Predictive Model for the Non-Invasive Diagnosis of Endometriosis Based on Clinical Parameters

Abstract

Objectives: Are other pain symptoms in addition to dysmenorrhea, dyspareunia, dyschezia, dysuria, and chronic pelvic pain correlated to endometriosis and suitable for a clinical prediction model?

Methods: We conducted a prospective study from 2016 to 2022, including a total of 269 women with numerous pain symptoms and other parameters. All women filled out two questionnaires and were examined by palpation and transvaginal ultrasound (TVUS). In cases of suspected deep endometriosis, magnetic resonance imaging (MRI) was performed. After the operation, endometriosis was diagnosed by histological examination.

Results: All in all, 30 significant parameters and 6 significant numeric rating scale (NRS) scores associated with endometriosis could be identified: 7 pain adjectives, 8 endometriosis-associated pain symptoms, 5 pain localizations, 6 parameters from the PainDETECT, consumption of analgesics, and allergies. Furthermore, longer pain duration (before, during, and after menstruation) was observed in women with endometriosis compared to women without endometriosis (34.0% vs. 12.3%, respectively). Although no specific pain for endometriosis could be identified for all women, a subgroup with endometriosis reported radiating pain to the thighs/legs in contrast to a lower number of women without endometriosis (33.9% vs. 15.2%, respectively). Furthermore, a subgroup of women with endometriosis suffered from dysuria compared to patients without endometriosis (32.2% vs. 4.3%, respectively). Remarkably, the numbers of significant parameters were significantly higher in women with endometriosis compared to women without endometriosis (14.10 ± 4.2 vs. 7.75 ± 5.8, respectively). A decision tree was developed, resulting in 0.904 sensitivity, 0.750 specificity, 0.874 positive predictive values (PPV), 0.802 negative predictive values (NPV), 28.235 odds ratio (OR), and 4.423 relative risks (RR). The PPV of 0.874 is comparable to the positive prediction of endometriosis by the clinicians of 0.86 (177/205).

Conclusions: The presented predictive model will enable a non-invasive diagnosis of endometriosis and can also be used by both patients and clinicians for surveillance of the disease before and after surgery. In cases of positivety, as evaluated by the questionnaire, patients can then seek advice again. Similarly, patients without an operation but with medical therapy can be monitored with the questionnaire.

Keywords: PainDETECT; endometriosis; neuropathic pain; pelvic pain; pre-operative diagnosis; prediction model; questionnaire.

Figure 1

The number of significant parameters (SP) with a cutoff of 8.5 could be used as the root of the decision tree. In summary, 11 significant parameters were useful for the prediction. From the PainDETECT questionnaire, we used the final score (FS), the pain course pattern, pain radiating to the legs (legs), and lumbar spine pain (LS). The “classical” endometriosis pain parameters (dysuria, dyschezia, dyspareunia, period pain (PP), and obstipation) gave good discrimination. Furthermore, the duration of period pain during menstruation (PPM) was useful as a parameter.