Combined Oral Contraceptives in Women with Systemic Lupus Erythematosus

Study Design

Investigators from the core SELENA group³ met before the recruitment of patients to revise the definitions of the instrument used, the Systemic Lupus Erythematosus Disease Activity Index (SLEDAI).⁹ The SELENA-SLEDAI flare composite^8,10 included three elements: the SELENA-SLEDAI score (range, 0 to 105, with 0 indicating inactive disease); an assessment of new or worsening disease activity, medication changes, and hospitalizations not captured with the use of the SLEDAI; and the score on the physician's global-assessment visual-analogue scale (range, 0 to 3, with 0 indicating inactive disease and 3 severe disease) (for details, see the Supplementary Appendix, available with the full text of this article at www.nejm.org). The physicians at each study site scored each element of the flare composite. Mild or moderate flares and severe flares were defined exactly as described in the recently published hormone-replacement study of the SELENA trial.⁸ Whether these definitions could identify flares accurately and uniformly was tested before enrollment with the use of scenarios involving subjects from the Hopkins Lupus Cohort.¹¹ In 2001, all site investigators completed this test with the use of case-report forms of SELENA subjects, which showed a high intraclass correlation (0.89).

Patient Population

A total of 183 women with systemic lupus erythematosus were enrolled from June 1997 through July 2002 (with follow-up through July 2003) at 15 U.S. sites (see the Supplementary Appendix). Institutional review boards at all sites approved the protocol and consent forms, and written informed consent was obtained from subjects before enrollment. Throughout the study, each site reported adverse events to its local institutional review board. In addition, all serious adverse events in the overall study were reported to all institutional review boards.

At enrollment, subjects fulfilled at least four of the criteria of the American College of Rheumatology for the classification of systemic lupus erythematosus¹² and were under 40 years of age if they were nonsmokers or under 36 years of age if they were smokers. All subjects had disease that was clinically stable or had improved in the previous three months. Subjects were stratified as having either inactive disease (defined as a SELENA-SLEDAI score of 4 or less and the receipt of a daily dose of prednisone of no more than 0.5 mg per kilogram of body weight that had not been increased in the previous three weeks) or stable active disease (defined as a SELENA-SLEDAI score of 5 to 12 and the receipt of a daily dose of prednisone of no more than 0.5 mg per kilogram that had not been increased in the previous three weeks). Additional immunosuppressive drugs at enrollment were permitted if the dose had been stable for the previous two months. Exclusion criteria were the use of oral contraceptives for more than one month after the diagnosis of systemic lupus erythematosus; a diastolic blood pressure of more than 95 mm Hg or a systolic blood pressure of more than 145 mm Hg on three determinations; a history of spontaneous deep venous thrombosis, arterial thrombosis, or pulmonary embolus; the presence of IgG, IgM, or IgA anticardiolipin antibodies (more than 40 IgG phospholipid units, 40 IgM phospholipid units, or 50 IgA phospholipid units), a demonstration of lupus anticoagulant by the dilute Russell's viper-venom time test, or both; a history of gynecologic or breast cancer; a history of myocardial infarction; hepatic dysfunction or tumors of the liver; uncontrolled diabetes; congenital hyperlipidemia; migraines associated with neurologic sequelae; unexplained vaginal bleeding; or a positive pregnancy test.

Potential subjects were informed orally by the recruiting physician and in the written consent form that they might receive either oral contraceptives or identical placebo pills. To enter the study, all subjects had to agree to use an alternative form of birth control throughout the study, and any subject who became pregnant was instructed to discontinue the study drug.

Randomization and Treatment

Subjects were randomly assigned in a ratio of 1:1 to receive oral contraceptives (triphasic ethinyl estradiol plus norethindrone) (Ortho-Novum 7/7/7, Ortho-McNeil) in four-week cycles (seven days of 35 μg of ethinyl estradiol plus 0.5 mg of norethindrone, seven days of 35 μg of ethinyl estradiol plus 0.75 mg of norethindrone, seven days of 35 μg of ethinyl estradiol plus 1.0 mg of norethindrone, and seven days of inert pills) or identical placebo (Ortho-McNeil). The randomization scheme was stratified according to study site and disease severity (stable active vs. inactive). Permuted blocks of variable size (2 through 8) were used within each stratum for treatment assignments. Study drug was taken for 12 cycles of 28 days each (“months”). The study drug (Ortho-Novum 7/7/7) and matching placebo were provided free of charge by the manufacturer, Ortho-McNeil; the company had no role in the initiation, planning, conduct, data assembly, analysis, or interpretation of the study.

At screening, a history was taken and physical examination performed. A gynecologic examination and Papanicolaou smear were required within the year preceding enrollment. Laboratory tests included complete blood count, metabolic panel, lipid profile, urinalysis, and 24-hour urine collection to measure levels of creatinine and protein excretion. Serologic profiles included antinuclear antibodies assayed by HEp-2 cell line, anti–double-stranded DNA (anti-dsDNA) antibodies by enzyme-linked immunosorbent assay (ELISA, Diamedix), and C3 and C4 by nephelometry (Dade Behring) at the Hospital for Joint Diseases in New York; lupus anticoagulant by dilute Russell's viper-venom time test with confirmatory mixing studies at Johns Hopkins University in Baltimore; and anticardiolipin antibodies by ELISA (standard β₂-glycoprotein I–dependent assay) at the Hospital for Special Surgery in New York. Subjects were seen at screening and qualifying visits, contacted 2 weeks after entry, and then seen at 1, 2, 3, 6, 9, and 12 months.

End Points

The primary end point was the occurrence of a severe flare. Secondary end points were the occurrence of mild or moderate flares and score on the SELENA-SLEDAI instrument. Pregnancy was not considered as a statistical end point or outcome measure but, rather, constituted a discontinuation criterion. In this noninferiority trial, our design criteria were to demonstrate that oral contraceptives did not increase the risk of the occurrence of a severe flare by more than a prespecified maximum margin that was clinically acceptable, as compared with placebo. We expected a 12-month severe-flare rate of 6 percent in the placebo group. Taking into consideration the potential benefits of oral contraceptives, the SELENA investigators deemed a priori that an absolute difference of less than 9 percent in severe-flare rates between the groups receiving oral contraceptives and placebo would be clinically acceptable.

Statistical Analysis

The criterion for the establishment of the safety of oral contraceptives was that the upper limit of the one-sided 95 percent confidence interval for the difference in severe-flare rates between the groups had to be less than 9 percent. It was determined that a sample size of 350 subjects would yield 95 percent power at a one-sided type I error rate of 0.05 to conclude that oral contraceptives are not inferior to placebo, assuming a noninferiority margin of 9 percent and a 6 percent severe-flare rate in both treatment groups. However, the SELENA data and safety monitoring board elected to close enrollment after 183 patients had been randomly assigned, because of difficulty in recruitment.

Distributions of time to the first occurrence of a severe flare were estimated with the use of the Kaplan–Meier method. The difference in 12-month severe-flare rates between treatment groups was computed from the difference in the corresponding Kaplan–Meier estimates.

The Cox proportional-hazards model was used to obtain estimates of the relative risk of severe flare. Times to first occurrences of mild or moderate flares and flares of any type were analyzed with similar approaches. The SELENA-SLEDAI instrument score was analyzed by computing the change from baseline at each follow-up visit and comparing the size of the changes between treatment groups with the use of the two-sample t-test. Linear mixed-effects models were also fitted to the repeated measures.

Analyses were based on the intention-to-treat principle. Data on subjects who were lost to follow-up were considered to be censored at the time of the subjects' last visit. A per-protocol analysis that was limited to subjects who were fully adherent to the treatment protocol was also performed to evaluate the primary end point. P values are two-sided and are based on the standard null hypothesis of no treatment difference.

Enrollment

Figure 1. Figure 1. Study Subjects.

Seven subjects who were assigned to receive oral contraceptives (OCs) had a severe flare. Of those subjects, five discontinued the study drug due to severe flare, one had a flare after randomization but before taking the drug, and one had a flare after discontinuing the drug (voluntary stop). Subjects who were lost to follow-up are a subgroup of the subjects who discontinued drugs. All subjects were included in the intention-to-treat analysis. A per-protocol analysis to evaluate the primary end point (the occurrence of a severe flare) was also performed; it included only subjects who completed 12 months of medication or who stopped medication because of the occurrence of a severe flare and completed 12 months of follow-up (57 in the oral-contraceptive group and 58 in the placebo group).

Table 1. Table 1. Baseline Characteristics of the Study Subjects According to Treatment Group.

A total of 91 subjects were randomly assigned to receive oral contraceptives, and 92 subjects to receive placebo. Figure 1 is a flow chart of the study subjects; Table 1 summarizes the subjects' clinical characteristics at enrollment. Racial and ethnic backgrounds, which were determined by the subjects, were similar for subjects in the oral-contraceptive group and the placebo group, respectively: white, 41 percent and 33 percent; black, 28 percent and 38 percent; Asian, 13 percent and 14 percent; Hispanic, 18 percent and 14 percent; and other groups, 1 percent and 1 percent.

The 12-month nonadherence rate was 37.4 percent for the oral-contraceptive group and 37.0 percent for the placebo group (P=0.95). Subjects who terminated the study drug early for any reason other than the occurrence of a severe flare were considered to be nonadherent. The 12-month rate of loss to follow-up (i.e., nonadherent subjects who dropped out completely and were not followed for the full 12 months) was 14 percent for the oral-contraceptive group and 20 percent for the placebo group (P=0.34).

Outcomes

Table 2. Table 2. Clinical Manifestations of Severe Flares According to Treatment Group.

Figure 2. Figure 2. Kaplan–Meier Estimates of the Cumulative Probability of a Severe Flare in Subjects Receiving Oral Contraceptives and Placebo.

The difference between treatment groups in the 12-month rate of occurrence of severe flare was –0.0028 (P=0.95).

The occurrence of a severe flare was infrequent in both groups: 7 of 91 subjects in the oral-contraceptive group (7.7 percent) and 7 of 92 subjects in the placebo group (7.6 percent) (Table 2). Two of the severe flares in the oral-contraceptive group took place when subjects were not taking the study drug (one subject had a flare after qualifying for the study but before taking the drug, and the other had a flare four months after voluntarily stopping the drug). The 12-month severe-flare rate estimated from the Kaplan–Meier approach was 0.084 for the oral-contraceptive group (95 percent confidence interval, 0.024 to 0.14) and 0.087 for the placebo group (95 percent confidence interval, 0.025 to 0.15), which was a difference of –0.0028 between the groups (P=0.95). The upper limit of the one-sided 95 percent confidence interval for the true difference was 0.069, implying that the data are consistent with an absolute difference in severe-flare rates of up to 6.9 percent, less than the prespecified equivalence margin of 9 percent. A Kaplan–Meier estimate for the cumulative probability of the occurrence of severe flare is provided in Figure 2. As calculated by the Cox proportional-hazards model, the estimated relative risk of the occurrence of severe flare while receiving oral contraceptives, as compared with placebo, was 0.93 (P=0.89; 95 percent confidence interval, 0.33 to 2.65). Although the sample size was reduced to 183 subjects, we were able to demonstrate that the difference in the risk of severe flare between the oral-contraceptive group and the placebo group was less than the maximum clinically acceptable difference. Furthermore, this finding persisted even with the hypothetical calculation of one or two additional severe flares in the oral-contraceptive group (data not shown).

The per-protocol analysis, which included 57 subjects receiving oral contraceptives and 58 receiving placebo, yielded an estimated 12-month rate of severe flares of 0.088 for oral contraceptives (95 percent confidence interval, 0.014 to 0.16) and 0.12 for placebo (95 percent confidence interval, 0.037 to 0.20). The difference in the 12-month rate of severe flares between groups was –0.033 (P=0.56). The upper limit of the one-sided 95 percent confidence interval for the true difference was 0.060, less than the prespecified margin of 9 percent and therefore consistent with the intention-to-treat result that oral contraceptives are not inferior to placebo with respect to risk of a severe flare.

In both study groups, subjects who entered the trial with stable active disease were at increased risk for severe flare as compared with those with inactive disease. The estimated relative risk, adjusted for treatment, was 3.50 (95 percent confidence interval, 1.23 to 9.99; P=0.02).

Rates of mild or moderate flares did not differ significantly between groups: 63 subjects randomized to receive oral contraceptives (69 percent) and 55 subjects randomized to receive placebo (60 percent) had one or more mild or moderate flares. The incidence rate for mild or moderate flares, which included multiple flares in the same subject, was 1.40 flares per person-year for the oral-contraceptive group and 1.44 flares per person-year for the placebo group (relative risk, 0.98; 95 percent confidence interval, 0.76 to 1.26; P=0.86).

The probability of having at least one flare of any type during the 12-month follow-up period was 76 percent for subjects receiving oral contraceptives and 69 percent for subjects receiving placebo, a treatment difference of 0.07 (P=0.34; upper limit of the one-sided 95 percent confidence interval, 0.19). The estimated relative risk in the oral-contraceptive group by the Cox proportional-hazards model based on the time to the first flare of any type was 1.09 (P=0.65; 95 percent confidence interval, 0.76 to 1.55).

Table 3. Table 3. Mean Change in the SELENA-SLEDAI Instrument Score from Baseline, According to Treatment Group and Month of Follow-up.

The mean change in SELENA-SLEDAI scores was not significantly different between groups at any follow-up visit (Table 3). Linear mixed-effects models fitted to these repeated measures found no significant effect due to treatment group (P=0.94).

Adverse Events

Table 4. Table 4. Additional Adverse Events According to Treatment Group.

Serious adverse events and adverse events requiring the discontinuation of study medications are listed in Table 4. There were four pregnancies. One subject in the placebo group had a positive pregnancy test after randomization but before receiving placebo and delivered a healthy infant at term. Two subjects stopped receiving the study drug after positive pregnancy tests: one subject stopped receiving oral contraceptives at the 2-month visit and later had a miscarriage of twins at 19 weeks' gestation, and one subject stopped receiving placebo at the 3-month visit and elected to terminate the pregnancy at 16 weeks' gestation. An ectopic pregnancy occurred in one subject two months after oral contraceptives had been discontinued owing to a rash.

Effective and safe birth control is essential to the care of premenopausal women with systemic lupus erythematosus. Hormone-based contraception remains the most effective reversible form of birth control and is the nonsurgical method used by most women (both whites and blacks) between the ages of 15 and 44 years in the United States.¹³ However, available data affirm that women with systemic lupus erythematosus use oral contraceptives less often than do age-matched healthy controls,³ reflecting prescribing practices. Yet, the impression that exogenous estrogens may negatively influence lupus disease activity is not derived from any reproducible direct evidence. The SELENA trial prospectively evaluated the effect of oral contraceptives on lupus disease activity. Although the sample was smaller than was initially planned, the trial demonstrated that contraception with the use of estrogen and progesterone does not increase a patient's risk of severe or mild–moderate flare beyond a clinically acceptable margin. The nonadherence rate was consistent with rates among first-time users of oral contraceptives in the general population.^13,14 In a similar manner, the pregnancy rate of 1 percent in subjects who were receiving oral contraceptives is consistent with national averages.¹³

Current decision making about the use of oral contraceptives in women with systemic lupus erythematosus is based on lupus models in mice and on anecdotal reports and retrospective case–control studies in humans. Among 121,645 women in the Nurses' Health Study, past users of oral contraceptives had a small increase in the risk that systemic lupus erythematosus would develop (relative risk, 1.4; 95 percent confidence interval, 0.9 to 2.1), as compared with those who had never used oral contraceptives.¹⁵ A recent case–control study showed a weak association between the risk of lupus and current or past use of oral contraceptives (adjusted odds ratio, 1.3; 95 percent confidence interval, 0.9 to 2.1).¹⁶ In a retrospective case–control study of subjects with lupus who had a history of nephritis, Jungers et al.⁴ noted flares in 9 of 26 women during the first three months after the initiation of combined oral contraceptives as compared with none of 11 women receiving progestogen-only formulations. In contrast, Julkunen¹⁷ observed no difference in flare rates among 31 patients with systemic lupus erythematosus who were taking oral contraceptives, as compared with 31 nonusers. In a retrospective survey of 55 women who used oral contraceptives after receiving a diagnosis of systemic lupus erythematosus, 7 reported an exacerbation of disease, which was usually confined to the musculoskeletal system (flare rate, 0.45 per 100 patient-months).³

The relevance of establishing the safety of oral contraceptives in systemic lupus erythematosus extends beyond birth control. Oral contraceptives may control cyclic lupus activity.¹⁸ Estrogens prevent bone loss and increase bone mass,^19,20 suppress bone-resorbing cytokines such as interleukin-1²¹ and interleukin-6,²² and exert positive changes in calcium homeostasis.²³ In a retrospective cohort of 702 women with lupus, fractures occurred in 12.3 percent, a rate nearly five times that in healthy women.²⁴ In 376 women with lupus who were observed prospectively, the use of oral contraceptives was associated with a significantly reduced risk of musculoskeletal damage.¹¹ The guidelines of the American College of Rheumatology Task Force on Osteoporosis recommend the administration of oral contraceptives to prevent glucocorticoid-induced osteoporosis in premenopausal women with oligomenorrhea or amenorrhea,²⁵ although this benefit has not been prospectively evaluated in women with systemic lupus erythematosus.

In sum, although murine models and retrospective human studies suggested that exogenous estrogens might exacerbate lupus disease activity, the results of the OC-SELENA trial do not support this inference. Oral contraceptives were not associated with an increase in severe, mild–moderate, or total flares over the course of one year, as compared with placebo. These data support the use of oral contraceptives containing estrogen as a birth-control choice for patients with inactive or stable, moderate systemic lupus erythematosus who are at low risk for thrombosis.

The following investigators and study coordinators participated in the OC-SELENA Trial: Johns Hopkins Hospital, Baltimore — C. Robinson; Hospital for Joint Diseases–New York University School of Medicine, New York — P. Louie; Hospital for Special Surgery, New York — V. Kaplan; Louisiana State University Health Sciences Center, Shreveport — L. Green and R. Brouillette; Medical College of Wisconsin, Milwaukee — J. Zrnic; Oklahoma Medical Research Foundation, Oklahoma City — J. James, B. Lee, and F. Shelton; Rheumatology Associates of Long Island, Port Jefferson Station, N.Y. — D. Kaell; St. Luke's–Roosevelt Hospital Center, New York — J.T. Merrill, R. Shriky, and A. Shriky; University of California, Los Angeles — A. Rapkin and W. Chen; University of California, San Francisco — M. Fitzpatrick; University of Michigan, Ann Arbor — C.J.M. Van De Ven, G. Christman, and B. Gilson; University of North Carolina at Chapel Hill — W. Meyer and B. Meier; University of Pennsylvania, Philadelphia — K. Barnhart and L. Loh; University of Texas, Houston — N.-H. Chiu.