Materials and Methods
Eligibility criteria and study selection
The present systematic review involved all published research articles that compared the efficacy of a continuous vs a cyclic OC regimen after surgery for ovarian endometriomas in the prevention of endometriosis recurrence. Studies assessing endometrioma recurrence at ultrasonography and/or pain recurrence at follow-up were included. Pain symptoms were evaluated separately for dysmenorrhea, noncyclic chronic pelvic pain, and dyspareunia. Studies evaluating only a cyclic or a continuous regimen were excluded. Studies reporting OC treatment after surgery for endometrioma recurrence were excluded.
Information sources and search strategy
An electronic database search was performed using PubMed, MEDLINE, and Embase for the identification of articles published through December 2014, using the combination of the following search terms: endometrioma, endometriosis, oral contraceptives, oral estroprogestins, laparoscopy, and surgery.
Only articles in English were included. Only full-length articles from peer review journals were considered. Congress abstracts were excluded.
Outcome measures
The primary analyses were aimed at determining the recurrence of endometrioma and the recurrence of pain after continuous OCs compared with cyclic OCs.
The recurrence of endometrioma had to be evaluated at transvaginal ultrasonography. The recurrence of pain, separately for dysmenorrhea, noncyclic chronic pain, and dyspareunia, had to be reported either as a dichotomous variable (symptom either present or absent) or as a continuous variable expressed with any validated scale (visual analog scale, for example). The pain outcome was evaluated separately for dysmenorrhea, noncyclic chronic pelvic pain, and dyspareunia. In the case of studies reporting all symptoms together, the authors were contacted to obtain the data for each symptom separately.
Secondary outcomes included discontinuation of treatment because of side effects, reoperation rates, patient satisfaction, and quality of life expressed with any validated method.
In the case of studies reporting on OC treatment after surgery for endometriosis in which the data from the population of patients with endometriomas were not reported separately from those of patients with endometriosis but without an endometrioma, the authors were contacted to obtain the missing data for the former population. In case the missing data could not be obtained, the studies were considered at the primary analysis for the evaluation of pain recurrence and secondary outcomes; a secondary analysis after the exclusion of these studies was planned for the evaluation of endometrioma recurrence.
A sensitivity analysis was planned after the exclusion of nonrandomized studies and randomized clinical trials (RCTs) at high risk of bias.
Data collection and extraction
Three investigators conducted the search independently. Following the electronic search, all articles considered pertinent on the basis of the title and abstract were retrieved, and their reference list was searched for additional potential studies. Subsequently, the same investigators independently read the full text to verify the pertinence of the article to the systematic review on continuous vs cyclic OCs after surgery of ovarian endometrioma. The data were extracted independently by each investigator and recorded on apposite forms. In the case of missing data, or data expressed as diagrams, plots, or rates with no absolute numbers, the authors were contacted to obtain the missing or incomplete data. Disagreements between authors were resolved by mutual discussion or by involvement of further investigators.
Assessment of risk of bias
Risk of bias assessment for the RCTs included in this review was performed using the Cochrane risk of bias assessment tool. Six domains were evaluated: random sequence generation, allocation concealment, blinding of outcome assessor, completeness of outcome data reporting, selective outcome reporting, and other potential sources of bias. The possibility of a publication bias was assessed visually using a funnel plot for asymmetry for the primary outcomes.
Data synthesis
The data were pooled using RevMan software (Review Manager version 5.1, the Cochrane Collaboration, 2011). Dichotomous outcomes from each study were express as a risk ratio (RR) with a 95% confidence interval (CI). Heterogeneity between studies was based on the results of the I 2 and X 2 statistics. A random-effect model was used at metaanalysis in case of high heterogeneity (I 2 > 50% or P < .10), whereas a fixed-effect model was used in case of low heterogeneity between studies. A value of P < .05 was considered statistically significant.
Trial registration
The systematic review was registered in the PROSPERO International Prospective Register of Systematic Reviews with the registration number of CRD42015016616. The Preferred Reporting Item for Systematic Reviews and Meta-analysis was followed.
Results
Study selection
The electronic search identified 13 potentially relevant papers. After removal of duplicates, 8 records were considered. On the basis of the title and abstract, 4 articles were included, whereas 4 were excluded ( Figure 1 ), for the following reasons: 2 were review studies, and 2 studies did not compare a cyclic vs a continuous OC regimen ( Figure 1 ).
After reading of the full text, 4 studies were included at final analysis, for a total of 557 patients evaluated. A total of 496 patients completed the assigned treatment and scheduled follow-up, 343 of which had ovarian endometriomas.
Study characteristics
The main characteristics of the included studies are detailed in the Table . In all the included studies, surgical treatment was performed by laparoscopic excision of the cyst wall. In no case nonexcisional techniques or oophorectomies were performed. Three studies were RCTs, and 1 was a prospective cohort trial. Comparable patient characteristics and exclusion rates for the 2 treatment arms are reported in the 4 included studies.
| Authors and year of publication | Study design | Patients, n | Patients with endometrioma, n | Minimum cyst diameter for inclusion, cm | Mean diameter ± SD, cm | Patients with bilateral cysts, n, % | Mean age ± SD, y | Oral contraceptive | Duration of treatment, mo | Follow-up, mo | ||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Cyclic | Continuous | Cyclic | Continuous | Cyclic | Continuous | Estrogen | Progestin | |||||||
| Muzii et al, 2011 | RCT | 57 | 57 | > 3 | 5.0 ± 0.9 | 5.1 ± 1.0 | NR | NR | 30.3 ± 2.9 | 30.6 ± 3.1 | 20 μg EE | 0.150 mg desogestrel | 6 | > 6 |
| Seracchioli et al, 2010 | RCT | 148 | 148 | > 4 | 4.9 ± 0.8 | 5.1 ± 1.1 | 8/75 (10.7%) | 11/73 (15.1%) | 29.7 ± 2.8 | 28.6 ± 2.4 | 20 μg EE | 0.075 mg gestodene | 24 | 6-12-18-24 |
| Seracchioli et al, 2010 | RCT | 187 | 187 | NR | NR | NR | NR | NR | 30.2 | 29.6 | 20 μg EE | 0.075 mg gestodene | 24 | 6-12-18-24 |
| Vlahos et al, 2013 | Prospective cohort | 293 | 138 | NR | 4.5 ± 1.4 | 4.8 ± 1.7 | 21/84 (25.0%) | 14/54 (25.9%) | 27.0 ± 2.7 | 28.0 ± 2.4 | 30 μg EE | 3 mg drospirenone | > 6 | > 6 |
None of the 3 RCTs was at high risk of bias ( Figure 2 ). Funnel plots for the primary outcomes did not reveal any publication bias. Two of the RCTs were conducted by the same principal investigator : the first of the 2 studies evaluated endometrioma recurrence at ultrasonography, whereas the second one evaluated the recurrence of pain in patients with ovarian endometriomas, with the latter study including additional patients with smaller endometriomas compared with the former study. The other 2 studies evaluated both endometrioma and pain recurrence.
Two studies compared a continuous vs a cyclic OC regimen, whereas 2 studies included also a third group of untreated patients, which was not included in this metaanalysis.
Three studies evaluated endometrioma recurrence at ultrasonograpy. Three studies evaluated pelvic pain: 2 studies analyzed dysmenorrhea, chronic pelvic pain, and dyspareunia separately, whereas 1 study reported pain symptoms together. In this latter case, the authors were contacted to obtain data for each symptom separately: separate data were provided by the authors for dysmenorrhea and noncyclic pelvic pain, whereas dyspareunia was not among the evaluated outcomes. In 2 studies, pain recurrence was expressed as a dichotomous variable (ie, presence or absence of pain), whereas in a third study, pain was expressed both as a categorical variable and as a continuous variable expressed with a visual analog scale scale.
A metaanalysis was performed for the presence or absence of recurrent pain as a dichotomous variable expressed as RR with 95% CI, considering separately dysmenorrhea, chronic pelvic pain, and dyspareunia. For the study reporting pain both as a dichotomous and as a continuous variable, only pain expressed as a dichotomous variable was included in the metaanalysis, for consistency with the other studies.
One study reported on patients with endometriosis, and, among the outcomes considered in the present systematic review, only the outcome of cyst recurrence was reported separately for the subgroup of patients with ovarian endometriomas. Both the first author and corresponding author were contacted to have separate data for the endometrioma patients and also for the other outcomes considered, but no response was obtained. Therefore, a secondary analysis was conducted, with the exclusion of this study. The same study is also the only nonrandomized study among the included studies. Therefore, the same secondary metaanalysis is valid as a sensitivity analysis performed after exclusion of nonrandomized studies.
The follow-up timing schedule varied between a minimum of 6 months and a maximum of 24 months after surgery. In most studies, multiple follow-up visits were planned, at either 6 or 12 month intervals ( Table ). For some outcomes, however, results were reported only at 12 months. For consistency, the 12 month visit data were therefore considered for all outcomes.
As concerns the primary outcomes, recurrence of endometrioma was not significantly different after a continuous vs a cyclic OC schedule (RR, 0.54; 95% CI, 0.28–1.05; P = .07) ( Figure 3 ). Heterogeneity for this comparison was low (I 2 = 0%). At the sensitivity analysis, after exclusion of the nonrandomized study, the endometrioma recurrence rates were still not significantly different between the 2 schedules (RR, 0.53; 95% CI, 0.22–1.31, P = .17) ( Figure 4 ). Heterogeneity for this comparison was low (I 2 = 0%).
For the pain recurrence rates, a continuous OC regimen was associated with a significantly lower RR for dysmenorrhea recurrence (RR, 0.24; 95% CI, 0.06–0.91; P = .04) ( Figure 5 ). Heterogeneity for this comparison was high (I 2 = 67%). After the exclusion of the nonrandomized study, a statistically significant difference was not reached for a lower RR risk after a continuous OC schedule (RR, 0.10; 95% CI, 0.00–2.70; P = .17) ( Figure 6 ). Heterogeneity for this comparison was high (I 2 = 80%).
Nonsignificant differences were present for chronic pelvic pain (RR, 0.61; 95% CI, 0.36–1.03; P = .06) ( Figure 7 ) and dyspareunia (RR, 0.77; 95% CI, 0.52–1.12; P = .17) ( Figure 8 ). Heterogeneity was high for the former metaanalysis (I 2 = 51%) and low for the latter (I 2 = 0%). A sensitivity analysis with the exclusion of the only non-RCT did not change significantly the results for chronic pelvic pain (RR, 0.78, with 95% CI, 0.56–1.07; P = .13, I 2 = 0%) ( Figure 9 ) and for dyspareunia (RR, 0.87, with 95% CI, 0.56–1.35, P = .53). In this latter analysis, only 1 study could be included.
