The burden of preterm birth (PTB, i.e., birth <37 completed weeks of gestation) is highest in low-resource settings such as Africa and Asia (1, 2) where the number of primary studies on the condition is lowest (3). In 2014, sub-Saharan African and Asian countries accounted for more than 80% of global PTBs, with rates in South Africa estimated at 12.4% (1). By 2020, southern Asia and sub-Saharan Africa accounted for approximately 65% of all PTBs globally (2). The PTB rate in sub-Saharan Africa has remained fairly unchanged (10.1%) between 2010 and 2020. However, the total number of preterm babies in this subregion increased by 563,100 between the same decade (2). PTB is a notable cause of neonatal mortality and morbidity (1, 4, 5) and accounted for approximately 18% of the 5.3 million deaths of children under 5 years that occurred in 2019 (4). Several modifiable risk factors including food insecurity, low socioeconomic status, smoking, and alcohol consumption during pregnancy are linked with poor birth outcomes including PTB (6, 7).

Intrauterine infection/inflammation (IUI) is responsible for >40% of PTBs (7–9). Inflammation-induced remodelling of the cervix and foetal membranes that underpins sPTB (10, 11) may vary by maternal demographics, environment (12), and vaginal microbiota (10, 13, 14). For example, in an Asian cohort sampled at the Thailand–Myanmar border, decreased interleukin (IL)-4, interferon gamma (IFN-γ), and tumour necrosis factor (TNF)-α were associated with vaginal microbiota deficient in Lactobacillus crispatus and Finegoldia but rich in Prevotella buccalis (15). In South African women, lactobacilli from a non-optimal microbiota (Nugent score 4–10) induced greater pro-inflammatory cytokine production than those from women with optimal microbiota (Nugent score 0–3) (16). In pregnant Caucasian women living with HIV infection, a suboptimal vaginal microbiota was associated with inflammation and release of matrix degrading enzymes—increased cervicovaginal fluid (CVF) MMP-9 which was related to elevated TIMP-1, IL-1β, IL-8, TNF-α, and polymorphonuclear leucocyte (17).

It has been hypothesised that ethnicity may influence PTB rates through gene-mediated variation in cytokine response to infection. African-Americans with a promoter polymorphism of the MMP-9 gene produce increased MMP-9 that increases the risk for preterm premature rupture of membranes (PPROM) (18). The patho-mechanisms of cytokine-induced matrix degradation and sPTB in South African women without HIV infection are poorly understood. This study reports the relationships between markers of inflammation-driven remodelling of gestational tissues associated with sPTB in South African women. We describe the expression profiles of CVF cytokine and cervical tissue remodelling proteins MMP-9, and its regulator TIMP-1, in pregnant women at risk of PTB, and their associations with sPTB.

This study was approved by the Health Research Authority (HRA) and Health and Care Research Wales (HCRW, 18/LO/2044); Research Ethics Committee of Faculty of Health Sciences, University of Pretoria (145/2019); and University of Cape Town, Faculty of Health Sciences Human Research Ethics Committee (UCT FHS HREC, 196/2019).

We studied the temporal changes in cervicovaginal cytokines and ECM remodelling proteins in a predominantly Black and mixed-race South African cohort in order to better understand the inflammatory responses/profiles that drive the pathway to sPTB in this region/context, which could guide biomarker discovery and formulation of preventive and therapeutic interventions. This study demonstrates that in vaginal swab samples from non-labouring South African women at high risk of sPTB obtained at mid- to late-second and third trimesters, elevated IL-8 at 20–22 weeks (GTP1); IL-1β, IL-8, MMP-9, MMP-9/TIMP-1 ratio, and MMP-9/GM-CSF ratio at 26–28 weeks (GTP2); and IL-1β, IL-8, MMP-9, and TIMP-1 at 34–36 weeks (GTP3) were associated with sPTB, whereas elevated GM-CSF at 20–22 weeks and IL-10 at 26–28 weeks of gestation were associated with delivery at term. Secondly, increased IL-1β assessed at 34–36 weeks was associated with delivery within 2 weeks of index assessment. Thirdly, between 20–22 and 34–36 weeks, MMP-9 increased significantly with associated increase in IL-1β in the preterm women but not in the term-delivered women. The increased MMP-9 particularly at GTP3 indicated a large and significant increase in the probability of sPTB. Furthermore, we observed that the positive (regulatory) correlation between MMP-9 and TIMP-1 in term-delivered women was absent in the preterm women at later gestation (GTP 2 and GTP3) near the onset of labour. Moreover, moderate to late PTB subtypes (i.e., delivery between 32 and <37 weeks) (2, 9) predominated the preterm group.

Increased IL-1β (the master cytokine in inflammation) (24–26) was related to increased MMP-9 and TIMP-1 in the term women at all GTPs. Conversely, increase in IL-10, an anti-inflammatory cytokine (25, 27), was associated with term delivery at GTP2. The positive relationship between IL-1β, MMP-9, and TIMP-1 was also observed in the preterm women but only at GTP1 and GTP2. The breakdown of the positive (regulatory) relationship between MMP-9 and TIMP-1 in the preterm women at later gestation (particularly at GTP3) may be one of the driving mechanisms of sPTB in this cohort. Overexpression of MMP-9 in gestational tissues without a commensurate increase in its inhibitor (TIMP-1) creates dysregulated extracellular matrix (ECM) remodelling that induces premature cervical ripening, PROM, and PTB (10, 28).

Our observation in this South African cohort differs from our previous study of a UK population of similarly asymptomatic high-risk women but predominantly of Caucasian ancestry, which showed no difference in individual CVF cytokine levels between term vs. preterm at 20–22 and 26–28 weeks (19). That is, while the South African women who experienced sPTB had higher concentrations of inflammatory cytokines compared with their term counterparts at 20–22, 26–28, and 34–36 weeks, the preterm-delivered women in the previous UK Caucasian cohort did not show such differences at 20–22 and 26–28 weeks’ gestation, although they were not studied at 34–36 weeks. Only an increase in RANTES and IL-1β from 20 + ⁰–22 + ⁶ to 26 + ⁰–28 + ⁶ weeks expressed as a ratio was higher in preterm vs. term women in the UK cohort (19). Additionally, CVF cytokine concentrations were generally higher at mid-pregnancy in the UK cohort than the South African women. These context-specific variations between South African (majorly Black African and mixed race) versus UK (mainly Caucasian) cohort is consistent with our hypothesis that the mechanism of inflammation-induced sPTB may vary by race and geographical location.

It is also noteworthy that in the current study, IL-6 did not differ significantly between the term vs. preterm women at any GTP. This is despite a recent report of the ability of IL-6 to predict the onset of labour regardless of gestational age at sampling (29). The disparity between the reports is not surprising because while we studied a cohort of predominantly Black African and mixed (White/Black) non-labouring women, the other study (29) reported that IL-6 was predictive of onset of labour in Mexican women in the active phase of labour (regular uterine contractions with cervical dilation >3 cm). However, this report is similar to our previous observation that elevated CVF IL-6 is predictive of sPTB in a cohort of women with established labour, albeit of Caucasian descent (20). This may imply that CVF IL-6 is a major mediator of the inflammatory responses leading to the premature onset of labour in White and Hispanic labouring women, but not implicated in the pathway to spontaneous labour in Black African or women African decent without symptoms of spontaneous preterm labour (PTL). Again, this supports our hypothesis that the inflammatory responses that trigger the onset of PTL are influenced by racioethnic and environmental factors.

Various studies have reported the role of genetic polymorphisms in the regulation of cervical ECM remodelling. Apart from African-Americans with PPROM-associated MMP-9 gene polymorphisms (18), Indian women with the IL-6 G allele express increased MMP-9 and sPTB (30). However, a definite association between TLR-4, IL-1α, IL-1β, IL-6, and IL-10 gene polymorphisms and PTB in diverse ethnic and environmental settings has not been demonstrated (12). Further studies of the impact of gene–environmental interactions on sPTB (31, 32) may clarify these contextual differences in PTB incidence and patterns. Moreover, the inflammatory responses to infectious and non-infectious stimuli (environmental exposure) may be altered by inherent genetic susceptibility mediated by such gene polymorphisms (31, 32).

Herein, we hypothesise that the proteolytic activity of MMP-9 may be elaborated by IL-1β above the anti-proteolytic activity of TIMP-1 in women destined to deliver prematurely. The cytokine inhibitory action of IL-10 may also be diminished especially at the second to third trimester when sPTB is imminent. Conversely, there appears to be a balanced MMP-9/TIMP-1 interplay in the term women as both proteins increased in tandem across all GTPs. This balanced interaction is defective (absent) in the preterm women who demonstrated higher levels of IL-1β, MMP-9, TIMP-1, and MMP-9/TIMP-1 ratio (at GTP2 and GTP3); a positive relationship between all three proteins at GTP1 and GTP2 only; and a decrease in the anti-inflammatory IL-10 (25, 27) at GTP2. The absence of a positive correlation between IL-1β and the regulatory TIMP-1 in the preterm women at GTP3 could possibly contribute to the premature stimulation of labour pathways.

Because we excluded women with clinical genital tract infection, and most of the preterm women experienced moderate to late sPTB (32 to <37 weeks) (2, 9), we hypothesise that the inflammation-associated sPTB in this cohort is more likely of the non-infectious nature. This is because earlier PTB (<30 weeks) is commonly triggered by lower reproductive tract infection ascending into the intrauterine/intraamniotic cavity (9, 33). That is, the lower the gestational age at presentation with spontaneous PTL (with intact or ruptured membranes), the higher the frequency of abnormal genital tract microbiota and intrauterine/intraamniotic infection (34–38). Infection and infection-driven inflammation are more prevalent (up to 79% in some studies) in extreme sPTB (<28 weeks) (39, 40).

Other acute pro-inflammatory cytokines such as IL-8 (25, 41), which was also elevated, may act synergistically with or independent of IL-1β to perpetuate this inflammatory sPTB drive regardless of the stimulus. Our observations are similar to other studies (42–45) that have also reported upregulation of this mediator in CVF, amniotic fluid, cervix, chorioamnion, and placenta of preterm-delivered women, and supported by strong predictive accuracy (>70% and predictive values between 42% and 100%) for imminent sPTB.

The predictive potentials of the markers for sPTB and delivery within 2 weeks of assessment at 34–36 weeks observed in this study appear comparable with clinically employed tests such as transvaginal ultrasound cervical length measurement and quantitative cervicovaginal foetal fibronectin (20, 46, 47), which are currently employed in high-resource settings, but not in settings with limited resources such as sub-Saharan Africa. Additional studies investigating whether these biomarkers (especially IL-1β) may predict delivery within 2 weeks of assessment of women presenting with symptoms of PTL to inform clinical care (48, 49) are required.

The patho-mechanisms of MMP/TIMP-induced ECM modulation during pregnancy and parturition have been extensively reported (10, 17, 50). Labour-associated inflammatory processes are mediated by the overexpression of IL-1β, IL-6, IL-8, MCP-1, RANTES, and TNF-α in gestational tissues including foetal membranes, cervix, amniotic fluid, CVF, and placenta (7, 19, 20, 25, 30, 51–53). Expression of pro-inflammatory mediators in turn causes (a) increased elaboration of prostaglandins (PG) and contraction-associated proteins, which promote uterine contractility; (b) degradation of the chorioamnion ECM; and (c) remodelling of the cervix by MMP-9 linked to decreased expression of TIMPs (7, 10, 51, 54–56).

Overproduction of IL-1β (as we observed) characterises labour with or without infection (57, 58) by inducing the labour pathway (57), especially in IUI-associated PTB (25). IL-1β increases production of MMPs such as MMP-9 in gestational tissues by increasing production of cyclooxygenase-2 and PGE₂, vascular permeability, and immunocyte infiltration (51, 59–62). IL-1β can increase the production/activity of MMP-9 independently (63, 64) or in synergy with TNF-α (65), suggesting a causal relationship between IL-1β and IUI-associated PTB (51)—both intra-amniotic (64, 66–68) and systemic (69) infusion of IL-1β induced PTL and PTB.

On the contrary, IL-10 obliterates infectious induction of MMP-2 and -9 in chorioamnion (70), and cytotrophoblast MMP-9 activity/invasiveness in an autocrine fashion (71). IL-10 decreases MMP-9 activity and increases TIMP-1 levels in coculture systems in a dose-dependent manner (72). IL-10 reduces intra-amniotic concentrations of pro-inflammatory mediators and incidence of prematurity in animals challenged with lipopolysaccharide, Escherichia coli, and IL-1β (67, 73, 74). In line with these anti-inflammatory roles, we observed a decrease in IL-10 in term-delivered women at GTP2, which distinguished them from the preterm-delivered women with high sensitivity, specificity, PPV, NPV of >62%, and a +LR of 4.8. The role of IL-10 in regulating labour-associated inflammatory responses at different GTPs requires more race/environment-specific investigations.

Additionally, an imbalance in M1/M2 macrophage activation/polarization, which determines the fate of tissues in inflammation or injury (75–77), can plausibly be deduced from our data. The decrease in GM-CSF, an activator of M1 macrophages (75, 78), in the term women at GTP1 could be associated with decreased or regulated inflammation in these women. GM-CSF can induce the expression of IL-10 at the human maternal–foetal interface (79). Moreover, women who deliver at term showed a similar but early- (6–12 weeks) to mid-gestation (24 weeks) decline in plasma GM-CSF (80). Although the gestation-associated changes in GM-CSF were not significant in the preterm group, a decrease in GM-CSF correlated with an increase in CCL2, an activator of M2 macrophages (78) at GTP1. However, this relationship was absent at GTP3. Furthermore, CVF concentrations of both cytokines increase in women with a short cervix (81). Nonetheless, the role of M1/M2 macrophage polarisation in sPTB mediated by GM-CSF-CCL2 interaction is currently unclear.

Our study may be the first to investigate longitudinal changes in inflammatory mediators and matrix remodelling proteins in relation to sPTB in an African cohort of non-labouring women at risk of PTB. Although our findings are intriguing, interpretation should be with caution due to the relatively small and limited sample size. For instance, due to the limited number of women who provided samples at all three GTPs, there were no significant differences in the proteins when we performed a pairwise comparison between GTP1 and GTP2, GTP2 and GTP3, or GTP1 and GTP3 for either term or preterm women. Future studies are required to explore such comparisons as well as our previous observations that women presenting with symptoms of PTL who delivered within 2 weeks of index assessment demonstrated elevated RANTES, IL-6, and TNFR1 (20). Additionally, a linear mixed-effects model of pro-inflammatory cytokines and ECM remodelling proteins was not performed as the dependent variable (outcome) in this study was binary (sPTB or term) instead of continuous.

In this cohort, sPTB is associated with a gestation-dependent increase in acute and chronic pro-inflammatory cytokines, a decrease in IL-10 and GM-CSF, and dysregulated MMP-9-TIMP-1 interaction. Levels of cytokine (especially IL-1β) and ECM remodelling proteins rise significantly in the final 2 weeks before the onset of labour when sPTB is imminent. Whether this imbalance is modulated by IL-1β and perhaps IL-10 and contributes to the extracellular tissue remodelling and collagen degradation that precede birth remains to be determined.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving humans were approved by Health Research Authority (HRA) and Health and Care Research Wales (HCRW, 18/LO/2044); Research Ethics Committee of Faculty of Health Sciences, University of Pretoria (145/2019); and University of Cape Town, Faculty of Health Sciences Human Research Ethics Committee (UCT FHS HREC, 196/2019). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

EA: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. NI: Data curation, Investigation, Methodology, Resources, Validation, Writing – review & editing. AO: Data curation, Investigation, Methodology, Project administration, Resources, Validation, Writing – review & editing. PS: Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Validation, Visualization, Writing – review & editing. MM: Data curation, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing – review & editing. DA: Writing – original draft, Writing – review & editing, Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization.