Participants The present study participants were patients who visited China Medical University Hospital. In a previous 8-week parallel trial with two intervention groups, namely acupuncture plus duloxetine versus duloxetin; with 64 subjects in each group, the HAM-D at baseline and sixth week (similar to our course before crossover), the effect size was 0.548 (Ma et al., 2015). Based on this study, we decided that the difference between pain-specific and depression-specific acupoints would make the effect size larger than 0.548, thus we set the effect size as 0.6. Taking a drop-out rate of 20% into consideration, we expected to enroll 75 eligible patients with comorbid chronic pain and MDD. Both outpatients and inpatients who met our criteria for chronic pain and MDD were considered candidates. We used the Brief Pain Inventory (BPI) Long Form (score for the fifth item ≥ 4), Beck Depression Inventory-Second Edition (BDI-II; total score ≥ 20) to screen for potential candidates. Subsequently, the eligible candidates were assessed by trained psychiatrists for psychiatric disorders through the Structured Mini-International Neuropsychiatric Interview (Sheehan et al., 1998). The patients who met the inclusion criteria were asked to provide written informed consent. Demographic information such as gender, age, education, marital status, psychiatric history, substance use history, and family history was also collected.
Inclusion criteria Patients were included if they (a) were aged ≥20 years, (b) were diagnosed as having MDD following the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, (c) received a score ≥18 on the 21-item HAM-D, (d) reported experiencing persistent pain for more than three months and obtained a score ≥4 on the fifth item of the BPI, (e) were assessed as being moderately ill score ≥4 based on the Clinical Global Impression (CGI) scale, (f) had a consistent (in terms of dosage and type of medication or treatment) analgesic and antidepressive medication regimen and psychotherapy regimen in the four weeks prior to enrollment, and (g) had not received acupuncture and herbal treatments in the two weeks prior to enrollment. The Brief Pain Inventory (BPI) user guide indicates that BPI is one of the most widely used measurement tools for assessing clinical pain. The BPI could assess pain severity and measure how much pain has interfered with seven daily activities. The third, fourth, and fifth items of the BPI long form assess pain at its "worst," "least," "average," within recall period of one week. Because the pain of the patients with comorbid chronic pain and depression always varied on and off, the fifth item of the BPI long form could help us to assess the subjects' average pain severity rather than the worst or least pain severity (Cleeland and Ryan, 1991; Keller et al., 2004).
Exclusion criteria Patients were excluded if they (a) had MDD caused by substance abuse or pregnancy, (b) had their first onset of MDD after the age of 65 years, (c) had experienced malignant pain caused by cancer pain syndrome, (d) had experienced visceral pain (e.g., stomachache), or (e) had experienced referred pain (e.g., back pain due to pancreatitis). Changes in the brain, such as stroke, often cause the first onset of MDD after the age of 65 years (Cojocaru et al., 2013).
Study Protocol Randomization, group allocation (crossover), washout, and blinding The computer-generated permuted block randomization was used so that the participating patients were randomly assigned on an exactly 1:1 ratio to either the depression-pain group, in which the depression-specific acupoints were first targeted (12 sessions over six weeks), followed by the pain-specific acupoints (12 sessions over six weeks), or the pain-depression group, in which the pain-specific acupoints (six weeks 12 sessions) were first targeted followed by the depression-specific acupoints (six weeks 12 sessions). Following other studies (Kong et al., 2013; Liang et al., 2013), after the first six-week intervention, a two week washout period was implemented (Liu et al., 2013) to eliminate the residual acupuncture effects from the first 6-week phase. The patients' original conventional treatments were maintained during the washout period, and no acupuncture was performed. Subsequently, they underwent a second 6-week intervention in which a different set of acupoints were targeted. According to the computer-generated permuted block randomization, 75 sealed envelopes containing notes with 1\~75 numbers assigned to the depression-pain group or pain-depression group. Before each new treatment course began, the acupuncturists opened one of the 75 envelopes in numerical order and treated the newly included participants with pain-specific or depression-specific acupoints, according to the notes. The intervention and research teams were physically segregated during the trial. The patients, the assessor, and the biostatistician were blinded to the allocation process until the data were analyzed. After the interventions were completed, the patients were asked to indicate the interventions they perceived to have received during the first and second six-week periods. Research nurses recorded the results.
Interventions The following depression-specific acupoints: Baihui (Du-20), Yintang (EX-HN3), Sishencong (EX-HN1), bilateral Cuanzhu (BL2), bilateral Toulinqi (GB15), bilateral Neiguan (PC6), bilateral Shenmen (HT7), bilateral Sanyinjiao (SP6), and bilateral Taichong (LR3) were selected according to Cochrane reviews and other meta-analysis studies (Smith et al., 2010; Wang et al., 2008; Zhang et al., 2010). These depression-specific acupoints could raise up the qi, calm the mind, smooth liver qi stagnation, promote blood circulation, remove blood stasis, and nourish yin. Similarly, to cover different kinds of chronic pain in the present study, the following pain-specific acupoints: bilateral Fengchi (GB20), bilateral Jianjing (GB21), bilateral Quchi (LI11), bilateral Shousanli (LI10), bilateral Hegu (LI4), bilateral Wangu (SI4), bilateral Shenshu (BL23), bilateral Huantiao (GB30), bilateral Weizhong (BL40), bilateral Feiyang (BL58), bilateral Yanglingquan (GB34), bilateral Yinlingquan (SP9), and bilateral Taixi (KI3) were selected according to different clinical trials which treated pain in specific parts of the human body (Leibing et al., 2002; Mavrommatis et al., 2012; Molsberger et al., 2002; Molsberger et al., 2010; Vas et al., 2006). These pain-specific acupoints could eliminate wind, relax muscles, promote blood circulation, and supplement the kidney. These two sets of entirely different acupoints were used to treat MDD or chronic pain. During each session, each patient received a standardized 20-min acupuncture treatment, during which disposable acupuncture needles with a 0.25-mm diameter (34 gauge) and 40-mm length (CASOON, Taichung, Taiwan) were inserted into all acupoints except Huantiao (GB30) to a depth of 10-30 mm in a direction oblique or vertical to the skin surface and manipulated to induce the De-Qi sensation. The acupuncture needles with a 0.30-mm diameter (30 gauge) and 75-mm length (CASOON, Taichung, Taiwan) were inserted into the Huantiao (GB30) acupoint to a depth of 30-50 mm to achieve the De-Qi sensation, too. The acupuncturists treated the patients' MDD or chronic pain separately while lying on their front or back. At this point, acupuncturists did not tell the patients the expected efficacy. Two licensed attending physician acupuncturists with more than ten years of clinical experience in China Medical University Hospital performed all interventions. Participating patients were briefly introduced to the acupuncture procedure during their first visit. The patients were then randomized to receive 24 depression-pain or pain-depression acupuncture sessions.
Outcome measures During the visits at weeks 0 (baseline), 2, 4, 6 (after the first 6-week intervention), 8 (before the start of the second 6-week intervention), 10, 12, and 14 (after the second 6-week intervention), trained research nurses assessed the patients for depressive symptoms, pain symptoms, and the quality of life by using the HAM-D (Hamilton, 1960), BDI-II (Beck et al., 1996), BPI, Neurotoxicity Rating Scale (NRS), Clinical Global Impression scale (CGI), and World Health Organization Quality of Life BREF (WHOQOL-BREF). The HAM-D and BDI-II are the most frequently used observer-rated and self-report scales of depression, respectively. The BPI rapidly assesses the severity of pain and its impact on functioning. The NRS is a reliable and valid self-report measure used in the evaluation of psychiatric and physical symptoms. The CGI are measures of symptom severity, treatment response and the efficacy of treatments. The WHOQOL-BREF contains four domains related to the quality of life: physical health, psychological, social relationships and environment.
Blood sample and quantification of cytokines At weeks 0 (baseline), 6 (after the first 6-week intervention), 8 (before the start of the second 6-week intervention), and 14 (after the second 6-week intervention), peripheral venous blood samples (20 mL per time) were collected from the patients between 8:30 a.m. and 9:30 p.m. after they had fasted overnight. The obtained plasma samples were centrifuged at 3,000 rpm for 10 min and then stored at -80°C for further analysis. Subsequently, the samples were analyzed in duplicate using a Bio-Plex Human 15-plex assay kit (Bio-Rad Laboratories, Hercules, CA, USA) following the manufacturer's instructions. The complete list of cytokines \[interleukin (IL)-1β, IL-6, and tumor necrosis factor (TNF)-α\] was quantified for these cohorts, and their detection limits and reproducibility were provided in the product manual. The Bio-Plex Protein Array System (Bio-Rad) was used to distinguish three distinct sets of fluorescently dyed beads. The three detected cytokine profiles had high sensitivity and a broad dynamic range (0-32,000 pg/mL).
Treatment response and complete remission in depression According to previous studies, the definition of treatment response (Leucht et al., 2013) was a ≥ 50% decrease in HAM-D from baseline to week six or from week eight to week 14. The definition of complete remission (Riedel et al., 2010) was a score ≤ 7 in HAM-D at the end of week six or week 14.
Incidence of severe adverse events Most adverse events associated with acupuncture are non-life-threatening and have been documented to have low incidence rates (e.g., local bleeding, \<1%; hematoma, \<3%; dizziness, \<1%, fainting, \< 1%; pain, \<10%; (Shen et al., 2012; Zhang et al., 2015). Therefore, in the present study, only major adverse events such as pneumothorax and infection were included in the analysis.
