Received: 17 January 2022 | Revised: 2 August 2022 | Accepted: 24 August 2022 DOI: 10.1111/aogs.14460 ORIGINAL RESEARCH ARTICLE Cervical pessary and cerclage placement for preterm birth prevention and cervicovaginal microbiome changes Mireia Vargas1 | Francisca Yañez2 | Andrea Elias2 | Andrea Bernabeu1,3 | Maria Goya1,4 | Zixuan Xie2 | Alba Farrás1 | Olga Sánchez4,5 | Zaida Soler2 | Carlos Blasquez2 | Leonor Valle6 | Anna Olivella7 | Begoña Muñoz8 | Maia Brik9 | Elena Carreras1,4 | Chaysavanh Manichanh2 1 Department of Obstetrics, Maternal-­Fetal Medicine Unit, Hospital Universitario Vall d'Hebron, Universitat Autònoma de Barcelona, Barcelona, Spain 2 Maternal and Child Health and Development Network (SAMID), Instituto de Salud Carlos III, Madrid, Spain 3 Department of Reproductive Medicine, Instituto Bernabeu, Alicante, Spain 4 Microbiome Lab, Hospital Universitario Vall d'Hebron; Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD); Department of Medicine, Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Spain 5 Women and Perinatal Health Research Group, Institut d'Investigació Biomèdica de Sant Pau (IIB-­Sant Pau). Hospital Universitari de Sant Pau, Barcelona, Spain 6 Maternal-­Fetal Medicine Unit, Department of Obstetrics, Hospital Universitario Materno-­Infantil de Canarias, Las Palmas de Gran Canaria, Spain 7 Department of Obstetrics, Hospital de Sant Pau, Barcelona, Spain 8 Department of Obstetrics and Gynecology, Hospital Universitari Sant Joan de Reus, Reus, Spain 9 Department of Obstetrics, Hospital Universitario de Torrejón, Universidad Francisco de Vitoria, Madrid, Spain Correspondence Maria Goya, Maternal-­Fetal Medicine Abstract Unit, Department of Obstetrics, Hospital Introduction: Our objective was to compare the vaginal microbiome in low-­risk and Universitari Vall d'Hebron, Universitat Autònoma de Barcelona, Passeig de la high-­risk pregnant women and to explore a potential association between vaginal Vall d'Hebron, 119-­129, 08036 Barcelona, microbiome and preterm birth. Spain. Email: maria.goya@vallhebron.cat Material and methods: A pilot, consecutive, longitudinal, multicenter study was con- ducted in pregnant women at 18–­22 weeks of gestation. Participants were assigned Funding information ANID BECAS chile, Grant/Award Number: to one of three groups: control (normal cervix), pessary (cervical length ≤25 mm) and Francisca Yáñez was supported by a cerclage (cervical length ≤25 mm or history of preterm birth). Analysis and compari- grant from AN; Beca Dexeus Mujer 2015; Horizon 2020 Framework Programme, son of vaginal microbiota as a primary outcome was performed at inclusion and at Grant/Award Number: Zixuan Xie 30 weeks of gestation, along with a follow-­up of pregnancy and perinatal outcomes. received a fellowship from the European; Instituto de Salud Carlos III, Grant/Award We assessed the vaginal microbiome of pregnant women presenting a short cervix Number: FIS PI15/02043; Universidad with that of pregnant women having a normal cervix, and compared the vaginal micro- Francisco de Vitoria biome of women with a short cervix before and after placement of a cervical pessary or a cervical cerclage. Abbreviations: CL, cervical length; CST, community state type; PCR, polymerase chain reaction; sPTB, spontaneous preterm birth; WG, weeks of gestation. Mireia Vargas, Francisca Yañez, Andrea Elias, Andrea Bernabeu, Elena Carreras and Chaysavanh Manichanh contributed equally. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes. © 2022 The Authors. Acta Obstetricia et Gynecologica Scandinavica published by John Wiley & Sons Ltd on behalf of Nordic Federation of Societies of Obstetrics and Gynecology (NFOG). Acta Obstet Gynecol Scand. 2022;101:1403–1413.  wileyonlinelibrary.com/journal/aogs | 1403 1404 | VARGAS et al. Results: The microbiome of our control cohort was dominated by Lactobacillus crispa- tus and inners. Five community state types were identified and microbiome diversity did not change significantly over 10 weeks in controls. On the other hand, a short cer- vix was associated with a lower microbial load and higher microbial richness, and was not correlated with Lactobacillus relative abundance. After intervention, the cerclage group (n = 19) had a significant increase in microbial richness and a shift towards com- munity state types driven by various bacterial species, including Lactobacillus mulieris, unidentified Bifidobacterium or Enterococcus. These changes were not significantly observed in the pessary (n = 26) and control (n = 35) groups. The cerclage group had more threatened preterm labor episodes and poorer outcomes than the control and pessary groups. Conclusions: These findings indicate that a short cervix is associated with an altered vaginal microbiome community structure. The use of a cerclage for preterm birth prevention, as compared with a pessary, was associated with a microbial community harboring a relatively low abundance of Lactobacillus, with more threatened preterm labor episodes, and with poorer clinical outcomes. KEYWORDS 16S rRNA gene, cerclage, cervical length, community state type, microbial community, microbial diversity, microbial richness, pessary, vaginal bacterial load 1 | I NTRO D U C TI O N Key message Spontaneous preterm birth (sPTB) remains the leading cause of neo- Our study suggests that a short cervix and cerclage inter- natal morbidity and mortality worldwide.1 Although improvements vention may be associated with preterm birth risk due to in neonatal care have led to higher survival rates, 2–­6 perinatal out- changes in the microbiome. comes can only be improved with a more accurate identification of pregnant women at risk of sPTB and appropriate interventions for prevention.7 Pregnant women with a short cervix (≤25 mm) before 24 weeks 2 | M ATE R I A L A N D M E TH O DS of gestation (WG) are at a high risk of sPTB. 8 Early identification of these pregnancies allows timely and targeted intervention Between March 2016 and April 2018, an observational, lon- with progesterone therapy, cervical cerclage or cervical pessary.9 gitudinal study was conducted in five hospitals across Spain Recent studies have highlighted the influence of the vaginal micro- (Hospital Universitari Vall d'Hebron, Hospital de Torrejón de biome composition on cervical length during pregnancy10–­12 and Ardoz, Hospital Sant Joan de Reus, Hospital Materno-­Infantil de different susceptibilities to adverse outcomes when Lactobacillus Canarias and Hospital de Sant Pau de Barcelona) to examine the crispatus is absent or present only at a low level in the vaginal mi- vaginal microbiome of pregnant women at risk of sPTB, having ei- crobiome and displaced by Lactobacillus iners, Gardnerella vagina- ther cervical pessary or cervical cerclage intervention, vs low-­risk lis or other bacterial species.10,11,13–­16 Similarly, recent data have patients. Cervical length (CL) was measured according to the Fetal shown the importance of vaginal bacterial load on determining Medicine Foundation guidelines at the routine second-­trimester preterm birth risk, which should be considered when interpreting ultrasonography. 17 study findings. The study participants were assigned to one of three groups: The primary aim of our study was to assess the vaginal micro- control, pessary and cerclage. Women with a CL >25 mm were as- biome of pregnant women with a short cervix and compare it with signed to the control group, and women with a CL ≤25 mm were as- that of pregnant women who had a normal cervix. A secondary aim signed either to the pessary or the cerclage group based on their was to compare the vaginal microbiome of women with a short obstetric history according to the hospital's protocol. Exclusion cri- cervix before and after placement of a cervical pessary or a cer- teria were major fetal abnormalities, painful regular uterine contrac- vical cerclage. tions, active vaginal bleeding, ruptured membranes, placenta previa, VARGAS et al. | 1405 history of cone biopsy, previous or current progesterone treatment Cervicovaginal swabs were collected from all participants at in- and any disorder or any medication, such as antibiotics, that might be clusion (between 18 and 22 WG) before the intervention (baseline, associated with alterations in the vaginal microbiome. second trimester swab) and at around 30 WG (third-­trimester swab). F I G U R E 1 Flow chart of the study design. 1406 | VARGAS et al. One-­size cervical pessaries (65 × 25 × 32 mm) for preventing Cervicovaginal swabs were collected (Deltalab Amies; Copan sPTB were purchased from Dr Arabin GmbH & Co. A pessary of a Italia, LQ Amines) during a speculum examination. Swabs were different size was used only in patients who had problems with the stored at −80°C until processed. For profiling microbiome composi- normal size. Cervical cerclage was performed with a Mersilene band tion, genomic DNA was extracted from the swabs and the V4 region by the modified Shirodkar technique.18 of the 16S rRNA gene was amplified and sequenced by polymerase Patients assigned to the control group were encouraged to follow chain reaction (PCR) following a previously described method. 21 a low-­risk approach during pregnancy. Patients assigned to the pes- Sequencing details are shown in Table S2. For microbial load de- sary or cerclage groups were given detailed information on potential termination, the V4 region of the 16S rRNA gene was amplified by side effects, such as vaginal discharge and other symptoms. Placement real-­time quantitative PCR (qPCR) following a previously described 19,20 of pessary and cerclage was guided by transvaginal ultrasound. method. 22 Pessaries and cerclages were removed at 37 WG. In this study, no pa- Differences in clinical variables were assessed using the SPSS 19 tient received progesterone treatment for preterm birth prevention. software. The chi-­square test was used for categorical variables and Indications for pessary/cerclage removal before term were: ac- the non-­parametric Mann–­Whitney U-­test was used for continuous tive vaginal bleeding, severe discomfort or persistent uterine con- variables, with statistical significance set at p < 0.05. Microbial se- tractions despite tocolytic treatment, suspected chorioamnionitis, quence data was analyzed using the QIIME 2 software. The alpha and onset of labor. group significance plugin to test for differences in alpha-­diversity The primary outcome was the comparison of the vaginal micro- was implemented in QIIME 2. Alteration of the microbiome com- biome in pregnant women with a short cervix (cerclage or pessary munity could also be appreciated at the alpha-­diversity level, as groups) vs pregnant women with a normal cervix (control group). calculated by the Chao121 and Shannon22 indices. The Chao1 rich- Secondary outcomes are shown in Table S1. ness estimator gives weight to the low-­abundant species, as it only TA B L E 1 Demographic characteristics Control Pessary Cerclage of all three study groupsa (n = 35) (n = 26) (n = 19) p value Mean maternal age (years) 30.6 (5.8) 31.0 (6.2) 33.8 (5.1) 0.125 Pre-­conception BMI (kg/m2) 24.4 (4.4) 24.7 (4.9) 26.0 (5.3) 0.460 Ethnicityb 0.250 African American 3 (8.6) 1 (3.8) 2 (10.5) Asian 1 (2.8) 0 0 Arab 0 3 (11.5) 0 Caucasian 24 (68.6) 20 (76.9) 14 (73.7) Hindu 0 0 1 (5.3) Maghrebi 1 (2.8) 0 1 (5.3) South American 6 (17.1) 2 (7.7) 1 (5.3) Smoking status during pregnancy 0.689 No 16 (45.7) 15 (57.7) 10 (52.6) <5 cigarettes per day 16 (45.7) 6 (23.1) 6 (31.6) 6–­10 cigarettes per day 1 (2.8) 1 (3.8) 1 (5.3) 11–­20 cigarettes per day 2 (5.7) 3 (11.5) 2 (10.5) >20 cigarettes per day 0 1 (3.8) 0 Assisted reproduction techniques 0 0 3 (15.8) 0.124 Obstetric history Parous with no previous preterm 15 (42.8%) 15 (57.6) 13 (68.4%) 0.176 birth Parous with a least one previous 1 (2.8) 6 (23.1) 8 (42.1) 0.002c preterm birth Gestational age at inclusion (weeks) 20.4 (0.9) 20.8 (1.3) 20.6 (1.5) 0.100 Cervical length at inclusion (mm) 39.3 (6.9) 19.1 (4.0) 20.9 (3.9) <0.0001c a Continuous variables are shown as the median (standard deviation); categorical variables are shown as the number of cases (%). b Race was self-­reported. c Statistical significance was set at p < 0.05. VARGAS et al. | 1407 accounts for singletons and doubletons, whereas the Shannon transform = “LOG”, correction = “BH”, analysis_method = “LM”, index accounts for both the abundance and evenness of the spe- max_significance = 0.25 (default significance threshold), min_abun- cies present, and therefore gives weight to high-­abundant species. dance = 0.0001, min_prevalence = 0.1. Age was added as a fixed The ANCOM plugin, embedded in QIIME 2, was used to test for effect. The participant identification number was added as a ran- differences in microbiota composition among groups according to dom effect. Results with a false discovery rate <0.05 were consid- treatment and/or time point. Community state types (CSTs) were ered significant. A logistic regression model (statistics package, glm assigned to each sample using hierarchical clustering with Euclidean function) was used to predict the effect of Lactobacillus abundance distance metrics and complete linkage methods as described in the and treatment type (pessary or cerclage) on timing of birth (term or literature. 23 preterm). to investigate the association between microbiome data and Differences in microbial load among groups were assessed clinical variables, we used linear mixed models as implemented using the non-­parametric Wilcoxon signed-­r ank test (paired in the Multivariable Association with Linear Models (MaAsLin2) pairwise comparisons) and the Mann–­Whitney U-­test (unpaired package (https://doi.org/10.1101/2021.01.20.427420). MaAsLin2 pairwise comparisons). The Kruskal–­Wallis test was used for con- was set up with the following parameters: normalization = “TMM”, tinuous variables and the chi-­square test for categorical variables. TA B L E 2 Obstetrical outcomes in all Control Pessary Cerclage three study groupsa (n = 35) (n = 26) (n = 19) p value Cervical length after treatment (mm) -­ 27.6 (6.2) 23.4 (4.6) 0.067 Cervical length at 30 weeks (mm) 38.9 (8.6) 26.8 (4.2) 16.3 (7.1) 0.001b Symptoms at 30 weeks 0.01b No 34 (97.1) 20 (76.9) 18 (94.7) Yes 1 (2.9) 6 (23.1) 1 (5.3) Cervical length at 37 weeks (mm) -­ 16.7 (12.4) 14.0 (19.8) 0.347 Symptoms at 37 weeks 0.222 No 34 (97.1) 21 (76.9) 18 (94.7) Yes 1 (2.9) 5 (19.2) 1 (5.3) Threatened preterm labor episode 1 (2.9) 5 (19.2) 7 (36.8) 0.005b Tocolytic treatment 1 (2.9) 5 (19.2) 7 (36.8) 0.005b Corticosteroid therapy 2 (5.7) 7 (26.9) 10 (52.6) 0.001b Duration of maternal hospital stay due 1.0 (1.2) 1.7 (2.5) 4.8 (6.7) 0.045b to threatened preterm labor Chorioamnionitis 1 (2.9) 1 (3.8) 2 (10.5) 0.387 Bleeding during pregnancy 3 (8.6) 0 1 (5.3) 0.315 PPROM 1 (2.9) 3 (11.5) 2 (10.5) 0.377 Antepartum death 0 0 1 (5.3) 0.197 Spontaneous delivery <24 weeks 0 0 1 (5.3) 0.197 Spontaneous preterm birth <28 weeks 0 2 (7.7) 3 (15.8) 0.068 Spontaneous preterm birth <32 weeks 0 2 (7.7) 5 (26.3) 0.005b Spontaneous preterm birth <34 weeks 0 3 (11.5) 6 (31.6) 0.002b Spontaneous preterm birth <37 weeks 4 (11.4) 7 (26.9) 7 (36.8) 0.082 Any type of delivery <34 weeks 0 3 (11.5) 6 (31.6) 0.002b Gestational age at delivery (weeks) 38.7 (1.8) 36.8 (3.8) 34.5 (5.9) 0.003b Type of delivery 0.704 Vaginal 22 (62.9) 19 (73.1) 14 (73.7) Instrumental 5 (14.3) 2 (7.7) 1 (5.3) Elective Cesarean 3 (8.6) 1 (3.8) 0 Emergency Cesarean 5 (14.3) 4 (15.4) 4 (21.0) a Continuous variables are shown as the median (standard deviation); categorical variables are shown as the number of cases (%). b Statistical significance was set at p < 0.05. 1408 | VARGAS et al. TA B L E 3 Neonatal outcomes among Control Pessary Cerclage the three study groupsa (n = 35) (n = 26) (n = 19) p value Birthweight (grams) 3137.0 (383.1) 2800.7 (814.5) 2532.6 (1084.4) 0.158 Number of days admitted to 0.5 (2.3) 6.1 (20.1) 11.9 (23.3) 0.056 NICU Total days of neonatal admission 1.4 (2.5) 9.5 (24.0) 12.1 (26.9) 0.002b Neonatal death 0 1 (3.8) 2 (10.5) 0.134 Birthweight <1500 g 0 1 (3.8) 6 (31.6) 0.03b Birthweight <2500 g 2 (5.7) 11 (42.3) 8 (42.1) 0.001b Necrotizing enterocolitis 0 0 2 (10.5) 0.031b Intraventricular hemorrhage 0 1 (3.8) 2 (10.5) 0.262 Respiratory distress syndrome 0 2 (7.7) 3 (15.8) 0.058 Retinopathy 0 0 0 -­ Treatment for sepsis 0 1 (3.8) 3 (15.8) 0.030 b Composite neonatal adverse 0 3 (11.5) 4 (21.0) 0.022b outcomes a Continuous variables are shown as the median (standard deviation); categorical variables are shown as the number of cases (%). b Statistical significance was set at p < 0.05. P-­values were subjected to multiple hypothesis testing correction p < 0.0001, respectively). CL increased in the pessary and cerclage using the Benjamini–­Hochberg method with a false discovery rate groups after intervention (Table 2); however, this increase was threshold of 0.05. More details about the sequence analysis can be greater in the pessary group (p = 0.001). The cerclage group had found in Table S2. more episodes of threatened preterm labor, greater need for treat- ment with tocolytics and corticosteroids, and longer maternal hos- pital stay due to threatened preterm labor (p = 0.005, p = 0.001 2.1 | Ethics statement and p = 0.045, respectively, Table 2). The preterm birth rate be- fore 32 weeks, 34 weeks and 37 weeks was higher in the cerclage The protocol was approved by the local Ethics Committee of each group than in the control and pessary groups (26.3% vs. 0 vs. 7.7%, participating hospital (registration number PR[AMI] 24/2016 and p = 0.005; 31.6% vs. 0 vs. 11.5%, p = 0.002; 36.8% vs. 11.4% vs. date of issue February 1, 2016). All participants provided their writ- 26.9%, p = 0.082). Gestational age at delivery was also lower in the ten informed consent. cerclage group than in the control and pessary groups (34.5 weeks vs. 36.8 weeks vs. 38.7 weeks, p = 0.003). Finally, the cerclage group had poorer perinatal outcomes (Table 3), such as more necrotiz- 3 | R E S U LT S ing enterocolitis, need for sepsis treatment and poorer composite neonatal outcomes; all differences were statistically significant Among 114 screened pregnant women, 80 met the inclusion cri- (p = 0.031, p = 0.03 and p = 0.022, respectively). teria (Figure 1): 35 women had CL >25 mm (control group) and 45 The microbiome in pregnant women with a normal CL was repre- were considered to be at risk of sPTB (CL ≤25 mm or history of sented by 10 genera that accounted for more than 99% of the total preterm birth) and were recommended to have either a pessary sequencing data. Among those, Lactobacillus (89%) was followed by (n = 26) or a cerclage (n = 19). The 80 included pregnant women Gardnerella (4.3%), Bifidobacterium (3.1%), and Enterococcus (0.5%). At had a gestational age ranging from 18 to 22 weeks. Vaginal swab the species level, Lactobacillus iners was the dominant species (46.3% samples were collected at around 20 WG (baseline, second tri- on average), followed by Lactobacillus crispatus (33.3%), Lactobacillus mester swab) and 30 WG (third trimester swab), with a total of mulieris (6.8%), Gardnerella vaginalis (4.8%), unknown Bifidobacterium 152 swabs collected. One-­s ize pessary (the recommended size) (4%), unknown Enterococcus (1.8%) and Lactobacillus gasseri (1%) was used for all patients included. For the seven women who (Figure 2A). L. iners and L. crispatus, the two most abundant species, delivered before 30 WG, only one sample was collected (at were exclusive of each other (rho = −0.56, p = 6.8E-­14). Microbiome baseline). composition in the control group did not change significantly over the Demographics (Table 1) among groups were similar (p > 0.05). As 10-­week period at any specific taxonomic level. Alpha-­diversity analy- expected, unfavorable obstetric history and short cervix (Table 1), sis indicated an increased diversity (according to the Shannon index of with at least one previous sPTB, were more prevalent in the pes- diversity) of the most abundant species. Furthermore, microbial load in sary and cerclage groups than in the control group (p = 0.002 and the control group did not change significantly between the second and VARGAS et al. | 1409 F I G U R E 2 Vaginal microbial community of healthy pregnant women in their second and third trimesters. (A) Taxonomic profiles of the swab samples at the second (first barplot from the bottom) and third trimester (second barplot from the bottom) for each participant based on sequence data of the 16S rRNA V4 region (V4-­16S). (B) Representation of clustering of the vaginal microbiome into the five Community State Types (CST I –­V ). (C) Microbial alpha-­diversity based on Chao1 and Shannon indices of the V4-­ 16S sequence data, (D) Bacterial load as assessed by amplification of the 16S rRNA gene using quantitative PCR. third trimesters (Figure 2C). Over the 10-­week follow-­up period, only correlation between Lactobacillus relative abundance and CL five participants had a shift in CSTs: one patient shifted from CST I to was observed only for CL >25 mm (Spearman correlation test, CST II, one patient from CST III to I, two patients from CST IV to III, and rho = 0.43; p = 0.004); however, no such correlation was observed one patient from CST V to IV (Figure 2B). for CL ≤25 mm (Spearman correlation test, rho = 0.227, p = 0.16) Regarding the effect of a short cervix on the vaginal micro- (Figure 3C). Lactobacillus relative abundance was lower in the biome (Figure 3A), pregnant women with a CL ≤25 mm had a sig- CL ≤ 25 group than in the CL > 25 group (Mann–­W hitney test, nificantly lower number of 16S rRNA gene copies/ng (p = 0.01), p = 0.02). Preterm birth was more prevalent in the cerclage group suggesting that a longer cervix favors microbial growth. However, (logistic regression model, p = 0.008) but was not associated with a short cervix was associated with a greater microbial richness any bacterial species, although a trend was found for Lactobacillus (p = 0.009) and also with a trend towards a greater microbial even- mulieris (logistic regression model, p = 0.15). ness (p = 0.15) (Figure 3B). Diversity was not significantly cor- Regarding the effect of treatment on vaginal microbiome dy- related with CL, although there was a certain trend (rho = –­0 .2, namics, microbial diversity analysis indicated an increased micro- p = 0.075). These differences in microbial load and alpha-­diversity bial richness (p = 0.03; Chao1 index, Wilcoxon test) and increased were not associated with differences in the microbial profile microbial load (p = 0.018, Wilcoxon test) at 30 WG in the cerclage (ANCOM method), as the relative abundances of L. iners and L. group as compared with the control group (Figure 4A). These results crispatus were similar to those of the control group. A positive were confirmed using multivariate association with linear models 1410 | VARGAS et al. F I G U R E 3 Association between cervical length and microbial load, microbial diversity and Lactobacillus relative abundance at 20 WG. (A) Microbial load as assessed by amplification of the 16S rRNA gene V4 region using quantitative PCR for cervical length (CL) >25 mm and CL ≤25 mm. (B) Microbial alpha-­diversity as assessed by the Chao1 and Shannon indices on 16S sequence data for CL >25 mm and CL ≤25 mm (Chao1 index, P = 0.009; Shannon index, P = 0.15). (C) Correlation between CL and Lactobacillus relative abundance based on 16S rRNA sequence data (rho = 0.43, P = 0.04 for CL >25 mm). (false discovery rate = 0.001 for bacterial load and false discovery abundance from 24 WG until birth. Our data did not support this rate = 0.02 for microbial richness). Microbial richness and microbial finding between 20 and 30 WG, but did agree with other studies load did not differ significantly between the pessary and control on the stability of Lactobacillus relative abundance over all three groups. Interestingly, a negative correlation was observed between trimesters. 27 However, this stability may be related to ethnicity, as microbial richness and gestational age at delivery (rho = −0.24, suggested by other studies. 28 Our analysis identified five CSTs. CST p = 0.03; Figure 4B,C). Regarding CST analysis, the odds of shifting I, II, III and V were dominated by Lactobacillus crispatus, Lactobacillus from CSTs I, II and V to CST IV were 15.1 times higher in the cerclage gasseri, Lactobacillus iners, and G. vaginalis, respectively. CST IV was group than in the control group (p = 0.005) and 10.2 times higher in dominated by different microbial groups, such as Lactobacillus mu- the cerclage group than in the pessary group (p = 0.02) (Figure 4D,E). lieris, unidentified Bifidobacterium or Enterococcus.19 These differ- ences may also be related to ethnicity. Analysis of the groups at risk of sPTB due to a short CL showed 4 | DISCUSSION that several variables, such as microbial load, Lactobacillus relative abundance and alpha-­diversity indices, differed significantly from Our study agrees with previous reports on microbiome composition those of the group with normal CL. The fact that a short cervix is in healthy pregnant women as being dominated by CST I (L. crispa- associated with low microbial load and high microbial richness sug- tus) and CST III (L. iners), with variations in CST IV (others) and CST V gests that a diverse vaginal microbiome may not favor the growth (G. vaginalis). These CSTs were relatively stable over time with few of usually dominant commensals, such as Lactobacillus. Regarding shifting from CST to another, but diversity increased slightly with diversity, our results agree with a previous study reporting a higher gestational age. A short cervix was associated with a lower micro- microbial richness in women at risk of sPTB.15 However, our results bial load and a higher microbial richness, the latter being negatively for microbial load seem to contradict those of Freitas et al.,15 re- associated with gestational age at delivery. Prevention of sPTB by ported a higher microbial load in the group at risk of sPTB. cervical cerclage was associated with a microbial community har- The cerclage group had a lower increase in CL and more threat- boring a relatively low abundance of Lactobacillus. Furthermore, the ened preterm labor episodes than the pessary group. In addition, cerclage group had more threatened preterm labor episodes and at the vaginal microbiome level, the risk of shifting to a community poorer clinical outcomes than the other groups. with a lower relative abundance of Lactobacillus was higher in the The vaginal microbiome of our control cohort was dominated cerclage group. These findings agree with a previous study compar- by Lactobacillus, as reported by previous studies. 24,25 Indeed, lacto- ing braided suture, such as Mersilene, which was used in our study, bacilli are key players in the vaginal microbiome, probably through with monofilament; these authors observed that braided suture in- the production of lactic acid and the resulting decrease in pH.12,26 duced a persistent shift towards vaginal microbiome dysbiosis char- Ramussen26 reported a gradual decrease in Lactobacillus relative acterized by a reduced Lactobacillus spp. abundance and pathobiont VARGAS et al. | 1411 F I G U R E 4 Microbial diversity, load and CST over time. (A) Evolution of microbial alpha-­diversity according to the Chao1 and Shannon indices as fold-­changes between the second and third trimesters. Only the cerclage group showed a significant increase in microbial richness. (B) Increase in microbial load at around 30 WG in the cerclage group vs the pessary group. (C) Chao1 index at baseline vs gestational age at delivery plot, showing a negative correlation (Spearman test, rho = −0.24, P = 0.03). (D) CST shift over time. The shift to CST IV contributed to the significance (P = 0.01). (E) CST (I, II, III, V) shift to CST IV was higher in the cerclage group than in the control or pessary groups (odds ratio 15, Wald test, P = 0.005) and pessary (odds ratio 10.2, Wald test, P = 0.02) groups. Gray cells highlight the missing samples owing to PTB before week 30; green cells indicate no shift in CST; yellow cells indicate shift to any CST; orange cells indicate shift from any CST to CST IV. enrichment. Vaginal dysbiosis was associated with excretion of in- knowledge, no other studies have investigated the vaginal microbi- flammatory cytokines and interstitial collagenase into cervicovaginal ome in pessary carriers using multiple regions of the 16S rRNA gene. fluid and premature cervical remodeling. In comparison, monofila- The vaginal microbiome, which differs by ethnicity and vaginal ment suture had minimal impact on the vaginal microbiome and its microbial load in pregnancy, has not yet been stratified by ethnic- interactions with the host.10 ity. 29 In our study, cerclage with Mersilene was associated with This study is the first study assessing the vaginal microbiome in a microbial community harboring a relatively low abundance of pregnant women undergoing pessary or cerclage intervention. One of Lactobacillus, more threatened preterm labor episodes and poorer the main limitations of our study was the absence of randomization, clinical outcomes, as also found by some authors. Due to the lim- which could lead to a bias such as selection bias, and could prevent ited cohort of our study, before extrapolating our results to other the use of probability theory. The small size of each cohort, which may populations, further assessment is required using additional robust lead to type II errors, was another limitation of our study. Indeed, al- prospective randomized studies. though the vaginal microbiome is less diverse than the microbiome of other body sites, its complexity is enhanced by the presence of CSTs. This hinders the identification of a microbiome profile associated 5 | CO N C LU S I O N with a specific intervention for preventing sPTB. Another limitation of our study was the use of a single variable region of the 16S rRNA Our findings indicate that a short cervix is associated with a lower gene to study microbiome composition. However, to the best of our microbial load and higher microbial richness as compared with a 1412 | VARGAS et al. normal cervix. In addition, microbial richness is negatively associated ORCID with gestational age at delivery. The use of a pessary for preterm Maria Goya https://orcid.org/0000-0003-4001-2017 birth prevention does not appear to have an adverse effect on the Alba Farrás https://orcid.org/0000-0003-2100-7693 Lactobacillus relative abundance or microbial diversity. By contrast, Maia Brik https://orcid.org/0000-0002-1840-5087 cerclage with Mersilene was associated with a microbial community harboring a relatively low abundance of Lactobacillus, more threat- REFERENCES ened preterm labor episodes and poorer clinical outcomes. 1. Chawanpaiboon S, Vogel JP, Moller AB. Global, regional, and na- tional estimates of levels of preterm birth in 2014: a systematic re- view and modelling analysis. Lancet Glob Health. 2019;7:e37-­e 46. AU T H O R C O N T R I B U T I O N S 2. Hack M, Taylor HG, Klein N, Eiben R, Schatschneider C, Mercuri-­ MV was the principal clinical investigator. MV contributed to the Minich N. School-­age outcomes in children with birth weights study design, follow-­up of patients and data collection, and pre- under 750 g. N Engl J Med. 1994;331:753-­759. pared the first draft of both the protocol and article. MV is the guar- 3. Hurley EG, DeFranco EA. Influence of paternal age on perinatal outcomes. Am J Obstet Gynecol. 2017;217:566.e1-­566.e6. antor for the paper. MG was the study coordinator, and contributed 4. Paules C, Pueyo V, Marti E, et al. Threatened preterm labor is a risk to the study design, patient follow-­up and data collection, and pre- factor for impaired cognitive development in early childhood. Am J pared the protocol and article along with MV. AB and AF contrib- Obstet Gynecol. 2017;216:157.e1-­157.e7. uted to the study design and data collection. FY was the principal 5. Wallace JL, Aland KL, Blatt K, Moore E, Defranco EA. Modifying the risk of recurrent preterm birth: influence of trimester-­specific changes laboratory investigator. CM was the microbiome analysis coordina- in smoking behaviors. Am J Obstet Gynecol. 2017;216:310.e1-­310.e8. tor and prepared the article along with MV and MG. FY, AE, ZX, OS, 6. Wallace ME, Mendola P, Kim SS. Racial/ethnic differences in ZS and CB were part of the microbiome analysis team led by CM. preterm perinatal outcomes. Am J Obstet Gynecol. 2017;216:306. The principal clinical investigators in the other hospitals were LV e1-­3 06.e12. (Hospital Materno-­Infantil de Canarias), AO (Hospital de Sant Pau), 7. Iams JD, Romero R, Culhane JF, Goldenberg RL. Primary, second- ary, and tertiary interventions to reduce the morbidity and mortal- BM (Hospital de Reus) and MB (Hospital de Torrejón de Ardoz). EC ity of preterm birth. Lancet. 2008;371:164-­175. reviewed both the protocol and the article. All authors had complete 8. Iams JD, Goldenberg RL, Meis PJ. The length of the cervix and the access to the data at the end of the trial and the decision to submit risk of spontaneous premature delivery. National Institute of Child the article for publication was taken at a joint meeting, where the Health and Human Development maternal fetal medicine unit net- work. N Engl J Med. 1996;334:567-­572. whole team reviewed and approved the final version. This final ver- 9. Alfirevic Z, Owen J, Carreras Moratonas E, Sharp AN, Szychowski sion was then submitted by the corresponding author. JM, Goya M. Vaginal progesterone, cerclage or cervical pessary for preventing preterm birth in symptomatic singleton pregnant AC K N OW L E D G M E N T S women with a history of preterm birth and a sonographic short cer- vix. Ultrasound Obstet Gynecol. 2013;41:146-­151. We thank Christine O'Hara and María del Mar Jiménez Quesada for 10. Kindinger LM, MacIntyre DA, Lee YS, et al. Relationship between their help with the English version of this manuscript. In addition vaginal microbial dysbiosis, inflammation, and pregnancy outcomes to the authors, the following institutions and investigators partici- in cervical cerclage. Sci Transl Med. 2016;8:350ra102. pated in the study: Azahar Romero, Begoña Vega, José Ángel García 11. Kindinger LM, Bennett PR, Lee YS, et al. The interaction between vaginal microbiota, cervical length, and vaginal progesterone treat- (Hospital Universitario Materno-­Infantil de Canarias, Las Palmas de ment for preterm birth risk. Microbiome. 2017;19(5):6. Gran Canaria, Spain); Elisa Llurba (Hospital de Sant Pau, Barcelona, 12. Witkin SS, Moron AF, Ridenhour BJ, et al. Vaginal biomarkers that Spain); C Casanovas, Belén Santacruz (Hospital Universitario de predict cervical length and dominant bacteria in the vaginal micro- Torrejón, Universidad Francisco de Vitoria, Madrid, Spain); Belén biomes of pregnant women. MBio. 2019;10:e02242-­e 02219. 13. Callahan BJ, DiGiulio DB, Goltsman DSA, et al. Replication and re- Lledo, (Instituto Bernabeu Biotech, Alicante, Spain); Francesc Rius. finement of a vaginal microbial signature of preterm birth in two (Hospital Universitario Sant Joan de Reus, Reus, Spain). racially distinct cohorts of US women. Proc Natl Acad Sci USA. 2017;114:9966-­9971. F U N D I N G I N FO R M AT I O N 14. Brown RG, Marchesi JR, Lee YS, et al. Vaginal dysbiosis increases This study was supported by the Instituto de Salud Carlos III/FEDER risk of preterm fetal membrane rupture, neonatal sepsis and is ex- acerbated by erythromycin. BMC Med. 2018;16:9. FIS PI15/02043; Beca Dexeus Mujer 2015. Francisca Yáñez was 15. Freitas AC, Bocking A, Hill JE, Money DM. VOGUE research group. supported by a grant from ANID, BECAS Chile, N° 72190278. Zixuan Increased richness and diversity of the vaginal microbiota and Xie received a fellowship from the European Union's Horizon 2020 spontaneous preterm birth. Microbiome. 2018;6(117):117. research and innovation program under the Marie Sklodowska-­ 16. Elovitz MA, Gajer P, Riis V, et al. Cervicovaginal microbiota and local immune response modulate the risk of spontaneous preterm deliv- Curie Action, Innovative Training Network: FunHoMic; grant num- ery. Nat Commun. 2019;10:1305. ber 812969. 17. Goodfellow L, Verwijs MC, Care A, et al. Vaginal bacterial load in the second trimester is associated with early preterm birth recur- C O N FL I C T O F I N T E R E S T rence: a nested case–­control study. BJOG. 2021;128:2061-­2072. 18. Wood SL, Owen J. Cerclage: Shirodkar, McDonald, and modifica- The authors have stated explicitly that there are no conflicts of inter- tions. Clin Obstet Gynecol. 2016;59:302-­310. est in connection with this article. VARGAS et al. | 1413 19. To MS, Skentou C, Liao AW, Cacho A, Nicolaides KH. Cervical 27. Dobrut A, Gosiewski T, Pabian W. The dynamics of vaginal and rec- length and funneling at 23 weeks of gestation in the prediction of tal Lactobacillus spp. flora in subsequent trimesters of pregnancy spontaneous early preterm delivery. Ultrasound Obstet Gynecol. in healthy polish women, assessed using the sanger sequencing 2001;18:200-­203. method. BMC Pregnancy Childbirth. 2018;18:350. 20. Goya M, Pratcorona L, Higueras T, Perez-­Hoyos S, Carreras E, 28. Husain SM, Wilks M, Mupita M. Diversity and stability of cultured Cabero L. Sonographic cervical length measurement in preg- vaginal lactobacilli in pregnant women from a multi-­ethnic urban nant women with a cervical pessary. Ultrasound Obstet Gynecol. UK population. J Appl Microbiol. 2014;117:258-­265. 2011;38:205-­209. 29. Serrano MG, Parikh HI, Brooks JP, et al. Racioethnic diversity in the 21. Lleal M, Sarrabayrouse G, Willamil J, Santiago A, Pozuelo M, dynamics of the vaginal microbiome during pregnancy. Nat Med. Manichanh C. A single faecal microbiota transplantation modulates 2019;25:1001-­1011. the microbiome and improves clinical manifestations in a rat model of colitis. EBioMedicine. 2019;48:630-­6 41. 22. Sarrabayrouse G, Landolfi S, Pozuelo M, et al. Mucosal microbial S U P P O R T I N G I N FO R M AT I O N load in Crohn's disease: a potential predictor of response to faecal Additional supporting information can be found online in the microbiota transplantation. EBioMedicine. 2020;51:102611. Supporting Information section at the end of this article. 23. Ravel J, Gajer P, Abdo Z, et al. Vaginal microbiome of reproductive-­ age women. Proc Natl Acad Sci U S A. 2011;108(Suppl 1):4680-­4687. 24. Rasmussen MA, Thorsen J, Dominguez-­Bello MG. Ecological suc- cession in the vaginal microbiota during pregnancy and birth. ISME How to cite this article: Vargas M, Yañez F, Elias A, et al. J. 2020;14:2325-­2335. Cervical pessary and cerclage placement for preterm birth 25. Romero R, Hassan SS, Gajer P. The composition and stability of the prevention and cervicovaginal microbiome changes. Acta vaginal microbiota of normal pregnant women is different from that Obstet Gynecol Scand. 2022;101:1403-1413. doi: 10.1111/ of non-­pregnant women. Microbiome. 2014;2:4. aogs.14460 26. O'Hanlon DE, Come RA, Moench TR. Vaginal pH measured in vivo: lactobacilli determine pH and lactic acid concentration. BMC Microbiol. 2019;19:13.