Background: Preterm birth (PTB), defined as birth before 37 weeks of gestation, accounts for most neonatal death and morbidity. Accurate prediction is a pre-requisite for the prevention and proper management of PTB. However, methods for prediction are unsatisfactory, although sonographic cervical length has a moderate predictive value. For clinical utility, adding a biomarker could increase the predictive accuracy. The immune system has an important regulatory role during pregnancy. Thus, presumptive predictive biomarkers may be searched for among immune-related molecules, such as cytokines, chemokines and other inflammation-associated mediators. 

The aims of this thesis were to identify clinical risk factors and immunological prediction markers for PTB, both in women at increased risk of PTB because of preterm labour (PTL) or preterm prelabour rupture of the membranes (PPROM), and in asymptomatic women in early pregnancy. An additional aim was to explore immune reaction patterns in PTL and PPROM compared to normal pregnancy. 

Material and methods: In a retrospective registry study, including 20,643 women who delivered during a five-year period, risk factors for Apgar score <7 at five minutes and risk factors for PTB <32 weeks were investigated using univariate and multivariate logistic regression. Furthermore, in a multi-centre mixed case-control and prospective cohort study of women with PTL <34 weeks of gestation (n=80), PPROM (n=40), as well as antenatal controls (n=44) and controls in labour at term (n=40), plasma levels of cytokines and chemokines representing different types of immune responses were analysed with a multiplex bead assay. In addition, an extended protein analysis exploring 92 inflammation-associated plasma proteins using proximity extension assay (PEA) was performed, as well as analysis of 67 different oxylipins by liquid chromatography coupled to tandem mass spectrometry. Finally, the PEA technique was used also to explore plasma proteins in a case-control study including 46 women with PTB and 46 women with normal pregnancies and delivery at term. 

Results and conclusions: A number of partly preventable clinical risk factors for PTB <32 weeks were identified, for example smoking (odds ratio (OR) 1.61 (95% confidence interval (CI) 1.07-2.41)); preeclampsia (OR 5.48 (95% CI 3.39-8.86)); and multiple gestation (OR 15 (95% CI 10-24)). The most evident risk factor for low Apgar scores was PTB; the more preterm the higher the risk. This provides important information for health care professionals, and offers motivations for preventive strategies regarding smoking cessation. 

Both PTL and PPROM were associated with a more pro-inflammatory profile compared to antenatal controls, with an increase in CXCL1. In addition, PTL showed higher CCL17 levels, and PPROM showed higher IL-6 levels compared with normal pregnancy. The inflammatory profile was even higher in labour at term, reflected by higher levels of CXCL1, CXCL8, and IL-6 compared with PTL, probably due to the more advanced stage of the parturition process in these women. 

To identify women with PTL and subsequent PTB <34 weeks of gestation, we found that a combination of the proteins IL-6, IL-17C, IL-10RB, and FGF-23 strongly correlated with PTB <34 weeks with an area under the curve (AUC) of 0.90; inferring a sensitivity of 90%, and specificity of 74%. For the prediction of delivery within 48 hours in women with PTL, the combination of IL-6 and IL-17C displayed an AUC of 0.88, with a sensitivity of 100%, and a specificity of 71%. Additionally, plasma levels of oxylipins were associated with time of birth. Lower levels of 9,10-DiHODE were associated with PTB <34 weeks (adjusted (a) OR 0.12 (0.024-0.62)) and with delivery within 48 hours (aOR 0.13 (0.019-0.93)), while higher levels of 11,12-DiHETre were associated with delivery <34 weeks (aOR 6.19 (1.17-32.7)) and higher levels of 8-HETE were associated with delivery within 48 hours (aOR 5.01 (1.13-22.14)). 

In asymptomatic women, plasma levels of combinations of inflammation-associated proteins in the first and the second trimester also revealed predictive information regarding subsequent risk for PTB <34 weeks. Combining MMP10^trim1, sCD40^trim2, MCSF^trim2, Flt3L^trim2, and FGF-21^diff (diff= difference in protein levels comparing the first and second trimesters) provided a prediction model with an AUC of 0.90. Proteins from the first trimester exclusively (sCD40 and MMP10) rendered an AUC of 0.76. 

This work provides valuable knowledge in the field of PTB and PTL with useful information on risk factors for PTB. Important associations between levels of inflammation-associated proteins and oxylipins with PTB following PTL were found. Before these findings can have clinical implications, they need to be validated in other cohorts. Additionally, in order to be clinically useful as a prediction tool for PTB, a bedside test is needed. Since the PEA technique is PCR-based, this might be achievable. For prediction of PTB in early pregnancy, we have interesting findings with acceptable accuracy based on samples from both the first and the second trimesters. However, as preventive interventions for PTB are preferably initiated early in pregnancy, a prediction tool has better value if it is based on plasma samples from the first trimester. Therefore, we plan to extend this study and evaluate other potential protein biomarkers.