Article Text


Acupuncture for joint symptoms related to aromatase inhibitor therapy in postmenopausal women with early-stage breast cancer: a narrative review
  1. Elizabeth J Halsey,
  2. Mei Xing,
  3. Rachel C Stockley
  1. Department of Health Professions, Manchester Metropolitan University, Manchester, UK
  1. Correspondence to Elizabeth J Halsey, Department of Health Professions, Manchester Metropolitan University, Birley Building, Manchester Campus, Bonsall Street, Manchester M15 6GX, UK; Elizabeth.j.halsey{at}


Background Aromatase inhibitor-induced musculoskeletal syndrome (AIMSS) leads to discontinuation of aromatase inhibitor therapy in a significant proportion of patients with breast cancer. Acupuncture is popular among cancer patients and has previously been shown to improve symptoms in a range of musculoskeletal complaints.

Aim To determine the effectiveness and safety of acupuncture for the management of AIMSS in postmenopausal women with early-stage breast cancer.

Methods A literature search was carried out for randomised controlled trials (RCTs) on acupuncture for AIMSS in postmenopausal women with early-stage breast cancer. Characteristics of trials and outcomes were extracted from the retrieved articles, which were also assessed for risk of bias and quality of reporting.

Results Four RCTs were retrieved of sample size 32–67 (totalling 190 participants). Compliance with treatment was high and rates of adverse events were low. Of the three two-arm RCTs, two found no difference between acupuncture and sham acupuncture and one found that acupuncture was statistically superior to sham acupuncture. The fourth RCT, which incorporated three arms, found acupuncture and sham acupuncture to be statistically superior to usual care but there was no difference between true and sham acupuncture. Three trials that used non-penetrating sham as the control found no effect of acupuncture over sham, but the one trial that used superficial needle insertion found acupuncture to be superior.

Conclusions Acupuncture is safe and results in improvement in AIMSS symptoms, but similar benefits are also elicited by non-penetrating sham acupuncture. Future research should seek to establish the durability of improvements.


Statistics from


Breast cancer is one of the most commonly occurring malignancies, with 42 489 new cases registered in England1 and approximately 1.7 million worldwide2 in 2012. Treatment of breast cancers may include surgery, radiotherapy, chemotherapy and hormone therapy. Around 70% of breast cancers are oestrogen receptor-positive,3 and the most effective hormone treatments in postmenopausal women are aromatase inhibitors (AIs).4–6

AIs are associated with a number of side effects including hot flushes, sleep disturbance, nausea, decreased libido, vaginal dryness and fatigue. One of the most clinically significant adverse effects of AI therapy is AI-induced musculoskeletal syndrome (AIMSS). AIMSS is characterised by pain, stiffness and decreased function of joints, and most commonly affects the hands and knees. AIMSS occurs in an estimated 44–54% of women receiving AIs, with around two-thirds of patients reporting moderate to severe symptoms.7–9 A survey of UK specialist oncology clinicians found that 83% considered the effect of AIMSS on their patients’ lives to be large or very large.10 Premature AI cessation is common, occurring at rates of 21–38%,11 ,12 with a substantial proportion of patients reporting musculoskeletal symptoms as their main reason for discontinuation.12

The pathology underlying AIMSS is not well understood, but oestrogen deprivation is suspected to play a part.10 ,13 Imaging studies indicate possible tenosynovial changes,14 joint effusions and carpal tunnel syndrome.15

The current treatment of AIMSS is based primarily on non-steroidal anti-inflammatory drugs, but there are few data in support of their effectiveness in this condition.10 ,16 ,17 Small single-arm trials have found short-term benefits from prednisolone,18 yoga19 and tai chi.20 One trial found 2 months of vitamin D supplements initially improved AIMSS symptoms, but not significantly more than placebo beyond 2 months.21 The rate of discontinuation remains high after switching to an alternative AI.22

A substantial body of research supports the use of acupuncture for musculoskeletal complaints.23 Acupuncture is a popular treatment option among patients with breast cancer,24 and there is evidence of varying levels that it can reduce a number of side effects of cancer treatments including fatigue, nausea, peripheral neuropathy and vasomotor symptoms.25 The aim of this study was to evaluate the evidence for the safety and efficacy of acupuncture as a treatment for AIMSS.


We conducted a clinically focused narrative review of randomised controlled trials (RCTs) of acupuncture as treatment for symptoms of AIMSS.

Literature searches

Free-text searches were carried out of AMED, CINAHL, Cochrane CENTRAL and Medline (PubMed) in January 2015. Three concepts were combined using the Boolean operators ‘AND’ and ‘OR’ with appropriate truncations: (1) acupuncture, electroacupuncture (EA), dry needling, Chinese medicine, complementary medicine; (2) AI, anastrozole, Arimidex, letrozole, Femara, exemestane, Aromasin, hormone therapy, breast; (3) arthralgia, joint pain, musculoskeletal, side effect, adverse event. A ‘Research Article’ filter was applied to CINAHL. Reference lists of retrieved papers were examined for additional studies. Abstracts retrieved from each database were compiled into one list before being screened and excluded if they were not in English, if they were a duplication, if they did not concern acupuncture or AIMSS, if they were not a trial or not a RCT.

Study selection

Retrieved trials were included if they fulfilled the following criteria: participants were women with early (stages I–III) breast cancer who were being treated with an AI and experiencing joint pain as a side effect; the intervention being tested was a course of acupuncture treatment, inclusive of body acupuncture, EA and auricular acupuncture (AA); a comparison group was also measured; outcomes were related to joint symptoms. Included trials were not limited to particular settings.

Data extraction

The following data were extracted from each trial and entered onto pre-prepared spreadsheets: sample size, characteristics of acupuncture, number and frequency of sessions, type of comparator, compliance, rate and nature of adverse events, outcome and outcome measurements at each time point.

Quality appraisal

A tool developed by the Cochrane Back Review Group26 was used to assess the risk of bias in each trial. This tool allows the systematic appraisal of selection bias (randomisation, allocation concealment and baseline similarity), performance bias (co-interventions, compliance and blinding of participants, personnel and outcome assessors), attrition bias, reporting bias and detection bias. Articles were also assessed for the repeatability of their intervention and controls using the Standards for Reporting Interventions in Clinical Trials of Acupuncture (STRICTA) guidelines.27

Data analysis

Standardised effects were calculated for selected data to measure sizes of changes in treatment groups compared to control groups. Standardised effects were determined by dividing the difference in outcome value between two experimental groups by the mean baseline value of the two groups. These figures were transformed into percentages by multiplying them by 100. Each standardised effect was given a positive sign if symptoms in the treatment group were less severe than in the control group, and a negative sign if the reverse was true.


Results of the literature search

Of 150 retrieved trials, four were judged to fulfil all inclusion criteria and were included in the review (figure 1). Two feasibility studies (both by authors of included trials) were excluded for not including a comparison group. Among the excluded articles was a protocol for an included trial. No trial was excluded because it was not written in English as the abstracts provided sufficient information to assess eligibility.

Figure 1

Selection of trials for inclusion in the review. AIMSS, aromatase inhibitor-induced musculoskeletal syndrome.

Description of studies

Table 1 presents the characteristics of the included trials. Three studies took place in the USA28–30 and one in Australia.31 In the two studies that reported settings,29 ,30 both took place in tertiary care cancer centres. Three studies29–31 measured outcomes at 12-week follow-up, between 4 and 6 weeks after cessation of treatment. Primary outcome measures were patient-reported outcome (PRO) questionnaires in all trials, with two trials also using objective measurements.30 ,31 Two studies performed blood assays.29 ,31

Table 1

Characteristics of included trials


The proportion of randomised participants who completed all of the acupuncture or control interventions ranged from 88% to 94%. Of the three studies that recorded follow-up data beyond the intervention period, only one30 reported the number evaluated at this time point, which was 88% of those randomised.

Adverse events

All four studies specifically reported that significant adverse events did not occur. Three studies reported minor side effects such as pain and minor bruising at rates of 7–16% of patients per study.28 ,30 ,31 Sham non-penetrating EA resulted in a similar rate of pain at the site of needling as true EA in one study,30 which also reported that participants’ arms had been needled on the same side as breast cancer surgery without causing infection or aggravating lymphoedema.30

Quality appraisal

Risk of bias

The outcomes of risk of bias appraisal are presented in table 2. Risk of bias from most sources was judged to be low. All studies attempted participant blinding and measured whether this had been successful by asking participants to guess the group to which they belonged. Table 3 gives the percentage of each experimental group who correctly guessed which treatment they had received. Participants were more likely to correctly guess their allocation if they had received acupuncture than if they had received sham acupuncture. Patient blinding was judged to be incomplete in three RCTs, with an unclear risk of bias in one trial.29 Two RCTs reported statistically significant differences between control and experimental groups at baseline.28 ,29 One RCT may have been subject to performance bias as both body acupuncture and AA were provided to the acupuncture group, but only sham body acupuncture was given to the sham group.28 All PRO questionnaires have been previously validated for other conditions and, while no outcome measure has been developed specifically for AIMSS, three of the PROs used have been found to correlate with changes in AIMSS symptoms.9

Table 2

Judgements of risk of bias in RCTs

Table 3

Participants’ correct guesses of experimental group

Reporting of intervention

All acupuncture and sham needling protocols were well-described and repeatable, with two articles explicitly citing STRICTA guidelines.28 ,31 The amount of information concerning non-needling aspects of interventions was variable. Two studies30 ,31 described the inclusion of Traditional Chinese Medicine (TCM) diagnostic procedures within their experimental protocols. One study29 used a script to guide interaction between practitioner and patient and one28 did not give any details other than needling protocols.

Description of data

Statistical significance

Table 4 presents the results of significance testing carried out by the authors of the trials for differences between groups in outcome measures at the end of interventions and follow-up periods. Three trials found no difference between true and sham acupuncture in any outcome measure at any time point29–31 and one found acupuncture to be statistically superior to sham for all outcomes except three subsections of the Functional Assessment of Cancer Therapy-General questionnaire.28 The three-arm trial found that both acupuncture and sham acupuncture were superior to usual care in all outcomes except the Physical Performance Test.30

Table 4

Significance testing of outcomes between groups

Bao et al29 also conducted statistical tests comparing serum concentrations of oestrogen, cytokines and β-endorphin within groups between baseline and end of the intervention. There were no significant changes other than a significant reduction in interleukin-17 in both the true and sham groups and a significant increase in interleukin-12 in the sham group.

Size of effects

Table 5 gives standardised effects for data from Mao et al30 to allow estimation of the size of effect of true electroacupuncture (TEA) and sham electroacupuncture (SEA) compared with waiting list control (WLC). They reported that 57.1% of the TEA group rated themselves ‘much improved’ or ‘very much improved’ on the Patients’ Global Impression of Change scale compared with 4.6% in the WLC group. This outcome was not reported for the SEA group.

Table 5

Standardised effects in outcome measures from Mao et al30

Sham acupuncture

The one trial28 that used superficial insertion of normal needles as a comparator was also the only trial of the four in which real acupuncture was found to be statistically superior to sham acupuncture. No difference was found between real and sham acupuncture in the three trials employing non-penetrating sham acupuncture.


The statistical superiorities of TEA and SEA over WLC found by Mao et al30 in the Brief Pain Inventory-Short Form were maintained at 12-week follow-up. Differences between the true and sham acupuncture groups were not statistically significant in any of the three studies measuring follow-ups. However, trends in the study by Mao et al30 suggest that the benefits from sham acupuncture were less persistent. The selected data presented by Oh et al31 followed similar trends. Bao et al29 found true and sham acupuncture groups followed similar trends to each other at follow-up: symptoms remained static according to the Health Assessment Questionnaire Disability Index questionnaire but Visual Analogue Scores indicated that pain decreased.


This review identified four RCTs addressing the use of acupuncture for AIMSS. One report found that a course of acupuncture resulted in improvements in patient-reported joint pain, stiffness and function that were significantly better than usual care.30 The review found that there were no differences between true and sham acupuncture on any outcome in three studies,29–31 indicating that non-needling elements of interventions may be responsible for therapeutic effects. However, one study conversely found that acupuncture was statistically superior to sham acupuncture,28 so it is unclear whether needling itself affects AIMSS symptoms.

Clinical relevance

On the basis that three of the four studies found acupuncture did not have any effect beyond placebo, public funds are unlikely to be spent on acupuncture for AIMSS. However, side effects from AIs remain a clinically important problem: the arthralgia can be sufficiently debilitating to cause patients to stop medication that could be life-saving, and conventional treatment is ineffective.17 This review has found that acupuncture was significantly better than usual care, with a large proportion of participants reporting substantial improvements in their symptoms. Patients suffering joint symptoms from AI therapy may choose to pay for treatment privately, and ought to be offered the information that acupuncture may help relieve their symptoms although its effects could be due to placebo responses. Charitable organisations offering complementary therapies may also be more concerned with the overall benefits patients experience from acupuncture than delineating the effectiveness of its individual components.

Recognising that acupuncture is a complex intervention and that elements in addition to needling may have effects on patients could help acupuncture practitioners to optimise the effectiveness of their treatments. While the tactile stimulus provided by sham acupuncture could contribute to its effects, it is also possible that elements completely outside needling/sham procedures confer therapeutic benefits.32 Indeed, others have shown that the patient–practitioner relationship can affect how patients respond to interventions to treat other conditions such as irritable bowel syndrome.33 TCM diagnostic procedures, which typically involve detailed questioning, palpation of the pulse and examination of the tongue, were included in the sham intervention in at least two studies and could have had a bearing on how some patients responded to their treatments. The recognition of the importance of the patient– practitioner relationship should encourage acupuncturists to continue to identify a broad scope of possibilities for their continuing professional development, including the development of their communication, counselling and interpersonal skills.

Collectively, the four studies tested a range of acupuncture modalities (body acupuncture, AA and EA) and frequency of treatment varied between once and twice weekly. These variables did not appear to influence the overall outcome of the studies. This finding agrees with the results of a recent meta-analysis of acupuncture for chronic pain conditions,34 which concluded that the use of electrical stimulation and frequency of sessions did not influence the likelihood of finding true acupuncture superior to sham acupuncture. One characteristic that was shared by the two trials with positive results was the use of point prescriptions that were at least partly determined by the patients’ symptoms,28 ,30 in contrast to the standardised protocols employed by the others.

Future research recommendations

A key limitation of the trials is their relatively short duration of a maximum of 12 weeks. This is far short of the 5 years recommended for AI therapy in early breast cancer.35 A primary aim for future research in acupuncture for AIMSS should therefore be to establish the persistence of symptomatic relief following treatment and to assess whether ‘top-up’ sessions at longer intervals could maintain symptom control at a level patients find manageable. Trials could also record rates of discontinuation of AIs and assess whether receiving acupuncture leads to decreased cessation. Longer trials could also examine whether the effects of real acupuncture are more durable, as indicated by non-significant trends in two of the three studies incorporating follow-ups.30 ,31

Patients can be trained to perform acupuncture on their own bodies, and this has been found to be beneficial in pain management36 and cancer-related vasomotor symptoms.37 Future research could explore whether self-acupuncture may provide a means of cost-effective symptom management for AIMSS.

If sham acupuncture is to be used in future trials of acupuncture for AIMSS, it should be noted that a sham consisting of shallow insertion of normal needles appeared less active than when non-penetrating needles were used. This finding contradicts the results of a meta-analysis of acupuncture for chronic pain.38 To address the ongoing uncertainty concerning which elements of acupuncture are therapeutically active, future studies should seek to systematically record interventions beyond needling protocols, including information such as settings, diagnostic procedures and the type of interaction between patient and practitioner.

Little is known about how patients regard evidence for interventions. Future research could explore whether patients with AIMSS would choose to use acupuncture if they were aware it has not been proven to be better than placebo. The fact that patients know that their treatment contains placebo elements may not necessarily inhibit its effectiveness, given that a previous study has found that placebo treatment for irritable bowel syndrome can be successful without any patient blinding at all.39


As with all reviews, the trials it includes are likely to be biased. Acupuncturists were not blinded and blinding of participants was incomplete, and this lack of masking may have led to exaggerated responses in treatment groups. One of the trials finding positive results may also have been subject to performance bias.28

The review itself may be subject to publication bias in favour of positive results as searches of grey literature were not conducted, study authors were not contacted and non-English journals were not searched. The retrieved studies are limited in number and are of small sample size, so this review can be said to provide only tentative evidence. The wide range of outcome tools used by the studies made comparison of results difficult and pooling of data was not attempted, limiting the precision and impartiality of the review.


AIMSS is a clinically important condition, the conventional treatment of which is unsatisfactory and can lead to early cessation of an important drug therapy as well as reduce quality of life for cancer patients. The data collected in the review provide promising evidence that a course of acupuncture treatment may improve joint pain, stiffness and loss of function secondary to AI therapy in women with breast cancer. However, non-needling components of the acupuncture intervention are likely to confer at least some of the benefits, and the evidence is equivocal as to whether needling components also have effects. Primary aims of future research should be to determine the durability of symptomatic relief gained from acupuncture and to assess whether patients would choose to undertake such a treatment in the knowledge that its effects have not been proved beyond placebo.

Summary points

  • Acupuncture is used for the musculoskeletal syndrome caused by aromatase inhibitors (aromatase inhibitor-induced musculoskeletal syndrome, AIMSS).

  • We reviewed four trials comparing acupuncture with sham and/or usual care.

  • Acupuncture was superior to sham in one trial and no different from sham in three.

  • Acupuncture was superior to no acupuncture.


We would like to thank Kay Hurst, Jill Bailey, Mark Burgess and Dr Mike Cummings for their support and advice.


View Abstract


  • Contributors All authors contributed to the manuscript.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.