Advertisement

Resting-State Functional Connectivity Explained Psychotic-Like Experiences in the General Population and Partially Generalized to Patients and Relatives

Open AccessPublished:September 08, 2022DOI:https://doi.org/10.1016/j.bpsgos.2022.08.011

      Abstract

      BACKGROUND

      Psychotic-like experiences (PLEs) are considered the subclinical portion of the psychosis continuum. Research suggests resting-state functional connectivity (rsFC) substrates of PLEs, yet it is unclear if the same substrates underlie more severe psychosis. We report the first study to build a cross-validated rsFC model of PLEs in a large community sample and directly test its ability to explain psychosis in an independent sample of patients with psychosis and their relatives.

      METHODS

      RsFC of 855 healthy young adults from the Human Connectome Project (HCP) was used to predict PLEs with elastic net. A rsFC composite score based on the resulting model was correlated with psychotic traits and symptoms in 118 patients with psychosis, 71 non-psychotic first-degree relatives, and 45 healthy controls from the psychosis HCP (P-HCP).

      RESULTS

      In the HCP, the cross-validated model explained 3.3% of variance in PLEs. Predictive connections spread primarily across the default, frontoparietal, cingulo-opercular, and dorsal attention networks. The model partially generalized to a younger, but not older, subsample in the P-HCP, explaining two measures of positive/disorganized psychotic traits (the Structured Interview for Schizotypy: β = 0.25, pone-tailed = 0.027, the Schizotypy Personality Questionnaire positive factor: β = 0.14, pone-tailed = 0.041). However, it did not differentiate patients from relatives and controls or explain psychotic symptoms in patients.

      CONCLUSIONS

      Some rsFC substrates of PLEs are shared across the psychosis continuum. However, explanatory power was modest and generalization was partial. It is equally important to understand shared versus distinct rsFC variances across the psychosis continuum.

      Key words

      Psychosis is one of the most debilitating symptoms shared by a range of psychiatric diagnoses, such as schizophrenia spectrum disorders and affective disorders with psychosis (
      • Jablensky A.
      Epidemiology of schizophrenia: The global burden of disease and disability.
      ,
      • Tienari P.
      • Wynne L.C.
      • Läksy K.
      • Moring J.
      • Nieminen P.
      • Sorri A.
      • et al.
      Genetic boundaries of the schizophrenia spectrum: Evidence from the Finnish adoptive family study of schizophrenia.
      ). A rapidly growing area of psychosis research is psychotic-like experiences (PLEs), i.e., experiences that resemble psychotic symptoms yet are not frequent, severe, or disturbing enough to warrant diagnoses. Estimated to be present in 5%-26% of the general population (
      • Bourgin J.
      • Tebeka S.
      • Mallet J.
      • Mazer N.
      • Dubertret C.
      • Le Strat Y.
      Prevalence and correlates of psychotic-like experiences in the general population.
      ,
      • Scott J.
      • Chant D.
      • Andrews G.
      • McGrath J.
      Psychotic-like experiences in the general community: the correlates of CIDI psychosis screen items in an Australian sample, 2005/11/23.
      ,
      • Van Os J.
      • Linscott R.J.
      • Myin-Germeys I.
      • Delespaul P.
      • Krabbendam L.
      A systematic review and meta-analysis of the psychosis continuum: Evidence for a psychosis proneness-persistence-impairment model of psychotic disorder.
      ), PLEs are considered the lower end of the psychosis continuum, sharing risk factors, genetic loadings, and neurobiological underpinnings with full-blown psychosis (
      • Esterberg M.L.
      • Compton M.T.
      The psychosis continuum and categorical versus dimensional diagnostic approaches.
      ,
      • Alemany S.
      • Arias B.
      • Aguilera M.
      • Villa H.
      • Moya J.
      • Ibanez M.I.
      • et al.
      Childhood abuse, the BDNF-Val66Met polymorphism and adult psychotic-like experiences.
      ,
      • Pain O.
      • Dudbridge F.
      • Cardno A.G.
      • Freeman D.
      • Lu Y.
      • Lundstrom S.
      • et al.
      Genome‐wide analysis of adolescent psychotic‐like experiences shows genetic overlap with psychiatric disorders.
      ,
      • Orr J.M.
      • Turner J.A.
      • Mittal V.A.
      Widespread brain dysconnectivity associated with psychotic-like experiences in the general population.
      ).
      The allure of studying PLEs for the neurobiology of psychosis is the ability to work with large population-based samples free of confounds such as antipsychotics and excessive substance use. One instance is the dysconnection hypothesis of psychosis (
      • Friston K.J.
      • Frith C.D.
      Schizophrenia: a disconnection syndrome?.
      ). In the WU-Minn Human Connectome Project (HCP), PLEs were associated with reduced global efficiency of the default and cingulo-opercular networks (
      • Sheffield J.M.
      • Kandala S.
      • Burgess G.C.
      • Harms M.P.
      • Barch D.M.
      Cingulo-opercular network efficiency mediates the association between psychotic-like experiences and cognitive ability in the general population.
      ) and higher connectivity within the default and lower connectivity within the frontoparietal network (
      • Blain S.D.
      • Grazioplene R.G.
      • Ma Y.
      • DeYoung C.G.
      Toward a neural model of the Openness-Psychoticism dimension: Functional connectivity in the default and frontoparietal control networks.
      ). PLEs were also associated with longer time dwelling in a state of increased visual and decreased default within-network connectivity while shorter time in a state of anticorrelation between the default and other networks (
      • Barber A.D.
      • Lindquist M.A.
      • Derosse P.
      • Karlsgodt K.H.
      Dynamic functional connectivity states reflecting psychotic-like experiences.
      ). In a non-HCP sample, PLEs were associated with hypoconnectivity between the dorsal striatum and the dorsolateral prefrontal, anterior cingulate, and primary motor cortices (
      • Sabaroedin K.
      • Tiego J.
      • Parkes L.
      • Sforazzini F.
      • Finlay A.
      • Johnson B.
      • et al.
      Functional connectivity of corticostriatal circuitry and psychosis-like experiences in the general community.
      ,
      • Pani S.M.
      • Sabaroedin K.
      • Tiego J.
      • Bellgrove M.A.
      • Fornito A.
      A multivariate analysis of the association between corticostriatal functional connectivity and psychosis-like experiences in the general community.
      ). In the Adolescent Brain Cognitive Development (ABCD) sample, PLEs were associated with decreased cingulo-opercular, default, and cinguloparietal network connectivity in 9- to 11-year-old children (
      • Karcher N.R.
      • O’Brien K.J.
      • Kandala S.
      • Barch D.M.
      Resting-state functional connectivity and psychotic-like experiences in childhood: Results from the adolescent brain cognitive development study.
      ). Besides these investigations in large-scale datasets, many studies have also examined functional connectivity associated with schizotypy in smaller samples (
      • Nelson M.T.
      • Seal M.L.
      • Phillips L.J.
      • Merritt A.H.
      • Wilson R.
      • Pantelis C.
      An investigation of the relationship between cortical connectivity and schizotypy in the general population.
      ,
      • Waltmann M.
      • O’Daly O.
      • Egerton A.
      • McMullen K.
      • Kumari V.
      • Barker G.J.
      • et al.
      Multi-echo fMRI, resting-state connectivity, and high psychometric schizotypy.
      ,
      • Zhu Y.
      • Tang Y.
      • Zhang T.
      • Li H.
      • Tang Y.
      • Li C.
      • et al.
      Reduced functional connectivity between bilateral precuneus and contralateral parahippocampus in schizotypal personality disorder.
      ,
      • Kozhuharova P.
      • Saviola F.
      • Diaconescu A.
      • Allen P.
      High schizotypy traits are associated with reduced hippocampal resting state functional connectivity.
      ,
      • Yang Z.
      • Zhang R.
      • Li Y.
      • Wang Y.
      • Wang Y.
      • Wang S.
      • et al.
      Functional connectivity of the default mode network is associated with prospection in schizophrenia patients and individuals with social anhedonia.
      ). Overall, findings echo conclusions from review and meta-analysis in chronic, first-episode, clinical high risk, and high genetic risk psychosis (
      • Dong D.
      • Wang Y.
      • Chang X.
      • Luo C.
      • Yao D.
      Dysfunction of large-scale brain networks in schizophrenia: A meta-analysis of resting-state functional connectivity.
      ,
      • Pettersson-Yeo W.
      • Allen P.
      • Benetti S.
      • McGuire P.
      • Mechelli A.
      Dysconnectivity in schizophrenia: Where are we now?.
      ). The psychosis continuum is associated with widespread brain dysconnectivity, especially in the frontoparietal, default, cingulo-opercular (also known as ventral attention or salience) networks and the fronto-striatal-thalamic loop.
      Despite this progress, research on resting-state functional connectivity (rsFC) in PLEs faces several challenges. First, most studies focused on a priori hypothesized slices of the brain connectome, leaving unanswered to what extent PLEs can be collectively explained by candidate connections (i.e., a multiple R2). Second, a recent report highlights the risk of irreproducible and inflated effects in brain-wide association studies (
      • Marek S.
      • Tervo-Clemmens B.
      • Calabro F.J.
      • Montez D.F.
      • Kay B.P.
      • Hatoum A.S.
      • et al.
      Towards reproducible brain-wide association studies affiliations.
      ), yet many existing findings were not cross-validated. Last, PLEs rsFC findings are rarely directly validated in other groups along the psychosis continuum, making it unclear if PLEs share rsFC correlates with more severe forms of psychosis.
      In this study, we built a cross-validated rsFC model for PLEs in a large community sample, following the success of a recent study (
      • Dubois J.
      • Galdi P.
      • Paul L.K.
      • Adolphs R.
      A distributed brain network predicts general intelligence from resting-state human neuroimaging data.
      ) which explained 20% of variance in general intelligence with rsFC (Study I). We then directly examined the ability of this model to explain psychotic traits and symptoms in patients with psychosis and their first-degree relatives (Study II). We hypothesized that the rsFC model will 1) encompass distributed hypo- and hyperconnectivity overrepresenting the frontoparietal, default, and cingulo-opercular networks and 2) generalize to patients and relatives.
      Study I. A rsFC model for PLEs in the HCP

      Study I Methods and Materials

      Participants

      We selected 1003 participants (age = 28.7 ± 3.7 years, range = 22 - 37 years, 53% female, White/African American/Other = 75.6%/13.9%/10.6%, mean relative root mean square (RMS) movement = 0.088 ± 0.037 mm (
      • Jenkinson M.
      • Bannister P.
      • Brady M.
      • Smith S.M.
      Improved optimization for the robust and accurate linear registration and motion correction of brain images.
      )) with completed rsfMRI scans (four runs, totaling 60 minutes) from the 1200 Subjects Data Release of the WashU–UMinn HCP. Participants were free of significant history of psychiatric, neurological, substance use, or cardiovascular disorders (
      • Van Essen D.C.
      • Smith S.M.
      • Barch D.M.
      • Behrens T.E.J.
      • Yacoub E.
      • Ugurbil K.
      The WU-Minn Human Connectome Project: An overview.
      ).

      PLEs

      We measured PLEs as the sum of four items in the Thought Disorder Subscale of the Achenbach Adult Self-Report (ASR (

      Achenbach TM (2009): The Achenbach System of Empirically Based Assessment (ASEBA): Development, Findings, Theory, and Applications. University of Vermont, Research Center for Children, Youth, & Families.

      ), Table 1). Cronbach’s alpha was 0.59 (
      • Cronbach L.J.
      Coefficient alpha and the internal structure of tests.
      ). Average inter-item correlation was 0.28. Previous studies (
      • Sheffield J.M.
      • Kandala S.
      • Burgess G.C.
      • Harms M.P.
      • Barch D.M.
      Cingulo-opercular network efficiency mediates the association between psychotic-like experiences and cognitive ability in the general population.
      ,
      • Blain S.D.
      • Grazioplene R.G.
      • Ma Y.
      • DeYoung C.G.
      Toward a neural model of the Openness-Psychoticism dimension: Functional connectivity in the default and frontoparietal control networks.
      ) show that this PLEs measure agreed with the literature on prevalence and correlation with demographic and personality variables. Table 2 shows the distribution of PLEs in the final sample.
      Table 1Endorsement of PLEs items in the Final HCP Sample (n = 855).
      Item0 (not true)1 (somewhat/sometimes true)2 (very/often true)
      Hear sounds/voices that others think aren’t there845 (98.8%)8 (0.9%)2 (0.2%)
      See things that others think aren’t there844 (98.7%)11 (1.3%)0 (0.0%)
      Do things that other people think are strange714 (83.5%)114 (13.3%)27 (3.2%)
      Have thoughts that others think are strange736 (86.1%)93 (10.9%)26 (3.0%)
      Table 2Distribution of PLEs in the Final HCP Sample (n = 855).
      PLEs0123456
      Frequency (%)679 (79.4%)70 (8.2%)76 (8.9%)12 (1.4%)15 (1.8%)0 (0.0%)3 (0.4%)

      RsFC

      RsFC derivation was previously described in full detail (

      Ma Y, Macdonald AW (2021): Impact of ICA dimensionality on the test-retest reliability of resting-state functional connectivity. Brain Connect ahead of print.

      ). Briefly, we applied group independent component analysis (ICA) to denoised, surface-based rsfMRI data to parcellate the cerebral cortex into 100 independent components (ICs) (
      • Smith S.M.
      • Hyva¨rinen A.
      • Varoquaux G.
      • Miller K.L.
      • Beckmann C.F.
      Group-PCA for very large fMRI datasets.
      ,
      • Beckmann C.F.
      • Smith S.M.
      Probabilistic independent component analysis for functional magnetic resonance imaging.
      ). We used dual regression (
      • Beckmann C.F.
      • Mackay C.E.
      • Filippini N.
      • Smith S.M.
      Group comparison of resting-state FMRI data using multi-subject ICA and dual regression.
      ) to derive a participant’s timeseries in each IC. RsFC between two ICs was the Fisher’s z transformed Pearson’s correlation between their timeseries, averaged across four runs. The result was a 100×100 rsFC matrix for each participant with adequate test-retest reliability as shown previously (

      Ma Y, Macdonald AW (2021): Impact of ICA dimensionality on the test-retest reliability of resting-state functional connectivity. Brain Connect ahead of print.

      ). Figure S1 summarizes the ICs, their distribution across seven canonical cortical networks (
      • Yeo B.T.
      • Krienen F.M.
      • Sepulcre J.
      • Sabuncu M.R.
      • Lashkari D.
      • Hollinshead M.
      • et al.
      The organization of the human cerebral cortex estimated by intrinsic functional connectivity.
      ), and rsFC reliability.

      Final Sample

      Within the 1003 participants, we excluded 1) 125 for positive drug screen on any day of study visit, including breathalyzer > .05 and positive urine screen for cocaine, THC, opiates, amphetamine, methamphetamine, or oxycontin; 2) 3 for incomplete PLEs; 3) 4 for incomplete cognitive tasks (see notes of Table S1); 4) 4 for incomplete education and income information; and 5) 12 for 3T functional preprocessing errors according to a recent HCP announcement (https://wiki.humanconnectome.org/display/PublicData/HCP+Data+Release+Updates%3A+Known+Issues+and+Planned+fixes). The final sample included 855 participants from 397 families (age = 28.8 ± 3.7 years, range = 22 - 36 years, 56.1% female, White/African American/Other = 78.5%/11.7%/9.8% (“Other” included Asian, Native Hawaiian, or other Pacific Islander; American Indian or Alaskan Native; more than one; and unknown or unreported), mean Relative RMS = 0.086 ± 0.030 mm).

      Model Training and Cross-Validation

      We trained and cross-validated elastic net models largely following (
      • Dubois J.
      • Galdi P.
      • Paul L.K.
      • Adolphs R.
      A distributed brain network predicts general intelligence from resting-state human neuroimaging data.
      ) (see Supplementary Methods). Briefly, we used 2320 rsFC features with higher than “poor” between-session reliability (intraclass correlation coefficient > 0.4 (
      • Cicchetti D.V.
      • Sparrow S.A.
      Developing criteria for establishing interrater reliability of specific items: Applications to assessment of adaptive behavior.
      )) to predict PLEs in leave-one-family-out cross-validation. We evaluated model performance with the coefficient of determination R2 (
      • Barrett J.P.
      The coefficient of determination—some limitations.
      ), which estimates out-of-sample predictionand can take negative values (i.e., not the “squared R”). We made statistical inferences based on 1000 multi-level block permutations that accounted for family structure (
      • Winkler A.M.
      • Webster M.A.
      • Vidaurre D.
      • Nichols T.E.
      • Smith S.M.
      Multi-level block permutation.
      ).

      Study I Results

      PLEs and Covariates

      Correlations between PLEs and potential covariates are shown in Table S1. Similar to previous reports (
      • Sheffield J.M.
      • Kandala S.
      • Burgess G.C.
      • Harms M.P.
      • Barch D.M.
      Cingulo-opercular network efficiency mediates the association between psychotic-like experiences and cognitive ability in the general population.
      ,
      • Blain S.D.
      • Grazioplene R.G.
      • Ma Y.
      • DeYoung C.G.
      Toward a neural model of the Openness-Psychoticism dimension: Functional connectivity in the default and frontoparietal control networks.
      ), PLEs were negatively correlated with age (r = -.15, p < .001) and household income (r = -.13, p < .001). Males (mean = 0.49, SD = 0.96) are more likely to endorse PLEs than females (mean = 0.32, SD = 0.85, Welch’s t(754) = 2.76, p = .006). Racial groups differed in PLEs (F(2,852) = 6.81, p < .001). A Tukey’s HSD test suggested that White individuals (mean = 0.33, SD = .81) reported fewer PLEs than individuals in Black or African American (mean = 0.59, SD = 1.13) or Other racial groups (mean = 0.63, SD = 1.22). PLEs did not correlate with g, handedness, years of education, mean relative RMS movement, or brain volume. The lack of correlation between g and PLEs was inconsistent with previous studies where cognitive ability was negatively associated with PLEs (
      • Sheffield J.M.
      • Kandala S.
      • Burgess G.C.
      • Harms M.P.
      • Barch D.M.
      Cingulo-opercular network efficiency mediates the association between psychotic-like experiences and cognitive ability in the general population.
      ,
      • Blain S.D.
      • Grazioplene R.G.
      • Ma Y.
      • DeYoung C.G.
      Toward a neural model of the Openness-Psychoticism dimension: Functional connectivity in the default and frontoparietal control networks.
      ). However, our measure of g (after (
      • Dubois J.
      • Galdi P.
      • Paul L.K.
      • Adolphs R.
      A distributed brain network predicts general intelligence from resting-state human neuroimaging data.
      ), see notes of Table S1) involved both fluid and crystal intelligence, and the latter did not correlate with PLEs in (
      • Sheffield J.M.
      • Kandala S.
      • Burgess G.C.
      • Harms M.P.
      • Barch D.M.
      Cingulo-opercular network efficiency mediates the association between psychotic-like experiences and cognitive ability in the general population.
      ). In this sample that passed drug screen on both days of scanning, PLEs did not correlate with tobacco use, drinking, marijuana use, and illicit drug use.

      Model Performance

      RsFC predicted R2 = 3.3% of variance in PLEs (p = .002, correlation between observed and predicted PLEs = .19) when controlling for age, sex, handedness, movement in scanner, total brain volume, and reconstruction algorithm version number (Model I, Figure 1A). R2 dropped to 2.0% (p = .007, correlation between observed and predicted PLEs = .16) when additionally controlling for g, race, years of education, and household income (Model II, Figure S2A).
      Figure thumbnail gr1
      Figure 1Performance and Predictive Connections of Model I. A) Distribution of R2 across 1000 permutations. Red line: observed R2 (3.3%). Dark gray area: p > .01. Light gray area: p > .05. B) Histogram of absolute standardized coefficients (|β|) in the final model. Dashed line: median. C) Predictive connections. For simplicity, only connections with an absolute standardized coefficient above the median (.01) were plotted. Exterior ring represents canonical networks by (
      • Yeo B.T.
      • Krienen F.M.
      • Sepulcre J.
      • Sabuncu M.R.
      • Lashkari D.
      • Hollinshead M.
      • et al.
      The organization of the human cerebral cortex estimated by intrinsic functional connectivity.
      ). Numbers index ICs. Red: connections with positive standardized coefficients. Blue: connections with negative standardized coefficients. TP/OFC: temporal pole/orbitofrontal cortex.

      Predictive connections

      Model I included 106 connections distributed across the brain (Table S2). Prediction was a result of many small effects: the absolute standardized coefficients (|β|s) ranged from 6e-4 to 0.038, with a median of 0.010 (Figure 1B). Figure 1C shows the top 50% connections with the highest |β|s. Predictive connections mainly involved the cognitive networks (frontoparietal, cingulo-opercular, dorsal attention, and default), and to a lesser extent the visual and somatomotor networks. This overrepresentation of cognitive networks was not a mere consequence that these networks were measured more reliably thus more likely included in modeling (Figure S3). Both hypo- (44.8%) and hyperconnectivity (55.2%) were associated with PLEs with no prevailing pattern for either type (Table S3). Spatial maps of ICs involved in these connections are shown in Figure 2.
      Figure thumbnail gr2
      Figure 2Spatial Maps of ICs Involved in the Predictive Connections of Model I. ICA z maps were normalized so that the maximal value was 1. Each map was thresholded at z/zmax = 0.3. IC: independent component.
      Model II included 97 connections that largely overlapped with Model I (Table S2). Differences between Model I and Model II were mainly due to smaller |β|s in Model II as more covariates were controlled for (Figures 1B & S2B, 1C & S2C).
      Study I demonstrated the feasibility of constraining a connectomics approach with reliability information and using cross-validation to account for a small proportion of variance (2-3%) in PLEs in a healthy sample. Next, we conducted Study II to determine whether a rsFC composite score based on these conservative assumptions generalized to independent samples enriched with psychosis.

      Study II. Generalization to the Psychosis HCP (P-HCP)

      Study II Methods and Materials

      Participants

      We included 118 patients with psychosis, 71 of their non-psychotic first-degree biological relatives, and 45 healthy controls from the ongoing Psychosis HCP (P-HCP). The P-HCP was approved by the University of Minnesota (UMN) Institutional Review Board. Details on study protocol are described elsewhere (
      • Demro C.
      • Mueller B.A.
      • Kent J.S.
      • Burton P.C.
      • Olman C.A.
      • Schallmo M.P.
      • et al.
      The psychosis human connectome project: An overview.
      ) and summarized in Supplementary Methods. Table 3 summarizes the demographic and clinical characteristics of the groups.
      Table 3Demographic and Clinical Characteristics of the P-HCP Patient, Relative, and Control Groups.
      Patient (n = 118)Relative (n = 71)Control (n = 45)Statistical Testing
      Age38.7 (13.3)44.5 (14.3)37.3 (12.3)F(2,231) = 6.9*

      Patient < Relative***
      % Male56.835.253.3χ2(
      • Tienari P.
      • Wynne L.C.
      • Läksy K.
      • Moring J.
      • Nieminen P.
      • Sorri A.
      • et al.
      Genetic boundaries of the schizophrenia spectrum: Evidence from the Finnish adoptive family study of schizophrenia.
      ) = 8.6*

      Patient > Relative*
      Racea (White:Black:Other%)70.3:16.1:13.691.5:2.8:5.686.7:6.7:6.7χ2(
      • Scott J.
      • Chant D.
      • Andrews G.
      • McGrath J.
      Psychotic-like experiences in the general community: the correlates of CIDI psychosis screen items in an Australian sample, 2005/11/23.
      ) =14.5**

      Patient ≠ Relative**
      Parental educationb5.1 (1.1)5.3 (1.3)5.5 (1.4)n.s.
      Handednessc59.0 (46.5)66.6 (45.3)67.2 (36.8)n.s.
      mean RMS movement (mm)0.13 (0.06)0.10 (0.04)0.09 (0.05)F(2,231) = 8.0***

      Patient > Relative**, Patient > Control**
      CPZ equivalent4.3 (5.0)---
      DAST total score1.8 (1.5)1.5 (1.1)1.2 (0.6)F(2,225) = 4.3*

      Patient > Control*
      WRAT-IV WR102.4 (14.0)102.3 (12.2)109.2 (12.9)F(2,231) = 4.9**

      Patient < Control*, Relative < Control*
      FSIQ-298.0 (11.3)102.4 (10.9)106.7 (11.6)F(2,230) = 10.6***

      Patient < Relative***, Patient < Control*
      BACS z score-0.73 (0.71)-0.01 (0.73)0.26 (0.66)F(2,231) = 42.7***

      Patient < Relative***, Patient < Control***
      BPRS total score45.8 (12.4)31.7 (5.9)27.4 (4.0)F(2,231) = 81.7***

      Patient > Relative***, Patient > Control***, Relative > Control*
      SPQ positive12.1 (7.5)3.6 (5.2)1.6 (2.9)F(2,231) = 68.2***

      Patient > Relative***, Patient > Control***
      SPQ disorganized7.0 (4.5)3.7 (3.6)1.9 (2.6)F(2,231) = 32.7***

      Patient > Relative***, Patient > Control***, Relative > Control*
      SPQ negative13.9 (7.2)7.2 (6.1)4.2 (4.8)F(2,231) = 45.5***

      Patient > Relative***, Patient > Control***, Relative > Control*
      PID5 psychoticism37.3 (20.1)12.2 (12.7)9.2 (9.7)F(2,226) = 73.0***

      Patient > Relative***, Patient > Control***
      SIS PLEsd-1.8 (1.4)1.1 (1.4)t(95.5) = 2.6*
      SAPS total scoree17.3 (16.3)---
      SANS total scoree27.9 (16.8)---
      Note: mean (standard deviation). aRace: the “Other” category included American Indian or Alaskan Native; Asian or Pacific Islander; Hispanic; other; and unknown or unreported. bParental education: 1: 7th grade or less; 2: between 7th and 9th grade; 3: between 10th and 12th grade; 4: high school graduate/GED; 5: partial college; 6: college graduate; 7: graduate degree. cHandedness: Edinburgh Handedness Scale. dOnly completed by relatives and controls. eOnly completed by patients. CPZ: chlorpromazine. DAST: Drug Abuse Screen Test. WRAT-IV WR: Wide Range Achievement Test, Fourth Edition, Word Reading. FSIQ-2: Full Scale IQ estimated with the Similarities and Matrix Reasoning subtests of the WAIS-IV. BACS: Brief Assessment of Cognition in Schizophrenia. BPRS: Brief Psychiatric Rating Scale. SPQ: Schizotypy Personality Questionnaire. PID5: Personality Inventory for DSM-5. SIS: Structured Interview for Schizotypy. PLEs: psychotic-like experiences. SAPS: Scaled Assessment of Positive Symptoms. SANS: Scaled Assessment of Negative Symptoms. p < .05, ∗∗p < .01, ∗∗∗p < .001, n.s.= not significant. p values were not corrected for multiple comparisons. Follow-up tests were based on Tukey’s HSD test (for F tests) or Bonferroni correction (for χ2 tests).

      Neuroimaging Acquisition and Preprocessing

      Participants completed two hour-long MRI scanning sessions (A & B) on a 3T Siemens Prisma scanner at the Center for Magnetic Resonance Research at the UMN during their second visit. With the exception of hardware limitations, scanning protocol was closely matched to the Phase 1b HCP Lifespan on teens and young adults (https://humanconnectome.org/study-hcp-lifespan-pilot/phase1b-pilot-parameters). Scanning parameters were identical to the HCP Lifespan data acquired on the same 3T Prisma scanner (https://humanconnectome.org/storage/app/media/documentation/lifespan-pilot/LSCMRR_3T_printout_2014.08.15.pdf). Briefly, two 6.5-min rsfMRI scans (TR = 0.80s, voxel size = 2.0mm isotropic, number of volumes = 488) with opposite phase encoding directions (anterior-to-posterior and posterior-to-anterior) were obtained each scanning session. This resulted in four rsfMRI runs totaling 26 minutes. Session A also included high-resolution T1w and T2w scans (voxel size = 0.8mm isotropic).
      To maximize the compatibility of rsfMRI data between the HCP and the P-HCP, we preprocessed the images with the HCP minimal preprocessing pipeline (v3.27.0) and kept all other preprocessing steps identical to Study I.

      RsFC Composite Score

      We computed a rsFC composite score for each participant by applying Model I in Study I directly to the P-HCP data. First, we computed the rsFC between the 100 HCP ICs for P-HCP participants using the procedure in Study I. Next, we removed the effects of Model I covariates (age, sex, handedness, movement in scanner, and total brain volume) from rsFC by estimating their linear effects in the HCP and subtracting them from the P-HCP data. We then normalized the P-HCP rsFC values with the corresponding means and standard deviations estimated from the (covariance removed) HCP rsFC. The rsFC composite score was the weighted sum of rsFC with the standardized coefficients in Model I.
      Benefits of using the large and generally healthy HCP as a normative sample include: 1) better estimates of parameters and 2) avoiding removing variance of interest due to confounding between psychosis and covariates (e.g., head motion) in the P-HCP. Direct application of the HCP model and parameters to the P-HCP was facilitated by the two samples’ high parallelism in measurement and image acquisition and preprocessing, resulting in compatible measures of rsFC and covariates (Figure S4). To minimize extrapolation, we also excluded participants (11 patients, 2 relatives, and 2 controls) whose mean RMS movement exceeded the highest mean movement in Study I (0.206 mm). The excluded participants did not differ significantly from the remaining sample in the demographic, cognitive, and clinical variables listed in Table 3 (all p’s >.05). Lastly, the P-HCP sample consisted of a wider age range (18-68 years) than the young-adult HCP (22-36 years), and participants within different age ranges were analyzed separately as described below.

      Statistical Analysis

      The HCP model was derived from young adults (22-36 years) and may not generalize to older adults in the P-HCP. Thus, we median-split the final sample into younger (18-38 years, N = 112) and older subsamples (39-68 years, N = 107). We focused on hypothesis testing in the younger subsample and repeated the analyses as explorative in the older subsample. Table S3 compares the two subsamples.
      Group Differences in the RsFC Composite Score. We first tested the a priori hypothesis that the rsFC composite score was elevated in the patients, followed by the relatives, and then the controls. To this end, we fit a linear model regressing the rsFC composite score on group membership. Statistical significance was based on a confirmatory one-tailed significance level of 0.05 in the younger subsample.
      RsFC Composite Score and the Positive/Disorganized Dimension of Psychosis. We then tested the a priori hypothesis that the rsFC composite score positively correlated with the positive/disorganized dimension of psychosis (five indices). Trait-level indices included the positive and disorganized factors of the Schizotypy Personality Questionnaire (SPQ (
      • Raine A.
      • Lencz T.
      • Scerbo A.
      • Kim D.
      Cognitive-perceptual, interpersonal, and disorganized features of schizotypal personality.
      ,
      • Raine A.
      The SPQ : A Scale for the Assessment of Schizotypal Personality Based on DSM-III-R Criteria.
      )) and the psychoticism domain of the Personality Inventory for DSM-5 (PID5 (
      • Krueger R.F.
      • Derringer J.
      • Markon K.E.
      • Watson D.
      • Skodol A.E.
      Initial construction of a maladaptive personality trait model and inventory for DSM-5.
      )), available in all participants. Additionally, we constructed a “PLEs equivalent” for the Structured Interview for Schizotypy (SIS (
      • Kendler K.S.
      • Lieberman J.A.
      • Walsh D.
      The structured interview for schizotypy (SIS): A preliminary report.
      ), available in controls and relatives), which was the sum of the illusions, magical thinking, and odd behavior symptoms/signs, matching the PLEs construct in Study I. Symptom-level index was the total score on the Scaled Assessment of Positive Symptoms (SAPS (

      Andreasen NC (1984): Scale for the Assessment of Positive Symptoms (SAPS). University of Iowa, Iowa City.

      )) in patients. See Tables S5 & S6 for correlation among these measures.
      We tested the relationship between the rsFC composite score and the indices of psychosis by regressing each index onto the rsFC composite score, controlling for group main effect and rsFC × group interaction when applicable. Statistical significance was based on a confirmatory one-tailed significance level of 0.05 in the younger subsample.
      RsFC Composite Score and the Negative Dimension of Psychosis. Lastly, we examined the relationship between the rsFC composite score and the negative dimension of psychosis (two indices) as negative controls. Trait-level index was the negative factor of the SPQ in all participants. Symptom-level index was the total score on the Scaled Assessment of Negative Symptoms (SANS (
      • Andreasen N.C.
      The Scale for the Assessment of Negative Symptoms (SANS): conceptual and theoretical foundations.
      )) in patients. See Tables S5 & S6 for correlation among these measures. Linear models were the same as described in the previous paragraph. Statistical significance was based on a two-tailed significance level of 0.05.

      Study II Results

      Group Differences

      In the younger subsample, no significant group effect was found on the rsFC composite score (F(2,109) = 0.34, p > 0.05, Figure 3A). In the older subsample, no significant group effect was found when considering three groups (F(2,104) = 2.61, p = .08). However, the patient group in the older subsample had significantly higher rsFC composite score than the controls and relatives combined (F(1,105) = 5.15, ptwo-tailed = 0.025, Figure 3B), suggesting power was a limiting factor.
      Figure thumbnail gr3
      Figure 3Group Differences in the RsFC Composite Score in the P-HCP Younger and Older Subsamples. A) Boxplot of the rsFC composite score in patients, relatives, and controls in the younger subsample (18-38 years). B) Boxplot of the rsFC composite score in patients, relatives, and controls in the older subsample (39-68 years). rsFC: resting-state functional connectivity. *p < .05.

      Correlation with the Positive/Disorganized Dimension of Psychosis

      Linear regression revealed no significant group × rsFC interaction on any indices of psychosis in either subsample (all ps > .05), and the interaction term was dropped from all analyses. In the younger subsample, consistent with our hypothesis, the SIS PLEs equivalent (available in relatives and controls) significantly positively correlated with the rsFC composite score across the two groups (β = 0.25, t(
      • Andreasen N.C.
      The Scale for the Assessment of Negative Symptoms (SANS): conceptual and theoretical foundations.
      ) = 1.98, pone-tailed = 0.027, Figure 4A). Moreover, the SPQ positive factor significantly positively correlated with the rsFC score across patients, relatives, and controls (β = 0.14, t(108) = 1.75, pone-tailed = 0.041, Figure 4B). The rsFC composite score did not correlate with the SPQ disorganized factor (all participants, β = 0.06, t(108) = 0.67), PID5 psychoticism (all participants, β = 0.03, t(107) = 0.43), or SAPS total score (patients, β = -0.06, t(
      • Cao H.
      • Zhou H.
      • Cannon T.D.
      Functional connectome-wide associations of schizophrenia polygenic risk.
      ) = -0.43), all one-tailed ps > 0.05.
      Figure thumbnail gr4
      Figure 4Scatterplots between the RsFC Composite Score and the SIS_PLEs (A) and SPQ_Pos (B) in the P-HCP Younger Subsample. The SIS_PLEs were only completed by relatives and controls. All values are residuals with the effect of group removed. β: standardized regression coefficient. SPQ: Schizotypy Personality Questionnaire Pos: positive. SIS: Structured Interview for Schizotypy. PLEs: psychotic-like experiences. CI: confidence interval. *p < .05. p values were not corrected for multiple comparisons.
      In the older subsample, no significant relationship was found between the rsFC composite score and the SIS PLEs equivalent (β = -0.02, t(
      • Oertel-Knöchel V.
      • Knöchel C.
      • Matura S.
      • Prvulovic D.
      • Linden D.E.J.
      • van de Ven V.
      Reduced functional connectivity and asymmetry of the planum temporale in patients with schizophrenia and first-degree relatives.
      ) = -0.12), SPQ positive (β = -0.09, t(103) = -1.12) or disorganized factors (β = 0.06, t(103) = 0.68), PID5 psychoticism (β = -0.06, t(99) = -0.80), or SAPs total score (β = -0.02, t(
      • Krueger R.F.
      • Derringer J.
      • Markon K.E.
      • Watson D.
      • Skodol A.E.
      Initial construction of a maladaptive personality trait model and inventory for DSM-5.
      ) = -0.14), all two-tailed ps > 0.05.

      Correlation with the Negative Dimension of Psychosis

      For the negative control tests, no significant group × rsFC interaction was found in either subsample (all ps > .05), and the interaction term was dropped from all analyses. The rsFC composite score was not associated with the SPQ negative factor across all participants (younger subsample: β = 0.03, t(108) = 0.41, older subsample: β = -0.13, t(103) = -1.56) or the SANS total score in patients (younger subsample: β = -0.02, t(
      • Carter C.S.
      • MacDonald A.W.
      • Ross L.L.
      • Stenger V.A.
      Anterior cingulate cortex activity and impaired self-monitoring of performance in patients with schizophrenia: An event-related fMRI study.
      ) = -0.12, older subsample: β = 0.14, t(
      • Raine A.
      • Lencz T.
      • Scerbo A.
      • Kim D.
      Cognitive-perceptual, interpersonal, and disorganized features of schizotypal personality.
      ) = 1.04, all two-tailed ps > 0.05), supporting that the rsFC composite score was specifically related to the positive/disorganized dimension of psychosis.

      Discussion

      In the present study, we predicted 3.3% of variance in PLEs in a large community sample of young adults with a cross-validated human connectome model. Consistent with our hypothesis, cortical connections predictive of PLEs involved both hypo- and hyperconnectivity, mostly of the brain’s cognitive (frontoparietal, cingulo-opercular, dorsal attention, and default) networks. This model partially generalized to a clinical sample of similarly aged adults, explaining trait-level psychosis in patients, relatives, and healthy controls in two of the four indices tested. However, the model did not differentiate patients and relatives from controls or explain psychotic symptoms. By and large, it also failed to generalize to older adults.
      These findings provide direct evidence that PLEs share some rsFC substrates with more severe forms of psychosis. Compared with existing literature, our findings are unique in providing a quantifiable rsFC indicator of psychoticism, showing a potential approach (albeit very preliminary) for precision neuroscience. Meanwhile, the findings highlight important differences in neurobiology across the psychosis continuum.

      A cross-validated rsFC model predicted a small amount of variance in PLEs

      We predicted 3.3% of variance in PLEs in a large community sample when controlling for age, sex, and known rsFC confounds including head movement. The cross-validated R2 dropped to 2.0% when further controlling for g, race, years of education, and household income. These modest multiple R2’s, while unsatisfactory, are in line with observations from large-scale datasets. In the HCP, a similar approach predicted 4.4% of variance in the Raven’s progressive matrix (
      • Bilker W.B.
      • Hansen J.A.
      • Brensinger C.M.
      • Richard J.
      • Gur R.E.
      • Gur R.C.
      Development of abbreviated nine-item forms of the Raven’s standard progressive matrices test.
      ), 2.4% in the personality trait Openness to Experience, and none in the other Big Five personality traits (
      • Dubois J.
      • Galdi P.
      • Han Y.
      • Paul L.K.
      • Adolphs R.
      Resting-state functional brain connectivity best predicts the personality dimension of openness to experience.
      ). In the ABCD, multivariate modeling in a discovery sample (N > 1,900) resulted in out-of-sample r’s below 0.4 for measures of cognitive ability and below 0.2 for psychopathology (
      • Marek S.
      • Tervo-Clemmens B.
      • Calabro F.J.
      • Montez D.F.
      • Kay B.P.
      • Hatoum A.S.
      • et al.
      Towards reproducible brain-wide association studies affiliations.
      ). Also in the ABCD, |β|s for the association between single rsFC measures and adolescent PLEs (N > 3,000) did not exceed 0.07 (
      • Karcher N.R.
      • O’Brien K.J.
      • Kandala S.
      • Barch D.M.
      Resting-state functional connectivity and psychotic-like experiences in childhood: Results from the adolescent brain cognitive development study.
      ). The measurement quality of a construct strongly affects prediction performance (
      • Dubois J.
      • Galdi P.
      • Paul L.K.
      • Adolphs R.
      A distributed brain network predicts general intelligence from resting-state human neuroimaging data.
      ). Future PLEs studies will benefit from more extensive measurement of PLEs beyond the few items used in the present study.

      PLEs and dysregulated self-generated thought, aberrant salience processing, and beyond

      The rsFC correlates of PLEs are notable for default network dysconnectivity within itself and with the frontoparietal network. This pattern is consistent with observations in first episode, clinical high risk, high genetic risk, and subclinical psychosis (
      • Gong Q.
      • Hu X.
      • Pettersson-Yeo W.
      • Xu X.
      • Lui S.
      • Crossley N.
      • et al.
      Network-level dysconnectivity in drug-naïve first-episode psychosis: Dissociating transdiagnostic and diagnosis-specific alterations.
      ,
      • Shim G.
      • Oh J.S.
      • Jung W.
      • Jang J.
      • Choi C.-H.
      • Kim E.
      • et al.
      Altered resting-state connectivity in subjects at ultra-high risk for psychosis: an fMRI study.
      ,
      • Whitfield-Gabrieli S.
      • Thermenos H.W.
      • Milanovic S.
      • Tsuang M.T.
      • Faraone S.V.
      • McCarley R.W.
      • et al.
      Hyperactivity and hyperconnectivity of the default network in schizophrenia and in first-degree relatives of persons with schizophrenia.
      ,
      • Lui S.
      • Deng W.
      • Huang X.
      • Jiang L.
      • Ma X.
      • Chen H.
      • et al.
      Association of cerebral deficits with clinical symptoms in antipsychotic-naive first-episode schizophrenia: an optimized voxel-based morphometry and resting state functional connectivity study.
      ,
      • Xia C.H.
      • Ma Z.
      • Ciric R.
      • Gu S.
      • Betzel R.F.
      • Kaczkurkin A.N.
      • et al.
      Linked dimensions of psychopathology and connectivity in functional brain networks.
      ,
      • Guo W.
      • Jiang J.
      • Xiao C.
      • Zhang Z.
      • Zhang J.
      • Yu L.
      • et al.
      Decreased resting-state interhemispheric functional connectivity in unaffected siblings of schizophrenia patients.
      ,
      • Li M.
      • Deng W.
      • He Z.
      • Wang Q.
      • Huang C.
      • Jiang L.
      • et al.
      A splitting brain: Imbalanced neural networks in schizophrenia.
      ). PLEs may be associated with altered self-generated thought (e.g., mind-wandering, mentalizing, self-referential thought) in the default network (
      • Andrews-Hanna J.R.
      • Smallwood J.
      • Spreng R.N.
      The default network and self-generated thought: Component processes, dynamic control, and clinical relevance.
      ) that is inadequately regulated by the goal-directed, externally-oriented processes in the frontoparietal network. While a meta-analysis in chronic patients indicated mainly hypoconnectivity between the frontoparietal and default networks (
      • Dong D.
      • Wang Y.
      • Chang X.
      • Luo C.
      • Yao D.
      Dysfunction of large-scale brain networks in schizophrenia: A meta-analysis of resting-state functional connectivity.
      ), this pattern may be due to antipsychotic medication rather than psychosis per se (
      • Anticevic A.
      • Hu X.
      • Xiao Y.
      • Hu J.
      • Li F.
      • Bi F.
      • et al.
      Early-course unmedicated schizophrenia patients exhibit elevated prefrontal connectivity associated with longitudinal change.
      ).
      The cingulo-opercular network stood out for disrupted connections with almost all other networks, highlighting its role in integrating sensory, motor, affective, and cognitive information and switching attention (

      Menon V (2015): Salience Network. Brain Mapping: An Encyclopedic Reference, vol. 2. Elsevier Inc. https://doi.org/10.1016/B978-0-12-397025-1.00052-X

      ). Aberrant monitoring and salience processing is likely a core pathophysiology of psychosis (
      • Kapur S.
      Psychosis as a state of aberrant salience: A framework linking biology, phenomenology, and pharmacology in schizophrenia.
      ,
      • Nekovarova T.
      • Fajnerova I.
      • Horacek J.
      • Spaniel F.
      Bridging disparate symptoms of schizophrenia: a triple network dysfunction theory.
      ,
      • Carter C.S.
      • MacDonald A.W.
      • Ross L.L.
      • Stenger V.A.
      Anterior cingulate cortex activity and impaired self-monitoring of performance in patients with schizophrenia: An event-related fMRI study.
      ).
      The dorsal attention and visual networks were implied in some predictive connections. Disruptions with attention control and visual processing have long been noted in schizophrenia (
      • Guo W.
      • Jiang J.
      • Xiao C.
      • Zhang Z.
      • Zhang J.
      • Yu L.
      • et al.
      Decreased resting-state interhemispheric functional connectivity in unaffected siblings of schizophrenia patients.
      ,
      • Khadka S.
      • Meda S.A.
      • Stevens M.C.
      • Glahn D.C.
      • Calhoun V.D.
      • Sweeney J.A.
      • et al.
      Is aberrant functional connectivity a psychosis endophenotype? A resting state functional magnetic resonance imaging study.
      ,
      • Oertel-Knöchel V.
      • Knöchel C.
      • Matura S.
      • Prvulovic D.
      • Linden D.E.J.
      • van de Ven V.
      Reduced functional connectivity and asymmetry of the planum temporale in patients with schizophrenia and first-degree relatives.
      ). Our findings suggest that these disruptions may also underlie PLEs. Interestingly, connectivity of a network comprised of frontoparietal, default, and visual regions correlated with polygenic risk for schizophrenia in the HCP (
      • Cao H.
      • Zhou H.
      • Cannon T.D.
      Functional connectome-wide associations of schizophrenia polygenic risk.
      ).

      Partial generalization to an independent sample enriched with psychosis

      In younger adults, the rsFC model for PLEs generalized to the SIS, which is a clinician-rated measure of schizotypy. In particular, the odd behavior sign in the SIS was based on clinical observation. The model also generalized to the SPQ positive factor, which involved traits of suspiciousness and ideas of references that were not assessed in the original PLEs questions. Thus, the PLEs appears to be a simple yet valid measure of the lower end of the psychosis continuum. On the other hand, the PLEs model was not correlated with the negative dimension of psychosis, suggesting its specificity as a marker for the positive/disorganized dimension. Lastly, the lack of correlation with the SPQ disorganized factor and the PID5 psychoticism domain may point to important nuances in the rsFC substrates of various psychotic traits.
      Inconsistent with our hypothesis, the rsFC model for PLEs failed to explain positive psychotic symptoms in patients. Symptom-level psychosis may involve distinct and more profound rsFC abnormalities that cannot be postulated by studying subclinical psychosis. Moreover, PLEs, as subclinical phenomena, may be endorsed due to psychopathologies other than psychotic disorders, such as depression and anxiety (
      • Legge S.
      • Jones H.
      • Kendall K.
      • Pardiñas A.
      • Menzies G.
      • Bracher-Smith M.
      • et al.
      Genetic association study of psychotic experiences in UK Biobank.
      ,
      • Varghese D.
      • Scott J.
      • Welham J.
      • Bor W.
      • Najman J.
      • O’Callaghan M.
      • et al.
      Psychotic-like experiences in major depression and anxiety disorders: a population-based survey in young adults.
      ,
      • Yung A.R.
      • Buckby J.A.
      • Cotton S.M.
      • Cosgrave E.M.
      • Killackey E.J.
      • Stanford C.
      • et al.
      Psychotic-like experiences in nonpsychotic help-seekers: associations with distress, depression, and disability.
      ). Lastly, psychotic symptoms in most P-HCP patients were actively managed by medication, and results may be different in a medication-naive sample.
      The fact that our model largely failed to generalize to a sample of older adults is a reminder against extrapolation. That the rsFC composite score was higher in older patients than relatives and controls possibly suggests neurodegenerative processes or impact of antipsychotic medication targeting rsFC substrates of the psychosis continuum. However, age-matched samples are necessary to test these postulations.

      Strengths

      To our knowledge, this is the first study to directly show that brain correlates of PLEs
      also underlie trait-level psychosis in patients with psychosis and their relatives. By joining the HCP and P-HCP datasets, we demonstrated the feasibility of combining independently collected HCP-style data in an actuarial fashion. This type of application (albeit very preliminary) is an important foundation for precision neuroscience.
      One strength of this study is incorporating reliability information in machine learning. While rsFC measured with low reliability can well be associated with PLEs, the cost of compromised model performance with noisy variables was deemed larger than the benefit of discovering these connections. To prove this point, we reran the models in Study I with all 4950 cortical connections. R2 dropped to 1.7% for Model I and 0.3% for Model II.

      Limitations

      The present analysis was restricted to cortical networks, and did not include subcortical and cerebellar regions despite their critical role in psychosis (
      • Woodward N.D.
      • Karbasforoushan H.
      • Heckers S.
      Thalamocortical dysconnectivity in schizophrenia.
      ,
      • Dandash O.
      • Fornito A.
      • Lee J.
      • Keefe R.S.E.
      • Chee M.W.L.
      • Adcock R.A.
      • et al.
      Altered striatal functional connectivity in subjects with an at-risk mental state for psychosis.
      ,
      • Bernard J.A.
      • Dean D.J.
      • Kent J.S.
      • Orr J.M.
      • Pelletier-Baldelli A.
      • Lunsford-Avery J.R.
      • et al.
      Cerebellar networks in individuals at ultra high-risk of psychosis: Impact on postural sway and symptom severity.
      ,
      • Zhou Y.
      • Shu N.
      • Liu Y.
      • Song M.
      • Hao Y.
      • Liu H.
      • et al.
      Altered resting-state functional connectivity and anatomical connectivity of hippocampus in schizophrenia.
      ). Simultaneous cortical and subcortical parcellation is challenging due to their disparity in signal-to-noise ratio (
      • Schaefer A.
      • Kong R.
      • Gordon E.M.
      • Laumann T.O.
      • Zuo X.-N.
      • Holmes A.J.
      • et al.
      Local-global parcellation of the human cerebral cortex from intrinsic functional connectivity MRI.
      ,
      • Yeo B.T.
      • Krienen F.M.
      • Chee M.W.L.
      • Buckner R.L.
      Estimates of segregation and overlap of functional connectivity networks in the human cerebral cortex.
      ,
      • Glasser M.F.
      • Coalson T.S.
      • Robinson E.C.
      • Hacker C.D.
      • Harwell J.
      • Yacoub E.
      • et al.
      A multi-modal parcellation of human cerebral cortex.
      ). Future studies can benefit greatly from reliable and comparable measurement of cortical and subcortical rsFC at the same time.
      Another limitation is that both the HCP and P-HCP comprised of mostly White participants. To make meaningful progress towards precision neuroscience, data from more diverse and representative cohorts are necessary.
      Like many data mining studies, interpretation of our rsFC model with many small effect sizes is not straightforward. Frist, knowledge of data-driven brain components derived at a high dimensionality is limited. Second, beta coefficients in multiple regression are affected by all other variables in the model and deviate greatly from zero-order correlations. We have made the brain components used in this study as well as their reliability information available online (https://balsa.wustl.edu/study/show/MxKx0) for readers who would like to develop further insights into our findings.

      Conclusions

      In this work, we cross-validated rsFC underpinnings of PLEs in the general population that partially generalized to patients with psychosis and first-degree relatives, providing evidence for shared rsFC mechanisms across the psychosis continuum. PLEs are a valid approach to probing the neurobiology of psychosis. Shared and distinct variances across the psychosis continuum are equally important.

      Acknowledgements and Disclosures

      The HCP data were provided [in part] by the Human Connectome Project, WU-Minn Consortium (Principal Investigators: David Van Essen and Kamil Ugurbil; 1U54MH091657) funded by the 16 NIH Institutes and Centers that support the NIH Blueprint for Neuroscience Research; and by the McDonnell Center for Systems Neuroscience at Washington University. The P-HCP was funded by the NIH (Principal Investigator: Scott Sponheim; 1U01MH108150-01A1). YM and AWMIII were funded by the University of Minnesota Informatics Institute, On the Horizon grant.

      Supplementary Material

      The authors report no biomedical financial interests or potential conflicts of interest.

      References

        • Jablensky A.
        Epidemiology of schizophrenia: The global burden of disease and disability.
        Eur Arch Psychiatry Clin Neurosci. 2000; 250: 274-285
        • Tienari P.
        • Wynne L.C.
        • Läksy K.
        • Moring J.
        • Nieminen P.
        • Sorri A.
        • et al.
        Genetic boundaries of the schizophrenia spectrum: Evidence from the Finnish adoptive family study of schizophrenia.
        Am J Psychiatry. 2003; 160: 1587-1594
        • Bourgin J.
        • Tebeka S.
        • Mallet J.
        • Mazer N.
        • Dubertret C.
        • Le Strat Y.
        Prevalence and correlates of psychotic-like experiences in the general population.
        Schizophr Res. 2020; 215: 371-377
        • Scott J.
        • Chant D.
        • Andrews G.
        • McGrath J.
        Psychotic-like experiences in the general community: the correlates of CIDI psychosis screen items in an Australian sample, 2005/11/23.
        Psychol Med. 2006; 36: 231-238
        • Van Os J.
        • Linscott R.J.
        • Myin-Germeys I.
        • Delespaul P.
        • Krabbendam L.
        A systematic review and meta-analysis of the psychosis continuum: Evidence for a psychosis proneness-persistence-impairment model of psychotic disorder.
        Psychol Med. 2009; 39: 179
        • Esterberg M.L.
        • Compton M.T.
        The psychosis continuum and categorical versus dimensional diagnostic approaches.
        Curr Psychiatry Rep. 2009; 11: 179-184
        • Alemany S.
        • Arias B.
        • Aguilera M.
        • Villa H.
        • Moya J.
        • Ibanez M.I.
        • et al.
        Childhood abuse, the BDNF-Val66Met polymorphism and adult psychotic-like experiences.
        Br J Psychiatry. 2011; 199: 38-42
        • Pain O.
        • Dudbridge F.
        • Cardno A.G.
        • Freeman D.
        • Lu Y.
        • Lundstrom S.
        • et al.
        Genome‐wide analysis of adolescent psychotic‐like experiences shows genetic overlap with psychiatric disorders.
        Am J Med Genet Part B Neuropsychiatr Genet. 2018; 177: 416-425
        • Orr J.M.
        • Turner J.A.
        • Mittal V.A.
        Widespread brain dysconnectivity associated with psychotic-like experiences in the general population.
        NeuroImage Clin. 2014; 4: 343-351
        • Friston K.J.
        • Frith C.D.
        Schizophrenia: a disconnection syndrome?.
        Clin Neurosci. 1995; 3: 89-97
        • Sheffield J.M.
        • Kandala S.
        • Burgess G.C.
        • Harms M.P.
        • Barch D.M.
        Cingulo-opercular network efficiency mediates the association between psychotic-like experiences and cognitive ability in the general population.
        Biol Psychiatry Cogn Neurosci Neuroimaging. 2016; 1: 498-506
        • Blain S.D.
        • Grazioplene R.G.
        • Ma Y.
        • DeYoung C.G.
        Toward a neural model of the Openness-Psychoticism dimension: Functional connectivity in the default and frontoparietal control networks.
        Schizophr Bull. 2020; 46: 540-551
        • Barber A.D.
        • Lindquist M.A.
        • Derosse P.
        • Karlsgodt K.H.
        Dynamic functional connectivity states reflecting psychotic-like experiences.
        Biol Psychiatry Cogn Neurosci Neuroimaging. 2018; 3: 443-453
        • Sabaroedin K.
        • Tiego J.
        • Parkes L.
        • Sforazzini F.
        • Finlay A.
        • Johnson B.
        • et al.
        Functional connectivity of corticostriatal circuitry and psychosis-like experiences in the general community.
        Biol Psychiatry. 2019; 86: 16-24
        • Pani S.M.
        • Sabaroedin K.
        • Tiego J.
        • Bellgrove M.A.
        • Fornito A.
        A multivariate analysis of the association between corticostriatal functional connectivity and psychosis-like experiences in the general community.
        Psychiatry Res Neuroimaging. 2021; 30711202
        • Karcher N.R.
        • O’Brien K.J.
        • Kandala S.
        • Barch D.M.
        Resting-state functional connectivity and psychotic-like experiences in childhood: Results from the adolescent brain cognitive development study.
        Biol Psychiatry. 2019; 86: 7-15
        • Nelson M.T.
        • Seal M.L.
        • Phillips L.J.
        • Merritt A.H.
        • Wilson R.
        • Pantelis C.
        An investigation of the relationship between cortical connectivity and schizotypy in the general population.
        J Nerv Ment Dis. 2011; 199: 348-353
        • Waltmann M.
        • O’Daly O.
        • Egerton A.
        • McMullen K.
        • Kumari V.
        • Barker G.J.
        • et al.
        Multi-echo fMRI, resting-state connectivity, and high psychometric schizotypy.
        NeuroImage Clin. 2019; 21101603
        • Zhu Y.
        • Tang Y.
        • Zhang T.
        • Li H.
        • Tang Y.
        • Li C.
        • et al.
        Reduced functional connectivity between bilateral precuneus and contralateral parahippocampus in schizotypal personality disorder.
        BMC Psychiatry. 2017; 17: 1-7
        • Kozhuharova P.
        • Saviola F.
        • Diaconescu A.
        • Allen P.
        High schizotypy traits are associated with reduced hippocampal resting state functional connectivity.
        Psychiatry Res Neuroimaging. 2021; 307111215
        • Yang Z.
        • Zhang R.
        • Li Y.
        • Wang Y.
        • Wang Y.
        • Wang S.
        • et al.
        Functional connectivity of the default mode network is associated with prospection in schizophrenia patients and individuals with social anhedonia.
        Prog Neuro-Psychopharmacology Biol Psychiatry. 2019; 92: 412-420
        • Dong D.
        • Wang Y.
        • Chang X.
        • Luo C.
        • Yao D.
        Dysfunction of large-scale brain networks in schizophrenia: A meta-analysis of resting-state functional connectivity.
        Schizophr Bull. 2017; : 1-14
        • Pettersson-Yeo W.
        • Allen P.
        • Benetti S.
        • McGuire P.
        • Mechelli A.
        Dysconnectivity in schizophrenia: Where are we now?.
        Neurosci Biobehav Rev. 2011; 35: 1110-1124
        • Marek S.
        • Tervo-Clemmens B.
        • Calabro F.J.
        • Montez D.F.
        • Kay B.P.
        • Hatoum A.S.
        • et al.
        Towards reproducible brain-wide association studies affiliations.
        BioRxiv. 2020; https://doi.org/10.1101/2020.08.21.257758v
        • Dubois J.
        • Galdi P.
        • Paul L.K.
        • Adolphs R.
        A distributed brain network predicts general intelligence from resting-state human neuroimaging data.
        Philos Trans R Soc B Biol Sci. 2018; 373https://doi.org/10.1098/rstb.2017.0284
        • Jenkinson M.
        • Bannister P.
        • Brady M.
        • Smith S.M.
        Improved optimization for the robust and accurate linear registration and motion correction of brain images.
        Neuroimage. 2002; 17: 825-841
        • Van Essen D.C.
        • Smith S.M.
        • Barch D.M.
        • Behrens T.E.J.
        • Yacoub E.
        • Ugurbil K.
        The WU-Minn Human Connectome Project: An overview.
        Neuroimage. 2013; 80: 62-79
      1. Achenbach TM (2009): The Achenbach System of Empirically Based Assessment (ASEBA): Development, Findings, Theory, and Applications. University of Vermont, Research Center for Children, Youth, & Families.

        • Cronbach L.J.
        Coefficient alpha and the internal structure of tests.
        Psychometrika. 1951; 16: 297-334
      2. Ma Y, Macdonald AW (2021): Impact of ICA dimensionality on the test-retest reliability of resting-state functional connectivity. Brain Connect ahead of print.

        • Smith S.M.
        • Hyva¨rinen A.
        • Varoquaux G.
        • Miller K.L.
        • Beckmann C.F.
        Group-PCA for very large fMRI datasets.
        Neuroimage. 2014; 101: 738-749
        • Beckmann C.F.
        • Smith S.M.
        Probabilistic independent component analysis for functional magnetic resonance imaging.
        IEEE Trans Med Imaging. 2004; 23: 137-152
        • Yeo B.T.
        • Krienen F.M.
        • Sepulcre J.
        • Sabuncu M.R.
        • Lashkari D.
        • Hollinshead M.
        • et al.
        The organization of the human cerebral cortex estimated by intrinsic functional connectivity.
        J Neurophysiol. 2011; 106: 1125-1165
        • Beckmann C.F.
        • Mackay C.E.
        • Filippini N.
        • Smith S.M.
        Group comparison of resting-state FMRI data using multi-subject ICA and dual regression.
        Neuroimage. 2009; 47: S148
        • Cicchetti D.V.
        • Sparrow S.A.
        Developing criteria for establishing interrater reliability of specific items: Applications to assessment of adaptive behavior.
        Am J Ment Defic. 1981; 86: 127-137
        • Barrett J.P.
        The coefficient of determination—some limitations.
        Am Stat. 1974; 28: 19-20
        • Winkler A.M.
        • Webster M.A.
        • Vidaurre D.
        • Nichols T.E.
        • Smith S.M.
        Multi-level block permutation.
        Neuroimage. 2015; 123: 253-268
        • Demro C.
        • Mueller B.A.
        • Kent J.S.
        • Burton P.C.
        • Olman C.A.
        • Schallmo M.P.
        • et al.
        The psychosis human connectome project: An overview.
        Neuroimage. 2021; 241118439
        • Raine A.
        • Lencz T.
        • Scerbo A.
        • Kim D.
        Cognitive-perceptual, interpersonal, and disorganized features of schizotypal personality.
        Schizophr Bull. 1994; 20: 191-201
        • Raine A.
        The SPQ : A Scale for the Assessment of Schizotypal Personality Based on DSM-III-R Criteria.
        Schizophr Bull. 1991; 17: 555-564
        • Krueger R.F.
        • Derringer J.
        • Markon K.E.
        • Watson D.
        • Skodol A.E.
        Initial construction of a maladaptive personality trait model and inventory for DSM-5.
        Psychol Med. 2012; 42: 1879
        • Kendler K.S.
        • Lieberman J.A.
        • Walsh D.
        The structured interview for schizotypy (SIS): A preliminary report.
        Schizophr Bull. 1989; 15: 559-571
      3. Andreasen NC (1984): Scale for the Assessment of Positive Symptoms (SAPS). University of Iowa, Iowa City.

        • Andreasen N.C.
        The Scale for the Assessment of Negative Symptoms (SANS): conceptual and theoretical foundations.
        Br J psychiatry. 1989; 155: 49-52
        • Bilker W.B.
        • Hansen J.A.
        • Brensinger C.M.
        • Richard J.
        • Gur R.E.
        • Gur R.C.
        Development of abbreviated nine-item forms of the Raven’s standard progressive matrices test.
        Assessment. 2012; 19: 354-369
        • Dubois J.
        • Galdi P.
        • Han Y.
        • Paul L.K.
        • Adolphs R.
        Resting-state functional brain connectivity best predicts the personality dimension of openness to experience.
        Personal Neurosci. 2018; 1https://doi.org/10.1017/pen.2018.8
        • Gong Q.
        • Hu X.
        • Pettersson-Yeo W.
        • Xu X.
        • Lui S.
        • Crossley N.
        • et al.
        Network-level dysconnectivity in drug-naïve first-episode psychosis: Dissociating transdiagnostic and diagnosis-specific alterations.
        Neuropsychopharmacology. 2017; 42: 933-940
        • Shim G.
        • Oh J.S.
        • Jung W.
        • Jang J.
        • Choi C.-H.
        • Kim E.
        • et al.
        Altered resting-state connectivity in subjects at ultra-high risk for psychosis: an fMRI study.
        Behav Brain Funct. 2010; 6: 58
        • Whitfield-Gabrieli S.
        • Thermenos H.W.
        • Milanovic S.
        • Tsuang M.T.
        • Faraone S.V.
        • McCarley R.W.
        • et al.
        Hyperactivity and hyperconnectivity of the default network in schizophrenia and in first-degree relatives of persons with schizophrenia.
        Proc Natl Acad Sci U S A. 2009; 106: 1279-1284
        • Lui S.
        • Deng W.
        • Huang X.
        • Jiang L.
        • Ma X.
        • Chen H.
        • et al.
        Association of cerebral deficits with clinical symptoms in antipsychotic-naive first-episode schizophrenia: an optimized voxel-based morphometry and resting state functional connectivity study.
        Am J Psychiatry. 2009; 166: 196-205
        • Xia C.H.
        • Ma Z.
        • Ciric R.
        • Gu S.
        • Betzel R.F.
        • Kaczkurkin A.N.
        • et al.
        Linked dimensions of psychopathology and connectivity in functional brain networks.
        Nat Commun. 2018; 0: 1-14
        • Guo W.
        • Jiang J.
        • Xiao C.
        • Zhang Z.
        • Zhang J.
        • Yu L.
        • et al.
        Decreased resting-state interhemispheric functional connectivity in unaffected siblings of schizophrenia patients.
        Schizophr Res. 2014; 152: 170-175
        • Li M.
        • Deng W.
        • He Z.
        • Wang Q.
        • Huang C.
        • Jiang L.
        • et al.
        A splitting brain: Imbalanced neural networks in schizophrenia.
        Psychiatry Res - Neuroimaging. 2015; 232: 145-153
        • Andrews-Hanna J.R.
        • Smallwood J.
        • Spreng R.N.
        The default network and self-generated thought: Component processes, dynamic control, and clinical relevance.
        Ann N Y Acad Sci. 2014; 1316: 29-52
        • Anticevic A.
        • Hu X.
        • Xiao Y.
        • Hu J.
        • Li F.
        • Bi F.
        • et al.
        Early-course unmedicated schizophrenia patients exhibit elevated prefrontal connectivity associated with longitudinal change.
        J Neurosci. 2015; 35: 267-286
      4. Menon V (2015): Salience Network. Brain Mapping: An Encyclopedic Reference, vol. 2. Elsevier Inc. https://doi.org/10.1016/B978-0-12-397025-1.00052-X

        • Kapur S.
        Psychosis as a state of aberrant salience: A framework linking biology, phenomenology, and pharmacology in schizophrenia.
        Am J Psychiatry. 2003; 160: 13-23
        • Nekovarova T.
        • Fajnerova I.
        • Horacek J.
        • Spaniel F.
        Bridging disparate symptoms of schizophrenia: a triple network dysfunction theory.
        Front Behav Neurosci. 2014; 8: 171
        • Carter C.S.
        • MacDonald A.W.
        • Ross L.L.
        • Stenger V.A.
        Anterior cingulate cortex activity and impaired self-monitoring of performance in patients with schizophrenia: An event-related fMRI study.
        Am J Psychiatry. 2001; 158: 1423-1428
        • Khadka S.
        • Meda S.A.
        • Stevens M.C.
        • Glahn D.C.
        • Calhoun V.D.
        • Sweeney J.A.
        • et al.
        Is aberrant functional connectivity a psychosis endophenotype? A resting state functional magnetic resonance imaging study.
        Biol Psychiatry. 2013; 74: 458-466
        • Oertel-Knöchel V.
        • Knöchel C.
        • Matura S.
        • Prvulovic D.
        • Linden D.E.J.
        • van de Ven V.
        Reduced functional connectivity and asymmetry of the planum temporale in patients with schizophrenia and first-degree relatives.
        Schizophr Res. 2013; 147: 331-338
        • Cao H.
        • Zhou H.
        • Cannon T.D.
        Functional connectome-wide associations of schizophrenia polygenic risk.
        Mol Psychiatry. 2020; : 1-9
        • Legge S.
        • Jones H.
        • Kendall K.
        • Pardiñas A.
        • Menzies G.
        • Bracher-Smith M.
        • et al.
        Genetic association study of psychotic experiences in UK Biobank.
        JAMA Psychiatry. 2019; : 1-22
        • Varghese D.
        • Scott J.
        • Welham J.
        • Bor W.
        • Najman J.
        • O’Callaghan M.
        • et al.
        Psychotic-like experiences in major depression and anxiety disorders: a population-based survey in young adults.
        Schizophr Bull. 2011; 37: 389-393
        • Yung A.R.
        • Buckby J.A.
        • Cotton S.M.
        • Cosgrave E.M.
        • Killackey E.J.
        • Stanford C.
        • et al.
        Psychotic-like experiences in nonpsychotic help-seekers: associations with distress, depression, and disability.
        Schizophr Bull. 2006; 32: 352-359
        • Woodward N.D.
        • Karbasforoushan H.
        • Heckers S.
        Thalamocortical dysconnectivity in schizophrenia.
        Am J Psychiatry. 2012; 169: 1092-1099
        • Dandash O.
        • Fornito A.
        • Lee J.
        • Keefe R.S.E.
        • Chee M.W.L.
        • Adcock R.A.
        • et al.
        Altered striatal functional connectivity in subjects with an at-risk mental state for psychosis.
        Schizophr Bull. 2014; 40: 904-913
        • Bernard J.A.
        • Dean D.J.
        • Kent J.S.
        • Orr J.M.
        • Pelletier-Baldelli A.
        • Lunsford-Avery J.R.
        • et al.
        Cerebellar networks in individuals at ultra high-risk of psychosis: Impact on postural sway and symptom severity.
        Hum Brain Mapp. 2014; 35: 4064-4078
        • Zhou Y.
        • Shu N.
        • Liu Y.
        • Song M.
        • Hao Y.
        • Liu H.
        • et al.
        Altered resting-state functional connectivity and anatomical connectivity of hippocampus in schizophrenia.
        Schizophr Res. 2008; 100: 120-132
        • Schaefer A.
        • Kong R.
        • Gordon E.M.
        • Laumann T.O.
        • Zuo X.-N.
        • Holmes A.J.
        • et al.
        Local-global parcellation of the human cerebral cortex from intrinsic functional connectivity MRI.
        Cereb Cortex. 2018; 28: 3095-3114
        • Yeo B.T.
        • Krienen F.M.
        • Chee M.W.L.
        • Buckner R.L.
        Estimates of segregation and overlap of functional connectivity networks in the human cerebral cortex.
        Neuroimage. 2014; 88: 212-227
        • Glasser M.F.
        • Coalson T.S.
        • Robinson E.C.
        • Hacker C.D.
        • Harwell J.
        • Yacoub E.
        • et al.
        A multi-modal parcellation of human cerebral cortex.
        Nature. 2016; 536: 171-178