Five-year follow up of a randomised controlled trial comparing NovaSure and ThermaChoice endometrial ablation

Authors


Dr MY Bongers, Department of Obstetrics and Gynaecology, Máxima Medical Centre, PO Box 7777, 5500 MB Veldhoven, the Netherlands. Email my.bongers@iae.nl

Abstract

Objective  We have previously reported that NovaSure® was more effective than balloon ablation at 12 months follow up in the treatment of menorrhagia. In this paper, we report the 5-year outcome of this study. The objective was to evaluate amenorrhoea rates, hysterectomy rate, and quality of life associated with the bipolar impedance-controlled endometrial ablation technique (NovaSure) in comparison with balloon ablation technique (ThermaChoice®) at 5 years after administration.

Design  Double-blind randomised controlled trial, 2:1 randomisation NovaSure versus ThermaChoice.

Setting  A teaching hospital with 500 beds in the Netherlands.

Population  A total of 126 premenopausal women suffering from menorrhagia with a pictorial blood loss assessment count ≥150 without intracavitary abnormalities.

Methods  Women were randomly allocated to bipolar radio-frequency ablation and balloon ablation in a 2:1 ratio.

Main outcome measures  The main outcome measures were amenorrhoea rate, hysterectomies, and health-related quality of life (HRQol) as reported at 5 year follow up.

Results  At 5 years of follow up, the total response rate was 96% in the bipolar group and 90% in the balloon group. Amenorrhoea was reported in the bipolar group by 48% of women and in the balloon arm by 32% (relative risk 1.6 [.93–2.6]). There were eight women in the bipolar group (9.8%) and five in the balloon group (12.9%) who had undergone a hysterectomy. Furthermore, there was a significant equal improvement of HRQoL over time in both groups.

Conclusions  At 5 years follow up, bipolar thermal ablation was superior over balloon ablation in the treatment of menorrhagia.

Introduction

One in 20 women suffers from menorrhagia.1,2 Provided that intracavitary abnormalities are absent, several endometrial techniques are available for treatment. Although many ablation techniques have been evaluated both in randomised and nonrandomised clinical trials, follow up is often limited to 1 year or less in most studies. However, women are seeking help because they want a structural solution for their problem. Studies with long-term follow up of women are therefore an essential need before definitive decisions on the preferred treatment can be made.

In 2004, we reported a randomised trial in which we showed that at 12 months after treatment, bipolar endometrial ablation (NovaSure®) is superior to balloon ablation (ThermaChoice®) in the treatment of menorrhagia.3 Women suffering from menorrhagia were randomly allocated to bipolar thermal ablation (bipolar group) or balloon ablation (balloon group) in a 2:1 ratio. A total of 126 women were included in the study, of which 83 were allocated to the bipolar group and 43 to the balloon group. There were no complications in either of the treatment groups. At the 12-month follow-up stage, amenorrhoea rates were 43% (34/83) in the bipolar group and 8% (3/43) in the balloon group (treatment effect in time P value <0.001). At this stage, 90% of the women in the bipolar group were satisfied with the result of the treatment against 79% in the balloon group (treatment effect in time = 0.003).4

As the aim of ablative therapies is to offer the woman a permanent solution, long-term follow up of treatments for menorrhagia is needed, before one can decide on the choice for one of the two treatments. Unfortunately, long-term follow up is lacking in most studies reporting on the effectiveness of endometrial ablation. In this study, we report the 5-year follow-up results of this randomised controlled trial (RCT). The primary outcome measure is amenorrhoea at 5-year follow up. Furthermore, we report hysterectomy rates and health-related quality of life (HRQoL).

Methods

The trial has been described in detail previously.3 In short, a RCT was performed in the Máxima Medical Centre, a teaching hospital with 500 beds in the south of the Netherlands. The ethics committee of the hospital approved the study. Women were eligible if they had a menorrhagia, indicated on a pictorial chart with a Higham score of 150 points or more;4 they had no intracavitary abnormalities; their family was complete; there was no endometrial atypia; and the uterine depth was between 6 and 11 cm. After informed consent, randomisation was performed in the operation room just before start of the procedure. The randomisation sequence was computer generated, and to conceal the allocation, opaque sealed envelopes were used. The ratio between women allocated to bipolar radio-frequency ablation and women allocated to balloon ablation was 2:1.

An increase of the amenorrhoea rate from 20% after balloon ablation5 up to 50% after bipolar radio-frequency ablation was assumed. A sample size of 82 women, 55 in the bipolar group and 27 in the balloon group, would enable detection of a 30% difference in amenorrhoea rate, at a 5% significance level and with a power of 80%. After randomisation of 44 women, a technical failure in the NovaSure generator was detected. Therefore, it was decided to extend the number of women to be included up to 126 (83 allocated to the NovaSure group, 43 to the balloon group).

To obtain insight in the long-term follow up, postal questionnaires were sent at 5 years after the original procedure. The women were kept unaware of the original treatment allocation. A resident dedicated to this study, who was also unaware of treatment allocation, coordinated follow up. All the women were asked to complete the questionnaires. Outcomes assessed were partially identical to those reported at 3, 6, and 12 months follow up; satisfaction with and acceptability of treatment, changes in HRQol (Short-Form 36 [SF-36]), as well as hysterectomy that was performed for persisting menorrhagia.

Subsequent operations were established from the questionnaires and from the patient charts. Women, who did not return the questionnaire, received a single reminder after 1 month. Nonresponse was mostly because of the fact that the women had moved houses.

Analysis

We used analytic methods that have been previously published.6–8 HRQol was studied on an intention-to-treat basis. Repeated-measure analysis of variance was used to establish changes in HRQol over time (time effect), differences in HRQol between both treatment groups (treatment effect), and interaction between changes in HRQol over time and treatment group (time by treatment effect).7 Dichotomous outcomes such as presence of amenorrhoea and presence of dysmenorrhoea were compared by calculating a relative risk (RR). Time to hysterectomy was compared with Kaplan–Meier analysis.

If a statistical difference in HRQol between both treatment groups, or an interaction between changes in HRQol over time and treatment group was found, Student’s t tests were used to examine differences between time groups at specific time points.

After treatment of 44 women, a technical failure in the NovaSure generator was discovered. Once the defect was detected, the prototype generator was replaced. No further technical problems occurred. No analysis of results was performed until the reason for the error was established. Because of this failure, it was decided to perform two analyses. In the first, all women who had entered the study were analysed, including those who were treated with the defective NovaSure controller. In the second analysis, the study group was limited to those women who were randomised after the failure of the NovaSure generator had been corrected.

Results

Between 1 November 1999 and 1 July 2001, 126 women were included in the trial, of which 83 were allocated to the bipolar group and 43 to the balloon group (Figure 1). The baseline characteristics of the two groups were comparable as reported in the 1 year follow-up study.3 The only difference noticed was more women with a retroverted uterus in the bipolar group (16 versus 9%).

Figure 1.

Trial profile. PBLAC, pictorial blood loss assessment count.

One woman who had already been randomised to the NovaSure treatment panicked in the operating room and refrained from the ablative treatment. Five years 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.

There were 67 (80%) women in the bipolar group and 31 (72%) in the balloon group who returned the questionnaires. There were no significant differences in baseline characteristics between women who returned questionnaires and those who did not. Among the 28 women who did not return questionnaires, 22 were contacted by telephone. These 22 women were asked about amenorrhoea, menstruation, dysmenorrhoea, and reintervention. Overall six women, two in the bipolar and four in the balloon group, were lost to follow up (Figure 1).

Table 1 shows the percentage of women with amenorrhoea after bipolar and balloon therapy at follow up. There was a significant decrease of the duration of menstruation after 5 years in both groups, as compared with the duration of menstruation at baseline (= 0.001). The percentages of women with amenorrhoea after 5 years of follow up in the bipolar group were 39/81 (48%), as compared with 12/39 (32%) in the balloon group. After exclusion of the women who were randomised before the failure in the bipolar device was detected, amenorrhoea rates in the bipolar B group were even better, 34/53 (64%) after 5 years. Both before and after exclusion of the women who were randomised before the failure in the bipolar device was detected, the amenorrhoea rates at specific time points were significantly higher in the bipolar device group (P values at 12 months and 5 years both <0.001).

Table 1.  Comparison of bipolar and balloon ablation: results after five year of follow-up
 Bipolar A group, = 81 (%)Balloon A group, = 39 (%)RR (95% CI)Bipolar B group, = 53 (%)Balloon B group, = 24 (%)RR (95% CI)
Presence of amenorrhoea
After 1 year34 (43)3 (8)5.9 (1.9–18)30 (56)2 (8)6.8 (1.8–26)
After 5 years39/81 (48)12/39 (32)1.6 (0.93–2.6)34/53 (64)9/24 (37)1.7 (0.98–3.0)
Median duration of menstruation (days, min–max)
After 1 year2 (0–8)5 (0–14)3 (1–6)0 (0–8)5 (0–11) 
After 5 years0 (0–10)3 (0–12)3 (1–6)0 (0–10)4 (0–11) 
Presence of dysmenorrhoea
After 1 year6 (8)6 (16)0.52 (0.18–1.5)1 (2)4 (17)11 (0.01–0.96)
After 5 years10/68 (14)7/29 (25)0.61 (0.26–1.4)6/45 (13)4/16 (25)0.53 (0.17–1.7)
Presence of clots
After 1 year2 (2)6 (14)0.17 (0.04–0.82)2 (4)5 (19)0.18 (0.04–0.87)
After 5 years6/63 (9)10/29 (32)0.28 (0.11–0.69)3/43 (6)7/18 (33)0.18 (0.05–0.62)

Table 1 also shows the median duration of menstruation at 12 months and 5 years after randomisation. Again, there appeared to be a strong effect over time (= 0.001) because of the differences between baseline and at 12 months and 5 years after randomisation. No significant interaction between changes in duration of menstruation over time and treatment group was noted. Both before and after exclusion of the women who were randomised before the failure in the bipolar device was detected, the duration of menstruation was significantly shorter in the bipolar device group (Table 1).

Severe dysmenorrhoea, present in >30% of the women at baseline, was reduced in both treatment groups (= 0.001), with bipolar ablation seeming to be more effective in the reduction of dysmenorrhoea than balloon ablation. The latter difference, however, was not significant (RR 0.66, 95% CI 0.26–1.4). When the analysis was limited to the women who were randomised after discovery of the failure in the bipolar group, however, the difference was statistically significant (= 0.001). There was no significant interaction between changes in duration of menstruation over time and treatment group (= 0.73).

At baseline, almost all women complained of clots. At 12 months and 5 years, the number of women reporting clots was not significantly lower in the bipolar group (RR 0.28, 95% CI 0.11–0.69). There was, however, a noticeable difference between the bipolar group and the balloon group (= 0.001), but no significant interaction between changes in women reporting clots over time and treatment group (= 0.78).

Eight women had a hysterectomy in the bipolar group and five in the balloon group after 5 years of follow up, resulting in cumulative risks of 9.8% and 12.9% after 5 years, respectively. (hazard rate ratio, 1.2 [95% CI 0.35–4.0]). Three of the eight hysterectomies in the bipolar group occurred in the 27 women who were treated before the failure in the device was detected, whereas the other five occurred among the 55 women who were treated after the failure had been repaired. The life table analysis is shown in Figure 2.

Figure 2.

Kaplan–Meier curves showing the cumulative hysterectomy rate after bipolar and balloon ablation.

Apart from the hysterectomies, one woman in the bipolar group and one in the balloon group were dissatisfied with the treatment result and had a re-ablation with a NovaSure. Both the women were satisfied with this second ablation treatment. Another woman in the bipolar group was dissatisfied at 9 months of follow up. A hysteroscopy showed an intrauterine fibroid type 1. After hysteroscopic removal, the woman was satisfied again, with a pictorial chart score of 58 points, and needed no further treatment. Finally, one woman in the bipolar group had amenorrhoea with cyclic abdominal pain. In this woman, a cervical stenosis was treated hysteroscopically and the pain disappeared thereafter.

Table 2 shows the results of the comparison between HRQol in the bipolar group and in the balloon group. In none of the dimensions, there was a significant difference between the bipolar group and the balloon group, nor was there a significant interaction between time and treatment effect. However, all measures showed a significant improvement of HRQol over time, except for the domain of general health in the SF-36 questionnaire.

Table 2.  Health-related quality of life in women in the bipolar group and in the balloon group
SF-36*TreatmentTime points after randomisationReference valueMultivariate analysis of variance P values
Intake (= 115)1 year (= 78)5 years (= 98)Time effectTreatment effectTime by treatment effect
  • Results are expressed as mean ± SD.

  • *

    Scores could range from 0 to 100, with higher scores indicating better functioning. Reference values are general population means.

Physical functionBipolar group82 ± 1991 ± 1886 ± 22850.040.800.69
Balloon group83 ± 1688 ± 2184 ± 25
Role physicalBipolar group79 ± 3094 ± 2894 ± 17810.040.480.22
Balloon group73 ± 2789 ± 2486 ± 28
Role emotionalBipolar group85 ± 2699 ± 590 ± 2181<0.0010.430.15
Balloon group80 ± 2695 ± 1599 ± 6
Social functioningBipolar group76 ± 1989 ± 1688 ± 1984<0.0010.780.62
Balloon group76 ± 2186 ± 2184 ± 24
Mental healthBipolar group72 ± 1880 ± 1576 ± 1781<0.0010.790.48
Balloon group72 ± 1880 ± 1881 ± 14
Energy/vitalityBipolar group56 ± 1973 ± 1865 ± 2061<0.0010.910.33
Balloon group54 ± 2064 ± 2168 ± 16
PainBipolar group62 ± 2076 ± 2478 ± 2575<0.0010.790.47
Balloon group63 ± 2277 ± 2583 ± 22
General healthBipolar group73 ± 1981 ± 1877 ± 38720.370.850.44
Balloon group76 ± 2175 ± 2371 ± 21

Discussion

We compared the treatment effects of two second-generation endometrial ablation techniques in women with menorrhagia in a follow-up period of 5 years. The bipolar radio-frequency impedance-controlled endometrial ablation system (NovaSure) performed better than the hot-water balloon ablation (ThermaChoice), both in terms of amenorrhoea rate and in hysterectomy rate.

In view of the 1-year follow-up results, the 5-year follow-up results show an increase in amenorrhoea rate for both groups (Table 1). This is partly because of the fact that some women became postmenopausal during the follow-up period. However, even at 5 years after the procedure, the bipolar group performed significantly better than the balloon group. For the short-term follow up, a high amenorrhoea rate implicates a thorough ablation of the endometrial tissue. For the long-term follow up, the satisfaction rate and the hysterectomy rate are more useful to evaluate ablation techniques. The amenorrhoea rate in the long-term follow up will be spoiled by the number of women who became postmenopausal. On the other hand, a high percentage of amenorrhoea after short-term follow up will predict long-term success in terms of low reintervention rate.

After inclusion of 44 women, the failure of the NovaSure generator was detected. Once the defect was detected, the prototype generator was replaced. No further technical problems occurred. Because of this failure, we decided to perform two analyses, one in which we applied the ‘intention-to-treat’ principle and one where women were excluded who 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 women, both analyses showed 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.

We observed lower amenorrhoea rates and less patient satisfaction in the balloon group after 1 year of follow up than reported previously (15–20%). The results after 5 years of follow up are comparable with the results in the literature and in our previous trial.

The performance of bipolar ablation in our study was similar to the results of an RCT comparing bipolar ablation with rollerball ablation.9 In that study, the reported amenorrhoea rates were 41% after 1 year of follow up. An observational study reported a 58% amenorrhoea rate after 1 year.10 Apart from our study, three comparable studies have reported on long-term follow up after endometrial ablation techniques. A prospective, single-arm, observational pilot study of the bipolar ablation system reported after a follow-up period of 3 years an amenorrhoea rate of 65%,11 with a hysterectomy rate of only 3%. However, a prospective observational study lacks the blinding of the patient and observer. The effectiveness of the ablation therapy might have been overstated in this study. Five-year follow up of women randomised to microwave endometrial ablation (MEA) or transcervical resection endometrial ablation (TCRE) for heavy menstrual loss reported a hysterectomy rate of 16% in the MEA and 25% in the TCRE arm.12 In a randomised trial of uterine balloon therapy (UBT) versus rollerball ablation for treatment of menorrhagia, the 5 years of follow-up results reported 42 (17%) hysterectomies (21 UBT, 21 rollerball), 5 repeat ablations (3 UBT, 2 rollerball), and 1 dilation and curettage (rollerball) in a study population of 255 women.13

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 amenorrhoea rates and lower hysterectomy rates, when compared with other second-generation ablation techniques.

The two second-generation ablation techniques that were used in this study are safe and easy to perform. Women 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 in longer follow up is superior to hot fluid balloon ablation in terms of amenorrhoea and hysterectomy rate, and is probably at least as effective as other second-generation devices.

Funding

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

Ancillary