Botulinum toxin for endometriosis-associated chronic pelvic pain: a randomised, double-masked, parallel, phase 2 trial

The trial protocol, list of protocol amendments and the statistical analysis plan will be made available on request. With publication of this article, de-identified study data will be made available on request to the authors with a signed data access agreement and after approval of a proposal.

This randomised, parallel, double-masked, phase 2 clinical trial, conducted at the National Institutes of Health (NIH) Clinical Center, evaluated the efficacy and safety of BoNT for endo-CPP. Participants were recruited from across the United States through advertisements and word-of-mouth referrals. Participants were non-pregnant women (cis-gender females) who had a history of surgically documented endometriosis, at least three months of pelvic pain, and palpable spasm in pubococcygeus, iliococcygeus, and/or obturator internus on examination. All participants had their hormonal and pain management optimised by their own health care providers; none had an indication for further endometriosis-related surgery. Those whose chronic pelvic pain could be attributed to another cause based on review of medical history and physical examination were excluded from study participation (full eligibility criteria: Supplementary Table S1 ). The study was approved by the Eunice Kennedy Shriver National Institute of Child Health and Human Development Institutional Review Board (Study 12N0083) with oversight from an independent data and safety monitoring board (DSMB). All participants gave written informed consent. The study was registered at ClinicalTrials.gov , NCT01553201 . After baseline evaluation, participants were randomly assigned 1:1 to receive onabotulinumtoxinA (Botox, Abbvie; 100 Units reconstituted in 4 mL saline) or normal saline (4 mL) using a permuted block design with block sizes of four and six ( Supplemental Figure S1 ). The NIH pharmacy generated the random allocation sequence, assigned participants to an intervention and released the assigned treatment to an unmasked nurse, who then prepared identical-appearing syringes with onabotulinumtoxinA or saline for the treating clinicians. After confirming a negative pregnancy test, masked study physicians injected the full 4 mL volume transvaginally, distributed into two to four areas of palpable muscle spasm and pain under electromyography guidance as previously described. 21 The primary outcome was assessed 1 month after the study injection. Participants were instructed to avoid changing hormonal or pain regimens including pelvic floor physical therapy and complementary treatments from 1 month before the study injection until after the 1-month primary outcome assessment. Participants could request a single open-label BoNT injection after outcome assessment at 1 month or at any later time over the next year. The same drug, dose, and procedure were used for all injections. In-person evaluations occurred 1 month after each injection and at 6 months and 12 months after the randomised injection. Follow-up calls were conducted three–14 days after each injection and three and nine months after the randomised injection. Participants, study clinicians and the research team assessing outcomes remained masked to the randomised intervention until all participants had completed the final 12-month evaluation. Participants would be removed from the study if they were unable to comply with study visits or home monitoring, had an adverse event that precluded further participation, or wished to withdraw. They could be withdrawn if urinary or fecal incontinence or pelvic organ prolapse developed after injection or if they became pregnant before the randomised study injection. Those with incontinence or prolapse or who became pregnant would have follow-up assessments but would not receive an open injection. The primary efficacy outcome, a binary self-report of benefit or no benefit, was assessed 1 month after the randomised injection. Secondary measures obtained at 1 month included participants’ classification of their pain relief as “none, minimal, mild, moderate, or excellent” and as percent improvement over baseline. A masked clinician assessed for palpable pelvic floor muscle spasm at the one-, six- and twelve-month visits. Adverse events were assessed during scheduled study phone calls and in-person visits and if participants called with concerns of a possible adverse event. Additional secondary outcomes evaluated at the one-, six-, and twelve-month visits included self-reported pain rating score (PRS; 0–10), pain medication usage, toxin effect duration, Endometriosis Health Profile (EHP), Oswestry Disability Index (ODI) score, and other participant-reported outcome measures (PROMs). A physiatrist evaluated the presence of myofascial trigger points at baseline and at study end. The EHP is a questionnaire assessing health-related quality of life specifically in women with endometriosis. Scores range from 0 (best) to 100 (worst) across five core domains (pain, control and powerlessness, emotional well-being, social support, and self-image) and additional modular items including sexual relationship. The ODI, a 10-item questionnaire, expresses the severity of disability as percentages (0–100%), with higher values indicating greater impairment. Participants classified their pain as focal or diffuse within the pelvis. We hypothesised that 70% of women receiving BoNT and 20% of those receiving placebo would report improvement at 1 month based on studies of BoNT effectiveness for pain available at the time of protocol development. 26 The sample size estimate thus required 14 women per group to detect benefit with 80% power (two-tailed test of significance) at α = 0.05. The primary endpoint was analysed using Fisher's exact test and results are reported as frequency (percentage). The number of adverse events were compared between groups using Fisher's exact test. Secondary endpoints of degree of benefit and percent improvement were analysed using Wilcoxon rank sum tests and are reported as frequency (percentage) or median (IQR). Continuous secondary endpoints listed above were compared by two-sample or paired t-tests for between and within groups, respectively; when data were not approximately normally distributed, non-parametric tests were used between (Wilcoxon rank sum test) groups and within (Wilcoxon signed-rank test) groups, respectively. Changes in measures are reported as delta and 95% CIs. Categorical data are reported as frequency (percentage) and used Fisher's exact tests (Cochran-Mantel-Haenszel or Kruskal–Wallis if singly ordinal) for between-group comparisons and McNemar's test for within groups. Logistic regression tested other associations of interest and adjusted for treatment and covariates; odds ratio (OR) 95% CI are reported. Kaplan–Meier survival analysis was used for analysis of time-to-event data and the log-rank test compared the treatment arms; results are reported as median (IQR). EHP total/subscale scores for the core questionnaire and sexual intercourse module were calculated. Persistent pain relief on EHP was analysed using mixed-model repeated-measures analysis, adjusting for concurrent benefit, active injections, disability, and pelvic floor muscles in spasm. The interrelationships among muscles in spasm, PRS score and ODI were tested using Spearman's rho and are reported by the correlation coefficient and its corresponding p-value. Data were analysed using SAS v9.4 (SAS Institute, Inc, Cary, NC). This clinical trial was supported by the Intramural Research Programs of the National Institute of Neurological Disorders and Stroke, the NIH Clinical Center, the Eunice Kennedy Shriver National Institute of Child Health and Human Development and the National Human Genome Research Institute at the US National Institutes of Health (NIH). Allergan provided study drug and funds for independent study monitoring under a Clinical Trial Agreement with the NIH and had no other role in the study.

Participants were recruited between July 24, 2014 and May 8, 2018. 49 women were assessed for inclusion. 15 declined participation, and four did not meet eligibility criteria. 30 women were randomly assigned and received the study intervention. No participant dropped out or was lost to follow-up. Two sisters were initially included; one was censored to limit bias on recommendation of the DSMB. Results of the remaining 29 participants are presented ( Fig. 1 ). Fig. 1 Study profile . Study profile . Demographics did not differ between cohorts ( Table 1 ). Participants were young (mean age 31 years; range 18–50) and had endured years of pelvic pain (mean 11.5 years; range 2–25). At enrolment, 22 of 23 women using hormonal therapy for endometriosis had menses suppression; six avoided hormones due to side effects. For pain control, at the first study visit, nine women were using only non-steroidal anti-inflammatory drugs (NSAIDs) and/or acetaminophen; 17 were using various combinations of NSAIDS, acetaminophen, opioids, muscle relaxants, cannabis, and benzodiazepines. One woman had an InterStim™ bladder device. Eight participants were undergoing pelvic floor physical therapy. Three participants in the toxin cohort and five in the placebo group included pelvic floor physical therapy in their pain management. Five others had tried pelvic floor physical therapy prior to study participation, found it ineffective, and did not use it during the study. Table 1 Baseline demographics and clinical characteristics. Characteristic Total (n = 29) BoNT (n = 15) Placebo (n = 14) Age—years, mean (SD) 30.9 (7.3) 31.0 (7.3) 30.7 (7.5) BMI, mean (SD) 25.9 (5.8) 26.5 (6.6) 25.2 (4.9) Race—no. (%) a  White 24 (82.8) 10 (66.7) 14 (100)  Black 3 (10.3) 3 (20) 0  Asian 1 (3.5) 1 (6.7) 0  Mixed race 1 (3.5) 1 (6.7) 0 Ethnicity- Hispanic or Latino—no. (%) 4 (13.8) 1 (6.7) 3 (21.4) Duration of pelvic pain- years, mean (SD) 11.5 (6.6) 12.4 (5.8) 10.6 (7.5) Current pain treatment:  Medication—no. (%)  None 3 (10.3) 2 (13.3) 1 (7.1)  NSAID/acetaminophen only 9 (31) 4 (26.7) 5 (35.7)  NSAIDS, acetaminophen, opioids, muscle relaxants, gabapentin, cannabis, and/or benzodiazepines 17 (58.6) 9 (60) 8 (57.1)  Device  Implanted bladder device (Interstim™) 1 (7.1) 0 1 (7.1)  Pelvic floor physical therapy 8 (27.5) 3 (20) 5 (35.7) Hormonal treatment—no. (%) 23 (79.3) 11 (73.3) 12 (85.7) Pain Rating Score—mean (SD) 4.9 (1.9) 4.7 (2.1) 5.0 (1.6)  PRS ≥ 4 - no. (%) 20 (69) 9 (60) 11 (78.6) Pelvic floor muscle spasm—no. (%)  2 6 (20.7) 5 (33.3) 1 (7.1)  3 3 (10.3) 2 (13.3) 1 (7.1)  4 12 (41.4) 5 (33.3) 7 (50)  5 4 (13.8) 1 (6.7) 3 (21.4)  6 4 (13.8) 2 (13.3) 2 (14.3) Disability based on Oswestry—no. (%)  Minimal 13 (44.8) 6 (40) 7 (50)  Moderate 13 (44.8) 8 (53.3) 5 (35.7)  Severe 3 (10.3) 1 (6.7) 2 (14.3) Pain characteristics—no. (%)  Menstrual pain b 26 (92.9) 13 (86.7) 13 (92.9)  Non-menstrual pain 28 (96.6) 15 (100) 13 (92.9) Among sexually active (n = 14) —no. (%):  Coital pain 12 (85.7) 5 (71.4) 7 (100)  Avoided intercourse due to pain 1 (7.1) 1 (14.3) 0  No coital pain 1 (7.1) 1 (14.3) Abbreviations: n sample size, BoNT botulinum toxin, SD standard deviation, BMI body mass index, no. number, NSAID non-steroidal anti-inflammatory drug, PRS pain rating score. a Self-reported race/ethnicity. b Data point missing from 1 participant. Baseline demographics and clinical characteristics. Abbreviations: n sample size, BoNT botulinum toxin, SD standard deviation, BMI body mass index, no. number, NSAID non-steroidal anti-inflammatory drug, PRS pain rating score. Self-reported race/ethnicity. Data point missing from 1 participant. At baseline, all participants had spasm in at least two of the six palpated pelvic floor muscles; 20 (69%) had spasm in more than three. Spasm was a main contributor to pelvic pain as palpation of these muscles reliably elicited each individual's typical pelvic pain. Sixteen of 29 (55%) participants were at least moderately disabled (ODI score ≥21%). 15 women were assigneed to BoNT; 14 to placebo. On the primary outcome measure assessed 1 month after injection, 11 (73%) of 15 women receiving BoNT reported benefit compared to four (29%) of 14 receiving placebo (p = 0.027; Fig. 2 A). The BoNT cohort reporting a greater degree of benefit (p = 0.048; Fig. 2 B) and a higher percent improvement (median 16.0 [IQR 6–51] vs 0.5 [IQR 0–18.5]; p = 0.034; Fig. 2 C) than those receiving placebo. Fig. 2 Self-reported benefit 1 month after injection. Panel A presents the number of participants with benefit and no benefit by cohort, showing that more participants in the botulinum toxin (BoNT) cohort reported benefit (p = 0.027). Panel B presents benefit using the self-assessed degree of benefit, which appeared greater in those receiving BoNT than those receiving placebo (p = 0.048). Panel C depicts the greater percent benefit reported by those receiving BoNT compared with those receiving placebo (p = 0.034). For Panel C, each box represents the middle 50% of the data, with the length representing the first (25th percentile) and third (75th percentile) quartiles; inside the boxes are the medians (line) and mean (marker symbols); whiskers represent the range of the data; outliers are the individual data points outside the whiskers. Self-reported benefit 1 month after injection. Panel A presents the number of participants with benefit and no benefit by cohort, showing that more participants in the botulinum toxin (BoNT) cohort reported benefit (p = 0.027). Panel B presents benefit using the self-assessed degree of benefit, which appeared greater in those receiving BoNT than those receiving placebo (p = 0.048). Panel C depicts the greater percent benefit reported by those receiving BoNT compared with those receiving placebo (p = 0.034). For Panel C, each box represents the middle 50% of the data, with the length representing the first (25th percentile) and third (75th percentile) quartiles; inside the boxes are the medians (line) and mean (marker symbols); whiskers represent the range of the data; outliers are the individual data points outside the whiskers. The mean PRS score over the 30 days before study intervention was 4.9 (range 0.7–9.1); 20 women had a mean PRS of 4 or more. The PRS scores were not different between the cohorts before or after injection. However, among those with at least moderate baseline pain (PRS ≥ 4), scores improved in the BoNT cohort only (BoNT median 6.2 [IQR 6.0–6.7] to 4.9 [IQR 4.0–6.7]; p = 0.028; placebo median 5.4 [IQR 4.1–6.2] to 6.0 [IQR 4.2–6.6]; p = 0.90; Supplemental Figure S2 ). The BoNT (baseline 3.5 muscles [95% CI 2.8–4.3] vs 1 month 2.5 muscles [95% CI 1.4–3.6]; p = 0.019) but not the placebo group (baseline 4.3 muscles [95% CI 3.7–4.9] vs 1 month: 3.8 muscles [95% CI 3.0–4.6]; p = 0.19) had fewer muscles in spasm. There was no difference in ODI scores before versus 1 month after injection either within (BoNT Δ 0.3, 95% CI −4.3 to 4.8; p = 0.90; placebo Δ 2.6, 95% CI −0.2 to 5.3; p = 0.066) or between groups (Δ −2.3, 95% CI −7.5 to 2.9; p = 0.37). By the 1-month visit, three participants (23%) in the BoNT group reduced pain medication use compared to one (8%) in the placebo group (p = 0.49). No participant reported urinary or fecal incontinence or any serious adverse event by the 1-month visit. The most common non-serious adverse events within 1 month of the study injection were transiently worsened pelvic pain (2 (13%) of 15 in the BoNT group; 0 (0%) of 14 in the placebo group; p = 0.48) and mild urinary symptoms (3/15 (20%) in the BoNT cohort; 0/14 (0%) in the placebo cohort; p = 0.22). The total number of non-serious adverse effects during the month following the assigned injection was not different between cohorts (p = 0.11; Table 2 ). There were no adverse events related to the study intervention 1–3 months after the assigned injection and before any open injection in either cohort ( Table 3 ). Three (20%) of 15 women in the BoNT group and three (21%) of 14 in the placebo cohort reported abdominal pain or an episode of worsened pelvic pain within 2 months of open injection (p > 0.99). One participant in each cohort had mild urinary symptoms within 2 weeks of open injection and one woman in the BoNT cohort had transient stress incontinence beginning 1 month after open injection. Table 2 Adverse events within 1 month of assigned injection. BoNT cohort Participant number Related Grade b Placebo cohort Participant number Related Dysuria without urinary tract infection BTX001 Possible 1 Ovarian cysts/bleeding/mood changes with Mirena IUD BTX007 No Transient worsened pelvic pain (2 patients) BTX006, BTX027 Possible 1 Migraine headache/tremor BTX031 No Urinary urgency, single episode BTX024 Possible 1 Intermittent double voiding ×1 day BTX024 Possible 1 Rash/folliculitis BTX033 No Hip pain from skiing accident BTX033 No Migraine headache with fecal incontinence a BTX023 No a History of similar events prior to study participation. b Grade not reported for unrelated adverse events. Table 3 Adverse events between 1 month and 3 months after assigned injection/before open injection, if any. Adverse event Participant number Related Onset after randomised injection BoNT cohort  Restless legs BTX018 No 6 weeks Placebo cohort  Benign breast cyst BTX025 No 6 weeks  Hand pain/swelling BTX031 No 6 weeks  Hospitalisation for depression/anxiety a BTX031 No 2.5 months a History of similar events prior to study participation. Adverse events within 1 month of assigned injection. History of similar events prior to study participation. Grade not reported for unrelated adverse events. Adverse events between 1 month and 3 months after assigned injection/before open injection, if any. History of similar events prior to study participation. Two participants, both in the BoNT group and each with more than 100 days of relief from the initial injection, declined open BoNT injection ( Supplemental Figure S3a ). 16 of the 27 remaining women (nine placebo, seven BoNT) elected open toxin injection at the 1-month timepoint ( Supplemental Figure S3b ); 11 chose later injection ( Supplemental Figure S3c ). Controlling for initial treatment, those without benefit more often requested open-label injection at 1 month compared with those with benefit (OR 7.94, 95% CI 1.19–52.34; p = 0.032). Adjusting for treatment and benefit, higher baseline ODI score predicted request for open injection at 1 month (OR 1.11, 95% CI 1.02–1.22; p = 0.022). 22 of 27 women reported benefit after open injection. Decrease in pain rating in individual pelvic floor muscles at the end of the study is shown in Supplemental Figure S4 . Seven (50%) of 14 women in the placebo group and nine (69%) of 13 in the BoNT group who requested open injection reported benefit at study end. Of participants without benefit from assigned injection at 1 month, nine of ten in the placebo group and two of four in the BoNT group had benefit with the open injection. Three participants (one placebo, two BoNT) had no benefit after either injection; one in the placebo group reported brief, transient benefit after masked injection. The duration of benefit from the masked injection was longer in the BoNT than placebo cohort (median duration BoNT group 57 days [IQR 0–188] vs placebo cohort median 0 days [IQR 0–60]; p = 0.023; Fig. 3 A). Similarly, before the open injection, benefit was more likely to have worn off after placebo than BoNT (p = 0.020; Fig. 3 B). Overall (p = 0.019; Fig. 4 A) and particularly in the placebo group (median 0 days [IQR 0–60] vs 49 days [IQR 0–346]; p = 0.0018; Fig. 4 C), benefit was briefer with the masked injection (median 7 days [IQR 0–110]) than the open injection (162 days [IQR 30–321]). There was no difference between the masked and open injection in benefit duration in the BoNT cohort (p = 0.31; Fig. 4 B). Women reporting focal pelvic pain had longer lasting benefit than those with diffuse pelvic pain (median 60 days [IQR 0–168] vs 0 days [IQR 0–7]; p = 0.020), especially in the BoNT group (p = 0.0041; Supplemental Figure S5 ). By the end of the study, 20 of 26 participants who used pain medication at baseline decreased its use (p < 0.0001), with 12 (92%) of 13 in the BoNT group and eight (62%) of 13 in the placebo group using less medication (p = 0.061). Fig. 3 Benefit duration after masked injection. Panel A shows the duration of benefit as a Kaplan Meier survival curve. The duration of benefit was longer in the toxin group than the placebo group (p = 0.023). The panel B Kaplan Meier survival curve shows that the time between injections without benefit was longer in the placebo than in the toxin group (p = 0.020). Fig. 4 Benefit duration: masked versus open injection. Panel A demonstrates that the duration of benefit was shorter after masked compared with open injection (p = 0.019). Panel B shows the lack of difference between benefit duration after the masked compared with the open injection in the botulinum toxin (BoNT) group (p = 0.31) while Panel C illustrates the significant difference between the masked and open injections in the placebo group (p = 0.0018). Benefit duration after masked injection. Panel A shows the duration of benefit as a Kaplan Meier survival curve. The duration of benefit was longer in the toxin group than the placebo group (p = 0.023). The panel B Kaplan Meier survival curve shows that the time between injections without benefit was longer in the placebo than in the toxin group (p = 0.020). Benefit duration: masked versus open injection. Panel A demonstrates that the duration of benefit was shorter after masked compared with open injection (p = 0.019). Panel B shows the lack of difference between benefit duration after the masked compared with the open injection in the botulinum toxin (BoNT) group (p = 0.31) while Panel C illustrates the significant difference between the masked and open injections in the placebo group (p = 0.0018). EHP total and subscale scores for pain, control/powerlessness and sexual intercourse improved over the course of one year ( Supplemental Table S2 ). Persistent pain relief assessed by the EHP was associated with ongoing benefit (p = 0.0075), fewer concurrent pelvic floor muscles in spasm (p = 0.0006) and lower baseline ODI score (p < 0.0001). Improvement in sexual intercourse was similarly associated with fewer concurrent pelvic floor muscles in spasm (p = 0.032) and lower concurrent ODI score (p = 0.016). The study measures of muscle spasm, pain level and disability level were interrelated as evidenced by correlations between the baseline timepoint and the timepoints with the lowest PRS score, lowest ODI score and fewest pelvic floor muscles in spasm. Baseline PRS correlated with baseline ODI score (r = 0.73, p < 0.0001) and lowest ODI score (r = 0.68, p < 0.0001). Baseline ODI score correlated with lowest number of muscles in spasm (r = 0.38, p = 0.040) and lowest PRS (r = 0.52, p = 0.0039). Baseline number of pelvic floor muscles in spasm correlated with lowest ODI score (r = 0.42, p = 0.024). Lowest number of pelvic floor muscles in spasm correlated with lowest PRS (r = 0.49, p = 0.0069). Lowest PRS correlated with lowest ODI score (r = 0.53, p = 0.0033). Non-serious adverse events at least possibly related to BoNT after open injection included abdominopelvic and leg pain, transient difficulty voiding and fatigue. Two patients were briefly hospitalised for worsened pelvic pain immediately after open injection ( Table 4 ). Adverse events were not more frequent in those initially randomised to BoNT compared to those who initially received placebo. Table 4 Adverse events up to 3 months after open injection. Adverse event Participant number Related Grade b Onset after open injection Comments BoNT cohort  Abdominal pain BTX032 Possibly 2 Day of open injection Hospitalisation  Abdominal pain/constipation BTX009 Yes 2 1 week Hospitalisation  Transient stress incontinence BTX027 Possibly 2 1 month When coughing or sneezing  Leg and pelvic pain BTX033 Possibly 1 1 day  Migraine, fever, abdominal cramping a BTX026 Possibly 1 1 day  Transient intermittent double voiding BTX024 Possibly 1 Day of open injection  Intermittent leg buckling a BTX015 No 5 days  Yeast infection BTX006 No 5 days  Diarrhoea, fatigue, lightheadedness BTX033 No 1 day Placebo cohort  Increased pelvic pain BTX022 Possibly 1 1 day  Unilateral groin pain BTX029 Possibly 1 2 weeks  Transient difficulty emptying bladder BTX030 Possibly 1 2 weeks  Piriformis syndrome/pain BTX022 Possibly 1 2 months  Depression/suicidality a BTX007 No 3 months a History of similar events prior to study participation. b Grade not reported for unrelated adverse events. Adverse events up to 3 months after open injection. History of similar events prior to study participation. Grade not reported for unrelated adverse events.

This randomised, double-masked, parallel placebo-controlled clinical trial demonstrates the efficacy of intramuscular pelvic floor BoNT for women with endo-CPP and pelvic floor spasm. Relief of pain was found on the primary outcome of self-reported benefit/no benefit at 1 month after masked injection and supported by multiple secondary outcomes including patient-assessed percent improvement, greater degree of benefit, and improvement in PRS among those with at least moderate baseline pain in the BoNT cohort. In addition, the participants receiving BoNT had fewer pelvic floor muscles in spasm, an objective finding. Furthermore, the two participants not requesting a second injection both received BoNT as their masked injection. Findings after the optional open injection further support efficacy. Seventy-five percent of participants with no benefit after randomised injection had sustained benefit following active treatment. Further, at one year, 66% of participants experienced sustained benefit and 77% of those initially taking pain medication used less. Importantly, even those experiencing an adverse event after the study injection requested open injection suggesting that they assessed the potential benefit of relief from persistent pelvic pain as outweighing the risk of the non-serious adverse events. We selected 1 month for primary outcome assessment based on the known time course for BoNT effect in other indications. In motor disorders, benefit typically peaks about 1 month after injection. 27 Although BoNT irreversibly inhibits neurotransmission, clinical effects are transient. The poisoned presynaptic nerve terminals sprout, re-establishing the neuromuscular junction over approximately 3 months. Thus, benefit from BoNT intramuscular injection lasts about 12 weeks leading to typical three-month re-treatment intervals for many indications including chronic migraine, dystonia, spasticity, and hemifacial spasm. For some indications, such as overactive bladder, BoNT has demonstrated effects lasting six-eight months. 28 The duration of benefit in pain may differ from muscle disorders due to differential effects on motor and pain pathways. Results in the active study injection cohort further indicate that repeat injection can sustain therapeutic benefit. Decades of experience with repeated BoNT injections across multiple indications confirm that effectiveness can be maintained for years without serious or cumulative adverse effects. 29 , 30 Our results differ from the three previously published placebo-controlled trials of BoNT for chronic pelvic pain in women primarily due to differences in study design, outcome assessment, and target population. Abbott et al. found a significant decrease in nonmenstrual pelvic pain with BoNT, but no difference in overall pain on their visual analogue scale (VAS) primary outcome measure. 23 Wide variability in their cohorts’ VAS ratings may explain this as well as a higher than anticipated placebo response that the authors noted may have led to their study being underpowered. They also used a lower toxin dose into fixed predetermined sites in fewer muscles and included women with CPP of any aetiology. Dessie et al. used a higher dose than ours, injecting pelvic floor trigger points. 24 BoNT was not more effective than placebo in decreasing pain in the most painful muscle group on VAS, however, they did observe a higher percentage of BoNT than placebo participants rating their overall pelvic pain as improved. Interpretation of their results is complicated by more severe baseline pain in the active drug cohort and the initiation of pelvic floor physical therapy at 4 weeks after BoNT injection. The randomised, masked trial by Spruijt et al. focused on women with CPP and increased pelvic floor muscle tension who had failed previous pelvic physical therapy. 26 They compared intravaginal pelvic floor muscle injection of 100U onabotulinumtoxinA to placebo. All participants also received a customised programme of pelvic floor muscle therapy including four in-person sessions and home exercises. Their primary outcomes, assessed 26 weeks after the randomised injection, included the painDETECT scale VAS pain scores over the prior four weeks and number of participants reporting feeling “much better” or “very much better” on the Patient Global Impression of Improvement. Both cohorts benefited from the study interventions, with no pain score difference at 26 weeks. The authors did find significantly more participants in the BoNT cohort reporting being “much better” or “very much better” at eight weeks. They questioned the significance of the eight week outcome as the difference was not sustained at 26 weeks. However, the typical time course of BoNT suggests that the toxin effects might have largely worn off by 26 weeks, their primary outcome timepoint. In addition, the authors noted that their power calculation assumed no response to placebo injection, leading to a potentially underpowered study. It is also important to note that both Dessie et al. 24 and Spruijt et al. 26 required all subjects to engage in pelvic floor physical therapy after the study toxin injection, adding a potentially confounding pain management strategy. Thus, we may have found benefit while other controlled studies did not because of differences in dosing, muscle selection, technique, timing of outcome assessment, and target population. Further, our participants were able to continue their pain management strategies but refrained from adding other pain interventions, like pelvic floor physical therapy, before the primary outcome assessment. We chose the dose of 100 Units onabotulinumtoxinA based on the authors’ long experience with therapeutic botulinum toxin injections, prior negative studies at lower doses, and the concern, supported by the literature, that higher doses increased the risk of urinary and/or fecal incontinence. 21 We chose the diluent and matching saline placebo volume of 4 mL to provide sufficient volume to inject multiple areas of spasm within our dose limit. Adverse events in our participants are consistent with other clinical trials of BoNT for pelvic pain with the most frequent adverse effects of new or worsening, transient urinary incontinence, usually stress incontinence 23 , 24 , 25 Similar to our study, the previous clinical trials found no significant difference in adverse events between the active and placebo cohorts. 23 , 24 , 25 The strengths of our study include its randomised, double-masked, parallel placebo-controlled design with a binary primary outcome measure and multi-modal corroborative secondary subjective and objective outcome assessments. Our participants, all premenopausal women who had undergone thorough clinical evaluation for potential causes of pelvic pain as well as medical and surgical treatment for endometriosis, also represent a more homogeneous study population. We achieved full cohort retention and obtained complete data on all participants throughout the year-long study. To minimise variability, we standardised BoNT dose and injection procedure and prohibited changes in pain and hormone treatments until after the 1-month primary outcome assessment. The study presents several limitations. First, pain assessment is, by its nature, subjective. There currently exists no validated objective way to detect pain or to measure its intensity or change in intensity over time. In this study, we used multiple different subjective approaches to assess pain and change in pain with the intervention and the objective evaluation of pelvic floor muscle spasm. While the randomised, double-masked, placebo-controlled design enhances the robustness of the findings, the sample size is relatively small. We also did not require a minimum baseline pain intensity, making it difficult to detect a lessening in pain intensity for those with low baseline scores. However, in the secondary analysis we were able to detect a change in PRS score in those with a baseline pain level ≥4. Furthermore, assessment of pain relief was based on self-reported measures that can be influenced by individual pain tolerance, sociological and psychological factors that were not accounted for in our data. Since pelvic pain in all our participants persisted despite treatments being used at the time of enrolment, they were allowed to continue medications, pelvic floor PT, and ancillary approaches throughout the study, although without changes from 1 month before randomisation until after the 1-month evaluation. Later changes in pain management could potentially confound the long-term outcome. However, we found that pain medication use decreased over time after the open injection. Focussing solely on a population with spasm and endo-CPP limits generalisability to other types of pelvic pain. Restricting the muscles for injection to the pelvic floor and selecting a single toxin type and brand, dose, and dilution leaves optimisation of these parameters unexplored. Despite these limitations, the study provides valuable insights into the efficacy of BoNT for treating endo-CPP. In conclusion, our phase 2 clinical trial provides compelling evidence of sustained pain improvement from BoNT in women with endo-CPP and pelvic floor spasm without serious adverse effects. BoNT can be safely administered as an office procedure and integrated into standard care for these patients. Decades of experience with therapeutic BoNT confirm that it can be combined safely with pain medications and ancillary pain treatments. These results merit confirmation in further clinical trials. Such studies will help define optimal dosage and other aspects of this approach and explore the BoNT effectiveness in women with pelvic floor spasm and pelvic pain associated with etiologies other than endometriosis.

Drs. Karp and Stratton conceived of the study and contributed to all aspects of the study and paper. Drs. Shah contributed to study design, participant evaluation, analysis of data, and writing the paper. Dr. Sinaii contributed to study design, development of statistical analysis plan, analysis of data, and writing the paper. Drs. Tandon, Aredo, and Phan contributed to the conduct of the study, data collection, analysis of data, and writing the paper. Dr. Merideth contributed to conduct of the study and writing the paper. Dr. Salmeri contributed to analysis of the data and critical review of the manuscript. Drs. Stratton, Tandon, and Phan directly accessed and verified the underlying data reported in the manuscript. All authors read and approved the final version of the manuscript. All authors confirm that they had full access to all the data in the study and accept responsibility to submit for publication.

Chronic pelvic pain (CPP) is a major cause of morbidity and lowered quality of life, affecting up to 27% of women worldwide. 1 The most frequently identified antecedent cause is endometriosis, present in 40–87% of women with CPP. 2 , 3 Endometriosis is characterised by hormone-sensitive endometrial tissue forming lesions outside of the uterus, with predominant symptoms of pain and infertility. Standard care encompasses surgery to remove lesions and hormonal therapy to suppress menses and lesion growth. Despite endometriosis-specific treatment, pelvic pain persists in 30% of those with a history of endometriosis. 3 Patients also use analgesic medications, pelvic floor physical therapy, and other ancillary treatments, which often provide inadequate pain relief. The presence and intensity of pain in patients with endometriosis do not correlate with the type, location, or extent of endometriosis lesions, 3 , 4 making it unlikely that lesions are the sole cause of pain before treatment. Further, although endometriosis lesions may initiate pain, they cannot explain the persistence of pain after surgical intervention. Endometriosis and its associated pain have complex and systemic pathophysiologies. 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 Endometriosis lesions develop their own blood supply and innervation. The lesions and adjacent peritoneal fluid contain immune cells and are associated with neurogenic inflammation and elevated levels of pain mediators, including substance P, calcitonin gene-related peptide (CGRP), prostaglandin E2, tumour necrosis factor-α, interleukin-8 and interleukin-1β. Lesion innervation includes pain afferent, sympathetic and parasympathetic fibres. This innervation provides direct access to pain neurotransmission pathways and the central nervous system. The transition to sustained abdominopelvic pain following endometriosis-specific treatment and in the absence of lesions likely arises from peripheral and central sensitisation. 6 , 7 , 11 Such pain may be best classified as “nociplastic pain,” defined by the International Association for the Study of Pain as pain in the absence of actual or threatened neuronal damage (neuropathic pain) or an identifiable non-neuronal cause (nociceptive pain). Myofascial pain is similarly associated with peripheral and central pain sensitisation. People with myofascial pelvic pain may have tight bands of skeletal muscle, hard tender nodules, and/or myofascial trigger points within pelvic floor muscles. The reported prevalence of myofascial dysfunction in women with CPP ranges from 14 to 93% 12 , 13 , 14 depending on the population studied and type of assessments performed. Uncertainty arises because evaluation for myofascial dysfunction, myofascial trigger points and pelvic floor tone are neither standardised nor included in a routine gynaecological examination. Myofascial dysfunction with pelvic floor hypertonia and spasm is also present in women with endometriosis. 14 , 15 Shared innervation and pain transmission pathways between the visceral organs and pelvic somatic musculature leads to viscerosomatic and viscerovisceral convergence. Convergence of sensory input can result in cross-visceral sensitisation and referred pain. Viscerovisceral interactions that exacerbate pain have been demonstrated in animals with pain initially triggered by endometriosis. 6 Rats with endometriosis exhibited increased pain behaviours with ureteral calculosis, indicating hyperalgesia when compared to those without endometriosis. Viscerovisceral interplay has similarly been demonstrated in women with dysmenorrhoea and a history of urinary stones who had both more severe dysmenorrhoea and greater central nervous system sensitisation than women with dysmenorrhoea without this history. 16 Cross-visceral sensitisation can contribute to pain severity or persistence in individuals with comorbid pelvic or lumbosacral pain conditions. Myofascial dysfunction may not only underlie endometriosis-associated pain but could also arise as a consequence of it through viscerosomatic convergence. While endometriosis lesions may play a role in the development of pelvic floor muscle spasm, the spasm then can contribute to ongoing endometriosis-associated CPP (endo-CPP). 17 Botulinum toxin (BoNT) is increasingly used for chronic pain management. Controlled studies demonstrated efficacy in chronic migraine, leading to FDA approval for onabotulinumtoxinA in 2010. Accumulating evidence supports BoNT effectiveness in many pain conditions, including painful bladder syndrome, chronic myofascial pain, neuropathic pain, post-herpetic neuralgia, and piriformis syndrome. 18 BoNT prevents acetylcholine release from presynaptic neurons at neuromuscular junctions, interfering with neurotransmission and causing functional muscle denervation. In clinical use, the intramuscular dose of toxin is titrated to weaken, but not paralyse muscles. However, pain relief does not rely solely on muscle relaxation. BoNT also acts on central and peripheral pain pathways, blocking release of pain neurotransmitters including substance P and calcitonin gene-related peptide (CGRP). 19 , 20 Given its effects on muscle hypertonia and, independently, on pain neurotransmission, the use of BoNT for chronic pelvic pain warrants consideration. Case series and uncontrolled studies of BoNT for pelvic pain in women provide preliminary evidence of efficacy and safety 21 , 22 although the three randomised placebo-controlled trials published to date did not demonstrate benefit compared to placebo. 23 , 24 , 25 These earlier studies lacked a clearly-defined antecedent cause of pelvic pain and did not focus on endometriosis-associated chronic pelvic pain. Focussing on the endo-CPP population provides context for an identifiable initiator of CPP with a putative mechanism of action for its chronification. In this Phase II clinical trial, women with endo-CPP and pelvic floor hypertonia were randomised to onabotulinumtoxinA or placebo saline injection into painful pelvic floor muscles with spasm. We hypothesised that BoNT might alleviate both spasm and chronic pelvic pain, thereby offering a novel therapeutic approach for endo-CPP patients.

Dr. Karp received compensation for serving as faculty for educational courses on the use of botulinum toxins. Dr. Stratton received royalties from UpToDate for a chapter on Diagnosis and Management of Acute Pelvic Pain and from Frontiers in Reproductive Health as Speciality Chief Editor, Gynaecology, participated in an Endometriosis Research Day at the Open Endoscopy Forum Cambridge, Massachusetts, and reviewed a book proposal on endometriosis for Elsevier. Dr. Salmeri has served in unpaid positions as an Early Career Ambassador for the World Endometriosis Society, on the Journal Club Editorial Team of the European Society of Human Reproduction and Embryology, and for the International Biometric Society—Italian Region. Dr. Aredo has received consulting fees from AstraZenca for work unrelated to this study, and payment or honoraria from the Binaytara Foundation, Kaplan, and IDEOlogy Health. She has participated on DSMB or advisory boards for Johnson & Johnson and Bristol Meyers Squibb. Drs. Tandon, Merideth, Phan, Shah, and Sinaii have no disclosures. This research was supported by the Intramural Research Program of the National Institutes of Health (NIH). The contributions of the NIH authors were made as part of their official duties as NIH federal employees, are in compliance with agency policy requirements, and are considered Works of the United States Government. However, the findings and conclusions presented in this paper are those of the authors and do not necessarily reflect the views of the NIH or the U.S. Department of Health and Human Services. Allergan, Inc provided study drug and funds for independent study monitoring through a Clinical Trials Agreement with the National Institutes of Health.