Discussion Questions

• ■
  

  What was this study’s primary hypothesis?

  
  
• ■
  

  What assumptions did the authors make?

  
  
• ■
  

  What approach did the authors take to test their hypothesis?

  
  
• ■
  

  Were the study populations appropriate?

  
  
• ■
  

  Were the statistical analyses suitable?

  
  
• ■
  

  What were the key findings?

  
  
• ■
  

  How might these findings be further explored?

Introduction

Endometriosis is a common disease associated with pelvic pain, dysmenorrhea and dyspareunia. It is diagnosed when endometrial tissue is discovered outside the uterine cavity. While histologic evidence is usually limited to the pelvis, endometriosis has been identified in unusual locations, such as the lung, liver, and brain. Strict criteria for the diagnosis of endometriosis exist, but symptoms and corresponding surgical findings connected with the disease vary, as do theories regarding pathogenesis of the disease. In this month’s Journal Club, we discuss a study by Treloar and colleagues, who investigated links between early menstrual history and the diagnosis of endometriosis.

Emily S. Jungheim, MD and George A. Macones, MD, MSCE, Associate Editor

Background

Jungheim: Can you describe the different theories of the pathogenesis of endometriosis?

Schon: The prevailing theory in the pathogenesis of endometriosis is the theory of retrograde menstruation put forth by Sampson, in which fragments of menstrual endometrium pass backwards through the fallopian tube and implant on peritoneal surfaces. Other proposed theories include the coelomic metaplasia hypothesis, which suggests that the endometriotic lesions in the peritoneum are the result of differentiation of mesothelial cells into endometrium-like tissue. A third hypothesis is that of blood or lymphatic spread, in which menstrual tissue travels from the uterine cavity to other sites in the body through veins or the lymph system. Finally, a fourth hypothesis is that circulating blood cells originating from the bone marrow can differentiate into endometriotic tissue at various sites.

Sampson’s theory is the most commonly described mechanism. However, it does not totally explain why endometriosis only develops in some women when retrograde or reflux menstruation is so common. Mechanisms suggested to explain the persistence of this tissue within the peritoneal cavity include molecular defects within the endometrial tissue or a failure of the immune system to clear implants from the peritoneal surface.

Study Design

Jungheim: What were this study’s hypotheses? What assumptions were made?

Doblado: The authors hypothesized that women who were younger at menarche, had shorter cycle lengths, and had heavier menstrual flow were at higher risk for endometriosis. The authors also hypothesized that a history of longer menses, tampon use, and sexual intercourse during menstruation were risk factors for endometriosis.

One assumption the authors make is that the controls do not have endometriosis, but 35% of the control patients reported having dysmenorrhea often. It is not clear how many of these women with frequent dysmenorrhea have had surgery to evaluate for endometriosis, so it is possible that some of these controls have endometriosis. While it is unlikely that these women have moderate to severe endometriosis, it cannot be ruled out. The authors do address this concern and feel it is unlikely to have affected the results significantly due to the low prevalence of severe endometriosis in the community.

An implicit assumption the authors made in their hypothesis is that retrograde menstrual flow is important in the pathogenesis of endometriosis. While Sampson’s theory of pathogenesis probably has the most support, other theories, such as the coelomic metaplasia theory, do not involve retrograde flow.

Jungheim: What approach did the authors take to test their hypotheses?

Jimenez: The authors did a case-control study. Cases were women aged 18-55 years with surgically confirmed endometriosis (stage III/IV) and no affected first-degree relatives. Controls were randomly selected from female twin pairs enrolled in the Australian Twin Registry; they had never been diagnosed with endometriosis. The subjects completed a self-administered questionnaire.

Jungheim: Can you discuss the authors’ previous study using the same dataset?

Schon: The authors’ previously-published case-control analysis used the same dataset to compare the relative weights of women when they were 10 years and 16 years of age, as reported by cases, controls, and their mothers, and their risk of endometriosis. Other studies suggested that women with endometriosis are more likely to be thinner and underweight, so the purpose of this study was to determine whether relative weight in early life is associated with a subsequent diagnosis of endometriosis.

The authors found that self-report of being overweight at age 10 years was associated with subsequent development of endometriosis with an odds ratio (OR), of 2.8 (95% confidence interval [CI], 1.1-7.5). Similarly, the concordant response among mother-daughter pairs also suggested a positive association between being overweight at 10 years and development of endometriosis, but these results were not statistically significant. There was moderate concordance between self-reports and mothers’ reports of weights for both the 10-year and 16-year age groups. There were no statistically significant associations between weight at age 16 and the development of endometriosis; however, there was the suggestion of a positive trend between self-reports and mothers’ reports of being underweight at 16 years and the development of endometriosis.

The authors concluded that women who reported being overweight at age 10 were more likely to be diagnosed with endometriosis and that there was no clear evidence of an association between relative weight at age 16 years and endometriosis.

Jungheim: Was the control population for this study a reasonable choice? Might it have influenced the study findings?

Jimenez: The control population was taken from the Australian Twin Registry, a voluntary registry of mono- and dizygotic twins established in the 1970s. It encompasses a broad amount of information regarding demographics, lifestyle, and personal and family medical history. Data for the study by Treloar et al were gathered from questionnaires mailed to participants. The benefits of using a registry like this are that the registry is well-established, having been used successfully for a number of different research questions, and there is a lot of information available on the subjects.

However, while it is useful to have lots of data, we do not know the accuracy of the reported information or what available participant information was not included in this study. One piece of information that might be particularly important for this study is whether or not the registry participants were conceived through assisted reproductive techniques (ART). Given that many twins are conceived through ART, it is not unreasonable to suspect that a good number of the twins in this database were. This is essential because menstrual irregularities and endometriosis are potential reasons to undergo ART, and they are genetic disorders. If a significant percentage of the twins were conceived through ART, the control population might not be a representative sample of women without endometriosis. Furthermore, the control population was defined as having never been diagnosed with endometriosis in earlier studies. We do not know if they had prior surgeries that confirmed no evidence of endometriosis.

Jungheim: Can you discuss how the controls and cases were matched?

Allsworth: Controls were matched to cases using frequency matching based on age (within 5 years) and region of residence (urban vs rural). In frequency matching, cases are not directly matched to controls; instead, the relative percentages are balanced. For example, in this study the percentage of urban dwellers would be the same among controls as was observed among cases. In a case-control study, the primary benefit of matching is the potential for improved efficiency or precision. Improved efficiency should not be assumed, however, as the magnitude of the relationship between the confounder and outcome is positively correlated with expected gains.

Jungheim: What are the benefits and the shortcomings of a case-control study design? Are there any specific concerns you have about using a case-control study to address this study’s hypotheses?

Doblado: Case-control studies are fairly quick and inexpensive if reliable patient data are available, and they are also useful in examining rare diseases for multiple possible risk factors. Potential shortcomings are recall bias and problems with selection of the control group. A case-control study is probably the most feasible method for the study under discussion because moderate/severe endometriosis has a low prevalence, and several risk factors were examined. Recall bias could be an issue, as it is plausible that women with pelvic pain severe enough to require surgery may be more likely to report dysmenorrhea. The authors analyze several characteristics, such as location, ethnicity, marital status, and smoking status in the affected and control groups to help minimize the risk of a confounding variable affecting the results.

Jungheim: The primary exposure was described as “early menstrual characteristics.” Is there more information regarding the questionnaire that would have been helpful in determining how menstrual characteristics were assessed?

Jimenez: This questionnaire was used in prior studies by this group. Recall bias may be decreased but not eliminated. In the prior studies, the authors also contacted the study participants’ mothers to confirm information. It is unclear if the characteristics of interest were collected at each defined age band. Furthermore, there is no commentary about whether changes in menstrual characteristics or sanitary protection use over time were evaluated or analyzed. A record of when the endometriosis diagnosis was made and what treatment followed would have been helpful in assessing whether menstrual characteristics changed in relation to treatment. Better characterization of the pain such as severity, assessment of dyspareunia, and pelvic pain unrelated to menstruation might also have been useful data to collect.