Clinical Predictors of Endometriosis in the Infertility Population: Is There a Better Way to Determine Who Needs a Laparoscopy?

Transcription

1 Gynaecology Clinical Predictors of Endometriosis in the Infertility Population: Is There a Better Way to Determine Who Needs a Laparoscopy? Kellie Whitehill, MD, FRCSC, 1 Paul J. Yong, MD, PhD, FRCSC, 1,2 Christina Williams, MD, FRCSC 1,2,3 1 Department of Obstetrics and Gynaecology, University of British Columbia, Vancouver BC 2 British Columbia Women s Centre for Reproductive Health, Vancouver BC 3 Division of Reproductive Endocrinology and Infertility, University of British Columbia, Vancouver BC Abstract Objective: Endometriosis is a known contributor to infertility, but the gold standard for its diagnosis is surgical. Therefore, it is important for clinicians to be able to predict which women with infertility are at high risk for endometriosis and thus should be offered laparoscopy. We sought to identify the clinical predictors of endometriosis in the infertility population. Methods: We conducted a retrospective review of patients at an academic infertility centre. The primary outcome was identification of endometriosis at laparoscopy, and we used logistic regression to test clinical variables for their ability to predict endometriosis. Results: Primary infertility, dysmenorrhea, and uterosacral/cul-de-sac nodularity were significant independent predictors of finding endometriosis at laparoscopy. Other clinical variables (including hysterosalpingogram findings) were not independent predictors of endometriosis, and physicians with an endometriosis-focused practice were more likely to diagnose endometriosis at laparoscopy. Conclusion: Key predictors of endometriosis in the infertility population are primary infertility, dysmenorrhea, and uterosacral/ cul-de-sac nodularity. These results will be used to develop and validate a formal clinical prediction model for endometriosis in infertile women. Résumé Objectif : L endométriose est l un des facteurs connus de l infertilité et la chirurgie constitue la meilleure façon de la diagnostiquer. Ainsi, il est important que les cliniciens soient en mesure d identifier les femmes infertiles qui sont exposées à un risque élevé d endométriose et qui devraient donc se voir offrir une laparoscopie. Nous avons cherché à déterminer les facteurs prédictifs cliniques de l endométriose au sein de la population des femmes infertiles. Key Words: Endometriosis, infertility, laparoscopy, clinical prediction Competing Interests: None declared. Received on November 26, 2011 Accepted on March 6, 2012 Méthodes : Nous avons mené une analyse rétrospective des patientes d un centre d infertilité universitaire. Le critère d évaluation primaire était l identification de l endométriose au moment de la laparoscopie. Nous avons fait appel à la régression logistique pour tester la capacité des variables cliniques à prédire l endométriose. Résultats : L infertilité primaire, la dysménorrhée et la nodularité utérosacrée / du cul-de-sac étaient des facteurs prédictifs indépendants significatifs du fait de constater une endométriose au moment de la laparoscopie. D autres variables cliniques (y compris les constatations de l hystérosalpingogramme) ne se sont pas avérées être des facteurs prédictifs indépendants de l endométriose et les médecins dont la pratique était axée sur l endométriose étaient plus susceptibles de diagnostiquer l endométriose au moment de la laparoscopie. Conclusion : Les facteurs prédictifs clés de l endométriose au sein de la population des femmes infertiles sont l infertilité primaire, la dysménorrhée et la nodularité utérosacrée / du cul-de-sac. Ces résultats seront utilisés pour concevoir et valider un modèle structuré de prévision clinique de l endométriose chez les femmes infertiles. J Obstet Gynaecol Can 2012;34(6): INTRODUCTION Endometriosis affects one in 10 women of reproductive age 1 and is responsible for $1.8 billion in total annual costs to Canadian society. 2 In addition to infertility, endometriosis presents with dysmenorrhea, chronic pelvic pain, and deep dyspareunia, as well as tenderness and nodularity of the uterosacral ligaments and cul-de-sac. 3 Identification of clinical predictors for endometriosis is important, as it would allow clinicians to determine which infertile patients are at high risk for endometriosis and should be offered laparoscopy. Surprisingly, there have 552 JUNE JOGC JUIN 2012

2 Clinical Predictors of Endometriosis in the Infertility Population: Is There a Better Way to Determine Who Needs a Laparoscopy? been few studies of clinical predictors of endometriosis in infertile women. 4 6 Thus, there is controversy about which patients with infertility should be offered laparoscopy for diagnosis of endometriosis or other pelvic pathologies. 7,8 The few published studies in this area have found severe dysmenorrhea, dyspareunia, pelvic pain, and uterosacral ligament/cul-de-sac nodularity and tenderness to be associated with a subsequent diagnosis of endometriosis in infertile women. 4 6 In addition, a few studies have noted an association between findings on hysterosalpingography of polypoid endometrium (i.e., hypertrophic endometrium with a polypoid appearance) or intrauterine filling defects (e.g., endometrial polyps), and a subsequent diagnosis of endometriosis in infertile patients. 9,10 In this retrospective study, we used logistic regression to determine which clinical factors, including symptoms, signs, and HSG findings, are independent predictors of finding endometriosis at laparoscopy in infertile women. METHODS We retrospectively reviewed the medical records of women with infertility who underwent laparoscopy by gynaecologic infertility specialists at the British Columbia Women s Centre for Reproductive Health, an academic infertility centre affiliated with the University of British Columbia. All patients are asked to complete a standard questionnaire before the initial visit, including questions about severity of dysmenorrhea (absent, mild, moderate, severe), deep dyspareunia (present/ absent), and chronic pelvic pain (present/absent). At the initial visit, routine pelvic examination is performed. Patients are also offered HSG, and the majority of hysterosalpingograms are performed at one radiology centre and read by the same radiologist, allowing a consistent and standardized report. Our centre does not have a standardized protocol to decide which patients with infertility should have surgery; the decision to undergo laparoscopy is made by the individual clinician and patient. In this study, we did not explore why clinicians and patients decided to proceed with laparoscopy in each case. Inclusion criteria for the study were being women with no prior laparoscopic diagnosis of endometriosis, having laparoscopy performed at our centre ( ), and having medical records available on site. Exclusion criteria were not having HSG performed and having incomplete medical records (questionnaire not completed or pelvic examination findings not available). ABBREVIATIONS ART assisted reproductive technology HSG hysterosalpingography The primary outcome was whether or not endometriosis was diagnosed at laparoscopy in our centre. We tested 11 predictor variables: demographic variables (age, type of infertility [primary versus secondary], and duration of infertility); symptoms (degree of dysmenorrhea [absent, mild, moderate, or severe], deep dyspareunia [present/ absent], and chronic pelvic pain [present/absent]); signs (uterosacral/cul-de-sac tenderness [present/absent] and uterosacral/cul-de-sac nodularity [present/absent]); HSG findings (intrauterine filling defects [present/absent] and polypoid endometrium [present/absent]); and physicianspecific factors (laparoscopy performed by gynaecologic infertility specialists [3 physicians] or by gynaecologic infertility specialists with an endometriosis-focused practice [2 physicians]). The gynaecologic infertility specialists with an endometriosis-focused practice uniformly excise all suspected lesions of endometriosis (whether typical or atypical in laparoscopic appearance) and confirm the diagnosis on histology. The other gynaecologic infertility specialists either biopsy suspected lesions of endometriosis (whether typical or atypical in laparoscopic appearance) and confirm the diagnosis on histology, or they make a visual diagnosis of endometriosis based on typical appearance. Statistical analysis was performed using SPSS 19.0 (IBM Corp., Armonk NY). Means are expressed as ± 1 standard deviation. The Mann-Whitney test was used for scaled or ordinal predictor variables (age, duration of infertility, and degree of dysmenorrhea), and the Fisher test for binary predictor variables (all others). Significance was set at P < Multiple logistic regression modelling was performed using likelihood ratio modelling. All squared terms (predictor variable squared) and 2 2 interaction terms (e.g., age type of infertility) (n = 55) were tested for, with significance set at P < 0.01 for multiple comparisons. For each variable in the final logistic regression model, the odds ratio represents the following: 1. for binary variables, it is equal to the odds with the variable present divided by the odds with the variable absent; 2. for scaled or ordinal variables, it is equal to the odds with the variable = n + 1 divided by the odds with the variable = n (for example, the odds with severe dysmenorrhea divided by the odds with moderate dysmenorrhea). The study was approved by the clinical research ethics committees of the University of British Columbia and the Children s and Women s Health Centre of BC. JUNE JOGC JUIN

3 Gynaecology Table 1. Pairwise comparisons of women with and without endometriosis Predictor variable No Endometriosis (n = 261) Endometriosis (n = 168) P Demographic Age, years, mean (SD) 33.7 ± ± Primary infertility, n (%) 122 (47) 109 (65) < Duration of infertility, years, mean (SD) 2.9 ± ± Symptoms, n (%) Dysmenorrhea < None 90 (34) 37 (22) Mild 82 (31) 40 (24) Moderate 60 (23) 53 (32) Severe 29 (11) 38 (23) Deep dyspareunia 20 (8) 26 (15) 0.02 Chronic pelvic pain 33 (13) 31 (18) 0.13 Signs, n (%) Uterosacral/cul-de-sac tenderness 10 (4) 20 (12) Uterosacral/cul-de-sac nodularity 9 (3) 23 (14) < HSG, n (%) Intrauterine filling defect 45 (17) 27 (16) 0.79 Polypoid endometrium 2 (1) 5 (3) 0.12 Physician specific, n (%) Endometriosis-focused practice 56 (21) 78 (46) < RESULTS The study sample size was 429, of which 168 cases of endometriosis (39%) were identified at laparoscopy and 261 cases (61%) were found not to have endometriosis. On pair-wise comparisons, the predictor variables significantly associated with endometriosis were primary infertility, increasing severity of dysmenorrhea (absent, mild, moderate, or severe), deep dyspareunia, uterosacral/ cul-de-sac tenderness, uterosacral/cul-de-sac nodularity, and having a gynaecologic infertility specialist with an endometriosis-focused practice (Table 1). On logistic regression modelling, the predictor variables with an independent effect on endometriosis were primary infertility, increasing severity of dysmenorrhea, uterosacral/ cul-de-sac nodularity, and gynaecologic infertility specialist with an endometriosis-focused practice (Table 2). There were no statistically significant squared or 2 2 interaction terms. In Table 3, the proportion of cases with endometriosis is listed according to whether a patient had primary or secondary infertility, moderate/severe or mild/absent dysmenorrhea, and uterosacral/cul-de-sac nodularity or no nodularity. Of note, the proportion with endometriosis was 88% in cases with primary or secondary infertility, moderate/severe dysmenorrhea, and uterosacral/cul-desac nodularity (Table 3). In contrast, the proportion with endometriosis was 21% in cases with secondary infertility, mild/absent dysmenorrhea, and no uterosacral/cul-de-sac nodularity (Table 3). DISCUSSION In this study, we found that primary infertility, increasing severity of dysmenorrhea, the presence of uterosacral/ cul-de-sac nodularity, and having a specialist with an endometriosis-focused practice were significant independent predictors of endometriosis diagnosed at laparoscopy. On pair-wise comparisons, deep dyspareunia and uterosacral/cul-de-sac tenderness were also associated with endometriosis, but they were not independent predictors on logistic regression modelling. HSG findings (intrauterine filling defects or polypoid endometrium) were not associated with endometriosis. Our findings for symptoms and signs are similar to previous studies. Fedele et al. found the severity of dysmenorrhea to be greater, and the presence of dyspareunia and pelvic pain to be more likely, in women with infertility and endometriosis than in women with infertility but not 554 JUNE JOGC JUIN 2012

4 Clinical Predictors of Endometriosis in the Infertility Population: Is There a Better Way to Determine Who Needs a Laparoscopy? Table 2. Logistic regression model for predictor variables with an independent effect on risk of endometriosis Predictor variable β-coefficient Odds ratio 95% CI P Primary infertility to Degree of dysmenorrhea to Uterosacral/cul-de-sac nodularity to Endometriosis-focused practice of gynaecologist to 4.60 < Odds ratio = Exp[β-coefficient]. For degree of dysmenorrhea, the odds ratio represents (1) the odds of endometriosis in severe dysmenorrhea divided by the odds of endometriosis in moderate dysmenorrhea, (2) the odds of endometriosis in moderate dysmenorrhea divided by the odds of endometriosis in mild dysmenorrhea, and (3) the odds of endometriosis in mild dysmenorrhea divided by the odds of endometriosis in absent dysmenorrhea. Table 3. Proportion of cases with endometriosis Primary infertility Moderate/severe dysmenorrhea Uterosacral/ cul-de-sac nodularity Proportion with endometriosis 95% CI Primary or secondary Moderate/severe Nodularity 88% (15/17) 66 to 98 Primary or secondary Mild/absent Nodularity 53% (8/15) 27 to 78 Primary Moderate/severe No nodularity 51% (52/102) 41 to 61 Primary Mild/absent No nodularity 39% (42/108) 30 to 49 Secondary Moderate/severe No nodularity 39% (24/61) 27 to 53 Secondary Mild/absent No nodularity 21% (27/126) 15 to 30 Proportion of cases with endometriosis depending on whether primary or secondary infertility, moderate/severe or mild/ absent dysmenorrhea, and nodularity or no nodularity. endometriosis. 4 However, logistic regression modelling was not performed to determine whether these were independent predictors. Matorras et al. determined that cul-de-sac nodularity was the key indicator of endometriosis in the infertility population, using a stratified analysis as a replacement for logistic regression. 5 Cheewadhanaraks et al. found both tenderness and nodularity of the uterosacral ligaments/cul-de-sac to be associated with endometriosis in infertility patients with severe dysmenorrhea, but did not perform logistic regression. 6 In contrast, our results for HSG findings disagree with two previous studies that found an association between endometriosis and intrauterine filling defects or polypoid endometrium on HSG The reason for the discrepancy is likely that the previous studies confirmed HSG findings with hysteroscopy, which was not the case in our study. However, our study does indicate that clinical use of HSG findings to predict endometriosis is not possible without follow-up diagnostic hysteroscopy. Interestingly, gynaecologic infertility specialists with an endometriosis-focused practice were more likely to diagnose endometriosis at laparoscopy. The endometriosis-focused specialists uniformly excised all lesions of endometriosis (whether typical or atypical in appearance), and confirmed the diagnosis on histology. Therefore, one explanation may be that these physicians were more likely to recognize atypical patterns of endometriosis on laparoscopy and to excise them, and also to excise larger areas of peritoneum (rather than just biopsy); they were thus more likely to identify endometriosis confirmed by histology even when laparoscopic findings were atypical. These physicians may also be referred patients with a higher suspicion for endometriosis, although the association was present even when controlling for markers of severity of disease (e.g., degree of dysmenorrhea). However, we did not have data on stage of endometriosis, and it may be that cases referred to these physicians were found to have a greater surgical burden of disease. False-positive diagnoses by these physicians are unlikely as they uniformly excise lesions and confirm the diagnosis of endometriosis histologically. Our study is limited in that it is a retrospective chart review. Also, in some cases the diagnosis of endometriosis was based purely on a typical visual appearance on laparoscopy, and the proportion of such cases was not recorded. Patients were also asked to quantify the degree of dysmenorrhea as absent, mild, moderate, or severe, but these terms are JUNE JOGC JUIN

5 Gynaecology subjective and may not mean the same thing to all patients. Ideally, any scale for dysmenorrhea or other pelvic pain should be standardized by a scale for the acceptable pain level for the patient. Furthermore, all gynaecologists in this study were infertility specialists, and thus the patient population and study results may not be applicable to general gynaecologists. On the other hand, none of the patients had a prior laparoscopic diagnosis of endometriosis, which would be similar to the patient population seen by general gynaecologists. The study s strengths include a larger sample size than in previous studies of clinical predictors for endometriosis in infertility, as well as the use of logistic regression modelling to determine which pair-wise associations were true independent effects and which pairwise comparisons were affected by confounding. In the randomized Canadian Collaborative Group on Endometriosis study of the laparoscopic treatment of endometriosis for infertility, the number needed to treat for the outcome of pregnancy > 20 weeks was eight. 11 Based on this, patients may choose to proceed directly to treatment with assisted reproductive technology rather than laparoscopy. However, there will be subsets of patients who will choose laparoscopy: those who have a strong desire to have a diagnosis and to know whether or not endometriosis is a factor in their infertility; those who are young and with a relatively short period of infertility; those who have other pelvic causes of infertility (e.g., tubo-ovarian adhesions or hydrosalpinges) that can be treated surgically; and those who prefer spontaneous conception, whether because of financial limitations, potential side-effects during ART treatment, concern about multiple pregnancy, or personal beliefs. In addition, the Canadian study involved only endometriosis that was typical in appearance (pigmented). In our study, both typical pigmented endometriosis and atypical appearing endometriosis (e.g., erythematous, vascular reaction, or white) were included, although we do not have data on the proportion of each. For this reason, the results of the Canadian study may not be directly applicable to this study sample or to cases with atypical endometriosis in general, and the number needed to treat to achieve pregnancy may be different in these cases. Furthermore, laparoscopy allows assignment of a score for the Endometriosis Fertility Index, 12 which provides patients with a prognosis for non-ivf pregnancy. In addition, although dysmenorrhea and uterosacral/cul-de-sac nodularity are traditionally known as clinical predictors of endometriosis, our study attempts to formalize their value as clinical predictors through calculation of odds ratios. Our study also leads to a more evidence-based, rational clinical assessment, by showing that deep dyspareunia and uterosacral/cul-de-sac tenderness are not independent predictors of endometriosis in the infertile population. Finally, the data in Table 3 allow for more precise counselling about the probability of endometriosis, allowing women to make a more informed choice about whether to undergo laparoscopy or to proceed directly to ART. This study will form the basis of a clinical prediction model for endometriosis in the infertility population. We plan to obtain another retrospective sample and to recruit patients prospectively, and then formally develop and validate a clinical prediction model. Predictor variables in this model are likely to include primary infertility, degree of dysmenorrhea, and uterosacral/cul-de-sac nodularity, as identified in this study. The goal of the clinical prediction model will be to have a practical, easy-to-use rule that will allow clinicians to stratify infertility patients quickly, based on risk of endometriosis, and therefore to determine who would benefit from laparoscopy. Recent studies, including a double-blinded study, 13 have also investigated the role of endometrial biopsy for the identification of nerve fibres in the diagnosis of endometriosis. Endometrial biopsy could be incorporated into a clinical prediction model for endometriosis, although some studies have also found endometrial nerve fibres in painful conditions of the uterus (e.g., leiomyomas and adenomyosis), thereby affecting the specificity of the test. 14,15 The proportion of women with endometriosis, based on whether they have primary or secondary infertility, moderate/ severe or absent/mild dysmenorrhea, and uterosacral/ cul-de-sac nodularity or no nodularity is summarized in Table 3. The physician-specific factor (whether or not the physician s practice was endometriosis-focused) was not included in Table 3, as it is highly institution-specific and is a modifiable factor. Until the development of a formal clinical prediction model, Table 3 could be used by clinicians to stratify patients into sub-groups with different estimates of risk for endometriosis; however, the utility of this table is limited by a small sample size in some sub-groups, resulting in wide confidence intervals. CONCLUSION Clinical features that are significant independent predictors of endometriosis in the infertility population are primary infertility, moderate-to-severe dysmenorrhea, and uterosacral/cul-de-sac nodularity. In contrast, hysterosalpingogram findings were not associated with endometriosis in our study. We plan to extend our findings by developing and validating a formal model for the clinical prediction of endometriosis in the infertility population. 556 JUNE JOGC JUIN 2012

6 Clinical Predictors of Endometriosis in the Infertility Population: Is There a Better Way to Determine Who Needs a Laparoscopy? ACKNOWLEDGEMENTS We would like to acknowledge Dr Michael Papsdorf from the Women s Health Research Institute at BC Women s Hospital for assistance with the statistical analysis, as well as the staff and physicians at the BC Women s Centre for Reproductive Health. REFERENCES 1. Leyland N, Casper R, Laberge P, Singh SS, Allen L, Arendas K, et al. Endometriosis: diagnosis and management. J Obstet Gynaecol Can 2010;32(7 Supp 12):S1 S Levy AR, Osenenko KM, Lozano-Ortega G, Sambrook R, Jeddi M, Belisle S, et al. Economic burden of surgically confirmed endometriosis in Canada. J Obstet Gynaecol Can 2011;33: Spaczynski RZ, Duleba AJ. Diagnosis of endometriosis. Semin Reprod Med 2003;21: Fedele L, Bianchi S, Bocciolone L, Di Nola G, Parazzini F. Pain symptoms associated with endometriosis. Obstet Gynecol 1992;79: Matorras R, Rodriguez F, Ignacio Pijoan J, Soto E, Perez C, Ramon O, et al. Are there any clinical signs and symptoms that are related to endometriosis in infertile women? Am J Obstet Gynecol 1996;174: Cheewadhanaraks S, Peeyananjarassri D, Dhanaworavidul K, Liabsuetrakul T. Positive predictive value of clinical diagnosis of endometriosis. J Med Assoc Thai 2004;87: Nezhat C, Littman ED, Lathi RB, Berker B, Westphal LM, Giudice LC, et al. The dilemma of endometriosis: is consensus possible with an enigma? Fertil Steril 2005;84: Olive DL, Prits EA. The treatment of endometriosis: a review of the evidence. Ann NY Acad Sci 2002;955: McBean JH, Gibson M, Brumsted JR. The association of intrauterine filling defects on hysterosalpingogram with endometriosis. Fertil Steril 1996;66: Kim MR, Kim YA, Jo MY, Hwang KJ, Ryu HS. High frequency of endometrial polyps in endometriosis. J Am Assoc Gynecol Laparosc 2003;10: Marcoux S, Maheux R, Berube S. Laparoscopic surgery in infertile women with minimal or mild endometriosis. Canadian Collaborative Group on Endometriosis. N Engl J Med 1997;337: Adamson GD, Pasta DJ. Endometriosis fertility index: the new, validated endometriosis staging system. Fertil Steril 2010;94: Al-Jefout M, Dezarnaulds G, Cooper M, Tokushige N, Luscombe GM, Markham R, et al. Diagnosis of endometriosis by detection of nerve fibers in endometrial biopsy: a double-blind study. Hum Reprod 2009;24: Zhang X, Lu B, Huang X, Xu H, Zhou C, Lin J. Endometrial nerve fibers in women with endometriosis, adenomyosis, and uterine fibroids. Fertil Steril 2009;92: Zhang X, Lu B, Huang X, Xu H, Zhou C, Lin J. Innervation of endometrium and myometrium in women with painful adenomyosis and uterine fibroids. Fertil Steril 2010;94: JUNE JOGC JUIN