Abstract
Preterm birth (PTB), defined as birth occurring before 37 weeks’ gestation, is the leading cause of deaths in neonates and the second in children under five globally. Spontaneous preterm birth (sPTB) accounts for approximately 70% of all PTBs. sPTB is associated with a combination of modifiable and non-modifiable risk factors, including cervicovaginal microbiota, inflammation/infection, stress, lifestyle, environment, ethnicity, and body mass index (BMI). The cervicovaginal environment, including the host immune response and resident microbiota, is postulated to play a critical role in influencing inflammation, cervical shortening, and ascending infection, thus contributing to risk of sPTB.The overarching focus of this thesis was to use bioinformatics approaches to explore the complex associations between cervicovaginal microbiota, the immune response, bacterial vaginosis (BV), ethnicity, clinical data, and sPTB. There were four main research aims addressed. Firstly, to better understand how BV in pregnancy is associated with cervicov-aginal microbiota, elafin, ethnicity, and sPTB. Secondly, to explore neutrophil (the most abundant immune cell in the cervix) gene expression in pregnancy, and to determine how this associates with resident cervicovaginal microbiota. Thirdly, to enhance understanding of the network of associations between bacterial species detected in the cervicovaginal microbiota in pregnant Black women, and how clusters of these species are related to metabolites, inflammatory markers, and clinical data. Finally, to determine how knowledge of antenatal factors could, using machine learning, define clusters of pregnancies, and predict risk of sPTB.
Early analysis explored 16S microbiota data and demonstrated associations between BV status and the abundance of specific cervicovaginal bacteria and metabolites. Subsequent analysis focusing on immune cells indicated that cervical neutrophil proportions decreased across gestation, and their gene expression was associated with the metagenome and possibly ethnicity and/or BMI. Several correlations were found within and between omics data layers (metagenome, metabolome, and immune markers) in Black British women, and the metagenome network was associated with data from other omics layers. Unsupervised clustering established two clusters of pregnancies, with one group having over double the PTB rate. Supervised learning models predicted pregnancies as PTB with moderate success, with cervical length being the most important variable in the classification.
In conclusion, the applied bioinformatics research undertaken in this thesis enhances our understanding of the factors influencing sPTB risk and establishes a stronger ground-work for future investigations aimed at understanding and reducing the risk of sPTB. By continuing to explore the complex interactions within the cervicovaginal environment, valuable insights will be gained towards improving pregnancy outcomes.
| Date of Award | 1 May 2024 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Rachel Tribe (Supervisor), James Mason (Supervisor) & Flavia Flaviani (Supervisor) |