Journal of Psychiatric Research 131 (2020) 220–227

Contents lists available at ScienceDirect

Journal of Psychiatric Research

journal homepage: www.elsevier.com/locate/jpsychires

Could “triple-therapy” considered as a novel-optimal treatment model for

acute bipolar depression? A prospective real-world research in China
Yaling Zhou a, Xu Zhang b, Ruhan A. c, Yuexin Chen a, Xueli Sun a, *
a
Mental Health Center, West China Hospital of Sichuan University, China
b
Sichuan Provincial Center for Mental Health, Psychosomatic Medical Center of Sichuan People’s Hospital, China
c
Sleep Medicine Center of University of Electronic Science and Technology Hospital, China

A R T I C L E I N F O A B S T R A C T

Keywords: Background: Results of researches of bipolar depression treatment are inconsistent and to our knowledge, no
Treatment model study has previously revealed an optimal treatment model for bipolar depression in the real-world through a
Bipolar depression prospective way.
Multi-center
Objective: To find out an optimal treatment model for bipolar depression in the real-world by evaluating the effect
of different treatment models: monotherapy, double-therapy and triple-therapy.
Design: and Intervention: This 12 or 16-week, multi-center, real-world clinical study was conducted at 15 study
sites (inpatient or outpatient department) in West China and a total of 573 patients completed the follow-up.
During the study weeks, all researchers could choose a most proper treatment model freely basing on the
evaluation of patient’s symptoms and complete the follow-up according to the procedure.
Main outcomes and measures: The primary outcomes were baseline-to–endpoint change in Montgomery-Asberg
Depression Rating Scale (MADRS) total score and the constituent ratio of effects. Total score change in Young
Mania Rating Scale (YMRS) and Clinical Global Impression (CGI) from baseline to endpoint, treatment-emergent
mania rate and severe adverse events rate were used as secondary outcomes.
Results: During all study weeks, all the 3 groups showed a statistically significant improvement in MARDS, YMRS
and CGI (P＜0.001), but the triple-therapy group showed much more effective in significant response and
response rates at endpoint than double-therapy group and monotherapy group (P＜0.001) with lower treatment-
emergent mania rates (P = 0.001). At week 4, mean scores of MARDS in triple-therapy group are statistically
significant lower than monotherapy group (P = 0.013) and at the endpoint, mean scores of MARDS in triple-
therapy group are statistically significant lower than both double-therapy and monotherapy groups (P =
0.011). The severe adverse events rates are rare in all the 3 groups at week 4 and endpoint, and the rate of dry
mouth in triple-therapy group at week 4 is statistically significant lower than the other 2 groups (P = 0.002).
Conclusions: Triple-therapy is more effective in treating bipolar depression than double-therapy and monotherapy
model with a lower risk of developing manic symptoms.
Trial registration: Chinese Clinical Trial Registry. Identifier: ChiCTR1800019064.

1. Introduction significant degree of psycho-social impairment and disability (Bauer

et al., 2012). However, the treatment recommendations of bipolar
Bipolar disorder (BD) is a serious chronic disease characterized by depression (BD-D) are not very consistent in worldwide. Based on the
high recurrent and suicidal rate. It affects 1–3% of the population latest opinions of Canadian Network for Mood and Anxiety Treatments
worldwide and is rated as the sixth most common cause of disability in (CANMAT) and International Society for Bipolar Disorders (ISBD), the
the world by the World Health Organization (Merikangas et al., 2011; use of antidepressants (AD) has been downgraded to second-line treat­
Mathers et al., 2008). Compared to the other states of this illness, ment recommendations in treating bipolar depression (Yatham et al.,
depressive symptoms are the most prevalent dimension with a 2018). This is an important change in bipolar treatment guideline but

* Corresponding author. Mental health center of West China Hospital of Sichuan University, No.28, South Telecom Street, Wuhou District, Chengdu, Sichuan,
China.
E-mail address: [email protected] (X. Sun).

https://doi.org/10.1016/j.jpsychires.2020.09.017
Received 15 June 2020; Received in revised form 10 September 2020; Accepted 14 September 2020
Available online 19 September 2020
0022-3956/© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Y. Zhou et al. Journal of Psychiatric Research 131 (2020) 220–227

not consistent with the clinical experience of most clinicians, neither registration number is ChiCTR1800019064. All patients provided writ­
with the current Chinese guidelines (Yu and Fang, 2015). It is generally ten informed consent after receiving a complete description of the study.
known that the US Food and Drug Administration has recommended Data were summarized and analyzed from January 1, 2020 to April 30,
several treatment plans for BD-D, including the compound 2020.
olanzapine-fluoxetine, quetiapine, lurasidone and cariprazine. There is
no doubt that the choice is limited and most of those are antipsychotic 2.2. Patients
(AP) medications or contain an AP drug without a priori disease model
(Paul et al., 2015). Taking together, it seems to be difficult to find a All patients were aged from 18 to 65 years with bipolar disorder
unified way to stable the mood while treating depressive symptoms. meeting DSM-5 criteria for a current major depressive episode. Eligi­
Researches for the treatment of BD-D have lasted for decades, most of bility criteria is just total scores of 22 or higher on the Montgomery-
which are randomized, double-blind and placebo-controlled to study the Asberg Depression Rating Scale (MADRS) (Mundt et al., 2006) and
effect of monotherapies, but unfortunately, the results are inconsistent total scores lower than 12 on the Young Mania Rating Scale (YMRS)
and couldn’t be replicated (Lafer and Soares, 2005). For instance, (Colom et al., 2002). Exclusion criteria include a history of alcohol or
lithium, which is universally acknowledged to be the main drug for BD substance dependence within the previous 3 months; suicidal behavior
proved to have the superiority in treating BD-D when compared to within the previous 3 months; unstable or severe physical disease;
quetiapine in a clinical study (Kessing et al., 2011) while another trial pregnant or lactating women.
failed to find a significant effect when compared to quetiapine (Young
et al., 2010). More importantly, the use of antidepressants is still a 2.3. Procedures
controversial issue. Post et al. (Post et al., 2001) found that using AD to
treat the depression in both bipolar I and bipolar II could lead a high General demographic data of each patient should be collected at
treatment-emergent mania rate, especially in bipolar I (51.2%) and the baseline, meanwhile, the evaluation of MARDS, YMRS and Clinical
treatment-emergent mania rate were more likely to occur in the first 16 Global Impression (CGI) (Jones et al., 2019) was completed by re­
weeks of treatment. However, when comparing the results of adding a searchers with unified training. Then the researchers make a treatment
second mood stabilizer (MS, either lithium or valproate) to the initial plan for the patient based on the patient’s own symptoms. At week 4, the
mood stabilizer versus adding the paroxetine, Young et al. found no researchers need to make a second evaluation of the patients with
significant difference in treatment-emergent mania rate (Young et al., MARDS, YMRS and CGI and complete the evaluation of Treatment
2000). As previous researches and controversy of AD using in bipolar Emergent Symptom Scale (TSEE) (NIMH, 1985). If the outcome is
depression mostly focused on its efficacy and treatment-emergent mania “Significant response” or “Response”, continue the treatment plan for 2
rate, few studies have reported the relationship between the severity of months and complete the endpoint evaluation of MARDS, YMRS and CGI
bipolar depression and using of antidepressant. and TSEE at week 12. If the outcome at week 4 is “No response”, re­
Given the above, we found that previous studies were almost ran­ searchers should change another treatment plan for one more month to
domized, double-blind, parallel-treatment and placebo-controlled trial re-evaluate, if the re-evaluation outcome shows “Significant response”
with strict inclusion and exclusion, but the fact is that environment of or “Response”, then continue the new treatment plan for 2 months and
clinical treatment is various and a combination-therapy is common in complete the endpoint evaluation of MARDS, YMRS and CGI and TSEE
the treatment of bipolar disorder in the real-world. Therefore, the pre­ at week 16; if the re-evaluation outcome still shows “No response”, the
vious research results from strict conditions had low external validity, endpoint of these patients is week 8 (Fig. 1).
that may at least partially explain why the previous results are incon­
sistent. Moreover, the quality of treatment cannot be the effect of any 2.4. Assessments
drug alone, we believe a correct and flexible drug-combination is the key
to the treatment. As we all know there are 3 main sorts in BD-D treat­ Our primary outcome was baseline-to–endpoint change in MADRS
ment: mood stabilizers, antipsychotics and antidepressants, few studies total score and the constituent ratio of effects. Secondary efficacy
have explored the effect of 2 or more sorts of drugs in depression measures included the baseline-to–endpoint total score change in YMRS
treatment. Here we sought to find out the optimal treatment models of and CGI, treatment-emergent mania rate (YMRS score＜15 at baseline
acute BD-D treatment in a real-world prospective research. but ≥15 at endpoint was defined “treatment-emergent mania” (Tohen
et al., 2003) and severe adverse events (TESS score ≥3). Clinical visits
2. Methods were conducted at baseline, week 4 and endpoint. “Significant response”
was defined as 50% or greater improvement in the MADRS total score
2.1. Study design from baseline to endpoint with YMRS total score ＜15. “Response” was
defined as 25%–50% improvement in the MADRS total score from
This study was conducted at 15 study sites (inpatient or outpatient baseline to endpoint with YMRS total score ＜15. Both “Significant
department) in West China between December 31, 2018 and December, response” and “Response” are considered as effective treatment. “No
31,2019 (Supplement 1). This prospective, multi-center and real-world response” was defined as less than 25% improvement in the MADRS
clinical study evaluated the efficacy, safety, and treatment-emergent total score which means ineffective treatment (Xueli and Xu, 2017).
mania rate of different treatment models for the treatment of patients Adverse events were recorded at each visit and were coded using the
with Diagnostic and Statistical Manual of Mental Disorders (DSM-5) TESS.
(APA, 2013) defined bipolar disorder I/II depression. The treatment
models could be divided into 3 big types when analyzing: I. Mono­ 2.5. Statistical analysis
therapy Group (each sort was a subgroup): Monotherapy of (a) MS; (b)
Second Generation Antipsychotic (SGA) or (c)AD; II. Double-therapy All analyses were performed using statistical software Statistical
Group (each combination was a subgroup): Combination of either 2 Product and Service Solution (SPSS version 23.0). Significant effects
foregoing drugs:(d) MS + SGA; (e) MS + AD; (f) SGA + AD; III. were tested at a 2-sided α level of 0.05. Descriptive statistic was used to
Triple-therapy Group: Combination of mood stabilizers, antipsychotics analyze demographic data and the top 3 frequency of final treatment
and antipsychotics (MS + AD + SGA). All the researchers could choose a medicine. The distribution of demographic classification variables such
most proper treatment model freely basing on the evaluation of patient’s as gender, marriage, education background, etc. was tested by 2-sided
symptoms by their own consideration. Our study was approved by the χ2 test; the continuous variables such as age and first onset age were
Institutional Ethics Board at Chinese Clinical Trial Registry and the analyzed by one-way ANOVA; the constituent ratio of effects among the

221
Y. Zhou et al. Journal of Psychiatric Research 131 (2020) 220–227

Fig. 1. Flow diagram.

(sub)groups were compared by 2-sided χ2 test; the scale scores among 3. Results
groups at baseline, week 4 and endpoint were compared by one-way
ANOVA; the scale scores within groups at baseline, week 4 and 3.1. Demographics and clinical characteristics
endpoint were compared by repeated measure ANOVA; the treatment-
emergent mania rate and adverse events at endpoint were compared A total of 688 patients were enrolled in this study, in which 573
by 2-sided χ2 test (Fisher’s test was considered when n＜40). patients finally completed the follow-up study, the drop-out rate was
16.72%. The main reasons of drop-outs included the later included time,
change of address, poor medication compliance and distrust of

222
Y. Zhou et al. Journal of Psychiatric Research 131 (2020) 220–227

researchers. At the endpoint, there were 316 patients treating with Table 2
triple-therapy (mean [SD] age, 37.17 [14.14] years; 181 females Effectiveness rates among each subgroup of Monotherapy Group at endpoint.
[57.28%]), 180 patients treating with double-therapy (mean [SD] age, Treatment Patients, No. (%) χ2 P
38.03 [13.87] years; 117 females [65%]) while 77 patients treating with Outcome Value
MS Only (n SGA Only AD Only (n
monotherapy (mean [SD] age, 36.43 [14.15] years; 52 females = 24) (n = 30) = 23)
[67.53%]). 152 of 316 (48.10%)triple-therapy patients, 78 of 180
Effective 12 (50) 18 (60) 18 (78.26) 4.110 0.128
(43.33%)double-therapy patients and 36 of 77 (46.75%) monotherapy No Response 12 (50) 12 (40%) 5 (21.74)
patients had a diagnosis of bipolar disorder I, the rest were diagnosed of
bipolar disorder II. Baseline characteristics of each group are given in Abbreviations: MS, mood stabilizers; SGA, second generation antipsychotic; AD,
antidepressants.
Table 1. The number of patients with depression as the first onset
symptoms was statistically significant more in triple-therapy group (P＜
0.001). No more statistically significant sociodemographic or clinical Table 3
differences among the 3 groups. Effectiveness rates among each subgroup of Double-therapy Group at endpoint.
Treatment Patients, No. (%) χ2 P
3.2. Effectiveness Outcome Value
MS + SGA MS + AD (n SGA + AD
(n = 55) = 79) (n = 46)
The treatment effectiveness (Significant response + Response) rates Effective 41 (74.54) 69 (87.34) 36 (78.26) 3.793 0.15
of different treatment models in each subgroup at endpoint were No Response 14 (25.46) 10 (12.66) 10 (21.74)
depicted in Tables 2–3. There were no statistically significant differences
Abbreviations: MS, mood stabilizers; SGA, second generation antipsychotic; AD,
among MS only, AD only and SGA only subgroups in monotherapy, also, antidepressants.
no statistically significant differences were found among MS + SGA, MS
+ AD and SGA + AD subgroups in double-therapy. Taken together, we
2 groups (χ2 = 67.790, P＜0.001) (Table 4). The constituent ratio of
compared the constituent ratio among Monotherapy group, Double-
effect of each group were showed in Figs. 2–4.
therapy group and Triple-therapy group, the results showed a statisti­
At baseline, the mean score of CGI in triple-therapy group was sta­
cally significant effectiveness difference among 3 groups: the effective­
tistically significant higher than monotherapy group (F = 3.204, P =
ness rate in triple-therapy group were significantly higher than the other
0.041) while no statistically significant differences were found in
MARDS and YMRS among the 3 groups. At week 4, statistically signifi­
Table 1 cant differences were found in MARDS and YMRS: the mean score of
Clinical and demographic characteristics.
MARDS in triple-therapy and double therapy were statistically signifi­
Patient Characteristic Mean ± SD/Patients, No. cant lower than monotherapy group (F = 4.373, P = 0.013), meanwhile
Triple- Double- Monotherapy P the mean score of YMRS in triple-therapy group was statistically sig­
therapy therapy group(n = 77) Value nificant lower than the other 2 groups (F = 64.957, P＜0.001). As in
group(n group(n week 4, the same statistically significant differences in MARDS were
= 316) = 180)
found in endpoint (F = 4.581, P = 0.011) while no statistically signifi­
Age,y 37.17 ± 38.03 ± 36.43 ± 14.15 0.81 cant differences in YMRS and CGI setting (Table 5).
14.14 13.87
Analysis of score change in MARDS, YMRS and CGI within groups by
First onset Age,y 28.47 ± 30.85 ± 28.69 ± 13.12 0.293
11.17 12.41 time were given in Table 6 and Fig. 5. Except for YMRS, there were
No. of attackness 3.11 ± 2.59 ± 1.78 ± 1.06 0.266 statistically significant differences in score change in MADRS and CGI by
0.46 3.31 time (P＜0.001). In the YMRS setting, there was a statistically significant
Course of disease,mo 45.21 ± 39.86 ± 19.79 ± 24.71 0.135 difference in score decreasing in triple-therapy (P＜0.001)at week 4 but
72.58 77.63
no statistically significant differences in the other 2 groups. Interest­
Gender Male 135 63 25 0.109
Female 181 117 52 ingly, there were statistically significant score decreasing in YMRS from
Address City 196 119 50 0.642 week 4 to endpoint in double-therapy and monotherapy group (P＜
Countryside 120 61 27 0.001)but no statistically significant change in triple-therapy group.
Educational High school 195 123 52 0.281
level or lower
College/ 121 57 25 3.3. Treatment-emergent mania
university or
higher
Marriage Unmarried 147 69 36 0.183 Treatment-emergent mania was defined as a YMRS score＜15 at
Married 169 111 41 baseline but ≥ 15 at endpoint. In all 3 groups, the incidence of
Personal No 140 84 29 0.411 treatment-emergent mania was no more than 20%. Rates of treatment-
income Yes 176 96 48 emergent mania were 4.11% (13/316) for the triple-therapy
Family No 292 167 73 0.764
group,10.56%(19/180) for the double-therapy group, and 15.58%(12/
history of Yes 24 13 4
psychosis 77) for the monotherapy group (Table 7). Switch rate in triple-therapy
First onset Depression 187 121a 44a ＜
symptoms 0.001
Table 4
（Hypo） 46 38 15
Effectiveness rates among Monotherapy Group, Double-therapy Group and
Mania
Mixed state 26 13 12 Triple-therapy Group at endpoint.
Other 57 8 6 Treatment Patients, No. (%) χ2 P
symptoms Outcome Value
Type of Bipolar I 152 78 36 0.591 Triple- Double- Monotherapy
bipolar disorder therapy (n therapy (n (n = 77)
Bipolar II 164 102 41 = 316) = 180)
disorder Effective 303 146 48 (62.34) 67.790 ＜
(95.89) (81.11) 0.001
Notes: a represents a significant difference compared with Triple-therapy Group
No Response 13 (4.11) 34 (18.89) 29 (37.66)
while in Pairwise comparison.

223
Y. Zhou et al. Journal of Psychiatric Research 131 (2020) 220–227

Table 5
Comparison of scores at baseline, week 4 and endpoint among 3 groups.
Time and Mean ± SD F P
Scale value
triple- double- monotherapy
therapy therapy group (n = 77)
group（n group (n
= 316） = 180)

Baseline
MARDS 25.84 ± 12.16 25.89 ± 23.97 ± 10.27 0.418 0.659
10.45
YMRS 7.98 ± 5.44 8.24 ± 9.12 ± 4.30 1.25 0.29
6.55
a
CGI 4.65 ± 1.67 4.58 ± 4.17 ± 1.65 3.204 0.041
1.03
week 4
MARDS 15.08 ± 14.14b 15.37 ± 19.53 ± 9.99 4.373 0.013
8.21b
Fig. 2. Constituent ratio of triple-therapy at endpoint. YMRS 1.50 ± 4.22 6.61 ± 9.62 ± 7.62 64.957 ＜
9.03 0.001
CGI 3.18 ± 1.14 3.11 ± 3.04 ± 1.28 0.403 0.668
1.10
Endpoint
MARDS 6.41 ± 7.61 9.08 ± 10.61 ± 9.99a 4.581 0.011
8.45a
YMRS 1.43 ± 6.12 0.84 ± 1.84 ± 0.63 0.497 0.609
2.67
CGI 2.34 ± 1.19 2.24 ± 2.01 ± 0.89 2.675 0.07
1.13

Notes: a represents a significant difference compared with triple-therapy group

while in Pairwise comparison; b represents a significant difference compared
with monotherapy group.

week 4, rates of dry mouth were no more than 15% (6.65% in triple-
therapy group, 13.89% in double-therapy group and 14.29% in mono­
therapy group), and the rate in triple-therapy group was statistically
significant lower than the other 2 groups (χ2 = 12.059, P = 0.002).
There were no statistically significant differences in constipation and
Fig. 3. Constituent ratio of double-therapy at endpoint. weight gain among groups at both week 4 and endpoint.

3.5. Top 3 frequency of treatment models

The top 3 frequency of treatment models using in our study were

given in Table 9. The dosage of all the medicine using in our study were
within the dosage range specified in the drug specification. From the
foregoing results, we found that the most frequently used treatment
model in our study was the triple-therapy model, namely mood stabi­
lizer combines antipsychotic and antidepressant agent. In the triple-
therapy group, the most frequently drug-combination were Lithium +
Venlafaxine + Quetiapine (18.99%), Valproate + Venlafaxine + Olan­
zapine (8.23%) and Valproate + Citalopram + Quetiapine (7.28%).
Also, the results implied that in all 3 groups, mood stabilizers Lithium
and Valproate, antidepressants Venlafaxine, Duloxetine and Sertraline,
antipsychotics Quetiapine and Aripiprazole were using the most
frequently in drug-combination or alone.

Fig. 4. Constituent ratio of monotherapy at endpoint. 4. Discussion

group was statistically significant lower than the other 2 groups (χ2 = To our knowledge, this is the first prospective research in the real-
14.554, P = 0.001). world to find out an optimal treatment model for acute bipolar
depression. Results indicate that all 3 treatment models could reduce
both YMARS and YMRS total scores, but the triple-therapy model do
3.4. Safety have faster and better effect in scale reduction rates. Moreover, the
constituent ratio of effect at endpoint in each group indicate that triple-
Severe treatment-emergent adverse event was defined as TESS score therapy has a much higher significant response and response rates than
≥3. Table 8 listed severe treatment-emergent adverse events reported by the other 2 groups with a much lower treatment-emergent mania rate.
5% or more of patients in any treatment group. In all 3 groups, the severe Taken together, the results reveal that the triple-therapy model has
treatment-emergent adverse events reported ≥5% were rare and similar. obvious advantages in both scale score reduction and emotional stability
No matter at any estimate-time, the most reported treatment-emergent than other treatment models.
adverse events included dry mouth, constipation and weight gain. At It is known to all that bipolar disorder is a disease with apparent

224
Y. Zhou et al. Journal of Psychiatric Research 131 (2020) 220–227

Table 6
Comparison of scores at baseline, week 4 and endpoint within groups.
Scale and Group Patients, No. Mean ± SD F P value

Baseline Week 4 Endpoint

MARDS
triple-therapy group 316 25.84 ± 12.16 15.08 ± 14.14 6.41 ± 7.61 221.382 ＜0.001
double-therapy group 180 25.89 ± 10.45 15.37 ± 8.21 9.08 ± 8.45 157.063 ＜0.001
monotherapy group 77 23.97 ± 10.27 19.53 ± 9.99 10.61 ± 9.99 35.054 ＜0.001
YMRS
triple-therapy group 316 7.98 ± 5.44 1.50 ± 4.22a 1.43 ± 6.12a 158.077 ＜0.001
double-therapy group 180 8.24 ± 6.55 6.61 ± 9.03 0.84 ± 2.67ab 62.049 ＜0.001
monotherapy group 77 9.12 ± 4.30 9.62 ± 7.62 1.84 ± 0.63ab 56.924 ＜0.001
CGI
triple-therapy group 316 4.65 ± 1.67 3.18 ± 1.14 2.34 ± 1.19 235.441 ＜0.001
double-therapy group 180 4.58 ± 1.03 3.11 ± 1.10 2.24 ± 1.13 212.922 ＜0.001
monotherapy group 77 4.17 ± 1.65 3.04 ± 1.28 2.01 ± 0.89 52.325 ＜0.001

Notes: a represents a significant difference compared with Baseline while in Pairwise comparison; b represents a significant difference compared with week 4.

Fig. 5. Mean score change in the scale MARDS, YMRS and CGI from baseline to endpoint.

Table 7
Comparison of treatment-emergent mania rate among 3 groups at endpoint.
State Patients, No.（%） χ2 P value

triple-therapy group（n = 316） double-therapy group (n = 180) monotherapy group (n = 77)

switch to mania 13（4.11%） 19（10.56%）a 12（15.58%）a 14.554 0.001

No switch 303（95.89%） 161（89.44%） 65（84.42%）

Notes: a represents a significant difference compared with Triple-therapy Group while in Pairwise comparison.

225
Y. Zhou et al. Journal of Psychiatric Research 131 (2020) 220–227

Table 8 our study, triple-therapy model performs an advantage in emotional

Comparison of the adverse event among 3 groups at week 4 and endpoint. (Only stability and treatment effect may probably due to regulating more
depicted serious adverse events reported by 5% or more). extensive circadian rhythm.
Time and Patients, No.（%） χ2/ P On the other hand, the constitution of triple-therapy model shows a
Adverse Event
triple- double- monotherapy
Fisher value balance of different sorts of medicine in treating bipolar disorder. The
therapy therapy group characteristic of triple-therapy model is that mood stabilizers are used as
group group core medicines with antipsychotics and antidepressants as supplements.
Week 4 The model not only treat the symptoms of bipolar disorders compre­
Dry mouth 21 25 11 (14.29%)a 12.059 0.002 hensively but also adjust dosage flexibly according to the symptoms
(6.65%) (13.89%)a changing so that the mood would not fluctuate too much in treating.
Constipation 34 11 11 (14.29%) 4.864 0.088 As our study is conducted in a real-world, all researchers could
(10.76%) (6.11%)
Weight gain 13 13 NA 2.231 0.135
choose a most proper treatment model freely basing on the evaluation of
(4.11%) (7.22%) patient’s symptoms by their own consideration. The clinical and de­
Endpoint mographic characteristic results show an obvious choosing tendency in
Dry mouth NA 7 (3.89%) 5 (6.49%) 0.822 0.352 treatment models. In our study, patients with depression as first onset
Constipation NA 4 (2.22%) 6（7.79%） 2.481 0.18
symptoms are more probably given triple-therapy treatment model. This
Notes: a represents a significant difference compared with Triple-therapy Group result indicate that researchers may consider patients with depression as
while in Pairwise comparison. first onset symptoms would have more recurrent and suicidal rate (Forty
et al., 2008; Rosa et al., 2008), so they tended to use triple-therapy
model rather than monotherapy or double-therapy for controlling
Table 9
symptoms and potential risk more positive.
Top 3 frequency of treatment models in each group.
Regarding the possibility of treatment-emergent mania, the present
treatment models (Patients, No.) Patients, No. (%) findings indicate no additional risk when using 3 sorts of medicine than
Triple-therapy(n=316) 2 or 1 sort while the use of antidepressants in treating BD-D is still a
Lithium + Venlafaxine + Quetiapine 60 (18.99%) controversy. To our surprise, the triple-therapy treatment model has
Valproate + Venlafaxine + Olanzapine 26（8.23%） lower switch rate during the treatment, as mentioned before, triple-
Valproate + Citalopram + Quetiapine 23（7.28%）
Double-therapy(n=180)
therapy treatment model could stable personal emotion more effec­
MS + AD(n = 79) tively, and this may be the reason why it has a lower treatment-emergent
Lithium + Venlafaxine 16 (20.25%) mania rate than other treatment models and previous random, double-
Lithium + Duloxetine 10（12.66%） mind study (Tohen et al., 2003). Researchers found that
Lamotrigine + Venlafaxine 10（12.66%）
treatment-emergent mania is more likely to occur in the first 16 weeks of
Valproate + Sertraline 6（7.59%）
MS + SGA(n = 55) treatment (Post et al., 2001). So the results of switch rate in our study
Valproate + Quetiapine 25（45.45%） have a certain value in antidepressants using.
Valproate + Aripiprazole 14 (25.45%) In terms of safety, only 3 sever adverse events were reported over 5%
Valproate + Clozapine 9 (16.36%) at week 4 and 2 at endpoint: dry mouth, constipation and weight gain. In
SGA + AD(n = 46)
Quetiapine + Venlafaxine 8（17.39%）
general, these side effects are very common in BD-D treatment and not
Quetiapine + Duloxetine 6（13.04%） difficult to handle. Previous studies reported some other side effects
Quetiapine + Sertraline 6（13.04%） such as nausea and diarrhea (Bhana et al., 2001; Sanger et al., 2001), but
Monotherapy(n=77) not in our study, that may partially due to these light side effects would
SGA(n = 30)
not affect one’s daily life.
Aripiprazole 20（66.67%）
Quetiapine 5（16.67%) In summary, our study was the first prospective real-world treatment
MS(n = 24) research for BD-D in China and break through the limitations of classical
Lithium 22 (91.67%） random, double-mind trial, the present results come from the real-world
AD(n = 23) and are closer to the actual situation in treatment for BD-D.
Citalopram 13 (56.52%)
Duloxetine 4 (17.39%)
Sertraline 41 (7.39%) 4.1. Limitations
Abbreviations: MS, mood stabilizers; SGA, second generation antipsychotic; AD,
There are some limitations of our study. First, although 12 or 16
antidepressants. All the kinds of drugs were used within the dosage range
specified in the drug specification. weeks are enough to evaluate the effect in acute BD-D treatment, the
longer the follow-up time is, the more convincing the results would be.
Second, it is well known that BD-D are often associated with abnormality
dysrhythmia emotion. The symptoms are related to abnormal expression
of some endocrine hormone and immune indexes (Belvederi et al., 2016;
of circadian genes. No matter mood stabilizer, antipsychotic or antide­
Joshua and Roger, 2017), besides scales, improvement of these objective
pressants, previous research found they are all related to modulate
laboratory indicators when treating may be another assessment indica­
circadian genes. Many studies have demonstrated that Valproate could
tor in the future study. Last, considering the safety and stability of this
adjust the circadian rhythm by increasing the amplitude of PER2 gene
study, patients with suicidal behavior within the previous 3 months
and changing the phase of rhythm while Carbamazepine could influence
were excluded, which might lead to a limitation of the sample which was
the expression of BMAL1, DBP and CRY1; By promoting the expression
not very large for a real-world study, but the results indeed provide a
of CRY1 and PER2 gene and enhance the amplitude of PER2, Lithium
novel treatment framework and direction in BD-D treatment and have
could modulate our circadian rhythm (Bradley et al., 2017; Patnaik
conceptual and clinical implications.
et al., 2018). In addition, Quetiapine was found to affect rhythm by
improving the expression of PER1 and PER2 rhythm gene’s mRNA and
5. Conclusions
Fluoxetine could also affect the expression of rhythm gene in hippo­
campus (Chenaf et al., 2017; Oyffe et al., 2015). The foregoing findings
Triple-therapy is more effective in treating bipolar depression than
imply that the core response mechanisms of MS, SGA and AD in treating
double-therapy and monotherapy with a lower risk of developing manic
bipolar disorders are properly stabling the personal circadian rhythm. In
symptoms.

226
Y. Zhou et al. Journal of Psychiatric Research 131 (2020) 220–227

Declaration of competing interest bipolar I disorder and clinical course of illness. Bipolar Disord. 11, 82. https://doi.
org/10.1111/j.1399-5618.2008.00654.x.
Jones, R.M., Patel, K., Moscovici, M., Mcmaster, R., Simpson, A.I.F., 2019. Adaptation of
None declared. the clinical global impression for use in correctional settings: the CGI-C. Front.
Psychiatr. 10, 687. https://doi.org/10.3389/fpsyt.2019.00687.
Acknowledgements Joshua, R., Roger, M.I., 2017. Bipolar disorder and immune dysfunction: epidemiological
findings, proposed pathophysiology and clinical implications. Brain Sci. 7, 144–150.
https://doi.org/10.3390/brainsci7110144.
The authors would like to thank Professor Bo Zhang for study design; Kessing, L.V., Hellmund, G., Geddes, J.R., Goodwin, G.M., Andersen, P.K., 2011.
thank Dr. Zaiquan Dong; Yang Liu; Wenjiao Min; Lei Tang; Chun Zhou; Valproate v. lithium in the treatment of bipolar disorder in clinical practice:
observational nationwide register-based cohort study. Br. J. Psychiatry 199, 57–63.
Zhijie Ding; Techeng Jiang; Li Kuang; Ping Liu; Kezhi Liu; Shanyong https://doi.org/10.1192/bjp.bp.110.084822.
Than; Jianjun Bao; Desui Wang; Fengchen Gao; Yinghui Zhang; Youguo Lafer, B., Soares, M.B.D.M., 2005. The treatment of bipolar depression. Rev. Psiquiatr.
Tan; Ming Luo and Qiang Chen for data collection and thank Dr. Xu Clínica 32, 49–55. https://doi.org/10.1590/S0101-60832005000700008.
Mathers, C., Fat, D.M., Boerma, T., 2008. World Health Organization.the Global Burden
Zhang, Ruhan A for their help with the data analysis. of Disease:2004 Update. Harvard School of Public Health on behalf of the World
Health Organization and the World Bank, Boston.
Appendix A. Supplementary data Merikangas, K.R., Jin, R., He, J.P., Kessler, R.C., Lee, S., Sampson, N.A., Viana, M.C.,
Andrade, L.H., Hu, C.Y., Karam, E.G., Ladea, M., Mora, M.E.M., Ono, Y., Villa, J.P.,
Sagar, R., Wells, J.E., Zarkov, Z., 2011. Prevalence and correlates of bipolar
Supplementary data to this article can be found online at https://doi. spectrum disorder in the world mental health survey initiative. Arch. Gen. Psychiatr.
org/10.1016/j.jpsychires.2020.09.017. 68, 241–251. https://doi.org/10.1001/archgenpsychiatry.2011.12.
Mundt, J.C., Katzelnick, D.J., Kennedy, S.H., Eisfeld, B.S., Bouffard, B.B., Greist, J.H.,
2006. Validation of an IVRS version of the MADRS. J. Psychiatr. Res. 40, 243–246.
Funding/support https://doi.org/10.1016/j.jpsychires.2005.05.002.
NIMH, 1985. TESS(Treatment emergent symptom scale-write-in). Psychopharmacol.
This study was supported by West China Psychiatric Association, Bull. 21, 1069–1072 (There was no available doi could be found).
Oyffe, I., Shwizer, R., Stolovy, T., 2015. The association between diagnosis, treatment
China. Grant ID:WCPASX-2016. delay and outcome among patients with bipolar disorders. Psychiatr. Q. 86, 1–11.
https://doi.org/10.1007/s11126-014-9316-4.
Author statement Patnaik, K.P., Menon, V., Chandrasekaran, V., Navin, K., 2018. Biological rhythm
impairment in bipolar disorder: a state or trait marker? Indian J. Psychiatr. 60,
404–408. https://doi.org/10.4103/psychiatry.IndianJPsychiatry_110_18.
Yaling Zhou, Ph.D: Data analysis and article writing. Paul, A Insel, Amara, Susan G., Blaschke, Terrence F., 2015. Introduction to the theme
Xu Zhang, Ph.D and Ruhan A, MM: Help for advise for data analysis. "precision medicine and prediction in pharmacology. Annu. Rev. Pharmacol.
Toxicol. 55, 11–14. https://doi.org/10.1146/annurev-pharmtox-101714-123102.
Yuexin Chen, MM: Help for data summarize and arrangement. Post, R.M., Altshuler, L.L., Frye, M.A., Suppes, T., Rush, A.J., 2001. Rate of switch in
Xueli Sun, Corresponding author: trial design and data quality bipolar patients prospectively treated with second-generation antidepressants as
control. augmentation to mood stabilizers. Bipolar Disord. 3, 259–265. https://doi.org/
10.1034/j.1399-5618.2001.030505.x.
Rosa, A.R., Andreazza, A.C., Kunz, M., Gomes, F., Santin, A., Moreno, J.S., Reinares, M.,
References Colom, F., Vieta, E., Kapczinski, F., 2008. Predominant polarity in bipolar disorder:
Diagnostic implications. J. Affect. Disord. 107, 45–51. https://doi.org/10.1016/j.
APA, 2013. Diagnostic and Statistical Manual of Mental Disorders: DSM-5™, fifth ed. jad.2007.07.021.
American Psychiatric Association, Washington, DC. Sanger, T.M., Grundy, S.L., Gibson, P.J., Namjoshi, M.A., Tohen, M.F., 2001. Long-term
Bauer, M., Ritter, P., Grunze, H., Pfennig, A., 2012. Treatment options for acute olanzapine therapy in the treatment of bipolar I disorder. J. Clin. Psychiatr. 62,
depression in bipolardisorder. Bipolar Disord. 14, 37–50. https://doi.org/10.1111/ 273–281. https://doi.org/10.4088/JCP.v62n0410.
j.1399-5618.2012.00991.x. Tohen, M., Vieta, E., Calabrese, J., Ketter, T.A., Breier, A., 2003. Efficacy of olanzapine
Belvederi, M.M., Prestia, D., Mondelli, V., Pariante, C., Patti, S., Olivieri, B., Arzani, C., and olanzapine-fluoxetine combination in the treatment of bipolar I depression.
Masotti, M., Respino, M., Antonioli, M., Vassallo, L., Serafini, G., Perna, G., Arch. Gen. Psychiatr. 60, 1079–1088. https://doi.org/10.1016/S0924-977X(13)
Pompili, M., Amore, M., 2016. The HPA axis in bipolardisorder: systematic review 70587-X.
and meta-analysis. Psychoneuroendocrinology 63, 327–342. https://doi.org/ Xueli, Sun, Xu, Zhang, 2017. Consensus of Identification and Optimize Therapeutic
10.1016/j.psyneuen.2015.10.014. Scheme of Bipolar Affective Disorder and its Atypical Symptoms, second ed. China
Bhana, N., Foster, R.H., Olney, R., Plosker, G.L., 2001. Olanzapine: an updated review of Science Press, Peking, pp. 195–196.
its use in the management of schizophrenia. Drugs 61, 111–161. https://doi.org/ Yatham, L.N., Kennedy, S.H., Parikh, S.V., Schaffer, A., Berk, M., 2018. Canadian
10.2165/00003495-200161010-00011. Network for mood and anxiety treatments(CANMAT)and international society for
Bradley, A.J., Webb-Mitchell, R., Hazu, A., Slater, N., Middleton, B., Gallagher, P., bipolar disorders(ISBD)2018 guidelines for the management of patients with bipolar
McAllister-Williams, H., Anderson, K.N., 2017. Sleep and circadian rhythm disorder. Bipolar Disord. 20, 97–170. https://doi.org/10.1111/bdi.12609.
disturbance in bipolar disorder. Psychol. Med. 47, 1678–1689. https://doi.org/ Young, L.T., Joffe, R.T., Robb, J.C., MacQueen, G.M., Siotis, I.P., 2000. Double-blind
10.1017/S0033291717000186. comparison of addition of a second mood stabilizer versus an antidepressant to an
Chenaf, C., Chapuy, E., Libert, F., Marchand, F., Courteix, C., Bertrand, M., Gabriel, C., initial mood stabilizer for treatment of patients with bipolar depression. Am. J.
Mocaër, E., Eschalier, A., Authier, N., 2017. Agomelatine: a new opportunity to Psychiatr. 157, 124–126. https://doi.org/10.1176/foc.1.1.34.
reduce neuropathic pain-preclinical evidence. Pain 158, 149–160. https://doi.org/ Young, A.H., McElroy, S.L., Bauer, M., Philips, N., William, C., Bengt, O., Paulsson, B.,
10.1097/j.pain.0000000000000738. Martin, B., 2010. A double-blind, placebo-controlled study of quetiapine and lithium
Colom, F., Vieta, E., Martínezarán, A., Garcia-Garcia, M., Salamero, M., 2002. Spanish monotherapy in adults in the acute phase of bipolar depression (EMBOLDEN I).
version of a scale for the assessment of mania:validity and reliability of the Young J. Clin. Psychiatr. 71, 150. https://doi.org/10.4088/JCP.08m04995gre.
Mania Rating Scale. Med. Clínica 119, 366–371. https://doi.org/10.1016/S0025- Yu, X., Fang, Y.R., 2015. Guidelines for the Prevention and Treatment of Bipolar Disorder
7753(02)73419-2. in China, second ed. Chinese Medical Multimedia Press, Peking.
Forty, Liz, Jones, Lisa, Jones, Ian, Smith, Daniel, Caesar, Sian, Fraser, Christine, Gordon
Smith, Katherine, Hyde, Sally, Craddock, Nick, 2008. Polarity at illness onset in

227