Active rTMS not superior to Sham for Treating Cognitive Impairment in Patients with Alzheimer’s Disease

doi:10.21203/rs.3.rs-3147268/v1

Download PDF

Research Article

Active rTMS not superior to Sham for Treating Cognitive Impairment in Patients with Alzheimer’s Disease

https://doi.org/10.21203/rs.3.rs-3147268/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 01 Apr, 2024

Read the published version in Neurotherapeutics →

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Background-

We report results of a large multisite double-blind randomized trial investigating the short and long-term efficacy of repetitive transcranial magnetic stimulation (rTMS) applied to patients with Alzheimer’s disease (AD) at mild to moderate stages, in doses of either 2 or 4 weeks of treatment (5 days/week), whilst compared with 4 weeks of sham TMS. Randomization to treatment group was stratified based on age and severity. The objectives of this study were to: 1) investigate the efficacy of active rTMS versus sham, 2) investigate the effect of dose of treatment (2 or 4 weeks), and 3) investigate the length of benefits from treatment.

Methods

- The rTMS pulses (20 Hz, 30 pulses/train, 25 trains, 10-s intertrain interval) were applied to the bilateral dorsolateral prefrontal cortex bilaterally using neuro-navigation. We compared the primary outcome measure’s (ADAS-Cog) score changes from pre- to post-treatment, with assessments at baseline and 4 more times up to 6 months post-treatment.

Results-

Data of 135 patients were analyzed. The mean total ADAS-Cog score at baseline did not differ between the active and sham treatment groups, nor across the three study sites. The overall results show significant cognitive improvement after treatment with either sham or active coils.

Conclusion-

The results do not support the benefit of active rTMS versus sham for cognitive treatment of mild/moderate AD. We discuss this finding in the context of the existing literature on rTMS therapy for AD, as well as evidence of the sham coil’s potential to induce a low-level current in the brain.

Alzheimer&rsquo

s disease

rTMS efficacy

dorsolateral prefrontal cortex

double-blind

randomized clinical trial

No significant difference between sham and active rTMS for cognitive treatment of patients with Alzheimer’s disease.
No significant difference between two- and four- week treatment durations of active rTMS.
About 68% of the participants showed a significant positive response (improvement) to either active or sham rTMS treatment at post-treatment assessment.
The positive response to treatment amongst the 68% lasted up to 8 weeks post-treatment.

Investigations of the figure-8 Magstim sham coil used shows the sham coil can also induces a low current in the brain that potentially may have a similar effect to that of transcranial electrical stimulation.

The use of repetitive transcranial magnetic stimulation (rTMS) treatment for improving or stabilizing cognition in patients with mild to moderate Alzheimer’s disease (AD) has been on the rise in recent years due to some encouraging results from pilot studies. Most of the studies applying rTMS as a treatment for AD have applied the pulses with a high-frequency (10–20 Hz); some have chosen the dorsolateral prefrontal cortex (DLPFC) either only the left side^[1–3] or bilaterally^[4–6] as the brain area(s) for stimulation, while some applied stimulation to additional sites such as the Broca and Wernicke areas^[5], and more recently some chose the precuneus as the area of stimulation^[7]. However, the majority of studies were conducted in small sample sizes mainly due to the demanding protocols of rTMS treatment. The maximum sample size reported to date was 109 (out of 130 enrolled), which has been the phase III randomized sham-controlled multisite study of the neuroAD™ system in patients with mild to moderate AD^[5].

Although a majority of rTMS studies have reported some positive outcomes on the cognitive status of people with AD, yet there is reason to be cautious in concluding that rTMS represents a breakthrough in the treatment of AD. Forty years of rTMS research has shown that its behavioural effects are influenced by a wide range of factors, including the targeted brain region as well as the frequency, duration and number of treatment sessions. Systematic reviews of a decade of rTMS studies in the context of AD have not yielded consistent conclusions about the optimal stimulation parameters for enhancing cognition. For example, the frequency of the applied pulses is well established as a key parameter influencing physiological and behavioural effects. High-frequency pulses (10–20 Hz) increase cortical excitability and synaptic plasticity through long-term potentiation mechanisms^[8]; however, a systematic review paper on the studies on rTMS efficacy for AD treatment did not find any statistically significant differences in outcomes whether using pulse frequencies < 10Hz or > 10Hz^[9].

The target stimulation area of the brain is another important variable whose effects are not very well established. Based on knowledge of functional neuroanatomy and the brain regions most affected by AD, previous studies have targeted the precuneus, Broca area, Wernicke’s area, the DLPFC, or even combinations of these regions. Nevertheless, it seems that neither the brain region targeted for stimulation, nor whether the rTMS pulses were delivered with cognitive training, played a significant role in rTMS treatment efficacy^[9].

Another important parameter is the duration of the treatment. Most commonly, researchers have used 2 to 4 weeks of everyday (5 days/week) treatment, while some have used maintenance treatment (either 1 or 2 sessions/week) up to 6 months^{[5, 10]}, and another study in a limited sample performed maintenance treatment every 3–7 months with two weeks treatment up to two years after the baseline^[4]. Other than the latter study in a very limited sample size (n = 10)^[4], no other previous study has followed up with the patients to investigate the duration of rTMS effect. On the other hand, the same review paper^[9] found a significant correlation between the total number of pulses delivered per protocol and the study effect size. The total number of the pulses obviously is correlated with the duration of the study; however, it is not known whether there is a difference in delivering the same number of pulses over a short period of time or over a longer period of time.

In the last 5 years, we have been running a large, multisite (Winnipeg and Montreal in Canada and Melbourne in Australia), randomized, placebo-controlled, dose finding double-blind clinical trial investigating the effect of rTMS treatment for improving or stabilizing cognition in patients with mild to moderate AD. The detailed protocol of the study can be found in^[11]. In this paper, we present the results of the study using data including 156 participants. To the best of our knowledge, this is the largest clinical trial to date using rTMS treatment in AD and the first exploring duration of treatment (dose − 2-weeks vs 4-week) as a variable of interest. In addition, this is the only larger study that has followed up measuring the long-term effect of rTMS treatment up to 6 months post-intervention.

Study Design and Participants

This was a multisite (Winnipeg and Montreal in Canada and Melbourne in Australia), randomized, placebo-controlled, double-blind randomized clinical trial (NCT02908815) for investigating the effect of rTMS treatment for improving or stabilizing cognition in patients in the mild to moderate stage of AD. The details of the protocol are described in^[11]. In brief, the study randomized the participants into three groups to receive one of the two doses of active treatment (either 2 or 4 weeks of 5 days/week) or 4 weeks (5 days/week) with a sham coil wherein 1500 pulses at 20 Hz were delivered in 1.5-second trains with 10-second intertrain intervals each session (each day); the pulses were applied to DLPFC bilaterally. Inclusion criteria were to have: age > 55 years; MoCA score between 7 and 25; a Clinical Dementia Rating (CDR) score of 1 to 2; a Cornell Scale for Depression in Dementia (CSDD) score of 18 or less to rule out moderate to severe depression; a diagnosis of probable mild or moderate AD as confirmed by the treating neurologist or psychiatrist, and be on a stable dose (or no dose) of an acetylcholinesterase inhibitor for at least 3 months prior to study entry with no plans to change medication for the duration of the study.

The study was approved by the Ethics board of each site of the study and all participants and their primary caregiver signed an informed consent form prior to enrollment according to Declaration of Helsinki. Participants were randomized based on their age (less or greater than 70 years) and severity (measured by CDR as 1 or 2) to be in three groups of active 2-weeks (R2), active 4-weeks (R4) or sham 4-weeks (S4) treatment. Participants were not informed that there was no 2-week sham group and we assessed all participants at Week3 (after two weeks of treatment) in order to mask the group assignment. Participants, their caregivers, assessors and those who analyzed data were all blind to treatment group assignment. The sham coil was also identical to the active coil in terms of produced sound and sensation over the scalp during the application.

Trial Procedures

Using the Super Rapid-2 Magstim system (including two air-cooled figure-8 active and sham coils), the rTMS pulses were administered at 90–100% intensity of the resting motor threshold (RMT) of each participant as described in^[11]; the majority received rTMS at 100% intensity and it was lowered down to 90% for those who could not tolerate 100% intensity. The DLPFC location was determined using the Brainsight Neuronavigation system software and each patient’s MRI scan using Talairach coordinates (x, y, z) = (Left: − 50, 30, 36; Right: 50, 30, 36) with the coil held at approximately 45 degrees relative to the horizontal axis. The same targeting procedures were used whether using the active or the sham coil.

Outcome Measures

The primary outcome measure was the change in the Alzheimer Disease Assessment Scale-Cognitive Subscale (ADAS-Cog) score from pre- to post-treatment at different times up to 6 months after the baseline. We used different versions of the ADAS-Cog word list for recall to reduce the practice effect between the assessment sessions. There were several secondary outcome measures as described in^[11]; however, in this paper we present the results of only the main primary (ADAS-Cog) and secondary outcome measures (changes in neuropsychiatric symptoms measured by Neuropsychiatric Inventory–Questionnaire (NPI-Q) and changes in activities of daily living measured by Alzheimer Disease Co-operative Study-Activities of Daily Living Inventory (ADCS-ADL)). We assessed participants within one week prior to the start of intervention (baseline) and at 3, 5, 12, 20, and 28 weeks after the start of the intervention.. The assessments on Weeks 3 and 5 were performed within 3–5 days post-treatment, while later follow-up visits (Week 12, 20, 28) were completed within a week of their target date. The initial sample size, to have a minimum of 80% statistical power and a significance level of 0.05 at Week 5 assessment, had been estimated as 208 considering 10% dropout^[11].

Objectives of the Study

The objectives of this study were to investigate 1) whether patients in either of the active treatment groups (R2, R4) respond better (shown by a decrease of ≥3 points of ADAS-Cog respect to baseline or the way it is defined in the “Response Analysis” section) than those in sham group (S4), 2) whether there is a difference between the effects of R2 and R4 treatments, and 3) to assess the duration of treatment effects, if any.

Statistical Analysis

The detailed statistical analysis has been described and discussed in^[12]. First, we tested the primary outcome measure (ADAS-Cog) for any significant differences between the three study sites at baseline, with the intention of combining their data, if appropriate, when comparing the treatment groups. Baseline comparisons were done using ANOVA or its non-parametric equivalent in case assumptions of normality and homogeneity of variances were not satisfied.

Then, we performed a complete analysis of the ADAS-Cog data. Logarithmic transformation of the scores was used to adjust for outliers and any departure from normality assumptions. A mixed effects model for longitudinal data was used to test for the effects of Group (R2, R4, Sham), Time and the interaction of Group and Time using the changes in ADAS-Cog scores from baseline to post-treatment assessments.

We conducted a repeated measure ANOVA to compare the changes of log(ADAS-Cog) scores (the effect of treatment) over Weeks. Then, we used a multivariate ANOVA (MANOVA) to compare the three groups within each study site. Finally, the three sites were compared based on the ADAS-Cog scores obtained in all assessments using a MANOVA approach.

Response Rate Analysis- While the ADAS-Cog score change with respect to baseline is the primary outcome measure of this study, we also considered a term called “Response Rate” using the following criteria that was derived based on similar literature monitoring improvement/decline in AD patients undergoing pharmaceutical treatments using similar outcome measures^[13–17]. We defined a “Response Rate” based on ADAS-Cog score along with caregiver scores on the patient’s symptoms and severity and functionality^[12].

The “Response Rate” of the participants was defined as the following: a “Marked Positive Response” if their ADAS-Cog score has ≥3 point improvement compared to baseline at either Week5 or Week8 assessment sessions. A “Moderate Positive Response” if the ADAS-Cog score has a non-significant improvement (< 3 points) in ADAS-Cog AND improvement or same in ADCS-ADL OR NPI_Q at either Week5 or Week8. If the AND part does not hold, then it is considered as a “Small/Stabilized Response”. Likewise, if there is a non-substantial decline in ADAS-Cog (< 3 points) AND there is an improvement in both NPI-Q and ADCS-ADL at either Week 5 or Week8, it is considered a “Small/Stabilized Response”; otherwise if the AND part does not hold OR the ADAS-Cog score shows a substantial increase (i.e., worsening performance), it is considered as “Non-Responsive”. All the AND/OR are logical notions. It worth noting that ADAS-Cog change with respect to baseline still plays the dominant role in the definition of the Response Rate. The main difference of this response rate definition compared to simply using ≥3 points reduction of ADAS-cog score (respect to baseline) as a response rate is for patients with a marginal ADAS-Cog change with respect to baseline.

Next, we investigated the effects of treatment groups (R2, R4 and Sham) given the site of the study and time of assessment on responders (the combined groups of Marked, Moderate and Small Responders) versus non-responders as defined above. We also analyzed the effect of Time on the response to treatment (the difference between assessments).

Data Availability

The de-identified performance data along with the corresponding information file about the data, will be made available to the public within one year after the study is fully finished. The data will be shared with researchers upon a signed data access agreement on the FTP server of the study PI.

With a target number of 208 participants based on our initial sample size estimation ^[11], 633 individuals were screened, of whom 156 were enrolled. The study was ceased at that point following the conduct of the analysis presented in this paper. Twenty-eight participants withdrew; however, seven of them had data up to 8 weeks post-treatment. Thus, a total of 135 (156 − 28 + 7) participants’ data were analyzed in this study (Fig. 1). Overall, the rate of withdrawal/drop-out from the study (18%) was higher than the 10% expected rate; however, a number of participants withdrew due to COVID-19 pandemic and lab closure in Australia site. The rTMS treatment overall was found tolerable without any serious adverse effects although two participants were withdrawn by the PI for safety concerns of an issue that could be related to rTMS treatment. Table 1 shows the demographics of the participants in different groups and sites for all enrolled patients, and Table 2 shows the participants’ withdrawal reasons.

Table 1

Demographic and clinical characteristics (frequencies or means ± SDs) of participants in each group and site.
Site	Group	N, #Males, # Females	Age (years)	CDR	MoCA	Baseline ADAS-Cog	CSDD
Winnipeg	R2	25 (14 M, 11 F)	72.9 ± 7.6	1.08 ± 0.37	13.8 ± 5.2	25.8 ± 9.9	3.5 ± 2.9
	R4	24 (15 M, 9 F)	72.4 ± 7.4	1.13 ± 0.34	17.0 ± 5.4	21.7 ± 9.8	3.0 ± 2.7
	S4	23 (15 M, 8 F)	73.5 ± 9.7	1.09 ± 0.29	17.3 ± 4.5	21.2 ± 8.9	4.2 ± 3.8
Montreal	R2	15 (7 M, 8 F)	72.8 ± 11.5	1.13 ± 0.35	16.1 ± 4.5	24.9 ± 10.3	5.2 ± 4.0
	R4	16 (7 M, 9 F)	72.4 ± 6.7	1.19 ± 0.40	13.4 ± 5.4	29.9 ± 11.2	5.6 ± 2.7
	S4	16 (6 M, 10 F)	72.9 ± 7.4	1.19 ± 0.40	16.1 ± 4.4	23.9 ± 11.8	5.4 ± 3.4
Melbourne	R2	12 (11 M, 1 F)	76.9 ± 3.9	1.17 ± 0.39	15.8 ± 4.1	21.4 ± 5.0	3.7 ± 2.2
	R4	13 (5 M, 8 F)	76.1 ± 5.9	1.23 ± 0.44	15.8 ± 5.2	24.3 ± 14.1	4.4 ± 4.7
	S4	12 (5 M, 7 F)	80.7 ± 8.2	1.17 ± 0.39	12.6 ± 3.9	24.1 ± 8.5	3.4 ± 2.9
All Sites	R2	52 (32 M, 20 F)	73.8 ± 8.3	1.12 ± 0.37	14.9 ± 4.8	24.4 ± 9.1	4.0 ± 3.2
	R4	53 (27 M, 26 F)	73.3 ± 6.9	1.17 ± 0.38	15.6 ± 5.5	24.9 ± 12.0	4.1 ± 3.4
	S4	51 (26 M, 25 F)	75.0 ± 9.1	1.14 ± 0.35	15.8 ± 4.6	22.7 ± 9.8	4.4 ± 3.5
Total	All	156 (85 M, 71 F)	74.0 ± 8.1	1.14 ± 0.36	15.4 ± 5.0	24.0 ± 9.0	4.2 ± 3.3

Table 2

Participants’ withdrawal reasons and their distribution among the sites and the treatment groups; AU, MQ and MB refer to Australia (Melbourne), Quebec (Montreal) and Manitoba (Winnipeg) sites; R2, R4 and S4 refer to 2 and 4 weeks active and 4 week sham treatment groups.
Withdrawal Reasons	Site / Group, #
Withdrawal Reasons	AU	MQ	MB
Finding treatment painful and/or exhausting		1 R2	1 R4
Other unrelated illnesses	1 R2, 1 S4	1 R2
Missing several treatments and/or assessments due to weather, commuting issues, etc.	1 S4	1 R2, 1 R4
Caregiver felt treatment is unproductive		1 S4	1 R4, 1 S4
Caregiver sickness		1 S4
Patient being non-compliant (too much movement under the treatment); PI decision to withdrawn			1 R2
Patient felt too anxious before starting the treatment	1 R4		2 R4
Patient changed medication during the treatment	1 R4		1 R2
Plausible side effect, PI decision to withdraw for safety (MQ Patient felt brain fog and chest pain after one session of treatment; MB patient felt numbness in the left side of face and left arm for a few minutes after the second treatment at night)		1 R4	1 R2
Due to Pandemic and labs’ closure	1 R2, 2 R4, 1 S4
Patient and carer become unresponsive to contact		1 (S4)	1 R4, 1 S4
Patient became ineligible by postponing the starting date			1 R2
Total (count, %)	9 (24%)	8 (17%)	11 (15%)

Using chi-square test of homogeneity of proportions over all enrolled patients, there was no significant difference between the proportion of females and males (\({\text{χ}}_{\left(\text{2}\right)}^{\text{2}}\text{=}\text{ }\text{4.05, }\text{p}\text{ }\text{=}\text{ }\text{0.132}\text{)}\) overall or between the treatment groups \(({\text{χ}}_{\left(\text{2}\right)}^{\text{2}}\text{=}\text{ }\text{1.56, }\text{p}\text{=}\text{ }\text{0.458}\)). Also, there was no significant difference between the proportion of females and males between the treatment groups in Montreal and Winnipeg sites \({\text{χ}}_{\left(\text{2}\right)}^{\text{2}}\text{=}\text{ }\text{0.280, }\text{p}\text{ }\text{=}\text{ }\text{0.869}\) and \({\text{χ}}_{\left(\text{2}\right)}^{\text{2}}\text{=}\text{ }\text{0.457, }\text{p}\text{ }\text{=}\text{ }\text{0.796}\), respectively). However, there was a significant difference in Melbourne site \({\text{χ}}_{\left(\text{2}\right)}^{\text{2}}\text{=}\text{ }\text{8.84, }\text{p}\text{ }\text{=}\text{ }\text{0.012}\): the proportion in group R₂ of Melbourne site was different from other groups. Note that sex was not a parameter for stratified randomization.

To compare the ages of participants between the sites and treatment groups, we used the ANOVA test. The mean ages of participants between the groups (R2, R4, and S4) were not significantly different \(\text{F }\left(\text{2, 153}\right)\text{ = 0.596, }\text{p}\text{ = 0.552}\) as age was a randomization parameter for group assignment. The sites and different treatment groups were found not significantly different in terms of baseline ADAS-Cog scores, which means we could combine data of the different sites for comparison between the treatment groups.

Figure 2 shows the boxplot of the ADAS-Cog total scores averaged amongst participants (N = 135 total analyzed) of each group at different assessment times. As can be seen, each group’s participants on average had an improvement in ADAS-Cog total scores after the treatment. This improvement lasted for two months post-treatment and then gradually returned to baseline values after 6 months. However, there were no differences in this trend between active and sham groups or the 2 dose groups.

Analysis of mixed effect models for comparing treatment and time interval effects

A mixed effect model for repeated measures was used to investigate the effect of treatment (the R2, R4, S4 groups), time intervals (weeks) and the interaction of treatment and time on the difference between log(ADAS-Cog) at baseline and a given time point in all sites together and each site separately. The results of the mixed effects model on all participants and all sites are shown in Table 3. The results showed that the ADAS-Cog scores of different weeks were significantly different but not different for different treatment groups, congruent with the results mentioned above. Figure 3 shows post-hoc paired comparison tests of the difference in scores of different time periods (weeks) in each site using the pairwise t-test. As can be seen, again the change of log(ADAS-Cog) score at any point of assessment time compared to baseline is not significantly different between the three treatment groups. There were some significant improvements with respect to baseline in some weeks post-treatment in all 3 groups (decreases and increases in mean scores) but no clear pattern of difference between the 3 groups (Fig. 3).

Table 3

Results of comparing the time (weeks) and treatment groups in all sites.
	# of DF	Den. DF	F-value	P-value
Weeks	4	528	17.68	< 0.00001
Groups (R2, R4, S4)	2	132	0.54	0.58
*WeeksGroups**	8	528	1.64	0.11

Comparisons based on the Response Rate

Table 4 shows the distribution of different responses across the sites and in total, averaged for the two active groups combined versus sham group. Using Chi-square test on the equality of different response proportions in different treatment groups, we observed that the Marked, Moderate, Small and Non-responders were not equal in all treatment groups, nor in any individual group, but the proportions were homogeneous between the sites. In all groups, “Marked Responses” were more frequent than moderate and small responses (Fig. 4). All the statistical analyses done on ADAS-Cog (and its log) changes as the primary outcome generated similar results when considering the “Response Rate” as defined in Statistical Analysis section instead of the ADAS-Cog score changes respect to baseline. This was expected since ADAS-Cog score plays a major role in the definition of the Response Rate as well and since the “Marked” Response (of which the ADAS-Cog score is the main determinant) was more frequent than the other types of responses.

Table 4

Distribution of the response rates (n = 135) across the sites for the active and sham groups; active group includes both R2 and R4 groups.
		Marked/Moderate	Mild/Stabilized	Non-responsive/declined
Active rTMS	Winnipeg (n = 40)	72.5%	15.0%	12.5%
	Montreal (n = 27)	77.8%	0.0%	22.2%
	Melbourne (n = 23)	65.2%	8.7%	26.1%
	All sites (n = 90)	72.2%	8.9%	18.9%
Sham rTMS	Winnipeg (n = 21)	81.0%	9.5%	9.5%
	Montreal (n = 13)	61.5%	7.7%	30.8%
	Melbourne (n = 11)	72.7%	9.1%	18.2%
	All sites (n = 45)	73.3%	8.9%	17.8%

Next, we partitioned the responders into two groups of Positive- and Non-Responders, and repeated the analyses. The results showed the proportion of the Positive Responders was similar across the three Groups (R2, R4 and S4) and much higher than the non-responders. It shows there were improvements across repeated assessments, but in all active and sham groups similarly; that can easily be observed by the distribution of the responses shown in Table 4.

In this paper, we present the results of a randomized double-blind placebo-controlled rTMS treatment study applied to AD patients with two doses of active treatments versus sham; participants were followed for a period of 6 months post-baseline. To the best of our knowledge, this study is the largest of its kind to date, with a sample size of 156 participants and the longest follow-up duration (6 months) post-treatment investigating the short and long-term effect of rTMS as a treatment for AD. Overall, our outcomes indicate that high frequency rTMS treatment applied to bilateral DLPFC does not yield greater cognitive improvement than the “sham” group either immediately post-intervention or in the longer term. On average, the two active treatment groups and the sham group show similar cognitive outcomes across the assessment times up to 6 months follow-up. In addition, and interestingly, on average all groups showed no decline compared to baseline at follow-up assessment. We briefly compare these results to those reported for other clinical trials of rTMS in AD, then move on to present evidence in support of a potential explanation for the improvement in cognitive status seen in sham-treated participants.

Our results are contrary to some recently reported findings on the effect of active rTMS treatment on AD. A recent study with a relatively high number of participants (a total of 109 in a randomized grouping)^[5] reports statistically significant benefit of rTMS treatment applied to DLPFC and to Broca and Wernicke areas when paired with simultaneous cognitive exercises for those who received active treatment versus sham (cognitive exercises were also given during sham). The benefit of active treatment was not seen immediately after the intervention but at 5 weeks post-treatment. In a separate study^[10], 25 AD patients randomized to receive active rTMS over the precuneus area showed better cognitive performance after treatment than the 25 participants in sham intervention. In both studies, cognitive performance of the patients in the active group remained stable during the period of the intervention, while cognitive performance in the sham group declined, and the difference between the two groups at post-intervention was statistically significant. Our study on 156 (135 analyzed) participants randomized into active and sham groups yielded very different results, i.e., improvements were seen in the sham and the active treatment groups, not only immediately post-intervention but also at the 6 months follow-up assessment. We attempted to control for practice effects by using alternate versions of the ADAS-Cog in which different word lists were presented for memory testing; however, the remaining test items are consistent across assessments and it is conceivable that growing familiarity with the assessment procedures in general from baseline through Week 5 may have led to better test outcomes, even in patients with a limited capacity to remember from one session to the next.

In examining the literature on rTMS treatment efficacy in AD, we note that many research publications do not specify exactly when the assessments were done. The timing of post-intervention assessment has been reported as “after the last rTMS session”^[6], “immediately after”[18], the week of assessment^[5] or simply “post-treatment”^[7]. Distinguishing between “immediately after” as in 5 minutes later on the same day, and three days later (as we did in this study), is crucial to separating out the potential acute effects of rTMS from the cumulative effects of several weeks of intervention. Future clinical trials should specify clearly the time elapsed between the last rTMS pulse and the assessment of cognitive function.

Notwithstanding important differences in the brain regions targeted with rTMS in our studies and those that yielded more positive outcomes, the fact that our sham participants on average did not decline even at 6 months post-intervention is intriguing. This led us to consider the possibility of a potentiating effect of the sham coil, which was intended to be an inactive control condition. The present study used the Magstim sham coil, as have the majority of trials of rTMS in Alzheimer’s disease. Experiments have shown^{[20, 21]} that significant perpendicular magnetic fields are produced by this coil, a finding that we have now replicated in our own laboratory (see Supplementary Materials). As suggested by^[19], the induced significant electric field can produce low-field effects that affect neuromodulation. The study in^[20] showed the weak electric fields from the sham TMS coil can modify the cortical excitability of the neurons. Therefore, it is possible that the induced weak electrical fields of the sham coil have an effect similar to that of transcranial alternating electrical stimulation (TACS), even though the TACS is given in a continuous mode and rTMS fields are induced within the duration of the given train of pulses. Our protocol had 10-second intertrain intervals between each train of rTMS pulses given at 20 Hz. It is possible that the induced electric currents in the brain may have had a similar effect of TACS at 20 Hz.

Earlier studies such as^{[1, 4]} used a different technique for sham stimulation; they used a 2-cm thick wooden piece with the shape of the same coil placed under the active coil. The wooden spacer beneath the active coil effectively reduces the penetration of the TMS pulses into the brain significantly while not changing the sound of the pulses. However, the wooden piece under the active coil significantly reduces the scalp sensation of the pulses and that can potentially unblind the participant. On the other hand, the authors of the study in^[4] argued that, since the participants had memory impairment, and there were wash-out periods between the two blocks of active and sham treatment, none of the 10 participants reported feeling any difference. The results of the study in^[4] showed the sham group’s cognitive status stayed the same respected to baseline, while the active group showed improvement.

While our results indicate that the active groups’ positive responses (improvements) were similar to that of the sham group, it does not necessarily mean that active rTMS treatment should not be used as a treatment for any AD patients. One plausible explanation for our results could be that the improvement is all due to improvement in depression. However, this explanation is unlikely because the majority of our participants were not depressed at baseline (see CSDD scores on Table 1). Out of the 135 patients, only 13 had a CSDD score of 10 or higher that could suggest a mild level of depression (see Table 5), and participants with higher levels of depression were excluded by design. The majority of participants with mild depressive symptoms (58.3%) were from the Montreal site, which showed better results for the Active group versus Sham for R2 group. Nevertheless, these numbers are small and represent less than 10% of the entire subset of patients who showed improvement in the course of this study. Further, in the absence of a post-intervention measure of depression we cannot test the hypothesis that a reduction in depressive symptoms underlies the observed cognitive improvement.

Table 5

Distribution of the 13 patients with major depression (CSDD ≥ 10) across the sites and groups, their response rate as Marked, Moderate, Small or Non-responsive.
Site	Group	# (Response Rate)
Winnipeg	R2	2 (1 Marked, 1 Moderate)
	R4	0
	S4	2 (1 Marked, 1 Small)
Montreal	R2	4 (3 Marked, 1 Non)
	R4	1 (Moderate)
	S4	2 (1 Marked, 1 Moderate)
Melbourne	R2	0
Melbourne	R4	1 (Marked)

One may attribute the improvement in both active and sham groups to a “care effect” of the interaction due to attending the treatment (either active or sham) but that is unlikely as such a care effect in our experience usually does not last more than a month after the interaction (treatment) time is over. We see the improvement lasted more than 2 months in the majority (about 68%) of patients of active and sham groups and some even up to 6 months post-treatment (see Fig. 5).

Our data show about 10% of patients either in active or sham groups cognitively declined after rTMS treatment; although the amount of decline measured by ADAS-Cog score changes with respect to baseline is higher in the active group compared to sham group (average values of 2.6 and 1.8 for R2 and R4, respectively, compared to -0.11 for S4), the numbers are too small to draw any meaningful conclusion.

A recent study on a subset of our data who also went through an exploratory assessment, called Electrovestibulography (EVestG), which showed a high (76%) accuracy in predicting at baseline whether a participant improves, declines or has no change after an active rTMS treatment^[21]. The feature that was most different between responders and non-responders was a lower frequency (efferent) modulation of the vestibular afferent firing pattern for responders, and this was hypothesized to be related to GABAergic changes^[21]. Unfortunately, the number of EVestG study participants who received sham rTMS treatment was too small to draw any conclusion on sham rTMS’ efficacy at baseline.

The results of our randomized clinical trial demonstrate that active rTMS is not superior to sham rTMS coil (Magstim coil used in this study) when following the stimulation protocol used herein. Our protocol was designed to match those most commonly used for studies in AD at the outset of this trial. The results do not preclude that rTMS may be found effective in treating AD when using different stimulation parameters or targeting other brain regions or using a sham coil that does not induce any neuromodulation (if it exists). Moreover, given the importance to society of finding novel interventions for AD, further studies should seek to identify patient characteristics that best predict response to interventions using brain stimulation.

Limitations of the study

This study despite, being carefully designed, is not free of limitations. One limitation is that the assessor personnel changed over the 5 years of the study. With training videos, regular video calls to discuss the assessment procedures, and regular site visits we made every effort to ensure consistency in administration of interventions and assessments; however, turnover of personnel could possibly affect the scoring. Second, our experiments on the magnetic fields induced by the sham coil suggest that the use of this coil may have reduced the ability to detect differences between the control and active treatment conditions. Finally, important differences between our study and recent work by others demonstrating efficacy of rTMS in AD include the target site and the absence of a cognitive training intervention. The current study was designed to provide a definitive estimate of the efficacy of rTMS in a protocol that has been used most commonly in the literature, i.e., stand-alone treatment with rTMS over the DLPFC. It is important to recognize these methodological differences, as they preclude generalization of our null results to the wider range of rTMS protocols that have yet to be fully explored for treating AD.

This study for the first time examined the efficacy of rTMS for treatment of cognitive impairment in AD in a randomized double-blind placebo-controlled study up to 6 months post-treatment. In summary, the results of our analysis of 135 participants in three groups, active treatment in two doses of 2 and 4 weeks versus sham, indicate 1) an overall significant improvement for all groups after treatment, and 2) no difference between the sham and active groups even at any point up to 6 months after the baseline. Since this result lasted till 6 months post-intervention, it is unlikely to be entirely due to placebo effects. We hypothesize that sham TMS coil also has a therapeutic effect, akin to a modified tACS. While active coil has shown positive therapeutic effect for the majority (> 68%) of patients, there was continued decline or no effect for other patients. We believe the future of rTMS application as a treatment for AD should be further explored with “true” sham coils.

Ethics Approval and Consent to Participate - Each site of the study received the Ethics approval from their respective university’s Ethics Board prior to starting the study. All studies were performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki, and all participants gave written consent prior to the experiments. All participants (and their caregiver in cases where the caregiver was a legal representative acting on behalf of the participant) read and signed an informed consent form prior to participate in the study.

Consent for publication - All participants (and their caregiver in cases where the caregiver was a legal representative acting on behalf of the participant) read and signed an informed consent form to publish the outcomes of the study.

Availability of Data and Materials - The de-identified performance data along with the corresponding information file about the data, will be made available to the public within one year after the study is fully finished. The data will be shared with researchers upon a signed data access agreement on the FTP server of the study PI.

Competing Interests – In the last 3 years, the co-author PBF has received equipment for research from Neurosoft, Nexstim and Brainsway Ltd. He has served on scientific advisory boards for Magstim and LivaNova and received speaker fees from Otsuka. He has also acted as a founder and board member for TMS Clinics Australia and Resonance Therapeutics.

Authors’ Contributions - ZM wrote the main text, supervised the entire research and analysis as she was the main PI. MU, GR and BL conducted the experiments on sham coil. MU wrote the supplementary file. GR and BL reviewed it. GR helped with running research and BL was also in charge of ensuring all sites follow the same protocol exactly. PBF and LK were the site PIs for Australia and Montreal sites, respectively. All authors reviewed the manuscript and contributed to its different sections.

Funding - This study was funded by Weston Brain Institute. PBF co-author is supported by a National Health and Medical Research Council of Australia Investigator grant (1193596).

Acknowledgements – The PI wishes to acknowledge the generous donation of Puchniak’s family to help with the travel expanse of out-of-town participants in Manitoba site.

Cotelli, M., et al., Improved language performance in Alzheimer disease following brain stimulation. J Neurol Neurosurg Psychiatry, 2010. 82: p. 794–7.
Li, X., et al., Cortical plasticity is correlated with cognitive improvement in Alzheimer’s disease patients after rTMS treatment. Brain Stimulation, 2021. 14(3): p. 503–510.
Zhang, F., et al., High-frequency repetitive transcranial magnetic stimulation combined with cognitive training improves cognitive function and cortical metabolic ratios in Alzheimer's disease. J Neural Transm (Vienna), 2019. 126(8): p. 1081–1094.
Rutherford, G., B. Lithgow, and Z. Moussavi, Short and Long-term Effects of rTMS Treatment on Alzheimer's Disease at Different Stages: A Pilot Study. Journal of experimental neuroscience, 2015. 9: p. 43–51.
Sabbagh, M., et al., Effects of a combined transcranial magnetic stimulation (TMS) and cognitive training intervention in patients with Alzheimer's disease. Alzheimer's & Dementia, 2019.
Ahmed, M.A., et al., Effects of low versus high frequencies of repetitive transcranial magnetic stimulation on cognitive function and cortical excitability in Alzheimer’s dementia. Journal of Neurology, 2012. 259(1): p. 83–92.
Koch, G., et al., Precuneus magnetic stimulation for Alzheimer’s disease: a randomized, sham-controlled trial. Brain, 2022. 145(11): p. 3776–3786.
Hallett, M., Transcranial magnetic stimulation and the human brain. Nature, 2000. 406(6792): p. 147–150.
Menardi, A., et al., Transcranial magnetic stimulation treatment in Alzheimer’s disease: a meta-analysis of its efficacy as a function of protocol characteristics and degree of personalization. Journal of Neurology, 2022. 269(10): p. 5283–5301.
Koch, G., et al., Transcranial magnetic stimulation of the precuneus enhances memory and neural activity in prodromal Alzheimer's disease. NeuroImage, 2018. 169: p. 302–311.
Moussavi, Z., et al., Repeated Transcranial Magnetic Stimulation for Improving Cognition in Patients With Alzheimer Disease: Protocol for a Randomized, Double-Blind, Placebo-Controlled Trial. JMIR Res Protoc., 2021. 10(1).
Moussavi, Z., et al., Repeated Transcranial Magnetic Stimulation for Improving Cognition in Alzheimer Disease: Protocol for an Interim Analysis of a Randomized Controlled Trial. JMIR Res Protoc, 2021. 10(8): p. e31183.
Schrag, A. and J.M. Schott, What is the clinically relevant change on the ADAS-Cog? Neurol Neurosurg Pscyhiatry, 2012. 83: p. 171–173.
Burns, A., et al., Defining treatment response to Donepezil in Alzheimer's disease. Drugs Aging 2008. 25(8): p. 707–714.
Lanctot, K.L., et al., Efficacy and safety of cholinesterase inhibitors in Alzheimer's disease: a meta-analysis. Canadian Medical Association Journal, 2003. 169(6): p. 557–564.
Cook, R.J. and D.L. Sackett, The number needed to treat: a clinical useful measure of treatment effect. BMJ, 1995. 310: p. 452–454.
Posch, M. and P. Bauer, Interim analysis and sample size reassessment. Biometrics, 2000. 56(4): p. 1170–1176.
Saitoh, Y., et al., Randomized, sham-controlled, clinical trial of repetitive transcranial magnetic stimulation for patients with Alzheimer’s dementia in Japan. Frontiers in Aging Neuroscience, 2022. 14.
Smith, J.E. and A.V. Peterchev, Electric field measurement of two commercial active/sham coils for transcranial magnetic stimulation. Journal of Neural Engineering, 2018. 15(5): p. 054001.
Opitz, A., et al., Is sham cTBS real cTBS? The effect on EEG dynamics. Front Hum Neurosci, 2014. 8: p. 1043.
Lithgow, B.J., Z. Dastgheib, and Z. Moussavi, Baseline Prediction of rTMS efficacy in Alzheimer patients. Psychiatry Research, 2022. 308: p. 114348.

Competing interest reported. The co-author, PBF, declares that in the last 3 years he has received equipment for research from Neurosoft, Nexstim and Brainsway Ltd. He has served on scientific advisory boards for Magstim and LivaNova and received speaker fees from Otsuka. He has also acted as a founder and board member for TMS Clinics Australia and Resonance Therapeutics.

SupplementaryFile.pdf

Download PDF

Journal Publication

published 01 Apr, 2024

Read the published version in Neurotherapeutics →

Version 1

posted

You are reading this older preprint version

Read the latest preprint version →

Active rTMS not superior to Sham for Treating Cognitive Impairment in Patients with Alzheimer’s Disease

Status:

Journal Publication

Version 1

Abstract

Background-

Methods

Results-

Conclusion-

Figures

Research Highlights

Introduction

Materials and Methods

Study Design and Participants

Trial Procedures

Outcome Measures

Objectives of the Study

Statistical Analysis

Data Availability

Results

Analysis of mixed effect models for comparing treatment and time interval effects

Comparisons based on the Response Rate

Discussion

Limitations of the study

Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1