عدم واکنش پذیری اینسولا به محرک های بد در اسکیزوفرنی

Abstract Patients with schizophrenia may have altered pain perception, as suggested by clinical reports of pain insensitivity, and recent neuroimaging findings. Here, we examined neural responses to an aversive electrical stimulus and the immediate anticipation of such a stimulus using fMRI and a classical conditioning paradigm, which involved pairing an electrical shock with a neutral photograph. Fifteen men with schizophrenia and 13 healthy men, matched for demographic characteristics, electrical stimulation level and scan movement, were studied. The shock induced robust responses in midbrain, thalamus, cingulate gyrus, insula and somatosensory cortex in both groups. However, compared to controls, the schizophrenic patients displayed significantly lower activation of the middle insula (pFWE = 0.002, T = 5.72, cluster size = 24 voxels). Moreover, the lack of insula reactivity in the schizophrenia group was predicted by the magnitude of positive symptoms (r = − 0.46, p = 0.04). In contrast, there were no significant differences between the two groups in the magnitude of neural responses during anticipation of the shock. These findings provide support for the existence of a basic deficit in interoceptive perception in schizophrenia, which could play a role in the generation and/or maintenance of psychotic states.

1. Introduction It has long been observed that some patients with schizophrenia are relatively insensitive to pain. Kraepelin reported that dementia praecox patients could burn themselves with cigarettes and experience needle pricks or injuries without showing adaptive withdrawal reactions ( Kraepelin and Robertson, 1919 and Bonnot et al., 2009). More recently, a meta-analysis of experimental pain studies indicated that schizophrenic patients show a blunted response to experimental pain ( Potvin and Marchand, 2008), a finding confirmed in a detailed review of cases, and clinical and experimental studies ( Bonnot et al., 2009). Pain insensitivity in schizophrenia is associated with increased morbidity and mortality, but the underlying pathophysiology is poorly understood ( Singh et al., 2006). Although antipsychotic medications may have an analgesic effect ( Seidel et al., 2010), alterations in pain perception in schizophrenia cannot be solely explained by medication effects ( Potvin and Marchand, 2008). Recent neuroimaging studies in schizophrenia found greater somatosensory activation, but diminished insula, posterior cingulate cortex and brainstem responses to thermal pain ( de la Fuente-Sandoval et al., 2010 and de la Fuente-Sandoval et al., 2012). These reports provide initial evidence that painful stimuli are processed differently in schizophrenia. But pain is a highly subjective experience, and emotional, anticipatory and/or sensory aspects of noxious processing may drive alterations in sensitivity. Considering the communicative and social impairments associated with schizophrenia, a more detailed dissection of aversive experiences in the disorder is warranted. Here we examined fMRI responses evoked by an aversive electrical shock stimulus in schizophrenic patients and healthy controls. Using a Pavlovian fear-conditioning protocol (Milad et al., 2007 and Holt et al., 2009) we examined neural and autonomic responses to conditioned (CS+) and unconditioned stimulus (US, an electrical shock) presentations in a partial reinforcement paradigm. The US was delivered at a 62.5% reinforcement rate, allowing us to compare neural responses to the US with responses to the immediate anticipation of the US (the moment just prior to the offset of unreinforced CS+ trials). In this way, the sensory component of a US response may be isolated from its expectancy related components (Linnman et al., 2011a, Linnman et al., 2011b and Dunsmoor and Labar, 2012). Responses to an electrical (Linnman et al., 2011a) or auditory (Dunsmoor et al., 2007, Dunsmoor et al., 2008, Knight et al., 2010 and Dunsmoor and Labar, 2012) US in healthy subjects are accompanied by increased activity in the brainstem and thalamus, as well as in the cingulate, sensory and insular cortices, structures also known to respond to noxious stimuli (Apkarian et al., 2005). In the current investigation, we hypothesized, based on previous evidence (de la Fuente-Sandoval et al., 2010 and de la Fuente-Sandoval et al., 2012), that schizophrenic patients would display impaired responses of the insula and brainstem and elevated responses of somatosensory cortex, compared to controls. We further sought to disentangle the sensory and anticipatory aspects of this response, and relate any observed alterations to the symptoms of schizophrenia.

3. Results 3.1. Skin conductance responses None of the skin conductance measures of the fear conditioning procedure differed between patients and controls. 3.2. Brain responses to the US The US elicited responses in midbrain, thalamus, cingulate gyrus, insula and somatosensory regions in both the controls and schizophrenic patients (see Fig. 2a and b, and Table 2a and b). The whole brain between-group comparison revealed that the schizophrenic patients exhibited lower BOLD reactivity to the US than the controls in a cluster encompassing the left mid insula (54% of voxels) and the left precentral gyrus (46% of voxels), with the peak at MNIxyz − 44, − 2, and 10 (pFWE = 0.002, T = 5.72, cluster size = 24 voxels) (see Fig. 3). There were no regions showing greater activation in the patients compared to the controls. In the contrast [controls (US > CS+end)] > [patients (US > CS+end)], i.e. controlling for anticipatory components of the shock response, the insula showed a similar trend towards hypoactivation in the schizophrenia group, but did not reach a whole brain-corrected level of significance (puncorrected = 0.004, T = 2.62 at MNI (− 44, − 2, − 10)). Shock (US versus CS− offset) responses in (a) healthy controls (n=13) and (b) ... Fig. 2. Shock (US versus CS− offset) responses in (a) healthy controls (n = 13) and (b) schizophrenic patients (n = 15). The contrast maps are displayed at a threshold of T > 3 and overlaid on a template structural MRI image. The color bar denotes T-values. Figure options Table 2. BOLD responses in healthy controls and in schizophrenic patients. Contrast Cluster p(FWE-cor) Cluster size (voxels) Peak T MNI Peak region x y z a. BOLD responses in healthy controls US–CS −end < 0.001 2985 10.51 − 50 − 28 16 L. postcentral gyrus 9.91 − 58 − 18 22 L. postcentral gyrus 8.05 − 40 − 34 20 L. insula < 0.001 1706 7.96 60 − 32 18 R. postcentral gyrus 7.28 54 − 28 30 R. inferior parietal lobule 7.12 56 − 16 16 R. postcentral gyrus < 0.001 863 7.06 − 2 16 32 L. cingulate gyrus 6.72 − 6 4 40 L. cingulate gyrus 6.35 4 6 40 R. cingulate gyrus < 0.001 420 6.65 − 52 − 22 54 L. postcentral gyrus 6.16 − 42 − 22 48 L. postcentral gyrus 5.98 − 44 − 14 50 L. precentral gyrus < 0.001 85 6.48 − 14 8 2 L. putamen 5.51 − 14 0 8 L. putamen < 0.001 51 6.24 10 − 14 0 R. thalamus 5.08 8 − 22 − 6 R. red nucleus < 0.001 69 5.88 36 0 − 4 R. claustrum 5.21 36 − 12 − 4 R. claustrum 0.007 11 5.55 − 6 − 22 − 6 L. red nucleus 0.005 13 5.33 − 40 − 14 62 L. precentral gyrus CS+end–CS −end < 0.001 305 7.96 − 50 − 34 30 L. inferior parietal lobule 6.22 − 58 − 18 22 L. postcentral gyrus 5.47 − 60 − 24 28 L. inferior parietal lobule < 0.001 217 7.16 36 18 8 R. insula 6.45 40 26 2 R. inferior frontal gyrus < 0.001 119 7.12 54 − 28 32 R. inferior parietal lobule < 0.001 58 5.82 − 32 20 8 L. insula < 0.001 45 5.63 0 20 32 L. cingulate gyrus 0.005 13 5.37 − 4 6 40 L. cingulate gyrus b. BOLD responses in schizophrenic patients US–CS−end < 0.001 1088 9.84 − 46 − 20 20 L. insula 7.89 − 50 − 30 24 L. inferior parietal lobule 6.39 − 62 − 22 20 L. postcentral gyrus < 0.001 678 7.67 − 10 12 36 L. cingulate gyrus 6.97 − 6 2 42 L. cingulate gyrus 6.58 − 2 − 8 46 L. cingulate gyrus < 0.001 418 7.33 − 52 − 24 60 L. postcentral gyrus 5.49 − 38 − 18 68 L. precentral gyrus < 0.001 844 6.92 50 − 28 30 R. postcentral gyrus 6.06 64 − 34 20 R. superior temporal Gyrus 6.05 38 − 20 18 R. insula < 0.001 240 6.82 − 58 2 0 L. superior temporal gyrus 6.25 − 52 − 2 8 L. precentral gyrus < 0.001 77 6.63 − 34 6 − 6 L. claustrum < 0.001 111 6.2 − 2 − 64 8 L. posterior cingulate 5.3 − 12 − 56 0 L. occipital lobe, lingual gyrus < 0.001 60 6.03 38 0 8 R. claustrum < 0.001 51 5.97 18 0 − 10 R. lateral globus pallidus < 0.001 39 5.91 − 46 − 68 6 L. middle temporal gyrus 5.28 − 48 − 58 10 L. middle temporal gyrus 0.001 28 5.83 58 6 6 R. precentral gyrus < 0.001 53 5.78 − 20 − 68 10 L. posterior cingulate 5.57 − 16 − 70 18 L. precuneus 0.001 33 5.71 36 − 18 4 R. claustrum CS+end–CS−end < 0.001 89 6.15 − 62 4 − 4 L. superior temporal gyrus 5.75 − 52 − 2 6 L. superior temporal gyrus Table options Greater shock related activation of the left middle insula was found in controls ... Fig. 3. Greater shock related activation of the left middle insula was found in controls as compared to schizophrenic patients. The insula cluster was significant at p = 0.002, corrected for multiple comparisons across the whole brain. The contrast map is displayed at a threshold of T > 3 and overlaid on a template structural MRI image. The color bar denotes T-values. Figure options 3.3. Brain responses to the CS+end At the time point immediately prior to the offset of the CS+ (CS+end derived from non-reinforced trials), healthy subjects displayed activation in the bilateral inferior parietal lobe, bilateral insula, the left postcentral gyrus and the left cingulate gyrus (Table 2a). Schizophrenic subjects appeared to display less activation, with a significant response in the left superior temporal gyrus only (Table 2b). However, there were no significant differences between the two groups, consistent with the observation that the responses to CS+end were qualitatively similar between control and schizophrenic subjects at a lower threshold. 3.4. Post-hoc analyses Within the schizophrenia cohort, we observed no significant differences in BOLD responses between medicated and non-medicated patients, and no significant correlation between antipsychotic dose (in chlorpromazine equivalents) and responses of the insula to the US (parameter estimates at peak between-group difference: r = − 0.04, p = 0.44), as well as across the whole brain. We observed a significant negative correlation within the patient group between responses of the insula to the US and positive symptom levels (r = − 0.46, p = 0.04) (see Fig. 4). There were no correlations between insula response to the US and levels of negative (r = 0.32, p = 0.24) or general symptoms (r = − 0.04, p = 0.44), trait anxiety (r = − 0.16, p = 0.28), state anxiety (r = 0.1, p = 0.36) or depressive symptoms (r = − 0.26, p = 0.17). A scatter plot displaying the negative correlation (r=−0.46, p=0.04) between ... Fig. 4. A scatter plot displaying the negative correlation (r = − 0.46, p = 0.04) between individual BOLD signal (mean beta weight across all voxels in cluster showing between-group difference, at MNIxyz − 44, − 2, 10) in the insula and scores on the PANSS Positive Symptom Subscale in the schizophrenia cohort is shown. PANSS, Positive and Negative Syndrome Scale.