Methods

This chapter reports the methods used to conduct the GaPP2 trial.

Trial design

The GaPP2 trial was a placebo-controlled, randomised, blinded, multicentre trial of gabapentin for the management of CPP in women with no known aetiology with a nested-mechanistic fMRI brain study (see Chapter 4). The trial had a favourable ethics opinion from Research Ethics Committee West Midlands – Coventry and Warwickshire (Multicentre Research Ethics Committee reference 15/MW/0036).

Recruitment

The GaPP2 trial participants were recruited from gynaecology outpatient departments in 39 participating NHS sites across the UK. The GaPP2 trial recruitment followed a two-step process (Table 1). Potential participants were referred, with their permission, to the local research teams by their attending clinician. All participants were told that participation in the trial was completely voluntary and that they could withdraw at any stage in the trial. This was part of the consent process. Participants were reassured that participation or withdrawal would not affect their normal clinical care. All women were approached, with permission, by researchers who were trained in Good Clinical Practice and specifically in taking consent for this trial. Potential participants were provided with a participant information sheet and given time to consider their involvement. If participants expressed an interest, written informed consent was sought and participants were invited to a screening visit at which they were assessed for eligibility. For the mechanistic substudy, eligible participants in Edinburgh only were given a separate participant information sheet and, if interested, signed a second consent form.

Eligibility criteria

Participants were assessed for eligibility by an appropriately trained doctor. The participants needed to meet the following criteria:

• women aged between 18 and 50 years
• experiencing CPP (non-cyclical with or without dysmenorrhoea or dyspareunia) of > 3 months duration
• having pain located within the true pelvis or between and below anterior iliac crests
• having no obvious pelvic pathology at laparoscopy (laparoscopy must have taken place at least 2 weeks prior to consenting to participation, but no more than 36 months prior to screening)
• using, or willing to use, effective contraception if necessary to avoid pregnancy
• able to give informed consent.

Participants could not be included if any of the following criteria were applicable:

• known pelvic pathology –
  • endometriosis (macroscopic lesions)
  • complex or > 5 cm ovarian cyst or fibroid > 3 cm
  • dense adhesions
• current malignancy under treatment
• current use of gabapentin or pregabalin
• taking gonadotropin-releasing hormone agonists, and unable or unwilling to stop
• surgery planned in the next 6 months
• history of significant renal impairment
• previous reaction to gabapentin
• breastfeeding
• pregnant
• planned pregnancy in the next 6 months
• pain suspected to be of gastrointestinal origin (positive Rome III Diagnostic Criteria)
• co-enrolment in another clinical trial of an investigational medicinal product
• metal implant/pacemaker/claustrophobia (fMRI mechanistic study only)
• receiving prohibited medications (e.g. pregabalin or high-dose opioids).

The final element of eligibility was a 4-week screening phase. Participants were asked to return numerical rating scale (NRS) pain scores weekly for 4 weeks on both the average and the worst scales (scores range from 0 to 10, where 0 is no pain and 10 is the worst pain imaginable). If at least three of four pain scores were returned on both scales, and at least two of the worst pain scores were ≥ 4, the woman was considered fully eligible for the trial and was invited to attend a randomisation visit. No study drugs were taken during this pre-randomisation screening phase, but participants were able to remain on any analgesics they were taking.

Randomisation method and minimisation variables

Once final eligibility was confirmed and consent obtained, women were randomised to the GaPP2 trial by the research staff at sites using a secure online randomisation service provided by the Birmingham Clinical Trials Unit (BCTU) (see Table 1). Participants were randomised in an equal (1 : 1) ratio to gabapentin or placebo, and a bottle number was allocated. The bottle number was sent via e-mail to the local principal investigator (PI), the trial pharmacist and the research nurse undertaking the randomisation. A ’minimisation’ procedure, incorporating a random element using a computer-based algorithm, was used to avoid chance imbalances in important prognostic variables. Strata used in the minimisation were:

• Presence or absence of dysmenorrhoea (yes, no); a pain score of ≥ 4 was considered as ‘presence of dysmenorrhoea’ (on a NRS of 0–10).
• Psychological distress measured by the General Health Questionnaire (short) (GHQ-12) (≥ 2 on a 0–12 scale).
• Use of sex hormonal treatments (yes, no) (e.g. combined oral contraceptives, progestogens and levonorgestrel-releasing intra-uterine system).
• Recruiting centre.

Investigation medicinal product information

The investigational medicinal product (IMP) was gabapentin in the form of an overencapsulated capsule. Each capsule contained 300 mg of gabapentin.

The placebo was lactose powder, which was encapsulated in the same way as the IMP to be identical in colour, shape and weight. The treatment regime was exactly the same as in the gabapentin group. For the gabapentin, it was assumed that given that the outer capsule disintegrates, the excipient falls away and exposes the original gabapentin capsule. Disintegration of the original capsule and subsequent bioavailability were not impacted by the overencapsulation.

Interventions were supplied by Sharp Clinical Services (Tredegar, UK), who procured the trial drug and manufactured the placebo capsule, overencapsulated the IMP and placebo, and dispensed into containers accordingly. This company had no role in the design, conduct, analysis or reporting of the trial.

A clinical trial pharmacist prepared the trial treatment bottle for dispense. Each trial treatment bottle contained 155 capsules. This was enough to see every participant through the dose-escalation phase. Bottles were then dispensed at each visit, depending on the optimal dose reached, up to a maximum of seven bottles.

Treatment allocations

Participants commenced the trial intervention on the day that they were randomised. They commenced on a dose of 300 mg and increased this by 300 mg every 3 days. Doses were split into three doses three times per day. Participants were given written instructions regarding dose escalation (Table 2). This took place in the first 4 weeks of treatment. Optimal dosing was determined by the participants, who were instructed to increase until they perceived adequate pain relief or intolerance to the perceived side effects. The optimal dose was then continued for 12 weeks. At the end of the treatment phase, the dose was reduced over a 2-week period (written instructions were provided; Table 3), unless there was a clinical decision to continue open-label treatment.

Blinding

Participants, investigators, research nurses and other attending clinicians all remained blind to the trial drug allocation for the duration of their participation. All participants were unblinded at the end of the trial after all data were collected. Women who perceived a benefit from gabapentin were able to discuss treatment continuance at their optimal dose on open-label treatment following discussion with their direct clinical care team.

In case of any serious adverse event (SAE), the general recommendation was to initiate management and care of the participant as if the woman was taking gabapentin. Cases that were considered serious, unexpected and possibly, probably or definitely related to the trial intervention (see Vincent et al.38) were unblinded as appropriate. In any other circumstances, investigators, research nurses and midwives remained blind to drug allocation while the participant remained in the trial. However, if the drug allocation was specifically requested immediately to assist the medical management of a participant, clinicians could contact the relevant pharmacy department where code-break envelopes were kept for each individual bottle that held the related allocation for that bottle.

Scheduled trial appointments

Trial participants completed five trial visits in total, which comprised the initial screening visit, randomisation and three follow-up visits that were conducted at weeks 4–5 (visit 3), 8–10 (visit 4) and 16–17 (visit 5 was the end of the trial) (Box 1). If no resupply of the IMP was required, women were able to complete visits 3 and 4 over the telephone. At each follow-up visit, adverse events (AEs), use of rescue analgesia and any side effects were captured, and, for visit 5 only, visits to a GP and other health-care professionals were recorded. For the mechanistic substudy, participants were asked to attend for a fMRI scan at the time of their second visit (pre randomisation), and were then asked to return at the time of their fifth visit (before unblinding).

Adherence monitoring

Adherence was evaluated by two methods. First, women were asked to complete a daily treatment diary, which documented how many capsules of IMP were taken. When women provided data for at least 5 days in 1 week, the weekly median number of capsules taken was calculated. Second, the participant was asked about adherence to the study medication at their final follow-up visit (visit 5). This was asked as a categorical response with the following groups: never (0%), hardly any (1–24%), some (25–49%), most (50–74%), almost always (75–99%) and every day (100%). Women were defined as adherent if they reported taking ≥ 50% of their study drug at visit 5 [most (50–74%), almost always (75–99%) or every day (100%)]. Women who were considered adherent as per this definition constituted the per-protocol cohort.

Participant withdrawal

A participant was considered for withdrawal from the trial treatment if, in the opinion of the investigator or the care-providing clinician or clinical team, it was medically necessary to do so. Participants could also voluntarily withdraw from treatment at any time; however, women were encouraged to continue follow-up after withdrawal from the trial treatment to minimise attrition bias.

Participants could voluntarily withdraw their consent to study participation at any time. If a participant did not return for a scheduled visit, attempts were made to contact them and, where possible, review adherence and safety data. Reasons for withdrawal were captured where possible. If a participant explicitly withdrew consent to have any further data recorded, their decision was respected and recorded on the electronic data capture system. All communication surrounding the withdrawal was noted in the patient’s medical notes, and no further data were collected for that participant.

Outcomes and assessments

Primary outcome measures

We employed dual primary outcome measures of average and worst pain scores recorded on a NRS. These were assessed and interpreted as separate outcomes. Weekly pain scores (ranging from 0, no pain, to 10, the worst pain imaginable) were recorded, during the final 4 weeks of treatment (weeks 13–16 post randomisation) in the form of (1) ‘average pain this week’ and (2) ‘worst pain this week’. The average pain score was taken as the average of ‘average pain this week’ and the worst pain score as the worst response from ‘worst pain this week’ over the 4 weeks of assessment.

Secondary outcome measures

Secondary outcomes are as follows:

• Numerical rating score of pain – to include an examination of the proportion of women who have a 30% or 50% reduction in average and worst pain scores from baseline to the end of treatment (pain scores ranging from 0 meaning no pain to 10 being worst pain imaginable).
• Short Form-12 (SF-12) quality of life – Short Form Health Survey provides summary information on physical and mental health status.39
• Brief Pain Inventory (BPI) – a comprehensive instrument for pain assessment.40
• Brief Fatigue Inventory (BFI) – to measure the severity of fatigue in adults.41
• GHQ-12 – to identify psychological distress.42
• Work and Productivity Activity Impairment Questionnaire (WPAIQ) – a valid questionnaire for assessing impairments in paid work and activities.43
• Pain Catastrophizing Scale (PCS) – one of the most widely used instruments for measuring catastrophic thinking related to pain.44
• Sexual Activity Questionnaire (SAQ) – a valid, reliable and acceptable measure for describing the sexual functioning of women in terms of pleasure and discomfort.45
• PainDETECT™ – a new screening questionnaire to identify neuropathic components in patients.46
• Pelvic Pain and Urinary/Frequency Patient Symptom Scale (baseline only) – a questionnaire that is predictive of treatment success.47
• Number of attendances to health-care professionals for CPP.
• Use of concomitant medications was recorded to identify any reductions in analgesic use.

Outcome assessment details

The schedule of outcome assessments is given in Table 4. Details of how outcomes were generated are given in Table 5.

Relevant trial data were transcribed directly into a secure web-based database. All personal information was treated as strictly confidential. Source data comprised the case report forms, questionnaires and hospital notes. Women were encouraged to report AEs occurring between clinic visits or visits to non-participating hospitals to the research nurse. Self-reports were verified against clinical notes. Pain scores were reported by women via text messages, which were imported directly into the trial database using a third-party company called ‘Textlocal’ (Chester, UK). There were validation methods built into this system to ensure data consistency and quality. Any text message sent by women that did not contain only a numerical digit between 0 and 10 generated an e-mail to the GaPP2 trial mailbox for interpretation or chasing up. All worst/average pain scores were checked on entry to ensure that they were the right way round (logically worst scores should be higher than average scores). Furthermore, any scores that had been inputted into the database manually following collection of the pain score over the telephone could not be overwritten by a subsequent text message.

Adverse events and serious adverse events

All AEs, from consent to the end of treatment (including a dose reduction if required) and whether observed directly or reported by the patient, were collected and recorded. Commonly known side effects of gabapentin were not reported as AEs but were captured directly into the database at each visit. Trial participants were asked about the occurrence of AEs and SAEs at each study visit. All SAEs were e-mailed or faxed to the sponsor’s office within 24 hours of the research staff becoming aware of the event. The local PI (or other nominated clinician) had to assign seriousness, severity, causality and expectedness (if deemed related) to the SAE before reporting. SAEs categorised by the local investigator as both suspected to be related to the trial drug and unexpected were classified as suspected unexpected serious adverse reactions (SUSARs), and were subject to expedited reporting. In the case of any SAEs, management and care of the participant was initiated as if they were taking gabapentin. The attending clinician and local PI were not made aware of the actual trial drug allocation.

Pregnancy reporting

Any participants who became pregnant while on treatment were withdrawn from treatment, and all pregnancies were followed up until delivery.

Statistical considerations

Sample size

The planned sample size of 240 women was estimated to provide 90% power to detect a minimally important clinical difference in NRS pain score of 1 point on a 0–10 scale,48 assuming a standard deviation (SD) of 2.5 (GaPP1), which is equivalent to a standardised difference of 0.4.1 To account for any increase in the risk of type-I error that may be associated with having dual outcome measures, a Bonferroni correction was applied (a two-sided alpha level of 0.025 was used). We planned to include 300 women in the trial to account for up to 20% loss to follow-up.

Statistical analysis

A comprehensive statistical analysis plan (SAP) was drawn up prior to any analysis and was provided to the independent Data Monitoring Committee (DMC) and Trial Steering Committee (TSC) for review.

In summary, categorical baseline data were summarised with frequencies and percentages. Normally-distributed continuous variables were summarised with means and standard deviations; otherwise, medians and interquartile ranges (IQRs) were presented. In the first instance, participants were analysed in the treatment group to which they were randomised (intention to treat), irrespective of adherence with the treatment protocol. All estimates of differences between groups were presented with two-sided CIs.

For the primary outcome (average and worst pain scores), means and standard deviations were reported alongside adjusted mean differences (with 97.5% CIs) that were estimated using a linear-regression model adjusting for baseline score and the minimisation parameters (presence of dysmenorrhoea, psychological distress defined by the GHQ, current use of hormonal contraceptive and recruiting hospital). Statistical significance of the treatment group parameter was determined from the p-value generated by the model. A Bonferroni correction was applied for multiplicity (differences considered statistically significant at a 2.5% level). A further analysis of pain scores was examined using a repeated-measures (multilevel) model adjusting for the minimisation parameters. All assessment times were included in the model (weeks 13–16 pain scores), with baseline score included as a covariate in the model. Time was included as a continuous variable in the model. Time-by-treatment effects were explored by including the corresponding parameter in the model; if significant (p < 0.025), a constant treatment effect was not assumed and estimates of effect size (and 97.5% CI) were generated at each time point (weeks 13–16). A general ‘unstructured’ covariance structure was assumed.

For continuous secondary outcome measures [SF-12, BPI, BFI, GHQ-12, WPAIQ, PCS, SAQ and PainDETECT], means and standard deviations were reported alongside adjusted mean differences (with 99% CIs) that were estimated using a linear-regression model adjusting for baseline score and the minimisation parameters. Binary outcomes (≥ 30% or 50% reduction in pain NRS pain scores) were summarised using frequencies and percentages. A log-binomial model was used to generate adjusted relative risks (and 99% CIs), adjusting for baseline score and the minimisation parameters. The number of attendances to health-care professionals for CPP and the use of concomitant medications were summarised descriptively only. Categorical data were summarised by frequencies and percentages. Continuous data were summarised by the number of responses, mean and SD if they were deemed to be normally distributed, and the number of responses, median and IQR if data appeared skewed. Formal statistical testing was not applied.

Sensitivity analysis was performed on the dual primary outcomes only. Every attempt was made to collect follow-up data from all participants. In particular, participants continued to be followed up even after protocol treatment violation where possible. Patients who returned zero or one NRS pain score were not included in the primary analysis; however, they were included in a sensitivity analysis using a multiple imputation approach. Missing responses were simulated using a Markov chain Monte Carlo method that assumes an arbitrary missing data pattern and a multivariate normal distribution. Variables, including treatment group and the three subgroup variables (listed below), were included in the model and were used to generate 20 simulated data sets. An analysis was then performed (as per the primary analysis) on each set, with the results combined using Rubin’s rules to obtain a single set of results (treatment effect estimate and CIs). Further sensitivity analyses were conducted to assess the effect of adherence; this was limited to the per-protocol cohort, as defined above, and an analysis was carried out to assess the effect of time between screening and randomisation.

Pre-planned subgroup analyses (limited to the dual primary outcome measures only) were completed for the following: (1) presence or absence of dysmenorrhoea (yes/no), (2) psychological distress measured by the GHQ-12 (0–1, 2–12) and (3) current use of hormonal contraceptives (e.g. combined oral contraceptives, progestogens and levonorgestrel intra-uterine system) (yes/no). The effects of these subgroups were examined by adding the subgroup by treatment-group interaction parameters to the linear-regression model; a chi-squared test was used to test the statistical significance of this parameter.

Interim analyses of effectiveness and safety end points were performed on behalf of the DMC (see Acknowledgements) on an approximately annual basis during the period of recruitment. These analyses were performed using the Haybittle–Peto principle;49 therefore, no adjustment was made in the final p-values to determine significance.

Trial oversight

Study oversight was provided by a TSC that was chaired initially by Dr Jim Thornton (University of Nottingham) and then Dr Patrick Chien (NHS Tayside), and a DMC that was chaired by Professor Mary Ann Lumsden (University of Glasgow).

The TSC provided independent supervision for the trial, and provided advice to the chief investigator and co-investigators on all aspects of the trial throughout the study. The DMC adopted the DAMOCLES charter50 to define its terms of reference and operation in relation to oversight of the GaPP2 trial.