The effects of repetitive transcranial magnetic stimulation and aerobic exercise on cognition, balance and functional brain networks in patients with Alzheimer's disease

Abstract

The purpose of this study was to investigate the effects of high-frequency repetitive Transcranial Magnetic Stimulation (rTMS) and aerobic exercises (AE) in addition to the pharmacological therapy (PT) in Alzheimer's Disease (AD). Twenty-seven patients with AD aged ≥ 60 years were included in the study and divided into 3 groups (rTMS, AE and control). All groups received PT. rTMS group (n = 10) received 20 Hz rTMS over dorsolateral prefrontal cortex (dlPFC) bilaterally and AE group (n = 9) received the structured moderate-intensity AE for 5 consecutive days/week over 2 weeks. Control group (n = 8) only received PT. Cognition, balance, mobility, quality of life (QoL), and resting state functional brain activity were evaluated one week before and one week after the interventions. (ClinicalTrials.gov ID:NCT05102045). Significant improvements were found in executive functions, behavior, and QoL in the rTMS group, in balance and mobility in the AE group, and in the visual memory and behavior in the control group (p < 0.05). Significant differences were found in the behavior in favor of the rTMS group, and balance in favor of the AE group (p < 0.05). There was a significant increase in activation on middle temporal gyrus, intra calcarine, central opercular cortex, superior parietal lobule, and paracingulate cortex in Default Mode Network (DMN) in the rTMS group (p < 0.05). High-frequency rTMS over bilateral dlPFC may improve executive functions and behavior and lead to increased activation in DMN, structured moderate-intensity AE may improve balance and mobility, and PT may improve memory and behaviour compared to pretreatment in AD.

Introduction

Alzheimer's Disease (AD) is the leading cause of dementia and progressive neurodegenerative disease associated with a gradual decline in cognitive functions such as memory, attention, perception, spatial skills, language, and executive functions (Frisoni et al. 2017). Pharmacological interventions are the main approach in the treatment of AD, however, their effectiveness diminishes in time. Medications approved by the US Food and Drug Administration include 3 Acetylcholinesterase inhibitors (AChEIs: donepezil, galantamine, and rivastigmine), memantine (that targets the glutamatergic system), and a combined treatment of AChEIs and memantine (Briggs et al. 2016). AChEIs is effective for mild-to-moderate disease, while glutamate regulators (e.g. memantine) are effective for moderate-to-severe disease stages (O’Brien et al. 2017). Therapies aimed at targeting altered pathways based on misfolding protein aggregation, mitochondrial dysfunction, oxidative stress, autophagy impairment, alterations in intracellular Ca²⁺ homeostasis, neuroinflammation, and neurogenesis impairment are among the new techniques being used in AD treatment (Bulck et al. 2019). Phosphodiesterase 5 (PDE5) inhibitors are used as an alternative approach in altering the nitric oxide pathway which plays a multifactorial role in several cellular processes leading to an improvement of learning and memory in AD (Zuccarello et al. 2020). Additionally, new insights into the genetics of AD recently revealed a number of druggable targets and signal transduction pathways that are only expressed in microglia and central nervous system-resident macrophages in the brain. It was shown that potential antibody-mediated triggering receptor of myeloid cells 2 receptor treatment seems to have a huge potential (Lewcock et al. 2020). Among the novel therapeutic approaches are anti-amyloid therapy, anti-tau therapy, anti-neuroinflammatory therapy, neuroprotective agents including N-methyl-D-aspartate (NMDA) receptor modulators, and brain stimulation (Yu et al. 2021b). These treatments have been shown to reduce the progression of the disease, but they cannot prevent its initiation. Despite the degenerative nature of the disease, there is currently no effective treatment to stop its harmful progression. Recent studies have suggested that moderate-intensity aerobic exercises (AE) and repetitive transcranial magnetic stimulation (rTMS) may be alternative treatment modalities for AD by increasing the neural plasticity which may slow down the cognitive decline (Ni and Chen 2015; Yu et al. 2021a).

In recent years, while drug repositioning does have the potential to greatly improve pharmacological development, non-pharmacological approaches, notably rTMS, also have clinical potential (Yu et al. 2021b). In recent years, rTMS has evolved as a potentially safe and cost-effective treatment for AD (Jiang et al. 2020). rTMS is a non-invasive technique that involves inducing weak electrical currents in a rapidly changing magnetic field, which produces changes in neuronal polarization and activity. While rTMS affects brain activity, its effects can last for minutes or hours, simulating the long-term potentiation (LTP) process. Learning new information and adjusting to changes are facilitated with LTP, which requires the plasticity of neural networks (Nevler and Ash 2015). Recent research has found that rTMS is shown to improve cognitive functions such as memory, and language ability in Alzheimer's patients, especially in the moderate or early stages of the disease (Lefaucheur et al. 2020). The effects of rTMS stimulation have been shown in previous research to depend on the position in which the coil is fixed. However, these effects are not limited to the location of the coil. The majority of the effects of rTMS involve changes in the brain networks of the cortex and subcortical regions, rather than changes in local inhibition or excitability in specific brain regions. The activity of the brain network is crucial, because altering a single point in the network can affect the entire network. Due to the spread of excitability or the internal links of the brain network, the influence on one area will affect other areas from a functional and overall perspective. Dorsolateral prefrontal cortex (dlPFC), which is also responsible for regulating working memory and cognition, is the target region for rTMS stimulation in some studies (Otani 2002; Hiser and Koenigs 2018). dlPFC stimulation has been demonstrated to be a suitable area for targeting Default Mode Network (DMN) activation in studies (Pievani et al. 2017).

AE has been shown to provide a positive effect on executive functions and general cognition in patients with AD, and can be an effective way to slow down or limit the progression of neurodegeneration in the brain (Öhman et al. 2016; Morris et al. 2017; Guitar et al. 2018). Additionally, exercise training may reduce decline in global cognition in older adults with mild-to-moderate AD dementia (Yu et al. 2021a). It has been shown that AE may improve cognitive function by enhancing brain vascularization and producing neurotrophic changes (Nagy et al. 2021). According to neuroimaging studies, AE is shown to affect the brain function and structure (Haeger et al. 2019). Furthermore, AE may lead to volumetric changes in the hippocampus and memory (Erickson et al. 2011). Neuronal activity in the prefrontal region was increased after AE in Alzheimer's patients. Interestingly, exercise-induced improvements in cardiorespiratory fitness have been linked to better memory performance and less hippocampal atrophy (Morris et al. 2017). AE tends to have beneficial effects on the right hippocampus, as well as the hippocampus in general and other regions, the cingulate cortex, and the medial temporal areas of the DMN, according to studies. Additionally, AE has been shown to increase functional connectivity or activation in the hippocampus, cingulate cortex, and parahippocampal gyrus in association with the DMN (Li et al. 2017).

Functional magnetic resonance imaging (fMRI) has been shown to be a particularly useful mean for detecting early alterations in brain function and may be a critical marker for the detection of physiological changes over a short interval, given the growing body of evidence that alterations in synaptic function are present very early in the course of the neurodegenerative disease process (Agosta et al. 2017). Resting-state fMRI (rs-fMRI) has the potential to detect minor functional abnormalities in brain networks that enable complex cognitive processes that diminish over time in AD (Agosta et al. 2012). Several studies of Alzheimer's patients have found alterations in the DMN and other resting-state networks (RSN) that also are related to cognitive functions (Badhwar et al. 2017).

There is a lack of evidence comparing the effects of rTMS and AE on cognitive functions and functional changes in the brain in addition to pharmacological therapy (PT) in AD. The purpose of this study was to investigate the effects of rTMS and AE on cognition, physical functions, and neurobiological changes reflected in the connectivity of functional brain networks in patients with AD. We hypothesized that rTMS treatment on dlPFC would improve executive function and behavior, AE would improve physical parameters, and PT would improve memory and behavior in Alzheimer's patients, and these effects would be reflected as activation in DMN in rs-fMRI.

Materials and methods

Study design

This study is a randomized controlled clinical trial. The study was approved by the Istanbul Medipol University Non-Interventional Clinical Research Ethics Committee (no. 575/18.11.2015). The study was registered in ClinicalTrials.gov (ClinicalTrials.gov ID: NCT05102045). All participants signed a written informed consent form, and the study was conducted in accordance with the principles of the Declaration of Helsinki.

Study population

The study was conducted with AD patients who were followed-up in the Istanbul Medipol University Hospital Neurology Outpatient Clinic and fulfilled the inclusion criteria. All participants were recruited from Istanbul Medipol  University Hospital between September 2015-May 2017. Incomplete data were collected from three patients who are not suitable for MRI. All patients were eligible for MRI and TMS procedures on standard MRI safety screening as well as on their answers to a TMS safety-screening questionnaire.

The patients who got clinical AD diagnosis according to the National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA) criteria and The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V), 60 years and older, with Clinical Dementia Rating Scale (CDR) scores 1 or 2, using AChEIs and Memantine and living independently were included in the study. Exclusion criteria were defined as having metallic implants, not being able to walk independently, having physical disabilities, a history of alcohol/substance abuse, having mental illnesses including schizophrenia and delirium, and having epileptic disease, seizures, brain tumors, or trauma.

Forty-five patients were screened, 30 patients who met the inclusion criteria were included in the study. Three of 30 patients withdrew their consent during the study and were excluded from the study. The study was completed with 27 patients (17 female).

Patients were randomly allocated to three groups as rTMS group, AE group and control group using block randomization. In the rTMS group (n = 10) rTMS was applied to the bilateral dlPFC at 20 Hz, for 5 consecutive days per week for 2 weeks (totally 10 sessions). AE group (n = 9) received a structured moderate-intensity AE program lasting 50 min per session, 5 consecutive days per week for 2 weeks (totally 10 sessions). No additional intervention was given to the patients in the control group (n = 8) and participants were only treated pharmacologically. All patients continued to receive their appropriate medication prescribed by the neurologist (Medicines Agency 2018). PT was selected from the approved pharmacological protocols for the treatment of AD, including AChEIs (Donepezil, Rivastigmine, Galantamine) and glutamate regulators (Memantine) (Medicines Agency 2018). There was no change in ongoing drug treatments during the study period for the rTMS and AE groups. After the evaluation process for the drug group, the drug doses (AChEIs and Memantine) were adjusted by the neurologist in the required patients, and the dose was increased if clinically necessary. Participants were requested not to change their present physical activities in any way except as directed by the research team. The treatment was administered by the same person throughout the therapy procedures. The participants were told that due to the course of the disease, only routine drug treatment would be continued for the time being. At the end of the study, following the recommendations of the neurologist and physiotherapist, AE was administered to the participants in the rTMS and control groups.

Experimental design

Participants in the rTMS group (n = 10) received 10 sessions of rTMS therapy in addition to the PT, in the AE group (n = 9) sessions of AE in addition to the PT, in the control group (n = 8) received only PT over two weeks. One week prior to interventions the following data was collected from patients: expanded Turkish version of neuropsychometric tests, physical tests (balance, functional mobility), quality of life, and anatomic and functional MRI scans. After one week from interventions, the same data were recollected from patients.

rTMS protocol

Individual maps of hippocampal rs-fMRI collected before rTMS were used to determine a stimulation coordinate for each patient. According to Cieslik et al., the coordinates of the right and left dlPFC were ± 30, 43, 23 (MNI coordinates) (Cieslik et al. 2013). The left dlPFC was used as a control area because one of the goals of the study was to confirm the involvement of the right dlPFC in error awareness. Consequently, we chose to target the left dlPFC using the same coordinates as the right dlPFC (MNI coordinates: 30, 43, 23), but only adjusting the x-parameter (MNI coordinates: − 30, 43, 23), a technique for selecting a control site that has previously been utilized in TMS studies (Herwig et al. 2003; Vallesi et al. 2007). After that, we used the “Brainvoyager TMS Neuronavigation” software suite's (Brain Innovation B.V., The Netherlands) instrument to apply rTMS to these coordinates. We obtained 3D brain imaging using the Power "Mag for CMS20" software. The participants in the rTMS group received baseline T1-weighted brain images recorded at a high resolution (1 mm³). The neuronavigation software received spatial position communication from the TMS coil and the patient's skull (brain) via a small sound-emitting apparatus.

rTMS pulses were delivered using a PowerMag TMS device (MAG & More GmbH, Germany) and an internally cooled Fig. 8-coil with a winding diameter of 2 × 70 mm. The targeted brain regions were consistently stimulated during sessions on consecutive days using neuronavigation markers. The hand area of the left primary motor cortex was scanned using TMS pulses to determine the cortical region providing a reaction on the "right abductor pollicis brevis muscle" depending on the obtained 3D images. Before each session, the resting motor threshold was determined, and the TMS pulse strength that evoked an electromyographic response higher than 50 microvolts in at least 5 of 10 trials was accepted as the threshold. The pulse frequency was 20 Hz, and there was a 5 s interval between trains of 75 pulses. Totally 3000 rTMS pulses on dlPFC (1500 pulses on the right and left hemispheres, respectively) were applied in each session. In addition to the PT, the rTMS protocol included ten 50-min sessions carried out over two weeks with a weekend break. The coil was placed on the skull at a 45⁰ angle over the application areas, parallel to the head's anterior–posterior axis (Ahmed et al. 2012). In terms of safety and ethics, the rTMS protocol was ensured to be compliant with the literature and guidelines (Rossi et al. 2009).

Fig. 8.

Central Opercular Cortex

AE protocol

The structured AE protocol included ten 50-min sessions carried out over two weeks, in addition to the PT. The protocol was based on World Health Organization (WHO) recommendations which consists of 10 min warm up exercises, 30-min moderate AE and 10-min cool down exercises (Waxman 2004). A E consisted of simple to difficult, repetitive movements and walking exercises without equipment. The progression of the exercises was determined by the BORG Rating of Perceived Exertion (RPE) scale. For older adults with AD, moderate intensity was prescribed as 12 to 14 on the 6 to 20 Borg RPE scale (Yu et al. 2015). Moderate-intensity AE was performed so that the amount of fatigue felt according to the BORG scale increased from 11–12 to 13–14. Both warm up and cool down exercises included neck muscle stretching, trunk right/left lateral flexion, biceps and triceps stretching, rowing, shoulder elevation and circumduction, and quadriceps and hip flexors stretching. AE included stepping exercises, stepping exercises with excessive hip and knee flexion, forward stepping, sideways stepping, moderate-intensity walking, and bilateral shoulder flexion and abduction exercises. The exercises were performed under the supervision of a physiotherapist in the first session. All subsequent sessions were followed daily by phone and video call using the telerehabilitation method. The participants followed the WHO's home exercise protocol under the supervision of a physiotherapist (Realdon et al. 2016). Based on the data from the American Heart Association, the AE included 30 min of moderate-intensity walking which is 50–70% of the maximal heart rate (Piercy and Troiano 2018). Exercise intensity was controlled against ventilation thresholds, and individuals were instructed to continue exercising at a rate at which they could talk but not sing, even if their heart rate was elevated.

Outcome measurements

One week prior to interventions the following data was collected from patients: expanded Turkish version neuropsychometric tests, physical tests (balance, functional mobility), quality of life, and rs-fMRI. After one week from interventions, the same data were recollected from patients. Mini-Mental State Examination (MMSE) was used to evaluate global cognition (Güngen et al. 2002). Digit Span Test was used to evaluate attention functions (Kurt et al. 2011). Wechsler Memory Scale (WMS) was used as part of a general neuropsychological evaluation. Memory functions were evaluated WMS Logical Memory (Immediate and Delayed), WMS Visual Reproduction Test (Immediate and Delayed), WMS Visual Reproduction Test Recognition (Karakas et al. 1999), and Oktem Verbal Memory Process Test (VMPT) (Bosgelmez et al. 2015); language skills were tested by Boston Naming Test(BNT) (Ekinci Soylu and Cangöz 2018); visual and perceptual functions were measured by Benton Facial Recognition Test ((BFRT) Schretlen et al. 2001); and executive functions were evaluated by abstraction, the Semantic Fluency, the Stroop Test (Karakas et al. 1999), and the Clock Drawing Test (CDT) (Shulman 2000). Patients were evaluated for non-motor and behavioral symptoms with the Geriatric Depression Scale (GDS) (Durmaz 2017), The Neuropsychiatric Inventory Questionnaire (NPI) (Trzepacz et al. 2013), and Frontal Behavioral Inventory (FBI) (Kim et al. 2014). Balance was evaluated with Berg Balance Test (BBT) (Sahin et al. 2008), functional mobility was evaluated with Timed Up and Go Test (Richardson 1991), and quality of life was evaluated with Quality of Life in Alzheimer's Disease Measure (QoL-AD) (Kahle-Wrobleski et al. 2017).

MRI protocol

The MRI scans were conducted using a 3 T Philips Achieva scanner with an 8 channels head coil (Philips Medical Systems, The Netherlands) at Istanbul Medipol University Hospital, Department of Radiology (Istanbul, Turkey). The anatomical images were obtained with T1-weighted imaging sequence (TR/TE: 7.7/3.7 ms, FA: 8°) covering the entire head and neck (FOV: 256 × 256) with high resolution (1 × 1 × 1 mm isotropic) including 190 sagittal sections. For the functional imaging studies, 150 dynamic scans were acquired at rest with a single-shot echo-planar imaging (EPI) sequence that consists of 40 slices of 3.5 mm thickness (TR/TE: 2000/30 ms, FA: 90°, matrix: 96 × 96). During functional scans, the participants were asked to fix their gaze on a spot inside the scanner while keeping their eyes open. The participants might close their eyes and rest during other times in the scanner. To minimize movement artifacts, the patients' heads were stabilized with spongy pads inserted between the temporals and the head coil. The imaging process took 30 min to complete, including preparation time.

Analysis

rs-fMRI data analysis

The tools included in the FMRIB FSL software package were used to accomplish the preprocessing steps of the functional image analysis and subsequent functional network analysis (Smith et al. 2004). The dcm2nii tool contained in the MRIcro-GL software package was used to convert DICOM files into NIFTI + format during the preprocessing procedures. The brain tissue was extracted from the brain images using the FSL BET program. Using the tools supplied in the FSL FEAT interface, the following tasks were completed. A 150-s high-pass filter was used to filter the functional data. The head motions between dynamic scans were corrected using the FSL MCFLIRT algorithm (Jenkinson et al. 2002). For group comparisons, the functional and anatomic data were spatially adjusted. The spatial normalization of functional and anatomical images was performed in three steps using the FSL FLIRT and FNIRT linear and non-linear registration tools, respectively. Using limited transformations, the functional brain images were first matched to the patients' own high-resolution anatomical images (6 DOF). The patients' high-resolution anatomical images were matched with MNI152 standard brain images utilizing 12 DOFs and non-linear transformations in the second step. The distortion resolution was ten millimeters. Using transformation matrices from previous transformations, low-resolution functional images were mapped to the MNI152 brain in the third step. In a separate step, each participant's functional data in the native space was processed to an exploratory Independent Component Analysis (ICA) using the FSL MELODIC tool to determine movement, physiological (heart, respiration, etc.) and other imaging artifacts (Beckmann and Smith 2005). Those compatible with the noise were manually marked and filtered from the functional data using the FSL reg_filt tool by analyzing signal features of ICA components such as spatial distribution, frequency spectrum, and temporal fluctuation. The functional data was transformed into the standard space after artifact removal in the native space. The FSL MELODIC tool was used to perform a group ICA with 30 components on the preprocessed artifact free data. The components' spatial correlation with canonical resting-state networks was evaluated, and ICA components with high correlation with these networks were determined. The subject's own spatial component maps and temporal data of the common ICA components were acquired using the FSL dual regression tool for statistical analysis (Qiao et al. 2018). The GLM matrix is constructed for a 2-way (3 Group by 2 Time) repeated measures design using a mixed ANOVA, and contrasts for intra- and inter-subject factors were assessed. Finally, non-parametric inference using 5000 random permutations was used to examine the significance of the differences (FSL randomize tool). Using cluster-based TFCE (Threshold-Free Cluster Enhancement) corrected for multiple comparisons (Family-wise Error Rate), the results were considered significant at p < 0.05. The reported mean z-values were calculated from the regions in each spatial map showing significant differences between groups.

Sample size planning

The sample size was determined using the G*power sample size calculator (Faul et al. 2007). Sample size was calculated using repeated measure design with an 80% power (α = 0.05, β = 0.20) and effect size of 0.5 for a sample size of 24 participants (8 per group).

Statistical analysis

SPSS (Statistical Package for Social Science) 25.0 for Windows was used for statistical analysis. The normal distribution of the variables was tested by the Kolmogorov–Smirnov Test. Regardless of the homogeneity of the variances, Time dependent differences within group were analyzed by Two Way Repeated Measure ANOVA, and Time*Group interactions were analyzed by MANOVA. Bonferroni correction was used for post-Hoc tests. Significance value was accepted as p < 0.05.

Results

Forty-five patients were screened, 30 patients who met the inclusion criteria were included in the study. Three of 30 patients withdrew their consent during the study and were excluded from the study. One out of ten patients in the AE group and two out of ten patients in the control group did not come to the second assessment, and one out of ten patients in the control group did not continue because of personal problems. The study was completed with 27 patients (17 female). The algorithm for patient allocation into the study groups was shown in Fig. 1.

Fig. 1.

Flow Chart

The average age was 72.00 ± 5.01 in the rTMS group, 73.60 ± 8.62 in the AE group and 74.90 ± 8.11 in the control group. The minimum and maximum age was 60–88 respectively. Seventeen of the participants were female. Nine of the participants were not literate, 12 of them were educated 1–5 years, and 6 of them were educated ≥ 8 years. All participants were right-handed. 6 of the participants used AChEIs, 9 of them used Glutamate regulators, and 12 of them used a combination of AChEIs and Glutamate regulators. There was no statistically significant difference between groups in terms of age, gender, years of education, dominant hand, and the drugs they used (p > 0.05). The distribution of demographic data is shown in Table 1.

Table 1.

Distribution of demographic data

	rTMS group (n = 10)	AE group (n = 9)	Control group (n = 8)	p value
Age (Avg ± SD)	72.00 ± 5.01	73.60 ± 8.62	74.90 ± 8.11	0.491
Gender
Female (n/%)	6/60	4/44.4	7/87.5	0.192
Male (n/%)	4/40	5/55.6	1/12.5
Education level
Not literate (n/%)	1/10	5/55.6	3/37.5	0.078
1–5 years (n/%)	5/50	3/33.3	4/50
≥ 8 years (n/%)	4/40	1/11.1	1/12.5
Dominant hand
Right (n/%)	10/100	9/100	8/100	1
Left (n/%)	0/0	0/0	0/0
Information of the pharmacological treatment
AChEIs (n/%)	2/20	3/33.3	1/12.5	0.726
Memantine (n/%)	3/30	2/22.2	4/50
Combination of AChEIs and memantine (n/%)	5/50	4/44.4	3/37.5

rTMS repetitive transcranial magnetic stimulation, AE aerobic exercise, AChEIs asetilcholinesterase inhibitors, Avg average, SD standard deviation. *p < 0.05

Within group changes are shown in Table 2, 3, and 4. There were statistically significant improvements in CDT, NPI, FBI, and QoL-AD in the rTMS group (p < 0.05) (Table 2), in BBT and TUG in the AE group (p < 0.05) (Table 3); and in the visual delayed recall and NPI in the control group (p < 0.05) (Table 4).

Table 2.

Within group findings in rTMS group

rTMS group	(n = 10)	Pre treatment	Post treatment	Mean difference	Confidence of interval (lower to upper)	F	Effect size (Cohen’s d)	p value
		Avg ± SD	Avg ± SD
Attention	Digit span forward	4.00 ± 1.94	3.80 ± 1.93	0.200	− 0.908 to 1.308	0.167	0.018	0.693
	Digit span backward	2.50 ± 1.64	2.30 ± 1.25	0.200	− 0.252 to 0.652	1.000	0.100	0.343
Executive functions	Stroop	29.35 ± 18.31	25.83 ± 19.82	3.526	− 4.890 to 11.942	0.898	0.091	0.368
	VF-fruit name pairs	4.49 ± 2.62	3.70 ± 2.58	0.796	− 0.555 to 2.147	1.777	0.165	0.215
	Categorical fluency	9.70 ± 3.46	9.80 ± 3.91	− 0.100	− 1.834 to 1.634	0.017	0.002	0.899
	Phonemic fluency	12.87 ± 5.50	10.49 ± 5.58	2.379	− 0.927 to 5.685	2.651	0.228	0.138
	CDT	1.10 ± 0.87	2.47 ± 1.59	− 1.370	− 2.274 to  − 0.466	11.750	0.566	0.008*
Memory	Visual immediate recall	1.92 ± 1.36	1.30 ± 0.83	0.620	− 0.286 to 1.526	2.398	0.210	0.156
	Visual delayed recall	0.76 ± 0.78	0.50 ± 0.97	0.260	− 0.494 to 1.014	0.608	0.063	0.456
	Visual recognition	0.50 ± 0.47	0.79 ± 1.22	− 2.292	− 1.069 to 0.485	0.724	0.074	0.417
	VMPT total	47.00 ± 20.3	45.10 ± 14.54	1.900	− 5.391 to 9.191	0.348	0.037	0.570
	VMPT immediate recall	2.50 ± 1.58	2.90 ± 1.44	− 0.400	− 1.477 to 0.677	0.760	0.073	0.423
	VMPT delayed recall	1.00 ± 2.49	0.70 ± 1.88	0.300	− 2.086 to 2.686	0.081	0.009	0.782
	VMPT recognition	5.00 ± 3.91	6.40 ± 3.83	− 1.400	− 3.667 to 0.867	1.951	0.178	0.196
Language	BNT	17.78 ± 5.15	18.64 ± 6.09	− 0.865	− 3.237 to 1.506	0.681	0.070	0.430
Visual spatial functions	BFRT	40.11 ± 4.66	40.47 ± 2.89	− 0.360	− 2.455 to 1.746	0.150	0.016	0.708
General cognition	MMSE	17.50 ± 4.50	19.5 ± 5.70	− 2.000	− 4.361 to 0.361	3.673	0.290	0.088
Behavior	NPI	41.00 ± 21.60	21.90 ± 13.60	19.100	8.747 to 29.453	17.419	0.659	0.002*
	FBI	18.20 ± 11.82	7.80 ± 3.76	10.400	2.284 to 18.516	8.403	0.483	0.018*
Balance	BBT	45.40 ± 9.55	46.90 ± 7.93	− 1.500	− 3.842 to 0.842	2.098	0.189	0.181
Mobility	TUG	27.10 ± 11.99	24.80 ± 9.50	2.300	− 2.458 to 7.058	1.196	0.117	0.303
Quality of life	QoL-AD	31.90 ± 8.25	34.90 ± 8.03	− 3.000	− 5.698 to  − 0.302	6.328	0.413	0.033*

Avg average, SD standard deviation, rTMS repetitive transcranial magnetic stimulation, VF verbal fluency, CDT clock drawing test, VMPT verbal memory process test, BNT Boston nemini test, BFRT beton facialar recognition test, MMSE mini-mental state examination, NPI neuropsychiatric inventory, FBI frontal behavioral inventory, BBT berg balance test, TUG timed-up and go test, QoL-AD quality of life-Alzheimer’s disease. *p < 0.05

Table 3.

Within group findings in AE group

AE group	(n = 9)	Pre treatment	Post treatment	Mean difference	Confidence of Interval (lower to upper)	F	Effect size (Cohen’s d)	p value
		Avg ± SD	Avg ± SD
Attention	Digit span forward	3.55 ± 1.58	4.33 ± 1.50	− 0.778	− 2.097 to 0.541	1.849	0.188	0.211
	Digit span backward	2.11 ± 1.16	2.55 ± 0.72	− 0.444	− 1.224 to 0.335	1.730	0.178	0.225
Executive functions	Stroop	22.50 ± 19.13	17.25 ± 13.82	5.251	− 0.714 to 11.217	4.120	0.340	0.077
	VF—fruit name pairs	4.11 ± 1.61	4.22 ± 1.39	− 0.111	− 1.160 to 0.938	0.060	0.007	0.813
	Categorical Fluency	10.55 ± 3.60	13.11 ± 4.75	− 2.556	− 5.460 to 0.349	4.117	0.340	0.077
	Phonemic Fluency	12.55 ± 7.76	13.66 ± 8.60	− 1.111	− 5.168 to 2.946	0.399	0.047	0.545
	CDT	0.88 ± 0.92	1.50 ± 1.40	− 0.617	− 1.268 to 0.035	4.766	0.373	0.061
Memory	Visual immediate recall	3.55 ± 3.57	2.44 ± 2.29	1.111	− 0.494 to 2.716	2.548	0.242	0.149
	Visual delayed recall	0.55 ± 1.01	0.88 ± 1.36	− 0.333	− 0.999 to 0.332	1.333	0.143	0.282
	Visual recognition	0.55 ± 0.88	0.88 ± 1.05	− 0.333	− 1.275 to 0.608	0.667	0.077	0.438
	VMPT Total	44.88 ± 9.73	47.66 ± 15.46	− 2.778	− 13.613 to 8.057	0.350	0.042	0.571
	VMPT immediate recall	2.66 ± 1.41	2.77 ± 0.97	− 0.111	− 1.572 to 1.350	0.031	0.004	0.865
	VMPT delayed recall	0.11 ± 0.33	1.66 ± 3.08	− 1.556	− 3.990 to 0.879	2.172	0.214	0.179
	VMPT recognition	7.44 ± 2.92	6.00 ± 1.93	1.444	− 0.733 to 3.622	2.339	0.226	0.165
Language	BNT	18.11 ± 6.67	18.88 ± 4.78	− 0.778	− 3.406 to 1.851	0.466	0.055	0.514
Visual Spatial Functions	BFRT	37.33 ± 5.72	36.77 ± 3.96	0.556	− 5.081 to 6.192	0.052	0.006	0.826
General Cognition	MMSE	18.66 ± 2.64	20.22 ± 4.23	− 1.556	− 4.434 to 1.323	1.552	0.163	0.248
Behavior	NPI	16.55 ± 22.91	15.00 ± 19.75	1.556	− 3.262 to 6.373	0.554	0.065	0.478
	FBI	19.33 ± 8.41	13.88 ± 12.08	5.444	− 1.496 to 12.385	3.272	0.290	0.108
Balance	BBT	42.77 ± 7.61	50.33 ± 4.79	− 7.556	− 11.382 to − 3.729	20.735	0.722	0.002*
Mobility	TUG	20.22 ± 2.58	18.66 ± 2.00	1.556	0.395 to 2.716	9.561	0.544	0.015*
Quality of Life	QoL − AD	33.77 ± 3.38	35.00 ± 3.53	− 1.222	− 4.802 to 2.358	0.620	0.072	0.454

Avg average, SD standard deviation, AE aerobic exercise, VF verbal fluency, CDT clock drawing test, VMPT verbal memory process test, BNT boston naming test, BFRT benton facial recognition test, MMSE mini-mental state examination, NPI neuropsychiatric inventory, FBI frontal behavioral inventory, BBT berg balance test, TUG timed-up and go test, QoL-AD quality of life -Alzheimer’s disease. *p < 0.05

Table 4.

Within group findings in Control group

Control group	(n = 8)	Pre treatment	Post treatment	Mean difference	Confidence of Interval (lower to upper)	F	Effect size (Cohen’s d)	p value
		Avg ± SD	Avg ± SD
Attention	Digit span forward	4.25 ± 0.88	4.25 ± 0.88	0.000	− 0.774 to 0.774	0.000	0.000	1.000
	Digit span backward	2.12 ± 1.12	3.00 ± 0.75	− 0.875	− 1.917 to 0.167	3.943	0.360	0.087
Executive functions	Stroop	17.38 ± 16.90	16.15 ± 13.85	1.230	− 6.355 to 8.816	0.147	0.021	0.713
	VF—fruit name pairs	4.87 ± 2.90	4.12 ± 1.55	0.750	− 1.023 to 2.523	1.000	0.125	0.351
	Categorical Fluency	11.17 ± 6.28	9.75 ± 3.95	1.428	− 1.668 to 4.524	1.189	0.145	0.312
	Phonemic Fluency	13.23 ± 13.98	15.24 ± 13.62	− 2.011	− 4.050 to 0.029	5.433	0.437	0.053
	CDT	1.00 ± 1.30	1.46 ± 1.76	− 0.462	− 1.199 to 0.274	2.205	0.240	0.181
Memory	Visual immediate recall	2.50 ± 2.07	2.62 ± 2.55	− 0.125	− 1.167 to 0.917	0.080	0.011	0.785
	Visual delayed recall	1.12 ± 2.03	1.75 ± 2.37	− 0.625	− 1.247 to  − 0.003	5.645	0.446	0.049*
	Visual recognition	0.50 ± 0.75	1.25 ± 1.03	− 0.750	− 1.822 to 0.322	2.739	0.281	0.142
	VMPT Total	47.25 ± 22.15	46.37 ± 14.96	0.875	− 6.582 to 8.332	0.077	0.011	0.789
	VMPT immediate recall	2.50 ± 1.30	2.87 ± 1.35	− 0.375	− 1.634 to 0.884	0.496	0.066	0.504
	VMPT delayed recall	2.00 ± 2.87	2.25 ± 3.15	− 0.250	− 1.322 to 0.822	0.304	0.042	0.598
	VMPT recognition	9.62 ± 2.97	7.75 ± 2.71	1.875	− 1.267 to 5.017	1.991	0.221	0.201
Language	BNT	17.50 ± 4.03	17.75 ± 4.83	− 0.250	− 2.381 to 1.881	0.077	0.011	0.790
Visual Spatial Functions	BFRT	36.37 ± 7.40	39.25 ± 6.36	− 2.875	− 5.786 to 0.036	5.454	0.438	0.052
General Cognition	MMSE	19.62 ± 4.77	21.37 ± 4.83	− 1.750	− 4.838 to 1.338	1.796	0.204	0.222
Behavior	NPI	13.25 ± 4.94	10.75 ± 12.08	− 7.500	− 14.622 to − 0.378	6.201	0.470	0.042*
	FBI	19.75 ± 13.33	19.62 ± 6.94	0.125	− 8.382 to 8.862	0.001	0.000	0.973
Balance	BBT	42.62 ± 10.95	33.62 ± 20.98	9.000	− 5.209 to 23.209	2.243	0.243	0.178
Mobility	TUG	26.37 ± 14.10	41.00 ± 43.72	− 14.625	− 50.978 to 21.728	0.905	0.114	0.373
Quality of life	QoL − AD	32.25 ± 6.38	33.37 ± 7.89	− 1.125	− 4.137 to 1.887	0.780	0.100	0.406

Avg average, SD standard deviation, VF verbal fluency, CDT clock drawing test, VMPT verbal memory process test, BNT boston naming test, BFRT benton facial recognition test, MMSE mini-mental state examination, NPI neuropsychiatric inventory, FBI frontal behavioral inventory, BBT berg balance test, TUG timed-up and go test, QoL-AD quality of life -Alzheimer’s disease. *p < 0.05

Between group changes are shown in Table 5. At the baseline, statistically significant difference was found in VMPT recognition in favor of the control group (p < 0.05). There were statistically significant differences in FBI and BBT in the AE group after treatment (p < 0.05). Significant differences were found in the NPI in favor of the rTMS group and BBT in favor of the AE Group in Time*Group interaction and post-Hoc analysis (p < 0.05).

Table 5.

Between groups findings

	Pre—treatment					Post—treatment					Difference
	rTMS group (n = 10)	AE group (n = 9)	Control group (n = 8)	F	p value	rTMS group (n = 10)	AE group (n = 9)	Control group (n = 8)	F	p value	Mean difference (confidence of interval)	F	Effect size (Cohen’s d)	p value
	Avg ± SD	Avg ± SD	Avg ± SD			Avg ± SD	Avg ± SD	Avg ± SD
Attention
Digit span forward	4.00 ± 1.94	3.55 ± 1.58	4.25 ± 0.88	0.428	0.657	3.80 ± 1.93	4.33 ± 1.50	4.25 ± 0.88	0.329	0.723	− 0.193 (− 0.775 to 0.390)	1.154	0.088	0.332
Digit span backward	2.50 ± 1.64	2.11 ± 1.16	2.12 ± 1.12	0.250	0.781	2.30 ± 1.25	2.55 ± 0.72	3.00 ± 0.75	1.181	0.324	− .373 (− 0.761 to 0.015)	2.804	0.189	0.080
Executive functions
Stroop	29.35 ± 18.31	22.50 ± 19.13	17.38 ± 16.90	0.984	0.389	25.83 ± 19.82	17.25 ± 13.82	16.15 ± 13.85	0.985	0.388	3.336 (− 0.572 to 7.244)	0.358	0.029	0.703
VF—fruit name pairs	4.49 ± 2.62	4.11 ± 1.61	4.87 ± 2.90	0.210	0.812	3.70 ± 2.58	4.22 ± 1.39	4.12 ± 1.55	0.191	0.828	0.478 (− 0.243 to 1.200)	0.723	0.057	0.496
Categorical Fluency	9.70 ± 3.46	10.55 ± 3.60	11.17 ± 6.28	0.240	0.785	9.80 ± 3.91	13.11 ± 4.75	9.75 ± 3.95	1.870	0.176	− 0.409 (− 1.730 to 0.911)	3.162	0.209	0.060
Phonemic Fluency	12.87 ± 5.50	12.55 ± 7.76	13.23 ± 13.98	0.011	0.989	10.49 ± 5.58	13.66 ± 8.60	15.24 ± 13.62	0.591	0.562	− 0.248 (− 1.988 to 1.492)	2.624	0.179	0.093
CDT	1.10 ± 0.87	0.88 ± 0.92	1.00 ± 1.30	0.098	0.907	2.47 ± 1.59	1.50 ± 1.40	1.46 ± 1.76	1.214	0.315	− 0.816 (− 1.228 to  − 0.405)	2.064	0.147	0.149
Memory
Visual immediate recall	1.92 ± 1.36	3.55 ± 3.57	2.50 ± 2.07	1.029	0.373	1.30 ± 0.83	2.44 ± 2.29	2.62 ± 2.55	1.213	0.315	0.535 (− 0.096 to 1.167)	1.300	0.098	0.291
Visual delayed recall	0.76 ± 0.78	0.55 ± 1.01	1.12 ± 2.03	0.394	0.679	0.50 ± 0.97	0.88 ± 1.36	1.75 ± 2.37	1.357	0.277	− 0.233 (− 0.596 to 0.130)	2.246	0.158	0.128
Visual recognition	0.50 ± 0.47	0.55 ± 0.88	0.50 ± 0.75	0.017	0.983	0.79 ± 1.22	0.88 ± 1.05	1.25 ± 1.03	0.395	0.678	− 0.458 (− 0.934 to 0.017)	0.383	0.031	0.686
VMPT Total	47.00 ± 20.3	44.88 ± 9.73	47.25 ± 22.15	0.045	0.956	45.10 ± 14.54	47.66 ± 15.46	46.37 ± 14.96	0.070	0.933	− 0.001 (− 4.520 to 4.518)	0.435	0.035	0.652
VMPT immediate recall	2.50 ± 1.58	2.66 ± 1.41	2.50 ± 1.30	0.040	0.961	2.90 ± 1.44	2.77 ± 0.97	2.87 ± 1.35	0.023	0.977	− 0.295 (− 0.953 to 0.362)	0.086	0.007	0.918
VMPT delayed recall	1.00 ± 2.49	0.11 ± 0.33	2.00 ± 2.87	1.578	0.227	0.70 ± 1.88	1.66 ± 3.08	2.25 ± 3.15	0.753	0.482	− 0.502 (− 1.629 to 0.626)	1.058	0.081	0.363
VMPT recognition	5.00 ± 3.91	7.44 ± 2.92	9.62 ± 2.97	4.296	0.025*	6.40 ± 3.83	6.00 ± 1.93	7.75 ± 2.71	0.791	0.465	0.640 (− 0.656 to 1.936)	2.804	0.189	0.080
Language
B NT	17.78 ± 5.15	18.11 ± 6.67	17.50 ± 4.03	0.027	0.973	18.64 ± 6.09	18.88 ± 4.78	17.75 ± 4.83	0.106	0.899	− 0.631 (− 1.887 to 0.625)	0.095	0.008	0.909
Visual spatial functions
B FRT	40.11 ± 4.66	37.33 ± 5.72	36.37 ± 7.40	0.994	0.385	40.47 ± 2.89	36.77 ± 3.96	39.25 ± 6.36	1.644	0.214	− 0.893 (− 2.876 to 1.090)	1.071	0.082	0.359
General cognition
MMSE	17.50 ± 4.50	18.66 ± 2.64	19.62 ± 4.77	0.613	0.550	19.50 ± 5.70	20.22 ± 4.23	21.37 ± 4.83	0.314	0.733	− 1.769 (− 3.192 to  − 0.345)	0.037	0.003	0.964
Behavior
NPI	41.00 ± 21.60	16.55 ± 22.91	13.25 ± 4.94	9.334	0.001*	21.90 ± 13.60	15.00 ± 19.75	10.75 ± 12.08	1.188	0.322	4.385 (0.149 to 8.621)	14.811	0.552	0.000*
FBI	18.20 ± 11.82	19.33 ± 8.41	19.75 ± 13.33	0.046	0.955	7.80 ± 3.76	13.88 ± 12.08	19.62 ± 6.94	4.594	0.020*	5.323 (1.224 to 9.423)	2.225	0.156	0.130
Balance
BBT	45.40 ± 9.55	42.77 ± 7.61	42.62 ± 10.95	0.259	0.774	46.90 ± 7.93	50.33 ± 4.79	33.62 ± 20.98	4.089	0.030*	− 0.019 (− 3.937 to 3.900)	6.111	0.337	0.007*
Mobility
TUG	27.10 ± 11.99	20.22 ± 2.58	26.37 ± 14.10	1.139	0.337	24.80 ± 9.50	18.66 ± 2.00	41.00 ± 43.72	1.883	0.174	− 3.590 (− 13.103 to 5.923)	1.361	0.102	0.275
Quality of life
QoL-AD	31.90 ± 8.25	33.77 ± 3.38	32.25 ± 6.38	0.222	0.802	34.90 ± 8.03	35.00 ± 3.53	33.37 ± 7.89	0.150	0.861	− 1.782 (− 3.395 to  − 0.170)	0.641	0.051	0.535

rTMS repetitive transcranial magnetic stimulation, AE aerobic exercise, Avg average, SD standard deviation, CDT clock drawing test, VMPT verbal memory process test, BNT Boston naming test, BFRT benton facial recognition test, MMSE mini-mental stateexamination test, NPI neuropsychiatric inventory, FBI frontal behavioral inventory, BBT berg balance test, TUG timed-up and go test, QoL-AD quality of life-Alzheimer’s disease. *p < 0.05

There was no statistically significant difference in the AE group and control group in the rs-fMRI analysis (p > 0.05). There were significant changes in activation levels on lingual fusiform gyrus, middle temporal gyrus, precentral gyrus, and middle frontal gyrus in the rTMS group at the baseline (p < 0.03) (Fig. 2, 3, 4, 5). It was found that after intervention, there was a statistically significant increase in activation on middle temporal gyrus, intra calcarine, central opercular cortex, superior parietal lobule, and paracingulate cortex in DMN the rTMS group (p < 0.05) (Fig. 6, 7, 8, 9, 10). The clusters of rs-fMRI activity are shown in Table 6.

Fig. 2.

Lingual fusiform

Fig. 3.

Middle temporal gyrus

Fig. 4.

Precentral gyrus

Fig. 5.

Middle frontal gyrus

Fig. 6.

Middle temporal gyrus

Fig. 7.

Intra Calcarine Cortex

Fig. 9.

Superior Parietal Lobule

Fig. 10.

Paracingulate Gyrus

Table 6.

Clusters of rs-fMRI activity in the rTMS Group

Brain region	Activity area	Cluster index	Voxels	1-p- MAX	1-p- MAX X (vox)	1-p- MAX Y (vox)	1-p- MAX Z (vox)	1-p- COG X (vox)	1-p- COG Y (vox)	1-p- COG Z (vox)	COPE MAX	COPE MAX X (vox)	COPE MAX Y (vox)	COPE MAX Z (vox)	COPE MEAN
Pre—treatment
Lingual fusiform gyrus	Lingual gyrus	1	2	0.961	33	17	20	33	16.5	20	7.46	33	17	20	7.14
	Temporal occipital fusiform cortex	2	4	0.982	42	24	17	42.5	24	17	8.89	42	24	17	7.45
Middle temporal gyrus	Middle temporal gyrus	1	10	0.988	14	24	27	13.8	23.2	27	7.94	14	24	27	6.44
Precentral gyrus	Middle temporal gyrus	1	3	0.977	46	29	23	45.7	29.3	23.7	8.41	46	29	23	7.85
	Precentral gyrus	3	7	0.985	14	43	36	13.7	42.6	36.9	8.35	14	43	36	7
Middle frontal gyrus	Cingulate gyrus, anterior division	1	2	0.991	31	41	34	31	41.5	34	9.84	31	41	34	9.25
	Middle frontal gyrus	2	6	0.967	44	44	36	43.5	44.7	36.2	6.61	44	44	36	6.02
Post—treatment
Middle temporal gyrus	Middle temporal gyrus, posterior	1	3	0.981	49	38	16	49	38.3	16.7	8.26	49	38	16	7.57
Intra calcarine	Right intra calcarine cortex	3	7	0.968	26	17	27	25.7	17.3	27.1	6.08	26	17	27	5.17
	Left intra calcarine cortex	5	41	0.996	32	15	26	32.8	15.4	26.5	8.66	34	15	27	5.51
Central opercular cortex	Left hippocampus	1	1	0.959	40	37	17	40	37	17	7.29	40	37	17	7.29
	Inferior temporal gyrus, posterior	2	1	0.956	13	33	16	13	33	16	7.3	13	33	16	7.3
	Temporal fusiform cortex,	4	2	0.969	41	30	18	41	29.5	18	7.51	41	30	18	7.18
	Middle temporal gyrus,	5	3	0.965	48	22	22	48	22.3	22.7	7.32	48	22	22	6.83
	Lateral occipital cortex, inferior	6	41	0.984	13	17	28	14.9	17.5	28.1	7.86	15	21	29	4.94
Superior parietal lobule	Supramarginal gyrus, anterior	1	1	0.951	43	33	38	43	33	38	0.951	43	33	38	0.951
	Supramarginal gyrus, anterior	2	1	0.961	47	30	38	47	30	38	0.961	47	30	38	0.961
	Lateral occipital cortex, superior	3	1	0.954	38	18	36	38	18	36	0.954	38	18	36	0.954
	Lateral occipital cortex, superior	4	3	0.963	39	18	33	39	17.3	33.3	0.963	39	18	33	0.959
	Postcentral gyrus, superior parietal lobule	5	28	0.981	42	29	38	41.8	29.9	37.8	0.981	42	29	38	0.963
	Supramarginal gyrus, posterior	6	49	0.994	46	26	36	47.8	28.3	33.9	0.994	46	26	36	0.968
Paracingulate cortex	Middle frontal gyrus	1	1	0.963	44	50	37	44	50	37	0.963	44	50	37	0.963
	Precentral gyrus	3	110	0.999	47	44	27	47.3	45.3	24.5	0.999	47	44	27	0.971
	Central opercular cortex, heschl's gyrus	4	133	0.986	44	36	28	43.8	36.5	24	0.986	44	36	28	0.963

Discussion

We investigated the efficacy of rTMS and AE in addition to PT in patients with AD. Over the course of two weeks, high frequency rTMS treatment over bilateral dlPFC improved executive functions, behavioral status, and quality of life. In addition to the expected clinical effects with medication in AD, structured moderate-intensity AE for 2 weeks provided significant improvements in balance and functional mobility. Memory and behavioral status of the participants in the control group who did not receive any intervention and only continued their PT was improved. Considering the Time*Group interactions, NPI in the rTMS group and BBT in the AE group were significantly improved. Finally, rs-fMRI analysis showed significant increase in the activation levels in the rTMS group in the nodes of DMN.

Mood and behavioral symptoms (e.g. apathy, anxiety, and agitation) are particularly common in AD, affecting nearly 80% of a typical clinical population (Clerici et al. 2012). AChEIs (Donepezil, Rivastigmine, Galantamine) may have a positive impact on aggressive behavior in AD patients (Forester et al. 2007). While they all elevate acetylcholine (ACh) levels in the brain, they differ substantially in mechanism of action, inhibitory potency, brain selectivity, and metabolism (Gauthier et al. 2007). The interhemispheric connectivity of different brain regions is affected differently by Donepezil treatment. In their study, Zaidel et al. found an increase in functional connectivity in bilateral dlPFC after treatment with 5 mg/day Donepezil for 8 weeks in 11 patients with mild AD (Zaidel et al. 2012). The dlPFC is especially crucial for cognitive control, and AChEIs' effects on attention and executive function may actually improve cognition. Holmes et al., found an improvement in agitated behavior after treatment with Rivastigmine for 4 weeks in 28 patients with severe AD (Holmes et al. 2007). Kavanagh et al., found in their study that Galantamine using in 6 months reduces behavioral symptoms (NPI) in patients with mild to moderate AD (Kavanagh et al. 2011). McGeown et al., found increased behavioral scores after 10 mg/day Donepezil over 20 weeks in 12 patients with AD (McGeown et al. 2010). In moderate-to-severe AD patients, glutamate regulators (Memantine) have been shown to improve agitation and aggression (Wilcock et al. 2008). Drug therapy studies in Alzheimer's patients include at least 4 weeks of treatment. In our study, participants performed approved pharmacological protocols for the treatment of AD, including AChEIs (Donepezil, Rivastigmine, Galantamine) and glutamate regulators (Memantine) (Medicines Agency 2018). Brassen et al., applied 2-weeks AChEIs treatment in patients with AD and found an increase in The Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-Cog) memory score (Brassen and Adler 2003). Sole-Padulles et al. administered daily treatment of Donepezil during 3 months in patients with AD (Solé-Padullés et al. 2013). They found that treatment with Donepezil is associated with enhanced connectivity of the medial temporal regions at rest and greater brain efficiency during a cognitive task. We found improved behavioral scores in NPI and visual delayed recall after PT in the control group. This improvement in behavioral and visual long-term memory can be explained by the medial temporal region activation of the short-term effect of PT. According to Chaudhary et al., AChEIs therapy increased cerebral blood flow in the lateral temporal lobe, posterior and anterior cingulate gyrus (including white matter), and prefrontal regions, independent of regional brain tissue volume (Chaudhary et al. 2013). We assume that boosting the activation of the prefrontal region with the prescribed medications we administered to Alzheimer's patients resulted in an improved behavioral symptom. The resting-phase studies consistently suggested that each treatment resulted in increased activation in multiple brain regions (such as dlPFC) and increased functional connectivity in multiple networks (such as DMN), and that the brain functional responses were often correlated with cognition (Guo et al. 2018). The improvement in behavioral functions was not reflected in the rs-fMRI results in our study. Due to the small sample size in the control group and the short duration of PT, we consider there was no functional brain alterations.

AE is a potential treatment option to slow cognitive decline and may improve cognition in adults with cognitive impairment such as AD. AE have been found to lead to improved executive functions, increased prefrontal brain volume and greater task-related activation of key nodes of the attentional system, as well as improved memory performance and increased hippocampal volume (Li et al. 2017). Yu et al., showed that supervised, moderate-intensity cycling for 20–50 min, 3 times a week over six months may reduce decline in global cognition in older adults with mild-to-moderate AD (Yu et al. 2021a). Yang et al., found that cycling training at 70% of maximal intensity for 40 min/d, 3 d/wk for 3 months can improve cognitive function in patients with mild AD (Yang et al. 2015). In our study, individuals received the structured moderate-intensity AE for 50 min sessions, 5 days a week over 2 weeks. We found increases in all measures in terms of cognitive functions, but they were not significant. There could be a variety of reasons for this result, small sample size making interpretation difficult. Whereas fitness improves after only a few weeks of training, the temporal evolution of the dose–response connection between training and potential cognitive benefits is unknown, and a longer intervention period may be required in theory. Another reason for the lack of significant differences in exercise-induced cognitive development, we assume, is that the intervention periods were too short to produce a wide range of scores and “downstream” cognitive effects that could be evaluated consistently in different domains. AE has been shown to have cardiovascular and metabolic benefits in individuals with mild AD and is likely to delay physical function and independence deterioration (Sobol et al. 2016). Gait impairment, including a decreased walking speed and greater step-to-step variability, is observed in the early stages of AD. Changes in gait pattern and increased risk in walking speed have been related to a deficiency in executive function and the ability to divide attention as well as interpret and integrate sensory information in people with AD. For example, TUG tasks might be useful for detecting the early signs of cognitive impairment (Cedervall et al. 2012). A minimum of moderate-intensity aerobic activity has been shown to improve brain function in inactive older people and those at risk of developing AD. Furthermore, research indicates that regular physical activity is crucial for people with AD at all stages, both to maintain motor function and to benefit from beneficial effects on cognitive function, behavioral problems, sleep, and well-being (Cedervall et al. 2012). Nagy et al. observed improvement in BBT scores of individuals with AD who performed moderate-intensity AE using the treadmill for 45–60 min, 3 days a week over 3 months (Nagy et al. 2021). Canocini et al. reported improvement in BBT scores of individuals with AD who exercised for 60 min three times a week over a 6-month period (Canonici et al. 2012). De Andrade et al., published improvement in TUG scores of individuals with AD who performed multimodal exercise intervention 1-h sessions three times a week for 16 weeks (De Andrade et al. 2013). Aguiar et al., resulted in improvement in TUG scores of individuals with AD who performed aerobic, flexibility, strength, and balance movements, twice a week for 6 months in addition to Rivastigmine therapy (Aguiar et al. 2014). We found improvements in BBT and TUG scores in the AE group. The outstanding difference of our study was that, while there was no significant improvement in cognitive functioning, there was a significant change in functional mobility and balance scores. This is due to the fact that the AE program we use only promotes a small amount of neuroplasticity in terms of functionality throughout the period. Exercise improves the structure and function of the brain, increases cognitive reserve, and reduces the neuropathologic alterations associated with AD (Yu and Kolanowski 2009). The hippocampus, located in the medial temporal lobe, is a significant location of neuroplasticity that is sensitive to the impacts of physical activity (Sinha et al. 2021). In our study, it was observed that the hippocampus of the individuals improved at the trend level after AE, but the result was not statistically significant. Having a trend level of activation can provide us with information regarding the neuroplasticity of exercise in the short term, even after only 2 weeks of AE. Even though regular AE caused a significant increase in cognitive test scores, the neuroplastic change in the short term was not reflected in the rs-fMRI results. The insignificance of our findings can be attributed to the neurodegenerative effects of AD and the fact that we were evaluating after such a short period of time.

Executive functions are affected from the onset of AD due to prefrontal cortex degeneration. Inhibitory abilities, attentional abilities, and visuospatial abilities, in particular, would be significantly affected. In a randomized controlled research, Ahmed et al. found that a high frequency rTMS significantly improved cognition in Alzheimer's patients with mild to severe cognitive impairment (Ahmed et al. 2012). dlPFC has a role in working memory even though it is related in executive functions of the rest of the brain, such as planning, organization, decision-making, and information storage and retrieval (Rutherford et al. 2015). AD affects the plasticity of the dlPFC, which is linked to poor working memory. In patients with AD, high frequency rTMS is shown to have facilitative effects on the bilateral dlPFC. Kumar et al., support the use of dlPFC plasticity as a measure of dlPFC function and as a potential therapeutic target to improve dlPFC function and working memory in patients with AD (Kumar et al. 2017). After two weeks of left lateral parietal rTMS stimulation, Velioglu et al. observed remarkably improved CDT scores, which were associated with elevated peripheral Brain Derived Neurotrophic Factor levels and decreased oxidant status (Velioglu et al. 2021). In our study we found significant improvements in CDT in the rTMS group. Poor CDT performance has been associated with alterations in frontal and parietal activation in dementia patients (Velioglu et al. 2021). In other words, our research has shown that even in the short term, an increase in CDT score can influence the executive functions of Alzheimer's patients. We assume that this finding is due to bilateral dlPFC stimulation, which plays an important role in executive functions. Although our study demonstrated the short-term effects of high frequency rTMS treatment applied to bilateral dlPFC in addition to PT, studies with long-term evaluation and follow-up are required to better understand the LTP mechanism and see the long-term effects of neuroplasticity. The dlPFC is associated with mood regulation as well (Heath et al. 2018). High-frequency rTMS administered to the right dlPFC area has been demonstrated to reduce anxiety symptoms, which are much higher in patients with mild to moderate AD than in age-matched healthy controls (Moussavi et al. 2021). Another study found that high frequency rTMS applied to the right or left dlPFC for up to three months improved behavior (Ahmed et al. 2012). In our study, we found that there was improvement in NPI scores. We assume that the improvement in behavioral status following rTMS treatment is a result of both improved executive function and the relationship between dlPFC and mood. The rate at which arterial blood is delivered to the tissue capillary bed is referred to as cerebral blood flow (CBF). The integrity of the cerebral vascular system is dependent on apolipoprotein E (ApoE) 4. People with one ApoE 4 allele have a two- to three-fold increased risk of AD, while those with two ApoE 4 alleles have a 12-fold increased risk. The ApoE 4 genotype and alterations in CBF in medial temporal lobe which is related to DMN are both associated with a higher risk of cognitive impairment, demonstrating that cerebrovascular pathways are crucial in maintaining cognitive performance. Hyperperfusion is linked to poor memory performance in ApoE 4 carriers, which could indicate vascular and/or cellular malfunction. Hyperperfusion appears to sustain memory performance in patients who do not have a hereditary risk of Alzheimer's disease, but it indicates that heightened CBF doesn't quite maintain memory performance in people who have ApoE 4 (Wang et al. 2019). The left dlPFC may act as an important hub for network integration in cognition and behavior, which is impaired in AD (Alcalá-Lozano et al. 2018). Consequently, applying rTMS in this region may improve network activity and integration, which could be connected to clinical improvement. The connections between the anterior part of the middle cingulate and the dlPFC, which is involved in cognition, indicate that this region may be involved in the executive function path utilized during CDT (Yu et al. 2011). Cui et al., showed that rTMS-induced hypoconnectivity within DMN is associated with clinical cognitive improvements (Cui et al. 2019). The medial septum provides cholinergic input to the hippocampal pyramidal neurons that are damaged in AD. When acetylcholine is insufficient, numerical results from simulations of cholinergic potentiated M-current and calcium ion-activated potassium ion current demonstrate that relative theta band power increases dramatically and firing rate reduces noticeably. Similarly, as beta-amyloid accumulates, the relative theta band power increases while the firing rate reduces dramatically when imitating beta-amyloid increased delay rectification potassium ion current. The etiology of AD is complicated by acetylcholine shortage and beta-amyloid accumulation (Jiang et al. 2020). Decreasing functional connectivity in beta bands is related to cognitive function (Utianski et al. 2014). In addition, dysfunction of the gamma- aminobutyric acid (GABA) signaling system leads to lower oscillatory gamma band activity, according to a study of neurotransmitter alterations in AD (Calvo-Flores Guzmán et al. 2018). Nobukawa et al. stated that the fractality of temporal scale-specific electroencephalography(EEG) signals for faster frequency ranges (the beta and gamma range) decreased in Alzheimer's patients compared to healthy individuals, and this decreased fractality was associated with cognitive decline (Nobukawa et al. 2019). In our study, there were significant changes in activation levels on lingual fusiform gyrus, middle temporal gyrus, precentral gyrus, and middle frontal gyrus in the rTMS group at the baseline. Activation decreases in these areas are also consistent with the clinical course of AD. In our clinical results in which we evaluated cognitive functions, we think that the impairments in attention, executive function and memory tests can be explained by the decrease in these areas. In our study, there were also significant increases in activation on middle temporal gyrus, intra calcarine, central opercular cortex, superior parietal lobule, and paracingulate cortex the rTMS group after treatment. In rs-fMRI analyses, we found that these areas were associated with DMN because it is medially anchored to the posterior cingulate cortex/precuneus and ventromedial prefrontal cortex, and to the bilateral parietal (inferior parietal lobule—IPL, which include the angular and inferior parietal gyri), temporal (lateral temporal cortex and hippocampus), and frontal cortex (dlPFC, roughly corresponding to the superior frontal gyrus) (Bagattini et al. 2021). The DMN appears to play a central role in this dynamic functional integration of brain networks at rest. The anterior and posterior components of the DMN have been revealed to be less functionally connected, and patients with AD had increased aDMN connectivity despite maintaining decreased pDMN connectivity (Klaassens et al. 2017). The right dlPFC could be involved in the activation of the posterior DMN, which could be connected to wandering episodes (Klaassens et al. 2017). In line with our results and this information, we observed that the activation in the areas that decreased in line with the clinical findings at the beginning of the disease showed a significant increase after rTMS application. We think that the effect of rTMS application on DMN may affect the clinical findings of AD. We are of the opinion that after rTMS application to the bilateral dlPFC, the clinical symptoms suffered by Alzheimer's patients may improve thanks to the increase in DMN-related areas. We assume that bilateral dlPFC stimulation produces significant activation in the DMN in this study, but much more studies are needed to determine which component is involved.

Limitations

Our first limitation was the duration of the intervention. Two weeks of rTMS can be accepted as the minimum length of intervention. However, considering the clinical situations that patients with AD would have difficulty adhering to the long duration of interventions, we planned to consider the short-term effects of rTMS to prevent any further drop-outs due to lack of follow up. Our second limitation was the small sample size. Problems in the transportation, difficulties in reaching the hospital research facility due to challenges in a megacity and hesitation of the AD patients to accept MRI scanning twice led to the small sample size. Our third limitation was that we did not plan a "seed to seed" analysis in rs-fMRI analysis.

Conclusions

In conclusion, high frequency rTMS treatment may be effective on executive functions and behavioral status, while moderate-intensity AE may be effective on balance and mobility in patients with AD. In addition, high frequency rTMS treatment can lead to activation in DMN in rs-fMRI. Importantly, increasing activation in DMN can be associated with executive functions and behavioral status. Our findings indicate that therapeutic methods that appear to be as an alternative in AD are in fact effective in executive functions and behavioral profiles. Considering rTMS, AE, and PT treatments have unique curative effects in AD, future studies should investigate the effects of a "multimodal" therapeutic approach that involves a combination of high frequency rTMS, structured moderate-intensity AE, and PT in AD.

Author contribution

MB: The conception and design of the study, acquisition of the data, analysis, and interpretation of data, drafting the manuscript, revising the manuscript critically for important intellectual content, final approval of the version to be submitted. ZB: The conception and design of the study, analysis, and interpretation of data, revising the manuscript critically for important intellectual content, final approval of the version to be submitted. LH: The conception and design of the study, revising the article critically for important intellectual content, final approval of the version to be submitted.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declarations

Conflict of interest

None.