Functional MRI-guided repetitive transcranial magnetic stimulation in cognitive impairment in cerebral small vessel disease

Full Text

Introduction

Cerebrovascular disease is the second most common cause of cognitive impairment after Alzheimer's disease [1]. Cerebral small vessel disease (CSVD) is a leading cause of vascular cognitive impairment [2, 3]. Currently, treatment options for vascular cognitive impairment are limited [4].

Mild cognitive impairment (MCI) is a decline in cognitive function that exceeds that of normal aging but does not meet the clinical criteria for dementia [5]. In the population aged 65 years and older, the MCI prevalence is equal to or greater than that of dementia and may reach 42% [6]. However, MCI not only deteriorates quality of life [7], but is also an independent risk factor of dementia; 20–30% of cases of mild cognitive impairment progress to dementia within 6 years [1].

Repetitive transcranial magnetic stimulation (rTMS) is a non-invasive brain stimulation method increasingly used to treat neurological disorders [8]. The mechanism of the therapeutic rTMS effect is usually associated with TMS-induced synaptic plasticity [9].

rTMS is best studied for treatment of Alzheimer's cognitive impairment. A meta-analysis included 12 stu-dies (n = 231) with different rTMS protocols, including multi-target stimulation and stimulation of the dorsolateral prefrontal cortex (DLPFC). A statistically significant improvement in cognitive function was found in the active stimulation group compared to the control group. The effect was greater in milder forms of Alzheimer's disease [10]. The most compelling evidence for efficacy was obtained using a multi-target rTMS protocol combined with target-specific cognitive training, called rTMS-COG. In the international expert recommendations, this protocol was assigned a level of evidence C [8]. A randomized, placebo-controlled study showed a statistically significant effect of high-frequency rTMS of the left DLPFC on memory in elderly patients with amnestic MCI [11].

A relatively large number of studies evaluated the use of TMS in Alzheimer's disease and amnestic MCI (as a predementia stage of Alzheimer's disease). The use of TMS in vascular cognitive impairment has been less studied. Most studies evaluated the diagnostic use of TMS, while only few studies evaluated the therapeutic effect of rTMS [12–14]. Two studies investigated the effect of a single rTMS session on the left DLPFC. I. Rektorova et al. showed that executive functions (EF) measured with the Stroop test improved after 1 rTMS session [12]. S. Sedlackova et al. found no statistically significant differences between stimulation of DLPFC and M1 (control target) [13]. One of more recent studies showed the effect of supplementary motor area rTMS on cognitive functions in CSVD patients with MCI [14]. There are no studies on the efficacy of multiple sessions of the left DLPFC rTMS in CSVD patients.

It is important to explore the potential of personalized targets for rTMS due to the structural and functional he- terogeneity and interindividual anatomical variability of the cerebral cortex. One way of personalization is to use structural neuroimaging to construct a 3D model of the head and then to overlay functional neuroimaging data such as resting-state fMRI and task fMRI [15]. For example, functional connectivity-based personalization is being actively studied in depression [16-19]. Task fMRI-guided navigated TMS is actively used in studies with healthy volunteers [20], but is rarely applied in clinical practice (J.P. Szaflarski et al. reported its use in patients with post-stroke aphasia [21]).

The aim of this study was to evaluate the immediate and delayed effects of 10 sessions of fMRI-guided high-frequency rTMS of the left DLPFC on cognitive functions in CSVD patients with MCI.

Materials and methods

This randomized, double-blind, placebo-controlled, parallel study was conducted at the Research Center of Neurology.

Inclusion criteria:

• Age of 45 to 80 years,
• CSVD diagnosed according to STRIVE guidelines (2013) [22],
• MCI diagnosed using VASCOG criteria [23],
• No changes in treatment for cognitive impairment or use of other central nervous system agents for 1 month before rTMS, during rTMS, and for 3 months after the stimulation.

Exclusion criteria:

• Contraindications to rTMS and/or MRI,
• History of stroke (except lacunar one),
• History of epilepsy or epileptiform discharges on EEG,
• Use of antidepressants and antipsychotics,
• Decompensation of severe somatic disease,
• Mental disorder or alcohol and/or drug abuse.

The study was conducted in accordance with the Declaration of Helsinki and approved by the local ethics committee of the Federal State Budgetary Scientific Institution “Research Center of Neurology” (Protocol No. 12-4/16 dated 14 December 2016). Prior to enrollment, all patients signed an informed consent form.

Design of the study

After enrollment, all patients were randomized using sealed envelope method in a 2:1 ratio to active stimulation (left DLPFC) or control stimulation (vertex area).

Prior to stimulation (T0), cognitive functions were assessed, and patients underwent task fMRI. Subsequently, 10 rTMS sessions (5 sessions per week) were performed. Cognitive functions were reassessed immediately (T1) and 3 months (T2) after the rTMS treatment. The investigator performing the initial clinical assessment for eligibility and repeated cognitive assessments at each stage was blinded to the stimulation protocols.

Cognitive function tests

The Montreal Cognitive Assessment Test (MoCA) was used to assess general cognitive status. Additional tests were used to assess EF and visuospatial functions [24]. EF was assessed using the Trail Making Test (TMT), with the TMT-A subtest assessing psychomotor speed and the TMT-B subtest assessing divided attention, and the Tower of London test assessing cognitive control. Visuospatial functions were assessed by copying the Rey–Osterrith Complex Figure Test (ROCF), and nonverbal memory was assessed by delayed recall of the CF 30 min after copying [25].

Neuroimaging

All patients had structural MRI and task fMRI on a 3T Magnetom Verio (Siemens) for determination of the stimulation target. The anatomical structure was visualized using a 3D-T1 gradient echo with multiplanar reconstruction (3D-T1 MPR) sequence consisting of 176 sagittal slices (TR = 1940 ms; TE = 308 ms; interslice interval = 0.5 mm; field of view = 250 mm; matrix = 256×256; slice thickness = 1 mm). The paradigm fMRI used an axial T2* gradient echo sequence (TR = 3000 ms; TE = 30 ms; slice thickness = 3 mm).

The paradigm had a block design and consisted of 4 acti- vation blocks and 4 rest blocks, with each block lasting 30 s. Before the start and at the end of each activation block, subjects were verbally instructed to start or stop the task. They were asked to count silently starting from 1 and skipping multiples of 3. When the next activation block began, the patient repeated counting from 1 [26]. Before the MRI, the patient was trained to perform the task outside the scanner under the supervision of the investigator.

SPM12 for MATLAB R2018a (Mathworks) was used for preprocessing and statistical analysis of individual fMRI data to determine the stimulation target¹. The first level analysis used a regressor with a value of 1 in the activation block, 0 in the rest block, and a T-contrast corresponding to the regressor with a voxel-wise significance threshold of 0.001 without correction. The data obtained were co-registered with the structural data and uploaded to the navigation system. The rTMS target was positioned within the left DLPFC (corresponding to the middle frontal gyrus) according to the visually detectable maximum activation.

Transcranial magnetic stimulation

The navigation system of the NBS eXimia Nexstim stimulator (Nexstim Plc) was used for fMRI-guided navigation rTMS. The target for active stimulation was located in the DLPFC while the control group used the vertex detected by visible anatomical landmarks. For rTMS, a Magstim Rapid 2 stimulator (Magstim Company Ltd.) was used with a figure-of-eight coil calibrated for na- vigation. Stimulation was performed at an intensity of 100% of the resting motor threshold of m. abductor pollicis brevis, determined using the Rossini–Rothwell algorithm [27]. In both groups, high-frequency rTMS was performed with a stimulation frequency of 20 Hz, 2-se-cond trains with a 28-second intertrain interval, 2400 stimuli per session, for a total of 10 sessions. Patients completed standardized TMS tolerability questionnaires (adverse events (AEs) during stimulation and within 24 hours after the stimulation).

Statistical analysis

MATLAB R2018a (Mathworks) was used for statistical analysis. Normal distribution was tested using the Shapiro–Wilk test. The data were non-Gaussian distributed, so non-parametric methods were used. The Friedman test was used to determine changes in cognitive test scores between the 3 intra-group measurements. The Wilcoxon test was used for paired intra-group comparisons. The Mann–Whitney test was used to compare quantitative characteristics between groups (comparison of effects). The Fisher's test (for binary characteristics) and the Fisher–Freeman–Halton test (for 3 levels of the Fazekas scale) were used for comparing qualitative parameters. Changes were considered significant at p < 0.05.

Results

Patients

A total of 96 patients were screened for the study, of which 30 patients were included into the final analysis (Figure 1). There were no statistically significant differences in gender, age, severity of neurological symptoms, or cognitive test scores between two groups (Table 1).

 

Fig. 1. Flow chart.

 

Table 1. Demographic, clinical, and neuroimaging characteristics of patients

Parameter	Active group (n = 20)	Control group (n = 10)	p
Sex (male), n (%)	10 (50%)	2 (20%)	0.24
Age, years, Ме [Q1; Q3]	60 [57.5; 66.5]	58 [57.5; 69.0]	0.94
Gait disorder, n (%)	14 (70%)	6 (60%)	0.69
Pseudobulbar palsy, n (%)	5 (25%)	2 (20%)	1.00
White matter hyperintensity (Fazekas scale)			0.73
Fazekas I	2 (10%)	0
Fazekas II	9 (45%)	4 (40%)
Fazekas III	9 (45%)	6 (60%)
White matter lacunes, n (%)	12 (60%)	4 (40%)	0.44
Lacunes in subcortical structures, n (%)	6 (30%)	5 (50%)	0.43
Brainstem lacunes, n (%)	9 (45%)	4 (40%)	1.00
Juxtacortical microbleeds, n (%)	5 (25%)	3 (30%)	1.00
Subcortical microbleeds, n (%)	11 (55%)	3 (30%)	0.26

 

All patients underwent paradigm fMRI with a target identification in the left DLPFC. Figure 2 shows the stimulation target localization for the active group.

 

Fig. 2. Localization of stimulation target within the left DLPFC in the active group.

 

Assessment of the intra-group effect

In the active group, using the Friedman test, a statistically significant effect of time point was reported for MoCA test (p < 0.001), TMT-A (р = 0.012), TMT-B (p < 0.001), the Tower of London (p < 0.001), and delayed CF recall (p = 0.009) test. For CF copying, the effect was not statistically significant (p = 0.929). In the control group, the effect of time point was not statistically significant for all tests (MoCA, р = 0.119; TMT-A, р = 0.368; TMT-B, p = 0.347; the Tower of London, p = 0.187; copying and recall of CF, p = 0.867 and p = 0.792, respectively).

In a pairwise comparison using the Wilcoxon test, a statistically significant improvement was reported immediately after the stimulation and 3 months after the stimulation for the MoCA, TMT-B, and Tower of London tests, while for the TMT-A and delayed CF recall, a statistically significant improvement was reported only immediately after the stimulation, and no significant changes were observed when copying the CF (Table 2). In the control group, no statistically significant changes were observed in any of the tests used, either immediately or 3 months after the stimulation (Table 2).

 

Table 2. Intra-group effect of rTMS, Me [Q₁; Q₃]

Test	Time points	Active group (n = 20)	p	Control group (n = 10)	p
MoCA, score	T0	24 [22; 26]	–	22 [22; 24]	–
	T1	26 [24; 28]	< 0.001	25 [22; 25]	0.094
	T2	26 [19.5; 27]	0.001	23 [21; 24]	0.329
TMT-A, sec	T0	69 [51.0; 91.5]	–	56 [47; 76]	–
	T1	53 [42.5; 72.5]	0.009	57 [52; 86]	0.91
	T2	58 [41; 87.5]	0.08	57 [52; 86]	0.093
TMT-B, sec	T0	162 [126.0; 256.5]	–	168 [135; 243]	–
	T1	138 [106.5; 219.5]	0.007	186 [109; 207]	0.071
	T2	119 [82; 173]	0.032	169 [138; 244]	0.889
The Tower of London Test, score	T0	12.5 [8.75; 16]	–	12 [11; 14].	–
	T1	13 [11.0; 18.5]	0.002	12 [11; 16]	0.083
	T2	14 [12.5; 16.5]	0.044	10 [9; 13]	0.724
Complex Figure, copying, score	T0	32 [27.5; 35]	–	34 [33; 34]	–
	T1	34 [27.5; 36]	0.279	33 [32; 35]	0.656
	T2	33 [28; 36]	0.612	27 [25; 31]	0.380
Complex Figure, delayed recall, score	T0	16 [8; 21.75]	–	15.5 [7; 21.5]	–
	T1	20 [10.25; 26.75]	0.003	17.5 [15.5; 26]	0.102
	T2	16 [9.75; 26.75]	0.690	14 [12; 21]	0.500

Note. T0, baseline testing; T1, testing immediately after the stimulation; T2, testing 3 months after the stimulation; p for T1 and T2 was calculated as a pairwise comparison for T1/T0 and T2/T0, respectively.

 

Comparison of inter-group effects

When comparing the effects between the active and control groups, the active group showed a statistically significant greater effect on the MoCA, TMT-A, TMT-B, and the Tower of London tests immediately and 3 months after the stimulation. For delayed recall of CF, the active group showed a statistically significant greater effect only immediately after the stimulation. For copying CF, no statistically significant difference was found between the active and control groups (Table 3).

 

Table 3. Effect sizes of rTMS (difference in cognitive test scores between T1 and T0. T2 and T0) and comparison of active and control groups. Me [Q₁; Q₃]

Test	Time point	Active group (n = 20)	Control group (n = 10)	p
MoCA	Т1	2 [1; 3]	1 [0; 2]	0.044
	Т2	2 [0; 3]	1 [0; 1]	0.044
TMT-A	Т1	–9 [–22.5; –3]	7 [–4; 12]	0.007
	Т2	–5 [–26.5; 7.5]	7 [–4; 21]	0.041
TMT-B	Т1	–15[–54; 6]	–5 [–10; 9]	0.014
	Т2	–46 [–56.5; –18.5]	–5 [–10; 9]	0.006
The Tower of London Test	Т1	2 [2; 3.5]	1 [0; 1].	0.019
	Т2	3 [2; 5]	1 [0; 3].	0.046
Complex Figure. copying	Т1	0 [–1; 3.5]	0 [–1; 1]	0.275
	Т2	0 [–1.5; 3]	0 [–1; 1]	0.270
Complex Figure. delayed recall	Т1	3 [0.75; 5.5]	1.5 [–0.5; 2.5]	0.043
	Т2	1.5 [–1.75; 4]	0.5 [–0.5; 4]	0.480

Note. T0. baseline testing; T1. testing immediately after the stimulation; T2. testing 3 months after the stimulation. A negative value indicates a decline in the test scores. A positive score indicates an improvement for all tests except TMT. where improvements are indicated by a negative score and declines are indicated by a positive score.

 

Tolerability

Data from 270 sessions were analyzed (180 sessions in the active group and 90 sessions in the control group). No serious adverse events were reported during rTMS in any group. Pain at the site of stimulation was observed in 11.7% of all rTMS sessions in the active group and in 9% of all rTMS sessions in the control group. Pain severity on the numerical rating scale was 1 in 61.9% of cases, 2 in 28.6%, 3 in 4.75%, and 5-6 in 4.75% in the active group and it was 2 in 62.5% of cases, 3 in 25%, 6 in 12.5% in the control group. Other discomfort sensations at the site of stimulation (muscle contractions, burning, itching, etc.) were observed in 15.5% of all sessions in the active group and in 2.2% in the control group. Patients reported somnolence in 11.7% of sessions in the active group and 17% of sessions in the control group. Headache within 24 hours of stimulation was reported in 3.9% of all sessions in the active group and 8.9% in the control group. When comparing the incidence of AEs between the active and control groups, there was no statistically significant difference for headache during (р = 0.539) and after the stimulation (р = 0.08), as for the somnolence (р = 0.26), and the statistically significant difference was found only for non-painful discomfort (р < 0.001).

Discussion

Our paper showed that 10 sessions of fMRI-guided rTMS of the left DLPFC significantly improved cognitive functions in CSVD patients with MCI. The active group showed the significantly greater effect compared to the control group, both at the general cognitive level and in specific domains (EF and non-verbal short-term memory), and this effect persisted across a range of tests for 3 months after the stimulation. The proposed rTMS protocol for MCI had a good safety and tolerability profile.

In our protocol, the left DLPFC was used as a stimulation target. The choice of the target area was based on the available data on the role of this area in vascular cognitive impairment [28, 29], as well as on previous studies showing the efficacy of stimulation of the left DLPFC in cognitive impairment of various etiologies [30]. The activity of the frontal lobes plays a key role in the EF, and in CSVD early decline in EF is observed [28]. Loss of frontal-parie- tal connectivity in white matter involvement is currently considered an important factor in the pathogenesis of cognitive impairment in CSVD [29]. A significant loss of interhemispheric and frontal connectivity in CSVD has been demonstrated by structural connectivity studies [28].

Along with the EF scores, nonverbal memory and overall cognitive level improved. However, no statistically signi- ficant changes in visuospatial functions (CF copying test) were found in our study, which may be due to the predominant role of the posterior parts of the cerebral cortex, such as the parietal and occipital cortex, for this domain [31]. Further studies are required to evaluate the efficacy of rTMS in cortical regions other than DLPFC or multisite TMS for the treatment of visuospatial deficits in MCI.

In our study, we analyzed individual paradigm fMRI data to determine target localization. fMRI-guided TMS is considered one of the potential methods to increase the efficacy of rTMS [32-34]. Comparing different approaches to target detection (using a 10-20% electrode positioning system, using structural MRI, fMRI, or a target in Talairach space), determination of a significant effect of rTMS of the right DLPFС on fMRI reaction time requires a 10-fold smaller sample size compared to using the 10-20% system [33]. A recent meta-analysis showed that fMRI-guided na- vigation for rTMS in healthy volunteers had a higher online effect compared to other methods of target selection [20].

Our choice of paradigm is primarily based on the switching task, but also involves other components of EF and limits the effect of learning [35].

Limitations of target selection using individual paradigm fMRI data include the low reproducibility of fMRI results at the individual level [36]. It is unclear what impact this may have on clinical efficacy. However, we did not compare our method with other methods of target selection. Therefore, we cannot conclude on the advantage of a personalized approach and its appropriateness for real clinical practice.

It should be noted that the effect of rTMS on some of the tests persisted for at least 3 months after treatment. M. Sabbagh et al. reported a positive effect after TMS-COG treatment that was more significant 12 weeks after rTMS compared to that after 7 weeks, which is consistent with the data obtained in this study [37]. Given the progressive nature of cognitive impairment in CSVD, this suggests that rTMS may have an effect on the course of the disease. However, this statement requires further investigation in separate studies. It is also useful to study the effect of repeated rTMS courses or maintenance sessions after the main treatment. The latter approach has been shown to be effective in other conditions, such as depression and pain syndromes [38, 39].

One of the most promising areas for future research is the development of effective combined protocols for rTMS and cognitive training. The potential enhancement of neuromodulation effects by combining it with various methods of cognitive interventions is being actively studied in several neurological and mental disorders [40]. This approach has been shown to be effective in Alzheimer's disease [38] and requires further study in vascular cognitive impairment.

No serious AEs were reported during the rTMS treatment. The AEs reported by patients in the study were mild and did not affect treatment adherence. The profile of AEs was comparable to previously published data [41]. Therefore, our data confirm the safety and good tolerability of rTMS of the left DLPFC in vascular MCI.

Limitations of our study include the small sample size, therefore, our results should be validated in studies with larger samples. The diagnosis of vascular cognitive impairment was based on clinical and neuroimaging data according to the VASCOG criteria [23]; however, given the lack of cerebrospinal fluid testing for beta-amyloid, we cannot exclude the presence of mixed cognitive impairment in some patients. We used vertex TMS in the control group, which is a common approach in the study of cognitive function using TMS. However, it should be noted that one session of vertex theta burst stimulation had an effect on reaction time similar to that of stimulation of a functionally significant cortical area in the study by D. Pizem et al [42]. On the other hand, our study showed no significant differences in any of the test points in the control group, so we can to some extent ignore the significance of the effect of vertex stimulation. Another limitation of our study is the randomization procedure with unequal distribution ratio between groups. This resulted in a mismatch in number of subjects in the main and control groups. Such an approach increases the probability of the second type of error (reduces the probability of finding an effect, if any), but does not affect the probability of the first type of error (a false positive result). It should be noted that the aim of our study was not to identify the cognitive tests on which rTMS has the greatest effect, so the р values are given without adjustment for multiple comparisons. However, if the study results are used in work where such selection is required, a correction should be applied to control the possibility of false positives.

Conclusion

This study provides promising results on the potential efficacy of fMRI-guided rTMS for treatment of MCI in CSVD patients. However, given the above limitations, further larger studies are needed for more definitive conclusions.

 

¹ Statistical Parametric Mapping; Wellcome Centre for Human Neuroimaging, UCL Queen Square Institute of Neurology, London, UK. URL: http://www.fil.ion.ucl.ac.uk/spm

About the authors

Dmitry Yu. Lagoda

Research Center of Neurology

Author for correspondence.
Email: lagoda.d@neurology.ru
ORCID iD: 0000-0002-9267-8315

Cand. Sci. (Med.), researcher, Noninvasive neuromo- dulation group, Institute of Neurorehabilitation and Rehabilitation Medicine, Research Center of Neurology

Russian Federation, Moscow

Ilya S. Bakulin

Research Center of Neurology

Email: bakulinilya@gmail.com
ORCID iD: 0000-0003-0716-3737

Cand. Sci. (Med.), senior researcher, Head, Noninvasive neuromodulation group, Institute of Neurorehabilitation and Rehabilitation Medicine, Research Center of Neurology

Russian Federation, Moscow

Alexandra G. Poydasheva

Research Center of Neurology

Email: alexandra.poydasheva@gmail.com
ORCID iD: 0000-0003-1841-1177

Cand. Sci. (Med.), researcher, Noninvasive neuromodulation group, Institute of Neurorehabilitation and Rehabilitation Medicine, Research Center of Neurology

Russian Federation, Moscow

Dmitry O. Sinitsyn

Research Center of Neurology

Email: d_sinitsyn@mail.ru
ORCID iD: 0000-0001-9951-9803

Cand. Sci. (Physics and Math.), researcher, Institute of neurorehabilitation and rehabilitation medicine, Research Center of Neurology

Russian Federation, Moscow

Alfiia K. Zabirova

Research Center of Neurology

Email: alfijasabirowa@gmail.com
ORCID iD: 0000-0001-8544-3107

postgraduate student, neurologist, Research Center of Neurology

Russian Federation, Moscow

Zukhra S. Gadzhieva

Research Center of Neurology

Email: zuhradoc@mail.ru
ORCID iD: 0000-0001-7498-4063

Cand. Sci. (Med.), neurologist, 3^rd Neurological department, Institute of Clinical and Preventive Neurology, Research Center of Neurology

Russian Federation, Moscow

Maryam R. Zabitova

Research Center of Neurology

Email: gadjieva@neurology.ru
ORCID iD: 0000-0003-2523-333X

Cand. Sci. (Med.), researcher, 3^rd Neurological department, Institute of Clinical and Preventive Neurology, Research Center of Neurology

Russian Federation, Moscow

Kamila V. Shamtieva

Research Center of Neurology

Email: kamila.shamt@gmail.com
ORCID iD: 0000-0002-6995-1352

Cand. Sci. (Med.), researcher, 3^rd Neurological department, Institute of Clinical and Preventive Neurology, Research Center of Neurology

Russian Federation, Moscow

Larisa A. Dobrynina

Research Center of Neurology

Email: dobrla@mail.ru
ORCID iD: 0000-0001-9929-2725

D. Sci. (Med.), chief researcher, Head, 3^rd Neurological department, Institute of Clinical and Preventive Neurology, Research Center of Neurology

Russian Federation, Moscow

Natalia A. Suponeva

Research Center of Neurology

Email: nasu2709@mail.ru
ORCID iD: 0000-0003-3956-6362

D. Sci. (Med.), Corresponding Member of the Russian Academy of Sciences, Director, Institute of Neurorehabilitation and Resto- rative Technologies, Research Center of Neurology

Russian Federation, Moscow

Michael A. Piradov

Research Center of Neurology

Email: mpi711@gmail.com
ORCID iD: 0000-0002-6338-0392

D. Sci. (Med.), Academician of the Russian Academy of Sciences, Director, Research Center of Neurology

Russian Federation, Moscow

References

Supplementary files