Auditory and Visual Thalamocortical Connectivity Alterations in Unmedicated People with Schizophrenia: An Individualized Sensory Thalamic Localization and Resting-State Functional Connectivity Study

Abstract

Background: Converging evidence from clinical neuroimaging and animal models has strongly implicated dysfunction of thalamocortical circuits in the pathophysiology of schizophrenia. Preclinical models of genetic risk for schizophrenia have shown reduced synaptic transmission from auditory thalamus to primary auditory cortex, which may represent a correlate of auditory disturbances such as hallucinations. Human neuroimaging studies, however, have found a generalized increase in resting state functional connectivity (RSFC) between whole thalamus and sensorimotor cortex in people with schizophrenia (PSZ). We aimed to more directly translate preclinical findings by specifically localizing auditory and visual thalamic nuclei in unmedicated PSZ and measuring RSFC to primary sensory cortices.

Methods: In this case-control study, 82 unmedicated PSZ and 55 matched healthy controls (HC) completed RSFC functional magnetic resonance imaging (fMRI). Auditory and visual thalamic nuclei were localized for 55 unmedicated PSZ and 46 HC who additionally completed a sensory thalamic nuclei localizer fMRI task (N = 101). Using localized nuclei as RSFC seeds we assessed group differences in auditory and visual thalamocortical connectivity and associations with positive symptom severity.

Results: Auditory thalamocortical connectivity was not significantly different between PSZ and HC, but hyperconnectivity was associated with greater positive symptom severity in bilateral superior temporal gyrus. Visual thalamocortical connectivity was significantly greater in PSZ relative to HC in secondary and higher-order visual cortex, but not predictive of positive symptom severity.

Conclusion: These results indicate that visual thalamocortical hyperconnectivity is a generalized marker of schizophrenia, while hyperconnectivity in auditory thalamocortical circuits relates more specifically to positive symptom severity.

Keywords: 22q11 deletion; Schizophrenia; auditory system; functional localization; functional magnetic resonance imaging; lateral geniculate nucleus; medial geniculate nucleus; positive symptoms; resting-state functional connectivity; sensory processing; sensory thalamic localizer task; thalamocortical; thalamus; visual system.

Figure 1.

T-statistic maps for thalamic localizer (TL) functionally defined regions of interest (fROIs) for group differences and associations with positive symptom severity in medial geniculate nucleus (MGN) and lateral geniculate nucleus (LGN) seed resting-state functional connectivity (RSFC) in the RS+TL sample (total N = 101). Significance was assessed grayordinate-wise using cortical surface GIFTIs in Permutation Analysis of Linear Models (PALM) within auditory cortex (AC) and visual cortex (VC) masks, using 10,000 permutations and threshold-free cluster enhancement. P-values were corrected for family-wise error rate; t-statistic maps were thresholded at α = 0.05. A) Probability density maps of MGN fROIs obtained from the TL task across participants in the RS+TL sample, in Montreal Neurological Institute 152 non-linear 6th-generation (MNI152NLin6) space. B) MGN RS seed connectivity associations with Positive and Negative Symptom Scale (PANSS) Positive subscale scores in auditory cortex (AC), controlling for age, biological sex, handedness, and schizophrenia diagnosis, as well as PANSS negative and general subscores. Gray underlay shows the auditory cortex mask used in PALM. 28 significant grayordinates. PANSS data were available for 52 patients with schizophrenia (PSZ) and 42 healthy controls (HC); total N = 94. C) MGN seed connectivity extracted from the significant grayordinates visualized in B, with MGN resting-state functional connectivity (RSFC) residualized to covariates, displaying the identified relationship between MGN seed connectivity and positive symptom severity. D) Probability density maps of LGN fROIs obtained from the TL task across participants in the RS+TL sample, in MNI152NLin6 space. E) LGN RS seed connectivity group differences between PSZ (n = 55) and HC (n = 46) in visual cortex (VC), controlling for age, biological sex, and handedness. 245 significant grayordinates. F) LGN seed connectivity extracted from the significant grayordinates visualized in E, displaying the distribution of LGN seed connectivity in HC and PSZ. Dashed lines in panels C and F indicate group means ± 1 standard deviation.

Figure 2.

Panels A-D show whole-cortex t-statistic maps for grayordinates exhibiting significant resting-state functional connectivity (RSFC) from medial geniculate nucleus (MGN; A, B) and lateral geniculate nucleus (LGN; C, D) seeds in healthy controls (HC; A, C) and people with schizophrenia (PSZ, B, D), from the RS+TL sample (55 PSZ, 46 HC; total N = 101). A) 25,166 significant grayordinates (2,971 positive, 22,195 negative). B) 23,177 significant grayordinates (2,012 positive, 21,165 negative). C) 27,497 significant grayordinates (4,417 positive, 23,080 negative). D) 21,820 significant grayordinates (5,312 positive, 16,508 negative). E) Whole-cortex t-statistic maps for group differences in LGN seed RSFC in the RS+TL sample, assessed using a design matrix with regressors for diagnosis, age, biological sex, and handedness. 1,824 significant grayordinates. Significance was assessed in Permutation Analysis of Linear Models, using 10,000 permutations or sign-flips and threshold-free cluster enhancement. P-values were corrected for family-wise error rate and t-statistic maps were thresholded at α = 0.05.

Figure 3.

Panels A-B show whole-cortex t-statistic maps for grayordinates exhibiting significant resting-state functional connectivity (RSFC) from whole thalamus seeds in healthy controls (HC; A) and people with schizophrenia (PSZ, B), from the RS-only sample (82 PSZ, 55 HC; total N = 137). A) 28,971 significant grayordinates (4,095 positive, 24,876 negative). B) 23,347 significant grayordinates (4,551 positive, 18,796 negative). Panels C-F show t-statistic maps for whole thalamus seed resting-state functional connectivity (RSFC) group differences and associations with positive symptom severity RS-only sample. C) Whole thalamus seed RSFC group differences between people with schizophrenia (PSZ; n = 82), and healthy controls (HC; n = 55), controlling for age, biological sex, and handedness. 6,547 significant grayordinates. Total N = 137. D) Whole thalamus seed RSFC associations with PANSS Positive subscale scores, controlling for age, biological sex, handedness, and schizophrenia diagnosis, as well and PANSS negative and general subscores. PANSS scores were available for 78 PSZ and 51 HC (total N = 129). 9 significant grayordinates. E) Whole thalamus seed connectivity extracted from the significant grayordinates visualized in D, with thalamic RSFC residualized to covariates, displaying the identified relationship between thalamic seed connectivity and positive symptom severity. Significance was assessed grayordinate-wise in whole cortex using cortical surface GIFTIs in Permutation Analysis of Linear Models in whole cortex, using 10,000 permutations and threshold-free cluster enhancement. P-values were corrected for family-wise error rate and t-statistic maps were thresholded at α = 0.05.