Vaginal microbial community state types fail to predict IVF outcomes, whereas Ureaplasma parvum and Lactobacillus iners are negative predictors of implantation, clinical pregnancy, and live birth

STUDY QUESTION: Are previously proposed vaginal microbial community state types (CSTs) valid predictors of IVF success, or do alternative microbial signatures provide stronger associations?. SUMMARY ANSWER: Previously proposed CSTs as predictors of implantation, clinical pregnancy and live birth were not confirmed, while an interaction between Ureaplasma parvum and Lactobacillus iners emerged as a strong negative predictor. WHAT IS KNOWN ALREADY: Infertility affects 17% of the global population. Only one-third of treatment cycles of assisted reproductive technologies result in embryo implantation, even fewer lead to clinical pregnancy or live births. While early findings have spurred the development of microbiome-based tests for success prediction, evidence on supporting their reliability remains inconclusive. STUDY DESIGN, SIZE, DURATION: This is a prospective, single-centre study to validate existing and to identify better microbial predictors of infertility treatment outcomes. 266 infertile female patients (age 18-45 years) undergoing a frozen-thawed embryo transfer cycle in an anovulatory regimen (i.e., a cycle with transfer of an embryo following a previous oocyte pick-up, fertilisation and freezing of zygotes or embryos) were recruited for the study in a timeframe from 05/2017 to 03/2019. PARTICIPANTS/MATERIALS, SETTING, METHODS: Female, infertile patients, aged 18 to 45 years from routine care. Vaginal swabs were taken prior to embryo transplantation and subjected to DNA isolation for 16S-based microbiota analysis. Extended demographic and treatment data were recorded. Clinical outcomes were defined as: (i) implantation, confirmed by a positive hCG test, (ii) clincial pregnancy, and (iii) live birth, the last defined as the birth of a viable infant. Sequencing data were processed in mothur following established pipelines, and microbial composition (taxonomy) as well as microbial diversity (dissimilarity analyses) were determined using the open-source software R. A prediction model for implantation success was built using binary logistic regression based on abundance of putativelye predictive microbial taxa. MAIN RESULTS AND THE ROLE OF CHANCE: Previous studies haveThis study suggestsed that vaginal microbial community state types, alpha-diversity, and the ratio of dominant Lactobacillus species do not correlate in statistical terms or in a clinically meaningful effect manner with implantation and clinical pregnancy,( as a surrogate for endometrial receptivity), or live birth,( as a surrogate for ongoing pregnancy viability). However, Ureaplasma parvum and Lactobacillus iners abundances were identified as negative predictors of embryo implantation, clinical pregnancy and live birth. A subset of women colonized by these taxa experienced drastically reduced embryo implantation and completely failed to achieve clinicial pregnancy or give birth to live offspring, suggesting a potential role of these organisms in implantation failure and reproductive outcome independent of other influencing factors such as age, oestradiol levels, endometrial thickness etc. LARGE SCALE DATA: The raw sequencing data used for this manuscript are publicly available at the European Nucleotide Archive under accession number PRJEB97062. LIMITATIONS, REASONS FOR CAUTION: This study is a single centre study warranting further validation cohorts. Given the variable nature of the vaginal microbiota, sample sizes need to enlarge for better refinement of the analyses. Further, underlying mechanistical basis of our findings is yet elusive and clinical translation has yet to be established. WIDER IMPLICATIONS OF THE FINDINGS: While this novel association warrants confirmation, the results caution against reliance on previously suggested CSTs as predictors and highlight the need for refined, reproducible microbiome-based diagnostics in reproductive medicine.