Endometriosis Foundation of America
Medical Conference – 2012
Endometriosis in Adolescents: What We Do Now Matters Later
Stacey A. Missmer, ScD
Analytic methods in the study of risk factors for and consequences of endometriosis. What she is going to talk about this morning is adolescent endometriosis, What We Do Now Matters Later, Professor Missmer.
Good morning and thank you of course to our gracious hosts, Dr. Seckin and Ms. Lakshmi. Also, as being in the clean up position for this first panel of the morning I am going to present somewhat of a summary of what my esteemed colleagues have already discussed but also try to put it in the context of what we need to be doing now, as both scientists and as people who care about women and girls with endometriosis, to move the field forward.
So we are still charged with the question of what really is endometriosis. It has been 15 years ago that there were another group of studies that suggested descriptions across typical symptoms and typical experiences of women with endometriosis. In 1996 there was the U.S. Interview Survey where women with endometriosis reported that on average they experienced 18 bed days per year. Within this study bed days were defined as days in which women were unable to go about their normal activities whether at work or at home because of the symptoms of disease. In this same year there was also a publication from a large study based in Chicago. They enrolled almost 700 patients, half of whom presented with pain symptoms, half of whom presented with infertility. They suggested, as has been shown in other work, that the symptom presentation really does differ; with 80 percent of those presenting with pain showing moderate to severe pain that again hits that bed day level of impairing quality of life and daily activities, while 15 percent of the infertile group reported moderate to severe pain symptoms. There were differences in diagnosis timing; the delay was much greater in women who presented with pain. They were diagnosed at a later stage and there were different patient characteristics. For example, a critical thing to take on board when we are thinking specifically about adolescents with endometriosis is that virtually all of these patients are presenting solely with pain. We often conflate research and clinical attention regarding infertile populations versus those presenting with pain symptoms.
In a study that we conducted, and has also been duplicated by some others, when we look at potential risk factors or etiologic factors with endometriosis if we separate the patients by presentation of pain or infertility we do see some differential associations. Here is just one example where a cigarette smoking history and current smoking levels we saw an increased risk with endometriosis diagnosis in those who present with pain but a decreased risk with infertility. This may speak to the complexity of cigarette exposures. There is a hypoestrogenic state, which should be protective for endometriosis; however there are obviously many toxins, phytoestrogens and other growth and immunologic factors that are present in cigarette smoking. But the key point here is that we have to think more critically about phenotypic variations in both symptoms and appearance of disease.
Going back to the key question that we are all charged with is what is endometriosis? What does variation in lesion appearance mean? Is the different appearance that is typical of adolescents versus women diagnosed in their late 20s and into their 30s actually indicative of earlier stage disease that will then progress had we visualized that woman at 30 instead of at 15. Or, does it indicate distinct disease pathology, a distinct disease pathway that needs to be considered more critically? What also has been mentioned earlier this morning, what is the significance of superficial peritoneal endometriosis versus deep infiltrating endometriosis versus endometrioma when we are thinking in respect of genetic, epigenetic, risk factor and other pathophysiologically important pathways. What is the difference between women among different age groups for the propensity for scarring and adhesions? What are the fertility risks that are inherent in girls who are diagnosed in their adolescence? As Dr. D’Hooghe has summarized we do not really know. We have to be very careful about a conflation of endometriosis with infertility. The majority of women with endometriosis is fertile and go on to have spontaneous conceptions. We need to think critically when we are making these distinctions about symptom presentation and phenotypic appearance. What are pain symptom patterns? As has already been described there is very little data looking at the natural progression and the variation in pain symptoms over time given different treatments and in the absence of treatment. And also, what are the multi-systemic variation in inflammatory and immune responses? Rheumatologic diseases, gastroenterologic diseases – there is some data looking at dermatologic factors and endometriosis. All of these very distinct, currently observed variations needs to be addressed in what we are doing in terms of data collection and sample stratification and interpretation of the results and our conclusions for the field.
We know that the current classification system, which is most broadly used, does not correlate well with symptoms and it is questionable depending on what circumstances we are looking at for outcomes. So a key challenge to the field is working on again, broadening this phenotypic definition and really looking at these nuanced variations of disease presentation.
As already has been mentioned there have been no case controlled cohort or randomized controlled trials specifically within adolescents. The questions remain, again, are there risk factor differences that are specific to the population of girls who during adolescents are experiencing severe pelvic pain and are receiving a laparoscopic visualization and treatment? What are the critical windows of exposure timing, do they differ for patients who are presenting with symptoms at these earlier ages versus women who are experiencing later evidence of pain symptoms and of infertility? Is there a phenotypic-specific treatment that is critical to develop for endometriosis? There likely is. There is this degree of variation as we again have already heard. There are 50 to 60 percent recurrent rates, particularly for pain. We do much better for the treatment of infertility but what could these nuances of phenotypic presentation and nuances in treatment response tell us about honing in on patient specific treatments? What is the reproductive prognosis and what is the ultimate co-morbidity risk? What are the risks overall of developing autoimmune diseases? What are the risks of developing cancer, which I will talk about in a moment, and other chronic conditions associated with endometriosis? Are the risks greater in girls who are presenting with symptoms versus women who are experiencing infertility or other presentation patterns?
Thinking specifically about different designs, one of the most rigorous designs of course is randomized controlled trials. There have not been randomized controlled trials in adolescents and of course the advantage of a trial is that we are assigning the exposures so we are randomizing across all of that human variation that we quantify poorly, and differences between people that may be influencing what the outcomes are. Also, we can determine the exact temporality of these exposures and the outcomes. However, there are important limitations for randomized trials in our current understanding of adolescent endometriosis. They really are an invalid method for discovery. You have to know what you are trying to randomize and the expected outcomes and effects before we can choose randomized designs. We cannot ever evaluate harmful exposures with randomized controlled trials. We cannot evaluate personal patient characteristics such as body size, age at menarche, characteristics like that that cannot be affected by the scientists. And, we cannot evaluate the impact of varying exposure timing. Once we commit to a trial we are presenting that exposure to the enrolled population at this specific point in time. If the critical window were when they were children or early adolescents then that window is already missed if we are enrolling women who are now in their 30s. We have to choose a specific level of exposure, which again goes back to the discovery issue. We really need to understand what the critical elements are before we define how we will handle this trial.
I have these highlighted because depending on the population we choose, randomized controlled trials have to be considered critically for generalizability, there can be compliance issues and typically they are of shorter duration. All of these are surpassable and typical within the field of trials but something critical to think about, particularly when we are involving adolescents and thinking over a long period of time.
We also have to consider that a randomized controlled trial really is a prospective cohort study where we are assigning the exposure. The majority of the literature on trials in endometriosis is in pharmaceutical treatments but also specific to surgical intervention within infertile populations prior to assisted reproductive therapy. Those trials have been very successful in design in that particularly for pregnancy outcomes; there is a short duration, there is a very specific treatment. But now we are attempting to utilize those populations to look over time and say, “Okay, we have this population of women who had surgical treatment, a population of women who did not, now two, three, five years later what is their pain profile? What are the other symptom profiles? How many have come back for repeat treatment?” The critical thing for those designs is to consider that now this is no longer in the realm strictly of a randomized trial. Now we are looking at additional outcomes and so it takes on the nature of a prospective cohort and we have to be collecting, as Dr. D’Hooghe has also said, we have to be collecting over time information about pregnancy changes, terminal treatment changes and other lifestyle factor changes over that time period that may also be affecting disease recurrence, repeat surgery and pain symptoms.
Speaking of prospective cohorts, much of my research is based in the two perspective cohorts of the Nurses’ Health Study where we have enrolled almost 120,000 female nurses and have been following them for about 20 years, but this is a population that has not been advantageous for us for thinking about adolescent endometriosis, all of these women were aged 25 or older. They were also enrolled in 1989 and so the time period during which they were children and adolescents really preceded modern understanding and modern techniques where related to endometriosis. We have now enrolled 15,000 of their daughters and so we have multi-generational data. They were 9 to 14 years at enrollment so we have had the ability to have a much more critical eye with respect to endometriosis. We are starting to build – they are getting into their early 20s and so to date we have had 250 incident surgically confirmed endometriosis cases reported. But, again, reviewing some of what has been said earlier today, among these girls and now young women, 3000 have reported moderate to severe dysmenorrhea. Again, moderate to severe being defined as influencing their quality of life and daily activity. So, really thinking more broadly about pelvic pain and what thresholds are critical to consider in terms of what is “normal” or typical pain versus atypical pain indicative of critical pathology is very important.
However, there are many international prospective cohorts primarily designed to look at cancer and cardiovascular disease outcomes. It is a critical opportunity for our community to utilize these data and think about some life course information for endometriosis. But the critical concern is that few of these cohorts, almost none, have collected information on endometriosis retrospectively or prospectively. A case in point is three weeks ago there was the publication in Lancet Oncology from the American Cancer Association Consortium on endometriosis and ovarian cancer risk. There were 13 studies. They were all case controlled studies designed to look specifically at ovarian cancer with 8000 ovarian cancer cases and 13,000 controls. This study reported that women with endometriosis had a three-fold risk of clear cell ovarian cancer and a two-fold risk of endometrioid or low-grade serous ovarian cancer. They focused a great deal on the sub-type differences. However, there has been a lot of media attention to this research and one critical issue that is a bit beyond our purposes today is that it has been lost somewhat that there are very small absolute risks nonetheless within this group. So, women with endometriosis, if there is a true association, this is not a huge population level cancer issue.
But thinking about our design and our contribution as the endometriosis scientific and clinical community is that all of this study, like most of the large cohorts, did not collect any endometriosis information other than to include on a questionnaire to their participants “have you ever been diagnosed with endometriosis?” We know that from that self-report with no qualification and no medical record evaluation, that self-report is actually very poor for endometriosis. It is poor in a complex way. There are just a few examples; there was a mid-western United States study of polychlorinated biphenyls and they had self-reported endometriosis. When they went back to the medical records only 47 percent of those who said “yes I have endometriosis” actually had a medical confirmation.
In the Black Women’s Health Study cohort, which is a population of women who self-identify as being of African-American ancestry in the United States, they reported having endometriosis and in that population only 52 percent could be confirmed within the medical records. Even within our Nurses Health Study cohort when we evaluated the women who did not also in addition to saying “yes, a physician told me that I have endometriosis”, if they did not also have a surgical confirmation the confirmation rate was again around this 50 percent, 54 percent. You really have to get to the level of having the documented surgical confirmation to reach accurate levels. The critical concern in terms of scientific and clinical interpretation is that in these 50 percent or so who were wrong about their self-reported endometriosis, it is not that it is just completely misclassified and erroneous, there are typically reports of conflation of other gynecologic pathology or testing and treatments. They report endometriosis instead of adenomyosis, endometritis, endometrial polyps, having had an endometrial biopsy, all of these gynecologic terms that are similar to, but are not specific to, endometriosis. That actually biases in a different direction from what we would expect if it was just wrong. It may bias because of a conflation of all these other pathologic factors that are now being summarized across to look at these associations.
We in the field have to critically demand valid case definitions, we have to be very specific about what we mean and also think through these nuances of phenotypic sub-types. We also need to prioritize investigation of establishment of less invasive methods and many groups are working on that at this time. There will be more discussion of that later today.
We also, when we think about these cancer or other chronic disease associations we really need to challenge the thought of a causal relationship with endometriosis. If this is a true association then there really are three possible pathways for this relationship. It is possible that endometriosis has some causal function in the future development of ovarian cancer. There have also been reports of associations with breast cancer, lymphomas, melanoma and it is biologically possible that there is some causal relation that we need to investigate further to determine. It is also possible that there is something related to the in particular long term hormonal and immunological treatments for endometriosis that could be influencing risk. So, it is not the disease state itself but rather the necessary disease symptom interventions.
What is the most probable pathway however is that there likely are common risk factors, common hormonal, immunologic, and other aberrant milieus that are impacting both the risk of endometriosis and separately in a multi-systemic factor, the risk for these other outcomes. All of these pathways are discernible and need the multi-disciplinary attention of our bench scientists, our epidemiologists, our clinical researchers to tease this apart. But whenever we see associations like this we have to think very critically about what the implications are and what that means for our next steps in the field.
Again, there is value in knowing if women with endometriosis are at high risk for other chronic diseases including cancers. It can affect screening patterns for those cancers that have solid screening information. It can also affect, again, life term monitoring and interaction with the health care field, but we need to be very careful about our interpretations.
There also have not been case control studies. Again a critical issue about case control studies is this case definition and thinking about stratifying across presentation by pain or infertility, what the differential lesion appearance is, and presence of adhesion and scarring. We also have to think critically about control selection. So do we compare women with infertility to women who are fertile? Do we compare women with pain symptoms for endometriosis to those with infertility presentation? Do we compare those with stage one and stage two disease to those with stage three and four disease? And when is it appropriate to use a general population sample? One nice thing about the very large prospective cohorts is that we can be certain that there are women within those cohorts who have not been diagnosed with endometriosis who have existing disease. But in a sample of 120,000 women if we have 10,000 confirmed endometriosis cases then the characteristics of those not yet diagnosed women gets diluted within the other 90,000 to 100,000 women. This is much more critical to consider when we are thinking in terms of smaller case control studies.
Really, the choice of controlled group has to be driven by the hypothesis you are addressing. If we are trying to discover diagnostics then comparing a group of women who present with symptoms and may or may not have endometriosis to comparing those who are determined to have endometriosis, to those who have other pathology, is very, very important because that is the diagnostic distinction. If we are looking for ideologic factors however, comparing women with endometriosis to women with uterine fibroids for example may conflate relationships when there are similar hormonal or immunologic milieu at work. We also need to think critically about the exposure window and I will talk about that in a little bit which is very important when we think of endometriosis in adolescents and what timing may be incurring in terms of establishing the disease and then also sample size.
So an example of controlled selection is in the study of persistent organochlorines this was one of the most addressed associations, particularly in the 1980s and into the 1990s. There have been 13 case controlled studies. They largely have had inconsistent results, but within these 13 studies there were seven very distinct control definitions so comparability and interpretation is really impaired. Also within this particular field for the persistent organochlorines the range of sample sizes in these studies went from 10 at the smallest to 96 at the largest, all of which were underpowered for looking at these specific associations. Even the larger studies would need to see a two to four-fold risk with an individual persistent organochlorine and that is a magnitude effect that is not typical for the field. There may be true relations here that we just have not yet sufficiently uncovered because of this variation and study design and critically, the sample size issues.
With regard to exposure timing Teresa Woodruff and Linda Giudice had published this schematic thinking of these critical windows of development in utero as an infant and during adolescence. When we think about adolescent endometriosis, if we are trying to determine risk factors that are pathophysiologically informative for treatment development, and also informative for potential prevention and influence on the disease, we really need to be thinking in terms of early life exposures. We and others have observed an association between diethylbestrol exposure and endometriosis risk with about an 80 percent increased risk in the women who were exposed in utero. We have also observed, and it has been confirmed within some French groups and also ____ Australian research, that those women and girls who they themselves had a lower birth weight were at higher risk of endometriosis diagnosis versus those who were born with a higher birth weight. Interestingly with respect to body size, which as been one of the most consistent epidemiological findings is that this relationship persists across the life course and is particularly strong in childhood and adolescence. With, again, the same pattern of thinner girls, this is an average of ages 5 and 10 body size, being associated with higher risk compared to the heavier girls who had lower risk of diagnosis. This also persists to age 18 and into adulthood. Thinking of these critical windows perhaps even prethelarchal windows of exposure and what these patterns may suggest for pathophysiology of the disease. Very consistent literature has also been mentioned today is this relationship of a later age of menarche being associated with lower risk. Thinking through all the genetic, epigenetic and hormonal factors related to age at menarche teasing out the details of what may be happening in these very early in utero and childhood exposures is critical for the field.
We have also been looking at adult diet and endometriosis. So in women within the Nurses Health Study, those who were consuming higher amounts of omega-3 fatty acids, the healthier fatty acids, had a lower risk of being diagnosed with endometriosis. We have to think critically in terms of these diet factors because these are associations with diagnosis. So if the diet is influencing pain symptoms or fertility it may not be a true association with the disease itself but rather presentation for surgical confirmation. Similarly we saw an increased risk with unhealthy or trans fat consumption. This data has been confirmed also within Kevin Osteen and Kaylon Bruner trans work within their rat models. They have seen very similar associations.
We also see a protective effect of vitamin A which is also what Rod Taylor’s group is working on. However, when we separate out the components of vitamin A we do not see a relationship with retinol itself but rather with the pro-vitamin A carotenoids, beta-carotene and alpha-carotene. The critical information for our purposes thinking about adolescents is that for several primarily hormonally based diseases, such as breast cancer, also some evidence with colon cancer, there have been associations with menstrual characteristics and polycystic ovarian syndrome and some other outcomes. What we have observed in the adult diet is actually more strongly associated when we evaluate adolescent diets. Again, thinking through windows of exposure and critical associations and potential alteration of disease within those different life course issues is critical.
What is important for us to do now? We need to establish standard surgical and pathologic case reporting details that can be used across all clinics and groups. One difficulty for studies where their medical record extraction is important is that there is such variation in the details of the phenotype of the disease that are being reported. We are very limited in that respect. Establishing a standard minimum baseline and longitudinal data collection tools, again, that was referred to earlier the critical importance of creating standard questionnaires that are used internationally so that these data can be combined and considered critically. We also need to form multidisciplinary teams. The focus of research and the expertise of just the panel this morning is very varied and comes at these questions from very different approaches. Our keynote speaker is later today and the rest of today will also highlight that importance. So forming these multidisciplinary teams, attacking these questions from many different angles is absolutely critical. And joining collaborative groups; sample size, sample size, sample size, we need to be increasing the variation of girls and women that we are looking at and also just powering our studies appropriately to be able to detect these differences. We have to be thinking really paradigm challenging thoughts. What really is endometriosis and what do we know for sure, and what is dogma that has very thin evidence behind it? Then again, collecting data across the life course; what we need to do now among children and adolescents is collect that critical information so we will have it to look at their fertility outcomes later, their pain patterns later, their response to treatment across the next ten to 15 and 20 years.
Among clinicians there is a similar list although we certainly need to be educating other specialists. We need to be collaborating with and having detailed discussions with rheumatologists, gastroenterologists, nutritionists and pain management specialists; thinking about not only appropriate varied interventions but also what types of data we can be collecting to move forward the field across this very multi-systemically impacting disease.
Again, multi-disciplinary teams and joining collaborative groups and questioning what we really do know about the treatment responses. Questioning what in the presence of surgical intervention what it means that there is still a 50 percent return to surgery rate, which is probably an under estimation of the return of pain symptoms. That is only the section of patients who come back for another invasive surgery. Again, thinking paradigm challenging thoughts and engaging patients across the life course. Following up and forming methods for keeping in contact with patients in terms of this life course outcome for this very chronic disease, collecting data.
And then everyone who cares about daughters, sisters, friends with endometriosis and pelvic pain – we have to believe that pelvic pain is meaningful. Much of the delayed diagnosis is a requirement of surpassing a certain threshold of pain to be taken seriously and terms of true underlying pathology. Continuing to bear witness as happened yesterday and is continuing to happen within this community in terms of what the experiences of endometriosis are and really helping the clinicians and scientists to focus on the critical issues.
Promoting and being very vocal about health care professionals who are excellent supporters and excellent clinicians and scientists within the endometriosis field. Demanding some specialty care, again thinking across gastroenterology, rheumatology, pain management and the other complexities of this disease. And it is very critical, particularly in adolescence, to be willing to enroll in research studies and contribute information to questionnaires and bio repositories and tissue banks, and to be able to more broadly capture the details of this disease.
In summary of hopefully the morning’s panel we need large, collaborative and geographically diverse studies. We need to focus on the critical stratified case definition that the current staging system has not achieved. We have to think about control selection and appropriate comparison groups, and exposure data collection and co-variant information from across these critical windows of development and across the life course.
Thank you.
