Effects of acupuncture on cancer-related fatigue and quality of life in breast cancer survivors: A systematic review and meta-analysis of randomized controlled trials

Abstract

Background:

Cancer-related fatigue (CRF) and impaired quality of life (QoL) are common consequences among breast cancer survivors. While acupuncture is increasingly applied as a complementary therapy, its efficacy remains unclear owing to inconsistent results across randomized controlled trials (RCTs) that have applied diverse outcome measures. This study aims to analyze how acupuncture affects CRF and both general and domain-specific QoL among breast cancer survivors.

Methods:

Seven databases were comprehensively searched until May 1, 2025. We included RCTs comparing acupuncture with control interventions in breast cancer survivors reporting CRF or impaired QoL. Methodological quality and evidence certainty were assessed with the Grading of Recommendations, Assessment, Development, and Evaluation framework and the Cochrane Risk of Bias tool.

Results:

Analysis of 13 RCTs (n = 963) showed acupuncture significantly reduced CRF severity on the Brief Fatigue Inventory (weighted mean difference [WMD]: −1.31, 95% confidence interval [CI] −2.04–−0.59, P = .0004, I² = 37%) and Cancer Fatigue Scale (WMD: −5.89, 95% CI −7.62–−4.17, P < .00001, I² = 0%), and improved clinical response rates (risk ratio: 1.46, 95% CI 1.14–1.87, P = .003, I² = 44%). Significant benefits were also observed for the Pittsburgh Sleep Quality Index (WMD: −1.65, 95% CI −2.30–−1.01, P < .00001, I² = 0%) and the Hospital Anxiety and Depression Scale (standardized mean difference: −0.54, 95% CI −0.79–−0.29, P < .0001, I² = 0%). QoL improvement was significant when measured by the breast cancer-specific Functional Assessment of Cancer Therapy-Breast (standardized mean difference: 0.33, 95% CI 0.09–0.56, P = .007, I² = 5%) but not with generic instruments. The evidence level was between moderate and very low. Acupuncture was found to be safe, with only minor adverse events reported.

Conclusion:

Acupuncture can effectively and safely alleviate CRF. Its benefits for QoL are best captured by disease-specific measures. Future high-quality RCTs with standard protocols and longer follow-up are warranted to confirm these benefits.

1. Introduction

Breast cancer is not only the most frequently seen cancer type but also a major factor for cancer-associated death in females globally.^[1,2] Progress in regimens to detect the disease early and treat it with multiple modalities has significantly elevated patient survival, increasing the number of breast cancer survivors.^[3] While this represents a major achievement in the field of oncology, addressing long-term sequelae of both cancer and treatments is increasingly warranted, as it can profoundly impair the daily functioning and overall well-being of survivors.^[4]

Among these persistent sequelae, cancer-related fatigue (CRF) serves as a frequent and destructive symptom for those surviving breast cancer.^[5] CRF exhibits the typical feature of an overwhelming, continuous sense of exhaustion, out of proportion to recent activities or unrelieved by rest.^[6] The symptom persists for months and years post-active treatment, severely compromising physical functioning, emotional stability, and the ability to resume normal social and occupational activities.^[7,8] The pathophysiology of CRF is multifactorial, involving inflammatory pathways, neuroendocrine dysfunction, and psychological distress; this complicates its clinical management.^[9] Conventional interventions, including pharmacological treatments and exercise therapy, exhibit variable efficacy and are frequently associated with limitations such as adverse effects or accessibility barriers.^[10,11]

Beyond specific symptoms like fatigue, the quality of life (QoL) of those who survive breast cancer encompasses a broader multidimensional construct, including functional, psychological, physical, and social well-being.^[12] Critical domains like sleep quality (commonly analyzed via the Pittsburgh Sleep Quality Index [PSQI]) or emotional state (frequently examined via the Hospital Anxiety and Depression Scale [HADS]) are QoL’s integral components. Importantly, while a survivor may not experience severe CRF, they may still suffer from diminished QoL owing to other factors such as pain, sleep disturbances, body image concerns, anxiety, depression, or financial toxicity.^[13,14] Subsequently, effective survivorship care warrants interventions that address both high-prevalence specific symptoms such as CRF, measured by tools such as the Brief Fatigue Inventory (BFI),^[15] Multidimensional Fatigue Inventory (MFI),^[16] and Cancer Fatigue Scale (CFS), and holistic well-being,^[17] as measured via the Functional Assessment of Cancer Therapy-Breast (FACT-B) or European Organization for Research and Treatment of Cancer QoL Questionnaire Core 30 (EORTC QLQ-C30).^[18,19]

Acupuncture, a cornerstone of traditional Chinese medicine, arouses wide interest as a complementary therapy in oncology supportive care. In this procedure, fine needles are inserted at certain body points to regulate physiological functions. The proposed mechanisms for its efficacy include neurotransmitter modulation, proinflammatory cytokine reduction, and autonomic nervous system regulation.^[20] For breast cancer survivors, acupuncture is investigated for its effect on alleviating a spectrum of treatment-induced adverse effects, including those affecting both specific symptoms and overall QoL.^[21]

Several randomized controlled trials (RCTs) analyze acupuncture's effects against CRF and QoL in this population. However, the results have been inconsistent. The variation in primary outcome measures across studies is a crucial factor contributing to this heterogeneity: some trials targeted survivors with significant fatigue, utilizing specific tools such as the BFI or MFI^[22,23]; in contrast, others recruited broader cohorts and employed general or cancer-specific QoL questionnaires, like FACT-B or EORTC QLQ-C30, which incorporate key subdomains such as sleep and mood.^[18,19] An intervention may demonstrate significant benefits for a specific symptom (e.g., fatigue measured by BFI) or a key QoL domain (e.g., sleep measured by PSQI) without significantly affecting the global QoL summary score in a heterogeneous sample, and vice versa. Previous systematic reviews have frequently pooled these diverse outcomes, potentially obscuring the precise effects of acupuncture.^[24]

Therefore, to provide a nuanced and clinically relevant evidence-based conclusion, the current work was carried out to combine RCT data to evaluate the functions of acupuncture in 2 distinct but related sets of outcomes in breast cancer survivors: the severity of CRF (using fatigue-specific measures) and QoL, encompassing both its broad dimensions and key specific domains, including sleep and emotional well-being (measured by FACT-B, EORTC QLQ-C30, PSQI, and HADS). By outcome stratification, we aimed to clarify the specific benefits of acupuncture and inform its targeted application in survivorship care.

2. Materials and methods

2.1. Study registration

This work was implemented as described by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.^[25] Our study protocol was registered in the International Prospective Register of Systematic Reviews (registration number, CRD420251136387).This systematic review was conducted in strict accordance with the preregistered protocol (International Prospective Register of Systematic Reviews CRD420251136387), with no material deviations.

2.2. Database and search strategy

The following databases were systematically searched until May 1, 2025: EMBASE, MEDLINE, PubMed, Science Citation Index, Cochrane Library, China National Knowledge Infrastructure, and Wanfang Database. Medical Subject Headings terms and population-related free-text keywords (“acupuncture” OR “electroacupuncture” OR “auricular acupuncture”) AND (“Breast Neoplasms” OR “breast cancer”) AND (“cancer-related fatigue” OR “CRF” OR “cancer fatigue”) AND (“Quality of Life” OR “QOL” OR “Health Related Quality Of Life” OR “HRQOL”) AND (“randomized clinical trial” OR “RCT”) were used. Besides, a search strategy for specific syntax in every database was adopted. There were no restrictions on publication status and language. The search formula in PubMed is as follows. (“acupuncture” [MeSH Terms] OR “electroacupuncture” [MeSH Terms] OR “auriculotherapy” [MeSH Terms] OR “auricular acupuncture” [Title/Abstract] OR acupressure [MeSH Terms] OR acupressure [Title/Abstract]) AND (“breast neoplasms” [MeSH Terms] OR “breast cancer” [Title/Abstract]) AND (“cancer-related fatigue” [Title/Abstract] OR “CRF” [Title/Abstract]) (“quality of life” [MeSH Terms] OR “quality of life” [Title/Abstract] OR “health related quality of life” [Title/Abstract]) AND (“randomized controlled trial” [Publication Type] OR randomized [Title/Abstract] OR “RCT” [Title/Abstract] OR trial [Title/Abstract]).

2.3. Eligibility criteria

Two independent researchers selected the studies to be included. First, they selected abstracts in the retrieved literature, obtained, and examined full texts of qualified studies based on our predefined criteria.

2.3.1. Inclusion criteria

The inclusion criteria were formulated according to the PICOS principles. Participants (P): Surviving adult breast cancer patients (≥ 18 years), regardless of cancer stage or current treatment status, who reported CRF or impaired QoL. Intervention (I): Any form of acupuncture, including but not limited to hand acupuncture, electroacupuncture, acupressure, or auricular acupuncture, used alone or as an adjunct to usual care (UC). Controls (C): Controls who received usual treatment, sham acupuncture (for example, non-penetrating acupuncture at non-acupoints), waitlist controls, no treatment, or active non-acupuncture treatment. Outcome (O): Primary outcome: severity of CRF measured by a validated scale such as the BFI, the MFI, or the Cancer Fatigue Scale (CFS); QoL measured using generic or cancer-specific instruments, such as the Functional Assessment of Breast Cancer (FACT-B), the Short Form-12 Health Survey (SF-12), or the EORTC QLQ-C30. Secondary outcomes included sleep quality (PSQI), emotional state (HADS), and clinical response rate of CRF. Study Design (S): RCT, including blinded and unblinded methods.

2.3.2. Exclusion criteria

Studies involving animal models or nonhuman subjects, studies where acupuncture was part of a complex intervention package and its effects could not be isolated, studies without a control or comparison group, duplicate publications or secondary analyses without original data, and studies that did not report relevant outcomes or provided insufficient data for analysis.

2.4. Study screening and data collection

Based on our preset eligibility criteria, our screened articles were evaluated through title and abstract-reading. Thereafter, full texts of qualified works were acquired and thoroughly assessed. The standard form was used to collect data, including the first author, publication year, country, sample size, mean age, breast cancer stage, details regarding the intervention and control groups, outcome measures, any reported adverse events (AEs), and results. Two independent reviewers (Zi Yang, Xinyue Sun) screened titles and abstracts and subsequently assessed the full texts of potentially eligible studies. Data were extracted independently using a standardized form. Any discrepancies in study selection or data extraction were resolved through discussion between the 2 reviewers. If consensus could not be reached, a third senior reviewer (Kuanyu Wang) was consulted for arbitration.

2.5. Risk of bias (ROB) evaluation

The ROB for every enrolled article was evaluated by the Cochrane Collaboration’s ROB evaluation tool covering 7 domains: random sequence generation (selection bias), allocation concealment (selection bias), participant and personnel blinding (performance bias), outcome assessment blinding (detection bias), insufficient outcome data (attrition bias), selective reporting (reporting bias), and additional possible bias sources. Every domain was assigned a low (+), high (−), or unclear (?) risk. Two independent researchers (Zi Yang, Xinyue Sun) evaluated the ROB. Any discrepancies were settled through negotiation with another reviewer (Qingquan Dai).

2.6. Statistical analysis

Review Manager (RevMan) software (Version 5.3; The Cochrane Collaboration) was used to analyze data. For continuous outcomes determined by this scale, mean difference (MD) was calculated. In contrast, for those measured using diverse scales, standardized MD (SMD) was determined. Risk ratio (RR) was determined for dichotomous outcomes. Effect estimates were evaluated based on their 95% confidence intervals (CIs). Among-study statistical heterogeneities were evaluated using I² statistics and the chi-square test. An I²-value of > 50% suggested obvious heterogeneities. Given the expected clinical and methodologic diversity among the studies (e.g., differences in acupuncture protocols, control interventions, and patient populations), all meta-analyses were performed with the use of random-effects models to provide more conservative and generalizable effect estimates. Subgroup analyses were prespecified according to the type of control intervention (sham acupuncture and UC/waitlist) to explore potential sources of heterogeneity. All P values are reported accurately. For the meta-analysis of QoL, although the included studies employed different instruments (e.g., FACT-B, SF-12, QLQ-C30), we pooled the SMDs to provide an overall estimate of the general QoL construct. We acknowledge that this approach introduces conceptual heterogeneity, and therefore, we prioritized subgroup analysis by specific instrument to interpret the findings. For missing data, the incomplete data required were acquired from original study authors to complete the dataset. Every article was excluded in sequence during sensitivity analysis for evaluating the pooled result robustness. This “leave-one-out” sensitivity analysis was performed to determine whether the pooled effect size of key outcomes (CRF measured by BFI and CFS, and QoL measured by FACT-B) was unduly influenced by any single study. The results are reported in the text and demonstrate the robustness of our primary findings. A funnel plot was used to test the risk of publication bias.

2.7. Certainty of evidence (CoE)

CoE was analyzed by the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework.^[26] This framework classifies confidence in estimated effects into 4 types: high, moderate, low, or very low. CoE for every key outcome was presented by generating a summary of findings table, facilitating transparent and structured result interpretation.

3. Results

3.1. Search outcomes and study features

First, during our search across multiple databases, we identified 1157 potentially relevant records. After removing duplicates, 389 unique citations were examined by reading their titles and abstracts. According to our preset criteria, the full texts of 87 studies were evaluated. After assessment, 13 RCTs meeting all these criteria were subjected to meta-analysis (Fig. 1).^{[22,23,28–38]}

Figure 1.

PRISMA 2020 flow diagram for new systematic reviews which included searches of databases and registers only. *Consider, if feasible to do so, reporting the number of records identified from each database or register searched (rather than the total number across all databases/registers). **If automation tools were used, indicate how many records were excluded by a human and how many were excluded by automation tools. Adapated from Page et al.^[27]

Table 1 offers an integrative overview of the baseline features of these 13 RCTs. They were published between 2011 and 2025 and exhibited considerable diversity in design and execution. The sample sizes ranged from 12 to 302. Geographically, these trials were conducted across several countries, including China (n = 6),^{[31,33,35–38]} the United States (n = 4),^{[22,29,30,32]} Germany (n = 1),^[28] Australia (n = 1),^[34] and the United Kingdom (n = 1),^[24] contributing to a multinational evidence base. The study populations exclusively included breast cancer survivors, and the mean patient age was 42 to 60 years across the different trials. Participants represented a broad spectrum of the cancer survivorship continuum, including individuals actively undergoing chemotherapy, those who had recently completed treatment, and long-term survivors. The universal inclusion criterion was CRF or impaired QoL. The intervention characteristics summarized in Table 1 highlight the various acupuncture modalities investigated. These involved manual acupuncture, acupressure, auricular acupuncture, acupoint catgut embedding, and electroacupuncture. The treatment regimens substantially differed in both frequency and duration. The number of acupuncture sessions administered across the studies ranged from 4 to 20, which were delivered over treatment periods lasting 4 weeks to 6 months. This heterogeneity reflects the differing clinical practices and experimental designs explored in the literature. The control groups were equally diverse, encompassing several comparison types. These included sham acupuncture (using superficial or non-penetrating needling at non-acupoints to control for placebo effects), UC (representing standard medical management), waitlist control (assessing the effect of time vs active intervention), and active non-acupuncture comparators (such as gabapentin in 1 study).^[28] This allowed for a nuanced analysis of the specific effects of acupuncture beyond common factors. The outcome measures systematically extracted and presented in Table 1 include both primary and secondary outcomes. Our primary outcomes included CRF, measured via validated measures such as BFI,^{[22,30,34,37]} MFI,^[23,30] and CFS.^[33,35] QoL was evaluated using comprehensive, well-validated measures like FACT-B,^[28,36,37] EORTC QLQ-C30,^[32] and SF-12.^[28] Our key secondary outcomes were sleep quality, evaluated using the PSQI,^{[22,29,33,35–37]} emotional state, assessed using the HADS,^[22,36,37] and clinically assessed efficacy rates for CRF.^[33,35,38]

Table 1.

Enrolled RCT features.

Author (Year) Country	Sample size/cancer stage/Population/mean age (Years)	Intervention (regimen, Randomized/Analysed)	Control (regimen, Randomized/Analysed)	Outcome measuremnts	Result: Mean (SD)/RR	AE
Brinkhaus (2019) Germany	150//I–III/female with newly diagnosed breast cancer A: 51.4; B: 50.6	(A) AT (NR, 6 months, 6 sessions, n = 75/75) + UC	(B) UC (n = 75/75)	1. FACT-B 2.SF-12	1. A: −2 (15.44) B: −5.3 (19.96) (P = .26) 2. A: −46.6 (10.65) B: 47.7 (11.08) (P = .54)	AT: hematoma, pain (n = 44); Chemotherapy: GI disorders, general disorders (n = 1188)
Garland (2017) USA	58/0–III/Breast cancer survivors showing hot flashes A: NR; B: NR	(A) EA (twice/week, 2 weeks, later once/week, 6 weeks, n = 30/30)	(B) Gabapentin (900 mg/day (300 mg TID) for 8 weeks, n = 28/28	1. PSQI	1. A: −2.6 (3.2) B: −0.8 (3) (P = .03)	None
Johnston (2011) USA	12/NR/Breast cancer survivors with CRF A: 55; B: 53	(A) AT (once/week, 8 weeks, n = 6/5) + UC	(B) UC (n = 7/7)	1. BFI	1. A: −4.2 (1.32) B: −1.62 (2.2) (P = .01)	None
LI (2020) China	40/I–III/Breast cancer survivors undergoing adjuvant chemotherapy A: 47.5; B = 42; C: 50.5	(A) AT (once/week, 20 weeks, n = 20/18)	(B) Sham AT (nonpoint superficial needling, n = 10/10) (C) UC (n = 10/9)	1. MFI-20	1. A: −3.3 (15.96) B: 9.7 (12.65) C: 11 (17.82) A vs B (P = .02) A vs C (P = .04)	No serious adverse events
Lu (2019) USA	40/I–III/Breast cancer survivors undergoing adjuvant chemotherapy A: NR; B: NR	(A) AT (18 times of acupuncture over 8 weeks, n = 20/16)	(B) WLC (n = 20/17)	1. EORTC QLQ-C30	1. A: −12 (15.2) B: −2 (12.3) (P = .05)	No serious adverse events
Mao (2014) USA	67/I–III/Postmenopausal women with breast cancer A: 57.5; B: 60.9; C: 60.6	(A) EA (twice/week, 2 weeks, then once/week, 6 weeks, n = 22/21)	(B) Sham AT (nonpoint superficial needling, n = 22/20) (C) WLC (n = 23/22)	1. BFI 2. PSQI 3. HADS	1. A: −1.4 (2.85) B: −0.6 (2.35) C: 0.5 (1.72) A vs B (P = .33) A vs C (P < .01) 2. A: −1.4 (3.41) B: −0.8 (2.46) C: 0.1 (2.82) A vs B (P = .52) A vs C (P = .12) 3.HADS-A A: −1.1 (2.41) B: −0.05 (1.6) C: 0.2 (2.82) A vs B (P = .12) A vs C (P = .02) HADS-D A: −1.2 (3.29) B: −0.8 (2.02) C: 1.2 (2.62) A vs B (P = .65) A vs C (P = .01)	NR
Meng (2024) China	60/III and IV/Advanced breast cancer with Spleen Qi Deficiency CRF A: NR; B: NR	(A) AT (3–5 times daily, 8 weeks, n = 30/30) + (B)	(B) Modified Huangqi Jianzhong Decoction (daily for 8 weeks, n = 30/30)	1. CFS 2. PSQI 3. CRF Clinical efficacy	1. A: 28.87 (3.88) B: 34.87 (4.31) (P < .00001) 2. A: −7.34 (2.82) B: −5.66 (3.6) (P = .04) 3. RR: 1.81 [1.29, 2.55] (P < .01)	NR
Molassiotis (2012) UK	302/I–III/Breast cancer survivors with CRF A: 52; B: 53	(A) Acupuncture (once/week, 6 weeks, n = 227/181) + UC	(B) UC (n = 75/65)	1. MFI	1. A: −3.72 (1.22) B: −0.62 (1.24) (P < .00001)	None
Smith (2012) Australia	30/NR/Breast cancer survivors with CRF A: 55; B: 53; C: 58	(A) AT (twice/week, 3 weeks; once/week, 3 weeks, n = 10/9)	(B) Sham AT (nonpoint superficial needling, n = 10/10) (C) WLC (n = 10/10)	1. BFI	1. A: −3.1 (2.19) B: −1.5 (1.76) C: −1.1 (1.65) A vs B (P = .08) A vs C (P = .03)	NR
Wang (2025) China	60/I–III/Breast cancer survivors with CRF A: 47; B: 46.27	(A) ACE (once every 7 days for 4 weeks, n = 30/30)	(B) Conventional AT (once daily, 6 times/week for 4 weeks, n = 30/30)	1. CFS 2. CRF Clinical efficacy	1. A: 19.07 (5.78) B: 24.73 (6.41) (P < .01) 2.RR: 1.23 [0.96, 1.57] (P = .1)	NR
Zhang (2021) HK, China	30/I–IV/Breast cancer survivors receiving or completing chemotherapy ≤ 6 months A: 52.5; B: 52.7	(A) EA + AA (Twice weekly for 6 weeks, n = 15/13)	(B) WLC (n = 15/15)	1. PSQI 2. HADS 3. FACT-B	1. A: −4.7 (2.57) B: −2.4 (2.8) (P = .02) 2.HADS-A A: −2.7 (1.74) B: −1.2 (1.9) (P = .04) HADS-D A: −2.9 (1.82) B: −0.8 (1.99) (P < .01) 3. A: 18.8 (10.18) B: 10.5 (11.11) (P = .06)	Mild: skin allergy (n = 1), pain (n = 1), bruising (n = 1) in (A); skin allergy (n = 1), bruising (n = 1), palpitations (n = 1) in (B)
Zhang (2023) HK, China	138/I–IV/Breast cancer survivors undergoing or have completed chemotherapy ≤ 6 months A: 52.5; B: 52.7	(A) EA + AA (15 times over 18 weeks, n = 69/67)	(B) Sham AT (non-insertive needles + sham AA, n = 69/56)	1. BFI 2. PSQI 3. HADS 4. FACT-B	1. A: −2.1 (1.84) B: −1.6 (1.86) (P = .14) 2. A: −4.6 (4.09) B: −3.7 (2.98) (P = .16) 3.HADS-A A: −2.1 (2.45) B: −1 (2.42) (P = .01) HADS-D A: −2 (2.66) B: −1.5 (2.61) (P = .3) 4. A: 14.7 (12.91) B: 9.5 (12.69) (P = .03)	mild: Bruising: n = 6 in (A); Auricular skin allergy: n = 4 in (B)
Zhou (2018) China	64/NR/Breast cancer patients with CRF A: 52; B: 50	(A) AT (Every 2 days for 1 month, n = 32/32)	(B) UC (n = 32/32)	1. CRF Clinical efficacy	1. RR: 1.5 [1.01, 2.24] (P = .05)	NR

AA = auricular acupressure, ACE = acupoint catgut embedding, AE = adverse events, AT = acupuncture therapy, avg. = average, BFI = Brief Fatigue Inventory, CFS = Cancer Fatigue Scale, CRF = cancer-related fatigue, EA = electroacupuncture, FACT-B = functional assessment of cancer therapy-breast, GI = gastrointestinal motility, HADS-A/D = hospital anxiety and depression scale-anxiety/depression, MFI = Multidimensional Fatigue Inventory, HK = HongKong, n = number of patients, NR = not reported, PSQI = Pittsburgh Sleep Quality Index, QLQ-C30 = European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30, RCT = randomized controlled trials, RR = risk ratio, SF-12 = Short Form-12 Health Survey, UC = usual care, UK = United Kingdom, USA = United States of America, WLC = wait list control.

Table 2 provides a detailed summary of our meta-analysis of all quantitatively pooled outcome measures. The table meticulously presents the study and participant number associated with every comparison. It specifies the statistical method employed: MD for outcomes measured using an identical scale, SMD for outcomes measured using diverse scales, and RR for dichotomous data. The pooled effect estimates were reported using respective 95% CIs.

Table 2.

Meta-analysis results.

Comparison	No. of Studies	No. of Subjects	Statistical Method	Effect Size	P
MFI	3	301	MD (IV, random, 95% CI)	−8.22 [−16.56, 0.12]	.050
MFI (AT vs UC)	2	273	MD (IV, random, 95% CI)	−6.49 [−16.58, 3.59]	.210
MFI (AT vs sham AT)	1	28	MD (IV, random, 95% CI)	−13.00 [−23.76, −2.24]	.020
BFI	6	257	MD (IV, random, 95% CI)	−1.31 [−2.04, −0.59]	< .001
BFI (AT vs WLC)	2	62	MD (IV, random, 95% CI)	−1.94 [−3.04, −0.84]	< .001
BFI (AT vs sham AT)	4	195	MD (IV, random, 95% CI)	−1.04 [−1.89, −0.19]	.020
CFS	2	120	MD (IV, random, 95% CI)	−5.89 [−7.62, −4.17]	< .001
CRF Clinical efficacy	3	184	RR (M−H,random, 95% CI)	1.46 [1.14, 1.87]	.003
HADS	6	388	SMD (IV, random, 95% CI)	−0.54 [−0.79, −0.29]	< .001
HADS-A	3	194	SMD (IV, random, 95% CI)	−0.55 [−0.84, −0.27]	< .001
HADS-D	3	194	SMD (IV, random, 95% CI)	−0.60 [−1.15, −0.05]	.030
PSQI	6	372	MD (IV, random, 95% CI)	−1.65 [−2.30, −1.01]	< .001
Qol	5	484	SMD (IV, random, 95% CI)	0.11 [−0.23, 0.46]	.520
FACT-B	3	301	SMD (IV, random, 95% CI)	0.33 [0.09, 0.56]	.007
SF-12	1	150	SMD (IV, random, 95% CI)	−0.10 [−0.42, 0.22]	.540
QLQ-C30	1	33	SMD (IV, random, 95% CI)	−0.71 [−1.41, −0.00]	.050

AT = acupuncture therapy, BFI = Brief Fatigue Inventory, CFS = Cancer Fatigue Scale, CI = confidence interval, CRF = cancer-related fatigue, FACT-B = functional assessment of cancer therapy-breast, HADS-A/D = Hospital Anxiety and Depression Scale-Anxiety/Depression, MD = mean difference, MFI = Multidimensional Fatigue Inventory, PSQI = Pittsburgh Sleep Quality Index, QLQ-C30 = Quality of Life Questionnaire Core 30, Qol = quality of life, RR = risk ratio, SD = standard deviation, SF-12 = Short Form-12 Health Survey, SMD = standardized mean difference, WLC = wait list control.

3.2. RoB evaluation

Methodological quality was assessed by the Cochrane RoB tool. Overall, significant methodological concerns impeded the evidence, with no study rated as low risk in each domain.

Most studies (n = 12) exhibited a low risk concerning random sequence generation. However, allocation concealment was rarely mentioned, with 4 articles rated as unclear and 1 as high risk. Participant and personnel blinding (performance bias) was a major limitation. Only 5 studies using credible sham acupuncture controls achieved low risk; in contrast, studies using active, waitlist, or UC controls (n = 8) were rated high or unclear risk because of the impossibility of blinding. The outcome measuring blinding (detection bias) was inconsistently reported, with 6, 5, and 2 studies with unclear, low, and high risks, respectively. Incomplete outcome data were well-managed in most studies (n = 12, low risk). Selective reporting was not a major concern for most studies (n = 10, low RoB). There were only 3 articles rated as unclear risk. Other potential biases were generally low risk (n = 13) (Figs. 2 and 3).

Figure 2.

Risk of bias graph.

Figure 3.

Risk of bias summary.

3.3. Risk of publication bias

The funnel plot analysis of CRF and QoL showed that the points were roughly distributed on both sides of the effect line, and the effect size point estimates were close. The distribution in the funnel plot should be more symmetrical, and the overall risk of publication bias was low (Figs. 4 and 5).

Figure 4.

Funnel plot for QoL. QoL = quality of life, EORTC QLQ-C30 = European Organization for Research and Treatment of Cancer QoL Questionnaire Core 30, FACT-B = functional assessment of cancer therapy-breast, SE = standard error, SF-12 = Short Form-12 Health Survey, SMD = standardized mean difference.

Figure 5.

Funnel plot for BFI subgroup analysis. AT = acupuncture therapy, BFI = Brief Fatigue Inventory, MD = mean difference, SE = standard error, WLC = wait list control.

3.4. Outcome measures

3.4.1. BFI

Four RCTs applied BFI scores to assess CRF. The combined results exhibited a significant decrease in scores (random-effects model, weighted mean difference [WMD]: −1.31 (points), 95% CI −2.04–− 0.59; P = .0004; I² = 37%). Subgroup analysis indicated that the reduction in BFI scores was more pronounced when acupuncture was compared to a waitlist control (MD: −1.94) than to sham acupuncture (MD: −1.04), suggesting that the therapeutic effect extends beyond nonspecific placebo effects. During subgroup analyses, the reduction in BFI scores remained statistically significant when comparing acupuncture and waitlist control groups (WMD: −1.94 points, 95% CI −3.04–− 0.84; P = .0006; I² = 0%) and the sham acupuncture group (WMD: −1.04 points, 95% CI −1.89–− 0.19; P = .02; I² = 35%). However, the effect size was more pronounced in the former comparison (Fig. 6).

Figure 6.

BFI scores of both acupuncture and control groups. AT = acupuncture therapy, BFI = Brief Fatigue Inventory, CI = confidence interval, IV = instrumental variables, SD = standard deviation, WLC = wait list control.

3.4.2. MFI

Two RCTs applied MFI scores to assess CRF. The combined results revealed that the scores were not significantly reduced (random-effects model, WMD: −8.22, 95% CI −16.56–0.12; P = .05; I² = 65%). The high heterogeneity and borderline significance warrant cautious interpretation. A prespecified subgroup analysis revealed a significant benefit of acupuncture compared to sham acupuncture (P = .02) but not compared to UC alone (P = .21). This pattern suggests that outcomes measured by the MFI may be particularly sensitive to patient expectations or the therapeutic context, highlighting the complexity of interpreting efficacy in acupuncture trials. Nonetheless, the acupuncture group was different from the Sham acupuncture group after subgroup analyses (P = .02). Nevertheless, the result remained nonsignificant when only UC was included in the control group (P = .21; I² = 61%) (Fig. 7).

Figure 7.

MFI scores of both acupuncture and control groups. AT = acupuncture therapy, MFI = Multidimensional Fatigue Inventory, CI = confidence interval, IV = instrumental variables, SD = standard deviation, UC = usual care.

3.4.3. CFS

Two RCTs applied CFS. Our meta-analysis revealed that the acupuncture group was different from the control group (random-effects model, WMD: −5.89; 95% CI −7.62–− 4.17; P < .00001; I² = 0%) (Fig. 8).

Figure 8.

CFS scores of both acupuncture and control groups. CFS = Cancer Fatigue Scale, CI = confidence interval, IV = instrumental variables, SD = standard deviation.

3.4.4. Clinical efficacy of CRF in those surviving Breast Cancer

Two RCTs evaluated the clinical efficacy of CRF. From our results, the acupuncture and control groups had an obvious difference (random-effects model, RR: 1.46; 95% CI, 1.14–1.87; P = .003; I² = 44%) (Fig. 9).

Figure 9.

Clinical efficacy of CRF in breast cancer survivors in both acupuncture and control groups. CI = confidence interval, CRF = cancer-related fatigue.

3.4.5. HADS

Three RCTs applied the HADS. Our meta-analysis revealed significantly improved HADS scores, favoring the acupuncture group over the control group (random-effects model, SMD: −0.54, 95% CI −0.79–− 0.29; P < .0001; I² = 26%). Subgroup analyses were conducted using the HADS subscales. For the anxiety subscale (HADS-Anxiety), 3 RCTs revealed markedly decreased scores (SMD:–0.55, 95% CI −0.84–− 0.27; P = .0002; I² = 0%). For the depression subscale (HADS-Depression), 3 RCTs also exhibited markedly improved scores (SMD: −0.60, 95% CI −1.15–− 0.05; P = .03; I² = 64%) (Fig. 10).

Figure 10.

HADS scores of acupuncture and control groups. CI = confidence interval, HADS = Hospital Anxiety and Depression Scale, IV = instrumental variables, SD = standard deviation.

3.4.6. PSQI

Six RCTs used the PSQI. Our results demonstrated that the acupuncture and control groups were markedly different (random-effects model, WMD: −1.65; 95% CI −2.30–− 1.01; P < .00001; I² = 0%) (Fig. 11).

Figure 11.

PSQI of both acupuncture and control groups. CI = confidence interval, IV = instrumental variables, PSQI = Pittsburgh Sleep Quality Index, SD = standard deviation.

3.4.7. QoL

Four RCTs evaluated QoL with comprehensive, well-validated measures. According to the pooled findings, the overall QoL scores showed no obvious difference in acupuncture compared with control groups (random-effects model, SMD: 0.11, 95% CI −0.23–0.46; P = .52; I² = 80%). This substantial heterogeneity likely stems, in part, from the conceptual and metric differences between the generic (SF-12, QLQ-C30) and breast cancer-specific (FACT-B) instruments pooled in this analysis. The considerable heterogeneity (I² = 80%) precludes a definitive conclusion from the pooled estimate. Subgroup analysis by instrument revealed a significant, positive effect on QoL when measured by the breast cancer-specific FACT-B (SMD: 0.33, P = .007, I² = 5%), but not with generic instruments (SF-12, EORTC QLQ-C30). This indicates that the benefits of acupuncture on QoL are most salient and consistently detectable in domains specifically relevant to breast cancer survivors, such as concerns about body image, arm symptoms, and sexual function, which are captured by FACT-B but not comprehensively by generic tools. Subgroup analysis was performed using the specific QoL instrument. Three RCTs with the FACT-B questionnaire reported markedly improved QoL in favor of acupuncture (SMD: 0.33, 95% CI 0.09–0.56; P = .007; I² = 5%). Differently, trials using SF-12 (SMD: −0.10, 95% CI −0.42–0.22; P = .54) and EORTC QLQ-C30 (SMD: −0.71, 95% CI −1.41–0.00; P = .05) did not reveal any differences (Fig. 12). A sensitivity analysis confirmed that the significant positive effect on FACT-B was robust and not driven by any single study; the pooled SMD remained statistically significant (range: 0.30–0.37) upon sequential exclusion of each trial.

Figure 12.

QoL of both acupuncture and control groups. CI = confidence interval, EORTC QLQ-C30 = European Organization for Research and Treatment of Cancer QoL Questionnaire Core 30, FACT-B = functional assessment of cancer therapy-breast, IV = instrumental variables, SD = standard deviation, SF-12 = Short Form-12 Health Survey

3.5. CoE

Nine outcomes were rated according to the GRADE system. They demonstrated moderate (n = 3), low (n = 4), and very low (n = 2) quality evidence (Table 3). The overall evidence was limited by methodological concerns, including the high or unclear RoB of some articles. Key limitations included inadequate participant and personnel blinding and unclear allocation concealment of certain trials. These factors downgraded the CoE for several outcomes. Despite these limitations, the populations, interventions, and comparators across the included articles exhibited a direct relation with our clinical question for this review, with no serious indirectness detected. However, there was inconsistency in some outcomes, including HADS-Depression (I² = 64%), indicating variability in effect estimates. In addition, imprecision was noted in several comparisons owing to wide CIs or small sample sizes, further decreasing the CoE. Table 3 elaborates on evidence quality evaluation for each outcome.

Table 3.

Summary of findings.

Outcomes	No of Studies (Participants)	CoE (GRADE)	Relative Effect (95% CI)	Absolute Effects (95% CI)	Comments
MFI (AT vs UC)	2 (273)	⊕⊕○○LOW	-	MD −6.49 (−16.58, 3.59)	RoB: serious Inconsistency: nonserious Indirectness: serious Inaccuracy: nonserious Publication bias: serious
BFI (AT vs WLC)	2 (62)	⊕⊕⊕○MODERATE	-	MD −1.94 (−3.04, −0.84)	RoB: serious Inconsistency: nonserious Indirectness: nonserious Inaccuracy: nonserious Publication bias: nonserious
BFI (AT vs sham AT)	4 (195)	⊕⊕○○LOW	-	MD −1.04 (−1.89, −0.19)	RoB: serious Inconsistency: nonserious Indirectness: nonserious Inaccuracy: serious Publication bias: nonserious
CFS	2 (120)	⊕⊕○○LOW	-	MD −5.89 (−7.62, −4.17)	RoB: serious Inconsistency: nonserious Indirectness: serious Inaccuracy: nonserious Publication bias: nonserious
CRF Clinical efficacy	3 (184)	⊕○○○VERY LOW	RR 1.46 (1.14–1.87)	270/1000 (82–511)	RoB: serious Inconsistency: serious Indirectness: serious Inaccuracy: nonserious Publication bias: nonserious
PSQI	6 (372)	⊕⊕⊕○MODERATE	-	MD −1.65 (−2.30, −1.01)	RoB: serious Inconsistency: nonserious Indirectness: nonserious Inaccuracy: nonserious Publication bias: nonserious
HADS-A	3 (194)	⊕⊕○○LOW	-	SMD −0.55 (−0.84, −0.27)	RoB: serious Inconsistency: nonserious Indirectness: nonserious Inaccuracy: serious Publication bias: nonserious
HADS-D	3 (194)	⊕○○○VERY LOW	-	SMD −0.60 (−1.15, −0.05)	RoB: serious Inconsistency: serious Indirectness: nonserious Inaccuracy: serious Publication bias: nonserious
FACT-B	3 (301)	⊕⊕⊕○MODERATE	-	SMD −0.33 (0.09–0.56)	RoB: serious Inconsistency: nonserious Indirectness: nonserious Inaccuracy: nonserious Publication bias: nonserious

GRADE Working Group grades of evidence: Moderate certainty (⊕⊕⊕○): We have moderate confidence in the estimated effect. The true effect is probably near the estimate, though it could potentially differ to a meaningful degree; Low certainty (⊕⊕○○): We have modest certainty in the effect estimate. The true effect differs considerably from what we have estimated; Very low certainty (⊕○○○): We have minimal certainty in the effect estimate. The true effect substantially differs from the reported estimate.

AT = acupuncture therapy, BFI = Brief Fatigue Inventory, CFS = Cancer Fatigue Scale, CI = confidence interval, CoE = certainty of evidence, CRF = cancer-related fatigue, FACT-B = functional assessment of cancer therapy-breast, GRADE = Grading of Recommendations, Assessment, Development, and Evaluation, HADS-A/D = Hospital Anxiety and Depression Scale-Anxiety/Depression, MD = mean difference, MFI = Multidimensional Fatigue Inventory, PSQI = Pittsburgh Sleep Quality Index, RoB = risk of bias, RR = Risk Ratio, SMD = standardized mean difference, WLC = wait list control.

3.6. AEs

These events were generally self-limiting and did not lead to study withdrawal or treatment discontinuation. For example, as discovered by Brinkhaus et al, the acupuncture group experienced minor hematoma and pain, whereas the chemotherapy group experienced gastrointestinal symptoms. Zhang et al found that the acupuncture group experienced mild skin allergy, pain, and bruising, with similar minor symptoms in the control group. According to Zhang et al, both acupuncture and sham acupuncture groups experienced bruising and auricular skin allergy, respectively.

No studies reported severe or life-threatening AEs directly attributable to acupuncture. Two studies explicitly stated that no serious AEs occurred (LI, 2020; Lu, 2019), whereas 1 study reported no AEs at all (Garland, 2017). However, AEs were not explicitly reported in several studies (e.g., Mao, 2014; Meng, 2024; Wang, 2025; Zhou, 2018).

4. Discussion

We combined findings from 13 RCTs to evaluate how acupuncture affected CRF and QoL in breast cancer survivors. Based on our findings, acupuncture is correlated with significant and clinically relevant improvements in CRF (BFI, MFI, and CFS), sleep quality (PSQI), and emotional state (HADS). Notably, the effect on overall QoL was contingent upon the measurement instrument used, with significant benefits observed only when a breast cancer-specific tool (FACT-B) was used.

4.1. Summary of main findings

Acupuncture significantly decreased fatigue severity across multiple validated instruments. Meta-analyses of studies using the BFI and CFS revealed consistent positive effects. While the overall pooled estimate for the MFI exhibited borderline statistical significance (P = .05), a prespecified subgroup analysis revealed a critical nuance: acupuncture conferred a significant benefit compared with sham acupuncture but not with UC alone. This suggests that nonspecific effects such as patient expectation and therapeutic attention may substantially contribute to the outcomes captured by this instrument, highlighting the complexity of interpreting efficacy in acupuncture trials. Furthermore, the clinical efficacy rate of CRF was significantly improved.

Beyond fatigue, significant improvements were also noted in key secondary outcomes. Acupuncture was associated with enhanced sleep quality, as determined using the PSQI, and alleviated anxiety and depression symptoms, as determined using the HADS, with a particularly robust effect on the anxiety subscale.

The effect on global QoL measures was heterogeneous. A significant positive effect was observed for trials using the FACT-B questionnaire. In contrast, no significant effects were observed when using SF-12 or EORTC QLQ-C30.

4.2. Interpretation of results

The benefits of acupuncture on CRF are probably associated with its proposed physiological mechanisms, which include modulating inflammatory factors (reduction of interleukin-1β, tumor necrosis factor-α, interleukin-6), regulating neurotransmitters (serotonin, dopamine), and restoring autonomic nervous system balance, all of which are pathways often implicated in the pathophysiology of CRF.^[9,20,39] The concurrent improvements in sleep and mood observed in our analysis further support the role of acupuncture in addressing a combination of interrelated symptoms that commonly affect cancer survivors, suggesting a potential holistic modulatory effect.

The discrepancy in QoL outcomes can be critically explained by the construct specificity of the measurement instruments. The FACT-B is a multidimensional tool designed specifically for patients with breast cancer, incorporating domains highly relevant to this population, including concerns about body image, sexual function, and arm swelling (lymphedema).^[18] Cancer treatment profoundly affects these issues; however, they are not comprehensively captured by generic tools such as SF-12. Regardless of its cancer-specificity, EORTC QLQ-C30 is designed for use with various types of malignancies. Therefore, it is probably insensitive in detecting alterations of breast cancer-specific concerns.^[19] Therefore, our study findings suggest that the benefits of acupuncture may be most salient in alleviating the unique symptom burden and functional limitations experienced by breast cancer survivors. This effect is more precisely quantified by a disease-specific instrument.

4.3. Comparison with previous literature

The above findings conform to previous meta-analyses supporting acupuncture application in managing symptoms among patients with cancer.^[21,40] However, our review adds a significant nuance by systematically differentiating among fatigue-specific outcomes, generic QoL, and disease-specific QoL. Furthermore, by including recent trials published up to 2025, we provide an updated and more comprehensive summary of the current evidence base. The moderate CoE of outcomes such as BFI, PSQI, and FACT-B strengthens the rationale for considering acupuncture in clinical practice guidelines to achieve survivorship care for breast cancer survivors.

4.4. Limitations

There are certain limitations. Firstly, our enrolled trials exhibited variable methodological quality. A notable RoB was noted across studies, particularly concerning participant and personnel blinding (performance bias) and incomplete allocation concealment. These limitations may overestimate the treatment effects. Secondly, substantial clinical heterogeneity was noted with respect to acupuncture protocols (e.g., technique, point selection, and treatment frequency and duration), the nature of control interventions (sham, UC, or waitlist), and patient characteristics (e.g., treatment status and time since diagnosis). This heterogeneity probably influences the combined result generalizability. Thirdly, a potential publication bias cannot be ruled out, and some subgroup analyses were limited by an insufficient study number, which may affect the estimate reliability. Finally, the application of different types of QoL instruments introduced an additional layer of heterogeneity, complicating the interpretation of the overall QoL results. Fourth, the assessment of safety was limited by inconsistent reporting of AEs across the included trials. While reported AEs were minor and self-limiting (e.g., bruising, mild pain), and no serious AEs were attributed to acupuncture, the lack of systematic AE reporting in some studies may affect the completeness of our safety profile. Future trials should adhere to standardized AE reporting guidelines.

4.5. Clinical and research implications

Clinically, based on our meta-analysis findings, acupuncture is a complementary therapy with high safety and effectiveness to manage CRF, sleep disturbances, and emotional distress among breast cancer survivors. Its positive effect on breast cancer-specific QoL, as determined by FACT-B, underscores its relevance in addressing the unique challenges faced by this cohort. Therefore, acupuncture should be integrated into multidisciplinary survivorship care plans.^[41]

A synthesis of the treatment regimens reveals that most studies applied acupuncture 2 to 3 times per week over a period of 4 to 8 weeks, with a total of 6 to 20 sessions. Commonly used acupoints included ST36 (Zusanli), LI4 (Hegu), PC6 (Neiguan), CV12 (Zhongwan), and BL23 (Shenshu), often selected based on traditional Chinese medicine principles for fatigue and Qi deficiency. However, due to the lack of a standardized protocol across studies, these parameters should be interpreted as common practices rather than definitive recommendations. This review did not compare acupuncture with surgical interventions or pharmacologic agents (e.g., chemotherapy or endocrine therapy) as direct competitors. Such comparisons were outside the scope of this meta-analysis. Regarding safety, acupuncture is generally contraindicated in patients with bleeding disorders, those receiving anticoagulation therapy, or at sites of infection, lymphedema, or needle phobia. No serious AEs attributable to acupuncture were reported in the included trials, supporting its favorable safety profile when administered by qualified practitioners.

In the future, we recommend larger, methodologically rigorous RCTs. These trials should prioritize adequate sample sizes, rigorous blinding procedures using validated sham acupuncture controls, and detailed reporting of allocation concealment. Furthermore, efforts should be made to develop and evaluate standardized, reproducible acupuncture protocols for CRF in breast cancer to facilitate consistency and clinical implementation, including optimal acupoint selection, treatment frequency, and duration of therapy for breast cancer survivors with CRF. Studies must include longer follow-up periods for assessing sustainability of the beneficial effects out of the immediate treatment period. In addition, future trials should consistently employ a generic QoL instrument (such as SF-12) and a disease-specific tool (FACT-B) to holistically assess the intervention effects. Mechanistic research should be conducted to elucidate the biological pathways of acupuncture in exerting its effects on fatigue, sleep, and mood in cancer survivors. Lastly, cost-effectiveness analyses should be incorporated to inform healthcare policy and reimbursement decisions.

5. Conclusion

In conclusion, this systematic review and meta-analysis suggests that acupuncture is associated with improvements in CRF, sleep quality, emotional distress, and QoL in breast cancer survivors. The magnitude and certainty of these benefits varied by outcome measure, with the most consistent evidence supporting effects on fatigue (BFI, CFS) and breast cancer-specific QoL (FACT-B). It is important to interpret these findings with caution given the methodological limitations of the included studies, the varying degrees of heterogeneity, and the overall low to moderate certainty of the evidence under the GRADE framework. Acupuncture can be considered a potentially safe and effective complementary therapy in multidisciplinary survivorship care. However, to confirm these benefits and establish strong clinical guidelines, more rigorous, high-quality RCTs, adequate blinding, longer follow-up, and consistent use of generic and disease-specific outcome measures are needed. The therapeutic benefits may be most readily detected by outcome measures that are consistent with the specific experiences of this patient cohort. Our results suggest the integration of acupuncture into supportive care plans for breast cancer survivors. Nevertheless, additional high-quality, standardized trials are warranted to solidify these findings and establish evidence-based clinical guidelines.

Author contributions

Conceptualization: Zi Yang, Xinyue Sun, Kuanyu Wang.

Data curation: Zi Yang, Xinyue Sun, Qingquan Dai.

Methodology: Zi Yang.

Supervision: Zi Yang, Xinyue Sun, Kuanyu Wang.

Visualization: Zi Yang, Kuanyu Wang.

Writing – original draft: Zi Yang, Xinyue Sun.

Writing – review & editing: Zi Yang, Xinyue Sun, Qingquan Dai, Gang Wang, Jia Luan, Kuanyu Wang.