Methylphenidate-mediated reduction in prefrontal hemodynamic responses to working memory task: a functional near-infrared spectroscopy study

INTRODUCTION

Methylphenidate (MP) is a piperidine-derived central nervous system stimulant that acts as a dopamine/norepinephrine agonist by binding to the monamine transporters in the presynaptic membrane and blocking reuptake (Kuczenski and Segal, 1997). It is used in the treatment of attention deficit hyperactive disorder (ADHD), narcolepsy (Swanson et al., 1991; Hirai and Nishino, 2011; Takon, 2011), depression, and co-morbid depression in the medically ill (Challman and Lipsky, 2000; Rozans et al., 2002). The therapeutic effects of MP are largely attributed to its effects on attention, working memory, and motivation. MP also improves attention and cognitive performance in healthy adults (Challman and Lipsky, 2000).

One of the major challenges in the use of MP is the inter-subject variability of its cognitive-enhancing effects (Volkow et al., 2002; Volkow et al., 2008). Although the mechanisms underlying this variation are not fully understood, evidence from positron emission tomography (PET) studies suggests that the variability in therapeutic effects of MP could be in part due to individual differences in the rate of dopamine release (dopamine tone) and regional brain activity (Volkow et al., 1997; Volkow et al., 2002). Although brain imaging technologies could be used to monitor and possibly predict treatment responsiveness to MP, technologies used in previous studies of MP such as PET and functional magnetic resonance imaging (fMRI) (Mehta et al., 2000; Schweitzer et al., 2004; Volkow et al., 2008; Schlosser et al., 2009; Tomasi et al., 2011) have limited value in clinical practice—they are expensive and not easily portable, and PET requires radioactive ligands.

The functional near-infrared spectroscopy (fNIRS) is an emerging non-invasive, less expensive, and easily portable technique to measure changes in cerebral hemoglobin concentrations, providing an indirect measure of neural activity in brain areas (Suto et al., 2004; Weber et al., 2007; Ehlis et al., 2008; Richter et al., 2009; Zhang et al., 2009; Pu et al., 2011; Schecklmann et al., 2011). fNIRS is an optical technique that uses infrared light with a wave length from 700 to 1300 nm. When infrared light passes through tissue, the light is absorbed by few biological chromophores including hemoglobin (Hb) and myoglobin. The oxy-Hb and deoxy-Hb have different absorption spectra. By measuring the transmitted light, we can estimate oxy-Hb and deoxy-Hb concentrations (Hoshi, 2003). Because the light passes through scalp, skull, cerebrospinal fluid, and brain, the total optical length (i.e., the length over which light travels through multilayered tissue) and optical attenuation due to scattering effects need to be taken into account in the calculation of absolute value of Hb concentrations. Continuous wave type instruments that are readily available commercially cannot measure total optical path length, limiting its use for absolute quantification of Hb changes. Neural activation is accompanied by increases in regional cerebral blood flow, regional cerebral oxygen metabolic rate, and cerebral blood volume resulting in increases in oxy-Hb and total-Hb and decreases in deoxy-Hb. However, deoxy-Hb and total-Hb do not necessarily show expected changes, and thus an increase in oxy-Hb is considered to be a sensitive marker of neural activity (Hoshi, 2003).

Recent literature suggests that fNIRS can be used as a substitute for fMRI for studying brain activity of cognitive tasks (Cui et al., 2011). However, the previously published fNIRS studies examining cognitive effects of MP in ADHD children were limited by open trial design without placebo comparisons (Weber et al., 2007; Schecklmann et al., 2011). Furthermore, there is no fNIRS study to date that investigates the effects of MP on working memory task-dependent hemodynamic changes in prefrontal regions in healthy adults.

The present pilot study aims to use fNIRS to evaluate the effects of MP on hemodynamic responses to a working memory task in adult healthy volunteers. Considering previous evidence for cognitive task-related reductions in regional cerebral blood flow and glucose metabolism in the prefrontal cortex after MP intake (Mehta et al., 2000; Schweitzer et al., 2004; Volkow et al., 2008), we predict that relative to placebo, MP treatment will decrease working memory task-dependent changes in oxy-Hb in lateral prefrontal regions as measured by fNIRS and improve working memory performance.

METHODS

RESULTS

Functional near-infrared spectroscopy findings

There were no significant differences between MP and placebo for baseline oxy-Hb (right: t = 0.04, df = 12, p = 0.97; left: t = −0.64, df = 12, p = 0.54), deoxy-Hb (right: t = −1.99, df = 12, p = 0.098; left: t = −1.20, df = 12, p = 0.26), or total-Hb (right: t = 0.135, df = 12, p = 0.90; left: t = 0.457, df = 12, p = 0.66) concentrations. Baseline-corrected oxy-Hb concentrations were significantly decreased during the 2-back task in the right frontal region after MP treatment compared with the placebo (t = −2.35, df = 12, p = 0.036; Table 2), whereas deoxy-Hb and total-Hb concentrations were not significantly different. Among the 13 participants, 10 showed reduction in the baseline-corrected oxy-Hb responses between MP and placebo conditions (Figure 1). The effect size of 2-back task-dependent oxy-Hb reductions in the right frontal region in the MP condition was medium (Cohen's d = 0.65). We used Tukey's test for familywise error correction for pairwise comparison of oxy-Hb differences in the right frontal region. The Tukey's test q was −6.82 with significance at p = 0.01. Because there was a large but non-significant decrease in oxy-Hb concentration in the left side (t = −1.78, df = 12, p = 0.13; Table 2), we examined the 2-back task-dependent oxy-Hb differences between MP and placebo conditions in the bifrontal region. The reduction in oxy-Hb was near significant in the bilateral prefrontal region (t = 2.04, df = 12, p = 0.06), suggesting that the effect of MP on neural activity during a 2-back task may be localized in the bifrontal regions but predominantly in the right side. The group differences between the subjects who received MP on the first day (N = 7) and the subjects who received placebo on the first day (N = 6) in 2-back task-dependent oxy-Hb reductions between MP and placebo conditions were not significant in the right frontal (t = −0.45, df = 11, p = 0.66), left frontal (t = −0.62, df = 11, p = 0.55), and bilateral frontal (t = −1.04, df = 11, p = 032) regions, suggesting that the impact of cross-over effects on oxy-Hb differences would be minimal.

Table 2. Hemoglobin (Hb) concentration changes after the administration of methylphenidate (MP) or placebo

		Right prefrontal cortex: baseline-corrected Hb concentration (µmol/L)			Left prefrontal cortex: baseline-corrected Hb concentration (µmol/L)
	Task	MP (SD)	Placebo (SD)	p	MP (SD)	Placebo (SD)	p
Oxy-Hb	0-back	−17.12 (24.06)	−7.05 (43.93)	0.064	−13.55 (23.82)	0.059 (38.00)	0.22
	2-back	−18.41 (28.20)	−0.51 (27.23)	0.036*	−15.85 (29.19)	0.45 (33.95)	0.13
Deoxy-Hb	0-back	−1.10 (17.03)	−5.70 (16.86)	0.40	−4.11 (13.21)	−7.53 (9.77)	0.44
	2-back	−1.47 (18.30)	−1.75 (12.05)	0.94	−4.85 (16.18)	−4.55 (12.29)	0.95
Total-Hb	0-back	−0.013 (0.020)	−0.02 (0.046)	0.56	−0.025 (0.033)	−0.028 (0.043)	0.82
	2-back	−0.013 (0.021)	−0.01 (0.022)	0.58	−0.025 (0.029)	−0.019 (0.023)	0.47

• * Significant difference at p < 0.05.

For the 0-back task, differences in baseline-corrected oxy-Hb measurements between MP and placebo was near significant (p = 0.06) in the right frontal region, suggesting N-back task with minimal working memory load also produced similar effect. There were no significant correlations between the baseline-corrected mean oxy-Hb concentration in the right frontal, left frontal, or bilateral frontal areas and performance measures (correct responses, missed responses, and reaction time) during the 2-back task after MP treatment (Table 3). The significant correlation between systolic blood pressure and correct responses during the MP condition (Table 3) suggests underlying sympathetic arousal induced by MP.

Table 3. Correlational analysis of performance measures, oxy-Hb changes, and cardiovascular measures

Performance and cardiovascular measures	Cortical Oxy-Hb changes
	Right prefrontal		Left prefrontal		Bilateral
	r	p	r	p	r	p
Correct response	0.063	0.84	−0.07	0.82	0.01	0.98
Missed response	−0.12	0.71	−0.059	0.85	0.17	0.63
Reaction time	−0.015	0.96	−0.12	0.69	−0.08	0.80
Systolic blood pressure	−0.42	0.20	−0.53	0.09	−0.51	0.11
Diastolic blood pressure	0.38	0.25	0.39	0.23	0.40	0.22
Performance measures	Cardiovascular measures
	Systolic blood pressure		Diastolic blood pressure
	r	p	r	p
Correct response	−0.64	0.032*	−0.085	0.80
Missed response	0.40	0.23	0.27	0.42
Reaction time	−0.44	0.17	0.29	0.39

• r, Pearson's product moment correlation coefficient.
• * Significant difference at p < 0.05.

DISCUSSION

This fNIRS study showed that a working memory task (2-back) performed after administration of a single dose of MP (20 mg) was associated with decreased oxy-Hb concentration preferentially in the right lateral prefrontal cortex as compared with placebo. This MP-induced decrease in oxy-Hb was accompanied by increased correct responses and decreased missed responses in a 2-back working memory task. Given that oxy-Hb is the most sensitive indicator of hemodynamic changes related to neural activity (Hoshi, 2003), the observed MP-related task-dependent decrease in oxy-Hb reflects a decrease in neuronal activity in the right lateral prefrontal region. This reduction is unlikely to be due to the known vasoconstrictor effect of MP on cerebral blood flow (Krimer et al., 1998) for several reasons: the effect was predominantly lateralized to the right side, there were no differences in baseline Hb concentrations between the MP and placebo conditions, and a previous study showed that MP-induced brain changes were not due to its effect on cerebral blood flow (Rao et al., 2000).

Our results are consistent with previous PET studies (Mehta et al., 2000; Schweitzer et al., 2004; Volkow et al., 2008) supporting the hypothesis that MP improves neuronal efficiency or signal–noise ratio, a well-documented effect of dopamine ((Foote et al., 1975). Contrary to our findings, fMRI studies reported working memory task-related increases in frontal activity following the administration of MP (Ramasubbu and Goodyear, 2008; Tomasi et al., 2011). The discrepancies between fMRI and NIRS could be in part due to differences in methodologies and wide variation in correlations between oxy-Hb and blood oxygen level-dependent responses. In the combined fMRI and fNIRS study, oxy-Hb was correlated better with blood oxygen level-dependent (BOLD) signals than deoxy-Hb in some cases, whereas in other cases, the deoxy-Hb was better correlated with BOLD signals than oxy-Hb during the cognitive tasks (Cui et al., 2011). Furthermore, given the large size of the dorsolateral prefrontal cortex (DLPFC), the differential activation pattern observed in fNIRS and fMRI studies could be due to differences in the regions that were examined. During the performance of cognitive tasks, areas of activation and deactivation in the DLPFC have been reported (Pfurtscheller et al., 2010). Simultaneous recording of multichannel fNIRS and fMRI would be helpful to examine the correlations between Hb changes and BOLD signals related to the combined effect of MP and N-back tasks on brain function.

Although the differential effect of MP on reductions in oxy-Hb during the 2-back task reached statistical significance only in the right frontal area, there was a near significant decrease in the bifrontal regions, suggesting that the effect was not unequivocally unilateral. A systematic review on this subject provides evidence that all N-back tasks (verbal/non-verbal and identity/location monitoring) broadly activate bilateral prefrontal and parietal regions, and differences in content of working memory tasks may be associated with differential activation of the cerebral hemispheres (Owen et al., 2005). Verbal identity monitoring (N-back task using numbers or letters) was associated with increased activation in the left DLPFC, a region known to be important for subvocal rehearsal. Non-verbal location monitoring (spatial working memory tasks) was associated with increased activation in the right DLPFC, a region important for spatial attention. However, contrary to expectations, Mehta et al. (2000) showed that the interaction effect of MP and spatial working memory tasks was lateralized to the left frontal region, even though the right hemisphere is preferentially involved in spatial N-back tasks. Similarly, our results showed preferential involvement of the right hemisphere during number N-back tasks in the MP condition, although verbal working memory processing is predominantly associated with activity in the left hemisphere. It is possible that the MP-induced enhancement in performance or efficiency is mediated by MP's effects on other cognitive processes through the contralateral hemisphere for content-specific working memory processing (Mehta et al., 2000).

Our study has several limitations including small sample size, large inter-subject variability of Hb responses, measurements limited to the dorsolateral prefrontal regions because of the use of two-channel optical system even though the dorsal cingulate and lateral parietal lobes are known to be involved in working memory processing (Owen et al., 2005), a fixed optical path length of 24 cm even though this may vary across the individuals (Wahr et al., 1996), and measurement of Hb responses as relative values. Furthermore, because this is a pilot study involving a small sample size, we performed hypothesis-driven planned comparisons without commonly used omnibus F-tests and Bonferroni corrections. Hence, our results should be considered as preliminary.

Despite these limitations, this is the first study to our knowledge that replicates the findings of PET studies showing MP-induced reduction in prefrontal metabolism during cognitive task performance using fNIRS imaging. These findings suggest that fNIRS can be developed as a bedside technology for monitoring and predicting treatment responsiveness to medications. fNIRS technology has the potential to predict clinical response to MP treatment, which will be crucial for personalized medicine in ADHD. As major psychiatric disorders are associated with abnormalities in the DLPFC and the DLPFC is essential for top-down regulation of emotions, thoughts, and actions (Arnsten, 2011; Gamo and Arnsten, 2011), fNIRS has wider clinical implications in the assessment of frontal lobe dysfunctions in several psychiatric conditions and during pharmacotherapy. However, there are several technical and practical problems with fNIRS technology including quantification of Hb changes and analysis of fNIRS data that need to be resolved prior to its application in clinical practice (Hoshi, 2003).

CONFLICT OF INTEREST

No conflict of interest declared.