The aim of the present case-control study was to develop a noninvasive adjuvant tool for the diagnosis of endometriosis. Serum samples from 100 patients undergoing intracytoplasmic sperm injection were split into two groups according to the cause of infertility: an endometriosis group (n = 50), consisting of samples derived from patients with Grade III and IV endometriosis, and a control group (n = 50), comprising samples derived from patients with isolated male factor infertility. The metabolomic profile of each sample was obtained, through mass spectrometry. Partial least squares discriminant analysis was able to clearly classify the endometriosis and control groups. Ten potential biomarkers were selected based on their importance for model prediction. These ions were used to build the receiver-operating characteristic curve, which presented an area under the curve of 0.904 (95% confidence interval: 0.796-0.985). To validate the model, 30 other samples from infertile women without any evidence of endometriosis were tested. Considering these ions as possible biomarkers, the model was able to correctly classify 84% of the patients. Finally, a similar prediction potential was observed in the model validated set, when samples from the disease-free group were tested. Serum metabolomics may be useful as a noninvasive adjunct tool for the selection of patients who must undergo laparoscopy for definitive endometriosis diagnosis.
Keywords: biomarker; endometriosis; mass spectrometry; metabolomics.
© 2019 Wiley Periodicals, Inc.