS-SCZ, polyconnectomic score for schizophrenia.

Figure 5.. Brain-behavior correlations using polyconnectomic score for schizophrenia.

A circle plot illustrates the association between the polyconnectomic score for schizophrenia (PCS-SCZ) and measures of cognition (dark blue), mental health (light blue), and self-reported medical conditions (pink), based on data from the UK Biobank. Pearson’s correlation coefficients are displayed only for associations that remain significant after FDR correction (gray indicates non-significant effects). Dichotomous variables measured with Cohen’s d are converted to Pearson’s correlation coefficient for visualization purposes. Elevated levels of PCS-SCZ are associated with reduced cognitive performance, increased neuroticism, and a higher incidence of mental health complaints. Similar associations are observed in subclinical populations. PCS-SCZ, polyconnectomic score for schizophrenia.