Introduction

Mild cognitive impairment (MCI) is a common medical condition with a prevalence in the elderly estimated between 4.9 and 25.2% [1, 2]. Patients affected by MCI have an annual risk between 5 and 15% of progression to dementia (mainly Alzheimer’s disease) [3, 4], with a cumulative risk in the longest studies estimated between 20 and 30% [5]. Our experience reported cumulative conversion rate to dementia of 18.4% (mean, 4.6% per year) [6].

According to international criteria, MCI is characterised by an under-average performance in at least one cognitive domain in such a way that everyday life is not affected [7, 8]. Executive and perceptual motor functions, complex attention, language, memory and social cognition represent the domains commonly assessed for diagnosis [7, 8]. Several pharmacological and physical interventions have been evaluated in MCI, and unfortunately, to date, no consistent reports of long-term efficacy have been demonstrated [9,10,11,12,13,14]. Transcranial direct current stimulation (tDCS) is a safe, low-cost [15], non-invasive neurophysiological technique that consists in the application of mild (1–2 mA) electrical current on the scalp [16]. Effects of tDCS vary based on current polarity, intensity, time and site of application [17, 18]. There are reports of improvement in different aspects of cognition after anodal stimulation, such as working memory, visuomotor coordination and naming in healthy subjects [19,20,21,22], recognition memory in Alzheimer’s disease [23] and working memory and mood in depression [24]. To our knowledge, there is only one report in literature that studied the neuropsychological effects of tDCS on MCI [25], as also confirmed by this recent review [26]. In this paper, we describe the neuropsychological effects of long-lasting anodal tDCS on MCI subjects.

Methods

Subject recruitment and procedure

We studied 34 patients admitted to our observation for mild cognitive impairment. Diagnosis was performed by different neurologists, according to the previously described criteria [8, 27]. Inclusion criteria were cognitive symptoms affecting one or more domains without loss of independence in everyday life. Exclusion criteria were taking medications affecting the central nervous system, brain MRI abnormalities, history of anxiety or depression. All patients had normal brain MRI and normal neurological examination performed before our observation. No patient in both groups was under medication treatment which could affect cognitive performances or was affected by mood or anxiety disorders. Patients underwent neuropsychological evaluation at baseline and then were randomly assigned to either the anodal or sham stimulation group. All patients underwent 5 days a week, up to a total of 20 days, of stimulation. A single-day session lasted 20 min in both the anodal and sham groups. At the end of the last day of stimulation, the patients underwent a second neuropsychological evaluation.

Neuropsychological evaluation

Neuropsychological evaluation consisted of Mini Mental State Examination (MMSE, range, 1.93–35.24; cut-off, < 23.8) [28, 29], Brief Mental Deterioration Battery (BMDB) with final result (FR) (cut-off, < 0) [6, 30], Rey Auditory Verbal Learning Test: immediate recall (range, 0–75; cut-off, < 28.53) and delayed recall (range, 0–15; cut-off, < 4.69) [31], Immediate Visual Memory (range, 0–22; cut-off, < 13.85) [31], Copy Design: simple (range, 0–12; cut-off, < 7.18) [31], Barrage test (time cut-off, ≥ 90; score cut-off, ≤ 9; errors cut-off, ≥ 2; result cut-off, > 2.5) [30, 32], Stroop test (time cut-off, > 27.5; errors cut-off, > 7.5) [33], verbal fluency: phonemic (range, 0–infinite; cut-off, < 17.35) [31] and semantic (range, 0–infinite; cut-off, < 25) [34], naming to description test (cut-off, < 15) [35], figure naming (cut-off, < 58.5) [36, 37], analogies (range, 0–20; cut-off, < 15.1) [30, 32], State-Trait Anxiety Inventory–Y (range, 20–80; cut-off, < 50 T points) [38] and Beck Depression Inventory (range, 0–63; cut-off, > 9) [39]. The figure naming test is composed of two different sets of 24 pictures derived from the list of Snodgrass and Vanderwart [37]. Each set was homogeneous for frequency of use and familiarity according to Laiacona et al. [36]. Test score for each set depended on response time: 3 points for correct answer within 5 s, 1 point between 5 and 10 s, 0 point for correct answer after 10 s or wrong answers.

The BMDB is based on results obtained by patients on the Rey Auditory Verbal Learning Test (RAVLT) with immediate and delayed recall, Immediate Visual Memory, visual search test (Barrage test) and verbal abstract thinking test (analogies test) and allows us to calculate a final result (FR) that expresses a measure of global cognition functioning of the subject; normality threshold is set at zero, and negative scores are considered pathological.

The neuropsychologist who administered the evaluation was blind to the stimulation received by patients. Parallel forms (same tests but with different subject matters) were used on follow-up evaluation to eliminate any learning effect.

tDCS

tDCS was administered by the HDC stimulator (Newronica s.r.l.). Two electrodes, (7 × 5) 35 cm2 and (7 × 6) 42 cm2 in size, were used as anode and cathode respectively. Electrodes were inserted into holding bags soaked with saline solution and placed over the left dorsolateral prefrontal cortex (DLPFC) and the right deltoid muscle, positions of anode and cathode respectively. The location of the latter electrode was selected to be extra-cephalic in order to avoid inhibitory effects due to cathodal stimulation. We used this site for the cathode in order to have the possibility to consider any possible subsequent measured neuropsychological result at follow-up as an effect of anodal stimulation. Conductivity gel was applied over the skin in contact with the electrode-holding bags. The location of the active electrode was selected in accordance with the EEG International 10–20 system, corresponding to F3 location.

Current was increased during the first 10 s to a maximum of 2 mA, then maintained constant for the rest of the 20-min stimulation. The subjects often complained, at the beginning of the procedure, of a slight itching sensation over the electrode sites, usually lasting only few seconds. In the sham group, the current was gradually decreased after the first 20 s in order to have patients perceive the same itching sensation.

Statistical analysis

Statistical analyses were performed using SPSS 24.0. We verified normal distribution of the variable using the Kolmogorov-Smirnov test. In the case of non-normal distribution, the variables were rank transformed.

The differences in the demographic and baseline neuropsychological data between the two groups were evaluated using the unpaired t test for continuous variables and by the χ2 test for categorical variables. The intervention effect of tDCS was evaluated by conducting a repeated measures analysis of variance (ANOVA) on the neuropsychological data, taking the condition (tDCS vs sham) as a between-subject factor and the time point (before vs after sessions) as a within-subject factor. We corrected for multiple comparisons using the false discovery rate (FDR). P values < 0.05 were considered significant.

Results

The two groups did not differ in age, sex, years of schooling and baseline neuropsychological evaluation (Table 1). Neuropsychological evaluation after 20 days of both anodal and sham stimulation groups is displayed in Tables 2 and 3. Compared with their baseline results, patients belonging to the anodal group showed at follow-up significantly better performance in Brief Mental Deterioration Battery (p < 0.0001), RAVLT: immediate recall (p < 0.001), figure naming test (p < 0.01) and Beck Depression Inventory (p < 0.01) (significant interaction between time, pre/post and intervention in the repeated measures ANOVA) (Fig. 1). Furthermore, MMSE, Immediate Visual Memory, Barrage test (time employed to perform) and Rey’s 15 Words: delayed recall showed a clear trend of improvement in the tDCS group (uncorrected p = 0.02, p = 0.02, p = 0.03 and p = 0.03 respectively), but non-surviving after FDR correction (p > 0.05).

Table 1 Demographic features of the two group studied
Full size table
Table 2 Neuropsychological evaluation (global cognitive function indices) of anodal and sham groups at baseline and follow-up
Full size table
Table 3 Neuropsychological evaluation (specific domains) of anodal and sham groups at baseline and follow-up
Full size table
Fig. 1
figure 1

Compared with their baseline results, patients belonging to the anodal group showed at follow-up significantly better performances in Brief Mental Deterioration Battery (p < 0.0001), RAVLT: immediate recall (p < 0.001), figure naming (p < 0.01) and Beck Depression Inventory (p < 0.01) (significant interaction between time, pre-post and intervention in the repeated measures ANOVA)

Full size image

Discussion

In this paper, we describe the effects on cognition of long-lasting anodal tDCS in patients affected by MCI.

Compared with the baseline and with the sham groups, after 20 days of stimulation of the anodal group, our patients showed better performance on global cognition functioning test (BMDB) and, more selectively, in neuropsychological assessments exploring immediate verbal memory and naming skills (RAVLT: immediate recall and figure naming respectively). In addition, mood evaluation through the Beck Depression Inventory disclosed lower scoring, suggesting improvement of depressive symptoms. Although several studies described tDCS effects on cognition, only one study [25] reported tDCS neuropsychological effects on MCI [26]. Meinzer and colleagues described improved semantic word-retrieval scoring after 5 days of 1-mA anodal stimulation over the left ventral inferior frontal junction (IFJ). A concomitant fMRI study of brain metabolism showed the reduction of bilateral prefrontal cortex (including the left IFJ) and right middle temporal gyrus metabolic activity. This observation was considered the result of a facilitatory effect in the stimulated neuronal network which causes a decrease requirement of energy [25]. However, this explanation partially contrasts with the results of a following work on MCI patients disclosing increased glucose metabolism in dorsolateral, ventrolateral, medial prefrontal and dorsal anterior cingulate, anterior and posterior insular regions, and hippocampal and parahippocampal regions on cerebral PET after 3 weeks of three times/week, 30 min, 2-mA anodal tDCS stimulation over the left DLPFC [40]. Compared with the study of Meinzer, our study differs in intensity of the current applied (1 mA vs 2 mA), site of stimulation (left DLPFC vs left IFJ) and days of stimulation (20 vs 5). Left DLPFC is the most common area targeted for transmagnetic stimulation (TMS) and tDCS stimulation studies concerning cognitive enhancement in the elderly [41]. Previous works describe improvement in episodic memory retrieval after anodal tDCS over the left DLPFC in healthy elderly adults [42, 43].

Furthermore, working memory and dual tasking skills, two functions which both decline early during ageing, showed also improvement after anodal TDCS over DLPFC in healthy-aged persons [41, 44].

In addition, the left DLPFC has been used as target for TMS study which showed improvement in Stroop test [45]. Some of our findings are consistent with the clinical-anatomical expectations based on the site of stimulation; on the other hand, some are less clear. Executive function networks which are thought to rely anatomically mainly on left prefrontal cortex are usually studied assessing skills in the subdomains of inhibition, set shifting, working memory and fluency [46, 47].

Unexpectedly, we did not observe improvement in Stroop test which evaluate the former subdomain neither in phonemic verbal fluency tests. We did not perform trial making test and digit span which selectively assess set shifting and working memory respectively. Despite analogies test depends upon proper function of pre-frontal areas and it allows to assess inhibition, working memory and selective attention subdomains [48], we did not found better scores in patients after anodal stimulation. Improvement in Beck Depression Inventory is consistent with previous studies which described, along with a definite positive effect of TMS on mood disorders, a possible therapeutic role of anodal stimulation of the left DLPFC in these conditions [49]. Whether the above effects are directly mediated by left DLPFC, or indirectly through its connection with the frontal insular, anterior cingulate and ventromedial pre-frontal cortex is not known. However, although baseline measurements in Beck Depression Test were not significantly different in anodal and sham groups, the latter showed a lower baseline score. Thus, it cannot be defined whether the above effects are mediated by left DLPFC, directly or indirectly through its connections, or they represent regression to mean in the anodal group. It is also to be defined the reason why selective stimulation of the left DLPFC area results in a better performance on general cognition, as evidenced by the Brief Mental Deterioration Battery score or why anodal stimulation of the left DLPFC improves cognitive performance thought to be mainly dependent upon other cortical areas, such as verbal episodic memory assessed through Rey’s 15 Words: immediate recall test, a function thought to rely mainly upon left mid temporal neuronal activity [50]. Moreover, the reason underlying selective improvement of general cognition index such as BMDB but not MMSE is also unclear. It could be speculated that using MMSE is less sensible to identify cognitive alterations that depended upon pre-frontal lobe functions; an improvement in BMDB could reflect better results in the subtest of analogies of BMDB. However, the latter view contrasts with the fact that in our patients, improvement of BMDB relates better with RAVLT results rather than with analogies test. On the other hand, a contribution of working memory in RAVLT results cannot be excluded. Nevertheless, as already expressed above, the anodal group did not show better scores in tests which selectively explore other subdomains of executive functions such as Stroop test or analogies test.

Besides, we cannot exclude a contribution of arousal reaction to our results; however, this interpretation would contrast with data that showed increased pain threshold in healthy individuals after anodal tDCS applied over the left DLPFC [51]. The neurological impact of tDCS stimulation has been theorised at different levels. It has been described that 13 min of anodal stimulation of the motor cortex is sufficient to induce threshold excitability modification lasting up to 1 h [16, 17, 52]. TMS-coupled studies showed tDCS during stimulation effects depends only on membrane polarity changes whilst after effects are consequences of glutamatergic- and GABAergic-mediated synaptic plasticity mechanisms [53]. Long-lasting effects of both anodal and cathodal stimulation are blocked by the use of NMDA antagonists suggesting a reliance mediated by long-term potentiation (LTP)-dependent mechanisms [52, 54].

An in vitro study showed that anodal tDCS increases AMPA receptor translocation in the hippocampus and a promoting effect of tDCS on S831 phosphorylation of the receptor. Interestingly, both phenomena relate with synaptic strength and LTP induction [55]. It has been reported that working memory can be impaired by temporary TMS-mediated inhibition of the left prefrontal cortex [56] and a subsequent work demonstrated improvement in working memory performances after anodal tDCS application over the left DLPFC [22]. We can speculate that using LTP important for learning [57] interventions facilitating its physiological activity could improve memory fruition.

Nevertheless, to date, some results concerning tDCS effectiveness in specific neurological disorders such as aphasia are still conflicting [58, 59] and additional studies are needed in order to better clarify biological aspects and clinical purposes of tDCS.

Conclusion

We found that long-lasting anodal tDCS over the left DLPFC stimulation improves overall cognition scores (BMDB), immediate verbal memory (RAVLT: immediate recall) and figure naming performance in patients affected by MCI. In addition, tDCS seems to improve a standardised measure of mood in the same population. Notwithstanding the foregoing, the following limitations of the study demand caution in interpretation: lack of long-term follow-up, lack of information about in vivo biomarkers of degenerative conditions (such as CSF tau, P-tau and beta amyloid) or functional imaging (FDG-PET). Furthermore, the biological mechanisms behind our observed results and the site of stimulation need to be clarified. Future studies are needed to confirm the promising therapeutical opportunities which tDCS seems to have in MCI. In this perspective, overcoming the aforementioned limitations will help to standardise more precisely the effects of the technique in the condition and to understand the exact mechanisms underlying improvement and its duration over time. tDCS, due to its low-cost and relative risk-free profile, could be a useful tool to improve cognitive symptoms in MCI. Considering the high prevalence of the condition, confirmation of effectiveness of tDCS application in MCI would have significant social and economic benefits.