Cingulate activity and fronto-temporal connectivity in people with prodromal signs of psychosis

Abstract

Schizophrenia is associated with fronto-temporal dysconnectivity, but it is not clear whether this is a risk factor for the disorder or is a consequence of the established illness. The aim of the present study was to use fMRI to investigate fronto-temporal connectivity in subjects with prodromal signs of schizophrenia using the Hayling Sentence Completion Task (HSCT). Thirty participants, 15 with an at risk mental state (ARMS) and 15 healthy controls were scanned whilst completing 80 sentence stems. The congruency and constraint of sentences varied across trials. Dynamic causal modelling (DCM) and Bayesian model selection (BMS) were used to compare alternative models of connectivity in a task related network. During the HSCT ARMS subjects did not differ from Healthy Controls in terms of fronto-temporal activation, i.e. there was neither a main effect of group nor a group-by-task interaction. However, there was both a significant main effect of group and a significant interaction in the anterior cingulate cortex (ACC), with greater ACC activity in the ARMS subjects. A systematic BMS procedure among 14 alternative DCMs including the ACC, middle frontal, and middle temporal gyri revealed intact task-dependent modulation of fronto-temporal effective connectivity in the ARMS group. However, ARMS subjects showed increased endogenous connection strength between the ACC and the middle temporal gyrus relative to healthy controls. Although task related fronto-temporal integration in the ARMS was intact, this may depend on increased engagement of the ACC which was not observed in healthy control subjects.

Fig. 1

Proportion of errors (SEM) for response Initiation and Suppression, High and Low CP by Group.

Fig. 2

Statistical parametric maps during (a) response Initiation > Word Repetition in controls, (b) Suppression > Word Repetition in controls, (c) Suppression > Initiation in controls, (d) Initiation > Word Repetition in ARMS, (e) Suppression > Word Repetition in ARMS, and (f) Suppression > Initiation in ARMS . All activations are reported at a whole-brain corrected cluster threshold of p < 0.05 (with a standard voxel-level cutoff of p < 0.001). Coordinates for activated regions in ARMS are presented in Table 2. Coordinates of activations in healthy control subjects are reported in Allen et al (2008).

Fig. 3

Statistical parametric maps showing (a) the main effect of group across both response Initiation and Suppression and (b) simple main effect of group during response Suppression in the ACC. All activations are reported at a whole-brain corrected cluster threshold of p < 0.05 (with a standard voxel-level cutoff of p < 0.001). Coordinates for activated regions are presented in Table 3.

Fig. 4

(a) Competing three-area DCMs of effective connectivity constructed with bidirectional connections to/from ACC, LMTG, LMFG. Models 1 to 4 specify different locations for modulatory inputs (b); an additional 10 models were constructed by combinations of these models (Model 5 = 1 + 2, Model 6 = 1 + 3, Model 7 = 1 + 4, Model 8 = 2 + 3, Model 9 = 2 + 4, Model 10 = 3 + 4, Model 11 = 1 + 2 + 3, Model 12 = 1 + 2 + 4, Model 13 = 1 + 3 + 4 and Model 14 = 2 + 3 + 4). (b) Exceedance probabilities for three-area Models 1–14 in controls. (c) Exceedance probabilities for three-area Models 1–14 in ARMS subjects. X-axis = model number, Y-axis = exceedance probability.