In “Promising molecular diagnostic approach to endometriosis” we learn about the use of decision trees using genomic data to diagnose endometriosis. The importance of this work is stated as:
‘Based on this gene expression, a simple test eventually could be performed in the doctor’s office to determine endometriosis and stage, Giudice said. In just minutes, a tiny, thin plastic catheter could be inserted through the cervix into the uterus to remove a sample of cells for analysis.
“Laparoscopy involves general anesthesia and making an incision in the abdomen,” said contributing author Louis DePaolo, PhD, chief of the Fertility/Infertility Branch of the National Institutes of Health (NIH) Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD). “These findings indicate that it may be possible to avoid the surgical procedure and diagnose endometriosis from a tissue sample obtained in the office setting without anesthesia.”‘
The entire article can be read here.
The article was based on the paper: “Molecular Classification of Endometriosis and Disease Stage Using High-Dimensional Genomic Data“. Below is the abstract.
‘Endometriosis (E), an estrogen-dependent, progesterone-resistant, inflammatory disorder, affects 10% of reproductive-age women. It is diagnosed and staged at surgery, resulting in an 11-year latency from symptom onset to diagnosis, underscoring the need for less invasive, less expensive approaches. Because the uterine lining (endometrium) in women with E has altered molecular profiles, we tested whether molecular classification of this tissue can distinguish and stage disease. We developed classifiers using genomic data from n = 148 archived endometrial samples from women with E or without E (normal controls or with other common uterine/pelvic pathologies) across the menstrual cycle and evaluated their performance on independent sample sets. Classifiers were trained separately on samples in specific hormonal milieu, using margin tree classification, and accuracies were scored on independent validation samples. Classification of samples from women with E or no E involved 2 binary decisions, each based on expression of specific genes. These first distinguished presence or absence of uterine/pelvic pathology and then no E from E, with the latter further classified according to severity (minimal/mild or moderate/severe). Best performing classifiers identified E with 90%–100% accuracy, were cycle phase-specific or independent, and used relatively few genes to determine disease and severity. Differential gene expression and pathway analyses revealed immune activation, altered steroid and thyroid hormone signaling/metabolism, and growth factor signaling in endometrium of women with E. Similar findings were observed with other disorders vs controls. Thus, classifier analysis of genomic data from endometrium can detect and stage pelvic E with high accuracy, dependent or independent of hormonal milieu. We propose that limited classifier candidate genes are of high value in developing diagnostics and identifying therapeutic targets. Discovery of endometrial molecular differences in the presence of E and other uterine/pelvic pathologies raises the broader biological question of their impact on the steroid hormone response and normal functions of this tissue.
Endometriosis (E) is a progressive, debilitating, estrogen-dependent, progesterone (P4)-resistant, inflammatory disorder associated with pelvic pain and infertility, with endometrium (uterine lining)-like tissue present outside the uterus (1, 2). By retrograde menstruation, endometrial tissue fragments/cells are transplanted to the pelvis (3), where they establish a blood supply, respond to cyclic hormones, grow, invade surrounding structures, become innervated (4, 5), and elicit a local inflammatory response and scarring (1, 2). E affects 6%–10% of reproductive-age women (6) and 50% of women with pelvic pain and/or infertility (>100 million women worldwide) (7) and is a major cause of disability and compromised quality of life (8, 9). Pelvic, lower abdominal and back pain, and urinary and gastrointestinal symptoms make diagnosis challenging, because many symptoms are nonspecific or are associated with other disorders (1). Pelvic inflammation and nerve infiltration result in pain (4, 5), and infertility is due to ovulatory dysfunction, poor egg quality, abnormal (P4-resistant) uterine endometrium, and compromised embryo implantation (1, 2, 10). In 2009, estimated United States healthcare costs for diagnosis and treatment of E-related pain and infertility totaled $49 billion (9).
The current gold standard for E diagnosis and staging is surgical visualization under general anesthesia with histologic confirmation of endometrial glands and stroma in biopsied pelvic lesions (1). Drawbacks of surgical diagnosis include procedural and anesthetic risks, time away from work and family, and cost (1, 9). The mean time from pain onset to surgical diagnosis is 6.7–11.0 years (8, 11), with attendant risk of disease progression (12) over this interval. Diagnostic delay may have deleterious consequences (11, 13), including progression of pain and infertility requiring more aggressive treatment approaches. In addition, recent data suggest radical removal of all visible disease is protective against later developing ovarian cancer (14), underscoring the importance of early diagnosis and intervention. When E is suspected, pain is empirically treated with contraceptive steroids, antiinflammatory drugs, and GnRH agonists, but these therapies are unsatisfactory in 20% of women because of side effects or resistance (1). Thus, a prompt, low-risk, low-cost diagnostic with high accuracy is needed (15) to shorten time to diagnosis, minimize disease progression and ovarian cancer risk, optimize timing and strategies for pain and infertility therapies, and monitor disease recurrence.
Because the endometrial transcriptome differs significantly in women with vs without disease (16–20), herein we applied machine learning and high dimensional analysis to leverage these observed differences towards developing classifiers for E diagnosis and stage and pursue further insights into the pathobiology of endometrium in women with disease. We report highly accurate diagnostic classifiers that use sequential binary decisions, each based on specific gene sets, that distinguish E (disease and stage) and are menstrual cycle stage specific or hormonal milieu independent. Furthermore, differential gene expression and pathway analyses, based on these binary decisions, provided insight into steroid hormone signaling and other molecular and cellular dysfunctions in endometrium of women with E and also dysfunctions with other uterine/pelvic disorders.‘