Impaired modulation of the saccadic contingent negative variation preceding antisaccades in schizophrenia

Abstract

Background: The contingent negative variation (CNV) is considered to reflect prefrontal functioning and can be observed before manual and ocular motor responses. Schizophrenic patients exhibit reduced CNV amplitudes in tasks requiring manual motor responses. A number of studies has also found normal prosaccades, but delayed antisaccades and an augmented rate of erroneous prosaccades during the antisaccade task in schizophrenia. In this study we examined the CNV during pro- and antisaccade tasks in schizophrenic patients and healthy control subjects.

Methods: Data of 17 medicated schizophrenics (ICD-10, F20) and 18 control subjects, matched with patients for age, gender, and education were analyzed. Horizontal pro- and antisaccades were elicited in four blocks, each consisting of 80 trials. Electroencephalogram was recorded from 32 channels with a DC amplifier.

Results: Patients exhibited delayed correct responses and more erroneous prosaccades during the antisaccade task than control subjects, but normal prosaccadic reaction times. In control subjects, the vertex-predominant saccadic CNV was generally larger than in patients, and larger during the anti- than during the prosaccade task. This task-related amplitude augmentation was absent in patients. Analyses of additional components suggested specificity of impaired event-related potential modulation to the saccadic CNV.

Conclusions: In accordance with the presumed prefrontal dysfunction, our results suggest deficient preparation and execution of antisaccades in schizophrenia.

Introduction

According to Fuster 1984, Fuster 1985, Fuster 1989, one of the functions of the prefrontal cortex is the mediation of cross-temporal contingencies for the temporal organization of goal-directed behavioral sequences. This function has been investigated most intensively with delay tasks. In delay tasks, the information provided by a stimulus, which is presented at the beginning of the trial, has to be retained during a delay period in order to be able to select the correct, rewarded response upon presentation of a second stimulus. Unit recordings in primates have revealed delay-related activity in the prefrontal cortex that seems to serve two complementary cognitive functions: working memory and preparatory set Funahashi et al 1993, Fuster 1984, Fuster 1989. Working memory enables the organism not only to retain behaviorally relevant information over time, but also to select an appropriate action on the basis of an internal representation; preparatory set involves the adjustment of the sensory and motor systems before an expected event in order to optimize the reception of stimuli and the anticipated (motor) response (Fuster 1989, p. 163). A rostro-caudal sequencing of motor response preparation seems to take place, because response-related units in the monkey principal sulcus (corresponding to Brodman’s area 46 in humans) increase their firing before units in the premotor and motor cortex (Fuster 1989, p. 103).

The two-stimulus paradigm is a variant of the delay task (Fuster 1989): A warning stimulus (WS), presented at the beginning of the trial, reliably signals the subsequent presentation of an imperative stimulus (IS), which is delayed by some seconds and associated with a certain task (Rockstroh et al 1989). Between WS and IS, a surface-negative potential arises referred to as the contingent negative variation (CNV; Walter 1964). For a number of reasons, the CNV is considered to reflect the delay-related activity of the prefrontal cortex; it is also seen to reflect the activity of additional cortical areas that are recruited by the prefrontal cortex as part of the behavioral structures of delay tasks Fuster 1984, Fuster 1985, Fuster 1989: 1) Between the WS and IS, a surface-negative cortical potential can be recorded at many prefrontal sites of the monkey cortex, including the dorsal and ventral banks of the principal sulcus (Sasaki and Gemba 1991); 2) subdural potentials similar to the CNV can be measured preceding the IS at different prefrontal sites in epileptic patients during presurgical evaluation Hamano et al 1997, Ikeda et al 1996; 3) the CNV amplitude is positively related to the working memory load imposed by a delayed matching-to-sample task (Klein et al 1996).

Slow surface-negative potential shifts have been recorded from the premotor, supplementary motor, motor, and somatosensory cortices of monkeys (Sasaki and Gemba 1991) and humans Hamano et al 1997, Lamarche et al 1995 before prewarned motor responses; they may reflect the aforementioned preparatory adjustment of the sensory and motor systems. These cortical potentials—with the exception of the prefrontal potentials (Rektor et al 1994)—have also been found preceding self-paced hand Hamano et al 1997, Ikeda et al 1994, Neshige et al 1988, Rektor et al 1994 or eye (Sakamoto et al 1991) movements. They are considered to be the source of the “Bereitschaftspotential” (“readiness potential,” RP; Kornhuber and Deecke 1965) that can be measured at the scalp. The RP has been suggested to be at least part of the terminal phase of the CNV (e.g., Brunia 1988; Roesler 1991). Despite this partial overlap between CNV and RP in their cortical generator structures, two arguments suggest that these components should be distinguished: 1) In neurological patients, degeneration of the basal ganglia reduces or abolishes the CNV, but preserves the RP (Ikeda et al 1997), whereas the reverse pattern is seen after decussation of the superior cerebellar peduncle (Ikeda et al 1994). The cerebellum projects to the premotor and motor cortex; the basal ganglia, however, also project to the prefrontal cortex (Kandel et al 1991); 2) the structure of the RP tasks (initiation of a movement) lacks the components of establishment of cross-temporal contingencies or sensory-motor integration that characterize delay tasks.

Saccadic eye movements are typically elicited within the two-stimulus paradigm: A central fixation point is presented as the WS. This is followed a few seconds later by the peripheral cue, which serves as the IS and the saccade goal, in the case of visually guided or “prosaccades.” Averaging time-locked to the onset of the saccade Evdokimidis et al 1992, Everling et al 1997, Klostermann et al 1994, or to the onset of the peripheral cue (e.g., Evdokimidis et al 1996, Gomez et al 1996) reveals surface-negative potential shifts with a topographical maximum at anterior central locations. Unit recordings in primates have revealed that neurons in the prefrontal frontal eye fields (FEF) and the premotor supplementary eye fields (SEF; Schall 1991, Schlag and Schlag-Rey 1987) show increased firing before visually guided saccades. In addition, during the execution of visually guided saccades as compared to a fixation control condition, in human subjects augmented blood flow in the FEF and the supplementary motor area (SMA) have been found in positron emission tomography (PET; Anderson et al 1994, Melamed and Larsen 1979, Petit et al 1993) and functional magnetic resonance imaging (fMRI; Darby et al 1996) studies.

During the “antisaccade” task Hallett 1978, Hallett and Adams 1980, subjects are instructed not to look at the cue that is presented in the visual periphery, but to look “voluntarily” in the opposite direction, that is, generate an antisaccade. The initiation of anti- as compared to prosaccades is typically delayed (e.g., Reuter-Lorenz et al 1995), and even normal subjects may happen to glance unconsciously (Mokler and Fischer 1999) at the peripheral cue in a number of trials. For some reason, this task seems to “stress” prefrontal or frontal cortical functions more than the prosaccade task: 1) Patients with large excisions of frontal lobe tissue (Guitton et al 1985), or with circumscribed lesions of the dorsolateral (Pierrot-Deseilligny et al 1991) or ventrolateral (Walker et al 1998) prefrontal cortex, but not of the FEF or the SMA (Pierrot-Deseilligny et al 1991), are impaired at inhibiting erroneous prosaccades during the antisaccade task; 2) patients with FEF lesions, however, exhibit delayed antisaccade initiation (Rivaud et al 1994). This clinical observation is complemented by physiologic results showing increased blood flow in the human FEF and SMA Doricchi et al 1997, Nakashima et al 1994, O’Driscoll et al 1995, Sweeney et al 1996, augmented neuronal firing in the monkey SEF Amador et al 1995, Schlag-Rey et al 1997, and greater slow negative potential shifts at anterior central sites in human subjects Evdokimidis et al 1996, Everling et al 1998 before anti- as compared to prosaccades; 3) inhibition of a peremptory response (looking at the peripheral cue) in favor of a “voluntary” response (looking at a position where no stimulus is present) on the basis of an instruction held in the working memory is per se a typical prefrontal function Frith et al 1991, Goldman-Rakic 1987, Roberts et al 1994, subsumed under the neuropsychological concept of “executive functions” Denckla 1996, Pennington and Ozonoff 1996.

So far, the CNV in schizophrenia has been investigated only using tasks that involve manual motor responses, with amplitude reductions being a frequently reported result (e.g., Abraham and McCallum 1976, Cohen 1989, Timsit-Berthier et al 1984, van den Bosch 1983). CNV amplitudes may normalize with remission from acute stages of the disease Knott et al 1976, McCallum and Abraham 1973 or with neuroleptic treatment Knott et al 1976, Tecce and Cole 1976. Indeed, neuroleptic treatment has been reported to increase the frontal brain metabolism in schizophrenia (e.g., Berman et al 1986, Buchsbaum et al 1987). The CNV amplitude reduction may be considered part of the frontal hypometabolism that has been supported by regional cerebral blood flow (rCBF) studies during the execution of tasks sensitive to frontal dysfunctions Andreasen et al 1992, Frith et al 1991, Lewis et al 1992, Paulman et al 1990, Weinberger et al 1986.

Impaired performance during antisaccade tasks (e.g., Crawford et al 1998, Sereno and Holzman 1995) seems to be another consequence of the frontal dysfunction in schizophrenia. Despite normal latencies of prosaccades, schizophrenic patients need significantly more time than healthy subjects to generate correct antisaccades Danckert et al 1998, Fukushima et al 1990, Karoumi et al 1998, McDowell and Clementz 1997. Schizophrenic patients also produce more erroneous prosaccades during the antisaccade task than control subjects but are apparently able to correct all or at least most of them Clementz et al 1994, Fukushima et al 1988, Fukushima et al 1990, Karoumi et al 1998, McDowell and Clementz 1997. There is some direct evidence that links the antisaccade deficit of schizophrenic patients with dysfunctions of the frontal lobes. First, the deficit is more frequently observed in schizophrenic patients with abnormal frontal computed tomography (CT) scans Fukushima et al 1988, Fukushima et al 1990. Second, during a task similar to the antisaccade task, the generation of “volitional” saccades was associated with an increase in FEF and left dorsolateral prefrontal cortex (DLPFC) metabolism in healthy participants but not in schizophrenic patients (Nakashima et al 1994). Finally, in schizophrenic patients the antisaccade task performance covaries with performance in other tasks sensitive to frontal dysfunctions, such as the smooth pursuit eye movement task Schlenker and Cohen 1995, Sereno and Holzman 1995 or the Wisconsin Card Sorting Test (WCST; Karoumi et al 1998, Nkam et al 1998, Rosse et al 1993; nonsignificant correlations with WCST were reported by Schlenker and Cohen [1995]).

The aim of our study is the investigation of the CNV preceding pro- and antisaccades in schizophrenic patients as compared to healthy control subjects. This study expects the following results: 1) A slow surface-negative potential shift, the saccadic CNV, with greater amplitudes preceding anti- as compared to prosaccades, should arise in healthy control subjects and patients; 2) schizophrenic patients should exhibit reduced CNV amplitudes during all saccade tasks, and a significantly smaller CNV amplitude modulation than control subjects when preceding antisaccades as compared to prosaccades; and 3) normal prosaccadic but augmented antisaccadic latencies along with augmented rates of erroneous prosaccades during the antisaccade task should be found in schizophrenic patients.