Clinical evaluation of paroxetine in post-traumatic stress disorder (PTSD): 52-week, non-comparative open-label study for clinical use experience


*Masaru Iwasaki, MD, PhD, Development and Medical Affairs Division, GSK Building 6-15, Sendagaya 4-chome, Shibuya-ku, Tokyo 151-8566, Japan. Email:


Aim:  The present study was a 52-week, non-comparative, open-label study of flexible dose paroxetine (20–40 mg) in 52 Japanese post-traumatic stress disorder (PTSD) patients in order to obtain clinical experience regarding efficacy and safety in regular clinical practice.

Methods:  Efficacy was measured using the Clinician-Administered PTSD Scale One Week Symptom Status Version (CAPS-SX).

Results:  The mean change from baseline in CAPS-SX total score was −19.1, −22.8 and −32.3 at weeks 4, 12 and 52, respectively, and that in the Clinical Global Impression (CGI) Severity of Illness score was −1.1 at week 12 and −1.7 at week 52. A total of 46.9% were CGI responders at week 12, while 67.3% were improved on the CGI at week 52. Of 52 subjects who entered into the drug treatment, 25 completed the study. Only one patient withdrew from the study due to lack of efficacy. In patients who were rated as ‘moderately ill’ or less at baseline, the proportion of CGI responders at end-point was higher at a dose of 20 mg/day than at higher doses, whereas in patients rated as ‘markedly ill’ or more, it was higher at 30 and 40 mg/day, suggesting that severely ill patients could benefit from higher doses.

Conclusion:  Paroxetine appeared generally tolerated in short- and long-term use, and the safety profile in this study was consistent with international trials and other Japanese populations (i.e. patients suffering from depression, panic disorder or obsessive–compulsive disorder). Although the study was not conducted in double-blind fashion, the current findings suggest that paroxetine may contribute to clinically meaningful improvement that is maintained during long-term use and is generally well tolerated.

POST-TRAUMATIC STRESS DISORDER (PTSD) is a relatively new diagnostic category that was first formally established in the Diagnostic and Statistical Manual of Mental Disorders (DSM) in 1980. PTSD occurs as the result of exposure to extreme traumatic stress, and is clinically characterized by persistent re-experiencing of the traumatic event, chronic avoidance of stimuli associated with trauma, numbing of general responsiveness, and persistent symptoms of increased arousal.1 PTSD has rapidly gained recognition as a mental disorder in Japan. The mass media frequently feature this disorder when natural disasters, criminal assaults, serious accidents and other traumatic events have happened. Still fresh in our memory are the Great Hanshin Earthquake, the sarin nerve-gas attack on the Tokyo subway system, the curry poison murder in Wakayama, and the stabbing of children in Ikeda Elementary School, and research has been done to report the PTSD arising after exposure to such prominent events.2

PTSD may arise from sexual assaults, traffic accidents and other traumatic events. In Japan, 8% and 16% of the severe traffic accident victims suffered from full and partial PTSD, respectively, 1 month after the incident.3 In the USA the lifetime prevalence of PTSD is estimated to be 5% of the population.4

The recommended treatment for PTSD is cognitive behavioral therapy (CBT), combined with drug therapy for more serious cases.4 In Japan, however, the availability of trained CBT therapists is limited and currently no medication is approved for the treatment of PTSD. At present, common treatments for PTSD in Japan include tricyclic antidepressants, benzodiazepines, antipsychotics, alpha-blockers, and beta-blockers, often administered as a combination of several drugs. Due to the apparent limitations and side-effects of such therapy,5 physicians have recently shifted towards the use of selective serotonin re-uptake inhibitors (SSRI), which are recommended as a first-line drug treatment for PTSD by the Expert Consensus Guideline Series.6

Paroxetine hydrochloride hydrate (paroxetine), a phenylpiperidine compound developed by SmithKline Beecham (now, GlaxoSmithKline), is a member of the class of SSRI and has a very low affinity for other neurotransmitter receptors and re-uptake proteins.7 Paroxetine was first approved in the UK in 1990, and has since then been approved for depression in more than 100 countries. Additionally, in many countries, paroxetine has been approved for the treatment of anxiety disorders, including panic disorder in more than 100 countries; obsessive–compulsive disorder (OCD) in more than 100 countries; social phobia/social anxiety disorder (SAD) in more than 80 countries; generalized anxiety disorder in more than 60 countries; and PTSD in more than 50 countries. In Japan, paroxetine was approved for the treatment of depression/depressive episodes, panic disorder in September 2000 and was launched for these indications in November 2000. Currently, paroxetine is the first line antidepressant for OCD in Japan since its approval in 2006.

Internationally, three key studies have been conducted to examine the efficacy and safety of paroxetine for the treatment of patients suffering from PTSD.8–10

In those studies, clinically and statistically significant reductions in the Clinician-Administered PTSD Scale-2 (CAPS-2) total score relative to baseline in the paroxetine treatment group compared with the placebo treatment group were observed at week 12. In addition, pooled analysis of the data obtained from the three studies together indicate the following: paroxetine improved a variety of PTSD symptoms, such as re-experiencing, avoidance and numbing, and increased arousal. The CAPS total score was statistically significantly lower in patients treated with paroxetine than in those treated with placebo at week 12. Among the three studies, the adverse events (AE) were mild–moderate in severity and, therefore, paroxetine appeared to be a well-tolerated agent for patients with PTSD.

Considering the difficulties and concerns of clinical studies in patients suffering from PTSD, and given that not a single placebo-controlled study on Japanese PTSD patients has been published, the current study did not use placebo and instead used an open design without an active comparator, because currently no drug with an indication for treatment of PTSD is available in Japan.

The main purpose of the present study was to obtain clinical experience regarding the efficacy and safety of paroxetine in flexible dosing during long-term administration (up to 52 weeks) in a regular clinical setting in Japanese patients suffering from PTSD.


Study population

The subjects were 52 adult patients with a primary diagnosis of PTSD according to the DSM-IV1 recruited from 11 centers in Japan. PTSD diagnosis was confirmed on the Clinician-Administered PTSD Scale-DX Current and Lifetime Diagnostic Version (CAPS-DX) and a score of 50 on Criteria B, C and D of the CAPS One Week Symptom Status Version (CAPS-SX) was required for study entry. All investigators underwent training for the CAPS rating before participating in the study.

Subjects with a primarily DSM-IV Axis I diagnosis other than PTSD within 6 months before week −1 were not allowed to enter the study. In addition, subjects with a current major depressive episode that preceded the diagnosis of PTSD were excluded from the study.

The study protocol was prepared in accordance with the Good Clinical Practice on Drugs (GCP), General Considerations for Clinical Trials,11 Guideline for Clinical Evaluation of Anxiolytics,12 and Statistical Principles for Clinical Trials13 and the medical ethics review board of the participating institutions approved this study in line with the requirements of establishing Institutional Review Board (IRB) according to GCP. Written informed consent was obtained from all patients according to the Declaration of Helsinki.

Study design and procedure

Baseline phase

The baseline phase was defined as the week before the first dose of treatment. During this period, subjects received placebo once daily after an evening meal while their eligibility to enter the treatment phase was determined. If subjects had been treated with any prohibited concomitant medication, it was washed out during this period. During the study only six medications, including hypnotics and anxiolytic medications were allowed concomitantly (i.e. brotizolam, etizolam, flunitrazepam, estazolam, zopiclone, and zolpidem).

Treatment phase

All patients received paroxetine starting with 20 mg/day after dinner for the first 2 weeks. If a sufficient clinical response based on the Clinical Global Impression (CGI) Scale was achieved (‘Very much improved’ or ‘Much improved’), the subject continued on the same dose level. When the clinical response was insufficient, the dose was increased to 30 mg/day and then, if necessary, to 40 mg/day at intervals of at least 2 weeks, until a sufficient response was reached and tolerated. Once response was obtained, the treatment was maintained at that dose, although for subjects who received 30 or 40 mg/day, dose reduction to the next lower level (20 or 30 mg/day) consequent to an AE was permitted at the investigator's discretion. The treatment phase was scheduled for a total of 52 weeks.

Taper phase

The medication was tapered off in 10-mg increments per week until dose 20 mg/day was reached, which required no further tapering.

Statistical analysis


The last available on-therapy observation for a patient was used to estimate missing data points (LOCF). Primary efficacy variables were the change from baseline in the CAPS-SX total score calculated as CAPS-SX total score at each timepoint minus CAPS-SX total score at baseline. Secondary efficacy variables were the proportion of responders based on the CGI Global Improvement, the change from baseline in the CAPS-SX re-experiencing cluster score, the CAPS-SX avoidance/numbing cluster score, the CAPS-SX hyperarousal cluster score, the CGI Severity of Illness score, and the proportion of patients remaining in the study.


The proportion of responders based on the CGI Global Improvement score was calculated and stratified by the initial severity of illness and by the dose level at the end of treatment (Table 1).

Table 1.  CGI responders at end-point
CGI Severity of illness (Baseline)Dose at week 52 (mg/day)Subtotal
  • Thirty-three patients rated as ‘very much improved’ and ‘much improved’.

  • CGI, Clinical Global Impression.

Moderately ill or less
n (%)
13 (86.7)2 (50.0)2 (28.6)17 (65.4)
Markedly ill or more
n (%)
4 (66.7)4 (80.0)8 (66.7)16 (69.6)

Safety assessment

All AE (whether or not considered related to the investigational product) and abnormal findings were summarized. The incidence of AE and withdrawals due to AE were calculated. AE with an incidence of ≥5% were tabulated and examined for relationship to the investigational product. Laboratory values and vital signs were collected at weeks −1, 12, 24, 36 and 52.


Study period

The study period was 2 year and 7 months (starting 17 May 2002 with the first subject's first visit and ending on 19 November 2004 with the last subject's last visit).

Study population results

A total of 56 patients gave written consent to participate, and 52 of them entered the treatment phase and received at least one dose of medication. Four patients who gave written consent were withdrawn prior to the treatment phase, three of them because of violation of the inclusion/exclusion criteria and one who was lost to follow up.

The most common reason for withdrawal was AE (n = 10). Four patients were lost to follow up and withdrew prematurely. These patients were subsequently contacted by the investigator and confirmed to be doing well.

Four patients (or their proxy consenters) requested an early discontinuation, three patients had deviations from the protocol, in two patients no further drug therapy was considered necessary due to substantial improvements of PTSD, one patient only took the study medication infrequently due to improvement of his symptoms, one patient was non-compliant for personal reasons, one patient was discontinued early because of concern for self-injurious behavior and one patient withdrew due to lack of efficacy. Three patients were withdrawn from the efficacy analysis due to the fact that one had no primary diagnosis of PTSD and the others had no efficacy assessment during the treatment phase. Therefore, the total number of patients used for the efficacy analysis was 49, while all 52 patients were included in the safety analysis. Of the 52 patients, 25 completed the study and 27 were withdrawn prematurely.

The mean ± SD of age was 35.0 ± 10.6 years (n = 52), and the duration of illness was 58.3 ± 65.9 months (range, 4–358 months). Among the 52 patients, 46 (88.5%) were female and six (11.5%) were male.

The type of traumas leading to PTSD were physical or sexual assault (n = 37), seeing someone hurt of die (n = 6), serious accident or injury (n = 4) and others (n = 2).

At study entry 33 patients received psychotropic drugs and an additional 28 patients received psychotherapy. Nine patients received no treatment. The major type of prior psychotropic medications used before study entry was hypnotics (brotizolam, flunitrazepam, zolpidem, nitrazepam) and anxiolytics (alprazolam, bromazepam, etizolam), tricyclic antidepressant (clomipramine), SSRI (paroxetine, fluvoxamine) and others (trazodone, chlorpromazine). Only one patient had a history of major depression disorder, but 23 suffered from psychiatric comorbidities (mood disorder, n = 11; anxiety disorder, n = 3; duplication of some psychiatric disorders, n = 8; and histrionic personality disorder, n = 1).

Extent of exposure

Among 52 patients, the percentage of patients on each dose level at end-point was 44.2%, 17.3%, and 38.5% for 20, 30 and 40 mg/day, respectively while it was 48.0%, 16.0% and 36.0% among the 25 study completers. The proportion of patients on 40 mg/day was largest at week 16 and declined thereafter. From week 40 onwards and at the end-point, more patients were on 20 mg/day than on higher doses.

Efficacy results

The mean ± SD of baseline CAPS-SX total score was 74.6 ± 18.2. The mean change from baseline (week 0) in CAPS-SX total score, the primary efficacy variable, was −19.1 (95% confidence interval [CI]: −14.1 to −24.2), −22.8 (95%CI: −16.8 to −28.7) and −32.3 (95%CI: −25.3 to −39.3) at weeks 4, 12 and 52, respectively (Fig. 1). The CAPS-SX total score fell considerably over the first 4 weeks of treatment compared to baseline and then continued to decrease up to week 24, with maintenance up to week 52. The proportion of CGI Global Improvement responders (i.e. patients rated as ‘very much improved’ or ‘much improved’) was 46.9% (23/49) and 67.3% (33/49) at week 12 and week 52, respectively. The mean change from baseline in CGI Severity of Illness score was −1.1 (95%CI: −1.3 to −0.8) and −1.7 (95%CI: −2.0 to −1.3) at week 12 and week 52, respectively. Furthermore, only one patient withdrew from the study due to lack of efficacy.

Figure 1.

Changes from baseline in Clinician-Administered Post-Traumatic Stress Disorder Scale One Week Symptom Status Version (CAPS-SX) total score (mean ± SD; last observation carried forward).

Sub-analysis indicated that in patients who were rated as ‘moderately ill’ or less at baseline on the CGI Severity of Illness, the proportion of CGI Global Improvement responders by dose at end-point was higher at a dose of 20 mg/day, whereas in patients who were rated as ‘markedly ill’ or more at baseline, it was higher at 30 and 40 mg/day (Table 1).

A post-hoc analysis was conducted in order to investigate whether patients being drug free at study entry (with respect to psychotropics) benefited more from the treatment, because the positive outcome of the present study might have been partially due to the clinical experience that drug-naïve patients frequently respond better to treatment. As a result, no differences were observed in CAPS-SX total score at the study end among patients who are naïve patients (−34.4; 95%CI: −c47.7 to −21.2), patients who received prior medication (−32.7; 95%CI: −42.1 to −23.2) and prior psychotherapy (−34.3; 95%CI: −44.3 to −24.3). Similarly, no difference was observed in patients who received both prior medication and psychotherapy (−32.6; 95%CI: −45.7 to −19.5).

A similar analysis was conducted in an international study.10 In the analysis of change from baseline in total CAPS-2 score, both subjects who had and had not received prior psychoactive medication had positive treatment benefits on paroxetine, in both the 20- and 40 mg groups. The observed treatment differences from placebo were slightly higher in the subjects who had received prior psychoactive medication.

Safety results

A total of 48 out of 52 patients (92.3%) reported AE after the start of the treatment phase; most frequently reported were nasopharyngitis, nausea, and somnolence (Table 2). Serious AE were reported in nine patients (17.3%), with the most frequently reported being non-accidental overdose (two patients), and exacerbation of PTSD (two patients). No fatal serious AE were reported. The nature of the most common AE (≥5% of patients) reported during the first 12 weeks was consistent with that observed during the whole study period. Twelve AE leading to premature discontinuation of the study were observed in 10 patients (19.2%): among which three patients experienced exacerbation of PTSD, two took a non-accidental overdose, two suffered from depression, and in one patient the AE of hyperthyroidism, somnolence, aggression, mania and self-injurious behavior led to discontinuation. No changes of potential clinical concern were noted for any vital signs or laboratory parameters. During the tapering period four AE occurred, one of them serious (abdominal pain/gastroenteritis), but was not considered to be related to the study procedures, furthermore tremor nasopharyngitis occurred twice. During the follow-up period 58 AE were reported, none of them serious.

Table 2.  Safety results
Most frequent AE on-therapyn (%)
  1. AE, adverse event; PTSD, post-traumatic stress disorder.

Subjects with any AE48 (92.3)
Nasopharyngitis29 (55.8)
Nausea27 (51.9)
Somnolence19 (36.5)
Vomiting10 (19.2)
Dizziness10 (19.2)
Headache9 (17.3)
PTSD (exacerbation of PTSD)7 (13.5)
Abdominal pain upper6 (11.5)
Stomatitis6 (11.5)
Malaise6 (11.5)
Hepatic function abnormal6 (11.5)



This open-label, uncontrolled study of paroxetine in Japanese subjects with PTSD suggests acute efficacy during a short-term (4 week) period that may be maintained during longer-term treatment (up to 52 weeks).

Paroxetine was generally well-tolerated, with only one patient withdrawing from the study due to lack of efficacy. These data are consistent with other studies showing the clinical usefulness of paroxetine for the treatment of PTSD.8–10

The mean baseline CAPS score of 74.6 was similar to scores seen in international studies8–10 but, because there is no approved medication for PTSD available in the Japanese market, with patients often treated with multiple drugs, it is possible that in this first Japanese study a higher proportion of treatment-resistant patients had been enrolled. Indeed, a sub-analysis of the present data suggests that patients classified as more severely ill may require higher doses of paroxetine (Table 1).

Looking at the sub-analysis of the data, Avoidance/Numbing is the largest cluster in terms of the number of questions, and accordingly the change from baseline for this cluster was higher than for the other two clusters, but percent reduction from baseline in the three clusters was similar (−42.8, −44.2 and −41.9% for Re-experiencing cluster score, Avoidance/Numbing cluster score, and Hyperarousal cluster score, respectively), suggesting that the efficacy was not one-cluster preferential (Table 3).

Table 3.  Change from baseline in CAPS-SX re-experiencing, avoidance and numbing, and hyperarousal scores (LOCF)
CAPS-SX ItemWeeknMeanSDMinimumMaximum95%CI (two-sided)
  • *

    Week 52 or withdrawal.

  • CI, confidence interval; CAPS-SX, Clinician-Administered Post-Traumatic Stress Disorder Scale One Week Symptom Status Version (CAPS-SX); LOCF, last observation carried forward.

Re-experiencing cluster scoreBaseline4920.17.7603717.922.3
Week 52*−8.69.16−2711−11.2−6.0
Avoidance/Numbing cluster scoreBaseline4932.69.95135029.735.4
Week 52*−14.411.73−3517−17.7−11.0
Hyperarousal cluster scoreBaseline49226.9393720.023.9
Week 52*−9.38.74−277−11.9−6.8


Paroxetine appeared to be generally well-tolerated in both acute and long-term use (up to 52 weeks). There were no specific AE that appeared with longer-term treatment, nor was there worsening of existing AE. The rates, spectrum and nature of the AE in the present study were consistent with the findings in the much larger and placebo-controlled international studies.8–10 In the previous study non-gender-related AE were reported in 83.0% of the patients in the paroxetine 40-mg group, 79.8% in the 20-mg group and 74.2% of patients on placebo (additionally there were gender-related AE of 24.6%, 34.5%, 1.6% in male and 12.7%, 8.7% and 10.5% in female patients, respectively) during the 12 weeks of the study.8

The withdrawal ratio in the present study was 3.8% at 12 weeks and 19.2% at 52 weeks (or early withdrawal), which is comparable to that of the three international studies.8–10

Furthermore the AE seen in the current study matched the safety profile of paroxetine seen in different Japanese populations, that is, patients suffering from depression, panic disorder or OCD.14–16


The present study was limited by the open treatment design, by a lack of placebo or active comparator group, and by the small number of patients studied. The design allowed enrollment of patients regardless of prior psychoactive medication (e.g. 33 patients had received psychoactive medication, among whom 15 patients received paroxetine as a prior medication), because the primary objective was to obtain the clinical experiences regarding not only the efficacy but also the safety of paroxetine with long-term use.

In conclusion, evidence of efficacy was obtained from the present study suggesting that the administration of a flexible dose of paroxetine between 20 and 40 mg/day was well tolerated and can contribute to clinically meaningful improvement during long-term therapy in Japanese PTSD patients.


The authors would like to acknowledge Gisa Gerstenberg and Masahiro Shimizu for contribution to the design of the study and conduct of the study and Gisa Gerstenberg for the preparation and revision of the manuscript. This work was funded and conducted by GlaxoSmithKline, Japan as a clinical trial in accordance to GCP. The following authors, Yoshiharu Kim, Nozomu Asukai, Takako Konishi, Hiroshi Kato, Hideto Hirotsune and Masaharu Maeda, have not financially or otherwise been reimbursed for their work on this paper. All have served as coordination committee members and received fees covering the expenses (e.g. investigator training) strictly according to GCP guidelines. All participating institutions received fees in line with GCP regulations. The authors had independence over the approval of the paper as it has been submitted. The authors Hirotaka Inoue, Hiroyasu Narita and Masaru Iwasaki, and the acknowledged contributors Gisa Gerstenberg and Masahiro Shimizu were employees of GlaxoSmithKline.