Organoids can be reliably established from cryopreserved endometrial tissue of infertile women and are similar to those derived from freshly biopsied tissue of the same patient. With the use of biobanked endometrial tissue, future organoid-based fertility research will no longer be limited to clinics situated near organoid laboratories.

The endometrial microbiota composition may be associated with implantation success. However, a ‘core’ composition has not yet been defined. This exploratory study analysed the endometrial microbiota by 16S rRNA sequencing (V1–V2 region) of 141 infertile women whose first IVF/ICSI cycle failed and compared the microbiota profiles of women with and without a live birth within 12 months of follow-up, and by infertility cause and type. Lactobacillus was the most abundant genus in the majority of samples. Women with a live birth compared to those without had significantly higher Lactobacillus crispatus relative abundance (RA) (p = 0.029), and a smaller proportion of them had ≤ 10% L. crispatus RA (42.1% and 70.4%, respectively; p = 0.015). A smaller proportion of women in the male factor infertility group had ≤ 10% L. crispatus RA compared to women in the unexplained and other infertility causes groups combined (p = 0.030). Women with primary infertility compared to secondary infertility had significantly higher L. crispatus RA (p = 0.004); lower proportions of them had ≤ 10% L. crispatus RA (p = 0.009) and > 10% Gardnerella vaginalis RA (p = 0.019). In conclusion, IVF/ICSI success may be associated with L. crispatus RA and secondary infertility with endometrial dysbiosis, more often than primary infertility. These hypotheses should be tested in rigorous well-powered longitudinal studies.

Study question After one failed IVF/ICSI cycle, does the endometrial microbiota composition differ between women who will or will not reach a live birth within 12 months? Summary answer The endometrial microbiota composition did not significantly differ in women with one failed IVF/ICSI cycle with or without live birth, but statistical power was low. What is known already Evidence for the presence of an indigenous endometrial microbiome is mounting, and its composition may be associated with implantation success. However, a ‘core’ endometrial microbiome has not yet been defined, and its role in embryo implantation is still poorly understood. Further investigation of this topic may allow improvement and personalisation of clinical care for infertile couples. Endometrial microbiome analysis in infertile women has not yet been performed using transcervically obtained endometrial tissue. Using endometrial tissue instead of swabs or fluid may increase the bacterial DNA yield and therefore the precision of microbiome analyses. Study design, size, duration Endometrial tissue was obtained from a cohort of 141 infertile women undergoing endometrial scratching within a randomised controlled trial (RCT) (SCRaTCH trial, NL5193/NTR5342). Briefly, women aged 18-44 years with failed implantation after one full IVF/ICSI cycle and planning a subsequent IVF/ICSI cycle, were eligible. Participants were followed-up until 12 months after randomisation, with the primary outcome being live birth, defined as the delivery of at least one live foetus after 24 weeks of gestation. Participants/materials, setting, methods Endometrial tissue was obtained with an endometrial biopsy catheter in the midluteal phase of a natural cycle preceding subsequent IVF/ICSI, snap-frozen and stored at -80 °C until use. Total DNA was isolated from these biopsies, followed by 16S rRNA sequencing (V3-V4 region) to determine the endometrial microbiota composition. Positive (mock communities) and negative controls (DNA extraction and PCRs) were included. QIIME2 and DADA2 were used for the data analysis, followed by statistical analysis in R studio. Main results and the role of chance During the 12-month follow-up, 61/141 women (43.3%) reached a live birth. While endometrial microbiota profiles of all 141 women were analysed, only samples with ≥100 reads were included in the analysis, resulting in a total of 46 samples (32.6%) that were included in the analysis, which consisted of samples from 25 women who did not have and 21 women who did have a live birth within 12 months. The median number of reads per sample was not significantly different between the two groups (respectively 2,317 (IQR 651-19,031) and 1,335 (IQR 296-3,180), p = 0.29 by Mann-Whitney test). The endometrial microbiota detected, were bacterial genera frequently reported within the vaginal microbiota (e.g. Lactobacillus, Atopobium and Gardnerella). A clear dominance of Lactobacillus (relative abundance 55-100%, n = 22) or an unclassified bacterium genus (relative abundance 52-76%, n = 18) was observed in the majority of the samples; however, this dominance was not associated with the outcome of live birth. In addition, the samples dominated by Lactobacillus genera were mostly dominated by one species of Lactobacillus each (L. crispatus, L. iners, L. gasseri or L. jensenii). Limitations, reasons for caution The low biomass and the low ratio of bacterial versus human DNA in endometrial tissue were limiting factors in endometrial microbiota analysis. Furthermore, tissue was obtained transcervically, and contamination with vaginal/cervical microbiota could therefore have occurred. In the SCRaTCH trial no vaginal swabs were taken to serve as internal controls. Wider implications of the findings Future endometrial microbiota studies should consider the use of samples with a lower proportion of human DNA to maximize bacterial DNA yield. Furthermore, for endometrial microbiota research, sampling devices avoiding cervicovaginal contamination are desirable and may be developed in the future. Trial registration number SCRaTCH trial, NL5193/NTR5342

Study question Does the endometrial transcriptome profile differ between infertile women with or without a recurrent implantation failure (RIF)? Summary answer Although two different clusters emerged from the endometrial transcriptome data, these were not associated with clinical phenotype (RIF vs non-RIF). What is known already Despite the transfer of morphologically ‘good-quality’ embryos in IVF/ICSI, implantation failure often occurs, which may be explained by impaired endometrial receptivity. In order to guide prognosis and use effective therapeutic interventions, identifying a gene expression profile predictive of endometrial receptivity as well as implantation failure, would be of great value. Additionally, transcriptome analysis may also shed light on alterations in biological processes responsible for the implantation failure. Thousands of potential biomarkers for endometrial receptivity have already been identified by transcriptomic approach, however due to differences in study methodology, there is little overlap of markers between studies. Study design, size, duration Endometrial tissue was obtained from a cohort of 141 infertile women undergoing endometrial scratching within a randomised controlled trial (RCT) (SCRaTCH trial, NL5193/NTR5342). Briefly, women aged 18-44 years with failed implantation after one full IVF/ICSI cycle and planning a subsequent IVF/ICSI cycle, were eligible. Participants were followed-up until 12 months after randomisation, with the primary outcome being live birth, defined as the delivery of at least one live foetus after 24 weeks of gestation. Participants/materials, setting, methods Endometrial tissue was obtained with an endometrial biopsy catheter in the midluteal phase of a natural cycle preceding subsequent IVF/ICSI. Biopsies were snap-frozen and stored at -80 °C until use. After thawing, total RNA isolation, library preparation and paired-end RNA-sequencing were performed. Raw data was preprocessed and mapped to GRCh38. Reads (counts per million) were normalised using library size. Differential gene expression (DGE) analysis was conducted using the EdgeR package with significance threshold FDR <0.05. Main results and the role of chance Out of 141 endometrium samples, 107 were included in the RNA-sequencing based on RNA quality. For DGE analysis, data of two groups were compared: the ‘fertile’ group, women with a live birth after ≤3 good quality embryo(s) transfers (n = 23), and the RIF group, women with no live birth after ≥3 good quality embryo(s) transfers (n = 23). Two clusters were visible in the principle component analysis (PCA) plot showing transcriptome data of the fertile and RIF samples (cluster 1, n = 29; cluster 2, n = 10), which was not explained by clinical phenotype, as both clusters contained samples of both the fertile and RIF group. DGE analysis between the fertile and RIF group resulted in respectively 3 significantly upregulated and 0 significantly downregulated genes, whereas DGE analysis between the two clusters resulted in 2,235 significantly upregulated and 2,162 significantly downregulated genes. Enrichment analysis of differentially expressed genes between both clusters demonstrated upregulation of enriched terms mainly annotated to cell migration and downregulation of enriched terms mainly annotated to lipid and mitochondrial metabolism. Limitations, reasons for caution A strength of the study is the large number of samples included. Bulk RNA-sequencing was conducted and there was a variation in LH-based timing of the biopsies (5-8 days after LH surge) for which adjustments of the transcriptome data for tissue cellular composition and menstrual cycle were performed. Wider implications of the findings Future studies investigating underlying biological mechanisms in the endometrium in (in)fertility by a (multi-)omics analysis approach with standardised methodology are required to obtain consistencies in relevant biomarkers/pathways, and in due course create possibilities to improve and personalise care for infertile couples. Trial registration number NL5193/NTR5342