Optimising and investigating biomarkers in Pregnancy of Unknown Location outcome prediction and Ectopic Pregnancy diagnosis

Abstract

Thesis abstract Following a positive urine pregnancy test, a pregnancy of unknown location (PUL) is classified when a pregnancy is not visualised using transvaginal ultrasound. Pending final diagnosis, biomarkers are utilised to manage a PUL classification and predict the presence of an ectopic pregnancy (EP). PUL triage is performed in various ways globally, using single or serial values of human chorionic gonadotrophin (hCG), or single values of progesterone to predict pregnancy location. There is limited evidence regarding their value as predictors of early pregnancy viability, and data have emerged to suggest prediction models developed using logistic regression statistical strategies are superior when predicting pregnancy location. Point of care devices now exist that process samples for hCG. Whilst novel biochemical markers continue to be investigated, none are currently appropriate or suitable for use in clinical practice.

Individual and/or serial use of hCG and progesterone in women with a PUL classification cannot be used reliably to predict pregnancy viability. Model protocols developed to utilise hCG, hCG ratio and progesterone in combination, with or without the clinical factors of bleeding scores, EP history, and, to a lesser extent, maternal age, provide PUL risk and outcome predictions with excellent performance and high accuracy. Further safe enhancement of the patient pathway with point of care testing technology is possible, with rapid hCG results compatible with model algorithms.

The kisspeptin protein, associated with placentation, has not been deemed useful in predicting PUL outcome. This is likely secondary to the early gestation of sample collection. MicroRNAs are exciting novel epigenetic markers, with two deemed useful when used as a ratio to differentiate viable intrauterine pregnancies with EP.

Overall, this work supports the use of high-quality prediction models as part of PUL triage, with or without point of care testing technology. Following in silico analysis, the two microRNAs proposed as novel markers of PUL outcome prediction and EP diagnosis following discovery and validation work are biologically plausible.