The Impact of Caffeine on Cognition in Schizophrenia

1.1 Background. Caffeine is the most used psychoactive drug in Canada, with regular consumption by 88% of the adult population and is unique because it is a "cradle-to-grave drug". In Canada, coffee is the most commonly consumed source of caffeine, while tea, chocolate and energy drinks are also significantly used. This significant rate of consumption is likely due in part to caffeine's actions on the human brain, resulting in increased arousal, elevated mood and cognitive enhancements. The effects of caffeine, at least at levels typically seen in human consumption, are primarily related to its actions at adenosine receptors. Caffeine's antagonistic effects on adenosine can be seen after even a single cup of coffee, with increased consumption resulting in increased blockade. Adenosine receptors are distributed in all brain cells, contributing to the caffeine's ability to affect a wide range of brain regions.

While levels of caffeine consumption are considered to be high in the general population, there is some evidence that they may be even higher within schizophrenia (SZ) patients. In a recent U.S. study of 146 community-dwelling individuals with SZ, daily consumption rates of caffeine were nearly double those observed in a healthy control population (471.6 mg/day vs. 254.2 mg/day; p < .001). Furthermore, 13% of the SZ population studied ingested more than 1000 mg/day of caffeine, with the highest observed intake at 2647.2 mg/day. To place this in context, Health Canada recommends adults consume no more than 400 mg of caffeine per day, and doses above 250 mg have the potential to induce symptoms of anxiety, agitation, irritability and insomnia . High doses of caffeine are particularly concerning for individuals with SZ as caffeine alters dopaminergic activity at post-synaptic neurons through its actions at adenosine A2A receptors; this enhances dopamine function, resulting in increased locomotor activity and appears to exacerbate positive symptoms, such as delusions and hallucinations . Another effect of caffeine with particular relevance to SZ involves its interaction with atypical antipsychotic drugs, such as clozapine. Both clozapine and caffeine are metabolized in the liver by the cytochrome P450 system and, in particular, the CYP1A2 isoenzyme. Competition between clozapine and caffeine has been reported to elevate antipsychotic blood levels, increasing the risk of clinically significant side effects at high dosage levels of caffeine. As such, given the high consumption levels and potential negative side effects of caffeine in SZ, it is important to understand the cortical mechanisms that underlie caffeine use.

As caffeine is mainly used because of perceived stimulant properties, including enhanced mental alertness and increased energy, research has been conducted to examine the effects of caffeine on cognitive performance in humans. These studies have identified beneficial effects of caffeine on a wide-range of cognitive processes including verbal working memory, sustained attention, and executive function. It must be noted, however, that while many studies report pro-cognitive effects of caffeine, others report no significant improvements or deficits in performance. This may be at least partially explained by evidence suggesting an 'inverted U-shaped' dose-response curve for caffeine, wherein lower doses have positive effects on performance, while high doses (i.e., above 500 mg) cause a decrease in performance.

Within the fields of attention and information processing, the electroencephalographically (EEG)-derived event-related potentials (ERPs) provide an exquisitely sensitive method of indexing cognition that can both complement and clarify behavioural observations. The ERP waveform is elicited in response to a specific stimulus, such as tones or light flashes, or cognitive events, such as recognition, decision making or response to specific stimuli events. Specifically, ERPs represent an average of the neural activity that follows the onset of a stimulus. In studies examining the effects of caffeine on the brain, EEG/ERPs appear to be preferable to neuroimaging techniques such as fMRI and PET as the vasoconstrictive nature of caffeine may confound these latter techniques, which model cortical blood flow. EEG/ERPs avoid these potential confounds and, furthermore, provide a temporal resolution far superior to some of the more sophisticated imaging techniques (i.e. PET, fMRI), making this methodology far more suitable for capturing instantaneous changes in information processing. While the literature of caffeine's effects on ERP-indexed cognitive processes is limited, there is evidence of caffeine enhancing ERPs. In one such study, not only was there quicker behavioural responding during a visual search task with caffeine (vs. placebo), but also quicker ERP measures of target detection, although caffeine did not alter ERP amplitudes.

While a hallmark characteristic of SZ is neurocognitive impairment, these deficits notably overlap with the cognitive domains that are improved by caffeine. Among the core deficts observed within SZ is a dysfunction of attentional systems, which can be observed in patients who are both acutely psychotic and in remission. Deficits of attention in SZ include both alteration of sustained attention, particularly as measured by continuous performance tasks and visual selective attention (Luck \& Gold, 2008), as measured by visual search tasks. It has further been suggested that within SZ these deficits of attention contribute to dysfunction of the central executive and working memory, the limited capacity system for storing and manipulating information for short time periods (up to ~30 s), in order to carry out complex cognitive operations such as planning, reasoning, and problem solving. Impairment in working memory processes, such as those indexed by the n-back task, are a robust feature of SZ, are stable over time, and independent of psychotic symptoms.

While there is a literature, albeit small, examining the effects of caffeine on cognition and brain activity in healthy individuals, such research reports are all but non-existent in SZ (based on extensive searches with both PsycINFO and PubMed). This gap in the literature is particularly notable given the rates of caffeine consumption observed within SZ. The sole studies to address how caffeine might impact cognitive processes in SZ have done so within animal models employing MK-801 (a potent NMDA receptor antagonist) infused mice. Within this behavioural model of SZ, chronic treatment with caffeine improved measures of long term memory, while reducing MK-801-induced perseverative errors in one study, while another study found no effects of chronic caffeine on cognition in mice given a significantly higher dose of MK-801.

1.2 Objective. This project proposes to examine how caffeine alters ERP-indexed aspects of cognition in SZ patients. Neural (ERPs) and behavioural (hits, reaction time) correlates of cognitive performance will be probed within a randomized, double-blind, placebo-controlled design with a battery of well-established cognitive paradigms that have both demonstrated deficits in SZ and are associated with caffeine-induced enhancement: a visual selective attention paradigm (visual search task), a working memory paradigm (N-back), and a sustained attention paradigm (AX-CPT). The primary outcome measures will be ERP amplitudes and latencies. Secondary indices will include behavioural measures of performance, including accuracy (% correct), reaction times and measures of signal detection (i.e. d'). While the investigators expect caffeine to improve behavioural and neural indices in both groups, the investigators expect this effect to be stronger in SZ patients due to expected lower baselines.

1. Materials and Methods. 2.1 Study Design and Procedure. This study will employ a randomized, placebo-controlled, double-blind, repeated measures design consisting of two experimental sessions separated by at least 24 hrs, with participants receiving caffeine one day and placebo on the other day. The order of caffeine administration will be counterbalanced, such that half of the participants are randomly administered caffeine in the first and placebo in the second session, while the remaining half are administered caffeine and placebo in the reverse order. Prior to the experimental sessions, participants will complete informed consent, having been given an opportunity to read the study description and have any questions answered, complete demographic questionnaires, and experimental sessions will be scheduled. Participants will attend the BIOTIC Neuroimaging Research Lab at the QEII Health Sciences Centre (Halifax, NS), for a morning (i.e. testing beginning between 9-11 am) testing session. Participants will be required to abstain from illicit drugs, over-the-counter medications, alcohol and cigarettes beginning at midnight of the previous day. SZ patients will not be asked to halt or pause any current medication regimen. Additionally, participants will be asked to abstain from caffeine (including coffee, tea and cola) for at least 6 hours prior to the test session to ensure adequate clearance of circulating caffeine (half-life = 2.5-4.5 hrs). Upon arrival at the laboratory, following self-report of adherence to pre-testing abstinence, drug treatment will be administered, EEG electrodes will be applied and, 30 minutes after drug administration, volunteers will be assessed with the test battery, presented in a randomized order to avoid order effects. All testing procedures will be carried out in accordance with the Declaration of Helsinki and following the approval of the relevant research ethics boards.

2.2 Caffeine. As in previous studies investigating the effects of caffeine on electrophysiological measures, in each session participants will be asked to swallow (with water) one of two identical pill capsules containing either caffeine (200 mg) or placebo. This method and dosage were selected as oral administration of caffeine results in efficient absorption, while the intermediate size dose, which is typical of most studies of caffeine on cognition, has been shown to exert widespread, strong cerebral effects. Furthermore, administration of caffeine in pill form allows for greater control over dosage and facilitates double-blind procedures, while controlling for potentially confounding sensory effects (e.g. smell, taste of coffee).

2.3 Test Battery. 2.3.1 Visual Search Task. The visual search task will follow methodology described by Lorenzo-Lopez and colleagues and previously used in psychopharmacology research. In short, participants will be instructed to perform a visual search, which consists of detecting a target stimulus (vertical bar) among an array of distractors (horizontal bars) by indicating whether the target was present or absent in each search array by pressing a green button on a response pad with one hand and a red button with the other hand, respectively. Behavioral measures will include the number of correct responses, and the reaction times (RTs) of correct responses. Neural (ERP) measures of interest include the N2pc, an index of visual spatial shifts of attention to the location of the target or distractor, and the P3b, an index of target detection. The amplitudes of these ERPs indicate the amount of resources allocated to the associated cognitive process, while latency represents processing speed. Hypotheses: Caffeine will significantly speed up behavioural indices of target detection (i.e. reduced RTs) and associated neural waveforms (i.e. reduced latency of N2pc \& P3b) in both groups, with an increase in amplitude only observed in SZ patients.

2.3.2 Visual Sustained Attention (AX-CPT). Sustained attention will be assessed within the A-X version of the continuous performance task (AX-CPT). Participants will be presented with a series of letters and will be instructed to respond to the letter "X" only when it was immediately preceded by "A." In order to test sustained attention, 400 letters will be presented over a span of 11 minutes consisting of 80 cues ("A"), 40 targets ("A" followed by "X"), 40 NoGos ("A" followed by any other letter), and 240 distractors (other letters or "X" not preceded by "A"). Behavioural endpoints include the number and RTs of correct responses, and false alarms (i.e., nontarget responses). Neural (ERP) measures of interest include the P3b to correct targets. Hypotheses: Consistent with previous reports in sustained attention, caffeine will not impact behavioural performance, but will increase P300 amplitudes, with this effect (vs. placebo) being seen in both participant groups.

2.3.3 Visual Working Memory. This paradigm, which has previously been used in psychopharmacology research, will employ four randomized conditions of the verbal N-Back task; each condition will have identical stimuli and response demands, but consist of increasing levels of working-memory load. A series of letters will be presented and participants will be required to respond as quickly as possible only when the letter on the screen matches the letter n stimuli back (i.e., for the 0-back condition, the target is any letter that matches a pre-specified letter (x), while in the 1-back, 2-back and 3-back conditions, a target is any letter that is identical to the letter presented one, two or three trials back, respectively). As the N-back task requires information storage, updating, and manipulation, it has become the dominant tool used in assessing WM control functions. Behavioural endpoint measures will include accuracy (% correct target detections) and reaction time (ms) to targets. Additionally, for each condition and each drug treatment I will calculate signal detection sensitivity (d') and response bias (C). Electrophysiological endpoint measures include amplitudes and latencies of the P3b to targets. Hypotheses: Similar to the other paradigms, and consistent with behavioural studies of caffeine's effects on n-back indexed working memory, the investigators expect caffeine will reduce RTs and will reduce P300 latency in both groups.

2.4 EEG Recording and ERP Computation. ERPs will be extracted from EEG activity recorded from an electrode cap with Ag+/Ag+-Cl- active electrodes at sixty-four scalp sites according to the 10-10 system of electrode placement, including: three midline sites (frontal [Fz], central [Cz], parietal [Pz]); three left hemisphere (frontal [F3], central [C3], parietal [P3]) and three right hemisphere (frontal [F4], central [C4], parietal [P4]) scalp sites; and bilateral mastoid activity. Electrodes will be also placed on the mid-forehead and nose to serve as ground and reference, respectively. Bipolar recordings of horizontal (HEOG) and vertical (VEOG) electro-oculogram activity will be taken from supra-/sub-orbital and external canthi sites, respectively. All electrode impedances will be kept below 5kohms. Electrical activity will be recorded with an amplifier bandpass of 0.1 and 30 Hz, digitized at 500 Hz, and stored on hard-disk for later off-line analysis. Stimuli (and resulting triggers for ERP analysis) will be generated by Presentation software (Neurobehavioural Systems, Berkeley CA) 2.5 Questionnaires. In order to assess clinical variables in SZ patients, as well as caffeine consumption, 2 questionnaires will be administered: 1) Psychotic Symptom Rating Scale (PSYRATS). The PSYRATS is designed to quantify auditory hallucinations and delusions, both of which have been associated with consumption of caffeine; 2) Caffeine Consumption Questionnaire (CCQ). Collects detailed information on caffeine consumption, including sources of caffeine (e.g., coffee, cola, etc…) and time period of caffeine consumption. As both of these questionnaires probe trait (vs. state) measures, they will only be administered once.

2.6 Data Analysis. Data will be subjected to separate ANOVA/ANCOVA procedures (SPSS, IBM Corp., Armonk NY). For each paradigm, endpoint measures will be analyzed by mixed ANOVAs, with between-group (2 levels: patients, controls) and within-group/repeated measures factors including drug (caffeine, placebo). Analysis of EEG/ERP endpoints will also include scalp site as a within-group factor. Daily caffeine consumption (as measured by the CCQ) will be used as a covariate. Follow-up of significant (Greenhouse-Geisser corrected) main or interaction effects (p < .05) will be carried out with Bonferroni-adjusted planned comparisons using separate (vs. pooled) error estimates. In order to examine the correlation between behavioural/electrophysiological endpoints and measures of caffeine consumption, two-tailed Spearman's rho correlations will be conducted between consumption measures and ERP amplitudes/latencies under placebo and drug conditions, as well as between consumption measures and measures of drug-associated change in ERP endpoints overall and within both groups.