Randomised controlled trial of bipolar radio-frequency endometrial ablation and balloon endometrial ablation

Authors


Dr M. Y. Bongers, Department of Obstetrics and Gynaecology, Máxima Medical Centre, PO Box 7777, 5500 MB Veldhoven, The Netherlands.

Abstract

Objective  To compare the effectiveness of two second-generation ablation techniques, bipolar radio-frequency impedance-controlled endometrial ablation (NovaSure) and balloon ablation (ThermaChoice), in the treatment of menorrhagia.

Design  Double-blind, randomised, controlled trial.

Setting  A large teaching hospital with 500 beds in The Netherlands.

Population  Women suffering from menorrhagia referred by their general practitioner.

Methods  Women suffering from menorrhagia, without intracavitary abnormalities, were randomly allocated to bipolar radio-frequency ablation (bipolar group) and balloon ablation (balloon group) in a 2:1 ratio. At follow up, both women and observers were unaware of the type of treatment that had been performed.

Main outcome measures  The main outcome measure was amenorrhea at 3, 6 and 12 months after randomisation.

Results  One hundred and twenty-six women were included in the study, of which 83 were allocated to the bipolar group, and 43 to the balloon group. No complications occurred in either of the treatment groups. At the one-year follow up stage, amenorrhea rates were 43% (34/83) in the bipolar group and 8% (3/43) in the balloon group (treatment effect in time P < 0.001). At this stage, 90% of the patients in the bipolar group were satisfied with the result of the treatment against 79% in the balloon group (treatment effect in time P= 0.003).

Conclusion  The bipolar radio-frequency impedance-controlled endometrial ablation system is more effective than balloon ablation in the treatment of menorrhagia.

INTRODUCTION

One in 20 women suffers from menorrhagia.1,2 Provided that intracavitary abnormalities are absent, several transcervical endometrial ablation methods are available to treat these women. The first-generation devices, used for the ablative treatment of endometrium, are endometrial laser ablation,3 transcervical resection of the endometrium and rollerball ablation.4

Despite reported success rates as high as 90%,5–8 these first-generation endometrial ablation techniques have not been implemented on a large scale in general practice. One reason for this limited popularity might be a fear of complications following hysteroscopic endometrial ablations, such as fluid overload and water intoxication.9,10 Second-generation techniques for endometrial ablation devices have overcome these disadvantages, and require less skill on the part of the surgeon. They also have in common the fact that a uterine distension medium does not have to be used under high pressure, thus minimising the risk of fluid overload and water intoxication.

The first second-generation ablation device was the hot water balloon ablation therapy (ThermaChoice, Gynecare, Johnson and Johnson, Somerville, New Jersey, USA) developed by Neuwirth in 1994.11 This treatment was found to be safe,12 and at least as effective as transcervical resection of the endometrium13,14 and rollerball ablation.15–17 The NovaSure Endometrial Ablation Device (Novacept, Palo Alto, California, USA) is a new second-generation device that uses a bipolar, radio-frequency, impedance-controlled endometrial ablation system to evaporate endometrial tissue. We performed a randomised controlled trial with women suffering from menorrhagia, but with an absence of intracavitary abnormalities. The proportion of women with amenorrhoea following treatment and patient satisfaction with the bipolar radio-frequency device was compared with that of a hot water balloon catheter.

METHODS

A randomised controlled trial was performed in the Máxima Medical Centre, Veldhoven, The Netherlands. The Máxima Medical Centre is a teaching hospital with 500 beds in the south of The Netherlands. The ethics committee of the hospital approved the study.

Women with menorrhagia as indicated on the pictorial chart described by Higham et al.,18 with a score of 150 points or more, were eligible for the trial. Saline infusion sonography or diagnostic hysteroscopy was required to confirm a normal uterine cavity with histological benign endometrium and a cavity length meaning 6–11 cm. All women had to have a normal Pap smear, a negative Chlamydia test of the cervix and a premenopausal follicular stimulating hormone (FSH) level of less than 40 IU/L. Exclusion criteria were documented as coagulopathies, patients treated with anticoagulants, a desire to preserve fertility and prior uterine surgery (except low segment caesarean section). All participants gave a written informed consent before enrolment. After a patient had given this consent, she was scheduled for surgical intervention. Randomisation was performed in the operating room by one of the authors (MB) just before the beginning of treatment. The randomisation sequence was computer generated, and to conceal the allocation, opaque sealed envelopes were used. Patients and investigating doctors were unaware of the result of the randomisation, and remained uninformed of the ablation method used during the study. The ratio between women allocated to bipolar radio-frequency ablation and women allocated to balloon ablation was 2:1. The reason for the 2:1 ratio was to enable rapid information to be obtained on the performance of the new bipolar radio-frequency technique.

The ablation treatments in both arms were performed by one gynaecologist (MB). Patients received no medical pretreatment, and the ablation was not timed in the menstrual cycle. All patients had Naproxen 250 mg 12 hours and 1 hour before treatment.

The NovaSure endometrial ablation system consists of a generator and a disposable NovaSure device (Novacept). The NovaSure radio-frequency generator is a constant power output generator with a maximum energy delivery of 180 W. The generator has a power cutoff limit set at 50 Ω tissue impedance. The vacuum pump is contained within the radio-frequency generator. When suction is applied, the endometrial lining is brought into contact with the electrode array. The device has an outer diameter of 7.5 mm and consists of a single use, conformable bipolar electrode mesh that can create a confluent lesion involving the entire interior surface area, within the cavity of the uterus. It is suitable for a uterus with a minimum of 2.5 cm cornu-to-cornu distance, and a depth of 6–11 cm as measured by uterine sounding. This distance and the uterine cavity depth are used to set the power delivery level of the radio-frequency generator.

The balloon ablation system (ThermaChoice I Gynecare) consists of a generator and a balloon catheter. The ThermaChoice catheter has an outer diameter of 4 mm, and is inserted through the cervix into the uterine cavity. The latex balloon is filled with dextrose; by increasing the pressure in the balloon it will make contact with the endometrium in the uterine cavity. A thermistor in the balloon is used to raise the temperature of the fluid to 87°C (170°F) for a period of 8 minutes, while keeping the pressure of the balloon 170 mmHg or higher. Endometrial thinning was performed by aspiration curettage prior to the balloon treatment procedure.

The duration of both treatments was recorded from the moment the gynaecologist begun with introduction of a speculum, until the end of the ablation. Patients in both groups were treated in a daycare programme, using either spinal anaesthesia or general anaesthesia.

Follow up visits were carried out at the outpatients clinic at 3, 6 and 12 month intervals after the initial treatment. At these consultations, the patients were seen by a doctor who was unaware of the treatment that had been performed. At each visit, duration of menstruation, presence of dysmenorrhea and presence of clots were registered. Patients also completed a pictorial chart,18 and expressed their satisfaction with the treatment result. Levels of satisfaction were categorised as completely satisfied, satisfied, doubtful satisfied or not satisfied. Furthermore, it was noted whether a reintervention had been performed, or whether a patient had started using oral contraceptives. If a patient had undergone a hysterectomy, satisfaction with the treatment result was categorised as ‘not satisfied’ at all follow up visits.

An increase of the amenorrhea rate from 20% after balloon ablation14 up to 50% after bipolar radio-frequency ablation was assumed. A sample size of 82 patients, 55 in the bipolar group and 27 in the balloon group, would enable detection of a 30% absolute difference in amenorrhea rate, with a one-sided χ2 test at a 5% significance level, and with a power of 80%.

After treatment of 44 patients, a technical failure in the NovaSure generator was discovered. No analysis of results was performed until the reason for the error was established. The precise moment at which the error occurred was not known, although it was clear that during the first five bipolar radio-frequency treatment procedures, the generator had worked properly. Because of this failure, it was decided to extend the number of patients to be included, to 126, 44 in the balloon group and 82 in the bipolar group. It was also decided to perform two analyses. In the first, all patients that had entered the study were analysed, including those who were treated with the defective NovaSure controller. The groups in this analysis are referred to as bipolar group A and balloon group A. In the second analysis, the study group was limited to those patients who were randomised after the failure of the NovaSure generator had been corrected. The groups in this analysis are referred to as bipolar group B and balloon group B.

The analysis was performed according to the ‘intention-to-treat’ principle. In other words, patients were analysed in the group to which they had been allocated. A relative risk for hysterectomy was calculated using Cox regression analysis.

Menstrual pattern and patient satisfaction at 3, 6 and 12 months after initial surgery were compared in the two groups. Repeated measures analysis of variance was used to establish changes in menstrual pattern and patient satisfaction over time (time effect), differences in menstrual pattern and patient satisfaction between both treatment groups (treatment effect) and interaction between changes in menstrual pattern and patient satisfaction over time and treatment group (time by treatment effect).19 Patients with missing measurements were included in the repeated measure analysis if data were available for at least two different time points.20P values less than 0.05 were considered to indicate statistically significant differences. Where a statistically significant difference in menstrual pattern and patient satisfaction between both treatment groups was found, or an interaction between changes in menstrual pattern and patient satisfaction over time and treatment group, the differences between treatment groups at specific points in time were examined. In case of dichotomous endpoints this was done by calculating relative risks and 95% confidence intervals. For the continuous outcome duration of menstruation, we calculate a difference of the medians with a 95% confidence interval, using a bootstrap procedure.

RESULTS

Between 1st November 1999 and 1st July 2001, 126 women were included in the trial. Eighty-three were allocated to the NovaSure group and 43 to the balloon group (Fig. 1). Table 1 shows the baseline characteristics of the two groups, both before and after exclusion of the patients that were randomised before the discovery of a technical failure of the NovaSure generator. There were more patients with a retroverted uterus in the bipolar group (16% vs 9%), but otherwise the two groups were comparable.

Figure 1.

Trial profile.

Table 1.  Baseline characteristics. Bipolar group A and balloon group A consists of all patients that had entered the study, including those who were treated with the defective NovaSure controller. Bipolar group B and balloon group B consists of those patients who were randomised after the failure of the NovaSure generator had been corrected. Values are presented as n (%), mean [SD], median (range), median [min, max].
 Bipolar group A (n= 83)Balloon group A (n= 43)Bipolar group B (n= 55)Balloon group B (n= 26)
Age (years)42.6 [4.9]43.1 [3.8]42.8 [5.3]43.3 [4.2]
Duration of menstruation in days7 (3–20)7 (5–15)7 (3–20)7 (6–15)
Patients with clots76 (92)38 (88)51 (93)24 (92)
Duration of clots (days)3.5 [2.0]3.3 [2.0]3.4 [1.6]3.6 [2.0]
Pictorial chart515 [150, 3401]660 [188, 3220]515 [210, 3401]700 [188, 3220]
 
Dysmenorrhea
Moderate17 (21)13 (30)11 (20)7 (27)
Severe34 (41)16 (37)19 (35)8 (31)
 
Uterus
Anteverted49 (60)33 (77)34 (62)22 (85)
Midposition10 (12)4 (9)4 (7)0
Retroverted13 (16)4 (9)8 (15)2 (8)
Missing10 (12)2 (5)9 (16)2 (8)
 
Haemoglobin (mmol/L)8.0 [0.83]7.9 [0.90]8.0 [0.81]7.9 [1.0]
FSH (IU/L)6.5 [3.5]6.1 [4.3]6.2 [4.6]6.3 [3.6]

One patient that had already been randomised to the NovaSure treatment panicked in the operating room and refrained from the ablative treatment. One year later she had not been treated. Her menstrual blood loss was still heavy but she wanted no further treatment. There were no complications during treatment in both arms of the study. While the mean duration of the NovaSure procedure was 9.0 minutes (range 5–32 minutes), that of the ThermaChoice was 14 minutes (range 9–40 minutes).

Four patients had a hysterectomy in the bipolar group and four in the balloon group (relative risk 0.47, 95% CI 0.07 to −3.3). Two of the four hysterectomies in the bipolar group occurred in the 27 patients that were treated before the failure in the device was detected, whereas the other two occurred among the 55 women that were treated after the failure had been repaired. The reasons for hysterectomy after the bipolar procedure were metrorrhagia (1), menorrhagia (2) and extreme growth of a fibroid (1). The latter patient had an intramural fundal fibroid of 8 cm at the start of treatment. Three months after the bipolar ablation, this fibroid had increased to a size of 15 cm and expulsed in the uterine cavity. Re-evaluation of the patients' history from the patient suffering of metrorrhagia showed an irregular bleeding pattern rather than heavy menstrual loss before the ablation therapy. The two patients with menorrhagia had a uterus depth of 11 and 12 cm. Evaluation by the pathologist showed adenomyosis in both cases.

Apart from the hysterectomies, one patient in the bipolar group was dissatisfied at nine months follow up. A hysteroscopy showed an intrauterine fibroid type one. After hysteroscopic removal the patient was satisfied again, with a pictorial chart score of 58 points, and needed no further treatment.

In the balloon group, all hysterectomies were performed for persisting heavy menstrual bleeding. Pathologic examination showed normal sized uteri in two cases, and a uterus with several small fibroids in two other cases.

Only one patient used oral contraceptives at six months after randomisation. This patient had been treated with balloon ablation. The patient reported to be dissatisfied with the treatment result at six months and one year, but no surgical reinterventions were performed.

Table 2 shows the percentage of women with amenorrhea after bipolar and balloon therapy at follow up. There was a significant decrease of the duration of menstruation after 3, 6 and 12 months in both groups, compared with the duration of menstruation at baseline (P= 0.001). There was also a strong treatment effect: after 3, 6 and 12 months the percentages of patients with amenorrhea in the bipolar group were 33/83 (40%), 36/83 (43%) and 34/83 (41%), respectively, compared with 5/43 (12%), 4/43 (10%) and 3/43 (8%) in the balloon group. There was no interaction between changes in amenorrhea over time and treatment group. After exclusion of the women that were randomised before the failure in the bipolar device was detected, amenorrhea rates in the bipolar B group were even better, 28/55 (52%), 30/55 (55%) and 30/55 (55%) after 3, 6 and 12 months, respectively. Both before and after exclusion of the women that were randomised before the failure in the bipolar device was detected, the amenorrhea rates at specific time points were significantly higher in the bipolar device group (P values at 3, 6 and 12 months all <0.001).

Table 2.  Results. Bipolar group A and balloon group A consists of all patients that had entered the study, including those who were treated with the defective NovaSure controller. Bipolar group B and balloon group B consists of those patients who were randomised after the failure of the NovaSure generator had been corrected.
 Bipolar group A (n= 83)Balloon group A (n= 43)Relative risk (95% confidence interval)Bipolar group B (n= 55)Balloon group B (n= 26)Relative risk (95% confidence interval)
  • *

    Difference in median (95% confidence interval).

Presence of amenorrhea
After 3 months33 (41%)5 (12%)3.4 (1.4 to 8.1)28 (51%)2 (8%)6.6 (1.7 to 26)
After 6 months36 (45%)4 (10%)4.7 (1.8 to 12)30 (55%)2 (8%)7.1 (1.8 to 27)
After 12 months34 (43%)3 (8%)5.9 (1.9 to 18)30 (56%)2 (8%)7.1 (1.8 to 27)
 
Median duration of menstruation (days, min–max)
After 3 months3 (0–18)5 (0–11)2 (1–6)*0 (0–13)6 (0–11)6 (3–7)*
After 6 months3 (0–13)5 (0–14)2 (1–6)*0 (0–10)6 (0–11)6 (3–7)*
After 12 months2 (0–8)5 (0–14)3 (1–6)*0 (0–8)5 (0–11)5 (3–7)*
 
Presence of dysmenorrhea
After 3 months4 (5%)3 (7%)0.69 (0.16 to 3.0)1 (2%)2 (8%)0.24 (0.02 to 2.5)
After 6 months8 (10%)6 (14%)0.69 (0.26 to 1.9)2 (4%)3 (12%)0.32 (0.06 to 1.8)
After 12 months6 (8%)6 (16%)0.52 (0.18 to 1.5)1 (2%)4 (17%)0.12 (0.01 to 1.0)
 
Presence of clots
After 3 months6 (7%)*10 (23%)*0.31 (0.12 to 0.80)2 (4%)9 (35%)0.11 (0.02 to 0.45)
After 6 months4 (5%)*5 (21%)*0.41 (0.12 to 1.5)2 (4%)5 (19%)0.19 (0.04 to 0.91)
After 12 months2 (2%)*6 (14%)*0.17 (0.04 to 0.82)2 (4%)5 (19%)0.19 (0.04 to 0.91)

Table 2 also shows the median duration of menstruation at 3, 6 and 12 months after randomisation. Again, there appeared to be a strong effect over time (P= 0.001), due to the differences between baseline and at 3, 6 and 12 months after randomisation. The treatment also had a significant effect on the duration of menstruation at 3, 6 and 12 months after randomisation, with a difference in the median of two to three days at each point in time. No significant interaction between changes in duration of menstruation over time and treatment group (P= 0.07) was noted. Both before and after exclusion of the women that were randomised before the failure in the bipolar device was detected, the duration of menstruation was significantly shorter in the bipolar device group (Table 2).

Severe dysmenorrhea, present in >30% of the women at baseline, was reduced in both treatment groups (P= 0.001), with bipolar ablation seeming to be more effective in the reduction of dysmenorrhea than balloon ablation. This difference, however, was not significant (P= 0.14). When the analysis was limited to the patients that were randomised after discovery of the failure in the bipolar group, however, the difference was statistically significant (P= 0.001). There was no significant interaction between changes in duration of menstruation over time and treatment group (P= 0.84). Examination of differences between treatment groups showed no significant difference at 3 and 6 months (P= 0.19 and 0.17, respectively), although at 12 months the difference was statistically significant (P= 0.02).

At baseline, almost all women complained of clots. While the number of women reporting these at 3, 6 and 12 months decreased in both groups, the difference was not significant (P= 0.13). There was, however, a noticeable difference between the bipolar group and the balloon group (P= 0.001), but no significant interaction between changes in patients reporting clots over time and treatment group (P= 0.59).

Figure 2 shows the pictorial chart scores at baseline, and at 3, 6 and 12 months after randomisation. There was a strong decrease of the pictorial chart score after treatment (P= 0.001), and in addition, the scores were significantly lower in the bipolar group A than in the balloon group A (P= 0.02). Scores were also significantly better in the bipolar group at each stage in the follow up. There was no interaction between changes in amenorrhea over time and treatment group (P= 0.91).Figure 3 shows the satisfaction with the treatment result at 3, 6 and 12 months after randomisation. The treatment effect was statistically significant (P= 0.003), but the time effect and the interaction between time and treatment group were not. In the bipolar group A, the percentage of women that were not satisfied with the treatment result was 9/83 (11%), 8/83 (10%) and 8/83 (10%), respectively. In the balloon group A, these rates were 9/43 (21%), 11/43 (26%) and 8/43 (19%), respectively. The differences at 3 and 6 months were statistically significant, whereas at 12 months was P= 0.07.

Figure 2.

Median pictorial chart score and range in the bipolar group A (circles) and the balloon group A (squares) at baseline, and after 3, 6 and 12 months. Patients that became amenorrhoeic after treatment are included.

Figure 3.

Patient satisfaction in the bipolar group A and the balloon group A after 3, 6 and 12 months.

After exclusion of patients that were randomised before the NovaSure defect was detected, the treatment effect remained statistically significant (P= 0.003). The percentage of women not satisfied with the treatment result being 2/55 (4%), 4/55 (7%) and 3/55 (6%) in the bipolar group B, against 7/26 (27%), 7/26 (27%) and 6/26 (23%) in the balloon group B (treatment effect P= 0.001; differences at each point statistically significant).

DISCUSSION

We compared the treatment effects of two second-generation endometrial ablation techniques in patients with menorrhagia. The bipolar radio-frequency impedance-controlled endometrial ablation system (NovaSure) performed better than the hot water balloon ablation (ThermaChoice), both in terms of amenorrhea rate as well as in terms of patient satisfaction.

Despite the fact that the balloon ablation therapy was performed using a similar method as in previous studies, we observed lower amenorrhea rates and less patient satisfaction than reported previously (15–20%).14,21 However, in contrast to the earlier prospective observational studies, the present study was a patient- and observer-blinded randomised clinical trial. Because of the lack of blinding in previous studies, the effectiveness of the balloon therapy might have been overstated in the earlier studies.

The baseline characteristics showed that a retroverted uterus was present in 9% of the patients in the balloon group compared with 16% in the bipolar group. In the balloon treatment, when compared with an anteverted uterus, a retroverted uterus is associated with a threefold increase of the risk of treatment failure.17,22 However, because the number of patients with a retroverted uterus was lower in the balloon group, this does not explain the inferiority of the balloon treatment found in the present study.

After inclusion of 44 patients, the failure of the NovaSure generator was detected. In several cases, treatment time was diminished to less than 50 seconds. The problem was caused by the bipolar electrode array of the device, which consisted of a gold-plated porous fabric mesh. This had been broken due to a too rapid increase of the voltage. The power cutoff limit protected against burning. Once the defect was detected, the prototype generator was replaced. No further technical problems have occurred. Due to this failure, we decided to perform two analyses, one in which we applied the ‘intention-to-treat’ principle, and one where patients were excluded that had been randomised before detection of the defect. Although analyses of the group that was treated with the non-defect NovaSure generator showed a better treatment effect than the ‘intention-to-treat’ analyses that included all patients, both analyses demonstrated the superiority of the bipolar device over the balloon device. Thus, we feel that the defect of the NovaSure generator does not affect the conclusions of the study.

Operation time for bipolar radio-frequency endometrial ablation was less than half that of the balloon procedure. This shortens operating room time and may be an important advantage in an outpatient setting, even though a cervical dilation up to 7.5 mm is necessary to perform the NovaSure technique, and no cervical dilation is needed for balloon ablation. Painless ablation under local anaesthesia is hampered due to the nerval innervation of the fundus uteri from thoracal 12, which cannot be blocked by paracervical anaesthesia. The very short ablation procedure, however, makes NovaSure a promising ablation technique for outpatient treatment.

Ablation of the endometrial layer is a low impedance process due to the high level of liquid in endometrial tissue. In this process, endometrium is vaporised and evacuated from the uterine cavity by continuous suction. Once all the endometrium is vaporised and the myometrium is reached, tissue impedance increases quickly. The bipolar radio-frequency technique is not based on time or temperature, but on the impedance (more than 50 Ω) of the coagulated tissue and, as such, on the physical characteristics of the endometrium. This approach enables a well-controlled ablation and ensures consistent results for each patient. Further studies will be needed to investigate the promising effects of this individualised treatment of the endometrium in the long term, compared with the other ablation techniques.

The performance of NovaSure ablations in our study was similar to the results of a randomised controlled trial comparing NovaSure with rollerball and an observational study. In those, the reported amenorrhea rates were 41% in the randomised controlled trial and 58% in the observational study after one year.23,24 Apart from our study, other devices from the second-generation have been evaluated only in two other randomised controlled trials.25,26 Microwave endometrial ablation compared similarly to transcervical resection of the endometrium, with an amenorrhea rate of 40% after microwave endometrial ablation at one year of follow up.25 The Hydro ThermAblator technique performed similarly to rollerball ablation treatment, and showed an amenorrhea rate of 40% after one year of follow up.26 Other second-generation devices, such as endometrial laser intrauterine thermo therapy (ELITT), Cavaterm balloon ablation and cryo ablation have not been evaluated in randomised clinical trials. Observational studies,27–31 however, do not indicate that any of these devices are superior, although the amenorrhea rate of the ELITT27 seems to be promising. In view of these data, and based on the results of the current randomised trial, bipolar radio-frequency endometrial ablation would appear to offer similar or even higher amenorrhea rates, when compared with other second-generation ablation techniques.

Both second-generation ablation techniques used in this study are safe and easy to perform. Patients consulting their gynaecologists with heavy menstrual blood loss should be offered the opportunity for an endometrial ablation. If there is insufficient experience with first-generation ablation techniques, a second-generation technique should be one of the treatment options offered, within the spectrum of medical treatment, the levonorgestrel-containing intrauterine device and hysterectomy.

The present study has shown that bipolar endometrial ablation is a safe and effective treatment in women suffering from menorrhagia. It has been found to be superior to hot fluid balloon ablation, and probably at least as effective as other second-generation devices.

Acknowledgements

The NovaSure devices were provided free of charge by Novacept (Novacept, Palo Alto, California, USA). The ThermaChoice devices were discounted. The study has not been otherwise supported.

Accepted 14 November 2003

Ancillary