The effect of microbiome and medications on maternal and child health

In this thesis, we used epidemiological approaches, utilized state-of-the-art sequencing tools, machine learning prediction models and performed an in-depth network meta-analysis to study the microbiome during pregnancy, microbiome modulating drugs and adverse outcomes including miscarriage, preterm birth and childhood cancer.

In Study I we utilized a large Swedish pregnancy cohort (SweMaMi) with extensive questionnaires and looked at women that had pre-pregnancy complications, defined as having experienced recurrent pregnancy loss, subfertility or the need of artificial reproductive technologies. We found that these women had different risk-profiles, where medication and weight were important modifiable risk factors. Furthermore, we found that women who had experienced recurrent pregnancy loss were associated with a higher risk for depression in their current pregnancy.

In Study II we used the same cohort and investigated women that had sent in vaginal and fecal microbiome samples in early pregnancy but had then suffered miscarriage (34 cases compared with matched controls). We found that having non-vaccine type HPV (aOR 5.17, 95%CI 1.27-21.02) or a vaginal microbiome dominated by less common lactobacilli (aOR 5.00, 95%CI 1.00-24.94) were associated with an increased risk of miscarriage. We also present a machine learning model which combines fecal and vaginal microbiome data, as well as selected questionnaire variables - reaching an AUROC value of 88%.

In Study III, we conducted a systematic review and network meta-analysis to study the association of the vaginal microbiome (as community state types, CSTs) and the risk of preterm birth. We found that women with "low lactobacilli" CST were associated with an increased risk of preterm birth (OR 1.69, 95%CI 1.15-2.49) compared to women with Lactobacillus crispatus dominant vaginal microbiome.

To continue our investigation of preterm birth, in Study IV we explored the fecal and vaginal microbiomes of women in the SweMaMi cohort that had delivered prematurely (132 cases) and compared them with matched controls. Using machine learning models, we propose a prediction model for preterm birth (AUROC 89%), which combines vaginal and fecal microbiome data, with selected questionnaire variables.

Lastly, in Study V we took advantage of the national Swedish registries to study the association of microbiome modulating drugs (antibiotics and proton pump inhibitors (PPIs)) taken in pregnancy or early life (before the age of two) and the risk of childhood cancer. We found that early life exposure to PPIs were associated with an increased risk of childhood cancer (aHR 3.68, 95%CI 2.24-6.06), similar to what has been observed in adult populations.

In conclusion, the microbiome has a lot of potential in the field of women's health as either a biomarker for early detection or as a therapeutic tool in the future. It should be highlighted that all included studies report either association or present a prediction model, and do not show causality. Future mechanistic or intervention studies on these topics are crucial before potential clinical applications.

List of scientific papers

I. Gudnadottir U, Du J, Hugerth L W, Engstrand L, Schuppe-Koistinen I, Wiberg-Itzel E, Fransson E, Brusselaers N Pre-pregnancy complications - associated factors and wellbeing in early pregnancy: a Swedish cohort study. BMC Pregnancy Childbirth 23, 153 (2023). https://doi.org/10.1186/s12884-023-05479-8

II. Gudnadottir U, Prast-Nielsen S, Wagner N, Hugerth L W, Alderheim V K, Antony A T, Du J, Guerreiro J R, Boulund F, Wiberg-Itzel E, Engstrand L, Schuppe-Koistinen I, Brusselaers N, Fransson E The role of the microbiome in miscarriage: a matched case-control study. [Manuscript]

III. Gudnadottir U, Debelius J W, Du J, Hugerth L W, Danielsson H, Schuppe-Koistinen I, Fransson E, Brusselaers N The vaginal microbiome and the risk of preterm birth: a systematic review and network meta-analysis. Scientific Reports 12, 7926 (2022). https://doi.org/10.1038/s41598-022-12007-9

IV. Wagner N, Gudnadottir U, Hugerth L W, Pavlenko T, Antony A T, Boulund F, Prast-Nielsen S, Ahlström G, Csitari B, Skalkidou A, Wiberg-Itzel E, Brusselaers N, Fransson E, Schuppe-Koistinen I, Engstrand L Prediction of spontaneous preterm birth based on the maternal vaginal and gut microbiome in combination with individual risk factors. [Manuscript]

V. Gudnadottir U, Fransson E, Ljungman G, Wikman A, Vlieghe E, Engstrand L, Brusselaers N Prenatal and early childhood exposure to proton pump inhibitors and antibiotics and the risk of childhood cancer: A nationwide population-based cohort study. [Submitted]