NMR analysis demonstrated high separation between endometriosis and control groups through an OPLS-DA model (R2Y = 0.854, Q2 = 0.649). However, drug metabolites, introduced via catheterisation during surgery, dominated the urinary profiles, masking any non-drug related metabolic signatures. Global profiling LC-MS analyses were similarly skewed by drug-derived signals, necessitating extensive annotated pre-processing and statistical filtering to reduce drug interference. Annotated LC-MS revealed significant differences (q <0.05) in 141 features between endometriosis and control cohorts. Purine metabolites, hypoxanthine, xanthine, and uric acid, were significantly higher in the endometriosis cohort in comparison to the control group. Feature-level variations of significance were identified between disease severities, including indoleacetic acid which had a lower abundance in the extensive endometriosis group in comparison to the minimal endometriosis group. Further, significant discriminating features between phenotype groups were identified including uracil, a pyrimidine metabolite.

Samples were utilised from two cohorts. Endometriosis samples (n = 82) were retrieved from the ENDORIGINS WA biobank, collected from patients undergoing laparoscopic surgery for the investigation of endometriosis. This allowed for further sub-classification of samples into disease severity and phenotype groups. Patients who did not receive a diagnosis of endometriosis during their surgeries were categorised into a nil group (n = 17). A control cohort (n = 57) was collected for this study, comprising of people assigned female at birth, with no diagnostic or clinical history of endometriosis.

These findings demonstrate the potential of urinary metabolomics for distinguishing endometriosis from non-endometriosis groups and for differentiating disease subtypes. Despite the influence of operative drug metabolites, biologically relevant signatures were still identified that support further investigation of urine as a non-invasive diagnostic approach for endometriosis. Importantly, this work highlights the value of considering disease severity and phenotypic variation when developing non-invasive diagnostic tools, rather than treating endometriosis as a single homogeneous condition.

Endometriosis is a chronic disease affecting 1 in 7 women in Australia. Currently, there is a delay to diagnosis of 8 to 11 years worldwide with diagnosis relying primarily on invasive laparoscopic surgeries. Thus, the development of non-invasive diagnostic methods is an urgent clinical priority. This thesis applied nuclear magnetic resonance (NMR) and liquid chromatography–mass spectrometry (LC-MS) to investigate urinary metabolite signatures associated with endometriosis.