Test–retest reliability of the BOLD pharmacological MRI response to ketamine in healthy volunteers

Abstract

The pharmacological MRI (phMRI) technique is being increasingly used in both pre-clinical and clinical models to investigate pharmacological effects on task-free brain function. Ketamine, an N-methyl-d-aspartate receptor (NMDAR) antagonist, induces a strong phMRI response and represents a promising pharmacological model to investigate the role of glutamatergic abnormalities in psychiatric symptomatology. The aim of this study was to assess whether the brain response to ketamine is reliable in order to validate ketamine phMRI as a mechanistic marker of glutamatergic dysfunction and to determine its utility in repeated measures designs to detect the modulatory effect of other drugs. Thus we assessed the test–retest reliability of the brain response to ketamine in healthy volunteers and identified an optimal modelling approach with reliability as our selection criterion.

PhMRI data were collected from 10 healthy male participants, at rest, on two separate occasions. Subanaesthetic doses of I.V. ketamine infusion (target plasma levels 50 ng/mL and 75 ng/mL) were administered in both sessions. Test–retest reliability of the ketamine phMRI response was assessed voxel-wise and on pre-defined ROIs for a range of temporal design matrices including different combinations of nuisance regressors designed to model shape variance, linear drift and head motion. Effect sizes are also reported.

All models showed a significant and widespread response to low-dose ketamine in predicted cerebral networks and as expected, increasing the number of model parameters improved model fit. Reliability of the predefined ROIs differed between the different models assessed. Using reliability as the selection criterion, a model capturing subject motion and linear drift performed the best across two sessions. The anatomical distribution of effects for all models was consistent with results of previous imaging studies in humans with BOLD signal increases in regions including midline cingulate and supracingulate cortex, thalamus, insula, anterior temporal lobe and ventrolateral prefrontal structures, and BOLD signal decreases in the subgenual cingulate cortex.

This study represents the first investigation of the test–retest reliability of the BOLD phMRI response to acute ketamine challenge. All models tested were effective at describing the ketamine response although the design matrix associated with the highest reliability may represent a robust and well-characterised ketamine phMRI assay more suitable for repeated-measures designs. This ketamine assay is applicable as a model of neurotransmitter dysfunction suitable as a pharmacodynamic imaging tool to test and validate modulatory interventions, as a model of NMDA hypofunction in psychiatric disorders, and may be adapted to understand potential antidepressant and analgesic effects of NMDAR antagonists.

Introduction

The effects of uncompetitive antagonists at the N-methyl-d-aspartate (NMDA) receptor, such as ketamine and phencyclidine (PCP), have led to a growing appreciation of the potential role of glutamatergic abnormalities in psychiatric symptomatology (Javitt, 2004, Luby et al., 1959). Numerous studies have shown that ketamine induces behavioural and cognitive perturbations in both experimental animals and healthy human volunteers (Cilia et al., 2007, Gilmour et al., 2011, Honey et al., 2004, Krystal et al., 1994). At sufficiently high doses (1500–2000 ng/ml plasma levels in humans; (Langsjo et al., 2005)) ketamine acts as a surgical anaesthetic, whereas at sub-anaesthetic doses (ca. 100–200 ng/ml) it induces symptoms including thought disorder, altered integration of sensory information, and disruption in working memory performance; all characteristics likened to those observed in patients with schizophrenia (Adler et al., 1999, Honey et al., 2003, Morgan et al., 2011). At still lower doses (ca. 50 ng/ml), the psychotomimetic symptoms are seen to a lesser extent (Krystal et al., 1994). Given the clear limitations of using one compound to ‘model’ a disorder, in this study ketamine is utilised as a pharmacological probe of one neurochemical component, the glutamatergic system, which is relevant for psychiatric conditions where dysfunction of this system may be evident.

Multiple imaging techniques have been shown to be sensitive to pharmacologically-induced changes in brain physiology following acute NMDA antagonist administration in both rodents and humans (Corlett et al., 2006, Duncan et al., 2000, Gozzi et al., 2008a, Vollenweider et al., 1997a). These provide neuroimaging correlates of the behavioural and subjective deficits induced by the compound. Moreover, a number of studies have demonstrated that the central haemodynamic or metabolic changes evoked by acute NMDA antagonist administration can be reversed by pharmacological pre-treatment, in particular by atypical antipsychotics and group-II metabotropic glutamate receptor (mGluR-II) activators (Chin et al., 2011, Dedeurwaerdere et al., 2011, Gozzi et al., 2008b, Hackler et al., 2010). These findings parallel work in experimental animals demonstrating that mGluR-II agonists block NMDA receptor antagonist evoked glutamate efflux (Lorrain et al., 2003, Moghaddam and Adams, 1998). Thus, with increasing interest in the development of therapeutics designed to ameliorate glutamatergic dysfunction, the neuroimaging signal evoked by compounds such as ketamine provides a potential translational biomarker, providing a physiologically-relevant pharmacodynamic signal to confirm “mechanistic engagement” of novel therapeutic compounds and inform dose selection for subsequent clinical trials (Schwarz and Tauscher, 2011). However, to enable the use of this imaging biomarker in crossover design studies – for example, to test the modulatory effects of other compounds – the reliability across sessions of the ketamine-evoked BOLD response and the effects of different signal modelling approaches need to be characterised.

Early studies demonstrated that sub-anaesthetic ketamine administration increased the rate of glucose utilisation, as measured by ¹⁴C-2-deoxyglucose (2-DG) uptake, in frontal, cingulate and hippocampal regions of the rat brain (Duncan et al., 1998a, Duncan et al., 1998b). Consistent with these findings, [¹⁸F] fluorodeoxyglucose (FDG) positron emission tomography (PET) and H₂¹⁵O PET studies have reported increases in cerebral glucose metabolism and cerebral blood perfusion respectively, in similar frontal, cingulate and insula areas in healthy volunteers (Holcomb et al., 2001, Langsjo et al., 2003, Langsjo et al., 2004, Vollenweider et al., 1997b). Breier and colleagues found increased FDG uptake in the frontal cortices alone and found that these were correlated with ketamine induced changes in conceptual disorganisation (Breier et al., 1997). Research into drug related changes in brain responses has increasingly turned to magnetic resonance imaging (MRI) methods, motivated by the absence of ionising radiation and the superior temporal resolution. In particular, pharmacological MRI (phMRI) approaches enable the direct effect of an administered compound on “resting” brain function to be assessed via the haemodynamic response in the absence of an experimental task (Breiter et al., 1997, Leslie and James, 2000).

PhMRI experiments commonly use the blood oxygenation level-dependent (BOLD) contrast to detect signal changes in a T₂*-weighted MRI time series covering both pre- (baseline) and post-compound administration conditions in a continuous acquisition (Shah and Marsden, 2004, Steward et al., 2005). Animal phMRI experiments with ketamine have shown increases in BOLD signal in frontal, cortical, hippocampal and thalamic regions (Littlewood et al., 2006), consistent with the previously described 2-DG and PET work. In analogous experiments using phMRI methods sensitive to changes in cerebral blood volume, haemodynamic changes evoked by PCP administration (Gozzi et al., 2008c) closely matched ketamine and MK-801 evoked changes in 2DG uptake (Duncan et al., 1999). In healthy humans, ketamine infusion induced BOLD phMRI signal increases in regions including the mid-posterior cingulate, thalamus and anterior temporal cortical areas, along with a decreased BOLD signal observed in the ventromedial prefrontal cortex (Deakin et al., 2008). Manifestations of dissociative and psychotic subjective effects, assessed by the Clinician Administered Dissociative States Scale (CADSS) and the Brief Psychiatric Rating Scale (BPRS) respectively were found to be correlated with changes in BOLD activation in the same study, supporting the connection between NMDA receptor blockade and psychotomimetic symptoms.

The temporal phMRI response to acute pharmacological challenge in the brain comprises a relatively low-frequency signal change through time. A range of temporal analysis approaches can be applied to model these signal changes. Data-driven methods such as clustering methods (Schwarz et al., 2007, Whitcher et al., 2005) or independent component analysis (ICA) can allow structure within the data to be explored. Non-linear regression methods can be used to provide a parametric model fit to the phMRI response in each voxel or brain area of interest; however, this approach can be sensitive to initial parameter estimates, may converge to local minima or fail to converge for signals with low signal-to-noise (SNR). Time courses of independently measured behavioural or biological (e.g., subjective scores or neurotransmitter concentration) parameters can be used as a temporal correlate for the central phMRI signal changes (Breiter et al., 1997, Littlewood et al., 2006, Stein et al., 1998); however, this approach assumes that the functional imaging changes closely parallel the reference time course and that the latter is available at a sufficient temporal resolution and SNR to enable such an analysis.

Alternatively, the phMRI response can be quantified within a general linear model (GLM) framework using a regressor (signal model) chosen pragmatically based on independent data with the same pharmacological compound. This approach yields a single numeric estimate of the amplitude of response to the ketamine challenge at each voxel, analogous to standard general linear model (GLM) functional MRI (fMRI) analyses of specified contrasts. This approach has several advantages: it enables more straightforward group analyses, the calculation of simple region of interest (ROI)-level summary measures and facilitates the combination of the phMRI data with behavioural, pharmacokinetic and other non-imaging variables. However, several factors can influence the measured BOLD phMRI time course and may lead to deviations from the hypothesised model: (1) phMRI often requires extended scan times compared to standard fMRI paradigms, and is thus susceptible to artefacts such as those generated by head motion and scanner drift; (2) the pharmacological agent administered can cause physiological and/or subjective effects that may also lead to additional head motion; (3) in phMRI data there is often a degree of variation in the shape (temporal profile) of the response between subjects and brain regions (Schwarz et al., 2007). The temporal analysis of ketamine BOLD phMRI time series data within a GLM framework thus relies on the choice of an appropriate design matrix to model the phMRI response of interest and also account for drift, head motion and potentially for plausible variations in temporal profile. Moreover, the utility of phMRI in understanding the effects of ketamine, and its relationship to subjective, pharmacokinetic or genotype profiles, depends on the within-subject reliability, between-subject reproducibility and dose sensitivity of the computed BOLD response amplitude. Assessing the reliability of the ketamine response across sessions is thus essential for (1) its application in repeated measures designs and (2) the degree of confidence with which it is used to understand the modulatory influences of other compounds.

In the present study we systematically examined the BOLD phMRI response to ketamine infusion in healthy human participants using an open-label, repeated measures study design. Test–retest reliability of the estimated BOLD response amplitudes from a range of temporal design matrices comprising different combinations of nuisance regressors designed to model shape variance, linear drift and head motion was assessed. Each model was fitted using a standard GLM approach and group-level effect sizes of the resulting parameter estimates were determined. These assessments were performed in order to guide the choice of model to be used in subsequent interventional studies. Due to the ubiquitous nature of glutamatergic projections in the human brain (Cotman et al., 1987, Storm-Mathisen et al., 1983), a relatively diffuse change in BOLD signal was expected, although stronger localised changes were expected in areas such as the cingulate, thalamus and frontal-temporal cortices (Deakin et al., 2008, Holcomb et al., 2001, Langsjo et al., 2003, Littlewood et al., 2006). The BOLD phMRI response to ketamine was then characterised more extensively using the best performing design matrix in terms of reliability, providing a sound quantitative basis for the prospective use of this assay in trials of novel pharmacological agents.