Repetitive transcranial magnetic stimulation (rTMS) for treatment-resistant major depression (TRMD) Veteran patients: study protocol for a randomized controlled trial

rTMS is a method of delivering brain stimulation with neither the seizures or risks associated with ECT, nor the potential side effects and risks of pharmacological augmentation strategies, such as MAOI therapy. It may offer a viable alternative to ECT for some patients. Several studies, including a systematic review and meta-analysis of the studies to date, appear to show a positive effect in TRMD [5] with an effect size of Cohen’s d of about 0.65 (moderate effect). This is comparable with that of contemporary antidepressant medications. The most recent large-scale clinical trial using advanced coil designs documented an effect size of 0.76 [6]. With a minimal side-effect profile, and the rarity of untoward events and side effects [5], safety concerns about the use of rTMS are considerably fewer than for ECT. Importantly, rTMS may be less expensive to administer than ECT (largely due to not requiring anesthesia) [7]. Thus, there is the potential for a significant advance in VA mental health care, with associated cost savings, if rTMS were to be shown effective in treating TRMD in VA patients.

Despite these positive studies of rTMS in civilian populations, research suggests that Veterans may experience a differential response to mental health treatments compared to the response seen in civilians. This finding has been well-documented in studies of treatment of posttraumatic stress disorder (PTSD) in which Veterans did not show evidence of the treatment gains seen in civilians in both pharmacological and psychotherapy trials [8, 9]. The reasons for this treatment disparity are most likely multi-factorial in nature. Clinical trials of new antidepressant treatments are initially tested in highly selected patients, free from comorbid conditions, and not taking other antidepressant medications. Yet, compared to civilian populations, Veterans experience a greater preponderance of medical and psychiatric comorbidities that complicate their clinical presentation and negatively impact their response to therapeutic intervention [10]. For example, in the national Veteran population that carries a diagnosis of a depressive disorder (N = 946,342 in the 2005 outpatient file), over 80% have at least one additional psychiatric diagnosis with the most common dual-diagnoses being PTSD (39%) and substance use disorder (45%). Most of these patients are already taking one or more psychotropic drugs. Furthermore, the VA has a special concern about treating potentially suicidal Veterans. Thus at this time, it is not clear if the results from the positive rTMS studies in civilian populations will translate into similar effects in the VA practice setting with all its comorbidities, multiple medications and risk for impulsive behavior. Given the fixed resources available for mental health treatment in the VA, rational allocation of resources is best guided by clinical trials in the exact population in which the treatments will be utilized, which led to several challenges in study design.

We define patients with TRMD as Veterans who meet the Diagnostic and statistical manual of mental disorders, 4 ^th edition (DSM-IV) criteria for MDD and who have failed at least two prior pharmacological interventions as defined by a modified and updated version of the Antidepressant Treatment History Form (ATHF) [11]. We will not exclude Veterans with PTSD or history of substance use and we will obtain detailed history on these disorders. Participants will remain stable on their antidepressant medication regimen throughout rTMS treatment.

We designed the inclusion/exclusion criteria to identify patients with TRMD who exhibit a full range of the manifestations of that condition. Furthermore, we intend our recruited population to represent the VA pool of patients with TRMD. Table 1 shows the detailed inclusion/exclusion criteria.

Potential participants will be recruited through a number of methods. These include, but are not limited to, referral by primary providers, referral by mental health providers, flyers posted in common areas such as canteens at VA hospitals, review of the VA administrative databases containing information for both outpatient and inpatient encounters, which are housed in the Austin Information Technology Center, sending IRB-approved messages to providers twice a year, and posting basic information about the study in local VA SharePoint sites. Local study staff will work with providers at their respective medical centers to identify Veterans who may be appropriate for participation in this study. Although most recruitment will occur in mental health clinics, recruitment within women’s health clinics will be used to try to maximize the enrollment of eligible women.

The Site Investigator (SI) will obtain informed consent (Additional file 2). Patients who sign the Informed Consent Form and meet the study eligibility criteria will be enrolled into the study and will be randomized to either rTMS or sham rTMS. We use an adaptive randomization scheme to approximately equate numbers of patients randomized to each treatment group within several important subsets. These subsets include enrollment site and whether patients have a substance use disorder and/or PTSD. We will make treatment assignments using a “biased coin” procedure that should improve overall balance across subsets. We calculate imbalance by summing the marginal totals for these three factors for each treatment group and calculating the difference, D. If the imbalance is < 3, we will assign to rTMS or sham rTMS with equal probability; otherwise, we will assign to the group that increases imbalance with probability, 1/D. We incorporate this approach into our electronic data capture system.

To randomize a patient into the study, the SI or the Study Coordinator (SC) will submit the electronic randomization form. This computerized system, after verifying eligibility, will randomize a patient to either the rTMS or to the sham rTMS treatment group. A non-sequential treatment number will be assigned. This unique treatment number will be key-entered into the device which will be associated with a treatment assignment and will enable the rTMS device to deliver the appropriate treatment (active or sham) to each patient. Every attempt will be made to randomize a participant so that he/she will receive his/her first rTMS treatment as soon as possible after randomization.

As this is a double-blinded study, both participants and on-site study staff (including site investigators, site coordinators, clinicians administering study treatment, and staff involved in assessing adverse events and study outcomes) will be blinded to group assignment. At no point will participants or study staff be unblinded to group assignment prior to study completion. The Data Monitoring Committee (DMC) will determine when they should be unblinded to treatment assignment for the reviewing of adverse event data. A questionnaire will be used before and after the first treatment session and again at the end of the final study visit to elicit patient and study staff perception of group allocation.

During the course of this study, all study procedures will adhere to the most current safety and application guidelines for administration of rTMS [12]. We will monitor adverse events per VA and US Food and Drug Administration (FDA) regulations. However, due to the complex psychiatric presentation of the typical VA patient with TRMD, we have developed additional safety protocols for the monitoring of suicidality and substance use. Study personnel will work closely with the participant’s primary mental health provider in situations in which increased suicidality or substance use is suspected to ensure full clinical coverage of these Veterans.

The Cooperative Studies Program Coordinating Center (CSPCC), located in Perry Point, MD, USA, will provide administrative, data management and statistical support for the study. CSPCC staff will provide guidance on completion of forms and data quality queries. They will develop editing software and manage the study database (Additional file 3). All reports generated during the ongoing phase of the study and the final statistical analyses will be the responsibility of the CSPCC.

The Executive Committee is the management and decision-making body for the operational aspects of the study. This committee is chaired by Dr. Jerome Yesavage, and includes the Study Biostatistician, the CRP, a minimum of three SIs, and outside consultants, if necessary. This committee monitors the performance of participating sites and the quality of data collected. The Executive Committee formulates plans for publications and oversees the publication and presentation of all data from the study. Permission from this committee must be granted before any study data may be used for presentation or publication. This group also does not receive outcome data during the course of the study. Executive Committee decisions that need to be made between regularly scheduled meetings will be made during periodic phone conferences.

The DMC is a group of outside experts in the area of TRMD, clinical trials and biostatistics that reviews the progress of the study and monitors patient enrollment, outcomes, adverse events, and other issues related to patient safety. The DMC makes recommendations to the CSP Director as to whether the study should continue or be modified or terminated. The DMC can consider patient safety or other circumstances as grounds for early termination, including either compelling internal or external evidence of treatment differences or unfeasibility of addressing the study hypotheses (e.g., poor patient intake, poor adherence to the protocol). The DMC will meet annually to review data reports prepared by the CSPCC. At the 6-month interval between the annual meetings, the DMC will receive a data report for their review. This group will receive outcome data during the course of the study. In order for the DMC to make its recommendation for continuation of the study, it will be necessary for them to see the analyses for the primary outcome measure every time that the report is run and it is possible to calculate the primary outcome measure. Periodic monitoring of interim results can significantly affect the probability of making an incorrect decision. A number of formal techniques have been developed for interpreting interim results. At the organizational meeting, the DMC will select the technique that it wants to use to monitor the study.

The SMART will conduct monitoring visits to each participating site to monitor investigative records and practices to ensure sites are in compliance with both the study protocol and GCP. These site visits will occur annually or more frequently if directed by the study Monitoring Plan. Independent quality assurance audits will also be conducted at selected sites if needed.

The VA CIRB will approve and oversee the ethical and human subjects aspect of the study. This includes review of adverse events, the protocol and its amendments, and annual reports. All CIRB protocol amendment approvals are shared with the entire Study Team.

We will administer rTMS using a modified MagPro R30 device with Cool-B65-A/P coil. Active treatment will consist of 4000 pulses per session, delivered over the left prefrontal cortex at 120% of motor threshold (MT). We will determine MT at screening and at the beginning of every treatment block (every 5 sessions) with a separate C-B60 MT coil using electromyographic measurement of the resting right thumb (right abductor pollicus brevis). Threshold will be determined using an adaptive procedure (ClinicalResearcher software, Knoxville, TN, USA). Treatment will be delivered at a pulse frequency of 10 Hz, length of each pulse train of 4 seconds, time between pulse trains of 10 seconds, and length of treatment of 25 minutes for each treatment session. Units of five treatment sessions will be delivered over a minimum of 5 calendar days and a maximum of 12 calendar days. Patients will receive a minimum of 20 sessions of treatment and, for those not clinically remitting, up to a maximum of 30 sessions. In addition, participants meeting criteria for remission after 20–30 sessions will receive six additional taper treatments delivered over a 3-week period. Thus, patients will receive up to a total of 144,000 stimulation (sham or active) pulses in the study.

The credibility of the study depends on the development of a highly believable sham treatment. Sham treatment will be accomplished by using the inactive side of the Cool-B65-A/P coil that functions both as an active (A) and placebo (P) coil. The coil is completely symmetrical and does not distinguish the A and P sides. Treatment condition (A vs P) is determined electronically in the software of the device, based on a predetermined random treatment assignment table hard-coded in each device’s memory and which is not accessible to study staff. The coil has a built-in position sensor and a symmetrical mechanical design with no markings that would differentiate the active or placebo sides. At the start of each session, a participant’s randomization number is entered into the device, triggering the device to instruct the TMS administrator on how to orient the coil based on a built-in position sensor and the treatment assignment (active/placebo) associated with the randomization number. Consequently, it is not possible for the operator to see or hear which side is used. Additionally, for each treatment session, whether sham or active, each patient will wear scalp electrodes through which a low-voltage, low electric current (2–6 mA at no more than 100 V) will be passed in order to provide cutaneous stimulation that mimics the sensation of actual rTMS. At the same time, we use a stimulation-synchronized white noise generator to mask the “click” noise from each stimulation. Both the patient and the treatment administrator receive the masking noise. We make every attempt to well-mask the patient, the treatment administrator, and all personnel at each treatment site to the treatment group assignment.

We define the standardized treatment location over the left dorsolateral prefrontal cortex by moving the TMS coil 6 cm anterior to the MT location along the left superior oblique plane with a rotation point about the tip of the patient’s nose [20‐22]. Our decision to move 6 cm rather than 5 cm reflects an evolution over several clinical trials. The optimal stimulation site, the dorsolateral prefrontal cortex is believed to be in Brodmann’s areas 9 and 46 and the vast majority of TMS studies move 5 cm rather than 6 cm anterior from the MT site. Traditionally this approach is called the “5 cm rule”. Unfortunately, a previous MRI study found that in a substantial number of patients the location was in the premotor rather than dorsolateral prefrontal cortex. Furthermore, these patients in whom the localization was inaccurate had an inferior treatment response [23]. The OP-TMS trial adopted a hybrid approach to targeting using this 5 cm rule [24] but also performing MRI and moving 1 cm anterior, to adjust for patients in whom there was inappropriate localization to the premotor cortex [25]. This hybrid approach led to fewer patients with inaccurately selected targets and an associated inferior treatment response. This current study adopts a 6 cm rule but also performs a separate study using MRI in patients from six of the nine sites for verification of accuracy and effect on treatment response (VA Funding CS R&D CX000604). Stimulation locations will also be related to potential functional MRI targets to assess this rule as an alternative to image guidance in patients in whom MRI is not feasible. Consistency in targeting across TMS sessions and in evaluating the accuracy in the MRI will be accomplished by marking the stimulation on a cloth cap (Magventure).

A separately funded MRI protocol (see above) will collect additional data on the 6 cm rule as implemented here. That study will use T1-weighted, structural MRI from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) to compute the accuracy of the location using the TMS stimulation location marked with a fiducial marker as in Herbsman et al. [23]. This T1 structural image will also allow measurement of the distance of TMS coil from the cortex (skull-to-cortex distance). This variable has been demonstrated to be an important potential moderator of treatment response because there is a substantial decrease in the strength of the magnetic field with increased distance of the coil from the brain in older patients with brain atrophy [26]. There has also been success with targeting based on image guidance [27], so that both active and resting state fMRI protocols from the functional Brain Imaging Research Network are being collected [28] and relative proximity of the 6 cm rule to these alternative targets will be computed.

The patient’s participation in the study may be terminated at any time if the SI deems that the patient has not been following the protocol. This will generally be done only when the protocol violation significantly increases the risk associated with continuing to participate in the study. Any female participant who becomes pregnant during the acute treatment phase of the study will discontinue the study treatments for safety reasons, as the effects of rTMS on the unborn fetus is not known at this time, and she will immediately enter the follow-up phase. Any female participant who becomes pregnant during the follow-up phase of the study will continue to be followed up in accordance with the protocol and will complete all assessments.

The primary outcome for this trial will be remission as defined by a score of 10 or less on the HRSD24. This HRSD is widely considered the “gold-standard” depression measure and is the most widely used assessment of depression in rTMS trials. Secondary outcomes will include reduction in depression as measured by the Montgomery-Asberg Depression Rating Scale (MADRS), reduction in suicide ideation as measured by the Beck Scale for Suicide Ideation (BSS), depression as measured by the Beck Depression Inventory (BDI), quality of life as assessed by the Veterans RAND 36-Item Health Survey (VR-36), and cognitive function as measured by a neuropsychological battery. The selection of these measures was based upon their use in previous studies of rTMS in TRMD. The detailed outcome collection scheme and data elements are shown in Tables 2 and 3.

There is growing evidence to suggest there are neurocognitive benefits of rTMS in TRMD. Specifically, improvements in verbal memory were demonstrated, seemingly independent of the antidepressant effects of rTMS. Additional improvements in verbal learning and psychomotor speed are also evidenced, though these are more likely the result of reduction of depressive symptoms [29, 30]. Based on these results, a brief neuropsychological battery was selected to examine as secondary outcomes in the present trial. Measures included in this battery include the Rey Auditory-Verbal Learning Test (RAVLT), Trail Making Test (TMT), Symbol Digit Modalities Test (SDMT), Judgement of Line Orientation (JLO), Controlled Oral Word Association Test (COWAT), Stroop Color and Word Test, and the North American National Adult Reading Test (NAART).

The economic analysis will take the perspective of the VA health care system, and include both short-term and longer-term analyses. All analyses will follow current standards for economic analyses of healthcare interventions [31‐33]. Study data will be used to determine rTMS treatment costs, and VA production costs will be used for all other treatment costs. The main outcome for the economic analysis will be the primary study outcome, but the Short Form 36-Item Health Survey for Veterans (SF-36V) will be converted to utilities to allow comparisons to the cost-effectiveness of interventions for other conditions. The short-term analyses will use study data and VA databases to follow subjects for up to a year. Decision modeling will be used to extend the analysis to a lifetime. A budget impact analysis will also be conducted to facilitate VA management decisions.

Better understanding of moderators of treatment effect defines in whom a treatment best works. Such information is important for “personalized treatment” of patients and for the proper allocation of resources. Better understanding of the mediators help explain how a treatment works. Understanding the mechanisms of action may be important for enhancing our ability to interpret clinical findings. If for example, we find little clinical impact of an “active” treatment and this were our only analysis, we would not be able to determine if the outcome were the result of “insensitive” outcome measures, a “weak” treatment, or a faulty design. However, if the treatment does in fact have significant effects on measures relevant to the proposed mechanism of action (e.g., changes in a relevant biological marker or psychological test), but is not accompanied by a change in clinical outcome, a strong argument can be made that the treatment “worked” on the proposed mechanism, but that it was not the appropriate mechanism to target.

While there is growing support for the use of rTMS in the treatment of TRMD, there are patient characteristics (e.g., patient age, gender, type of depression, and severity of psychiatric symptoms) that potentially moderate the treatment response. Several studies have shown that older patients have poorer outcomes after rTMS treatment compared to younger patients and younger patients also required fewer rTMS treatments to achieve remission [34]. It has also been suggested that gender may impact treatment response, as some studies have shown the antidepressant effect of rTMS to be greater in female than male subjects [35]. To date, most studies of rTMS have been conducted in those with unipolar depression, though there is some suggestion that participants with more complex presentations or severe psychiatric symptoms may evidence a different treatment response. Specifically, patients with psychotic depression and those with more severe symptoms of depression may experience significantly less improvement with rTMS treatment [36, 37]. These are important points for the current study as the typical Veteran with TRMD is a man in his mid-50s with a complex presentation of depression.

Based on the literature review, the study planning committee felt that a 10% difference between treatments would be of clinical relevance given the severity of the illness. With a sample size of 180 per group, the proposed study will have a power of 81% to detect an absolute difference between groups of 10% in the percentage of those participants who remit (6% sham and 16% rTMS). Thus, a total of 360 patients will be required from nine potential VA sites. Each site will be expected to randomize approximately 40 eligible patients over the recruitment period of the study.

In addition to the main analysis, using the entire randomized or intent-to-treat cohort, logistic regression models will be used for “completers” and also for “fully compliant” subjects to provide further information about treatment effects. For example, it would be expected that if rTMS had a significant clinical effect, its effect would appear greater in completer and totally compliant cohorts, than in the entire randomized cohort. Other analyses will be performed on secondary measures to further provide useful clinical information. Some secondary outcome measures, such as sustained response rate (“recovery”) and response on secondary outcome measures, can also be analyzed using logistic regression models. Other potential secondary analyses include change in suicidality, change in cognitive function, and change in QOL. The effect of rTMS on such measures will be determined using random regression and similar techniques that maximize the use of available data in repeated measures designs. At the end of the study, cost-effectiveness analysis will be performed.

The primary endpoint analysis will be based on the primary outcome measurement at the end of the treatment. Secondary outcome analyses are focused on the three time points, i.e., baseline, end of the treatment, and end of the follow up. However, the measurements from all time points will be used to illustrate the trend of outcome measurement changes across time.

Subjects who drop out during the treatment phase will be considered as treatment failures for the purpose of the primary analysis and the missing values will not be imputed. Multiple imputations will be used to handle missing data using Rubin’s method for secondary analyses.

While the current study is addressing several important potential moderators of treatment response and proposed mechanisms of action, we acknowledge there are additional potential mechanisms of action. Response to treatment may depend on several genetic and biochemical moderators, including brain-derived neurotrophic factors (BDNF) [38], Apolipoprotein E (APOE) [39], serotonin transporter genes [40], and polymorphisms in Catechol-O-methyl transferase (COMT) [40]. While these targets are beyond the scope of the current trial, they are worthy of future investigation as an understanding of these moderators could help determine both the best course of treatment for an individual and the mechanisms of rTMS in cell signaling.

We also acknowledge that there will be groups of Veterans who do not qualify for this trial, but could potentially obtain great benefit from rTMS. These groups include those Veterans with more mild levels of depression and those Veterans with significant cognitive impairment due to various forms of dementing illnesses and the residual effects of traumatic brain injury (TBI). Thus, it will be important for future trials to include these groups in their design.