ethinyl estradiol for treating premenstrual dysphoric disorder

Contraception 85 (2012) 19 – 27

Original research article

Continuous oral levonorgestrel/ethinyl estradiol for treating premenstrual dysphoric disorder☆,☆☆,★,★★ Uriel Halbreicha,⁎,1 , Ellen W. Freemanb , Andrea J. Rapkinc , Lee S. Cohend , Gary S. Grubbe,1 , Richard Bergeronf , Lynne Smithe , Sebastian Mirkine , Ginger D. Constantinee,1 a

Departments of Psychiatry and Gynecology and Obstetrics, University of Buffalo (SUNY-AB), Buffalo, NY; 14214, USA b Department of Obstetrics and Gynecology and Psychiatry, University of Pennsylvania, Philadelphia, PA; 19104, USA c Department of Obstetrics and Gynecology, David Geffen School of Medicine at University of California­Los Angeles, Los Angeles, CA; 90095, USA d Department of Psychiatry, Harvard Medical School, Boston, MA; 02114, USA e Pfizer, Inc., Collegeville, PA; 19426, USA f Departments of Psychiatry and Cellular and Molecular Medicine, Ottawa Health Research Institute, Ontario, Canada, CAN K1H 8L6 Received 24 March 2011; revised 6 May 2011; accepted 13 May 2011

Abstract Background: The study was conducted to investigate continuous daily levonorgestrel 90 mcg/ethinyl estradiol 20 mcg (LNG/EE) on premenstrual dysphoric disorder (PMDD). Study Design: In this multicenter, randomized, double-blind, placebo-controlled study, women with PMDD received LNG/EE (n=186) or placebo (n=181) daily for 112 days and completed the Daily Record of Severity of Problems (DRSP). Results: Mean DRSP change from baseline to late luteal phase was significantly greater with LNG/EE than placebo at the late luteal phase of the first estimated cycle (−30.52±1.73 [SE] vs. −22.47±1.77; pb.001) and the worst 5 days during the last on-therapy estimated cycle (−26.77±1.83 vs. −20.89±1.82; p=.016). Other primary end points were not statistically significant. Significantly more subject taking LNG/EE (52%) than placebo (40%) responded (≥50% improvement in the DRSP 7-day late luteal phase score and Clinical Global Impression of Severity score of ≥1 improvement) at last on-therapy cycle (p=.025). Conclusions: Continuous daily LNG 90 mcg/EE 20 mcg was well tolerated and may be useful for managing the physical, psychological and behavioral symptoms and loss of work productivity related to PMDD. © 2012 Elsevier Inc. All rights reserved. Keywords: Continuous; Ethinyl estradiol; Levonorgestrel; Oral contraceptive; Premenstrual dysphoric disorder

☆

Clinical trial registration: ClinicalTrials.gov, NCT00128934. Source of financial support: This study was sponsored by Wyeth, which was acquired by Pfizer, Inc., in October 2009. ★ Data presentation: 64th Annual Meeting of the American Society of Reproductive Medicine, San Francisco, CA; November 8–12, 2008. ★★ Financial disclosures: Dr. Halbreich is president and chief executive officer of and holds equity in IN-CLINE RE&D and has received grant support from Corcept, Eli Lilly Labs and Wyeth (acquired by Pfizer, Inc., in October 2009). Dr. Freeman has received research support from Wyeth (acquired by Pfizer, Inc., in October 2009); Pfizer, Inc.; and Xanodyne Pharmaceuticals, and honoraria for consulting and presenting for Wyeth (acquired by Pfizer, Inc., in October 2009), Forest Pharmaceuticals, Pherin Pharmaceuticals and Bayer HealthCare Pharmaceuticals. Dr. Rapkin has received research support from Wyeth (acquired by Pfizer, Inc., in October 2009) and is a consultant to Bayer Health Care Pharmaceuticals. Dr. Cohen has received research support from National Alliance for Research on Schizophrenia and Depression (past); Astra-Zeneca Pharmaceuticals; Bayer HealthCare Pharmaceuticals; Bristol-Myers Squibb; Eli Lilly & Company (past); Forest Laboratories, Inc.; GlaxoSmithKline; National Institute on Aging; National Institutes of Health; National Institute of Mental Health; Ortho-McNeil Janssen; Pfizer, Inc.; Sepracor, Inc. (past); Stanley Foundation (past); and Wyeth (acquired by Pfizer in October 2009) (past). He has been a consultant/advisor for or received honoraria from Eli Lilly & Company (past); Forest Laboratories, Inc. (past); GlaxoSmithKline (past); JDS/Noven Pharmaceuticals; PamLab LLC; Pfizer, Inc. (past); Sepracor, Inc. (past) and Wyeth (acquired by Pfizer in October 2009) (past). Dr. Bergeron has no conflict of interest. Drs. Grubb and Constantine were employees of Wyeth Research (acquired by Pfizer, Inc., in October 2009) at the time of the study. Dr. Mirkin and Ms. Smith are employees of Pfizer, Inc. ⁎ Corresponding author. Biobehavior Program, University of Buffalo School of Medicine and Behavioral Sciences, Hayes Annex C, Ste 1, 3435 Main St. Bldg 5, Buffalo, NY 14214. Tel.: +1 716 829 3811, +1 716 829 3808; fax: +1 716 829 3812. E-mail address: [email protected] (U. Halbreich). 1 Employees of Wyeth Research, which was acquired by Pfizer in October 2009, at the time of the study. ☆☆

0010-7824/$ – see front matter © 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.contraception.2011.05.008

20

U. Halbreich et al. / Contraception 85 (2012) 19–27

1. Introduction Approximately 75% to 80% of women experience some physical and affective symptoms during the late luteal phase of the menstrual cycle [1,2]. Among women of reproductive age, 3% to 8% meet the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSMIV-TR) criteria for premenstrual dysphoric disorder (PMDD), a severe premenstrual syndrome comprising primarily emotional and/or behavioral symptoms that interfere with social or occupational functioning, quality of life and family interaction [3,4]. Lifestyle and nutritional modifications are helpful in some patients [5], but pharmacotherapy is often needed in women with moderate-to-severe symptoms [6,7]. An effective PMDD management strategy is suppressing ovulation-related hormonal cyclicity with combined oral contraceptives (OCs). OCs containing drospirenone plus ethinyl estradiol (EE) significantly improved PMDD symptoms and measures of functioning [6], with a response rate (≥50% decrease in decrease in the daily symptom score) of approximately 60%. Nearly all OC regimens include a hormone-free interval, which causes hormonal fluctuations that can be avoided with a continuous daily OC. Continuous daily levonorgestrel (LNG) 90 mcg/EE 20 mcg has been shown to suppress ovarian activity and eliminate cyclic fluctuations in estradiol, progesterone, luteinizing hormone and follicle-stimulating hormone [8]. Therefore, we hypothesized that this OC would effectively treat PMDD. The objective of this study was to evaluate the safety and efficacy of a low-dose continuous OC containing LNG/EE for PMDD management. 2. Materials and methods A multicenter, randomized, double-blind, placebo-controlled study was conducted at 71 sites in the United States and 14 sites in Canada in accordance with the guidelines of International Conference on Harmonisation Guideline for Good Clinical Practice and the Declaration of Helsinki. The protocol was approved by an institutional review board at each study site before initiation. All subjects gave written informed consent before enrollment. 2.1. Subjects Healthy women aged 18 to 49 years who met DSM-IV-TR [3] criteria for PMDD over the preceding year were screened over two or more 2 menstrual cycles. Subjects had to have regular menstrual cycles and, during the late luteal phase on the Daily Record of Severity of Problems (DRSP) during one screening and one placebo run-in cycle, mild-to-moderate symptoms on ≥4 days, mean late luteal phase DRSP score ≥50 with ≥50% increase from mean follicular phase scores and at least moderate functional impairment on ≥2 days. Mean daily follicular phase scores on the DRSP had to

indicate absent or mild problems during the screening and run-in cycles. Exclusion criteria were the presence or history of thromboembolic or coagulation disorder, cardiovascular or cerebrovascular disease, neuro-ocular disorder, porphyria, otosclerosis, headache with focal neurologic symptoms, cholestasis, liver tumor, uncontrolled diabetes or diabetes with observable vascular involvement, nonmenses-related genital bleeding within 180 days before screening, hypertension, alcoholism or drug abuse within 12 months or major depressive disorder within 3 years, requiring antidepressants, hospitalization or associated with suicide risk. Also excluded were subjects diagnosed with psychiatric or antisocial/borderline/schizotypal personality disorders, elevated liver enzymes or cholesterol, or possible breast malignancy. Subjects aged ≥34 years who smoked ≥15 cigarettes/day were excluded; in Canada, all smokers aged ≥34 years were excluded. Prohibited medications before screening and during the study were depot medroxyprogesterone within 6 months; any sex hormone or gonadotropin-releasing hormone administered via intrauterine device, injection or orally within 60 days; fluoxetine within 30 days; or bromocriptine, methylphenidate, hepatic enzyme-inducing drugs, antidepressants/anxiolytics, sleep medication N3 times/week, and vitamin B6 N50 mg/day, or calcium N1000 mg/day within 10 days. Women with chronic use of oral anti-infectives, antipsychotics, lithium or psychotherapy other than family/ couples therapy were also excluded. 2.2. Treatment After the placebo run-in cycle, subjects were randomized to continuous daily oral LNG 90 mcg/EE 20 mcg tablets or matching placebo for 112 days (4 consecutive 28-day pill packs), taken at the same time each day. Randomization was generated for double-blind treatment using a computerized randomization/enrollment system. Blinding was accomplished with tablet in capsule pill packs labeled with the randomization sequence schedule. At the initiation of double-blind treatment, study medication began on cycle Day 1 (beginning of menses) but not later than Day 5. Subjects were instructed on how to handle missed pills. 2.3. Assessments Because of the expected absence of regular cyclic bleeding with LNG/EE during treatment, actual cycle length and luteal phase days could not be determined. Therefore, each subject's mean cycle length during the screening and run-in cycles was used to estimate cycle length and late luteal phase days for each treatment cycle. Bleeding start dates and DRSP scores were recorded in daily diaries. Subjects completed the DRSP questionnaire [9] daily during screening, placebo run-in, and the four double-blind 28-day pill packs. The DRSP has been validated for daily collection of 21 emotional and physical items within 11

U. Halbreich et al. / Contraception 85 (2012) 19–27

domains of the DSM-IV-TR criteria [3] and 3 items assessing functional impairment. DRSP ratings are 6-point severity scales indicating the degree to which problems are experienced: 1 (not at all), 2 (minimal), 3 (mild), 4 (moderate), 5 (severe) and 6 (extreme). Four primary efficacy end points assessed the difference between LNG/EE and placebo for the mean change in daily DRSP score from baseline to the late luteal phase of the (1) first estimated cycle and (2) the last on-therapy cycle, and the worst 5 symptomatic days during (3) the first estimated cycle and (4) the last on-therapy cycle. The first treatment cycle was a primary end point because of potentially different menstrual bleeding between the groups following the end of treatment cycle 1, possibly jeopardizing blinding. Secondary efficacy end points included percentage of responders defined as ≥50% improvement in the DRSP 7day late luteal phase score and Clinical Global Impression of Severity (CGI-S) score of ≥1 improvement, and remitters defined as responders who no longer met PMDD entry criteria at the last on-therapy cycle. Mean change from baseline in DRSP symptom (depressive symptoms, anger/ irritability and physical symptoms) and impairment (work/ school, social and relationship) clusters, subjects' CGI-S scores, and Work Limitations Questionnaire (WLQ) index scores were also evaluated. The WLQ [10], a 25-item, self-reporting questionnaire, was administered to employed subjects to assess work productivity loss on Days 1 and 12 of screening cycle 2, the placebo run-in cycle and each double-blind estimated cycle. Global evaluations using the CGI-S were completed during screening and once for each double-blind 28-day pill pack. Safety was evaluated via adverse event (AE) reporting (daily diary cards), vital sign and weight assessments (all visits), physical and gynecologic examinations, hematology (hematocrit, hemoglobin, white blood cell count, platelet count) and biochemistry (fasting glucose, total cholesterol, triglycerides, liver function) determinations, and cervical cytology smear (pretreatment and posttreatment). 2.4. Statistical analysis Assuming a mean baseline daily DRSP score of 75 for each group, a decrease of 37.5% with LNG/EE and 25% with placebo and a common standard deviation (SD) of 23, a sample size of 120 subjects/group would provide approximately 90% power to detect between-group differences at an α level of .0498. For the DRSP primary end point analyses, baseline symptom severity was determined for each subject by taking the average of her late luteal phase scores during the screening and placebo run-in cycles. Each subjects' daily DRSP total score and late luteal phase score for each cycle were determined by taking the average of the 5 days, with the highest daily scores within her late luteal 7-day phase. For the worst 5-day analyses, days with the highest DRSP scores from the entire on-therapy cycle was used. Because

21

the estimated cycle length could be shorter or longer than 28 days, subjects who completed the full 112 days of doubleblind treatment could have three to five on-therapy estimated cycles. For primary end points, changes from baseline late luteal phase scores for each subject were calculated for the first double-blind and last on-therapy estimated cycles using both late luteal and worst 5-day DRSP score calculations. Changes from baseline within-group comparisons were calculated using paired t tests. Analysis of covariance (ANCOVA) was used for comparisons between LNG/EE and placebo (factors=treatment and site and covariate=baseline). Each a priori clinically defined symptom cluster score and scores on the three impairment questions were analyzed separately for each double-blind estimated cycle. The changes in average daily score over the five most symptomatic luteal phase days for the items specific to each cluster were compared between treatments using ANCOVA. If any of the cluster items were missing, the data for that day were not used. Effect size was calculated as the difference between groups divided by the SD. The WLQ index score was computed as the weighted sum of the scores from individual WLQ scales (Time, Physical, Mental-Interpersonal and Output). The WLQ scores from Day 1 of the first cycle and from each subsequent doubleblind estimated cycle were compared with baseline scores. Responders and remitters were analyzed comparing baseline with last on-therapy late luteal phase scores, for subjects having baseline and double-blind on-therapy values for the DRSP during the late luteal phase and a CGI score during the same estimates cycle. The proportions of responders and remitters were compared between groups using Fisher's Exact Test. 3. Results 3.1. Subjects Of 4100 subjects screened, 3714 did not meet the studyentry criteria, 386 were randomized to LNG/EE or placebo and 367 subjects received one or more 1 dose of study medication during the double-blind period (Fig. 1). Of the 244 subjects who completed the full 112 days of the doubleblind treatment and completed the DRSP, 87 (36%) had 3 estimated cycles, 154 (63%) had 4 estimated cycles and 3 (1%) had 5 estimated cycles, due to varying cycle lengths. Fifty-nine (31.7%) subjects taking continuous LNG/EE and 53 (29.3%) taking placebo discontinued during the double-blind period. The most common reasons for study discontinuation in the LNG/EE and placebo groups were AEs, subject request and protocol violation (Fig. 1). One (0.5%) and 2 (1.1%) subjects taking LNG/EE and placebo, respectively, discontinued due to unintended pregnancy. Subject characteristics were similar between the 2 treatment groups, except for mean subject age (p=.019; Table 1).

22

U. Halbreich et al. / Contraception 85 (2012) 19–27

Subjects screened (n=4100) Excluded Did not meet DRSP-PMDD criteria (n=1158) Visit outside of schedule (n=636) Unspecified (n=529) Abnormal laboratory values (n=412) Psychiatric history/abnormal screen (n=396) Withdrew consent (n=380) Medical history (n=203)

Randomized (n=386)

LNG/EE (n=195)

Placebo (n=191)

No study drug taken (n=9)

Took ≥1 study drug dose (n=186)

No study drug taken (n=10)

Took ≥1 study drug dose (n=181)

Discontinued

Discontinued

Adverse event (n=4)* Subject request (n=14) Protocol violation (n=14) Lost to follow-up (n=15) Unintended pregnancy (n=2) Investigator request (n=2) Other (n=2)

Adverse event (n=23) Subject request (n=15) Protocol violation (n=12) Lost to follow-up (n=7) Accidental pregnancy (n=1) Investigator request (n=1)

Completed (n=127)†

Completed (n=128)†

Fig. 1. Subject disposition. *pb.001 for LNG/EE versus placebo; †subjects who completed the study had three to five estimated menstrual cycles with corresponding data on the DRSP.

3.2. Efficacy Mean total daily DRSP scores during the late luteal phase and worst 5 days decreased significantly from baseline with both LNG/EE and placebo for all estimated double-blind treatment cycles (p≤.001; Table 2). Adjusted mean DRSP score change from baseline during the late luteal phase was significantly greater with LNG/EE versus placebo for the first treatment cycle (pb.001; effect size, −0.38) and

Table 1 Subject demographics Characteristic

LNG/EE (n=186) Placebo (n=181)

Age (years), mean (range) 36.9 (19–49) Ethnicity, n (%) White 147 (79.0) African American 31 (16.7) Asian, Native American 8 (4.3) Body mass index (kg/m2), mean (SD) 27.7 (6.4) Cigarette smoker, n (%) 29 (15.6)

35.3 (18–49) 153 (84.5) 20 (11.0) 8 (4.4) 27.2 (6.5) 30 (16.6)

approached significance (p=.051; effect size, −0.22) for the last on-treatment cycle (Fig. 2A). For change from baseline using the worst 5 days of the first and last estimated cycles, mean change was significantly greater with LNG/EE than placebo for the last cycle (p=.016; effect size, −0.27), but not for the first cycle (p=.167; effect size, −0.16). In a post hoc analysis of the worst 5 days during the first cycle that excluded cycle Days 2 to 7 (i.e., initial menses days), the decrease from baseline with LNG/EE was significantly greater than that with placebo for this cycle (p=.043; Fig. 2B). Significantly more subjects taking LNG/EE versus placebo were responders at the last on-therapy estimated cycle [82 (52%) and 65 (40%), respectively (p=.025; effect size, 0.04)], but the percentage of subjects in remission with LNG/EE versus placebo [67 (43%) vs 53 (32%)] was similar (p=.065). DRSP cluster scores for depressive symptoms, anger/ irritability and physical symptoms significantly decreased from baseline with both LNG/EE and placebo during all cycles. A significantly greater decrease from baseline with

U. Halbreich et al. / Contraception 85 (2012) 19–27

23

Table 2 Mean change from baseline DRSP scores during the late luteal phase and worst 5 days Estimated cycle

1st 2nd 3rd 4th Last on-therapy

n

Late luteal phase, adjusted mean change (SE)

Worst 5 days, adjusted mean change (SE)

LNG/EE

PBO

LNG/EE

PBO

Diff

Effect size

LNG/EE

PBO

Diff

Effect size

160 143 133 82 161

168 148 132 76 168

−30.53 (1.78)* −36.42 (2.01)† −41.91 (1.97) −46.05 (2.56)† −41.25 (1.73)‡

−22.47 (1.77) −29.78 (2.03) −36.77 (2.03) −39.50 (2.65) −36.75 (1.73)

−8.06 −6.64 −5.14 −6.55 −4.50

−0.382 −0.298 −0.243 −0.336 −0.218

−14.40 (1.59) −23.46 (1.98)† −28.51 (2.00)† −31.17 (2.77)† −26.77 (1.83)§

−11.48 (1.58) −16.27 (2.00) −22.34 (2.06) −22.24 (2.87) −20.89 (1.82)

−2.92 −7.19 −6.17 −8.93 −5.88

−0.155 −0.327 −0.287 −0.423 −0.271

Effect size was calculated by dividing the difference in the adjusted mean change between the groups by the standard deviation (the square root of the mean square error term from the ANCOVA). Subjects completed three to five estimated cycles during the double-blind, 112-day treatment period. Diff, difference; PBO, placebo; SE, standard error. P values based on ANCOVA model: change=site+treatment+average baseline cycle length+baseline value. *pb.001; †p≤.05; ‡p=.051; §p=.016 for LNG/EE versus placebo.

LNG/EE versus placebo was seen for all three clusters at cycle 1 (pb.05). Only the physical symptom cluster was significant between groups at the last on-therapy cycle (p=.038; Fig. 3A). DRSP impairment scores also significantly improved from baseline in both groups at all time points. Significantly greater improvement for continuous LNG/EE over placebo occurred at cycles 1 and 2 in work/school impairment and at cycles 1, 2, and 3 in relationship impairment (pb.05; Fig. 3B). The WLQ index score decreased significantly from Day 1 of baseline to estimated Day 1 within both groups at doubleblind cycles 2, 3, and 4 and the last on-therapy cycle. A greater improvement with continuous LNG/EE compared with placebo was seen for WLQ scores, which was statistically significant at cycle 2 (adjusted change −6.86 vs. −4.65; p=.036) and cycle 4 (−9.78 vs. −7.44; p=.047; Fig. 4). There were no significant differences between continuous LNG/EE and placebo in the CGI-S scores. 3.3. Safety During double-blind treatment cycles, treatment-emergent AEs were reported by more subjects receiving continuous LNG/EE [130 (70%)] than those receiving placebo [108 (60%); p=.049]. Headache was the most commonly reported AE with both treatments. Significantly more subjects in the continuous LNG/EE than placebo group reported vaginal bleeding (Coding Symbols for a Thesaurus of Adverse Reaction Terms [COSTART] terms of metrorrhagia, menorrhagia and vaginal and uterine hemorrhage) and flulike syndrome. There were no other between-group differences in the incidence of individual AEs (Table 3). Two subjects receiving continuous LNG/EE and three receiving placebo had serious AEs. One subject taking placebo developed breast cancer. One subject in the LNG/EE group had a deep vein thrombosis and another subject a pulmonary embolism; these subjects were withdrawn from the study. Study discontinuation due to an AE occurred in 24 (12.9%) subjects in the LNG/EE group and 6 (3.3%) subjects in the placebo group (pb.001). Of these, the most common

AEs in the LNG/EE group were vaginal hemorrhage in 5 (2.7%) subjects, menstrual disorder in 4 (2.2%), menorrhagia in 3 (1.6%) and hypertension in 2 (1.1%). There were no clinically relevant changes from baseline or between groups in serum lipids, fasting glucose, liver function tests, hemoglobin and hematocrit, or white blood cells or platelets during the last on-therapy pill pack, nor were there clinically significant changes from baseline to posttreatment in mean vital signs or body weight. One subject taking active medication had a cervical neoplasm, which was reported as an AE.

4. Discussion The main hypothesis of this study was that a continuous OC would be effective for PMDD treatment, by eliminating the hormonal fluctuations inherent in currently approved treatment modalities. We found that continuous LNG/EE treatment is comparable to previously published results for other PMDD treatments [6,7,11–13]. Specifically, continuous LNG/EE for up to 112 days decreased PMDD symptoms in the late luteal phase as well as on worst symptomatic days, reduced severity of the predefined physical, depressive and irritability symptom clusters and reduced work limitation. Total premenstrual DRSP scores with continuous LNG/ EE compared with placebo significantly decreased from baseline to the first (but not the last) estimated cycle in the luteal phase analysis and during the last on-therapy cycle (but not the first) in the worst 5 day analysis. Because all subjects had menses at the beginning of the first treatment cycle (commencing on Day 1 of menses), and the primary end point worst-5-day analysis included these possibly symptomatic days, it was unlikely that an immediate response to the continuous LNG/EE treatment would be shown. This was illustrated in the post hoc analysis, where Days 2–7 of the first treatment cycle were excluded, and a significantly greater symptom decrease was observed in the worst-5-day analysis with continuous LNG/EE versus placebo in all treatment cycles.

24

U. Halbreich et al. / Contraception 85 (2012) 19–27

A Late Luteal Phase

Mean Change Total DRSP Score

0

-2

-4

-6 LNG/EE Placebo

-8

* -10

* †

-12

-14

Cycle LNG/EE (n) Placebo (n)

1

2

3

4

LOT

160 168

143 148

133 132

82 76

161 168

B Worst 5 Days

Mean Change Total DRSP Score

0 -5 -10 -15 LNG/EE Placebo

-20

* -25

*

-30

*

* *

-35

Cycle LNG/EE (n) Placebo (n)

1

1‡

2

3

4

LOT

160 168

160 168

143 148

133 132

82 76

161 168

Fig. 2. Adjusted mean decrease from baseline in total scores of the DRSP for the late luteal phase (A) and worst 5 days (B). *pb.05 for LNG/EE versus placebo; † p=.05 for LNG/EE versus placebo; ‡post hoc analysis, with Days 2–7 excluded. LOT, last on-therapy.

Both the response rates and effect sizes helped demonstrate the clinically important improvements in PMDD with continuous LNG/EE. Treatment response was achieved in 27% of subjects with continuous LNG/EE during the first cycle and in 52% of subjects during the last estimated cycle. These response rates were defined more stringently than in most clinical studies of approved pharmacotherapies for PMDD

such as selective serotonin reuptake inhibitors (SSRIs) and an OC, for which response rates of approximately 51% to 65% for overall symptoms have been reported, in conjunction with generally high rates of placebo response [6,7,11–13]. In this trial, effect sizes for the decrease in DRSP scores were −0.34 to –0.38 compared with those ranging from −0.26 to −0.67 for symptom improvement in a meta-analysis of SSRI trials [7].

U. Halbreich et al. / Contraception 85 (2012) 19–27

25

A Physical Symptoms 0

Mean Change DRSP Score

-1 -2 -3

LNG/EE Placebo

-4 -5

*

-6

*

-7

*

-8

1

2

3

4

LOT

160 169

143 149

133 133

82 76

161 169

Cycle LNG/EE (n) Placebo (n)

B Relationship Impairment 0

Mean Change DRSP Score

-0.5

-1

LNG/EE Placebo

-1.5

-2

* * *

-2.5

-3

Cycle LNG/EE (n) Placebo (n)

1

2

3

4

LOT

160 169

143 149

133 133

82 76

161 169

Fig. 3. Adjusted mean decrease from baseline in DRSP symptom cluster scores. (A) Physical symptoms and (B) relationship impairment. *pb.05 for LNG/EE versus placebo. LOT, last on-therapy.

A greater proportion of subjects taking LNG/EE reached remission versus those taking placebo, but the difference was not statistically significant. Consistent with previous studies, this protocol emphasized a decrease in the mean total score of symptoms measured by the DRSP but did not emphasize remission or improvement from entry diagnostic criteria as a predetermined primary end point. Furthermore, previous studies do not define/provide a threshold indicative of partial response or remission, nor do they define

response in terms of the subject no longer meeting PMDD diagnosis criteria. To our knowledge, this is the first report of diagnosis-based remission rates, which may become a standard in future studies. Further studies addressing specific symptoms may be helpful in defining patients' clinical characteristics most likely to respond to PMDD therapies. PMDD symptom diversity among individuals and lack of universally accepted diagnosis criteria create variability between

26

U. Halbreich et al. / Contraception 85 (2012) 19–27

0

Mean Change WLQ Score

-2

-4

LNG/EE

-6

Placebo

*

-8

-10

*

-12

Cycle LNG/EE (n) Placebo (n)

1

2

3

4

LOT

114 113

83 85

76 74

67 66

118 118

Fig. 4. Adjusted mean change from baseline in WLQ index scores at cycle Day 1. *pb.05 for LNG/EE versus placebo. LOT, last on-therapy.

results of clinical trials that do not select subjects based on individual symptoms. For example, data suggest that SSRIs may most effectively treat predominantly anxiety/irritability symptoms (up to 90% response in this selected population) [14,15], with less of an effect on physical PMDD symptoms. Study design limitations may also have impacted the study's outcomes, indicating that PMDD trial design needs

Table 3 Number and percentage of subjects reporting treatment-emergent AEs (≥5% of subjects) and those significantly different between treatment groups AEs, n (%) of subjects

LNG/EE (n=186)

Placebo (n=181)

Any Headache Metrorrhagia Nausea Pain Breast pain Back pain Upper respiratory infection Infection Abdominal pain Flu syndrome Menorrhagia Vaginal hemorrhage Arthralgia

130 (69.9)⁎ 30 (16.1) 17 (9.1)† 14 (7.5) 13 (7.0) 13 (7.0) 11 (5.9) 11 (5.9) 10 (5.4) 9 (4.8) 9 (4.8)‡ 9 (4.8)‡ 8 (4.3)§ 5 (2.7)

108 (59.7) 21 (11.6) 2 (1.1) 7 (3.9) 5 (2.8) 6 (3.3) 13 (7.2) 7 (3.9) 14 (7.7) 13 (7.2) 0 0 0 11 (6.1)

Vaginal hemorrhage is the Coding Symbols for a Thesaurus of Adverse Reaction Terms (COSTART) dictionary term for vaginal bleeding. ⁎ pb.05 versus placebo. † pb.001 versus placebo. ‡ pb.005 versus placebo. § pb.01 versus placebo.

to be reassessed. A high number of subjects failed screening; however, this is typical of PMDD studies, likely because most women who believe they have PMS or PMDD do not actually meet the diagnostic criteria with prospective monitoring. Similar to limitations of other PMDD studies, subject retention could have been challenging with the dedication required to complete a daily-rating questionnaire for 7 months. In addition, a high placebo response is well known in clinical trials of PMDD, and the placebo run-in cycle only partially addressed this issue. The safety profile of continuous LNG/EE here was consistent with the safety in large Phase 3 studies of 2457 subjects taking continuous LNG/EE for up to 1 year [16,17]. Except for vaginal bleeding, this study demonstrates a safety profile with continuous LNG/EE comparable with those with cyclic regimens. This was directly demonstrated in a previous report of continuous LNG/EE versus 28-day cyclic LNG/EE [16]. In conclusion, findings of this randomized, prospective clinical trial indicate that a continuous OC containing LNG/EE is safe and may be effective in decreasing symptoms of PMDD. Acknowledgments The authors acknowledge the assistance of professional medical writers, Naomi Plisko, MD, and Kathleen Ohleth, PhD, in the preparation of this manuscript; they revised the manuscript under the supervision of the authors and assisted with journal submission. Support for medical writers was received from Wyeth, which was acquired by Pfizer, Inc., in October 2009.

U. Halbreich et al. / Contraception 85 (2012) 19–27

References [1] Hylan TR, Sundell K, Judge R. The impact of premenstrual symptomatology on functioning and treatment-seeking behavior: experience from the United States, United Kingdom, and France. J Womens Health Gend Based Med 1999;8:1052. [2] Johnson SR, McChesney C, Bean JA. Epidemiology of premenstrual symptoms in a nonclinical sample. I. Prevalence, natural history and help-seeking behavior. J Reprod Med 1988;33:340–6. [3] American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders; 2000. p. 771–4. [4] Yang M, Wallenstein G, Hagan M, Guo A, Chang J, Kornstein S. Burden of premenstrual dysphoric disorder on health-related quality of life. J Womens Health (Larchmt) 2008;17:113–21. [5] Kroll R, Rapkin AJ. Treatment of premenstrual disorders. J Reprod Med 2006;51:359–70. [6] Lopez LM, Kaptein AA, Helmerhorst FM. Oral contraceptives containing drospirenone for premenstrual syndrome. Cochrane Database Syst Rev 2009;CD006586. [7] Brown J, O'Brien PM, Marjoribanks J, Wyatt K. Selective serotonin reuptake inhibitors for premenstrual syndrome. Cochrane Database Syst Rev 2009;CD001396. [8] Archer DF, Kovalevsky G, Ballagh SA, Grubb G. Ovarian activity and safety of a novel levonorgestrel/ethinyl estradiol continuous oral contraceptive regimen. Contraception 2009;80:245–53. [9] Endicott J, Nee J, Harrison W. Daily Record of Severity of Problems (DRSP): reliability and validity. Arch Womens Ment Health 2006;9: 41–9.

27

[10] Lerner D, Rogers WH, Chang H. The work limitations questionnaire. Qual Life Newsl 2002;28:9–10. [11] Halbreich U. Selective serotonin reuptake inhibitors and initial oral contraceptives for treatment of PMDD: effective but not enough. CNS Spectr 2008;13:293–300. [12] Landen M, Eriksson O, Sundblad C, Andersch B, Naessen T, Eriksson E. Compounds with affinity for serotonergic receptors in the treatment of premenstrual dysphoria: a comparison of buspirone, nefazodone and placebo. Psychopharmacology (Berl) 2001;155:292–8. [13] Halbreich U, Bergeron R, Yonkers KA, Freeman E, Stout AL, Cohen L. Efficacy of intermittent, luteal phase sertraline treatment of premenstrual dysphoric disorder. Obstet Gynecol 2002;100:1219–29. [14] Halbreich U, O'Brien PM, Eriksson E, Backstrom T, Yonkers KA, Freeman EW. Are there differential symptom profiles that improve in response to different pharmacological treatments of premenstrual syndrome/premenstrual dysphoric disorder? CNS Drugs 2006;20: 523–47. [15] Landen M, Nissbrandt H, Allgulander C, Sorvik K, Ysander C, Eriksson E. Placebo-controlled trial comparing intermittent and continuous paroxetine in premenstrual dysphoric disorder. Neuropsychopharmacology 2007;32:153–61. [16] Teichmann A, Apter D, Emerich J, et al. Continuous, daily levonorgestrel/ethinyl estradiol vs. 21-day, cyclic levonorgestrel/ ethinyl estradiol: efficacy, safety and bleeding in a randomized, open-label trial. Contraception 2009;80:504–11. [17] Archer DF, Jensen JT, Johnson JV, Borisute H, Grubb GS, Constantine GD. Evaluation of a continuous regimen of levonorgestrel/ethinyl estradiol: Phase 3 study results. Contraception 2006;74:439–45.