A Randomised Phase II Clinical Trial of Acupuncture Plus Standard Care Versus Standard Care Alone for Chemotherapy- Induced Peripheral Neuropathy (CIPN)

Background: Chemotherapy-induced peripheral neuropathy (CIPN) is a dose limiting toxicity that poses a major clinical challenge when managing patients receiving specic chemotherapy regimens (e.g. taxanes and Bortezomib). There is a growing body of literature that suggests acupuncture can lead to improvements in CIPN symptoms. Objective: To evaluate the effectiveness of acupuncture as a management tool for CIPN by comparing use of acupuncture with standard care (Acupuncture) against standard care alone (Control), to reduce symptoms of CIPN. Methods: This two-site clinical trial used a phase II, randomised, parallel group design to investigate the effectiveness of a 10-week course of acupuncture. Patients experiencing CIPN of ≥ Grade II (CTCAE v4.03), recording a ‘Most Troublesome’ CIPN symptom score of ≥ 3 using the "Measure Yourself Medical Outcome Prole" (MYMOP 2), were randomised (1:1) to ‘Acupuncture’ or ‘Control’ (note: Patients in Control were offered acupuncture off trial at the end of their study period). Primary endpoint was a ≥ 2-point improvement (success) in MYMOP2 score at week 10. 100 participants (120 randomised to allow for attrition) were required (90% power; 10% one-sided alpha) for a hypothesised improvement in success proportions from 30% to 55% using a primary analysis model with logistic regression adjusted for stratication factors and baseline MYMOP2 scores. Results: Primary MYMOP2 outcome data at week 10 was available for 108 of 120 randomised participants. There were 36/53 (68%) successes in ‘Acupuncture’ compared to 18/55 (33%) in ‘Control’. The adjusted success odds ratio was 4.3 (95% CI 1.9 to 9.6; 2p < 0.001; Acupuncture vs Control). Benecial effects were also seen in the secondary outcome data, including clinicians’ grading of neuropathy, EORTC QLQ-CIPN20, QLQ-C30 summary scores and patient reported pain scores. Conclusion: A 10-week course of resulted in measureable improvement in participants symptoms of CIPN. a phase was to investigate the ecacy of a 10-week course of acupuncture in the management of patients’ most troublesome symptoms relating to CIPN measured by a self-reported outcome scale. Secondary aims included use of clinicians’ functional assessments of neuropathy (graded in accordance with CTCAEv4.03) to evaluate effectiveness of acupuncture, in addition to patient reported pain related scores and quality of life assessments. The study was set over two sites, which form part of a large tertiary cancer centre in North West of England.

Given that standard medication for CIPN is not consistently effective [[v], [vi]], it is necessary to investigate interventions that would allow patients to continue on chemotherapy regimens at appropriate doses as well as alleviate CIPN symptoms once the course of chemotherapy has concluded. Acupuncture is a non-pharmacological treatment, which has attracted attention as a clinical management tool for complex symptoms such as neuropathy. There is an increasing eld of The exact mechanism of acupuncture analgesia remains unclear with multiple studies offering possible explanations.
Experimental data from functional magnetic resonance imaging (fMRI) studies have con rmed that areas of the limbic system, such as the nucleus raphe magnus, locus ceruleus and periaqueductal gray, become deactivated during acupuncture stimulation whilst regions within sensorimotor structures become activated [[xiv], [xv]]. This may be why acupuncture can reduce perception of pain and increase sensation -two problematic manifestations of peripheral neuropathy. Further research has pointed to the involvement of a variety of endogenous ligands as possible biochemical mediators in the acupuncture response including opioids (e.g. β-endorphin), neurotransmitters (e.g. GABA, glutamate), neuropeptides (e.g. oxytocin) and neurotrophins (e.g. nerve growth factor) [[xvi], [xvii]], suggesting acupuncture may cause the release of biochemicals that work directly on the nervous system. The ndings from these research studies provide a possible theoretical framework to explain how and why acupuncture might work in the management of CIPN symptoms.
A service evaluation at the study site (n=18), utilising an established acupuncture protocol reported improvements in CIPN symptoms for 14 participants whose symptoms were refractory to standard management practice [[xviii]]. Additional bene ts included reduction in use of analgesia and improved sleep. The current trial was designed to build on the clinical experience of an acupuncture service that has been continuous in its provision for CIPN since 2009. The primary research question being: "Does the addition of acupuncture to standard treatment reduce the level of CIPN experienced by patients during or following treatment with neurotoxic chemotherapy?"

Methods
The primary aim of the study was to investigate the e cacy of a 10-week course of acupuncture in the management of patients' most troublesome symptoms relating to CIPN measured by a self-reported outcome scale. Secondary aims included use of clinicians' functional assessments of neuropathy (graded in accordance with CTCAEv4.03) to evaluate effectiveness of acupuncture, in addition to patient reported pain related scores and quality of life assessments.
The study was set over two sites, which form part of a large tertiary cancer centre in North West of England.

Participants
Patients with breast cancer and multiple myeloma were initially eligible for randomisation if their neuropathy was of grade II or above (CTCAE v4.03). To increase recruitment the protocol was amended to widen patient eligibility to include gastrointestinal and gynaecological cancer diagnoses. The rationale for using patients from these four cancer disease groups was to allow for comparison of effect across pathologies/different drug regimens, without making the trial design too complex. Additional key inclusion criteria included a platelet count of ≥ 30 x 10 9 /L, a neutrophil count of ≥ 0.5 x 10 9 /L and a MYMOP2 score of their most troubling CIPN symptom of ≥ 3. The blood counts stipulated for entry to the study are lower than would be recommended in standard clinical practice. Study processes monitored the safety of these amended parameters, although this was not a documented aim of the trial.

Design
This study (Trial No. NCT02275403) was designed as a parallel group, open label, randomised phase II clinical trial. Acupuncture was added to standard care for CIPN, in patients randomised to the intervention (Acupuncture) arm. Patients received medication to manage symptoms of CIPN in accordance with local trust policy in addition to 10 weekly sessions of acupuncture from a trained acupuncturist. Patients randomised to the control Arm (Control) received only medication to manage symptoms of CIPN in accordance with local trust policy.
Eligible patients were randomised in a 1:1 ratio to either Acupuncture or Control. A strati ed permuted block (size 6-10) allocation scheme was implemented by the study statistician using a bespoke computer system which had username/password control and a full audit trail. Researchers telephoned a central number whereupon trained CTU staff recorded the participant's details in the system and the allocated trial ID and arm were only revealed after commitment of these details ensuring "allocation concealment" from both researchers and CTU staff. The system also sent an automatic con rmatory e-mail to site. The following two factors were controlled for in the algorithm: 1. Cancer diagnosis (Breast Cancer vs. multiple myeloma vs. gastrointestinal cancer vs. gynaecological cancer)

Treatment intention (one of the below):
Patient was currently on chemotherapy or any other cancer treatment (to end during the 10-week study or beyond) Patient was on chemotherapy, but this had been suspended due to neuropathy or continued but the neurotoxic agent omitted (but could restart in the 10-week study period) Patient was due to start their next line of chemotherapy during the 10-week study period Patient had nished treatment for their cancer and no further treatment is planned to start within the 10-week study period Recruitment for the study began in April 2015 and ended in November 2018. Participants remained on the study for 10 weeks, with those allocated to Acupuncture attending the hospital for acupuncture every week. In addition, they were followed up one week after the nal acupuncture session to document any adverse events. Patients in Control were only required to attend for the compulsory assessment visits at baseline, 6 and 10 weeks. Whilst not a documented aim of the study, participants were assessed at week 6 and 10 to provide some explorative data on whether a 10-week course of acupuncture is required to provide a signi cant level of bene t or whether 6 weeks might be su cient (Appendix 2).

Intervention -Acupuncture
The needles used in this study were for single use, size 32 gauge (0.25mm) and 15-30mm in length. The points were punctured perpendicularly to a depth of approximately 1.0cm (0.5 inch), dependent on the patients' size, sensitivity, state of health and the particular point. There was exibility in cases of lymphoedema for example, for the selection of alternative points to maintain an equal dose of treatment. Acupuncturists had exibility to add and record points for related symptoms, re ecting naturalistic current acupuncture practice within this NHS setting (see Table I).
Conversation between therapist and patient was limited to facilitation of treatment only. Although appropriate care was shown to the patient, any conversation about the effects of acupuncture was kept to the minimum and in response to questions the patients raised.
Participants randomised to Control were able to access the site's existing acupuncture service provision after the 10 week study period was over.

Outcome Measures
All outcome data was collected at baseline, 6 and 10 weeks. The primary outcome measure chosen for the study was the Patient Reported Outcome Measure (PROM): "Measure Yourself Medical Outcome Pro le" (MYMOP2), see appendix 1 [[vii]]. With this validated scale, patients identi ed and graded their own worst symptom, thus data capture was exible enough to address the complexity of CIPN. MYMOP2 is scored on an integer 0-6 scale with higher scores denoting greater issues. The eligibility criteria for the study required a 'Most Troublesome' CIPN symptom score of ≥ 3 at baseline, with the primary end-point being a subjective ≥2 point improvement (success). The necessary sample size required to appropriately power the study for a hypothesised improvement in success proportions from 30% to 55%, was calculated at 100 patients with a plan to randomise 120 to allow for attrition (90% power; relaxed 10% one-sided alpha).
MYMOP2 data was triangulated with assessment of functionality, which was completed by a clinician at the set time points. The clinicians used standardised questions and visual assessments of function, in accordance with CTCAE v4.03.
The key secondary endpoint was functional improvement set at reduction of CIPN to grade ≤ I (CTCAE v4.03).
Additionally, a broader measurement of change in symptom burden and quality of life was assessed through completion of the EORTC QLQ-C30 and associated CIPN20 module.
Patients were also asked to complete weekly diaries throughout the study period to capture daily pain scores and CIPN medication usage. These supplementary measures allowed for evaluation of the following secondary endpoints: Change in quality of life and symptom burden (EORTC QLQ-C30 + CIPN20 module) Alteration in pain related scores

Data Analyses
Both of the randomisation strati cation factors were collapsed to binary factors in the analysis as some levels had low frequencies. Approximately 50% of patients in both arms had a diagnosis of breast cancer, (see Table 1 below) therefore a decision was made to stratify diagnosis as Breast cancer (yes/no). Equally approximately 40% of participants had completed treatment on study entry therefore adjustment was made for treatment complete status (no/yes) (see Table 1 below). All analyses were carried out on an intention to treat basis i.e. as randomised.
Primary Outcome (Week 10) The analysis of the primary outcome was undertaken with a logistic regression, the focus being on the trial arm effect after adjustment for diagnosis, treatment status and the baseline MYMOP2 score. In addition, a worst-best case sensitivity analysis was conducted in which any missing outcomes in 'Acupuncture' were imputed as "failures" and those in 'Control' as "successes".

Secondary outcomes (week 10)
Physician assessed CTCAE CIPN grade ≤ 1: analogous analyses to those for the primary outcome were conducted for this binary outcome variable i.e. logistic regression with adjustment for diagnosis, treatment status and baseline grade. Again a worst-best case sensitivity analysis was conducted in which missing outcomes in 'Acupuncture' were imputed as "failures" and those in 'Control' as "successes".
EORTC QLQ-CIPN20: During the data collection period of the study, we sought and received advice from the EORTC that previously proposed subscales for this PROM (sensory, motor and autonomic), had been found unreliable and their current recommendation is to use an overall score based on the rst 18 of the 20 items in the questionnaire. This maps to a 0-100 scale with lower scores being better. An analysis of covariance (ANCOVA) model was used with adjustment for baseline score, diagnosis and treatment status.
EORTC QLQ-C30 Summary Score: The Summary Score was calculated from the mean of 13 of the 15 QLQ-C30 scales (the Global Quality of Life scale and the Financial Impact scale were not included). Prior to calculating the mean, symptom scales are reversed to obtain a uniform direction of all scales. The summary score was only calculated if all of the required in that scale have been completed). This results in a 0-100 scale in which higher scores represents better overall quality of life. An analysis of covariance (ANCOVA) model was used with adjustment for baseline score, diagnosis and treatment status.
Patient reported pain scores: Participants were asked to complete diaries in which they recorded their worst pain daily on an integer 0-10 scale with landmarks (0= "no pain at all", 10= "the most intense pain I can imagine"). Within patient weekly mean scores were calculated and the week 1 means were taken as baseline as there was no true baseline i.e. no prerandomisation diary. An analysis of covariance (ANCOVA) model was used with adjustment for baseline score, diagnosis and treatment status.
Longitudinal models: Outcome data were also recorded at week 6 and longitudinal models with both week 6 and week 10 outcomes were tted in further exploratory analyses. These models again adjusted for baseline value, diagnosis and treatment status. In addition, they included the week effect and the full interaction of week with each of the other model terms. With this approach we were able to assess differences in the trial arm effect between weeks 6 and 10.
Descriptors used for 'Most Troublesome Symptom' on baseline MYMOP2: As anticipated, symptom burden and impact of CIPN was expressed in a variety of different ways. The descriptors used by participants are laid out in Table II below: Primary Outcome (Week 10) Data was available for 108 of 120 randomised participants with 36/53 (68%) successes in Acupuncture compared to 18/55 (33%) in Control. The adjusted success odds ratio from the logistic regression model was 4.3 (95% CI 1.9 to 9.6; 2p < 0.001; A vs C) after adjustment. The worst-best case sensitivity analysis, using the pessimistic assumptions described above, showed 36/61 (59%) successes in Acupuncture compared to 22/59 (37%) in Control and the adjusted success odds ratio was 2.4 (95% CI 1.1 to 5.0; 2p = 0.02; A vs C). Therefore, the substantive nding of bene t of the intervention for the primary outcome is unaffected by missing data.

Secondary outcomes (week 10)
Physician assessed CTCAE CIPN grade ≤ 1: Data was available for 109 participants with 27/53 (51%) successes in Acupuncture compared to 4/56 (7%) in Control. The adjusted success odds ratio from a logistic regression model was 13.1 (95% CI 4.1 to 42; 2p < 0.001; A vs B). Using the worst-best case sensitivity analysis, the 7 missing outcomes in Acupuncture were imputed as "failures" and the 3 in Control as "successes". With these pessimistic assumptions there were 27/61 (44%) successes in Acupuncture compared to 7/59 (12%) in Control and the adjusted success odds ratio was 5.9 (95% CI 2.3 to 15; 2p < 0.001; A vs C). The substantive nding of bene t of the intervention for this secondary outcome is therefore unaffected by missing data.
EORTC QLQ-CIPN20: Data was available for 107 participants; (A =53 vs B =54). In an analysis of covariance (ANCOVA) model the adjusted mean difference in week 10 outcome values using overall score from the rst 18 questions was estimated to be -11.7 (95% CI -17.3 to -6.1; 2p < 0.001; A vs C).
EORTC QLQ-C30 Summary Score: Data was available for 103 participants and in an analysis of covariance (ANCOVA) model the adjusted mean difference in week 10 outcome values was estimated to be 9.5 (95% CI 5.0 to 14.0; 2p < 0.001; A vs C).
Patient reported pain scores: Data was available for 97 participants and the adjusted mean difference in week 10 outcome values was estimated to be -1.45 (95% CI -2.25 to -0.65; 2p = 0.001; A vs C).
Longitudinal models: Week 6 trial arm effects were also statistically signi cant but tended to be somewhat smaller in magnitude than the week 10 effects though not statistically signi cantly so.

Safety Data
Blood parameters for eligibility into the study were reduced compared to standard clinical practice for both platelets and neutrophils. However, no serious adverse events and only 16 acupuncture related adverse events (11 pts), were documented (tingling, ache/pain, bruising & 'spotting' of blood). Of these 16, none required action and 12 (75%) were reported as mild.

Discussion
In recent years, acupuncture has been identi ed as a possible management tool for CIPN [[i], [ii]]. Whilst data so far would suggest acupuncture may be bene cial, because of the complexity of symptom load it is a challenge to validate e cacy in all presentations (e.g. numbness and pain). In order to provide preliminary data encompassing all presentation of symptoms, the primary outcome data from this study was from the patient reported measure; MYMOP2. Patients identi ed their own symptoms, described each symptom in a way that was meaningful to them and scored the symptom. 'Success' was set as a ≥ 2 point improvement in symptom score, with this being accepted as a clinically signi cant reduction in symptom burden. The results showed a highly signi cant level of bene t, with double the number of patients in Acupuncture showing a 'success' at week 10 compared to Control. Secondary outcome data in the form of clinician assessment was collected to triangulate with the patient reported primary outcome measure. As with the primary outcome data, the results showed a signi cantly higher level of bene t for participants in the Acupuncture arm than in the Control.
Consequently, the results from the clinician rated outcome, go some way to validating the clinical signi cance of a 2 point reduction in MYMOP2 scores. Additional secondary outcome data from the EORTC-QLQ-C30 and associated CIPN-20 was collected in order to further triangulate the data, as it is recognised that the impact of CIPN is notoriously di cult to capture in one measure [1]. These measures also indicated a high level of bene t from adding acupuncture into routine care for patients with CIPN, thus strongly suggesting that acupuncture improved quality of life and reduced the level of reported pain in this patient cohort.
In order to clarify the length of treatment required to ensure bene t from acupuncture, longitudinal models were used to compare effects of the acupuncture at week six and ten. Whilst signi cant bene t was seen at both time points, the magnitude was less at week six. It will therefore be appropriate to undertake further exploratory analyses to better understand whether there are certain sub-sets of patients, who will gain the same level of bene t with a shorter course of acupuncture, such as those on certain chemotherapy protocols, or with a speci c symptom burden, for example: numbness vs pain.

Limitations
The patient reported primary outcome measure was key to the study design as it ensured that the data captured by the study was patient speci c for all participants However, MYMOP2 has a limitation as an outcome measure as it is subjective rather than objective. Reported bene t to participants was part-validated within the study by the clinician's grading of motor skills but, due to funding limitations, no completely objective outcome measure was included within the study design. For any future study, it is recommended that data be triangulated with an objective outcome such as nerve conduction studies.
A related issue concerns the decision not to incorporate sham acupuncture as a control within the study design. This decision was based an ethical rationale. A proportion of the study participants were still undergoing chemotherapy for their underlying cancer diagnosis. It was therefore decided that it would be unethical to provide sham acupuncture to patients who were potentially at higher risk of bleeding and infection due to low blood counts only to provide a higher degree of control to the study design. It is recommended that this decision would require review in any future study design.

Clinical Implications and Future Work
Data from this study can be used to inform the treatment options available to patients with debilitating and painful symptoms of CIPN. The results clearly suggest clinical bene ts from acupuncture and will add to the growing evidence base of this non-pharmacological treatment as a clinical management tool. The hope is that this study will help increase patients' and clinicians' con dence to actively consider acupuncture as a means of reducing the symptom burden associated with CIPN. The study goal was to maintain physical functioning / mobility, reduce symptoms relating to sensation (including pain and dexterity), enhance wellbeing and when appropriate, facilitate ongoing treatment with drugs designed to improve mortality from cancer.
The data from this study has provided evidence, which healthcare professionals can use to inform the advice offered in primary, secondary and tertiary care to this cohort of patients regarding the possible use of acupuncture. However, the design of the study was such that only short-term bene t (10 weeks) was evidenced. For assessment of longer-term value, a de ned follow-up period would be required.
To address some of the issues highlighted by this study, it is proposed that a larger multicentre study be undertaken. It is suggested that the design incorporates maintenance protocols to monitor continuation of effect and identi cation of factors, which would provide clarity regarding whether patients require a 6 or a 10 week protocol. A full cost-bene t analysis is also proposed to strengthen the case for the adoption of acupuncture as a patient care pathway within the NHS.

Conclusion
In summary, the immediate impact of the study was the corroboration of results from a previous service evaluation con rming that this cohort of patients bene ted from the acupuncture they received. However, sustainability of any improvements in symptoms requires further investigation.
Long-term, this study is the initial phase of a larger programme of work in which the data collected will be used to inform the design of a large multicentre trial. Written consent was obtained from all participants.

Consent for publication:
Not applicable Availability of data and materials: The datasets used and analysed during the current study are available from the corresponding author on reasonable request Competing interests: