Transcranial magnetic stimulation

Transcranial magnetic stimulation to treat 101 patients with depression and comorbid personality disorders in a real-world naturalistic clinical setting: feasibility, tolerability and effectiveness

Highlights

•Standard treatments often fail in MDD; comorbid personality disorders may impede treatment response.

•This naturalistic study assessed feasibility and efficacy of accelerated vs. standard iTBS in 101 patients with MDD and PD.

•Both iTBS protocols significantly reduced depressive symptoms, independent of PD status.

•Remission rates were 87% for accelerated and 63% for standard iTBS.

•iTBS was well tolerated; younger and antidepressant-naïve patients showed better outcomes.

•Comorbid PD did not negatively affect antidepressant efficacy of either iTBS protocol.

Abstract

Introduction

Many patients with depression fail to achieve adequate response or remission with standard treatments, prompting an urgent need to identify factors influencing treatment outcomes. Among these, comorbid personality disorders (PDs) have been identified as a significant contributor to poorer treatment responses.

Methods and materials

In this open-label, naturalistic study, we evaluated the feasibility, tolerability, and effectiveness of intermittent theta burst stimulation (iTBS) in a sample of 101 patients with depression and comorbid PD. Eighty-six patients received a standard iTBS protocol (one session per day over several weeks), while fifteen opted for an accelerated protocol involving two to four sessions per day. Symptom severity was assessed using the Beck Depression Inventory (BDI) and the Symptom Checklist-90-R (SCL-90-R), administered at baseline, weekly during treatment, and at four follow-up points (1, 2, 3, and 4 months post-treatment).

Results and conclusion

Both treatment protocols led to significant symptom reduction, which was sustained over the follow-up period. Remission rates were 87 % in the accelerated group and 63 % in the standard protocol group. iTBS was well tolerated, with no serious adverse events reported. Notably, patients who were not taking antidepressant medication and those who were younger showed better treatment responses. iTBS appears to be a feasible, safe, and effective treatment option for depressed patients with comorbid PD. The findings suggest the benefits of iTBS treatment and support further exploration of patient characteristics that may predict response to rTMS.

1. Introduction

Major Depressive Disorder (MDD) is characterized by diverse physical, psychological, cognitive, and psychomotor disturbances, affecting an estimated 5 % of the global adult population (American Psychiatric Association, 2000; World Health Organization, 2023). By 2030, MDD is projected to become the leading cause of disability worldwide (World Health Organization, 2022). The first line of treatment for MDD involves a combination of pharmacological medication and psychotherapy (Malhi et al., 2020). However, the clinical efficacy of pharmaceutical treatments is highly variable and often accompanied by side effects (Penn & Tracy, 2012). Alarmingly, approximately 30 % of patients do not show improvement after two cycles of treatment (i.e., the period from when a round of treatment ends until the beginning of the next one) and are subsequently diagnosed with treatment-resistant depression (TRD) (Lundberg et al., 2023).

Many patients with MDD fail to achieve adequate response or remission with standard treatments (Kolovos et al., 2017), prompting an urgent need to identify factors influencing treatment outcomes. Among these, comorbid personality disorders (PDs) have been identified as a significant contributor to poorer treatment responses (Berk et al., 2018; Van & Kool, 2018). Despite the profound impact of PDs on MDD prognosis, they are often overlooked in randomized controlled trials, potentially due to the unstable behaviors that characterize PDs, which increase treatment dropout rates (Ingenhoven et al., 2010).

PDs are exemplified by enduring behavioural patterns that deviate significantly from cultural norms and are not better explained by other mental disorders or substance use (American Psychiatric Association, 2000). Globally, PDs have a prevalence of 7.8 % (Winsper et al., 2020), with affected individuals experiencing heightened morbidity and mortality rates (Fok et al., 2012). These disorders are often characterised by a distorted sense of one’s self, which may exacerbate depressive symptoms. For instance, avoidant personality disorder (AVDPD) is characterised by low self-esteem, feelings of inferiority and shame, susceptibility to criticism and rejection, lack of energy and social desirability, and inability to feel pleasure (Young, 2018). Patients with borderline personality disorder (BPD) often experience unstable mood and relationships and feelings of loneliness (Ceresa et al., 2021).

The management of PDs is complex; while psychotherapy has shown promise, particularly for BPD (Kramer et al., 2020), the efficacy of pharmacological treatments remains contentious (Stoffers-Winterling et al., 2021). Interestingly, antidepressants can alleviate depressive symptoms and thereby potentially influencing PD symptoms and diagnosis (Fava et al., 2002), suggesting the possibility of shared pathophysiological mechanisms between MDD and PDs. The presence of comorbid PDs in MDD poses significant challenges for treatment, as it has been associated with poor prognosis (Alnæs & Torgersen, 1997; Grilo et al., 2000; Mulder, 2002) and an increased likelihood of TRD (McIntyre et al., 2023; Young, 2018). The impact of specific PDs on the course of MDD may vary; while BPD is the most common of the PDs co-occurring with MDD (Ceresa et al., 2021), all PDs can serve as predictors of MDD persistence. Among these, BPD is the most frequent predictor of MDD persistence, followed by schizoid, schizotypal, histrionic, and AVDPD (Skodol et al., 2011).

Repetitive transcranial magnetic stimulation (rTMS) is well-established as a treatment for MDD due to its capacity to modulate cortical excitability either directly, though excitation or inhibition of targeted cortical areas, or indirectly via broader interconnected neural networks (Hallett, 2007). However, the presence of comorbid PDs can complicate the treatment of depression (Newton-Howes et al., 2006). The efficacy of rTMS in patients with both MDD and PD remains underexplored. While some studies have reported clinical improvements in MDD patients with comorbid BPD (Feffer et al., 2022; Kafas & Umla-Runge, 2024), most of the existing literature focuses specifically on this subtype. Furthermore, clinical trials often exclude participants with psychiatric comorbidities, including PDs, to reduce confounding variables. Although this approach improves internal validity, it significantly limits the generalizability and real-world applicability of findings (Blanco et al., 2008). In routine clinical practice, patients with MDD frequently present with TRD and/or other comorbidities such as PDs, emphasizing the need for more inclusive research. Further studies are needed to evaluate the effectiveness of TMS in MDD with comorbid PDs.

In light of these challenges, our naturalistic, open-label study primarily aimed to investigate the effects of iTBS on depressive symptoms in MDD patients with comorbid PDs. Participants chose between a standard six-week one session per day iTBS protocol and an accelerated iTBS protocol with multiple sessions, ranging from two to four per day. We also evaluated the safety, tolerability, and practical feasibility of both the standard and accelerated iTBS protocols for treating MDD with comorbid PD.

2. Method

2.1 Procedure

Our sample consisted exclusively of outpatients who had received an MDD diagnosis with comorbid PD. All patients provided informed consent at the Medical Psychotherapeutic Center (ΙΨΚ) in Thessaloniki, Greece before undergoing treatment. The treatment sessions were conducted at ΙΨΚ between 2018 and 2023. Considering that this was an open-label naturalistic study in a clinical population sample, we were unable to control for variables such as age, sex, medication use and psychotherapy before treatment, and therefore included those as potential covariates in our statistical analyses.

2.2 Participants

2.2.1 Eligibility criteria

Data from patients were included in the study if they met the following inclusion criteria: 1. DSM-5 criteria for MDD as their primary diagnosis; 2. ICD-10 criteria for a depressive episode (F32) or recurrent depressive disorder (F33); 3. DSM-5 criteria for PD; and 4. agreed not to modify their medications during the two weeks prior and the period of iTBS treatment.

Exclusion criteria were: 1. primary diagnosis other than depression; 2. standard rTMS contraindications (Rossi et al., 2021); 3. use of medication known to lower seizure threshold; 4. co-initiation of new medication; 5. high-dose benzodiazepines (≥4 mg daily lorazepam equivalent) (Hunter et al., 2019).

Before rTMS therapy, patients completed the Beck Depression Inventory (BDI) (Beck et al., 1961) and the Symptom Checklist 90-R (SCL-90-R) (Derogatis, 1992) to assess severity of depressive symptoms. Patients from the standard group completed the BDI and SCL-90-R up to 11 time points: once at baseline (pretest), weekly (up to 6 weeks) on Fridays during the treatment period, and at four follow-up assessments approximately 1, 2, 3, and 4 months post-treatment. For follow-up assessments, a window of ±5 days was permitted to accommodate scheduling constraints. Patients in the accelerated group completed the BDI and SCL-90-R up to eight time points: once at baseline (pretest), weekly on Fridays during the treatment period (up to three weeks), and at four follow-up assessments conducted approximately 1, 2, 3, and 4 months post-treatment. Four patients received four iTBS sessions per day and completed assessments after every 10 sessions. Two of these patients completed 36 and 40 sessions, respectively; however, for consistency and comparability with the remaining 99 patients, we included only their data up to the 30th session in the analyses.

2.2.2 Sample demographics

Table 1 depicts the demographic characteristics of the patient sample, which included 101 patients (45 females), ranging in age from 18 to 81 years. Of these, 5 patients were experiencing a first episode of depression, while the remaining 96 were treated for TRD. All patients received iTBS treatment, and 81 also had psychotherapy. Psychotherapy types varied (e.g., Cognitive Bevarioural Therapy, Cognitive Analytic Therapy, Intensive Short Term Dynamic Psychotherapy etc.) and were not systematically recorded.

Characteristic	Standard protocol (N = 86)	Accelerated protocol (N = 15)	All patient sample (N = 101)
Age (M/SD)	42.98 (14.88)	43.06 (15.10)	43.00 (14.84)
Sex (males/females)	45/41	11/4	56/45
Baseline BDI score (M/SD)	28.23 (10.02)	23.60 (8.45)	27.54 (9.90)
No. of sessions (M/SD)	25.23 (5.91)	24.13 (5.46)	25.06 (5.83)
Mean TMS dose (120%) (M/SD)	44.14 (7.88)	44.46 (8.79)	44.18 (7.98)

Table 1

Demographic characteristics of the patient sample that comprised the standard and accelerated protocols, and the overall patient sample.

Antidepressant medication type	No. of patients taking medication in the standard iTBS protocol	No. of patients taking medication in the accelerated iTBS protocol	Total No. of patients
SSRI or SNRI	26	5	31
SSRI or SNRI augmented by other class (e.g., antipsychotics, mood stabilisers)	22	3	25
Other (e.g., antipsychotics, mood stabilisers)	10	0	10
Subtotal	58	8	66
None	28	7	35
TOTAL	86	15	101

Table 2 Classes of medication prescribed for depression for the patient sample that comprised the standard and accelerated protocols, and the overall patient sample.

Personality Disorder	All patients	No. of patients with PD type for the standard iTBS protocol	No. of patients with PD type for the accelerated iTBS protocol
Paranoid	0	0	0
Schizoid	1	0	0
Schizotypal	1	1	0
Antisocial	0	0	0
Borderline	3	2	1
Histrionic	10	8	2
Narcissistic	2	2	0
Avoidant	3	3	0
Dependent	1	1	0
Obsessive-Compulsive	1	2	0
Mixed	79	67	12

Table 3
Distribution of personality disorders for all patients and by protocol type.

iTBS Protocol	Response	Remission
Accelerated (N = 15) Males (N = 11) Females (N = 4)	80% 73% 100%	87% 82% 100%
Standard (N = 86) Males (N = 45) Females (N = 41)	57% 64% 49%	63% 66% 55%
All patients (N = 101)*	64%	66%
PD cluster
Cluster A (N = 42)	61%	64%
Cluster B (N = 67)	60%	63%
Cluster C (N = 77)	58%	62%

Table 4
Response and remission rates based on BDI scores at end of treatment, by iTBS protocol, sex, and PD cluster.
Note:
* Includes all patients treated with iTBS, regardless of protocol type (standard or accelerated).

2.3 Materials

2.3.1 TMS

Patients included in our open-label naturalistic study had been treated at the Medical Psychotherapeutic Centre. rTMS was administered using a MagVenture R30 stimulator (MagVenture A/S, Farum, Denmark) and a cooled MC-B70 figure-eight coil (inner radius = 10 mm, outer radius = 50 mm).

The TMS coil was positioned at a 45-degree angle to midline over the left dorsolateral prefrontal cortex (DLPFC), which was localized using the Beam F3 method with the handle pointing backwards (Beam et al., 2009). To determine the resting motor threshold (RMT), single TMS pulses of gradually decreasing intensity were applied over the left motor cortex. The RMT was defined as the minimum stimulation intensity that produced a visible muscle contraction in the first dorsal interosseous (FDI) muscle of the right hand in at least 5 out of 10 consecutive trials. For all patients, the stimulation intensity of the TMS treatment was set at 120 % of the RMT.

2.3.2 iTBS protocol

Each intermittent theta-burst stimulation (iTBS) session consisted of 600 pulses, delivered in bursts of three pulses at 50 Hz, repeated at a frequency of 5 Hz. These bursts were administered in 2-second trains, followed by 8-second inter-train intervals, for a total of 20 trains per session. This protocol is consistent with standard clinical iTBS procedures. All patients (N = 101) were offered a choice between a standard (non-accelerated) and an accelerated iTBS protocol, depending on personal preference and logistical considerations such as work obligations or travel distance. The majority of patients (86 out of 101) chose the standard protocol, receiving one session per weekday over a period of 4–6 weeks, resulting in a total of 20–30 sessions. Fifteen patients opted for the accelerated protocol. Among these, 11 received two sessions per day with a 50-minute break in between, over a period of 2–3 weeks, resulting in a total of 20–30 sessions. Four patients, who were living abroad, completed four sessions per day with 50-minute breaks between each, over a period of one week to 2 weeks resulting in a total of 20–40 sessions.

2.3.3 Beck depression inventory

The BDI was used to evaluate patients’ response to iTBS and remission of depressive symptoms. BDI is a 21-item self-report questionnaire with each item scored on a scale from 0 to 3, with a maximum score of 63. Scores of 29 or higher indicate severe depression (Beck et al., 1961), while remission is defined as a score below 13 (Mann, 2005).

2.3.4 Symptom checklist 90-R

The SCL-90-R (Derogatis, 1992) is a 90-item self-report questionnaire assessing nine psychopathological dimensions, including depression and anxiety. Items are scored on a 5-point scale from 0 (Not at all) to 4 (Extremely), reflecting symptom frequency over the past week. For this study, only the depression dimension was analyzed. While reliable, the validity of the SCL-90-R has been questioned (Holi, 2003).

2.3.5 International personality disorder examination (IPDE)

The International Personality Disorder Examination (IPDE) is a 77-item, semi-structured clinical interview used to assess personality disorders (PDs) as defined by DSM-IV (American Psychiatric Association, 2000). PDs are categorized into three clusters: Cluster A (odd-eccentric), which includes paranoid, schizoid, and schizotypal PDs; Cluster B (dramatic), which includes antisocial, borderline, histrionic, and narcissistic PDs; and Cluster C (anxious), which includes obsessive-compulsive, avoidant, and dependent PDs. The IPDE evaluates PD types on a continuum and determines whether diagnostic criteria are met. All patients completed the full IPDE.

2.4 Statistical analysis

A linear mixed model (LMM) analysis was used to assess iTBS treatment effects on depression levels over time, measured by BDI scores and the depression subscale of the SCL-90-R. Time (pre-treatment vs. post-treatment) was a within-subject factor, while Protocol Type (accelerated vs. standard) was a between-subject factor. Covariates included: sex, age, total number of classes of medications, number of benzodiazepines, adjunct psychotherapy, TMS dosage (i.e., stimulation intensity), and number of sessions. As our data lacked independency, in view of our repeated measures design, and due to the fact that the two protocol groups were unbalanced, we chose LMM as the preferred analysis method. LMMs can model unequal sample sizes and are also appropriate for continuous outcomes, while they allow for random effects, accounting for unexplained variance within our patient sample. Data was normally distributed as skewness and kurtosis for both BDI (skewness =.59; kurtosis = −.29). and SCL-90-R (SCL-90-R skewness =.52; kurtosis =.15) were within normal ranges. Q-Q plots indicated that both assumptions of normality and linearity were met for BDI and SCL-90-R. PD effects were analyzed through four models examining PD clusters and individual PD diagnoses. A backwards selection process simplified the initial comprehensive model, retaining relevant fixed factors; PD clusters (centered); and interaction terms involving Time, Protocol, and PDs. Individual PD models included centered PD diagnoses and interaction terms up to four-way interactions, with BPD and AVDPD prioritized due to prevalence. Antisocial PD was excluded due to insufficient observations. Post-hoc Bonferroni-corrected pairwise comparisons identified time points of significant changes between protocols (p < .05). We used a Likelihood Ratio (RT) test and chose the Bayesian Information Criterion (BIC) for selecting the best fitting model. A random intercept of person was fitted to the data by default to account for individual variation in baseline scores, but a random slope for time was removed due to too few observations. In addition, we conducted post hoc independent t-tests to explore pre-, and post-treatment changes in patients’ depression levels, as measured by the BDI, between the two protocol groups (i.e., accelerated vs. standard). Paired student’s t-tests were conducted for investigating pre-, and post- treatment changes in BDI scores for each PD cluster and each PD separately. A correlation analysis and a one-way ANOVA examined practical feasibility as well as the association between TMS dosage and medication prescription. All analyses were conducted using JASP v.0.18.3 with a standard covariance matrix for random effects.

3. Results

3.1 Safety and tolerability

iTBS treatment was overall well tolerated, and our clinic adhered to the updated safety guidelines for the clinical and research use of TMS (Rossi et al., 2021). Among 389 iTBS sessions, 11 instances of headache occurred: 10 were mild to moderate and one was severe, requiring treatment with 1 g of paracetamol (acetaminophen) daily for one week. Tolerability was monitored daily on-site and by phone over weekends by a psychiatrist. No serious adverse events were reported, including seizures or manic/hypomanic episodes. Three patients (2.9 %) discontinued treatment. One patient in the standard protocol group withdrew during the third treatment week after refusing to continue care under an alternative psychiatrist, as their primary psychiatrist, who was performing iTBS therapy, was on leave. Two patients discontinued TMS treatment due to psychiatric hospitalization, before the end of the first week of their treatment. Specifically, one patient was hospitalized following suicidal ideation that emerged after the end of a romantic relationship. The second patient was admitted for alcohol intoxication after being dismissed from his work. Both were evaluated by treating psychiatrists, who determined the hospitalizations were unrelated to TMS.

The mean stimulation intensity, expressed as a percentage of stimulator output, was comparable between the two treatment protocols. Patients in the standard iTBS group received an average maximum stimulator output (MSO) intensity of 44.14 % (SD = 7.88), while those in the accelerated group received an MSO of 44.46 % (SD = 8.79), indicating no meaningful difference in dosing levels. All participants were stimulated at 120 % of their individual resting motor threshold (rMT), meaning that % MT was held constant across individuals. However, as individual rMTs vary, the corresponding % of stimulator output required to reach 120 % rMT also varied. Since higher stimulator output may increase the likelihood of physical discomfort (e.g., scalp sensations, auditory intensity), reporting absolute stimulator output remains relevant for assessing tolerability. Stimulation intensity was not associated with adverse events or treatment discontinuation. Further analyses exploring the relationship between stimulation intensity and treatment response are reported below (see 3.6).

3.2 Practical feasibility

All patients (N = 101) were offered a choice between the standard and accelerated iTBS protocols. Eighty-six patients (86 %) chose the standard protocol, which involved one session per weekday over four to six weeks. Fifteen patients (15 %) opted for the accelerated protocol, receiving two to four sessions per day, typically due to work obligations or travel distance. This distribution reflects the role of logistical factors in patient decision-making.

Both protocols were well adhered to. Only three patients (2.9 %) discontinued treatment, two during the first week, none due to tolerability issues. Completion rates were high across both groups, indicating that standard and accelerated iTBS protocols are both feasible in a routine outpatient setting.

3.3 Response and remission rates

Treatment outcomes were evaluated using the BDI. A treatment response was defined as a reduction of 50 % or more from baseline, and remission as a post-treatment BDI score below 13. Across the full sample (N = 101), 64 percent of patients met criteria for response and 66 percent for remission at the end of treatment.

To assess outcomes in patients with moderate to severe depression, those with mild baseline symptoms (BDI ≤18) were excluded. In the resulting subsample (N = 76), 58 percent achieved both response and remission following treatment.

Follow-up assessments at one, two, three and four months post-treatment indicated that symptom reductions were largely sustained over time. This pattern was observed across both the BDI and the depression subscale of the SCL-90-R (see Figs. 1 and 3).

Fig. 1 Change in patients’ BDI scores from baseline to end of treatment [week 3 for the accelerated group (left panel); week 6 for the standard group (right panel)] and from baseline to post-treatment follow-up month (first through fourth post months) assessments after the end of each patient treatment. Dots represent individual patient scores; lines indicate group means with standard error bars.

Fig. 2 Change in BDI scores across treatment sessions for the accelerated and standard iTBS protocols, shown separately. The x-axis reflects the number of sessions since start of treatment. The accelerated protocol group (left panel) was assessed at baseline and after every 10 sessions (sessions 10, 20, 30). The standard protocol group (right panel) was assessed at baseline and every 5 sessions. Dots represent individual patient scores; lines indicate group means with standard error bars.

Fig. 3 Change in patients’ SCL-90-R scores from baseline to end of treatment [week 3 for the accelerated group (left panel); week 6 for the standard group (right panel)] and from baseline to post-treatment follow-up month (first through fourth post months) assessments after the end of each patient treatment. Dots represent individual patient scores; lines indicate group means with standard error bars.

3.4 BDI outcomes: effects of iTBS treatment

To examine changes in depressive symptoms over time, we conducted two linear mixed models (LMMs) using BDI scores as the outcome variable. The first model included PD clusters as fixed effects, and the second model examined specific PD diagnoses

3.4.1 Model 1: PD clusters

The first LMM (BIC = 4992.008) evaluated BDI score changes over time while accounting for treatment protocol, medication use, and PD cluster (see Table 5). There was a significant main effect of time (χ² = 369.275, df = 10, p < .001), indicating a reduction in depression severity across the treatment period. A significant effect of protocol type (χ² = 4.833, df = 1, p = .028) showed that patients in the accelerated protocol had lower overall BDI scores compared to those in the standard protocol. Medication use also had a significant effect (χ² = 15.596, df = 6, p = .016), with patients taking three medication classes reporting higher BDI scores.

Fixed Effect	df	X2	p
Time	10	369.275	< .001**
Protocol type	1	4.833	.028*
Medications	6	15.596	.016*
Cluster A	1	0.077	.781
Cluster B	1	0.079	.779
Cluster C	1	0.799	.372

Table 5

Fixed Effects from the Linear Mixed Model examining BDI scores over time with PD clusters, protocol type and medication use as predictors.

Note:

*p value is significant at < .05

**p value is significant at < .001

There were no significant effects for PD cluster (Cluster A: p = .781; Cluster B: p = .779; Cluster C: p = .372). Other covariates, including sex, age, number of benzodiazepines, adjunct psychotherapy, stimulation intensity, and number of sessions, were non-significant (all ps > .05).

Bonferroni-corrected post-hoc comparisons revealed a significant reduction of 8.50 BDI points (36.4 %) between baseline and the second assessment (z = 10.417, p < .001). The effect of time remained significant at all subsequent time points. On average, patients in the standard protocol scored 5.50 points (30.6 %) higher than those in the accelerated group (z = 10.331, p < .001). Patients taking three medication classes scored 10.71 points (58.3 %) higher than those not taking medication (z = –3.635, p = .002).

3.4.2 Model 2: specific PD diagnoses

A second LMM (BIC = 5114.577) examined BDI score changes while including individual PD diagnoses as fixed effects (Table 6). Again, time showed a strong main effect (χ² = 195.053, df = 7, p < .001), and medication use remained significant (χ² = 14.488, df = 6, p = .025). A significant three-way interaction between time, protocol, and BPD diagnosis was detected (χ² = 20.402, df = 7, p = .005), though this effect did not remain significant after Bonferroni correction.

Fixed Effect	df	X2	p
Time	7	195.053	< .001**
Medications	6	14.488	.025*
Time*BPD	7	9.072	.248
Protocol*Time	7	9.736	.204
ProtocolTimeBPD	7	20.402	.005*

Table 6

Fixed effects from the linear mixed model examining BDI scores over time with individual PD diagnoses, protocol type, and medication use as predictors.

Note:

*p value is significant at < .05

**p value is significant at < .001

Post-hoc analysis showed a significant reduction of 9.92 BDI points (29.4 %) between baseline and the second assessment (z = 8.895, p < .001). Patients not taking any medication scored 10.08 points (60.7 %) lower than those taking three classes of medication (z = –3.387, p = .004). As in Model 1, covariates such as protocol type, sex, age, benzodiazepine use, psychotherapy, stimulation intensity, and number of sessions were not significant (all ps > .05).

3.5 SCL-90-R outcomes: effects of iTBS treatment

To evaluate changes in general psychological distress, we conducted two linear mixed models (LMMs) using SCL-90-R depression subscale scores as the outcome variable.

3.5.1 Model 1: PD clusters

The first model (BIC = 5105.698) included PD clusters, treatment protocol, medication use, and age as fixed effects (see Table 7). There was a significant main effect of time (χ² = 391.093, df = 10, p < .001), showing that depression symptoms decreased over the treatment period. A significant effect of protocol type (χ² = 9.849, df = 1, p = .002) indicated that patients in the accelerated protocol had lower overall SCL-90-R scores compared to those in the standard protocol. Medication use (χ² = 16.340, df = 6, p = .012) and age (χ² = 7.668, df = 1, p = .006) were also significant predictors. Cluster C traits had a small but significant effect (χ² = 5.345, df = 1, p = .021). Bonferroni-corrected post-hoc analyses confirmed a significant reduction of 6.78 points (24.2 %) in SCL-90-R scores between baseline and the second assessment (z = 7.583, p < .001), with sustained improvement across follow-up assessments (Figs. 3 and 4). Patients in the standard protocol group scored 7.81 points (38.9 %) higher than those in the accelerated group (z = 3.207, p = .001). Patients scoring low on Cluster C traits showed significantly greater reductions in SCL-90-R scores than those with average or high Cluster C scores (37.62-point and 39.70-point differences, respectively; both p < .001). Medication use was again associated with outcome: patients taking three medication types scored 9.50 points (45.3 %) higher than those not on medication (z = –3.280, p = .006). Age was also a significant predictor: patients in the youngest group (mean age = 28.2) reported higher SCL-90-R scores than those in the middle-aged (mean age = 42.6; difference = 2.37 points, z = 2.821, p = .010) and older group (mean age = 57.1; difference = 4.57 points, z = –3.280, p = .006). Other covariates, including sex, benzodiazepine use, psychotherapy, stimulation intensity, and number of sessions, were not significant (all ps > .05).

Fig. 4 Change in SCL-90-R scores across treatment sessions for the accelerated and standard iTBS protocols, shown separately. The x-axis reflects the number of sessions since start of treatment. The accelerated protocol group (left panel) was assessed at baseline and after every 10 sessions (sessions 10, 20, 30). The standard protocol group (right panel) was assessed at baseline and every 5 sessions. Dots represent individual patient scores; lines indicate group means with standard error bars.

Fixed Effect	df	X2	p
Time	10	391.093	< .001**
Protocol type	1	9.849	.002*
Medications	6	16.340	.012*
Age	1	7.668	.006*
Cluster C	1	5.345	.021*

Table 7

Fixed effects from the linear mixed model predicting SCL-90-R depression subscale scores over time, with protocol type, medication use, age, and PD clusters as predictors.

Note:

*p value is significant at < .05

**p value is significant at < .001.

3.5.2 Model 2: specific PD diagnoses

The second LMM (BIC = 5300.593) evaluated SCL-90-R changes over time while including individual PD diagnoses (see Table 8). There were significant effects of time (χ² = 166.036, df = 8, p < .001), medication use (χ² = 15.351, df = 6, p = .018), and age (χ² = 6.641, df = 1, p = .010). AVDPD traits showed a trend-level effect (χ² = 3.833, df = 1, p = .050), and a significant time × AVDPD interaction (χ² = 20.135, df = 10, p = .028). A three-way interaction between time, protocol, and BPD diagnosis was also observed (χ² = 17.082, df = 8, p = .029), though it did not survive Bonferroni correction.

Fixed Effect	df	X2	p
Time	8	166.036	< .001**
Medications	6	15.351	.018*
Age	1	6.641	.010*
AvD PD	1	3.833	.050
Time*AvD	10	20.135	.028*
TimeBPDProtocol	8	17.082	.029*

Table 8

Fixed effects from the linear mixed model predicting SCL-90-R depression subscale scores over time, including individual personality disorder diagnoses and their interactions.

Note:

*p value is significant at < .05

**p value is significant at < .001.

Post-hoc comparisons revealed a reduction of 15.89 points (66.3 %) in SCL-90-R scores at Time 8 compared to baseline (z = 10.016, p < .001), and a 15.88-point reduction (46.3 %) at the two-month follow-up (z = 9.365, p < .001; Figs. 4 and 5). Patients taking four types of medication scored 8.57 points (92.2 %) higher than those not on any medication (z = 2.937, p = .017). Younger patients reported significantly higher SCL-90-R scores than older individuals, with the youngest group scoring 2.19 points more than the middle-aged group (z = 2.618, p = .027) and 4.39 points more than the oldest group.

Fig. 5 Bar plots of BDI scores at pre-treatment and post-treatment for the standard (left) and accelerated protocols (right). Whiskers indicate the full data range. Asterisks denote statistical significance.

Post-hoc contrasts further supported the time × AVDPDinteraction: patients low in AVDPDtraits showed a 43.59-point reduction in SCL-90-R scores between baseline and the second assessment (z = 14.268, p < .001), and a 19.80-point reduction at the one-month follow-up (z = 9.930, p < .001). No other covariates were significant (ps > .05).

3.6 Impact of personality disorder traits on treatment efficacy

We examined whether PD traits affected treatment outcomes using linear mixed models. PD clusters (A, B, C) had no significant effects on BDI outcomes (Table 5). However, Cluster C traits were significantly associated with higher SCL-90-R depression scores (Table 7), indicating higher symptom burden in these individuals across time.

When considering individual PD diagnoses, AVDPD traits showed a significant interaction with time in predicting SCL-90-R outcomes (Table 8). Patients low in AVDPD traits demonstrated larger symptom reductions, particularly at early follow-up. For BDI, a Time × Protocol × BPD interaction was initially significant but did not survive Bonferroni correction (Table 6).

Together, these findings suggest that, while PD traits had limited influence on core depressive symptoms, they may modestly affect broader psychological distress, particularly in patients with high Cluster C traits or low AVDPD features.

3.7 Protocol group differences in pre- and post-treatment BDI scores

To assess treatment-related changes in depressive symptoms between the standard and accelerated iTBS protocols, we conducted independent and paired-samples t-tests on BDI scores. There were no significant between-group differences at baseline (p = .095). However, post-treatment scores differed significantly: patients in the accelerated group reported lower BDI scores (M = 7.20, SD = 6.13) than those in the standard protocol (M = 13.38, SD = 11.52), t(99) = -2.022, p = .046.

Within-group comparisons revealed significant reductions from pre- to post-treatment in both groups. In the standard protocol, BDI scores declined from M = 28.23 (SD = 10.02) to M = 14.79 (SD = 11.30), t[42) = 8.931, p < .001. In the accelerated group, scores dropped from M = 23.60 (SD = 8.45) to M = 7.20 (SD = 6.13), t(14) = 6.503, p < .001 (Fig. 5).

We additionally examined sex-based differences. At baseline, females reported significantly higher BDI scores (M = 30.02, SD = 10.56) than males (M = 25.55, SD = 8.94), t(99) = 2.301, p = .023. Post treatment analyses revealed no significant differences between sexes (p > .05). However, both groups showed significant within-subject improvements. Males improved from M = 25.55 (SD = 8.94) to M = 10.76 (SD = 10.37), t(55) = 12.034, p < .001; females improved from M = 30.02 (SD = 10.56) to M = 14.57 (SD = 11.70), t(44) = 10.032, p < .001.

Mean BDI differences by PD cluster and individual PD diagnoses are reported in Table 9.

PD Clusters/PDs	Pre-treatment (M, SD)	Post-treatment (M, SD)	% Mean Difference	df	t	p
Cluster A	30.71 (10.05)	15.71 (12.50)	48.84	23	6.808	< .004
Cluster B	29.71 (10.10)	14.86 (10.86)	49.98	27	8.178	< .001
Cluster C	30.49 (9.16)	15.54 (11.59)	49.03	36	8.532	< .001
Paranoid	30.31 (10.12)	18.13 (13.99)	40.18	15	4.442	< .001
Schizoid	31.18 (8.14)	15.82 (9.85)	49.26	16	5.671	< .001
Schizotypal	29.78 (7.59)	14.56 (8.35)	51.10	8	4.069	.004
Histrionic	27.42 (10.09)	9.75 (8.23)	64.44	11	6.522	< .001
Narcissistic	30.67 (14.67)	14.67 (15.70)	52.16	8	3.913	.004
Borderline	31.27 (10.36)	16.14 (10.93)	48.38	21	6.742	< .001
Antisocial	23.33 (1.52)	11.33 (10.06)	51.43	2	2.110	.169
OCPD*	28.79 (8.66)	10.54 (7.49)	63.39	23	8.955	< .001
Dependent	30.42 (11.61)	15.47 (11.87)	49.14	18	5.969	< .001
Avoidant	30.41 (8.80)	15.75 (11.61)	48.20	31	7.629	< .001

Table 9

Pre- and post-treatment differences at BDI scores between per cluster PD and each PD.

Note: Obsessive-compulsive personality disorder

4. Discussion

With this open-label naturalistic study, we aimed to assess the practical feasibility, tolerability and clinical effectiveness of TMS to treat depressive symptoms in patients suffering from depression with comorbid PD in a real-world clinical setting. Replicating previous research (Chen et al., 2021; Fitzgerald et al., 2018) our clinical data showed that TMS therapy, both standard as well as accelerated using iTBS, is highly tolerable in a heterogeneous real-world patient population. Tolerability was examined by daily self-reported side effects. Practical feasibility was defined by the compliance degree with each of the two protocols. Only three out of 101 patients terminated their treatment earlier than scheduled. Hence, these observations suggest that both TMS protocols were safe and tolerable for our group of patients.

Our results demonstrated a significant reduction in reported depression levels as early as two weeks into TMS for both protocol groups. However, considering the limited number of patients in the accelerated group (N = 15), larger clinical trials are needed in order to establish the effectiveness of accelerated iTBS. Interestingly, the total number of sessions was not predictive factor of change in depression levels in our patient sample. This finding is inconsistent with previous literature that has shown significant associations between TMS sessions and clinical improvement (Hutton et al., 2023). This observation may be due to the fact that all patients did a sufficient number of sessions (on average 25 sessions for all patient sample, and each protocol type).

Several factors may influence the current outcomes. Specifically, the standard iTBS protocol resulted in 64 % of the patients responding and 66 % remitting, whereas the accelerated iTBS group reached higher rates in terms of response (80 %) and remission (87 %). However, since we did not randomize patients into one of these two protocol types, the selective and smaller group of patients who opted for an accelerated protocol may be differently motivated or characterized by different expectations and thus represent a certain selection bias. For instance, participants who travelled to receive iTBS treatment may have had higher expectations and hoping for faster results, leading to a potential higher placebo effect.

Differences in baseline depression severity may also have influenced our findings. Some studies suggest that greater severity prior to rTMS is associated with stronger treatment effects (Bouaziz et al., 2023; Taylor et al., 2017), whereas others report the opposite, that higher baseline severity predicts poorer outcomes (Trevizol et al., 2020). In our sample, patients in the standard protocol group reported higher baseline depression levels than those in the accelerated group. Although this difference was not statistically significant, it may nonetheless have contributed to observed differences in outcomes. As LMMs analysis revealed a main effect of protocol type, but not a protocol by time interaction, this also suggests that the accelerated group may have lower scores across all time points, rather than the accelerated protocol being more effective than the standard one.

Age may have influenced current observations. Studies have suggested that depression tends to decrease with age (Kessler et al., 2010), while younger patients may respond better to rTMS therapy (Rostami et al., 2017); although others have observed the reverse outcome (Fitzgerald et al., 2016). Older individuals may respond better to the standard rTMS protocol than iTBS (Slan et al., 2024). Thus, clinicians should consider age when choosing and evaluating depression treatments. Sex may have mediated our results. Prior studies have suggested that females are more likely than males to respond to rTMS and achieve remission (Sackeim et al., 2020), possibly reflecting greater sensitivity to neuromodulation. Nevertheless, other studies have not observed any sex differences in rTMS treatment outcomes (Feyrer et al., 2025). In our sample, both sexes showed significant reductions in BDI scores following iTBS treatment. However, our sample may not be fully representative, as males outnumbered females in both protocols. The existing literature suggests that both MDD and BPD (the most prevalent PD in our sample) are more common in females (Qian et al., 2022; World Health Organization, 2023). This may partly reflect the higher likelihood of females seeking psychiatric compared to males. Future studies systemically comparing age and sex differences in rTMS treatment responses are required.

Concerning the effects of comorbid PD diagnosis/type, we took into consideration both BDI and SCL-90-R scores in order to explore any changes in MDD levels in more depth. Initially, when we considered each PD cluster in our analysis, no significant associations were observed between clusters and BDI score changes. However, a significant effect of Cluster C PD diagnosis on SCL-90-R scores was identified, with Cluster C patients reporting increased depression levels as measured by the SCL-90-R scale. However, as no such effect was detected in BDI scores, the gold standard for depression assessment, this result should be viewed with caution.

To explore PD effects on depression levels and MDD treatment, we examined each PD’s impact on BDI and SCL-90-R scores. PD effects on SCL-90-R scores differed slightly. An initial three-way interaction between time, protocol, and BPD disappeared after Bonferroni corrections. Additionally, patients with low AVDPD traits had lower SCL-90-R scores at both the first and second follow-up compared to baseline. AVDPD patients may prefer rTMS over psychotherapy due to difficulty forming therapeutic relationships and avoiding conflict.

4.1 Limitations and future research

Although we aimed to control for several potential confounding variables, including concurrent psychotherapy, antidepressant use, and demographic factors, the naturalistic design of our study limits both its generalizability and replicability. Future research should consider employing controlled experimental designs with clearly defined groups (e.g., sham-control, rTMS, and iTBS), while also accounting for baseline depression severity and treatment-resistant depression (TRD) status.

Another limitation lies in the relatively small sample size, which restricts the statistical power and interpretability of our findings. Although no significant baseline differences in depression severity were observed between the protocol groups, it is important to note that the group sizes were unequal, with most patients receiving the standard iTBS protocol. Future studies with larger and more evenly distributed samples across experimental conditions would offer greater validity and reliability.

Additionally, our findings were influenced by the choice of clinical assessment tools. Discrepancies between outcomes on different depression scales highlight the limitations of self-report measures and underscore the importance of using well-validated, disorder-specific instruments. While the SCL-90-R assesses general psychological distress, we recommend that future research prioritize depression-specific measures such as the Beck Depression Inventory (BDI), the Hamilton Depression Rating Scale (HAM-D) (Hamilton, 1960), or the Montgomery-Åsberg Depression Rating Scale (MADRS) (Montgomery and Åsberg, 1979).

5. Conclusions

Our open-label naturalistic study assessed the feasibility, tolerability, and effectiveness of a repetitive transcranial magnetic stimulation in 101 patients with depression and comorbid PD. The findings indicate that both the standard and accelerated protocols are feasible, well-tolerated and clinically effective in this population. Larger studies are needed to confirm the effectiveness of the accelerated protocol. Given that iTBS requires less time than conventional rTMS protocols, it offers greater flexibility and reduced time burden for both patients and clinicians. These promising results underscore the importance of follow-up research using randomized controlled trial designs.

CRediT authorship contribution statement

Theodoros Koutsomitros: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Panagiota Koutsimani: Writing – review & editing, Writing – original draft, Visualization, Formal analysis. Teresa Schuhmann: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Project administration, Methodology, Investigation, Formal analysis, Data curation. Alexander T. Sack: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Conceptualization.

Ethics statement

Ethics Review Committee Psychology & Neuroscience (ERCPN): ERCPN OZL_279_38_03_2024

Funding

This research did not receive external funding.

Declaration of Competing Interest

ATS is the Chief Scientific Advisor for PlatoScience Medical, Scientific Advisor for Alpha Brain Technologies, Founder and CEO of Neurowear Medical, Director of the International Clinical TMS Certification Course (www.tmscourse.eu), and President of the Academy of Brain Stimulation (www.brainstimulation-academy.com). He has received equipment support from MagVenture, Magstim, and Deymed Diagnostics.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Acknowledgements

We would like to thank all the psychologists and psychiatrists at the Medical Psychotherapeutic Centre in Thessaloniki, Greece, who referred patients for this study.

Appendix A Supplementary material (1)

Document (18.50 KB)

Supplementary material

Data Availability

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

References

1.American Psychiatric Association

Diagnostic and Statistical Manual of Mental Disorders. Fourth

American Psychiatric Association, Washington DC, 2000

You May Also Like

Early Career Psychiatrists Section World Psychiatric Association

Volume 32 – Supplement 2021 – Advances in transcranial magnetic stimulation (TMS) and its applications in resistant depression

+30 2310 83 83 23

+30 6978 777 303