This systematic review and meta-analysis are registered on PROSPERO (No. CRD42022372513).

The study was performed according to the Cochrane Handbook for Systematic Reviews of Interventions and followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Statement (PRISMA) guidelines. The research data in this review were drawn from eight databases: PubMed, EMBASE, Web of Science, the Cochrane Library, Chinese Biomedical Literature Database (CBM), China National Knowledge Infrastructure (CNKI), Wanfang Database, and VIP Database. RCTs published from inception to October 2022 were searched. There was no limitation on the language.

The search terms used were as follows: (“acupuncture” OR “electroacupuncture” OR “needl*”) AND (“depressive symptoms” OR “emotional depression” OR “depression”) AND (“cancer” OR “benign neoplasm*” OR “malignancy” OR “malignant neoplasm*” OR “neoplas*” OR “tumor*” OR “neoplasms”).

Two authors worked independently on the selection and extraction processes, and disagreements were settled by discussion with a third author. First, in light of predetermined inclusion and exclusion criteria, all of the titles and abstracts were screened. Full-text articles were then obtained for further assessment. Data extraction incorporated basic information about the article (the first author, publication year, country, sample size), general materials about patients (age, course, type of tumor, and current anti-tumor therapy), traits of intervention and control groups (retention, frequency, course), outcomes, and adverse events.

Risk of bias assessments were performed with the Cochrane risk of bias tool (24). Two researchers separately evaluated the following domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias. Each sphere was rated as having a low, high, or unclear risk of bias.

The meta-analysis was tackled by Review Manager version 5.3 software. HAMD, SDS, and QLQ-C30 scores belonging to continuous data are expressed as mean difference (MD) with a 95% confidence interval (CI). Effective rate and adverse effects belonging to dichotomous data are represented as risk ratios (RR) with 95% CI. We first assessed heterogeneity, and then subgroup analysis was conducted for the results with high heterogeneity. The selection of a fixed-effect model (p ≥ 0.1 and I ² ≤ 50%) or a random-effect model (p < 0.1 and I ² > 50%) was determined by the values of p and I ². In addition, a sensitivity analysis was implemented to test the stability of the results. If necessary, publication bias analysis was depicted by a funnel plot.

A total of 1,513 studies were searched from databases, of which 308 duplicate literature were removed and 1,156 publications were excluded for not conforming to the inclusion criteria according to the title and abstract. After examining the whole text, 10 articles were accessible for analysis ( Figure 1 ; Table 1 ).

The numbers of participants ranged from 58 to 120, and a total of 725 patients were included, with 361 in the acupuncture group and 364 in the control group. It was reported that two participants dropped out of the investigation owing to long distances and intense needling sensations (20). Among the 10 studies, two consisted of breast cancer (18, 20), and the rest incorporated various types of neoplasms. The distribution of multiple tumors in the 725 samples was as follows: 179 breast cancer, 107 gastrointestinal tumors (gastric cancer, intestinal cancer, esophagus cancer, anal cancer); 112 lung cancer; 75 gynecological tumors (cervical cancer, fallopian tube tumor, endometrial cancer, ovarian cancer, vulvar cancer); 36 liver cancer; 13 nasopharyngeal carcinoma; 11 prostate cancer; 9 lymphoma; 6 testicular cancer; 6 thyroid cancer; 4 pancreatic cancer; and 167 other tumors. All of the studies performed in China and tumor-related depression is common in middle-aged and elderly people. Five trials specified the tumor stage (16, 19, 20, 26, 27). The participants were on anti-tumor treatment in three trials (16, 17, 28). The anti-tumor therapeutic schemes mainly consisted of chemotherapy, surgery, chemotherapy, radiotherapy, endocrinotherapy, biotherapy, and targeted therapy. Of all the included studies, three compared acupuncture with usual care (18, 20, 25), and the others compared the combination of acupuncture and usual care with usual care. A host of RCTs used manual acupuncture in the experimental group, except for two articles that used electroacupuncture (19, 28). The usual care primarily covered conventional antidepressants. The top acupoints used for tumor-related depression were GV20, GV24⁺, LR3, LI4, PC6, and HT7. Patients in the studies received acupuncture from twice a week to once a day for 4 to 12 weeks, each session varied from 20 to 40 min. Moreover, it took at least one month for acupuncture to exert effects on tumor-related depression. Three studies chose the HAMD and SDS as the depression measurement (17, 18, 25), six studies assessed depression with the HAMD alone (16, 19, 20, 26–28), and one study used the SDS alone (29). QLQ-C30 was implemented in three trials (16, 19, 29). Three studies reported effective rates (20, 25, 27). Three studies (18, 20, 27) reported side effects (pain, bleeding, dizziness, headache, etc.) resulting from acupuncture and other symptoms (nausea, perspiration, dry mouth, etc.) triggered by antidepressant drugs. The adverse events were slight, and medical interventions were dispensable. There were no adverse events in the two articles (16, 17), and the remaining studies did not mention relevant information.

The major accounts of the risk of bias were relevant to the blinding of participants and personnel. The risk of bias in the randomization process was quite low ( Figure 2A ).

For random sequence generation, five studies (16, 18, 25, 27, 29) used a random digit table; two studies took advantage of a computer to generate random numbers (17, 20); and the remaining did not point out techniques in details but generated randomly (19, 26, 28); all of them were considered to have a low risk of bias. For allocation concealment, one study (20) had a low risk of bias, and the others were regarded as having an unclear risk of bias. With regard to the blinding of participants and personnel, only two trials (20, 27) had a low risk of bias due to the specificity of acupuncture therapy. For blinding of outcome assessment, there was one article (17) having a low risk of bias; the rest were assessed as an unclear risk of bias. As for incomplete outcome data, one study (29) was deemed to have a high risk of bias because the safety indexes were absent, which had been mentioned in outcome measures. Another one (27) only illustrated the adverse events of the observation group but not the reported details of the control group. We cannot be certain about the data integrity; consequently, it is considered an unclear risk of bias. For selective reporting, five trials (19, 25, 26, 28, 29) had an unclear risk of bias as they did not report side effects; all the other trials had a low risk of bias. In light of other biases, one trial (26) did not explain the frequency of acupuncture. We cannot tell whether the observation group has the same frequency as the control group; thus, it has an unclear risk of bias ( Figure 2B ).

Previous investigations described acupuncture and acupressure as being effective as drug treatments, and acupressure treatments were mainly auricular points (15). Our work provides updated evidence to inspect the effect and safety of conventional acupuncture for cancer tumor-related depression. There were superior effectiveness and fewer adverse reactions to acupuncture compared to usual care. The analysis stated that acupuncture was an effective and safe therapy for tumor-related depression regardless of patients’ tumor stages or types. Acupuncture as a supplement can significantly enhance HAMD, SDS, and QLQ-C30 scores, particularly acupuncture plus usual care. The effect of electroacupuncture on HAMD scores remained to be discussed since there were only two pieces of literature.

The diagnostic criteria for tumor-related depression were different among the included studies. Three studies (16, 19, 25) applied the Chinese classification and diagnostic criteria of mental disorders (CCMD-3); the same amount of studies (17, 20, 28) used the diagnostic and statistical manual of mental disorders (DSM-5); and one study (18) utilized the international classification of disease (ICD-10). In different studies, HAMD and SDS were applied to diagnose and classify depression. The above contents may be the sources of high heterogeneity. Furthermore, heterogeneity might be correlated with various features of patients (type, status, current anti-tumor treatment, etc.) and different selective serotonin reuptake inhibitor (SSRI) antidepressants like fluoxetine hydrochloride capsule (25), sertraline hydrochloride (17, 18, 20), escitalopram oxalate tablets (18, 27), or all of the above (20). Most studies recruited manifold categories of tumor, except for two studies that involved breast cancer patients. While the subgroup analysis performed to explore whether acupuncture was more beneficial for breast cancer patients was inaccessible. Hence, we were incapable of judging the differentiation in multiple tumor types.

Due to special characteristics of acupuncture manipulation, major trials failed to blind participants and personnel. To a certain extent, this could lead to subjective consequences. Traditional Chinese medicine emphasizes treatment based on syndrome differentiation. The acupoints, duration, frequency, and course varied widely depending on individuals. Unfortunately, we were confined to conducting a subgroup analysis to determine the effect of different acupuncture regimens on tumor-related depression because few articles were available.

GV20 was the most frequently used point, in accordance with the result of metrological analysis (36), followed by GV24⁺, LR3, LI4, PC6, and HT7. Acupuncture stimulation at GV20 and GV24⁺ could prevent and treat depression by modulating the expression of multiple neurotrophic factors (37, 38). Acupuncture at LR3 and LI4 could raise the level of 5-HT and NE (39), as well as upregulate BDNF (40). LR3 and PC6 also had an advantage in regulating glial cell line-derived neurotrophic factor (GDNF) production (41). Stimulating at the PC6 acupoint inhibited the pathological state of the HPA axis in the depressive rat model (42). Acupuncture stimulation at HT7 can ameliorate depression in rat models by increasing 5-HT expression and BDNF levels (43).

Additionally, different RCTs employed different HAMDs (17 or 24 points). These were concerned with heterogeneity. Though a multitude of studies did not take the efficacy of acupuncture as an observation index, relevant depression scales were put into use, indicating that a state of depression was prevalent in tumor patients. As to safety, acupuncture is inevitable to produce adverse events on account of insertion. However, they can recover quickly without being tackled deliberately, as long as properly operated by a professional acupuncturist.

Psychological problems like depression have a great affection on the treatment of tumors. Previous studies have reported various acupuncture techniques for managing various tumor-related complications (44–47), such as pain, fatigue, insomnia, and so on. Although the conclusions indicated the effectiveness and safety of acupuncture, the effect of conventional acupuncture on tumor-related depression alone was lacking. It may be easier to understand the effect of acupuncture by concentrating exclusively on the treatment itself. We performed more rigorous inclusion criteria to screen eligible RCTs. Unlike those studies, this one demonstrated that acupuncture had a beneficial effect on tumor-related depression and improved characters’ quality of life with fewer side effects. However, in terms of improving HAMD and SDS, acupuncture was a complementary therapy, not an alternative.

There were a number of deficiencies in this review. First, a variety of studies did not design a double-blind trial with sham acupuncture as a placebo control, and several studies had small sample sizes. It is universally acknowledged that various accompanying symptoms of tumors, such as pain, sleep disorders, and other complications, may aggravate depression degree; however, the included studies did not explain the impact on tumor-related depression triggered by these factors, which may have a modest effect on our work outcomes. The overall qualities of integrated studies were not extremely compelling to support our results. Second, there were merely 10 studies; we were confined to processing subgroup analysis of various types, stages, current anti-tumor treatment, etc. High heterogeneity in each data would make a difference to the accuracy of the results. Third, the publication bias analysis described via funnel plot was infeasible because smaller quantities of studies were contained. Fourth, due to the specificity of acupuncture therapy, most studies cannot completely assure the implementation of the blinding process. Depression as a mental disorder is usually assessed by self-rating scales, which could magnify the placebo effect. Thus, it is essential to conduct a more rigorous design, like using sham acupuncture, to ensure the blinding process. Last but not least, all the trials we included were conducted in China, and there may exist ethnic differences. In conclusion, the results of meta-analysis should be treated carefully, and it is crucial to adopt additional outstanding studies for further research to support the evidence.