Approximately 75% of menopausal women experience hot flushes and other symptoms of the climacteric.1–3 Estrogen is an effective treatment for hot flushes and associated menopausal symptoms.4,5 Recent data, however, indicate that administration of estrogen to menopausal women may be associated with some risk. Both primary and secondary prevention trials6,7 have not only concluded that estrogen is ineffective in preventing heart disease among older menopausal women, but that estrogen increases the risk of coronary and thromboembolic events during the first year of treatment. Moreover, evidence from the Women’s Health Initiative points to an increased risk of breast cancer associated with estrogen use6 and possibly an increased risk of Alzheimer’s disease.8 Although future studies of estrogen in younger, perimenopausal women9 may or may not confirm these estrogen-associated health risks, many women entering menopause are now reluctant to take estrogen. An efficacious and safe alternative is needed.
Gabapentin is a γ-aminobutyric acid analogue approved in 1994 for the treatment of seizures. Gabapentin is thought to have nociceptive properties due to high affinity binding sites located on calcium channels, resulting in alteration of calcium currents. In an uncontrolled case series, Guttuso initially reported that gabapentin appeared to be effective in treating hot flushes in postmenopausal women.10 He later performed a randomized, placebo-controlled, clinical trial11 using 300 mg of gabapentin three times daily followed by an open-label component with autotitration to a maximum of 2,700 mg daily. His data showed a significantly greater reduction in hot flush composite score with gabapentin compared with a placebo (54% versus 31%, P = .01). In a recent larger trial, Pandya et al12 studied 360 women with a history of breast cancer who were experiencing tamoxifen-induced hot flushes. He found a 30% reduction in hot flushes using 100 mg of gabapentin three times daily and a 46% reduction using 300 mg three times daily, compared with an 18% reduction with placebo.
Although these data suggest that gabapentin is effective in the treatment of menopausal hot flushes relative to placebo, its efficacy relative to estrogen is unknown. Therefore, the objective of this study was to assess whether gabapentin is an effective alternative to estrogen, the gold standard, in the treatment of hot flushes and other climacteric symptoms. To achieve this objective, we conducted a randomized, placebo-controlled trial to test three comparisons: gabapentin versus placebo, estrogen versus placebo, and gabapentin versus estrogen.
PARTICIPANTS AND METHODS
Participants were recruited voluntarily through distribution of study flyers, letters addressed to physicians, Web site announcements, and newspaper advertisements. Women who identified themselves as potential participants were screened over the phone and at visit 1 in a standardized fashion. Eligibility for enrollment in the study was based on the inclusion and exclusion criteria shown in Table 1. Hormonal therapy and other medications for the treatment of hot flushes were discontinued at least 1 month before enrolling in the study.
All subject visits, physical examinations and laboratory evaluations were performed in the General Clinical Research Center of the University of Rochester School of Medicine and Dentistry. The University of Rochester Research Subjects Review Board and General Clinical Research Center approved the protocol. All study subjects provided written, informed consent. Given that gabapentin was to be used for an off-label indication, we obtained an Investigational New Drug application (#64008) from the U.S. Food and Drug Administration.
The study design was that of a randomized, double blind, placebo-controlled trial. Figure 1 depicts recruitment, screening, randomization, and enrollment into the three treatment arms: estrogen, gabapentin, and placebo. All study participants were required to make 5 visits. A 2-week baseline hot flush diary and subsequent diary compliance for the 12 weeks after randomization were assessed. Pretrial laboratory studies included serum follicle stimulating hormone (FSH) and estradiol (E2) concentrations, complete blood count, and a chemistry-14 panel. Also, vital signs, height, and weight were assessed. Similar posttrial laboratory studies were obtained, with the exception of serum FSH and E2 concentrations.
Sixty eligible women were recruited, 20 each for the gabapentin, estrogen, and placebo groups. The Office of Investigational Drug Services in the Department of Pharmacy at the University of Rochester performed the randomization, in 5 blocks of 12, via a random number table. Study drugs for each of the treatment arms were formulated, encapsulated, and packaged individually by the Office of Investigational Drug Services.
Hot flush frequency and severity were recorded in a validated hot flush diary.13 Patients were instructed to carry the diary at all times and to mark down the incidence and severity of their hot flushes as they occurred throughout the duration of the day. The hot flush composite score took into account severity as well as frequency. The severity of each hot flush was recorded in the diary on a visual analogue scale. A numerical score, ranging from 1 to 4, was designated for each hot flush based on the level of severity (1 for mild, 2 for moderate, etc). Four days of recorded data within a week were necessary for inclusion into the analysis. The composite score was the weekly summation of the number of hot flushes in each severity category multiplied by the severity score assigned to each hot flush.
The primary outcome measure was the percent change from baseline in hot flush composite scores. Each data point was the composite score for that week divided by the baseline composite score. Secondary outcome measures were other symptoms associated with the climacteric. The Greene Climacteric Scale was used in the study to quantify symptoms and complaints of the climacteric.14 This scale is based on factor analysis of four studies that had attempted to identify facets of the climacteric. Greene’s analysis indicated that climacteric symptoms fall into three independent categories: vasomotor, somatic, and psychological. In his scale, each category is measured independently as three separate subscales.14
The Zung Depression Scale used in this study is one of the most widely used self-evaluation measures of depression.15 Its validity has been tested in comparisons with other standard depression scales that require an interviewer, such as the Hamilton Rating Scale for Depression, the Beck Depression Inventory, and the Minnesota Multiphasic Personality Inventory. These comparisons report a good correlation, with a high positive predictive value, between the Zung Depression Scale as a screening instrument and interviewer-administered assessments of depression.16
All subjects completed the Greene Climacteric Scale and the Zung Depression Scale questionnaires at pre- and postrandomization. The scales were analyzed for group differences at the exit visit, controlling for baseline scores.
All study participants were required to make five visits to the study site: a baseline visit, a visit at randomization, and visits after 4, 8, and 12 weeks of treatment as shown in Figure 1. The treatment code was not revealed until all patients had completed the study. Adverse events were monitored at each visit and during bimonthly phone contacts.
In the initial trial by Guttuso et al,11 subjects in an open-label arm autotitrated to a maximum of 2,700 mg of gabapentin. However, a plateau in efficacy occurred around 1,800–2,400 mg, with no significant improvement in efficacy beyond 2,400 mg. A dosage of 2,400 mg/d was therefore chosen for subjects randomized to the gabapentin arm of this study.
Each gabapentin capsule contained 400 mg of drug. The treatment schedule was titrated to increase the dose of gabapentin gradually over a 12-day period, beginning at a dose of one capsule per night and reaching a final dose of two capsules three times per day.
Subjects in the placebo arm received placebo pills resembling gabapentin and estrogen, while using the same titration schedule for “placebo gabapentin” as that used in the gabapentin arm. Women in the gabapentin arm received gabapentin and a daily “placebo estrogen” pill. Those in the estrogen arm received daily estrogen (conjugated estrogens, Premarin [0.625 mg; Wyeth Pharmaceuticals, Philadelphia, PA] and “placebo gabapentin.”
All randomized study participants who received at least one dose of blinded study drug were included in the analysis. Because there were three a priori comparisons, we set the P value for statistical significance at .05/3 = .017, in accordance with the Bonferonni correction. Each comparison (gabapentin versus estrogen, gabapentin versus placebo, and estrogen versus placebo) was tested by computing t tests of difference scores, subsequent to an analysis of covariance, controlling for baseline differences of climacteric symptoms.
In the case of hot flushes, for which we collected data at additional time points, we performed a 2-factor repeated-measure analysis of variance (ANOVA), where the repeated-measures factor was week of study and the grouping factor was treatment. The groups across all visits were compared using t tests after the analysis of variance. Given the small sample size, the results were also evaluated with Wilcoxon rank-sum tests. Adverse events were evaluated using Fisher exact test for pairwise group comparisons. All statistical analyses were performed with Windows NT workstation and SAS 8.2 software (SAS Institute, Cary NC).17
We estimated our needed sample size based on Guttuso’s data11 showing a 30% reduction in hot flush frequency due to placebo. Across all groups the standard deviation was 3.0, which we used as the common standard deviation for purposes of power analysis. To detect a treatment-associated reduction in hot flushes that was 30% greater than that associated with placebo, at P = .05, we estimated that a sample size of 20 subjects provided 70% power.
Study participants were screened, consented, and enrolled from May 2002 to September 2004 as depicted in the flow diagram (Fig. 2). From the recruiting sources described above, 1,584 calls were received. Of these, 589 callers were screened by means of our designated telephone screening protocol; 106 subjects were consented based on the initial phone screen, and 483 were excluded. Women were excluded if they experienced few or no hot flushes, did not want to take or stop estrogen, were not interested, had menses, or were not qualified for medical reasons. Of the 106 consented subjects, 46 were subsequently excluded from the study because of lack of hot flushes, abnormal laboratory values, and/or medical or physical findings.
Sixty subjects were randomized into the three study groups and were included in the intent-to-treat analysis. Seven study participants withdrew from the study. The reasons were loss to follow-up (did not show up for visit and no contact could be made), noncompliance with completing diaries, and adverse effects. Of the 53 subjects who completed the double-blind study, 17 of 20 (85%) of gabapentin-treated, 17 of 20 (85%) of estrogen-treated, and 19 of 20 (95%) of the placebo-treated groups provided complete data through the 12-week duration of the study.
Baseline characteristics of subjects, summarized in Table 2, were similar among the three treatment arms. At the onset of study, all three groups were balanced in terms of age, body mass index, race, estimated frequency of hot flushes, and laboratory values. During the baseline week, women in the three study groups did not differ significantly in hot flush frequencies, hot flush composite scores, Greene Climacteric, and Zung Depression scores.
Figure 3 depicts a comparison of reduction of hot flush composite scores over a 12-week period between the three treatment arms: placebo, gabapentin, and estrogen. The baseline composite score is plotted at week 0. For weeks 1–12, the hot flush composite score is plotted as a percentage of the baseline hot flush composite score.
A substantial placebo effect was associated with a 54% reduction in the hot flush composite score (Fig. 3). Both estrogen and gabapentin resulted in a reduction in hot flush composite score that was approximately 20% greater than the reduction associated with placebo. Two-factor, repeated-measures ANOVA revealed statistically significant effects for treatment (P = .008) and for the treatment-by-visit interaction (P < .001). Tests of differences in change scores, over the entire time period, between estrogen and placebo (t = 2.50, P = .016) and between gabapentin and placebo (t = 3.03, P = .004) indicated that both estrogen and gabapentin were effective in treating hot flushes. In contrast, there was no difference between gabapentin and estrogen change scores (t = 0.53, P = .63), indicating possible comparability in their reduction of composite hot flush scores.
Wilcoxon rank-sum test results did not change the data interpretation of the parametric test results. All weekly time points were evaluated, and although most visits beyond week 1 showed significant differences for one or both treatment arms when compared with placebo, both treatment groups failed to achieve significant differences at week 12 (estrogen versus placebo, P = .019; gabapentin versus placebo, P = .079). This was not true at week 11, where both estrogen and gabapentin were significantly different from placebo (P = .004 and P = .009, respectively) (Fig. 3).
Graphic presentation of the percentage of the weekly hot flushes that were recorded at the different severity levels, shown in Figure 4, indicates that the percentage of weekly hot flushes in the “very severe” category declined in all groups (including placebo), whereas the percentage in the “mild” category increased.
The least-squares estimate of the group differences on the somatic subscale of the Greene Climacteric Scale, controlling for baseline scores, was the only subscale score to reveal statistically significant differences between gabapentin and the other groups. The difference between gabapentin and placebo was 2.056 (t = 3.01, P = .004), and the difference between gabapentin and estrogen was 1.982 (t = 2.77, P = .008). With the exception of the somatic cluster, no other group differences were found in the other clusters of the Greene Climacteric Scale or Zung Depression Scale.
The pattern of adverse events (Table 3) was similar among the three groups. The Headache, Dizziness, or Disorientation cluster symptoms visually appears with greater frequency in the gabapentin group than the placebo group, but the two-tailed Fisher exact test was not significant (P = .10). Estimation of the number needed to harm in this cluster suggests that these symptoms may occur with every fourth patient treated with gabapentin; thus, this adverse event may be common.
The results of this study confirm previous findings that, in comparison with placebo, gabapentin is effective for the treatment of hot flushes in postmenopausal women. A new finding is that gabapentin appears to be comparable with estrogen in reducing the frequency and severity of hot flushes. Although adverse events were similar in all groups, it remains to be seen whether this will be true in a larger study.
The hot flush composite score, which takes into account both hot flush frequency and severity, appears to be a more useful measure of efficacy than the sole measurement of hot flush frequency. The composite scoring system allowed the opportunity to record either worsened or improved hot flush severity over the course of the study. Because of this feature, we were able to show (despite the small sample size) that the subjective experience of hot flush severity improved among women taking either gabapentin or estrogen. Moreover, the comparable efficacy of estrogen and gabapentin in the reduction of hot flush composite scores is striking. The composite score in the early weeks was reduced in the gabapentin study group to an even greater extent than in the estrogen study group. During screening, subjects reported significantly fewer hot flushes than the number recorded on the hot flush diary, often discounting mild hot flushes, suggesting the importance of taking into account both the perception of the hot flush and the reporting of the hot flush. This improvement in the composite score suggests that both estrogen and gabapentin have a significant effect on the patient’s subjective improvement of hot flushes.
The results of studies of hot flushes that use separate self-report measures of either symptom frequency or severity may be influenced by memory and recall bias. In our study, subjects could carry the diary easily, record the hot flush as it was occurring, and rate its severity all at once. At the onset of our study, we believed that by encouraging diary compliance and following up with weekly phone calls, the placebo effect seen in most hot flush efficacy studies would be mitigated. Instead, a substantial placebo effect was still observed. A recent meta-analysis of all double-blind, randomized, placebo-controlled studies using oral hormonal replacement therapy for the treatment of hot flushes (N = 2,511 participants) revealed a large placebo effect of 50.8%.18 A similar observation was seen in the Women’s Health Initiative, with over 10,000 participants.7
One suggestion for a large placebo effect is that a natural dissipation of symptoms occurs over time. In our study, a majority of women had been suffering from hot flushes for several years with either persistent or worsening hot flushes. Thus, it is unlikely that our study group had dissipation of symptoms through the course of the study solely due of the natural progression of hot flushes. Nevertheless, a longer treatment time may help discern the effect of waxing and waning of hot flush symptoms to help distinguish the placebo effect from the natural attrition of symptoms. This issue becomes one of interest, given the results at visit 12 when compared with those of visit 11. Whether this reflects the natural progression of hot flushes or reflects an artifact of small sample size remains to be seen.
An improved understanding of hot flush triggers and how they mediate the placebo effect is also needed. Some evidence suggests that specific physiologic, endocrinologic, and external factors are related to the occurrence or persistence of the hot flush. Several studies have documented behavioral changes in subjects just with the suggestion of enrolling into an efficacy trial. It is certainly plausible to think that a patient may institute changes that may reduce triggers or external factors that contribute to hot flushes, which may in essence mediate the placebo effect.
The Greene Climacteric Scale is a validated standard measure of core menopausal complaints.14 Our findings suggested that women receiving gabapentin suffered from more somatic complaints than the estrogen and placebo groups. Although we found the incidence of adverse events to be statistically similar for gabapentin and estrogen, there was a greater frequency of adverse events in the Headache, Dizziness, or Disorientation cluster, which supports the significant increase seen in the somatic complaints of the Greene Climacteric Scale. A larger study may be necessary to confirm a difference between gabapentin, estrogen, and placebo in these somatic symptoms. We found that slow titration of gabapentin and taking the medication with meals were effective in minimizing adverse effects.
This study was undertaken to address the need for nonhormonal alternatives to estrogen for the treatment of hot flushes. Despite the small scale of the study, we conclude that gabapentin appears to be an efficacious alternative to estrogen for the treatment of postmenopausal hot flushes.
1. O’Bryant SE, Palav A, McCaffrey RJ. A review of symptoms commonly associated with menopause: implications for clinical neuropsychologists and other health care providers. Neuropsychol Rev 2003;13:145–52.
2. Dennerstein L, Dudley EC, Hopper JL, Guthrie JR, Burger HG. A prospective population-based study of menopausal symptoms. Obstet Gynecol 2000;96:351–8.
3. Groeneveld FP, Bareman FP, Barentsen R, Dokter HJ, Drogendijk AC, Hoes AW. Vasomotor symptoms and well-being in the climacteric years. Maturitas 1996;23:293–9.
4. Vestergaard P, Hermann AP, Stilgren L, Tofteng CL, Sorensen OH, Eiken P, et al. Effects of 5 years of hormonal replacement therapy on menopausal symptoms and blood pressure-a randomized controlled study. Maturitas 2003;46:123–32.
5. Warren MP, Shortle B, Dominguez JE. Use of alternative therapies in menopause. Best Pract Res Clin Obstet Gynaecol 2002;16:411–48.
6. Anderson GL, Prentice RL, LaCroix AZ, Kooperberg C, Stefanick ML, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women’s Health Initiative randomized controlled trial. JAMA 2002;288:321–33.
7. Hulley S, Grady D, Bush T, Furberg C, Herrington D, Riggs B, et al. The Heart and Estrogen/Progestin Replacement Study (HERS) Research Group. Randomized trial of estrogen plus progestin for secondary prevention of coronary heart disease in postmenopausal women. JAMA 1998;280:605–13.
8. Espeland MA, Rapp SR, Shumaker SA, Brunner R, Manson JE, Sherwin BB, et al. The Women’s Health Initiative Memory Study Investigators. Conjugated equine estrogens and global cognitive function in postmenopausal women: Women’s Health Initiative Memory Study. JAMA 2004;291:2959–68.
9. Machens K, Schmidt-Gollwitzer K. Issues to debate on the Women’s Health Initiative (WHI) study. Hormone replacement therapy: an epidemiological dilemma? Hum Reprod 2003;18:1992–9.
10. Guttuso TJ Jr. Gabapentin’s effects on hot flashes and hypothermia. Neurology 2000;54:2161–3.
11. Guttuso TJ Jr, Kurlan R, McDermott MP, Kieburtz K. Gabapentin’s effects on hot flashes in postmenopausal women: a randomized controlled trial. Obstet Gynecol 2003;101:337–45.
12. Pandya KJ, Morrow GR, Roscoe JA, Zhao H, Hickok JT, Pajon E, et al. Gabapentin for hot flashes in 420 women with breast cancer: a randomized double-blind placebo-controlled trial. Lancet 2005;366:818–824.
13. Sloan JA, Loprinzi CL, Novotny PJ, Barton DL, Lavasseur BI, Windschitl H. Methodologic lessons learned from hot flash studies. J Clin Oncol 2001;19:4280–90.
14. Greene JG. Constructing a standard climacteric scale. Maturitas 1998;29:25–31.
15. Zung WW. A self-rating depression scale. Arch Gen Psychiatry 1965;12:63–70.
16. Fountoulakis KN, Iacovides A, Samolis S, Kleanthous S, Kaprinis SG, St. Kaprinis G, et al. Reliability, validity and psychometric properties of the Greek translation of the Zung Depression Rating Scale. BMC Psychiatry 2001;1:6.
17. Moser EB, Saxton AM, Geaghan JP. Biological applications of the SAS system: an overview. Comput Appl Biosci 1988;4:233–8.
18. MacLennan A, Lester S, Moore V. Oral estrogen replacement therapy versus placebo for hot flushes: a systematic review. Climacteric 2001;4:58–74.