NEW YORK – Researchers have identified two subtypes of polycystic ovary syndrome (PCOS) that could help with the development of new treatments for the condition. Using unsupervised clustering of clinical, metabolic, and hormonal data from women of European ancestry with PCOS, the two subtypes were classified as reproductive and metabolic, and linked to novel gene regions.
A team of scientists from the Icahn School of Medicine at Mount Sinai, Northwestern University, and Penn State College of Medicine carried out the study and published the results in PLOS Medicine on Tuesday. The study demonstrates that PCOS has different genetic contributions and that the syndrome is associated with both metabolic and reproductive abnormalities, said Richard Legro, one of the authors and a specialist in endocrinology as well as obstetrics and gynecology at Penn State Health.
Data from a PCOS genome-wide association study (GWAS), previously published by senior author Andrea Dunaif at Mount Sinai, was used to analyze clinical, biochemical, and other traits for the new study. The researchers performed a cluster analysis on 893 PCOS cases from the GWAS and replicated the clusters in an independent group of 263 PCOS cases that had no genomic data available.
PCOS is the most common endocrine disorder of premenopausal women and affects up to 15 percent of this population worldwide, said Dunaif. Not only is this syndrome the leading cause of infertility, but also a major risk factor for type two diabetes, she added.
Of the 893 PCOS cases, 207 were characterized as reproductive and 329 as metabolic, while the remaining cases fell into an "indeterminate" group. The "reproductive" group had higher luteinizing hormone (LH) and sex hormone binding globulin (SHBG) levels, as well as a low body mass index (BMI) and low fasting insulin levels. The "metabolic" group had a higher BMI and higher fasting glucose and insulin levels, with low SHBG and LH levels.
The use of genetic analyses was unique to the study, said Dunaif. The team is the first to present an objective and unbiased mathematical approach to classifying PCOS, as well as to demonstrate this method by identifying reproductive and metabolic subtypes associated with novel and distinct gene regions, she said.
"The gene signals associated with the PCOS subtypes, particularly those associated with the reproductive subtype, are exceptionally strong, conferring threefold to almost sixfold increases in risk, suggesting the genes causing these signals have large effects on disease risk," said Dunaif in an email. "Signals this strong should make the genes easier to identify than typical GWAS genes, which have very weak signals."
She added that even though these findings are "exceptionally intriguing and exciting," they still need to be confirmed. With further investigation, if it turns out that certain genes have a significant effect on PCOS risk, then these genes could be used for genetic testing to predict PCOS, she said. Between the two subtypes, alleles in four loci were associated with the reproductive subtype and one locus was significantly associated with the metabolic subtype, according to the study.
The researchers also demonstrated that the reproductive and metabolic subtypes clustered in families by analyzing a separate, family-based cohort of 73 women. This familial analysis also pointed to carriers of rare variants in DENND1A, a gene that regulates androgen biosynthesis, who were significantly more likely to have the PCOS reproductive subtype.
Until recently, the diagnostic criteria for PCOS were not very scientific and based on expert opinion rather than knowledge of disease mechanisms because the causes of PCOS are unknown, said Dunaif. Current diagnostic criteria for PCOS include irregular and less frequent periods, evidence of elevated testosterone levels, and PCOS morphology. In a 2018 study, she and other researchers conducted the largest GWAS meta-analysis so far of more than 10,000 PCOS cases to compare different diagnostic criteria. They found no significant differences in genetic architecture between the current diagnostic criteria, suggesting they do not identify biologically distinct subtypes of PCOS.
The new PCOS subtypes identified in their recent study, on the other hand, were based on objective clustering of hormone levels and metabolic parameters, which demonstrated how these subtypes are associated with biological differences, Dunaif said. None of the three current PCOS criteria were found in the study's clusters, suggesting that the researchers' methods of looking at other hormonal data, such as LH and SHBG, may be better at identifying women who may have reproductive problems. Altogether, the team discovered novel and unique genetic signals that are not identified using current PCOS diagnostic criteria.
"We shouldn't be basing the diagnosis of PCOS on what 20 experts in a room vote, which is essentially the way it was decided," said Legro. "It should be based on scientific data. Clearly, these types of scientific data say that we should be peeling off the layers of PCOS according to the presenting complaints, and we shouldn't just lump in all diagnoses and all treatment of PCOS under this grand rubric."
So far, GWAS have suggested certain pathways, such as gonadotropin secretion, that are important in causing PCOS. Also, DENND1A appears to be an important gene regulating testosterone production in women and it likely plays a major role in the development of PCOS, Dunaif said.
"I think we're going to see a lot more of this kind of research as a way to understand disease better," she said. "The impact of these biologic approaches is already evident in oncology where cancer treatment is now based on molecular characteristics of tumors rather than their clinical presentation."
Dunaif said it's hard to say if a genotype will be a better way of diagnosing people with PCOS, but that the findings of the recent study are encouraging and show that measuring hormone levels associated with each PCOS subtype may result in more accurate diagnoses of this syndrome than the current PCOS diagnostic criteria.
"Identifying the genes associated with the PCOS subtypes promises to uncover the causes of PCOS and identify pathways to target for innovative PCOS therapies," she said. "The hormonal and genetic profiles of the PCOS subtypes can be used for personalized medical approaches to PCOS."
Based on this study, and with additional research, there is great potential for this to impact clinical applications and treatment if there is a strong, reproducible link established between gene expression and risk factors, said Renate Meier, a nurse practitioner and an assistant in obstetrics and gynecology at Vanderbilt Health University Medical Center, in an email.
"If it can be shown that these results are reproducible in other populations and in larger cohorts, genotyping could potentially help to improve PCOS management," said Meier, a specialist in PCOS at the Vanderbilt Polycystic Ovarian Syndrome Program who was not involved with the study.
One limitation of the new study is that it only included European ancestry PCOS cases diagnosed by one set of diagnostic criteria. With further studies, Dunaif and her team plan to investigate if similar subtypes are present in PCOS patients of diverse racial and ethnic backgrounds, as well as those diagnosed by other diagnostic criteria.
The challenge going forward is finding the exact gene or genes that generate the strong genetic signal within PCOS. Extensive gene mapping is needed to identify this signal and then figure out how it disrupts gene function.
Next month, Dunaif will be starting a new project that is a follow-up to the subtyping study. For that study, funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, the researchers will obtain additional groups of women with PCOS. They will be collaborating with co-investigators based at Imperial College London and the University of Oxford in the UK, Erasmus University Medical Center in the Netherlands, Ewha Womans University in South Korea, as well as at Duke University.
Genetic analyses are finally beginning to offer researchers insights into the pathways causing PCOS, she said. In terms of translating this research into new treatments for PCOS, Legro said he doesn’t think this will happen for another 10 years, though he hopes it will come within that time frame. With more studies and compelling enough data, Dunaif anticipates that changing the treatment of PCOS may take less than 10 years.