Alzheimer’s disease (AD) is characterized by a progressive deterioration of cognitive functions, impacting the autonomy of patients in performing their activities of daily living (ADLs; Buschert et al., 2010). Pharmacological approaches only provide moderate symptom control, provisionally improving cognition or slowing down cognitive decline (Massoud and Gauthier, 2010). In this context, the earlier the diagnosis of probable AD is performed, the better the patient prognosis will be as cognitive intervention can be applied in the early stages of the disease (Fernández-Calvo et al., 2015; Kawashima et al., 2015). For example, memory, executive functions and attention may be relatively improved when adequate support is provided (García-Alberca, 2015). However, intervention specificity, best candidate selection for different types of protocols, and whether benefits can transfer to untrained cognitive domains are still controversial topics (Alves et al., 2013; Fernández-Calvo et al., 2015).

Thus, multimodal intervention makes it possible to prevent (Olanrewaju et al., 2015) or delay the progression caused by the disease (Olazarán et al., 2010). Currently, non-invasive neurostimulation techniques such as transcranial direct current stimulation (tDCS) have been used to delay the cognitive deficits and to decrease the functional impairment of patients with dementia (Andrade et al., 2016; André et al., 2016; Gonsalvez et al., 2017). tDCS is a low-cost, portable and safe tool that induces plasticity via electrical stimulation using a low-intensity current and a continuous flow through the brain. This pattern of electricity modulates cortical activity without directly acting on neurons, which decreases the risk of serious adverse effects (Boggio et al., 2011).

The neurobiological basis of AD recovery after neurostimulation is the ability of tDCS to induce long-term synaptic changes. The effects perceived during stimulation are apparently associated with changes in the subliminal membrane resting levels that modulate the input-output curve. Conversely, prolonged effects would act on GABAergic activity, the glutamatergic system, long-term potentiation (LTP) and long-term depression (LTD) mechanisms related to plasticity (Yu et al., 2015). Studies in mice with a similar disease to AD show tDCS effects on several of these factors, also including axonal activation and alteration in the cellular transmembrane potential, and consequently neurotransmitter balance (Ruohonen and Karhu, 2012).

In a critical review of the use of tDCS combined with cognitive rehabilitation, Cappon et al. (2016) concluded that many studies failed to assess whether improvement in a specific task transfers to ADLs in AD treatment. The authors indicate it is imperative to discriminate between increased performance on a specific cognitive task and recovery in more general ADLs demanding that cognitive function. This suggests the importance of including other cognitive domains in both the evaluation and the treatment protocol, such as memory and executive function. Consistent results reported by systematic reviews and meta-analyses have shown that overall cognitive stimulation, a cognitive intervention based on a range of different cognitive activities (Clare and Woods, 2004), promotes better cognitive and neuropsychiatric benefits in patients with AD than cognitive training of specific domains (e.g., memory; Buschert et al., 2010; Spector et al., 2012; Woods et al., 2012; Bahar-Fuchs et al., 2013; Huntley et al., 2015).

Nevertheless, some studies only evaluated the effect of different forms of neurostimulation without associating neurostimulation with cognitive intervention protocols (Boggio et al., 2009, 2011; Bystad et al., 2016b). To date, three studies have evaluated the combined use of tDCS and cognitive rehabilitation (Cotelli et al., 2014; Penolazzi et al., 2015; Andrade et al., 2016). A common limitation to these studies is the use of a brief protocol of tDCS (only 10 sessions), without long-term follow-up of participants (longer than 6 months). Because AD is a progressive disease, monitoring participants for long periods of time is crucial to assess whether the therapeutic effect is sustained to enable the tested protocol to be transferred to clinical practice. However, a thorough search of the relevant literature yielded no articles analyzing tDCS efficacy associated with cognitive intervention in producing physiological and behavioral changes in patients over time (12 months).

Additionally, the current application moment, meaning whether it is applied before, during, or after cognitive intervention, may also cause variability in the results (Cappon et al., 2016). In a systematic review and meta-analysis assessing the effects of non-invasive neurostimulation on healthy elderly individuals and patients with AD, the authors observed that “the optimal timing for administration of tDCS” varies according to the physiological and pathological aging (Hsu et al., 2015). The use of neurostimulation before task execution can produce robust effects on healthy elderly individuals. Conversely, AD group shows more pronounced effects for studies that apply the stimulation during task execution compared to studies that uses the stimulation before task execution.

Another key point is that most studies involving tDCS in AD typically assess neurostimulation effects on only one cortical area. This aspect is worrying because the pathological characterization of dementia differs from other neurological diseases, wherein only one brain region should be modulated such as stroke or depression (Iannone et al., 2016; Ilić et al., 2016). In AD, the clinical symptoms are not restricted to memory loss, suggesting the involvement of more complex neural networks (Lesuis et al., 2016; Ossenkoppele et al., 2016). Thus, choosing only the frontal or temporal region could limit the scope of treatment and result in the absence of cognitive gains, as found by Bystad et al. (2016a) who observed no improvement in the cognitive domains of patients when applying tDCS over a short period (10 days) exclusively on the left temporal cortex. Thus, it is important that the therapeutic protocol of tDCS in AD should encompass another cortical locus.

Related to this, other factors must be analyzed such as the stimulation duration. Although the standard tDCS protocol only involves application in one cortical area per session of 20 min (Nitsche et al., 2008), some authors also reported that 10 min of tDCS application in one (Drummond et al., 2017) or two cortical regions, one after the other (frontal and parietal cortex; Jones et al., 2015), is sufficient to promote neuromodulatory effects. In addition, Fricke et al. (2011) prove that the use of repeated applications of tDCS generates long-lasting effects.

Although there are several studies involving non-invasive neurostimulation and cognitive training in the treatment of AD, the cognitive and functional effects remain unclear. Recent reviews suggest larger samples to investigate the synergistic effects of these two therapies, with control of confounding variables and with realistic parameters, aimed at clinical practice (Hsu et al., 2015; Zhao et al., 2017).

Accordingly, we propose that stimulating different brain regions specifically related to the neuropathology of AD along with repeated applications of 10-min tDCS in each session may be an option to achieve short- and long-term effects. However, at this time we have only just drawn up a study protocol and these hypotheses can only be tested in the main clinical trial that we aim to perform. In addition, no reports of factorial randomized clinical trials using with tDCS and cognitive intervention stimulating different cortical regions with a 10-min interval in each, and with follow-up for AD-related outcomes have been found to date.

In this way, we designed the Neurostimulation combined with Cognitive Intervention in AD (NeuroAD) clinical study, a phase II/III trial, to address the safety and efficacy of tDCS and cognitive intervention in the treatment of AD. Based on the aforementioned data, which suggest beneficial effects of both cognitive intervention and tDCS on patients with AD, we hypothesize that combining both treatments modulates cortical activity and improves the long-term clinical symptoms of patients, manifested as increased cognition and functional capacity and reduced affective and behavioral changes.

Figure 2 shows the stages of NeuroAD.

The phase I is a double-blind, randomized, factorial clinical trial in which tDCS will be applied in parallel to cognitive intervention in AD patients. This stage includes two measurement time points: baseline (week 0) and post-treatment (week 8). The data collection sessions will be performed by treatment-blinded investigators including the primary and secondary outcomes assessments. The pre-measurement consists of a screening and measurement part. The screening part contains data of the DSM-IV-TR, NINCDS-ADRDA criteria, CDR, MMSE and health questionnaire (including questions about disease duration, comorbidities and use of medication). Adverse effects will be assessed at each session, wherein the tDCS rater will ask the patients whether they feel any discomfort or pain and whether this could be related to tDCS.

Phase II is an open-label, crossover phase in which non-responders from groups (2) and (4) (i.e., those receiving sham tDCS) receive a course of active tDCS using the same parameters and montage of phase I, that is, a direct current of 2 mA (current density = 0.13 mA/cm²) for 30 min/day for 24 days, in respective six brain regions.

Phase III is the longitudinal study of NeuroAD in which all active-tDCS responders—that is, responders from groups (1) and (3) from phase I and responders from phase II will be enrolled in a bi-monthly 1-year follow-up. Thus, two months after the last session, the participants will receive ten consecutive tDCS sessions (from Monday to Friday), which will be repeated every other month, until completing a one-year period (totaling 70 follow-up sessions). The tDCS parameters will be the same as those applied during the period between baseline and post-treatment, and the same cortical areas will be stimulated.

The expected significant improvement is reaching 3.76 points in the mean ADAS-Cog, with a SD of 1.32 points, according to Rabey et al. (2013). Thus, we consider the difference of at least three points in the ADAS-Cog and in the pre- and post-treatment measures to be clinical improvement. All participants with clinical improvement will be considered as respondents.

In all steps, the primary and secondary outcomes assessments will be repeated. The pre-intervention measurement is planned to be performed in the week before the intervention starts and the post-intervention measurement is planned for the first week after the intervention.

Considering there is a serious lack of factorial designs for clinical trials in non-invasive cerebral stimulation for AD patients, we aim to ameliorate that situation by observing the effect size and clinical effectiveness of each procedure, not only comparing this intervention against a sham (which could prove the efficacy, but not the clinical effectiveness), but also against a cognitive intervention. Further, we hope to broaden the treatment scope, promoting improvement in cognitive and functional aspects by stimulating different cortical regions, in addition to memory. Therefore, our study may generate key results for clinical practice and patient management throughout the disease.

Then, to the best of our knowledge, our study may be the first to apply a factorial design in non-invasive cerebral stimulation for AD patients, and not only comparing this intervention against a sham (which could prove the efficacy, but not the clinical effectiveness), but also against a cognitive intervention. Next, this factorial design will be able to observe the effect size and clinical effectiveness of each procedure, and will also be able to elucidate whether it would actually be feasible to apply the therapies alone or in combination. Further, we hope to broaden the treatment scope, promoting improvement in cognitive and functional aspects by stimulating different cortical regions, in addition to memory. Therefore, our study may generate key results for clinical practice and patient management throughout the disease.

An additional relevant issue is the longer-lasting effects. Evidence shows that multisession protocols promote neural plasticity and after-effects in AD and other conditions (Fregni et al., 2015; Bystad et al., 2016b). Although a case study that evaluated the effects of tDCS up to 8 months has been reported in the literature (Bystad et al., 2017), we hope to find that bimonthly stimulation sessions amplify the produced effects. Likewise, the use of long-term physiological markers will allow us to know what kind of non-pharmacological therapies produce greater neuroplasticity effects in AD patients, which are responsible for reaching and maintaining better long-term overall performance. The implementation of more effective procedures of plasticity induction in clinical settings is essential—in fact, as a maintenance treatment for bimonthly sessions, tDCS can offer the advantage of a cost-effective approach (reducing missed visits) and a less expensive alternative compared to daily sessions in the long-term. In this way, our findings may also help improve the recommendation management of AD.

Fundamentally, our study presents some methodological limitations that should be discussed, including adherence. The participants must come to our laboratory to receive 24 sessions three times a week. This may result in absences due to transportation problems or personal difficulties. However, different strategies will be adopted to minimize these problems, such as flexible hours and frequent phone calls, encouraging and reinforcing the importance of treatment. Also, we will set a limit of missed appointments so that the treatment process is less rigid, and the patient can resume the missed sessions without compromising the treatment follow-up if necessary. Second, applying the neuropsychological tests could elicit a learning effect. The practice-adjusted RCI of the different cognitive tests used in the clinical trial will be used to minimize this limitation (Goldberg et al., 2015), as described in the statistical analysis. Although some tests may have increased susceptibility to practice effects (Calamia et al., 2012), patients with AD show no significant effects in practice regarding repeated administrations of different cognitive tests after a test-retest interval (Cooper et al., 2004; Foley et al., 2015). A third consideration is the inclusion of only mild AD patients, thereby limiting the generalization to other phases of the disease. Considering cognitive and functional differences between the different stages of AD, we chose to evaluate patients at only one stage of the disease for improved data control. However, this does not preclude future studies comparing patients at different stages of the disease to specifically clarify the best dosage, procedure and number of tDCS sessions and cognitive intervention, considering the clinical symptoms and disease progression of the patients. Finally, although our proposition is to stimulate different regions affected by AD, we cannot assure that there will be no cumulative effects between them. In view of intersubject and intrasubject variations, many studies claim to have the focal power on tDCS, so we cannot rule out the possibility of effect propagation from one region to another (Boros et al., 2008; Cunningham et al., 2015). As our study does not aim to evaluate or compare the efficacy of tDCS at each site separately, cumulative effects from synaptic connections between the involved regions may result in improved cognitive and functional performance, and not cause harm to patients. Nevertheless, it is important to emphasize that we will use several strategies to control bias, such as reduced electrode size, use of neuronavigation to localize target regions in each patient, and the use of standardized coordinates from previous studies for localizing target regions.

In general, we hope that our results can elucidate the relevance of such potential limitations. Additionally, we consider that the current protocol together with our results will have important implications for AD treatment, both theoretically in providing support for new trials and for clinical management, thus providing evidence applicable to the routine care of these patients.

SA and BF-C developed and drafted this study protocol in consultation with EO, NA, MQCG and MR. RN and IC provided inputs to the design of the customized group neurostimulation and cognitive program intervention. AS, CM, DS, ES, EF, ER and GL prepared the human research ethics. JC, JS, MT, MO, ML, NL, PR, RF and RC were involved in the overall research design and selection of outcome measures. RA, RS, PI, TP and WM were involved in the design and facilitation of the group program intervention. TF and VP were involved in the mixed-methods research methodologies, outcome measures and structured interview questions. All authors listed have made a substantial, direct and intellectual contribution to the work and approved it for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.