An SR was conducted according to the PRISMA statement (Liberati et al., 2009) and performed using three biomedical literature databases, MEDLINE, EMBASE, and Web of Science, selecting relevant studies concerning mycoplasma presence and preterm labor regardless of delivery at term or preterm (SPTL), SPTL followed by PTB (PTB), or preterm prelabor rupture of the membranes (PPROM). We sought studies between 1980 and 2019 using the following terms: Ureaplasma urealyticum OR U. parvum OR Mycoplasma hominis OR Mycoplasma genitalium AND pregnancy. This study was supported by the Strategic Research Commitment fund from the University of Texas Medical Branch at Galveston, UTMB, and 1R01HD100729 (NIH/NICHD) to R Menon.

Case-control, cross-sectional, prospective, and retrospective cohort studies were considered eligible in this SR, presenting the following criteria: subjects were pregnant women with the target outcomes either SPTL that leads to preterm birth (PTB) or PPROM; detection of genital mycoplasma presence in either maternal or fetal biological compartments during pregnancy. Cross-sectional studies were used in estimating the pooled prevalence of genital mycoplasma among patients with PTL, PTB, and PPROM. On the other hand, case-control and cohort studies were used in determining the association between genital mycoplasma and PTL, PTB, and PPROM.

Reviews, editorials, letters, case reports, conferences, summaries, books, opinions, in vitro studies, studies of diagnostics methodologies, and studies in animal models were excluded, and a search for unpublished studies was not performed. Other characteristics of excluded studies were maternal exposure to infection(s) caused by genital mycoplasmas during pregnancy was not directly measured but mentioned as a possible pathogen; infection by genital mycoplasmas in term infants; only term deliveries; PTB caused by other associated complications, such as preeclampsia, intrauterine growth restriction, stillbirth, diabetes, placental abruption, multiple pregnancies; the full-text of the study was not written in English, Portuguese, Spanish or French.

Two authors (NN-N and MGS) independently performed the screening of the records, first by titles and then by abstracts. Full-text analysis was performed by seven authors (NN-N, JP, OT, MCS, GB, GC, and CM) for articles that fulfilled the predetermined selection criteria for this SR. Discrepancies between the reviewers were discussed until consensus was reached. For each included study, we extracted the following primary data: author’s names, year of publication, type of study, the country where the study was conducted, study objective and design; the total number of participants, the pregnancy outcome, number of cases and controls (when applicable), gestational age or trimester at the time of sample collection, source of the biological sample, species of genital Mycoplasma detected, methodology used to identify genital mycoplasmas, the method for detecting inflammatory biomarkers, type of samples used for biomarker analysis and analytical approaches. The data collected were compared and discussed by the reviewers, and disagreements were resolved by consensus.

Six authors (NN-N, JP, OT, MCS, GB, and GC) assessed the quality of each included study, which was evaluated according to a set of parameters defined for this review using the Newcastle Ottawa Scale for case-control, cohort, and cross-sectional studies (Wells et al., 2013; Lo et al., 2014). The criteria for quality assessment are shown in Supplementary Tables 1–3.

A meta-analysis (MA) was carried out using random-effects models to determine the prevalence of genital mycoplasma with SPTL, PTB, and PPROM, considering all types of studies. For association analysis, only case-control and cohort studies were included. Forest and funnel plots were produced. A p-value of <0.05 was considered statistically significant. Higgins I² was used to assess heterogeneity with I² value of >50% regarded as high heterogeneity. All analyses were performed using the “metafor” package within R (R Core Team, 2020).

We identified 4699 articles, and after the screening process, we retained 121 articles and included these in the qualitative synthesis of this SR (Figure 1). This included 93 articles showing the prevalence and association of genital mycoplasma and adverse pregnancy outcomes, including SPTL, PTB, and PROM, and 51 articles reporting the levels of inflammatory biomarkers in pregnant women with genital mycoplasma and women with any of these adverse pregnancy outcomes. Excluded studies are described in Supplementary Table 4.

Regarding the quality of the included studies, we observed that most of the cross-sectional studies were assessed as satisfactory (17/74), good (33/74), and very good quality papers (20/74). Only four studies were evaluated as unsatisfactory due to low sample size, lack of description on the assessment of pregnancy outcomes, and the analysis did not control for confounding factors. For cohort and case-control studies, most of the papers were assessed as satisfactory (4/47), good (30/47), or very good quality (12/47). Only one paper was evaluated as unsatisfactory due to the low sample size and lack of operational definitions of primary outcomes. However, those unsatisfactory papers were retained in the analysis, as their results regarding mycoplasma prevalence or biomarkers did not influence our study (Supplementary Tables 1–3).

This SR and MA showed a 28% prevalence of U. parvum mycoplasmas among women with PTB, 19% prevalence of U. urealyticum in SPTL, and 27% in women with PPROM. We also showed that detection of genital mycoplasmas was significantly associated with a one to two-fold higher risk for PTB and a two to three-fold higher risk for SPTL and PPROM. However, after subgroup analysis, the prevalence and the association between genital mycoplasma and adverse pregnancy outcomes changes depending on the diagnostic test and the sample used. We observed that genital mycoplasmas have a higher prevalence in maternal samples than fetal samples. Genital mycoplasma presence of the cervix and vagina was associated with lower odds of PTB and PTL.

In contrast, colonization of the AF, fetal membrane, and the placenta was associated with higher odds of PTL. In addition, the association (odds ratio) of GM with adverse pregnancy outcomes changed after subgroup analysis (Tables 1, 2). This further adds to the growing evidence that genital mycoplasma presence in the cervix and the vagina is insufficient to promote preterm birth. Unless the cervix is compromised, it will serve as a barrier protecting the fetus in the intraamniotic cavity (Motomura et al., 2020; Pavlidis et al., 2020). Genital mycoplasma in the cervicovaginal space may promote a mild inflammatory response to prime the cervical tissues to mount an immune response to potentially pathogenic organisms that may try to ascend to the amniotic cavity (Tantengco and Menon, 2022).

Ascending genital mycoplasma to the fetal side very well may cause adversity during pregnancy as the presence in the amniotic cavity or invasion of the fetal organs can mount a massive fetal inflammatory response that can cause SPTL and PTB. This is unlikely for infections limited to the maternal side alone. We propose that the fetal inflammatory response is the final trigger to promote genital mycoplasma-induced PTB, while maternal inflammation due to GM presence helps maintain immunologic homeostasis during pregnancy and promote fetoplacental growth.

Our data partly corroborate the results of two recent SRs, which showed that M. genitalium, but not M. hominis, U. urealyticum, or U. parvum was significantly associated with PTB (Lis et al., 2015; Ma et al., 2021). Those SRs analyzed only three articles reporting PTB with M. hominis or U. urealyticum presence and one article regarding U. parvum. At the same time, our SR also included more recent papers and retried respectively 10, 16, and 5 articles reporting an association between these microorganisms and PTB. It is also important to highlight that one of the two referred SRs included only studies based on NAATs to detect genital mycoplasmas. In contrast, our SR had studies that used other diagnostic tests such as culture and PCR-based tests.

Most of the included studies were from Europe (33/93) or North American (30/93). This limited number of studies from some regions, such as low- and middle-income countries, impaired an estimative of a global prevalence and a more reliable association between the genital mycoplasmas and adverse pregnancy outcomes. Therefore, more studies are required in these regions, especially following a recent SR that showed a higher prevalence of M. genitalium infection in the general population in developing countries compared to developed countries (Baumann et al., 2018). Based on cross-sectional studies, the prevalence of M. hominis among women with SPTL or PTB was >10% in AF samples from United States (Gravett et al., 1986) and Switzerland (Nguyen et al., 2004), in vaginal samples from United States (Gravett et al., 1986; Harger et al., 1991), Israel (Horowitz et al., 1995), and Korea (Choi et al., 2012), and in fetal membranes from United Kingdom (Cahill et al., 2005). With PPROM, over 10% prevalence of M. hominis was reported in vaginal samples from France (Grattard et al., 1995) and Israel (Horowitz et al., 1995), in AF samples from United States (Perni et al., 2004) and Czech Republic (Kacerovsky et al., 2014), and cord blood samples from Czech Republic (Kacerovsky et al., 2014). In contrast, the prevalence of U. urealyticum in SPTL and/or PTB was more than 30% in AF samples from United States (Gravett et al., 1986; Athayde et al., 2000; Romero et al., 2019) and Israel (Horowitz et al., 1995; Bashiri et al., 1999), in vaginal samples from United States (Gravett et al., 1986; Harger et al., 1991), Israel (Holst et al., 1994), France (Grattard et al., 1995; Goffinet et al., 2003), Germany (Abele-Horn et al., 1997), Korea (Choi et al., 2012), Japan (Usui et al., 2002), Denmark/England (Thomsen et al., 1984), and Czech Republic (Koucký et al., 2016), while more than 30% of U. urealyticum in PPROM was reported in AF samples from United States (Romero et al., 1992a; Font et al., 1995; Athayde et al., 2000; Pacora et al., 2000; Perni et al., 2004), Israel (Horowitz et al., 1995), and Austria (Witt et al., 2005), in vaginal samples in France (Grattard et al., 1995), and placenta in Austria (Witt et al., 2005). These data demonstrate that the most prevalent microorganisms have been reported in diverse populations. This review observed significant associations in only two studies from low-income countries. These data demonstrate that the most prevalent microorganisms have been reported in diverse populations. This review observed significant associations in only two studies from low-income countries.

Analyses of cord blood cultures in the Alabama Preterm Birth Study demonstrated that 27.9% of samples from non-white women and 16.8% from white women were positive for U. urealyticum and/or M. hominis (Goldenberg et al., 2008). There was also a lower prevalence of M. hominis in white women, albeit that U. urealyticum presence was unrelated to ethnicity (Witt et al., 2005). Racial or ethnic disparities in adverse pregnancy outcomes associated with the genital mycoplasmas are difficult to establish based on current data as many studies do not report population ethnicity and we strongly recommend inclusion of race/ethnicity of populations in future studies to determine the contribution of racial and genetic factors in the determination of risk.

The relationship between intraamniotic infection and adverse pregnancy outcomes has been examined extensively over the last four decades. In this MA we were able to confirm the prevalence and importance of the genital mycoplasmas in SPTL, PTB and PPROM. The human Ureaplasma spp. are divided into U. urealyticum and U. parvum with 14 known serotypes. In the raw prevalence analysis, ureaplasmas were predominant in the three studied outcomes and was more prevalent in SPTL and PPROM, while U. parvum showed a higher prevalence in PTB. Based on in vitro models (Novy et al., 2009; Motomura et al., 2020) and in vivo (Shimizu et al., 2008; Novy et al., 2009; Dando et al., 2012; Uchida et al., 2013), it has been argued that injection of Ureaplasma and or its antigens may elicit a fetal inflammatory response contributing to PTB, which supports our findings.

Polymicrobial presence demonstrated in PTB (Thomsen et al., 1984; Elliott et al., 1990; McGregor et al., 1990; Hillier et al., 1991; McDonald et al., 1992, 1994; Holst et al., 1994; Grattard et al., 1995; Kundsin et al., 1996; Usui et al., 2002; Goffinet et al., 2003; Goldenberg et al., 2008; Onderdonk et al., 2008; Montenegro et al., 2019), SPTL (Minkoff et al., 1984; Gravett et al., 1986; Lamont et al., 1987; Hillier et al., 1988; Romero et al., 1989, 1992c; McGregor et al., 1990; Harger et al., 1991; Toth et al., 1992; Watts et al., 1992; Holst et al., 1994, 2005; Font et al., 1995; Horowitz et al., 1995; Marconi et al., 2011; Choi et al., 2012; Kim et al., 2012; Lee et al., 2013), and PPROM (Minkoff et al., 1984; Romero et al., 1992b,c; Averbuch et al., 1995; Font et al., 1995; Grattard et al., 1995; Horowitz et al., 1995; Kundsin et al., 1996; Perni et al., 2004; Shim et al., 2005; Nasution et al., 2007), was mostly based on a combination of U. urealyticum and M. hominis (Ma et al., 2021). Co-infections were also associated with decreased gestational age at birth and birth weight, and a significant increase in the incidence of histologic chorioamnionitis (Kwak et al., 2014). However, most of these studies examined only genital mycoplasmas. They did not examine other more virulent Gram-negative or positive associated with mycoplasmas, which may have biased some reports and conclusions.

One of the limitations of this SR is the heterogeneity shown by the large I² values from our MA. This may be due to several factors, such as the variability in the source and timing of biological samples and diagnostic methods for detecting genital mycoplasma.

Like other organisms associated with SPTL, PTB, or PPROM such as G. vaginalis, quantitative analysis of the genital mycoplasmas may be more important than simply the presence or absence of the organism, whether detected by cultivation-dependent or independent techniques (Lamont et al., 2020). Using cultivation-dependent technology in women delivering between 26 and 34 completed weeks of gestation, 80 and 24% of those in SPTL were colonized by U. urealyticum and M. hominis, respectively. In contrast, women who were not in SPTL but delivered electively for maternal-fetal indications such as preeclampsia or intrauterine growth restriction at the same gestational age, 46 and 8%, were colonized by U. urealyticum and M. hominis, respectively. The qualitative analysis was statistically significantly different for U. urealyticum (p < 0.01) but not for M. hominis. Nevertheless, when only heavy presence of M. hominis was considered [>10⁵ color change units (the unit of quantity at the time)], 13 women in the study group (18%) and none of the controls were identified (p < 0.05), suggesting that it was not the presence or absence of the organism that was important but the quantity present (Lamont et al., 1987). This was also supported by various in vitro and in vivo studies showing a dose-dependent increase in inflammation and PTB rate with higher doses of Ureaplasma spp. infections (Glaser et al., 2017, 2018; Motomura et al., 2020; Tantengco et al., 2021).

Only specific subtypes of the genital mycoplasmas are pathogenic. Using molecular-based techniques, it was found that Ureaplasma species consist of 14 serovars from two biovars. The majority of human Ureaplasma isolates belong to U. parvum (biovar 1), comprising four serovars (the predominant biovar in patients with genital tract infections), with U. urealyticum (biovar 2) containing ten serovars isolated much less often. However, the data are limited and conflicted because of the difficulties with traditional genotyping methods (Dhawan et al., 2012). This may be relevant to the confusion about the role of ureaplasmas in adverse outcomes of pregnancy.

The most common method of identifying microbes reported was culture, followed by PCR. Both these approaches are used for both qualitative and quantitative detection of bacteria. Most recent studies used molecular detection of genital mycoplasma instead of culture, as only four studies used microbiological culture after 2010 (Kim et al., 2012; Lee et al., 2013; Wang et al., 2013; Suzuki et al., 2018). Accordingly, detection of genital mycoplasmas may have been underestimated in former studies as the detection rate of mycoplasmas by PCR is higher than that by cultivation (Choi et al., 2012). Molecular analysis has the advantage of differentiating mycoplasma species and allowing the bacterial load to be measured, although bacterial load was reported only in a few included studies. The microbial load can alter the type of biomarker and their quantities (Flores-Herrera et al., 2012; Noda-Nicolau et al., 2016). Sample collections for mycoplasma detection were mainly performed at admission with clinical indications of labor and/or at the time of labor and delivery.

One of the objectives of this study was to determine the association between the genital mycoplasmas and adverse pregnancy outcomes with specific inflammatory biomarkers. However, the data were too heterogeneous to perform a MA. It is still unclear if the genital mycoplasmas can mount a proinflammatory surge that can lead to adverse pregnancy outcomes. Most biomarker studies examined only proinflammatory markers, and few studies report anti-inflammatory mediators. Among PPROM patients, Oh et al. (2010) have demonstrated higher maternal blood WBC and CRP during mycoplasma AF positivity, mainly Ureaplasma spp. (Oh et al., 2010).

Conversely, one major study examined 26 proteins in AF and reported several anti-inflammatory markers (IL-10, sTNFR1, sIL-6r) elevated in cases associated with the genital mycoplasmas, also predominantly by Ureaplasmas (Cobo et al., 2012). In vitro models have suggested that unbiased analysis of pro- and anti-inflammatory markers is required to determine the actual role of diverse mycoplasmas in inducing an inflammatory response during pregnancy (Noda-Nicolau et al., 2016; Polettini et al., 2018). Some genital mycoplasmas such as Ureaplasma spp. may present themselves as proxies for other virulent microbial pathogens such as E. coli, Group B streptococcus (GBS), or Gardnerella species. Therefore, genital mycoplasma might be insufficient to cause SPTL and PTB but can assist other pathogens or modify the tissue environment in a polymicrobial situation.