Of 213 patients, 59 were treated using COCs (27.6%) and 154 were treated by administering DNG (74.6%). Baseline characteristics of patients were evaluated (Table 1). The DNG group included a greater number of patients with stage 4 and/or deep infiltrating endometriosis than the COC group. The mean duration for which medication was administered was 44.5 ± 22.6 months in the COC group and 23.6 ± 13.5 months in the DNG group.

The incidence rates of AEs in the DNG group were 29.2%, 15.7%, 14.0%, 19.2%, and 0% at Y1, Y2, Y3, Y4, and Y5, respectively. The most frequent AEs were lower abdominal pain (7.2%, n = 11/154) and frequent bleeding (6.5%, n = 10/154) after one year of DNG use. Lower abdominal pain decreased after two years; however, frequent bleeding was observed continually up to the fourth year of medication. The patients also gained weight up to the third year, though it was not reported afterward.

The incidence rates of AEs in the COC group were 27.1%, 20.1%, 10.0%, 10.5% and 7.4% at Y1, Y2, Y3, Y4, and Y5, respectively. The most common AEs after one year of COC use were lower abdominal pain (11.9%, n = 7/59), frequent bleeding (5.1%, n = 3/59), and mood change (5.1%, n = 3/59). Most AEs decreased with time, although frequent lower abdominal pain was reported even after the second year.

Further, each AE and the reason for discontinuation of COCs and DNG were analyzed (Table 2, 3). In the COC group, lower abdominal pain was the most frequently observed AE, while frequent bleeding, weight gain, and lower abdominal pain were the common AEs among patients in the DNG group. Patients in both the groups exhibited discontinuation rates that were much lower than the AE occurrence rates. The discontinuation rates by AEs were 11.7%, 4.1%, 4.0%, 3.8%, and 0% at Y1, Y2, Y3, Y4, and Y5, respectively in the DNG group, while they were 3.4%, 3.8%, 4.0%, 0%, and 3.7% in the COC group at the same times.

Reasons for patient drop-out per year in the study included follow-up loss, intolerable AEs and immediate plan for pregnancy. In the COC group, the numbers of follow-up-loss patients were 3, 1, 12 and 11 per year. In DNG group, they were 11, 62, 21 and 13 per year.

Recurrence rates observed through sonography were 3.4% and 1.9% at one year in the DNG and COC groups, respectively. After two years, there was no recurrence among patients in the COC group. In the DNG group, the recurrence rates were 1.9%, 2.5%, 4.0%, 3.8%, and 0% at Y1, Y2, Y3, Y4, and Y5, respectively.

Endometriosis is a complex clinical condition, and despite extensive research over the past 160 years, several controversies and dilemmas related to its pathogenesis, diagnosis, clinical management, prognosis, and epidemiology remain.16 However, what is certain is that endometriosis is a chronic disease with a 50% chance of recurrence five years after surgery, and therefore, long-term surveillance and management of the condition are required.4 Considering that hormone therapy is the mainstream treatment indicated for successful long-term patient care for endometriosis, it is important for physicians to understand its efficacy and possible complications.

In this study, the incidence rates of AEs were relatively higher than those in previous studies. For instance, Park et al. discussed the incidence of AEs after one year of DNG use in Korean women; the incidence rate was 18.1%, and the most frequently observed AEs were frequent or prolonged uterine bleeding (3.2%), insomnia (2.7%), nausea (2.1%), weight gain (2.1%), and acne (2.1%).17 Further, according to the study of Kim et al., the incidences of headache, breast discomfort, nausea, and weight gain were 6.0%, 8.0%, 5.3%, and 24.0%, respectively, after 18 months of DNG use.18

It is widely known that frequent bleeding is an adverse effect of DNG; however, its dramatically higher incidence rates in the current study might have been because of the time at which DNG was initiated. To reduce instances of irregular bleeding, patients are advised to start medication on the second or third day of their menstruation.8, 10 However, in this study, most patients experienced altered periods because of recent ovarian surgeries; thus, there was no strict control over the initiation timing of DNG. Further, the higher incidences of lower abdominal pain might have been due to discomfort at the operation site after surgery, inflammatory changes during the post-surgical healing process, and/or altered menstrual periods and related dysmenorrhea. These AEs may be reduced by adjusting the initiation time of medication. Additionally, it has been recently reported that cyclic DNG administration or the preceding administration of GnRHa could reduce the symptoms of abnormal vaginal bleeding.19, 20

According to the official product information of COCs, during their administration, the incidence of pelvic pain is less than 10%, that of genital tract bleeding is 14%, and that of premenstrual syndrome is 13.2%.

Moreover, we found that the discontinuation rates were lower than the incidence rates, which suggests that most AEs were negligible. Nevertheless, severe AEs, such as thromboembolism and possible bone loss, are clinically important factors that have to be considered when deciding whether to continue medication. However, considering that the occurrence of such severe AEs is rare, managing negligible AEs is one of the major factors in determining a patient’s compliance with long-term application of medical therapy.

A comparison between the incidence rates of AEs and medication discontinuation rates between the two groups for five years were not statistically significant (data not shown). Such results require careful interpretations; the current study did not intend to compare the two drugs, and its study groups are distinctively composed of patients with different characteristics. Therefore, it is hard to state that the two study groups have no significant differences; rather, well-designed prospective studies regarding such issue are warranted.

This study has several limitations. First, it was a simple observational study wherein direct evaluation of the patients was limited. Second, some of the AEs might not have been related to hormonal treatment. Third, the number of patients included in this study was relatively small; especially in the DNG group, the number of patients rapidly decreased after two years of medication. This may have been due to the initial approval of DNG use only for 18 to 24 months in pivotal studies.

Despite these limitations, this study provides important practical information on endometriosis management in real-world clinical settings and presents relatively long-term data of up to five years. Our findings showed that most AEs were not severe, and the medication discontinuation rates were much lower than the incidence rates of AEs in patients treated with DNG and COCs for secondary prevention of endometriosis. Although the AEs were very diverse, the incidence rates were relatively high. These results could serve as a guide for patients to decide when to start medical treatment. When considering long-term hormonal treatment for endometriosis patients, timely, detailed explanation of the expected AEs and options to control them is essential.

In conclusion, through our clinical experience, the present study findings present valuable information on long-term hormonal therapy to prevent the recurrence of endometriosis, which could consequently aid physicians in coming up with personalized treatment plans for patients. Furthermore, we found that the associated AEs were not severe enough to warrant discontinuation of medication. It is important that clinicians provide information regarding minor AEs to patients in advance to effectively increase their compliance and thus successfully control the recurrence of endometriosis.