Immediate and delayed add-back hormonal replacement therapy during ultra long GnRH agonist treatment of chronic cyclical pelvic pain


Dr M Al-Azemi, Academic Unit of Reproductive and Developmental Medicine, Level 4, The Jessop Wing, Royal Hallamshire Hospital, Sheffield S10 2SF, UK. Email


Objective  To assess the safety and efficacy of long-term use of long-acting GnRH agonist in women with chronic cyclical pelvic pain using immediate versus delayed add-back hormonal replacement therapy (HRT).

Design  A prospective randomised trial.

Setting  Reproductive and Developmental Medicine, Academic Unit, University Teaching Hospital and NHS Hospitals.

Population  Thirty-eight premenopausal women with chronic cyclical pelvic pain were recruited.

Methods  Women were given Zoladex 10.8 mg over 18 months and randomised to receive HRT (tibolone 2.5 mg) either immediately or after 6 months. Follow up was 12-month post-treatment.

Main outcome measures  Bone mineral density at 6 months, the end of treatment (18 months), and 12 months later, pain and quality of life.

Results  Women treated with immediate HRT add-back showed less bone mineral density loss at 6 months and less vasomotor symptoms compared with those who had delayed HRT add-back treatment. Long-term follow up showed both groups experienced equivalent bone mineral density loss. Pain and health-related quality-of-life assessment showed improvement in both groups but there was evidence of a return to baseline levels after ending treatment.

Conclusion  Long-term use of GnRH agonist plus immediate add-back HRT is a safe and acceptable approach to intractable cyclical pelvic pain. Given the delay in reactivation of the hypothalamo-pituitary-ovarian axis after long-term suppression, an intermittent dose regime with GnRH agonist might warrant investigation.


Chronic cyclical pelvic pain (CCPP) is a common and debilitating condition. The community prevalence of this condition was found to be very high at 24% in women aged 18–49.1,2 It might account for as much as 10% of all outpatient gynaecological consultations and 20% of all laparoscopic procedures. CCPP causes psychological as well as functional disability and significantly impairs a women’s quality of life.3 Endometriosis is found in 71–80% of women with chronic pelvic pain.4 Recent studies suggest that a number of mechanisms exist that may contribute to cyclical pelvic pain, including an influence of ectopic endometrial tissue on the efferent nerve supply from the pelvis.5 Several forms of therapy have been used to treat CCPP associated with endometriosis, including GnRH agonists. The efficacy of the GnRH agonists in the treatment of endometriosis-associated CCPP is due to desensitisation of GnRH receptors, with resultant reduction in serum gonadotropin and ovarian steroid levels, causing a reversible state of hypoestrogenism.6,7 This hypoestrogenic status is associated with several adverse effects. In the short term, women usually experience significant vasomotor symptoms that may result in discontinuation of therapy. Prolonged treatment with these agents is associated with acceleration in loss of bone density. Therefore, the use of these agents has been generally restricted to 6-month exposure because of the concern of long-term effects on bone density which could limit safety and compliance.

The use of add-back therapy in women taking GnRH agonist is aimed to prevent vasomotor symptoms so to improve compliance, and also to protect against significant loss of bone density. The theoretical basis for use of add-back hormonal replacement therapy (HRT) is that endometriosis lesions are dependent on estradiol for continued growth and that end organ tissues exhibit varying sensitivity to estradiol so that bone density can be more or less preserved at serum concentrations of estradiol that are too low to stimulate proliferation of endometriotic lesions. This forms the basis of the estrogen threshold hypothesis, that is, the concentration of estradiol that will partially prevent bone loss may not stimulate endometrial growth. Thus, there is a hierarchy of organ response to estradiol such that bone metabolism is most sensitive while endometriosis is least sensitive. These differences may allow the design of regimens with a GnRH agonist that maintain a therapeutic response and ameliorate potential adverse effects.8

A variety of steroidal and nonsteroidal agents have been employed as ‘add-back therapy’ with variable results in an effort to maintain the therapeutic efficacy of the agonists while suppressing such side effects.9,10 The objective of this study was to compare the safety and efficacy of long-term use of long-acting GnRH agonist in women with CCPP combined with immediate versus delayed add-back HRT on bone mineral density (BMD), pain and health-related quality-of-life (HRQoL) outcomes. In addition, we also aimed to assess the reversibility of bone density loss during a long follow-up period of 12 months after the end of treatment.


The study was conducted in the Academic Unit of Reproductive and Developmental Medicine, University of Sheffield. Participants were recruited from attendees at gynaecology clinics at the participating centres: Royal Hallamshire Hospital Sheffield and Rotherham District General Hospital. We included premenopausal women aged between 21 and 45 years having CCPP of at least 9-month duration and who had had diagnostic laparoscopy or laparoscopic treatment to endometriosis but with recurrence of symptoms within the 3 years preceding the recruitment visit. Recurrence of pain following endometriosis surgery could reflect a combination of factors such as extent of nociception (recurrence of disease), peripheral and central sensitisation, current efficacy of natural modulating mechanisms, and a variety of psychological and social factors. Surgical nerve injury may mimic previous pain experiences leading to incorrect attribution to endometriosis.11 However; these causes of recurrence are less likely to cause cyclical pain, which remains the most common recurrent presentation. All women suffered from pelvic pain that followed their menstrual cycle. In all participants, pain was worst immediately before and during menses, and then subsided to a low or undetectable level for at least 7 days after cessation of menses before recurring again as the next period became due. We deliberately included women with and without visible deposits of endometriosis at laparoscopy as deep endometriosis may not be visualised laparoscopically and such patients form a large part of clinical practice in the context of a CCPP clinic. In our opinion, the cyclicity of pelvic pain is the cardinal sign that the disease process is driven by ovarian steroids and should therefore be amenable to treatment with long-term GnRH agonists. The participants were either willing to employ barrier contraception or using nonhormonal intrauterine contraceptive device for the duration of the study, unless either partner had been sterilised, and were willing to accept postponement of fertility for up to 12 months after the last injection of GnRH agonist. We excluded women who were breast feeding; having GnRH agonist in the previous 6 months; having used sex hormones such as oral contraceptive pills or danazol in the previous 3 months or received any other drug which likely to suppress the hypothalamo-pituitary-ovarian axis. Women known to have metabolic bone disease or who had history of bone fracture in the preceding 12 months were excluded from the study as were women who had uterine fibroids or adenomyosis at time of laparoscopy or on pre-recruitment transvaginal ultrasound. The study was approved by the South Sheffield Ethics Committee and the institutional review board.

A total of 38 women were recruited into this study. After obtaining written consent to the trial, women were initially randomly allocated into two treatment groups. They were randomised using a computer-generated random number table, which was held off site. The research nurse or research fellow undertaking the randomisation contacted the offsite facility by telephone and the participant was allocated to one or other of the groups. Allocation was confirmed in writing. The women were randomised into: group 1: women received tibolone (2.5 mg) as immediate add-back HRT (Tibolone, 2.5 mg; Organon Laboratories, Oss, the Netherlands) with long-acting GnRH agonist (Zoladex depot, 10.8 mg; AstraZeneca, Alderley Edge, UK) whereas group 2 received placebo in the first 6 months followed by the addition of tibolone 2.5 mg (Organon UK Ltd) with long-acting GnRH agonist as a delayed co-administration of add-back HRT. Long-acting GnRH agonist was administered to both groups as subcutaneous injection every 12 weeks. The treatment course was 18 months (72 weeks) and the follow-up period was up to 12 months (48 weeks) post-treatment. A total of 38 women were recruited and 25 women completed the study.

The initial visit included a complete medical, gynaecological, and drug history and included provision of a menstrual diary for recording vaginal blood loss. This was collected at each study visit and validated by the research nurse. Participants also completed structured pelvic pain and quality-of-life assessments at each visit. Throughout the active treatment and follow-up visits, woman’s requirement for analgesia for pelvic pain was recorded on the menstrual diary. Any adverse event was recorded including its severity, duration and its outcome. The participants were seen initially and then 3-monthly for 18 months comprising seven visits. The following visits were 6 and 12 months following the final injection of Zoladex 10.8 mg.

Health-related quality-of-life assessment

Health-related quality-of-life (HRQoL) is a multidimensional concept, which encompasses physical, social and psychological aspects associated with a disease and provides important information on the benefits of medical therapies or interventions from the patient’s perspectives. At present, no disease-specific instrument exists to assess the severity of CCPP and its impact on patient’s lives. Therefore, to assess pain the Endometriosis Health Profile-30 (EHP-30), the Chronic Pain Grade (CPG) questionnaire and the Short-Form McGill Pain Questionnaire (SF-MPQ) were used.

The EHP-30 is currently the only reliable, valid and responsive disease-specific questionnaire measuring the quality of life of women with endometriosis.12–14 It consists of a core questionnaire containing 30 items and five scales: pain, control and powerlessness, emotional well-being, social support and self-image. EHP-30 scale scores are standardised on a range of 0–100. Each scale was calculated as follows: scale score = the total of the raw scores of each item in the scale divided by the maximum possible raw score of all the items in the scale, multiplied by 100. A lower score indicates better health status.

The CPG15 is a simple, validated questionnaire based on measures of pain intensity and pain-related disability. It contains numerical self-rating scales, which can be scored to provide a characteristic pain intensity, disability score and disability points.

The SF-MPQ is a modification of the full MPQ16 and includes three components. The first component is the Pain Rating Index (PRI) in which the patient is shown 15 descriptors (11 sensory and 4 affective). Secondly, the SF-MPQ includes a 6-point Present Pain Intensity index. The participant is asked to select the word that best describes their pain right now. Thirdly, the SF-MPQ includes a 10-cm visual analogue scale (VAS) to measure Total Pain Score. The VAS is completed by the woman, which ranges from ‘No Pain’ to ‘Worst Possible Pain’.

Bone mineral densitometry

Bone densitometry was measured by dual-energy X-ray absorptiometry using a HOLOGIC device (Hologic MDM, Waltham, MA, USA). BMD of the lumbar spine (L2–L4), femoral neck as well as total hip were measured (ISCD recommendation17) at baseline and at 6 and 18 months of treatment as well as 15 months after last GnRH agonist injection. An experienced investigator blinded to the individual women analysed the scans.

Statistical analyses

Using data from the publication concerning the use of tibolone as add-back by Lindsay et al.18, it was calculated that 21 women would be required in each arm to give an 80% power for detection a difference of 2.6% in bone density between the two groups at 5% significance level. The total number needed was 42 women. Over the duration of the study we recruited 38 patients, but the dropout rate was high due to the long treatment course and follow-up period. Funding and logistic factors led to premature discontinuation and analysis of the study.

Statistical analysis was performed with spss software (SPSS for windows; SPSS, Chicago, IL, USA). Student’s t-test for unpaired data and paired t-test for paired data were used as appropriate. Mann–Whitney test was used to compare two independent groups for skewed data. One-way repeated measures of variance (ANOVA) were used for longitudinal comparison. Fisher’s exact test was used for discontinuous variables. < 0.05 was defined as statistically significant. Bonferroni correction was used for multiple pairwise comparisons. All analyses were performed on those who completed the study (per-protocol analysis).

The primary endpoints were the efficacy, assessed objectively by significant reduction in pain scoring system, and the safety, assessed by measurement of BMD. Primary analysis compared data from women who completed the study. We also undertook an intention-to-treat analysis, including 13 women who withdrew during the study. We replaced missing data using method of last observation (LOCF),19 which substitutes the last observed value of the variable.


Of the 25 women who completed the study, 14 received immediate co-administration of add-back HRT with GnRHa (group 1), whereas 11 women were assigned to delayed co-administration of add-back HRT with GnRHa (group 2). Ten women out of 14 in HRT group and ten out of 11 in Placebo group had endometriosis. The two groups were comparable in terms of demographic and clinical characteristics. There were no significant differences among the groups with regard to age, body mass index (BMI), smoking habits and AFS scores (Table 1).

Table 1.   Characteristics of studied patients according to the group
 HRT group (n = 14)Placebo group (n = 11)P-value
Mean ± SDMean ± SD
  1. P-values are generated by *independent t-test, **Mann–Whitney test, ***Fisher’s exact test for the comparison between the two groups.

Age (year)34.8 ± 5.335.9 ± 4.80.589*
Height (m)1.63 ± 0.41.64 ± 0.50.659*
Weight (kg)66.4 ± 13.368.2 ± 10.60.714*
BMI (kg/m2)24.8 ± 4.925.3 ± 3.60.784*
AFS score32.1 ± 30.419.5 ± 14.60.565**
 n (%)n (%) 
Smoking1 (7.1)1 (9.1)0.999***
Previous surgical treatment2 (14.3)1 (9.1)0.999***

Comparison of pain and HRQoL questionnaire scores between baseline and 6, 18 and 30 months for the two groups was made using EHP-30, SF-MPQ and CPG. For the EHP-30, the HRT group showed a significant improvement in all domain scores (with the exception of the emotions domain) from baseline to 6 months and this persisted until 18 months (< 0.05). At 30 months, all domains except social support and self-image significantly improved (< 0.05). The placebo group also showed significant improvement in all domain scores from baseline to 6 and 18 months. However, at 30 months only control and powerlessness remained significantly improved (= 0.012). A similar trend was observed for the SF-MPQ and CPG questionnaire. For the SF-MPQ, the placebo group showed a significant improvement in all domain scores from baseline to 6 and 18 months (< 0.05), except for the total PRI score at 6 months. At 30 months, all scores were returning to baseline levels. However, in the HRT group, all domain scores showed a significant improvement from baseline to 6 and 18 months and lasted until 30 months for the total PRI score. Significant improvements were also observed in the chronic pain intensity and disability scores of the CPG up to 30 months. However, the disability points score was only significant from baseline to 6 and 18 months.

Comparisons of percentage change in questionnaire scores, from baseline, between both groups after 6, 18 and 30 months are shown in Tables 2 and 3 for the EHP-30 and the SF-MPQ and CPG respectively. Independent t-tests revealed only significant differences in percentage changes on the EHP questionnaire. This was for the self-image domain between both groups at 18 and 30 months (< 0.05) and for the emotions domain at 30 months too (< 0.05) (Table 2).

Table 2.   Comparison of percentage change in EHP scores, from baseline, between placebo and HRT groups after 6, 18 and 30 months
EHP domainsPlaceboHRTMean Diff95% CIP-value
  1. A negative change = an improvement.

  2. P-values are for the comparison between group 1 and 2 by independent t-test.

After 6 months
Pain9−72.913−78.05.1−22.8 to 33.10.70
Control and powerlessness9−75.713−58.7−17.0−46.8 to 12.70.28
Emotions9−34.712−37.12.3−50.6 to 55.20.93
Social support8−43.213−18.4−24.8−91.5 to 42.00.45
Self-image9−60.812−37.2−23.6−74.0 to 26.90.34
After 18 months
Pain7−81.213−86.45.2−31.5 to 41.90.76
Control and powerlessness8−85.413−79.7−5.7−23.7 to 12.30.51
Emotions8−54.612−47.5−7.1−43.7 to 29.50.68
Social support8−76.813−68.3−8.4−42.0 to 25.1 0.60
Self-image8−86.313−57.3−29.0−58.5 to 0.40.05
After 30 months
Pain4−22.56−54.832.3−12.1 to 76.80.13
Control and powerlessness5−29.36−59.229.9−14.6 to 74.30.15
Emotions52.56− to 83.20.03
Social support5−25.86−35.39.5−47.6 to 66.70.71
Self-image5−7.16−55.648.510.1 to 86.80.02
Table 3.   Comparison of percentage mean change in SF-McGill (SF-MPQ) and Chronic Pain Grade (CPG) scores, from baseline, between placebo and HRT groups after 6, 18 and 30 months
 PlaceboHRTMean Diff95% CIP-value
  1. A negative change = an improvement.

  2. P-values are for the comparison between group 1 and 2 by independent t-test.

After 6 months
 Total pain rating8−54.310−54.0−0.3−57.2 to 56.61.0
 Visual analogue score7−81.811−73.7−8.1−38.7 to 22.50.58
 Present pain intensity8−81.310−55.0−26.3−75.6 to 23.10.28
 Chronic pain intensity9−61.99−36.925.0−60.6 to 10.70.16
 Disability score9−66.612−50.1−16.5−60.5 to 27.50.44
 Disability points7−85.710−70.0−15.7−57.9 to 26.50.44
After 18 months
 Total pain rating 7−87.410−55.6−31.9−96.1 to 32.40.30
 Visual analogue score6−83.611−31.9−51.7−137.1 to 33.80.21
 Present pain intensity7−71.410−66.7−4.8−59.8 to 50.30.86
 Chronic pain intensity8−86.813−70.2−16.6−39.6 to 6.50.15
 Disability score8−87.913−82.1−5.8−36.2 to 24.50.68
 Disability points6−100.012−94.4−5.6−17.8 to 6.70.34
After 30 months
 Total pain rating3−58.35−72.514.2−137.9 to 166.20.75
 Visual analogue score 3−20.76−28.88.1  −214 to 231.00.93
 Present pain intensity3−50.06−33.3−16.7−123.1 to 89.80.72
 Chronic pain intensity6−30.25−42.312.0−36.4 to 60.40.59
 Disability score6−27.35−58.030.7−35.4 to 96.70.32
 Disability points5−11.74−75.063.3−59.6 to 186.20.26

Using repeated measures ANOVA, the data revealed a significant change in HRQoL over time on all questionnaire domains for both treatment groups (P-value < 0.05). However, an interaction between group and time was only significant for the EHP self-image dimension (= 0.04) (Data not shown).

Table 4 shows mean ± SEM of BMD (g/cm²) as a comparison of baseline versus 6-, 18- and 30-month visits within each group during the follow-up period. There was significant BMD loss at the lumbar spine at 6-, 18- and 30-month visits in both groups. For total hip measurement, there was no significant loss in HRT group whereas there was significant loss at 6- and 18-month visits in placebo group. For femoral neck measurement, there was significant loss only at 6-month visit in HRT group whereas in placebo group there were significant losses at 6- and 18-month visits (Figure 1).

Table 4.   Comparison of baseline bone mineral density (g/cm2) with the same at 6-, 18- and 30-month visit for the two groups and for the total
 Baseline6 MonthsP-value*18 MonthsP-value**30 MonthsP-value***
  1. Values are expressed as mean ± standard error of the mean (SEM).

  2. P-values are generated using paired t-test for the comparison between *baseline and after 6 months, **baseline and after 18 months, ***baseline and after 30 months.

HRT group
Total spine1.098 ± 0.0331.076 ± 0.031<0.0011.046 ± 0.028<0.0011.054 ± 0.0300.001
Total hip0.947 ± 0.0330.943 ± 0.0340.3640.937 ± 0.0340.1350.938 ± 0.0330.615
Femoral neck0.861 ± 0.0260.846 ± 0.0250.0450.836 ± 0.0260.0540.838 ± 0.0250.061
Placebo group
Total spine1.166 ± 0.0291.122 ± 0.0290.0011.110 ± 0.0280.0041.130 ± 0.0320.016
Total hip1.027 ± 0.0281.001 ± 0.0290.0030.996 ± 0.0290.0170.994 ± 0.0210.059
Femoral neck0.934 ± 0.0240.909 ± 0.0250.0400.894 ± 0.0240.0110.903 ± 0.0290.076
Total spine1.128 ± 0.0231.096 ± 0.022<0.0011.074 ± 0.021<0.0011.087 ± 0.023<0.001
Total hip0.982 ± 0.0230.968 ± 0.0230.0050.963 ± 0.0230.0050.962 ± 0.0210.125
Femoral neck0.893 ± 0.0190.874 ± 0.0190.0030.862 ± 0.0190.0010.867 ± 0.0200.008
Figure 1.

 Mean and 95% CI for bone mineral density (g/cm2) between HRT group and placebo group during the follow-up period.

Table 5 shows the comparison of mean ± SEM percentage changes in BMD (%bsl00066BMD) from baseline for both groups during the follow-up period. Percentage change in BMD of the total hip at 6-month visit in HRT group was −0.533 ± 0.591 compared with −2.539 ± 0.661 in placebo group and this difference was statistically significant (= 0.034). However, there was no difference in percentage changes in BMD of the lumbar spine or femoral neck between both groups at the 18- and 30-month visits (Figure 2).

Table 5.   Comparison of percentage change in bone mineral density (g/cm2) between HRT group and placebo group after 6, 18 and 30 months
 HRT groupPlacebo groupP-value
  1. Values are expressed as mean ± standard error of the mean (SEM).

  2. P-values are for the comparison between HRT group and placebo group by independent t-test.

After 6 months
Total spine−1.966 ± 0.383−3.790 ± 0.8680.075
Total hip−0.533 ± 0.591−2.539 ± 0.6610.034
Femoral neck−1.685 ± 0.799−2.631 ± 1.1300.489
After 18 months
Total spine−4.617 ± 0.802−4.730 ± 1.2980.939
Total hip−1.167 ± 0.734−2.990 ± 1.0570.158
Femoral neck−2.780 ± 1.328−4.249 ± 1.4020.458
After 30 months
Total spine−3.865 ± 0.890−3.132 ± 1.0580.599
Total hip−0.783 ± 1.871−2.992 ± 1.3360.372
Femoral neck−2.532 ± 1.323−3.290 ± 1.6850.723
Figure 2.

 Percentage change in bone mineral density (g/cm2) between HRT and placebo groups during the follow-up period.

As there was no difference between the two groups in percentage changes of the three BMD parameters (total lumbar spine, total hip and femoral neck) at 30 months, we calculated the percentage changes of the three BMD parameters for two groups collectively (total 25 patients). Percentage change in BMD total lumbar spine, total hip and femoral neck at 30 months were −3.543 ± 0.671, −1.755 ± 1.200 and −2.866 ± 1.028 respectively.

Using repeated measures ANOVA, the data revealed significant change in total lumbar spine, total hip and femoral neck over study period with P-value <0.001, 0.026 and 0.015 respectively (Data not shown). Interaction of groups over time is not significant, = 0.270 for total spine, = 0.228 for total hip, and = 0.585 for femoral neck.

The analysis of other adverse effects of treatment showed that placebo group was affected by a higher rate of hot flushes (= 0.034) and emotional change (= 0.011). Other adverse effects reported were headache, lethargy, skin reactions and vaginal dryness with no significant differences between the two groups.

Assessment of the time elapsed between GnRHa discontinuation and resumption of menses revealed that the median time was 8 months. Six women remained amenorrhoeic 12 months post-treatment. Four of these patients resumed menses within the subsequent 6 months.

Comparison between those who completed the study (25 women) and those who dropped out (13 women) revealed that the latter group was significantly younger (31.6 ± 5.1 versus 35.3 ± 5.0). Other characteristics were comparable in both groups such as BMI, AFS, smoking habit, previous surgical treatment and stage of endometriosis. Further analysis of drop-out group revealed that three women withdrew following randomisation but before first Zoladex injection (one from placebo and two from HRT group). Of those who received Zoladex, five withdrew after first injection. Two of these women received placebo and withdrew because of peripheral neuralgia and pruritis, and three women received HRT and withdrew because of chest pain, migraine and weight gain. Further women withdrew later in the study because of calf cramps, joint pain or were seeking pregnancy.

Intention-to-treat analysis confirmed that the total study group performed similarly to those who completed the study. There were minor changes in P-value but this had no effect on overall level of significance compared with the per-protocol analysis.


Our study demonstrated a rapid and sustained improvement in level of pelvic pain within 3 months of initiation of GnRH agonist therapy. The effect was not lessened by concomitant immediate HRT. The reduction in pain showed a tendency to reverse at 12-month post-treatment follow up.

This is the first prospective comparative study to use long-acting GnRHa in the treatment of CCPP and/or endometriosis. In this study, we used a long-acting depot formulation (10.8 mg Zoladex) that is administered 3-monthly improving likelihood of compliance.20

The aim of medical treatment for endometriosis is to induce atrophy in the ectopic endometrial tissue with the use of GnRH agonists. A recent Cochrane Review has shown that GnRH agonists are an effective management option for endometriosis-associated pain and their short-term side effects can be ameliorated by the addition of add-back therapy.21 However, long-term use may lead to decreased BMD as a consequence of hypoestrogenism.22 Therefore, the standard duration of medical treatment using GnRHa is usually restricted to 6 months. In an attempt to extend the duration of treatment, different add-back therapies were used including estrogen–progestogen combination and tibolone.

The concept of add-back HRT as a means of preventing hypoestrogenic effects of GnRHa has been explored previously. Gangar et al.23 and Barbieri8 used GnRH agonist and estrogen–progestogen HRT to treat pelvic pain and reported abolition of hypoestrogenic side effects without prominent endometrial stimulation, thus supporting its use in endometriosis. In the current study, tibolone was used as add-back treatment. Tibolone is a synthetic steroid that demonstrates weak estrogenic, androgenic and progestational activity.24 Its established role in postmenopausal symptom control has extended to use in preventing hypoestrogenic complications of GnRHa and the preparation is licensed in UK for this indication.25 Several recent studies have been published investigating the use of HRT during GnRHa treatment in patients with endometriosis26–29 and have shown diminished postmenopausal type symptoms without adversely affecting GnRHa efficacy in treating endometriosis. The results of the present study show that appropriate immediate steroid add-back can sufficiently limit hypoestrogenic side effects to allow for 18 months of continuous GnRH agonist therapy without compromising efficacy, or leading to progressive loss of bone density, while providing satisfactory pain control. These efficacy-related findings are consistent with the above studies.

This is the first study which has properly assessed pain as part of overall HRQoL of women with endometriosis undergoing long-term treatment with GnRHa plus add-back HRT. The assessment involved physical, social and psychological aspects associated with endometriosis. It also evaluated the impact of endometriosis and associated treatments from the women’s perspective.30,31 This was performed systematically using well-validated questionnaires.32,33 Overall, no significant differences between the treatment arms were observed, although the HRT group had slightly better outcomes at 30 months compared with those on placebo. A strong effect of time on all questionnaire domains was found that is both groups did well on treatment but there was evidence of a return to baseline levels following the cessation of treatment. Therefore, this temporary effect should be emphasised during woman’s counselling for such treatment.

The main concern when using long-term GnRHa relates to loss of BMD. Women received a treatment course of 18 months of GnRHa with either immediate or delayed concomitant add-back HRT followed by a follow-up period of 12 months off medication to study the safety of long-term treatment. The results showed that both groups experienced loss of BMD at end of 18-month treatment, which was more pronounced in those who received delayed add-back HRT. Our findings also revealed that there is gradual recovery of bone mass on completion of treatment although we did not observe complete reversal at 15 months after last Zoladex injection. These findings are inconsistent with those of Zupi et al.34 who compared patients receiving GnRH agonist alone with those having add-back therapy in addition. In their study, the treatment duration was 12 months and follow-up period was 6 months. Both groups had significant BMD loss, which was more pronounced in the GnRHa alone group. These changes were still evident at 6-month post-treatment. Similar findings were also reported by other investigators.18,35–37 Hence, it seems likely that the hypoestrogenism induced by long-term use of GnRHa is only partially attenuated by add-back therapy, but that the impact of this treatment on BMD is small and may not be of clinical significance.

Interestingly, our data show stabilisation of BMD after initial fall after commencement of Zoladex. This suggests that long-term GnRHa therapy with addition HRT may result in a resetting of osteoblast and osteoclast activity to a new steady state rather than inducing a progressive bone loss. We did not observe any adverse events that might have been associated with loss of BMD over the long post-treatment follow up in our study. The sustained BMD loss at 12 months post-treatment at lumbar spine was of minor degree. The percentage change in BMD (Table 5) is equivalent to or less than the effect of moderate cigarette smoking on BMD, which has been estimated to be −4% in lumbar spine38 and up to −8% in both lumbar spine and neck of femur.39 Furthermore, the sustained deficit between baseline value and follow-up value should be adjusted to the bone age (30 months = 2.5 years). It has been estimated that there is an annual loss of 1% in BMD in longitudinal studies performed on western Caucasian women indicating that advancing age is associated with bone loss.40

Bone mineral density loss was reported in patients receiving GnRHa in spite the addition of different add-back therapy regimes and this showed only partial recovery after treatment cessation. If such deficit is genuine, one explanation is that higher doses of estrogen might prevent bone loss. However, this might be at the expense of pain control. A study to identify such a threshold level is needed. Another explanation is that estrogen suppresses the expression of osteoclastogenic cytokines and the differentiation of osteoclasts from their precursors.41,42 It was reported that the increase in osteoclastogenesis, which resulted in increase in the number of osteoclasts, was one of the mechanisms of acceleration of bone resorption induced by hypoestrogenism induced by GnRHa. Reduction in osteoclastogenesis after cessation of GnRHa treatment may take time to allow the number of osteoclasts to decrease to pretreatment levels. Matsuo43 reported the differences in the rate of recovery between patients according to different patients’ characteristics including pretreatment deoxypyridinoline level.

Our study also showed that the median time to resumption of menses was 8 months after GnRHa discontinuation with six women remaining amenorrhoeic 12-month post-treatment. Further follow up showed that two women were still amenorrhoeic after 18 months of follow up. These women were aged 42 and 44 years at the commencement of GnRH agonist. It is not likely that long-term GnRH agonist either accelerates or retards the process of primordial follicle depletion and it is possible that these two women either experienced natural menopause or will continue to metabolise the GnRH vehicle and resume ovulation later. Pain relief was also maintained for a similar duration after cessation of GnRHa. This indicates significant prolongation of the therapeutic effect after cessation of GnRHa treatment.

There are several drawbacks to our study. Recruitment was slow, mainly due to the long duration of the protocol, and we failed to achieve the predetermined target number. The drop out rate from both arms was significant, although not untypical over such a long study period. We did not encounter unexpected side effects and most women tolerated GnRHa plus tibolone well. Secondly, other confounding factors affecting the rate of bone recovery such as lifestyle (exercise, alcohol, smoking and caffeine consumption) and genetic factors, which may influence the rate of loss and recovery of BMD. The evaluation of biochemical markers of bone turnover may be useful in future studies.


Long-term use of GnRH agonist plus immediate add-back HRT is a safe and acceptable approach to intractable cyclical pelvic pain. One approach that might minimise risk of significant loss of bone density while maintaining good relief from pain might be to utilise the prolonged post-treatment ‘hangover’ effect of the long-acting GnRH agonists. This would involve treatment duration of 12 months followed by a second 12-month period off treatment, during which BMD might be expected to recover before resumption of menses and pelvic pain. Further courses of treatment with a break for recovery could then be instituted. This suggestion might merit further investigation.

Disclosure of interests

This study was supported by unrestricted educational grant from AstraZeneca.

Contribution to authorship

M. Al-Azemi contributed by performing the statistical analysis, literature review and writing the manuscript. G. Jones contributed in HRQoL part of the manuscript. F. Sirkeci and Mrs M. Houdmont participated in conducting the clinical part of the study. S. Walters participated in the statistical analysis of HRQoL data and Professor W. Ledger participated in the initial conception of the work idea, supervising the clinical work at all stages, reviewing the statistical analysis and editing the drafts of this manuscript.

Details of ethics approval

The study was approved by the South Sheffield Ethics Committee and the institutional review board.


This study was supported by unrestricted educational grant from AstraZeneca. Tibolone and placebo were donated by Organon.


We would like to thank Dr Nicola Peel for providing BMD data, Participating Consultants at Royal Hallamshire Hospital, Sheffield and Rotherham District General Hospital and our patients who participated in this study. We also would like to thank Dr Joseph Gomez from Community Medicine Department at Kuwait University for statistical analysis advice.