Adverse events of repetitive transcranial magnetic stimulation in older adults with depression, a systematic review of the literature

Abstract Objective In the last decade, repetitive transcranial magnetic stimulation (rTMS) has been introduced as a non‐invasive neuromodulation therapy for depression. Little is known, however, about (serious) adverse events (AE) of rTMS in older adults with a depression. In this article, we want to study what is known about (serious) AE of rTMS in older adults (>60 years) with late‐life depression (LLD). Methods A systematic search has been performed according to the PRISMA guidelines in PubMed, EMBase and PsycInfo. We have screened 622 articles for eligibility. Eleven studies, evaluating 353 patients in total, were included in this review. Results AE were reported in 12.4% of the older adults with a LLD treated with rTMS, serious AE in 1.5%. Headache (6.9%) and discomfort at the stimulation site (2.7%) are the most commonly reported AE. Serious AE reported are: psychiatric hospitalization (three times), a combination of posterior vitreous detachment and retinal tear, and increased suicide ideation (both once). Conclusions rTMS in older adults with LLD was concluded overall to be safe due to the low frequency of AE reported in trials and observational studies. In case‐reports, however, more serious AE have been described. To tailor use of rTMS in older adults with LLD, more research is needed in larger samples to optimize tolerance.


| INTRODUCTION
Today's society is aging rapidly. In the upcoming 30 years, the proportion of the world's population aged above 60 years will probably rise from 12% to 22%. 1 The number of older adults who suffer from mental illness is rising as well, based on long-term cohort studies that report 22%. 2 One of the most common mental disorders in this age group is depression, next to dementia. Depression most often causes substantial suffering and a decrease in quality of life. Late-life depression (LLD) is a primary diagnosis of major depression, dysthymia or minor depression according to DSM-IV criteria, in patients aged 60 years and over. 3 Especially in older adults, treatment of LLD is challenging as vulnerability for adverse events (AE) of anti-depressive medication increases with age. 3 Treatment options for this vulnerable group should be carefully assessed. A rapidly increasing new treatment option for depression is repetitive transcranial magnetic stimulation (rTMS), a noninvasive brain stimulation technique.
With rTMS, a rapidly changing magnetic field is used to generate an electric current in the brain tissue just below the skull, to alter the cortical excitation of this brain region and its interconnected brain network. 4 rTMS was approved by the Food and Drug Administration for treatment of mild to moderate treatment-resistant depression in 2008, but the first TMS device has been developed already during the early 1980s. 5 To date, rTMS is not a standard treatment option in the general population and only recommended by some guidelines with caution, 6 as for example in the Dutch multidisciplinary depression guideline. 7,8 A Cochrane review published in 2002 mentioned that there was no strong evidence for benefit from using rTMS to treat depression, although the small sample sizes did not exclude the possibility of benefit. 9 There are no open consensus guidelines available from the American Psychiatry Association. 10 The clinical TMS society stated some treatment recommendations based on a literature of three RCT's and a user survey of 68 members of the clinical TMS society.
They suggest that TMS therapy is recommended and should be considered as an acute treatment for symptomatic relief of depression in patients who have failed to receive satisfactory improvement from prior antidepressant medication in the current episode. 11 For the best results of an rTMS protocol, knowledge of the pathology is important. For example, there is an imbalance between the left and right hemisphere in depression; the left dorsolateral prefrontal cortex (DLPFC) is known to be hypoactive. 12 The application of highfrequency left rTMS (HF left rTMS) or low frequency right rTMS (LF right rTMS) both have anti-depressive effects. Different stimulation frequencies are thought to exert their effects through a differential influence, that is, increasing or decreasing excitability. 12 The use of rTMS as a treatment is increasing, also in older adults.
Nevertheless, extensive clinical research in the older adult population is sparse, especially when it comes to (serious) AE. 6 Higher prevalence of AE in the older population is expected as this patient group suffers far more than younger adults from physical comorbidity and polypharmacy. 13 Comorbidity and polypharmacy are common exclusion criteria in clinical trials, but frequently present in clinical practice with older adults. The exclusion of comorbidity and polypharmacy causes a possible underestimation of prevalence of (serious) AE in clinical trials.
Research on the efficacy of rTMS has shown response rates of 20% to 50% in older adults with LLD, 14 similar to response rates in adults with treatment-resistant depression. 15 Studies on the tolerability of rTMS in adults has reported several AE such as headache (9.7%), local pain and discomfort (9.3%), and neck pain, toothache, and paresthesia (together 4.7%). 16 Rare serious AE (<1%) are seizures and induction of hypomania, hearing changes and burns from the coil. 4,14,[17][18][19][20][21] Most often, people aged over 60 are excluded from trials. The National Institute for Health Care Excellence (NICE)-guideline of rTMS for depression is based on studies with a mean age between 38.4 and 50.5 years, the remaining studies used in the NICE-guideline did not mention a mean age. 6 To date, several publications exist about efficacy of rTMS in older adults with LLD, but no studies have systematically examined tolerability and safety of rTMS for LLD. 14 The rationale for this review is to give an overview about the (serious) AE of rTMS in LLD.
In this review, we consider reports of patients that are older adults with a LLD, use rTMS as intervention, use sham rTMS (when available) as comparison and report (serious) AE as outcome.

| MATERIAL AND METHODS
The study was performed according to the PRISMA-guidelines.

| Eligibility criteria
Study characteristics: We included studies that investigated patients who were older adults with LLD, used rTMS as intervention, used a comparison consisting of (when available) sham rTMS with real rTMS, in which (serious) AE was reported as an outcome or had a study design that is a review of the existing literature on this topic to check the references of that review. Studies that included patients with brain damage, such as tumors and brain contusions, were excluded.
We included articles of the last 15 years. We excluded meta-analyses, and articles written in another language than English as well.
A search was executed on January 16th 2019 in PubMed, PsycInfo/Ebsco, and EMBase on the and updated on 23th of November

Highlights
� Little is known about (serious) AE of rTMS in older adults (>60 years) with LLD.
� A relatively low percentage of AE (12.4% in total) and serious AE (1.5% in total) in response to rTMS occurs in older adults with LLD.
� rTMS is a safe and well-tolerated treatment option for older adults with LLD.

-
2019. Mesh Terms and free text terms [tiab] were used. Also, the option 'Similar articles' in PubMed was used. For the search we used the following keywords: adverse effects, side effects, harmful effects, AEs, safety, headache*, nightmare*, somnolence*, pain, mania, convulsion, transcranial magnetic stimulation, repetitive transcranial magnetic stimulation, rTMS, aged, elderly, older adult*, elder and geriatric*. For the study selection we removed the duplicates first. Second, we screened the records on title and abstract. Then we read full-text articles assessing for eligibility. Finally, we performed a snowball search to select articles, by checking the references of the articles used for this review, that may have been missed in the primary search.

| Data collection process
Two authors performed data extraction (RJ & GO) independently, that was subsequently compared when different results were found.
The following data items were acquired: year of publication, number of patients, age of patients, stimulation parameters used during rTMS (type of rTMS, location of rTMS, frequency and duration of rTMS, and rTMS details), and (serious) AE. We included all (serious) AE as reported as such in the publications found by the search as the principal summary measure.

| Synthesis of results
To combine the results of the studies, we counted all the reported (serious) AE and divided the number by that of all the included patients.

| Risk of bias of individual studies
We assessed the risk of bias of the included studies by using the recommended Cochrane Collaboration's Risk of Bias evaluation tool, that evaluates the bias in the conducted studies on reporting results, in this case: AEs. Using this tool, two independent authors in a double blind fashion (RJ & GO) scored six types of bias (selection bias, performance bias, detection bias, attrition bias, reporting bias and other types of bias) as low, high or unclear on potential risk of bias. 22

| Study selection
A total of 994 Articles was found. After removing duplicates, 811 articles remained. We screened these articles for eligibility by reading the title and abstract, resulting in 92 articles. These articles were full-text screened, eight articles were identified as eligible.
Through snowball search two additional articles and two case-reports were found. We excluded one study, 23 because the study population appeared to be a subsample of another included study. 24 In total, 11 studies were included. There were two RCT's, four open label studies, two retrospective studies, two case series, and two case-reports ( Figure 1).
A total of 331 included older adults received rTMS and 49 older adults received a form of sham treatment. Most older adults (n ¼ 246) received high frequency rTMS (HF-rTMS), 70 older adults received low frequency rTMS (LF-rTMS), and 15 older adults received both HF-rTMS and LF-rTMS. None of the studies mentioned the use of theta burst stimulation in their methods. 25 Sham condition was used in three studies, and performed by placing the coil at a 90°angle with the scalp, or the intensity setting was put on 0,0, or was not described. Specifying AE between stimulation frequencies was difficult due to the low number of older adults who received LF-rTMS and the lack of specificity of reported AE in some studies. We therefore lumped the findings of serious AE of HF-rTMS and LF-rTMS together. Some studies reported the use of H1-coils 26 ; although an H-coil can effectively stimulate deeper targets, it might activate different regions compared to standard figure-of-eight coils. 12 in the group treated with unilateral HF-rTMS; one patient reported headache and one patient reported insomnia. One patient dropped out due to intolerance to the treatment and three patients dropped out due to lack of treatment response (see Table 1).

| Randomized controlled trials
Kaster and colleagues 26  stimulation, one patient dropped out because of a worsening of the depressive symptoms, three patients dropped out due to circumstances not related to HF-rTMS.

| Open label studies
Leblhuber and colleagues 28  adults. There were some dropouts (due to anxiety, insomnia, induced mood elevation, increasing discomfort from the stimulation of the scalp, and the need for hospitalization during the protocol period); no details such as the exact number of these dropouts were available.

| Retrospective studies
Desbeaumes and colleagues 31  The older adults were treated with HF-rTMS at the left DLPFC (5 or 10 Hz). AE were retrospectively identified. Out of 75 older adults, three were admitted to a psychiatric hospital, although it was not mentioned if this was due to the psychiatric disease or due to the AE as a result of rTMS. One older adult was hospitalized unrelated due to the HF-rTMS.

| Case series and case reports
In the study of Milev and colleagues, 24     headache (n ¼ 23, 6.9%), followed by pain at stimulation site (n ¼ 10, 3.0%). Insomnia, nasopharyngitis, aphthous ulcer, corneal abrasion, dermatitis, sinusitis, nausea and fatigue were all mentioned once (0.3%). Serious AE, however, were reported in 5 of the 331 cases (1.5%) [32][33][34] ; posterior vitreous detachment and retinal tear (once), increased suicidal ideation (once), and psychiatric hospitalization (three times, 0.9%). Most of the (serious) AE were in the HF-rTMS group, except of the one (serious) AE of the study of Kung and colleagues, 34 that was the only (serious) AE in the LF-rTMS group.

| Risk of bias assessment (Table 2)
We did not only assess the studies on content, but also on quality. By using the recommended Cochrane guidance, 22  Serious AE were reported in 5 of the 331 cases (1.5%) [32][33][34] ; posterior vitreous detachment and retinal tear (one older adult), increased suicidal ideation (one older adult), and psychiatric hospitalization (three older adults). It is still unclear, however, whether a causal relationship between rTMS and some of these serious AE exists, like the posterior vitreous detachment and retinal tear. There may be an indirect effect through increased intra ocular pressure, caused by rTMS evoking these ophthalmological problems. 34 With respect to mortality as the most serious AE event that can occur, no deaths have been reported as (serious) AEs in any of the studies. This makes the mortality rate zero in older adults with LLD receiving rTMS.
The number of serious AE is low if we compare it with the absolute risk for all-cause mortality over 1 year in older adults with a depression not taking antidepressants (7.0%), for those taking tricyclic antidepressants (8.1%), for those taking selective serotonin reuptake inhibitors (10.6%), and for taking other antidepressants (11.4%). 13 Mortality rates in rTMS are not described or structurally studied until now. It could be that the mortality rates in rTMS are so low that it is missed in the current studies. We did not find anything about mortality rates in rTMS in the literature while we performed the study.
In general practice, older adults do not meet the exact criteria of a research protocol, as they have more comorbidity, polypharmacy, and long-term treatment. Older adults use more medication, have more somatic comorbidity, and experience more frailty. 3 It is reasonable that an interaction of these factors can cause AE, although scientific evidence is lacking. Until now, there was no structural overview of (serious) AE in older adults with depression receiving rTMS. In some studies, AE were described as a second outcome measurement. In case reports (serious) AE were described as the main topic of the case report. 33,34 The incompatibility between on the one hand the results of the trials and on the other hand the results of the case reports is remarkable. Some study protocols mentioned dropouts, but did not mention the reason for dropouts (except of the study of Trevizol and colleagues 27 ). In total, there percentage of registered dropouts was 3.9%. These dropouts were not included them in their follow-up.
These patients could be dropouts due to (serious) AE. On the other hand, one might expect that patients who drop out of studies due to (serious) AE are usually reported as such in the study or referenced to an appendix of an institutional review board. Fourthly, the number of sessions given in most studies is not representative for the number given in daily practice. The protocol of most included studies comprised 10-20 treatment sessions versus 29 sessions in the case report of Elmaadawi et al. 33 A lot of AE appear early in the treatment phase of rTMS, but some AE may appear later on in the treatment phase.
Finally, stimulation intensity, location of stimulation, position of stimulation and number of sessions and stimuli potentially influence the occurrence of (serious) AE. In addition, patients specific factors like age, medication use, and brain morphology (e.g., functional connectivity and neurodegeneration) will also influence the occurrence of (serious) AE. 35 Consistent reports on AE are needed to describe such potential relationships. In addition, further research is warranted to find predictors for (serious) AE in rTMS. can be that the ageing process protects the older adults from AE.
Another possible cause is that older adults report fewer AE, because they are used to it, regard rTMS as a last treatment option for help, and accept AE better. 36 A systematic comparison between adults and older adults has never been done and it is hard to determine results on this matter.
Slotema and colleagues 16  found a percentage of dropout of 10.6% in the HF-rTMS DLPFC group and of 8.3% in the LF-rTMS DLPFC group. Based on the studies done in older adults, the dropout percentage of 3.9% seems to be lower than in the adult population, although research to compare these results has never been performed.

| CONCLUSION
In conclusion, rTMS is a safe and well-tolerated treatment option for older adults with LLD with a relatively low percentage of AE (12.4% in total) and serious AE (1.5%), based on the findings in this review. The most commonly reported AE are headaches, and pain at the stimulation site. Serious AE found in trials and case reports are psychiatric hospitalization, suicide ideation, retinal tear and posterior vitreous detachment. Routine assessment and registration of (serious) AE during and after treatment in (inter)national databases will help determine best practice in rTMS. Such registrations will help to identify potential relationships between (serious) AE and rTMS dependent factors (target area, frequency, intensity, number of sessions), psychiatric and somatic comorbidity, age, medication status. Future research in larger sample sizes is needed to tailor use of rTMS in LLD, and optimize efficacy while ensuring tolerance.