Demystifying the Diagnosis of Polycystic Ovary Syndrome

QUEST WEBINAR 1

DR. ANDREA DUNAIF

00:00:07 Hi, everyone. My name is Trisha Winchester and I am the Senior Manager for the Clinical and Education team for the Cardiometabolic and Endocrine franchise with Quest Diagnostics. I have here today our esteemed speaker, Dr. Andrea Dunaif. She is the Professor and Chief of the Division of Endocrinology, Diabetes and Bone Disease at the Icahn School of Medicine at Mount Sinai. Today she is going to be talking to us about demystifying the diagnosis of Polycystic Ovarian syndrome. Dr. Dunaif, take it away.

00:00:37 Hello. Today I’d like to talk to you about challenges in the diagnosis and management of Polycystic Ovary syndrome. We are going to cover the diagnostic criteria, which it’s very important to know that these are based on expert opinion. They are not data-driven.There is a lack of knowledge among endocrinologists and primary care providers of the appropriate diagnostic evaluation for PCOS and its comorbidity and there is a lack of high-quality randomized clinical trials of optimal therapies for the management of nonreproductive comorbidities when they occur in women with PCOS. 

Why should you care about PCOS?  Well, it’s highly prevalent. It affects 5-15% of reproductive-age women worldwide and the healthcare costs just of the reproductive abnormalities are estimated to be around $8 billion per year. It presents early in adolescence which means if we can detect it, we could intervene to prevent some of the long-term adverse health outcomes. We now know that it’s a major metabolic disorder. There is a 4-fold increased risk for type 2 diabetes and this is independent of obesity.  The type 2 diabetes has a significantly younger age of onset and there is emerging evidence that youth onset type 2 diabetes may be more aggressive. Fifty to 80% of these women with PCOS are obese and this varies a bit by location but in all global locations women who have PCOS are more obese than women without PCOS, and since it’s got a genetic susceptibility, male as well as female first-degree relatives can be affected, can have features of PCOS. 

So, what is PCOS? Well, it’s a constellation of interrelated hormonal traits. It’s characterized by gonadotropin secretory abnormalities within increased GnRH which selectively increases LH while simultaneously suppressing FSH. We have increased LH drive of the ovary. LH stimulates ovarian androgen production and decreased ovarian estradiol production because of the FSH deficiency. The levels of the hormone anti-Müllerian hormone are elevated. These reflect the increased number of small growing follicles in the ovary which are what are called the cysts when they become arrested in development about 8-10mm in size. So, they are not cysts. Those are normal follicles that contain the eggs and with the appropriate hormonal stimulation, they can ovulate and lead to normal live births. We also then have feedback abnormalities on the hypothalamic-pituitary axis from the higher testosterone and that blocks the ability of estradiol to slow the GnRH pulse generator and we get a self-sustaining vicious cycle, it’s been called, of these reproductive abnormalities. And it was noted in about 1980 that women with PCOS had high circulating insulin levels, basally in response to glucose.  This was found to be result of severe insulin resistance about the same magnitude as we see in type 2 diabetes, so it’s a metabolic disorder. We also know that obesity is increased. And sex hormone bonding globulin is an interesting protein because it’s very important for modulating the bioavailability of testosterone, but it’s also regulated by insulin.  Insulin decreases it and so it’s a metabolic marker in addition to being a reproductive binding protein. 

And we don’t know what the cause of much more likely causes of PCOS are.  So, our current diagnostic criteria are completely based on expert opinion.  They are not based on data.  We have the so-called NIH criteria and those are hyperandrogenism HA, ovulatory dysfunction which means irregular, usually less frequent menstrual cycle – fewer than six to eight per year. In 2003, the Rotterdam conference – again, expert opinion – added polycystic ovarian morphology to the diagnostic criteria, said that you only needed two out of three of these three features, and that added two new classes of PCOS: hypoandrogenism and polycystic ovarian morphology with normal ovulation and ovulatory dysfunction, irregular periods and polycystic ovarian morphology with no androgen excess. I think the most important take-home message is that the NIH phenotype, which is quite easy to diagnose, is the phenotype with increased metabolic risks.  These other two phenotypes – and this has now been extensively studied – have much lower to absent evidence for insulin resistance and its associate metabolic morbidities. 

So, now we view PCOS as a disorder that affects women across the lifespan, presents with menarche in the reproductive years and we have mainly its reproductive features in the early reproductive years of irregular menses, fertility issues, symptoms of androgen excess. But the metabolic abnormalities also begin in adolescence and the adolescence prevalence of type 2 diabetes and impaired glucose tolerance is identical to what we see in adult women. Then as the women go into their thirties, reproductive abnormalities tend to ameliorate. The periods become regular, these women can have normal fertility, and in fact if you look at fertility overall in women with PCOS it’s not decreased.  They just have difficulty conceiving in their early reproductive years. Importantly, this disorder emerges in adolescence, so we have opportunities to prevent long-term health consequences if we can detect it. 

Now, let’s turn to practical approaches, the diagnosis and management of PCOS.  Let’s look at a series of 5,500 women with PCOS and how they presented.  These are women ultimately shown to have PCOS by Rotterdam criteria – by the broader criterion.  About 80% of women present – 78 – with ovarian dysfunction, and about 20% with regular menses. This is probably skewed by the fact that women with irregular menses are more likely to seek medical attention. Importantly, only about 60% of women with PCOS have hirsutism. Hirsutism is a function of whether or not the pilosebaceous unit is sensitive to androgens and how many there are. These are separate genetic traits, so not all women with androgen excess have hirsutism.  In terms of your diagnostic evaluation, the most consistent abnormality is elevated free testosterone. About 30% of women have elevated DHEAS, which is a marker of adrenal androgen secretion. If you look overall in the series of transvaginal ultrasounds, about 75% have polycystic ovarian morphology. This can vary a lot. If you get a highly skilled ultrasonographer who does repeat ultrasounds, they can get up to 100% and perhaps in a clinical radiology practice where they’re not used to doing this, you’re going to have a much lower detection rate. 

What is our diagnostic evaluation?  It really can be quite simple  First, we want to exclude other conditions. Any women with irregular menses should have pregnancy excluded. What are the other disorders that are common in the differential diagnosis?  Probably, the most common disorder – and it really isn’t that common and I’ll show you in a second – is the non-classic form of adrenal 21-hydroxylase deficiency. This is a genetic defect recessive disorder in the biosynthesis of cortisol at the 21-hydroxolase step. Currently, we screen for it with an early morning 17 hydroxyprogesterone level. Everybody gets a prolactin. Everybody gets a TSH. I think if you have a patient with irregular menses coming in to see you, particularly if she is not obese and not hirsute, you do have to think about other disorders in the differential diagnosis like hypothalamic amenorrhea and premature ovarian insufficiency.  In your initial evaluation, you can include an LH and FSH and an estradiol. 

Again, looking in a large series of women with hyperandrogenism or PCOS, how many have abnormalities in these things that we are excluding? About 4% have increased prolactin, about 1% have thyroid dysfunction, about 2% overall have non-classic congenital adrenal hyperplasia. However, this varies and certain racial and ethnic groups such as Ashkenazi Jews can have much higher prevalence so you should have a higher index of suspicion. What’s very fortunate is that serious disorders like androgen secreting tumors and Cushing syndrome are very, very rare. There are frequently calls from the Cushing syndrome community to be excluding Cushing syndrome in all women with PCOS, but this is incredibly low yield and you really want to have a high index of Cushing syndrome to proceed with doing an evaluation for androgen-secreting neoplasms only in those with markedly elevated testosterones need to be evaluated for a neoplasm.

What is the diagnostic evaluation in terms of measuring androgens?  The androgen that you want to measure definitely is testosterone and the accurate way to do that in the current era is by liquid chromatography tandem mass spec. 

Radioimmunoassays or ELISA assays do not have adequate specificity and sensitivity in the low ranges of testosterone that we see in children or in women. One needs a very precise high-sensitivity assay methodology and that is liquid chromatography tandem mass spec. This methodology is recommended by the Endocrine Society and the Centers for Disease Control. So, this is the gold standard assay that you should be doing.  You’ll often have to read carefully through the lab test offerings to ensure you’re getting the right type of assay, but it’s important to make this effort because one common reason I get referrals is for very abnormal testosterone levels that when they are repeated in the correct assay turn out to be normal, so it’s important. What about free testosterone? I showed you that free testosterone is the most common abnormality in hyperandrogenic women. How you measure this is also very important. There is a widely available direct-free testosterone assay. However, this is highly inaccurate and not reproducible. It really should not be used. This was stated by the Endocrine Society decades ago. Just to remind you that there is a subset of women who don’t have high circulating androgen levels – they just seem to have increased androgen action in their target tissues.  They’re hirsute, they probably have either increased conversion of testosterone to dihydrotestosterone by 5 alpha reductase or perhaps more sensitive androgen receptors. They are primarily presenting with just androgenic symptoms – hirsutism, acne, alopecia, and not with irregular menses – but every so often we can’t detect androgen excess in a woman who has hirsutism, clearly is hyperandrogenic and irregular cycles. So, clinical hyperandrogenism defined by hirsutism is sufficient to diagnose hyperandrogenism. And then, there has been a lot of publications recently on this class of adrenal androgens – 11-oxygenated C₁₉ steroids – and it’s become appreciated. This is really a large circulating pool of potent androgen and androgen precursors. However, there is no data to date that these should be measured clinically, so all you’re going to do is add expense to your evaluation if you can even find a laboratory that measures these androgens.

What about ovarian ultrasound? For those people who don’t do their own ovarian ultrasounds, which is unless you’re a reproductive endocrinologist or a gynecologist probably you don’t do your own, and you send this to a clinical radiology, you get a report that’s unfortunately not useful most of the time.

Even the guidelines agree on this. The international evidence-based guidelines state that you don’t need an ultrasound for the diagnosis of PCOS in women who have the NIH criteria – that is hyperandrogenism and ovulatory dysfunction. What about women with hirsutism? Do you need to worry about whether or not they have Rotterdam PCOS? For the management of their hirsutism and other potential hyperandrogenic symptoms such as acne or alopecia you don’t. This is from the Endocrine Society: “Unless fertility is an issue, demonstrating polycystic ovarian morphology to diagnose ovulatory PCOS is unlikely to affect management.” Bottom line, for the endocrine management of these patients either with NIH PCOS or with androgenic symptoms and regular menses, you do not need to assess their ovarian morphology.

What about the metabolic evaluation of PCOS?  There really aren’t the kind of robust large datasets that we need, but you can say that about almost everything with PCOS.  But certainly, the high-risk group for abnormalities of glucose tolerance – either prediabetes or type 2 diabetes – are the obese women with PCOS. Obesity really exacerbates these problems. There is agreement that we should test all obese women with PCOS with an oral glucose tolerance test, and people who have a first-degree relative that’s a parent or a sibling with type 2 diabetes should be tested. 

The controversy comes in overweight and lean PCOS and there are a variety of opinions – perhaps only those with metabolic syndrome.  It is true that there is an increased risk for impaired glucose tolerance in particular in lean PCOS, but it’s a modest increase. Importantly, HbA1C is not sufficient to detect dysglycemia in PCOS because these women are primarily insulin resistant and so their defect is in disposing of a glucose load. They have mostly postprandial hyperglycemia and it’s only late in their decompensation of glucose tolerance that they become having to have fasting hyperglycemia. We want to get a lipid profile. There is an increased risk for non-alcoholic fatty liver disease in women with PCOS, but how we should screen for it is not clear – whether we should be getting liver function in all women…we really don’t know how accurate that is for detecting women at risk. And then there is really no evidence for other biomarkers in PCOS such as C-reactive protein as aiding in the assessment.

What we’re looking at here is fasting glucose on the x-axis, 2-hour postprandial glucose on the y-axis. We have our fasting dysglycemia at 100. This is impaired fasting glucose – 126 is type 2 diabetes; 140 2-hour is impaired glucose tolerance; and 200 is type 2 diabetes. So, you see here there is a large number of patients who have completely normal fasting glucose, who have impaired glucose tolerance. So, you’re going to miss impaired glucose tolerance with a HbA1C which reflects mainly your ambient glucose levels and does not really reflect the postprandial excursion and so it really misses a lot of these patients. 

What about the management? Because we don’t know the cause, it’s symptomatic and directed at the patient’s concerns and goals and risk factors. It really sorts out into three areas: addressing hyperandrogenic symptoms – hirsutism, acne, alopecia – metabolic, obesity, dysglycemia, dyslipidemia, and reproductive and ovulation infertility. These women are at high risk for pregnancy complications. So, our management is personalized. I’m not going to go into tremendous detail but androgenic symptoms are most commonly treated with oral contraceptives and the off-label antiandrogen of choice in the United States, because we don’t have cyproterone acetate available, is spironolactone in high doses. We always have to remember that in women with oligomenorrhea, they are at risk for unopposed estrogen effect and endometrial neoplasia so they need to have endometrial shedding induced. This can be done with oral contraceptives, with intermittent administration of progestin such as 14 days of Prometrium every month or every several months. Progestin IUDs achieve this and are a great alternative in women who have contraindications to estrogen-containing contraceptives. And if metformin results in ovulatory cycles, it also results in endometrial protection. I do want to highlight the non-reproductive morbidities and the fact that these are tremendously exacerbated by obesity

What are our approaches here? Weight management is really a first-line approach because of the important role that obesity plays in the manifestations of these morbidities. We know that lifestyle alone is very difficult to achieve lasting weight loss and so now that we have these powerful new medications, this is an area that really is very promising for the management of PCOS. Bariatric procedures also have a role. Really, for metabolic syndrome and prediabetes, the same management, perhaps with the addition of metformin. There are very few PCOS-specific trials. There is no diabetes prevention trial in PCOS, so we just extrapolate the effectiveness of lifestyle in metformin from the diabetes prevention program in the normal population. Again, type 2 diabetes – there is no PCOS-specific data for management of type 2 diabetes in PCOS, but certainly since metformin is effective – this is a first-line drug in type 2 diabetes in general and in women with PCOS – but as GLP-1 receptor agonists become in greater use, these really may move up to first-line since the effect on weight is so much greater than that of metformin. Weight loss also reverses fatty liver disease. There is data that metformin improves it in PCOS and perhaps the GLP-1 receptor agonist. This is just a nice trial of the efficacy of metformin. At six months, about 42-47% of subjects resume regular menses with metformin, so it’s effective in almost half of women with PCOS at restoring regular menstrual cycles. We have much less data on the GLP-1 receptor agonist and as of my last check of the literature no data on the dual antagonist…agonist Tirzepatide. Unfortunately, most of the trials in PCOS have been done with lower doses of liraglutide with and without metformin, very few trials with the obesity dose and just a few trials with semaglutide and, as I said, not Tirzepatide. Other weight loss medications probably are most data with orlistat, which is effective but not preferred by patients, and other drugs such as phentermine, topiramate, naltrexone, bupropion – very minimal data. Combination therapy in the trials that there are seems to be superior to monotherapy and some data that SGL2 inhibitors may also be beneficial. In a recent meta-analysis, this is how they stratified from least effective to most effective.

These are very interesting data from a meta-analysis of bariatric surgery. Number one, the prevalence of PCOS among women coming for bariatric surgery is on average 36% and that postoperatively there was apparent complete resolution of PCOS in almost 100% of patients. It’s important to note that pregnancy after bariatric surgery…there are data that the infants can be small for gestational age and we really need much larger populations to know the effect of large amounts of weight loss prior to pregnancy and how much time a woman should be weight-stable before becoming pregnant.

In summary, PCOS is easy to diagnose with baseline testing. You need a detailed history and physical exam, limited number of static hormonal tests and serious disorders are very rare.

Thank you, Dr. Dunaif. That was a really great overview, really simplifying and obviously with the title of demystifying the diagnosis of PCOS, really important for providers to acknowledge what that path looks like and how to identify individuals with PCOS. One of the things that you mentioned a little bit about was AMH. Can you talk about the role of AMH in the diagnosis of PCOS?

AMH is made by developing follicles and correlates with the number of developing follicles. There is extensive data that it can be a proxy for polycystic ovarian morphology. Now, there is variability in the assays for AMH – different antibodies recognize different portions of the molecule, therefore there can be lab-to-lab variation in the absolute number that you’re getting. So, that’s one problem – how the AMH is being measured. I think there is a move to kind of standardize the assay that detects the bioactive form. Then there is some argument over what the best cut point is to say that there is polycystic ovarian morphology, but that said, a high AMH probably means that there is polycystic ovarian morphology and can be used in lieu of an ultrasound. However, that statement begs the question of why do we need to know there’s polycystic ovarian morphology. For most features of PCOS, we don’t need to know it. Where the assessment of polycystic ovarian morphology is critical is in the management of assisted reproductive therapy – so exogenous gonadotropin therapy for in vitro fertilization. Because polycystic ovaries are polycystic because those are multiple follicles – they’re not cysts – and they tend to all respond at the same time and you can get a syndrome which can be very serious called ovarian hyperstimulation. Women with polycystic ovaries, even if they have no other features of the polycystic ovary syndrome, are at risk for that so you want to monitor the ovaries. It takes around two years for women who present with symptoms of PCOS to get a diagnosis and they usually have to see three or more healthcare providers. There is just a huge lack of information and so everything we can do to make this a disorder that’s easy to diagnose… You saw in the data I presented you from the genetic analyses that women who self-report PCOS – so they’ve been given that diagnosis or they’ve read about the symptoms online – have the same genes as women in who we do an intense workup. So, it’s not difficult to diagnose; it’s pretty straightforward.

I think you outlined it great. It’s a side part – it’s not the main piece of the puzzle in terms of the morphology of the ovaries as being not that important when you’re already counting in those other criteria for a PCOS diagnosis. With that, do you think that women who might have missed a previous PCOS diagnosis who are then visiting with their primary care provider – does this make it more applicable for PCPs to also make the diagnosis of PCOS?

I think there has been a movement that menses should be a vital sign – that every woman of reproductive age should, right next to her blood pressure, have her menstrual history…last menstrual period. Because if a woman is having either long menstrual cycles – so, let’s say the normal cycle range is 21 to 35 days, so greater than 35 days or they are bleeding more frequently than 21 days – there is a problem that has to be diagnosed. For long menstrual cycles, that’s most likely to be PCOS. It can be other things but those other things aren’t good to have either. Hypothalamic amenorrhea which is a central shutting down of the reproductive axis leads to low levels of estrogen and is associated with risk for bone loss for osteopenia as is if there is premature ovarian insufficiency. And abnormal bleeding can be related to lesions in the uterus and polyps and fibroids. These symptoms…you need to have a diagnosis. It can’t be, “Oh, well.  It’s nothing to worry about.” Primary care providers have to be in tune to ask about this and to know what to do next. It’s fine to say, “I don’t feel comfortable with this,” and refer to somebody who does, but they need to be asking about the symptoms.

I agree.  Asking the right questions is the first tool in terms of really getting the right information to help improve patient care and to identify individuals with PCOS. With the associated risk that PCOS has, including with type 2 diabetes as well as non-alcoholic fatty liver disease, it’s important for all the providers in a woman’s life to be aware of this diagnosis. What would be your message to either a patient with PCOS…? How would you get this information to any provider that they’re visiting? Is that important as a patient as well for a provider to be asking those questions?

What’s needed is a large public education campaign.  Moms need to know that if their daughters are continuing to have irregular periods – and the definition is straightforward – that beyond two years after menarche, they need to be evaluated. Ideally, we would like to diagnose PCOS in girls once they’re two years into their adult reproductive stage.  Then we have the opportunity… It’s very hard to reduce weight but it’s easier to prevent weight gain. One thing that we can do early is prevent weight gain. We also have diabetes prevention measures that we can start… Making an early diagnosis and having that message out there is so important for population health because we do genetic studies and we often find we’ll diagnose PCOS in the daughter and then, because of our research, we’ll screen the dad and he’ll have type 2 diabetes. So, it means something to the men as well as the women and it’s a very important disorder for there to be increased awareness of. It’s a bad brand. The name PCOS is a misnomer. You don’t need to have PCOS to have the disorder, and this was said in the 2012 NIH conference evidence-based methodology workshop – that the name should be changed. Unfortunately, for lots of interesting non-scientific reasons, it hasn’t been and it’s an impediment to progress. 

I think you’re doing your due diligence here in terms of educational awareness and really simplifying the diagnosis of PCOS. Thank you, Dr. Dunaif, for your time and effort today giving us this important talk about PCOS and simplifying the diagnosis.

Thank you so much for giving me the opportunity to discuss this important disorder and to explain how it can be detected quite easily by providers.