Effects of High Frequency Repetitive Transcranial Magnetic Stimulation (HF-rTMS) on Delay Discounting in Major Depressive Disorder: An Open-Label Uncontrolled Pilot Study

Background: Delay discounting (DD) refers to the decrease of a present subjective value of a future reward as the delay of its delivery increases. Major depressive disorder (MDD), besides core emotional and physical symptoms, involves difficulties in reward processing. Depressed patients often display greater temporal discounting rates than healthy subjects. Repetitive transcranial magnetic stimulation (rTMS) is a non-invasive brain stimulation technique applied in several countries to adult patients with treatment resistant depression. Studies suggest that this technique can be used to modulate DD, but no trial has assessed its effects on depressed patients. Methods: In this open-label uncontrolled trial, 20 patients diagnosed with MDD and at least stage II treatment resistance criteria underwent 20 HF-rTMS sessions over the dorsolateral prefrontal cortex (dlPFC; 10 Hz, 110% MT, 20 min). Pre-post treatment DD rates were compared. Effects on impulsivity, personality factors, and depressive symptoms were also evaluated. Results: No significant effect of HF-rTMS over the left dlPFC on DD of depressed individuals was observed, although rates seemed to increase after sessions. However, treatment resulted in significant improvement on cognitive impulsivity and depressive symptoms, and was well-tolerated. Conclusion: Despite the limitations involved, this pilot study allows preliminary evaluation of HF-rTMS effects on DD in MDD, providing substrate for further research.


Introduction
Major depressive disorder (MDD) is a widespread psychiatric disease that will become the leading worldwide cause of years living with disability by 2030 according to the World Health Organization (WHO). Episodes of depression, in addition to core emotional and physical symptoms, also involve difficulties in reward processing [1]. A reduction of reward anticipation, along with diminishment Studies assessing TMS effects on the DD of clinical populations remain scarce. Sheffer and colleagues [25] delivered 900 pulses at either 10 or 20 Hz over the left dlPFC, with a 110% motor threshold (MT), in a population of 66 smokers and non-smokers. Discounting rates decreased and effects were more intense as frequency was higher. Recently, Sheffer and colleagues [26] demonstrated improvement on discount rates in a population of 29 tobacco smokers after 8 sessions of 20 Hz (900 pulses per sessions, 110% MT) also over the left dlPFC. Nonetheless, in another study by Zack and colleagues [27], no significant effect on nine men suffering from pathological gambling was reported after a single rTMS session (10 Hz, 80% AMT, 450 pulses) over the mPFC. Post-rTMS effects actually headed towards the opposite direction and were highly variable [27].
As far as we know, no study has assessed rTMS effects on DD in depressed individuals. We have therefore conducted an open-label uncontrolled pilot study to evaluate whether rTMS applied to the dlPFC could reduce DD in MDD patients who have not shown satisfactory clinical progress with medication. We hypothesized that rTMS would decrease temporal impulsivity of these patients, along with the expected improvement on their depressive symptoms. The primary objective was to analyze changes in baseline DD rates after 20 rTMS sessions. As secondary outcome measures, we also investigated (i) the neuromodulatory effects of a single rTMS session on baseline DD scores; (ii) the effects of 20 rTMS sessions on general impulsivity, (iii) on personality traits, and (iv) on depressive symptoms; (v) possible correlations of DD rates with the other psychiatric measures; and (vi) DD rates correlations with sociodemographic characteristics.

Participants
The study was conducted at the psychiatric department of the University Hospital of Besancon, France, and aimed to enroll all eligible volunteers that met inclusion criteria. The inclusion criteria were as follows: (i) men and women over 18 years old, with (ii) diagnosis of MDD according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria, and (iii) at least stage II treatment resistant criteria [28,29]. Exclusion criteria were as follows: (i) age less than 18 years old, (ii) contraindication for brain magnetic resonance imaging (MRI) or TMS, (iii) presence of psychotic features and/or (iv) neurological comorbidity. Semi-structured psychiatric interviews to establish MDD diagnosis according to DSM-IV criteria and stage of treatment resistance were conducted by trained psychiatrists (E.H. or D.B.). Concomitant use of psychiatric medication other than antidepressants did not interfere with participation in the study. Twenty patients were included and submitted to brain MRI prior to rTMS sessions.

Study Design
In this open-label uncontrolled trial, all subjects underwent the same procedure. Patients received 20 sessions of high frequency rTMS (HF-rTMS), delivered twice a day over a period of two weeks (sessions were performed by E.H. or D.B. from Monday to Friday). The protocol was approved by the French Committee of Protection of Persons (CCPPRB Reference No.: 04/380) and authorization was given by the General Health Administration (DGS 2005/0030). Investigations were conducted in line with the principles from the Declaration of Helsinki. Signed written informed consent was obtained from all patients before study enrollment. and after sessions, and through a Visual Analogue Scale (VAS) applied by the end of sessions. The VAS assessed the patient's perception on pain, mood, fatigue, and motivation.

Outcome Measures
Measures on DD rates were collected at three points in time: prior to stimulation sessions (baseline scores), immediately after the first session, and after 20 sessions. DD was assessed by the 27-item Monetary Choice Questionnaire (MCQ), based on Kirby's DD inventory [8]. The MCQ is a pen-and-paper task composed of 27 items of hypothetical monetary choices between smaller immediate rewards and larger delayed ones. Items are classified into three categories: small, medium, or large rewards [8]. Monetary values were converted into local currency (EUR; €). Delays varied between seven and 186 days, and rewards ranged from 11€ to 85€.
Furthermore, measures of impulsivity, personality traits, and severity of depressive symptoms were collected prior to stimulation sessions and at the end of 20 rTMS sessions. Impulsivity was evaluated by the French version of the Barratt Impulsiveness Scale (BIS-10) [30,31], a 34-item self-rated checklist that gives overall and specific impulsivity scores based on three subscales: cognitive-, motor-, and non-planning-impulsivity. Assessment of major personality traits was obtained through the French version of the Big Five Inventory (BFI-Fr) [32]. This self-rated questionnaire independently evaluates the following personality factors: conscientiousness, neuroticism, extraversion, openness to experience, and agreeableness.

Data Analysis
DD rates are based on the hyperbolic function V = A/1 + kD, where V represents the subjective value of the delayed reward, A is the amount of the delayed reward, D is the delay, and k is the coefficient that estimates the subjective discounting rate for the given delayed reward. The hyperbolic function is more efficient than the exponential model of discounting, since it better integrates discount variations for decisions concerning rewards in the short and in the long run [34]. K-values were generated by the 27-item MCQ Automated Scores [35] for small, medium, and large rewards, as well as for overall discounting rates of each subject. Given the small sample size, the Wilcoxon sign rank test (non-parametric test) was performed for all hypothesis testing. Spearman's rho was used to explore correlations. The alpha value for significance was set at 0.05. Bonferroni's correction for significance was applied for multiple comparisons, with the significance level adapted for each explored objective (primary and secondary outcome measures). The ratings were analyzed with SAS 9.4 T5 Level MM3.

Participants Characteristics, Missing Data, and Outliers
Our sample was composed of 9 women (45%) and 11 men (55%), with a mean age of 54.4 (SD ± 9.854) years. All patients had been using antidepressant medications for at least 4 weeks: 18 of them were on antidepressant monotherapy (selective serotonin reuptake inhibitor (SSRI) = 7, serotonin and norepinephrine reuptake inhibitor (SNRI) = 7, tricyclic antidepressant (TCA) = 3, tetracyclic antidepressant (TeCA) = 1), and two patients had a combination of two antidepressants (SNRI with TeCA = 2). Mean baseline scores and demographic data are displayed on Table 1. Far outliers were removed based on the z-score method with a SD 3.29 (0.01% of data was removed). Additionally, data Brain Sci. 2019, 9, 230 5 of 11 were missing completely at random for some participants. Hence, the number of subjects whose data were considered for analysis is detailed in the "Subjects (n)" column.

Impact of 20 rTMS Sessions on DD
The effect on small, medium, large rewards, and overall discount rates (four comparisons; p < 0.0125) was assessed. The median values apparently increased after rTMS sessions (Figure 1), especially for medium rewards (p = 0.0293). However, analysis showed that there was no statistical significance ( Table 2).
Brain Sci. 2019, 9, x FOR PEER REVIEW 5 of 11 TeCA = 2). Mean baseline scores and demographic data are displayed on Table 1. Far outliers were removed based on the z-score method with a SD 3.29 (0.01% of data was removed). Additionally, data were missing completely at random for some participants. Hence, the number of subjects whose data were considered for analysis is detailed in the "Subjects (n)" column.

Impact of 20 rTMS Sessions on DD
The effect on small, medium, large rewards, and overall discount rates (four comparisons; p < 0.0125) was assessed. The median values apparently increased after rTMS sessions (Figure 1), especially for medium rewards (p = 0.0293). However, analysis showed that there was no statistical significance ( Table 2).

Impact of One rTMS Session on DD
As a secondary outcome, the effect of a single rTMS session on the same variables was also analyzed (p < 0.0125), but no significant change was observed (Figure 1).

Impact of rTMS on the BIS-10
The analysis of the impact of 20 rTMS sessions on the BIS-10 considered four scores: BIS-10 total score and cognitive-, motor-, and non-planning-impulsivity subscales scores (p < 0.0125). There was a significant reduction on cognitive-impulsivity scores (p = 0.0071) as displayed in Table 2. No significant change was observed in the other scores.

Impact of rTMS on the BFI-Fr
No significant effect of 20 rTMS sessions on the five personality traits assessed by the BFI-Fr (p < 0.01) was observed ( Table 2).

Impact of rTMS on Depressive Symptoms
Depressive symptoms assessed by the MADRS and the QIDS-SR16 (p < 0.025) were significantly reduced in both assessments (p = 0.0234; p = 0.0240, respectively) at the end of 20 rTMS sessions ( Table 2).

Correlation between DD and Other Psychiatric Measures
Spearman's rho test revealed significant correlations (p < 0.05) between pre-post 20-rTMS-session discounting rates and scores of other psychiatric measures. The agreeableness score on the BFI-Fr was negatively correlated to k-values for medium rewards (rho = −0.62658; p = 0.0071), while the non-planning-impulsivity subscale from the BIS-10 was positively correlated to k-values for large rewards (rho = 0.49898; p = 0.0491). No other significant correlations were found. Baseline psychiatric scores, age, and gender of participants had equally no significant influence on pre-post 20 rTMS discount rates.

Side Effects
All treatment sessions were well tolerated and subjects reported no side effects.

Discussion
The present study found no significant effect of 20 HF-rTMS over the left dlPFC on temporal impulsivity of patients diagnosed with MDD and meeting at least stage II criteria for treatment resistance. We expected that rTMS sessions would improve temporal discount rates in patients with MDD, based on previous studies with healthy participants. However, our results did not support this simple hypothesis. However, treatment significantly improved cognitive-impulsivity scores and depressive symptoms with no side effects reported.
There are limitations of this trial that must be acknowledged before further discussion. A major limitation is our small sample size, which may compromise the generalizability of outcomes. In addition, confounding factors could influence the results due to the open-label uncontrolled study design and potentially compromise internal validity. On the other hand, these preliminary findings might be useful for directing further research in the field because our study is strengthened by well-defined eligibility criteria for the investigated population, outcome assessment by validated instruments only, and the adjustment of confidence intervals for multiple comparisons.
Interestingly, as we analyzed data from pre-post 20 HF-rTMS sessions, an apparent increase on temporal impulsivity-especially for medium rewards (p = 0.0293)-was observed, although no significance of the effect was obtained. Methodological differences between our trial and previous trials with healthy participants [23] make direct comparison of results unfeasible, whereas data on the assessment of clinical populations remain insufficient. Factors that may influence rTMS efficacy such as the cortical target, stimulation parameters, and inter-individual variations require re-examination in light of previous evidence.
The choice of cortical target and stimulation parameters were in line with prior evidence regarding rTMS application on depression [19,20,36]. When it comes to stimulation parameters, neuromodulation by rTMS has shown therapeutic effects and is usually performed at 10 Hz when applied over the left dlPFC of depressed individuals, with an increase in responsiveness observed when the number of sessions is greater than 10 [19]. The chosen parameters in our study have shown themselves to be effective on this population since expected improvements on secondary outcomes were observed.
An important aspect to be considered while evaluating neuromodulatory results is the influence of inter-individual features. A large variability of outcomes is frequently reported by non-invasive modulation experiments and has sometimes been associated with characteristics such as age, gender, baseline cortical activation, brain metabolic activity, neural tissue heterogeneity, and even smoking status [37][38][39][40][41][42].
MDD is frequently associated with changes in brain functionality and structure [43,44], such as progressive neurodegeneration in the left dlPFC [45]. Pathological regions may promote changes in TMS-induced currents, modifying the magnitude, location, and orientation of current distribution, due to local perturbation in conductivity [46]. Drysdale and colleagues [47] have recently observed four depression biotypes in a cluster analysis of resting-state functional MRI (fMRI) data. Although the biotypes shared a common neuroanatomical pathological core, they displayed different abnormal connectivity patterns, especially regarding frontoamygdala, thalamic, and frontostriatal networks, as well as anterior cingulate and orbitofrontal areas. The biotypes were also associated with different a responsiveness to rTMS when applied to the dorsomedial PFC (dmPFC) [47].
Moreover, trials merging TMS and brain-imaging methods have demonstrated that HF-rTMS can induce changes on dopamine release in the striatum [48][49][50][51][52]. In the context of MDD, Pogarelli and colleagues [52] described an induction of endogenous dopamine release on single photon emission computed tomography (SPECT) of patients with moderate to severe depression, as a result of a three week HF-rTMS standard treatment over the left dlPFC. Considering these findings, the rTMS impact on DD behavior of depressed individuals needs to be carefully assessed by further studies, since relation between increased striatal dopaminergic activity and immediacy has been previously described [12,14].
As a secondary outcome, we have detected a significant decrease in baseline cognitive impulsivity. This result is in accordance with previous findings on the promising effect of neuromodulation on impulsive behavior [18,23]. Stimulation sessions also showed a significant effect on MDD symptomatology. Clinical improvement was both perceived by the specialist (MADRS) and reported by the patients (QIDS-SR16). This effect is in accordance with solid literature on the field [19,20].
Correlation analysis revealed that the more the participants scored on the non-planning-impulsivity subscale, the steeper they discounted on large delayed rewards. A noteworthy observation is that not all BIS-10 subscales were correlated to DD behavior, and the significant improvement on the cognitive-impulsivity subscale occurred independently from other impulsivity dimensions. These results reinforce the hypothesis of impulsive behavior as a multifaceted complex phenomenon [53,54], and that tools employed to assess its dimensions probably reflect separate underlying processes, which results in a general lack of intercorrelation [13,17,55,56]. Regarding the BFI-Fr personality factors, data suggested that the less the subjects scored on characteristics of agreeableness, the steeper they discounted on medium delayed rewards. This negative correlation is in accordance with previous findings [57], albeit little attention has been given in the literature to the relation between these variables.
Since high discount rates may interfere with long-term plan fulfillment and quality of life, further investigation is necessary to clarify the effect on DD associated with rTMS application in MDD. Randomized controlled trials with larger population samples should be performed in order to strengthen internal and external validity. In addition, comparison between online and offline scores and/or association with brain imaging techniques (SPECT, fMRI) may lead to a better understanding on rTMS effects. Other targets could also be explored in accordance with depression biotypes and their particular connectivity dysfunction [47] as therapeutic alternatives for both depression and impulsivity. Lastly, given the complex nature of impulsivity, researchers could consider combining instruments to simultaneously measure other impulsive dimensions (such as response inhibition, planning) and obtain a more comprehensive assessment of the phenomenon.

Conclusions
Despite the limitations involved in this study, the results allow a preliminary evaluation of rTMS effect on temporal impulsivity in depressed individuals meeting treatment-resistant criteria. Nevertheless, further research concerning MDD and DD is required. Merging TMS with brain-imaging methods, performing randomized controlled trials, and associating measure instruments on impulsivity may help us better understand the effect of neuromodulation on temporal impulsivity in depression.