Endometriosis Disease Study

Identification of genetic risk factors for better clinical decision support

View all

233

disease-risk loci

136

risk-associated genes mapped to disease-associated mechanisms

32

druggable targets

Overview

Endometriosis is a painful, debilitating condition that is poorly detected and often misdiagnosed. Endometriosis is estimated to affect 10% of all women, yet it can take an average of 7.5 years before a clear diagnosis is received. The disease and search for an accurate diagnosis has a significant negative impact on a patient’s physical and emotional well-being, quality of life, and economic productivity.

There is a need to identify disease subtypes and understand the mechanisms involved to improve the diagnosis and treatment of this disease and reduce costly and potentially harmful surgical interventions and downstream health consequences.

A recent analysis of data using our combinatorial analytics approach identified disease-associated single nucleotide polymorphisms (SNPs) from genes strongly associated with the risk of developing endometriosis and which have a clear mechanism of action connected to disease symptoms. Of the 233 loci discovered, one SNP was present in every endometriosis patient, indicating a potentially strong role in disease.

Four other SNPs that we identified had already been directly linked to endometriosis in the scientific literature, validating our results. Our analysis generated an initial patient stratification to enable the development of better, more personalized predictive risk scores and clinical decision support tools for a disease that affects at least 200 million patients around the world.

Endometriosis is a painful, debilitating condition where tissue similar to the lining of the womb starts to grow in other places, such as the ovaries and fallopian tubes. It can affect women of any age and has a significant impact on quality of life and, in some cases, fertility.

Women wait an average of 7.5 years before receiving a clear endometriosis diagnosis, and up to 75% of affected patients may be missed. Endometriosis affects 10% of all women and up to 35% of infertile women¹.

We used the PrecisonLife platform to find and statistically validate combinations of SNPs that together are strongly associated with endometriosis. These SNPs were mapped to the human reference genome to identify disease-associated and clinically relevant target genes.

FEMaLe Consortium

Finding Endometriosis using Machine Learning (FEMaLe) is a Horizon2020 project with focus entirely on improving diagnosis, prevention and care in endometriosis. The FEMaLe project builds bridges across disciplines and sectors to translate genetic and epidemiological knowledge into clinical tools that support decision-making in terms of diagnosis and care via machine learning and artificial intelligence.

UK Biobank

We conducted a hypothesis-free case : control study using genotype data from the UK Biobank, aiming to identify new risk variants associated directly with endometriosis, and gain further insights into the underlying pathology and disease mechanisms in relation to this patient group.

Endometriosis Disease Architecture

233

disease risk loci (SNPs)

136

risk- associated genes mapped to disease-associated mechanisms

druggable targets

tool compound

We performed high-resolution patient stratification analyses on genetic, clinical, and other ‘omics data to identify biological sub-types of endometriosis. Our insights into the combined effects of multiple genetic variants and clinical characteristics have been used to develop predictive risk scores and clinical decision support tools for clinicians.

Clustering the SNPs by the patients in whom they co-occurred allowed us to generate a disease architecture of endometriosis, providing insights into patient stratification. We used this to find genes and biological pathways associated with particular patient subpopulations and comorbidities, enabling the development of disease biomarkers and precision medicine strategies.

When the SNPs were mapped to genes, we identified >130 protein-coding genes strongly associated with the risk of developing endometriosis. We identified many genes involved in cell migration, with many linked to cancer in the context of promoting metastases. Several of these genes were also estrogen responsive and with differential expression in endometrial and ovarian cancers. Other genes we found regulated key processes such as cell adhesion, angiogenesis, and pro-inflammatory cytokine cascades.

Four genes we identified have already been directly linked to endometriosis in scientific literature, providing validation for our initial hypothesis-free analysis. In addition, we identified a glutamate receptor subunit that is involved in the amplification of neuropathic pain, which may indicate a genetic basis for pain associated with endometriosis in a subgroup of patients.

The Finding Endometriosis using Machine Learning project (FEMaLe)

We have translated our deep understanding of disease biology into clinical tools that support clinical decision-making aimed at both general practice and highly specialized endometriosis clinics.

We’re developing diagnostics and risk scores for the EU Horizon 2020 FEMaLe (”Finding Endometriosis using Machine Learning) project. Our work with the FEMaLe consortium will generate higher resolution stratification of endometriosis patient subgroups and greater insights into the underlying genetic factors relating to specific phenotypes of the disease. We will also run adenomyosis (ICD-10 code N80.0) and endometriosis (ICD-10 codes N80.1–8) as two separate cohorts against the same controls used in this study, to gain further understanding of the genetic differences underlying both diseases.