Anodal Transcranial Direct Current Stimulation Temporarily Reverses Age-Associated Cognitive Decline and Functional Brain Activity Changes

Study rationale.

The two fMRI paradigms (task related vs resting state) were included to address the impact of atDCS on two different phenomena that have both been associated with healthy aging. Task-related fMRI specifically addressed potential atDCS effects on (reduced) word-retrieval performance and previously described hyperactivity in bilateral prefrontal during language tasks in aging (for review, see Crosson et al., 2013). The RS-fMRI paradigm was included to provide additional evidence for atDCS effects at the functional network level. Please note that the stimulated left IFG is not only involved in language processing but also part of brain networks supporting different cognitive functions (e.g., working memory, attentional processing). Moreover, numerous studies provided evidence for age-associated changes in RS connectivity that have been linked to impairment of the latter functions as well (Goh, 2011; Ferreira and Busatto, 2013). Given that prefrontal atDCS resulted in modulations in a number of different functional networks in younger adults (Keeser et al., 2011; Meinzer et al., 2012b), the RS analysis aimed to explore beneficial effects of atDCS on network configuration on a much broader scale compared with the highly specific approach used for the task-related analysis (see below for details of our respective data analysis strategies).

However, despite the fact that the two fMRI paradigms addressed different phenomena, we aimed to address three common issues. First, we wanted to determine differences in performance and task-related activity (during semantic word generation) and resting-state connectivity between younger and older adults (both age groups in their “normal” state, i.e., during sham). Second, we wanted to assess the impact of atDCS on performance, task-related activity, and RS connectivity in the older group (i.e., the within-group comparison of atDCS vs sham). Third, we wanted to assess whether atDCS in the older group would result in a more youth-like pattern of behavioral performance, task-related activity, and RS connectivity (i.e., by directly comparing data of the younger control group acquired during sham with data of the older adults acquired during atDCS).

Participants.

Twenty healthy older adults were recruited for this study (all native German speakers; 10 females and 10 males; age mean ± SD, 68.0 ± 5.7 years; range, 60–76 years). Each older subject completed a standard health questionnaire to exclude any previous or current neurological or psychiatric condition and the neuropsychological test battery of the Consortium to Establish a Registry for Alzheimer's Disease (CERAD-Plus; www.memoryclinic.ch), an established test to differentiate normal aging from dementia and its precursors. It comprises the Mini-Mental State Examination (MMSE; Folstein et al., 1975) and subtests assessing verbal fluency, naming, constructional praxis, and verbal memory. None of the older participants reported subjective memory complaints in everyday life or had a MMSE score below 27 (mean ± SD, 28.9 ± 0.9). All participants scored within 1.5 SD of the mean for the CERAD normative sample in all subtests, indicative of normal cognitive functioning. Ten older participants did not take any medication. Of the remaining 10 individuals, five substituted thyroxine, five took blood pressure-lowering medication, and three took statins.

As a control group, we used the data of 20 younger participants (age mean ± SD, 26.4 ± 3.4 years; range, 19–31 years; none reported use of psychoactive medication or recreational drugs) acquired during two previous studies of our work group that assessed the impact of atDCS versus sham on language (n = 16; Meinzer et al., 2012b) or language and motor functions (n = 4; Lindenberg et al., 2013) using crossover designs. Younger subjects were matched to the older subjects for sex (10 females, 10 males) and education (years of education mean ± SD: young, 15.6 ± 1.9; old, 15.9 ± 1.2) and were scanned at the same scanner using the identical fMRI setup. Only scans acquired during sham tDCS were used. To account for possible effects of repeated scanning, 10 of the younger subjects received sham tDCS during the first scanning session and the remaining 10 during the second session. All participants were strongly right hand according to the Edinburgh inventory (Oldfield, 1971) and participated for the first time in a tDCS study.

For all younger and older subjects, T1- and diffusion-weighted MRI images were acquired and inspected by a board-certified neuroradiologist for any clinically relevant pathology (none found). In particular, there was no evidence of brain infarcts, including lacunar lesions, which are frequently associated with cognitive impairment in old age (Galluzzi et al., 2008). In older adults, an additional FLAIR (fluid-attenuated inversion recovery sequence) was acquired and rated for presence of white matter hyperintensities (WHM) according to the revised Fazekas Scale (Pantoni et al., 2010; List et al., 2011). No WMH was evident in eight participants (Fazekas score 0), and 12 older participants showed mild WMH (Fazekas score 1).

tDCS.

The stimulation protocol was identical to that of a previous study (Meinzer et al., 2012b). A constant direct current (1 mA) was administered by an MRI-compatible stimulator (DC-Stimulator Plus; NeuroConn) during RS-fMRI and task-related fMRI. The stimulating electrode was inserted in a saline-soaked synthetic sponge (5 × 7 cm²) and centered over the left ventral IFG (vIFG) using the 10–20 EEG system (Meinzer et al., 2012b). Electrode positions were verified on the T1-weighted images in every subject. A 10 × 10 cm² reference electrode was positioned over the right supraorbital region. The large size (10 × 10 cm²) of the reference electrode renders the stimulation over the contralateral orbitofrontal cortex ineffective (Nitsche et al., 2007), and there is abundant behavioral evidence suggesting the specificity of this setup (for a recent review, see Nitsche and Paulus, 2011).

The current was turned on approximately 1 min before the resting-state scan and increased in a ramp-like fashion over 10 s, eliciting a tingling sensation on the scalp that fades over seconds. The current was turned off after 30 s (sham) or continued for a total of 20 min in all subjects during atDCS (i.e., active stimulation outlasted the task by ∼20–120 s). The study was conducted in a double-blind fashion (i.e., the subject and the researcher who conducted the MRI scans, interacted with the subjects and conducted the behavioral assessments, were blind to the stimulation condition). A separate investigator administered the stimulation. Two self-report rating scales were administered immediately before and after the two scanning sessions to assess mood and positive and negative effect [Visual Analog Mood Scales (VAMS) (Folstein and Luria, 1973); Positive and Negative Affect Scales (PANAS) (Watson et al., 1988)].

fMRI setup and acquisition parameters.

Scanning was conducted using a 3 tesla Siemens Trio MR System at the Berlin Center for Advanced Neuroimaging. Initially, a continuous RS-fMRI sequence was acquired (3 × 3 × 4 mm³; TR/TA, 2300 ms; TE, 30; flip angle, 90°; 34 transverse slices; no gap; interleaved acquisition; FOV, 192 × 192; acquisition matrix, 64 × 64; 150 functional volumes). Subjects were instructed to keep their eyes closed, relax, think of nothing particular, and move as little as possible. A T2*-sensitive echoplanar imaging BOLD sequence was acquired to assess task-related functional activity during overt semantic word retrieval. Details of the task and stimulus selection have been described previously in detail (Meinzer et al., 2012b). In short, the paradigm used a temporal sparse-sampling design, in which the overt verbal response was assessed during a scanner off phase, and the hemodynamic response was acquired after a short time delay to avoid movement artifacts during articulation (3 × 3 × 3 mm²; TR, 6000; TA, 2000; TE, 30; flip angle, 90°; 32 transverse slices; gap, 0.75 mm; slice thickness, 3 mm; interleaved acquisition; FOV, 192 × 192; acquisition matrix, 64 × 64; 104 functional whole-brain images). During each scanning session, participants were visually presented with six semantic categories (10 consecutive trials, 3.8 s per trial; i.e., total of 60 word-generation trials) and asked to overtly generate different exemplars for each category. After each trial, the stimulus disappeared and was replaced by a black screen (2.2 s), and a whole-brain functional volume was acquired. Task blocks alternated with a baseline condition (five consecutive trials of saying the word “rest”). Two different sets of categories (matched for linguistic criteria) were used during the two fMRI sessions and counterbalanced across subjects (for details, see Meinzer et al., 2012a). Verbal responses were recorded and subsequently transcribed. Responses were scored by two independent raters blinded to the stimulation condition. In case of disagreement, a consensus was reached. Incorrect responses (exemplars that did not belong to a given category), omissions, and repetitions (e.g., same response or synonyms) were scored as errors. Consecutive answers that started with the same letter were not scored as errors (except if they were repetitions or synonyms).

Task-related fMRI data analysis.

Statistical Parametric Mapping (SPM5; Wellcome Department of Imaging Neuroscience, London, UK) was used for task-related fMRI analysis. Preprocessing comprised functional image realignment, coregistration with the individual participants' anatomical images, unified segmentation and registration to MNI standard space, and spatial smoothing with an 8 × 8 × 8 mm Gaussian kernel. The design matrix for the statistical analysis comprised the covariates of interest (semantic word-generation and baseline trials) and movement parameters. Only correct trials were included in the analysis; incorrect trials were included as additional covariates of no interest in the statistical model. Before model estimation, a high-pass filter (128 s) was applied and data were modeled with a finite impulse response (Meinzer et al., 2012b). Planned contrasts of interest were estimated on the individual subject level for each session (semantic word-generation vs baseline trials) in both age groups. A random-effects whole-brain comparison of correct semantic word-generation trials with the baseline condition that comprised data of younger and older adults during sham (i.e., the main effect of task; n = 40) was calculated to assure that (1) the task elicited a similar pattern of activity as in our previous studies (Meinzer et al., 2009, 2012c) and (2) brain areas chosen for a subsequent a priori region-of-interest (ROI) analysis were located in areas active during the task.

This ROI analysis focused on bilateral prefrontal brain regions where activity differences between younger and older subjects are most pronounced (Park and Reuter-Lorenz, 2009). Left vIFG was selected because it has been implicated specifically with semantic retrieval processes (Thompson-Schill et al., 1997) and showed decreased activity in our previous study in younger adults during atDCS compared with sham (Meinzer et al., 2012b). Left dorsal IFG (dIFG), implicated with more general selection or phonological retrieval processes, was chosen to assess the specificity of the stimulation [i.e., in an area that was located in the vicinity of the active electrode but was not expected to show task-related activity changes because of the stimulation (Meinzer et al., 2012b)]. Right vIFG and middle frontal gyri (MFG) assessed potential remote effects of the stimulation on activity in right frontal areas in the older adults. Both areas have frequently been shown to exhibit increased activity in older compared with younger subjects during language (Meinzer et al., 2009, 2012c; Goh, 2011) and nonlanguage (Park and Reuter-Lorenz, 2009) tasks possibly caused by enhanced demands placed on executive control processes in aging (Spreng et al., 2010).

For the ROI analysis, 6 mm spherical ROIs were created centered around Talairach coordinates −48/23/2 (left vIFG; BA 45/47) and −42/30/23 (left dIFG; BA 46/9) and the contralateral homolog of the left vIFG (Meinzer et al., 2012b). The right MFG (Talairach coordinate 33/42/17) was selected based on a previous study that used the same task (contrast, semantic fluency > rest) in an independent sample of older native German speakers (Meinzer et al., 2009). To account for individual variability in anatomy and functional activity, mean beta activity was extracted from spherical ROIs (6 mm) centered around the individual participants' peak voxel from the contrast “word generation > baseline” in the above-described ROIs for each session in younger (sham) and older subjects (atDCS; sham). All ROIs were located within areas activated by younger and older subjects during the word-generation task (i.e., the main effect of task). We conducted repeated-measures ANOVAs that assessed activity differences in the four a priori ROIs between younger and older adults and comprised the within-subjects factor REGION and the between-subjects factor AGE GROUP. This analysis was conducted separately for the comparison of the younger subjects' data (acquired during sham) with that of the older subjects acquired during sham and atDCS. An additional analysis assessed the impact of atDCS compared with sham in the older group and comprised the within-factors REGION and STIMULATION. We report Greenhouse-Geisser corrected results of significant effects of interest. Post hoc analyses comprised paired or unpaired t tests as appropriate. Statistical analyses were performed using SPSS (version 21; IBM). Associations between activity (during sham tDCS) or activity changes (sham vs anodal tDCS) in the four ROIs in the older group with performance or tDCS-induced performance improvements, respectively, were explored using Pearson correlation coefficients.

To gain additional information about the distribution of stimulation effects in the older adults, we also conducted an exploratory whole-brain analysis that directly compared data acquired during atDCS and sham using a paired t test as implemented in SPM. We report results of significant voxels within significant clusters (voxel level, p < 0.001 uncorrected; cluster level, p < 0.05 family-wise error corrected).