P266. Novel diagnostic models for endometriosis including serum CA-125 levels, body mass index, cyst pathology, dysmenorrhea or dyspareunia

Tamara Knific (SI), Katja Vouk (SI), Andrej Vogler (SI), Joško Osredkar (SI), Manuela Gstöttner (AT), René Wenzl (AT), Tea Lanišnik Rižner (SI)

[Knific] Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, 1000, Ljubljana, Slovenia, [Vouk] Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, 1000, Ljubljana, Slovenia, [Vogler] Department of Obstetrics and Gynaecology, University Medical Centre Ljubljana, Slovenia, [Osredkar] Clinical Institute of Clinical Chemistry and Biochemistry, University Medical Centre Ljubljana, Slovenia, [Gstöttner] Department of Obstetrics and Gynaecology, Medical University Vienna, Austria, [Wenzl] Department of Obstetrics and Gynaecology, Medical University Vienna, Austria, [Lanišnik Rižner] Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, 1000, Ljubljana, Slovenia

Context: Diagnostic models for endometriosis based on serum biomarkers and patient’s clinical data. Objective: To evaluate the diagnostic potential of preoperative serum levels of CA-125 and HE4 in patients with endometriosis-like symptoms, and to construct diagnostic models with potential clinical applicability. Methods: Preoperative serum levels of CA-125 and HE4 were evaluated using specific electrochemiluminescent assays. Logistic regression was used to construct diagnostic models with optimal diagnostic accuracy based on serum levels of CA-125 and HE4 and clinical data. Patients: In total 221 patients (124 endometriosis patients; 97 control patients) from University Medical Centre Ljubljana, Slovenia and Medical University Vienna, Austria. Interventions: Collection of serum samples and clinical data from women undergoing laparoscopy or laparotomy. Main Outcome Measures: Serum levels of CA-125, body mass index, information on cysts and dyspareunia or dysmenorrhea. Results: CA-125 levels were significantly elevated in endometriosis patients compared to control patients (p = 1.3 ×10-10), while the difference in HE4 levels did not reach significance (p >0.05). The most important confounding variables were body mass index (BMI) and binominal variables, as the presence/ absence of cysts, presence/ absence of dyspareunia, and score for dysmenorrhea. A model with CA-125, BMI, information on cysts and dyspareunia separated patients from controls with sensitivity of 74.0%, specificity of 81.3% and an AUC of 0.836. The second model included CA-125, BMI, information on cysts and dysmenorrhea and had similar diagnostic accuracy with sensitivity of 74.8%, specificity of 79.2% and AUC of 0.819. Conclusions: The constructed models have the potential for noninvasive diagnosis of endometriosis, and might be translated into clinical practice after additional validation.