Identifying a panel of nine genes as novel specific model in endometriosis noninvasive diagnosis

Abstract

Objective: To study biomarkers to develop a novel diagnosis model for endometriosis and validate it using clinical samples.

Design: We used publicly available data sets and weighted gene coexpression network analysis to identify differentially expressed genes. Ten machine learning algorithms were used to develop an integrative model for predicting endometriosis. The accuracy and robustness of the model were validated using data sets and clinical samples.

Setting: Department of Obstetrics and Gynecology, Tangdu Hospital, Air Force Medical University, Xi'an, Shaanxi, China.

Patient(s): The study included clinical patients between the ages of 20 and 40 years who required laparoscopic surgery and who had not undergone hormone therapy within the previous 3 months. All the healthy individuals had given birth to a child at least once in their lives. Patients with inflammatory conditions, malignant diseases, immune diseases, myoma, or adenomyosis were excluded. Paraffin blocks of the samples were collected (case, n = 5; control, n = 5). Blood samples of 58 individuals were collected (case, n = 28; control, n = 30).

Intervention(s): None.

Main outcome measure(s): The areas under the receiver operator characteristic curve of our diagnostic model were measured for data sets and clinical samples. Multiplex immunohistochemical staining and real-time quantitative polymerase chain reaction assays were used for the validation of the model from tissue slides and peripheral blood samples.

Result(s): A nine-gene panel endometriosis messenger RNA score (EMScore), was constructed to distinguish the patients with endometriosis from healthy individuals using algorithms. The EMScore accurately predicted endometriosis, and the areas under the receiver operator characteristic curve of our diagnostic model were 0.920, and 0.942 for tissue and blood samples, respectively. Moreover, the EMScore outperformed other acknowledged signatures for predicting endometriosis across seven clinical cohorts. Overall, the EMScore constitutes a sensitive and specific noninvasive diagnostic method for endometriosis.

Conclusion(s): We developed the EMScore, a novel model that can aid in the diagnosis of endometriosis using peripheral blood samples. This study will contribute to the development of improved clinical noninvasive and sensitive diagnostic tools for endometriosis. These nine genes might be potential target molecules for treating endometriosis.

Keywords: EMScore; Endometriosis; biomarkers; diagnostic model; machine learning.