Skip to main content
Premium Trial:

Request an Annual Quote

UCSF Researchers Combine DNA Microarrays, Bioinformatics to Diagnose Endometriosis

NEW YORK (GenomeWeb) – Scientists have identified patterns of gene expression that could be used to create a non-invasive diagnostic test for endometriosis, using machine learning to analyze transcriptome data obtained from microarray testing of uterine tissue.

Results from the study, which was led by researchers from the University of California, San Francisco and funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development, were published this month in the journal Endocrinology. The researchers, led by Linda Giudice, an Ob/Gyn physician at UCSF, said in the paper that they could identify endometriosis with 90 percent to 100 percent accuracy, using relatively few genes.

Endometriosis is a condition that occurs when tissue from the endometrium, the uterine lining, flows backward through the fallopian tubes and into the pelvic cavity. Once in the cavity, the cells attach to organs such as the ovaries, the intestines, or the exterior of the uterus and continue to follow the monthly menstrual cycle, resulting in bleeding, inflammation, scarring and pain. The condition affects up to 11 percent of women of reproductive age and can take years to diagnose.

Currently, the only way to diagnose endometriosis is through laparoscopy, an invasive surgery. "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," Louis DePaolo, chief of the Fertility/Infertility Branch of the NICHD, said in a statement. Giudice said such a procedure could take as little as five minutes. The new microarray-based approach works because gene expression in the uterine lining is different in women with endometriosis.  

The researchers used Affymetrix HU133 Plus 2.0 high-density oligonucleotide arrays to characterize gene expression in 148 total samples; 77 were taken from women with endometriosis, 37 from women without endometriosis but with other uterine and pelvic problems, and 34 controls from women without any uterine conditions. The scientists validated gene expression using an Agilent Mx3005 Pro RT-qPCR system and Fluidigm 96.96 Dynamic Array Integrated Fluidic Circuits running on the Fluidigm BioMark system.

Machine learning algorithms then classified the samples based on the expression of specific genes. In addition to being able to accurately distinguish between patients with and without endometriosis, the researchers could also identify different stages in the disorder's progression.

The NICHD Reproductive Medicine Network has begun a multisite clinical trial that will validate the test in a larger cohort.