In this study, we sought to determine the presence of a genuine microbial fingerprint in the upper and lower female genital tract of benign and endometrial cancer patients highlighting potential differences and trying to elucidate the putative role of the microbiota in endometrial cancer generation or suppression.
Contamination from patient and non-patient sources may obscure actual microbial signatures from low bacterial biomass sites. Our study integrated four different sets of technical controls for air and reagent/equipment contaminants during sample collection in the pathology laboratory, DNA extraction and 16S rRNA gene sequencing and their sequence reads (15% of total sequencing data) were removed to deduct background DNA contamination from further data analysis. Furthermore, we demonstrated potential intra-patient contamination in one quarter to one third of samples in laparoscopic procedures. A previous report described high bacterial similarity of endometrial samples between transcervical and transuterine collection methods, indicating low contamination28.
We calculated actual microbial abundances using quantitative real time PCR and showed that endometrial, fallopian tube and ovarian microbiota are encountered in much lower quantities (2–4 log10) than vaginal and cervical microbiota and are similar to bacterial abundances of controls. Our findings are consistent with those of Mitchell and co-workers18, who reported that mean bacterial quantities in the endometrium and upper endocervix are lower than vaginal levels by 2–4 log10 rRNA gene copies, which was further corroborated by the Chen study28 showing that vaginal sites contained about four orders of magnitude more bacteria (1010-1011) than the endometrium and peritoneal fluid, still exhibiting much lower cycle threshold (Ct values) than negative controls. Another study confirmed high bacterial signals in vagina and comparable bacterial load in cervix and endometrium with most cervical and endometrial samples (60–72%) exceeding the bacterial copies of blank controls17.
Overall, we observed a microbiota signature above background contamination in 62% of benign endometrium, 50% of malignant endometrium, 85% of benign fallopian tube and 95% of benign ovary based on sequencing reads. Winters and colleagues also reported a 60% (15/25) bacterial recovery above background DNA controls from the endometrium of women with benign pathology, mainly fibroids, undergoing transabdominal hysterectomy17. Mitchell and co-workers concluded that 95% (55/58) of non-cancer patients having total laparoscopic or laparoscopically-assisted vaginal hysterectomy without an intracervical manipulator had endometrial colonisation with at least one species18, while Chen and colleagues enrolled 95 reproductive age women and isolated bacteria from the endometrium, fallopian tubes and peritoneal fluid during either laparotomy or laparoscopy28. With regards to bacterial composition, we identified a prominent microbial signature in benign endometrium, fallopian tube and ovary, while the microbiome of malignant endometrial samples and technical controls was largely shared. In the Mitchell and co-workers study, signature endometrial species were L. iners (45%), Prevotella spp. (33%) and L. crispatus (33%)18, while in the Chen study Lactobacillus, Vagococcus, Acinetobacter and Pseudomonas, which were also identified in the fallopian tube28. Notably, our results are predominantly deduced from post-menopausal women > 50 years of age, while the aforementioned reports involved mainly premenopausal women < 50 years of age. The data suggests that a microbiota signature is detectable in the upper reproductive tract of a proportion of women regardless of age and menopausal status.
Investigation of a microbial continuum within the female genital tract revealed that in 75% of benign patients the most abundant species of the lower genital tract were also recoverable from all sites of the upper genital tract, lending credence to the theory of bacterial colonisation through ascension. Conversely, bacterial continuum was less cohesive in the endometrial cancer cohort. Similarly, intra-individual microbial correlations between the lower and upper genital tract consistent with anatomical contingency have been reported by Chen and colleagues28, whereas Walsh and co-workers reported that the microbiome of lower genital tract is correlated with uterine microbiome in patients with endometrial cancer10. In addition to this, associations of vaginal with rectal microbiota have been suggested based on reports of bacterial species overlap between the two sites and the presence of enteric oestrobolome, where the gut microbiome metabolises oestrogens and indirectly facilitates vaginal Lactobacillus growth41–47. We found that vaginal-rectal correlations were stronger in endometrial cancer patients compared to benign, still accounting for only 15% sharedness of microbiota.
We subsequently took cross-sectional snapshots of the FGT microbial structure in thirty-seven women with and twenty-four women without endometrial cancer. Our findings suggest that there is an unequivocal decrease of Lactobacillus abundance in the vagina, cervix and endometrium of endometrial cancer patients and an enrichment of Anaerococcus, Porphyromonas, Prevotella, Fusobacterium, Bacteroides and Peptoniphilus, which is in keeping with previous studies10,11. Lactobacillus reduction lends support to the more broadly described association between Lactobacillus depletion and adverse gynaecological and obstetric outcomes, including cervical cancer and precancerous lesions13 as well as miscarriages48, preterm labour49 and in vitro fertilisation failure24. The biological significance of Lactobacillus communities includes pathogen antagonism by lactic acid, hydrogen peroxide and bacteriocin production but also immune and metabolic pathway as well as epigenetic regulation50–59. Anti-cancer effects of Lactobacillus species have been reported for L. casei, L. plantarum, L. rhamnosus and L. acidophilus through natural killer cell activation, dendritic cell maturation or probiotic-derived ferrichrome release60–68 inducing cancer cell apoptosis in vitro61,67,69 and increasing anticancer drug efficacy70–71,82. Continuous administration of L. reuteri strain to tumour-prone mice for several months reduces the frequency of intestinal pre-cancerous lesions68. Caution should be exercised, however, when analysing metataxonomic findings because routinely relative abundances are used, which conceal actual abundances. Therefore, the depletion in the relative abundance of a prevalent species will lead to an increase of the relative abundance of the least prevalent species. This could be misinterpreted as an enrichment of those taxa, even though their actual abundance has remained the same making the use of quantitative PCR imperative to confirm the enrichment of the taxa in question.
We sought to determine whether the disparities observed between benign and endometrial cancer patient microbiota can be attributed to malignancy, suggesting a cause-effect relationship or other confounders impacted on our findings. Ethnicity70, menopausal status10,71, obesity72, use of sex hormones71, smoking73, sexual behaviour74 and vaginal pH11 have all been recognised as modulators of microbial composition in the female genital tract. Our group has previously demonstrated that obesity drives the vaginal microbiome towards a highly diverse, Lactobacillus-depleted eco-structure characterised by higher levels of Dialister, Anaerococcus vaginalis and Prevotella timonensis, and lower levels of Lactobacillus crispatus compared to non-obese women72. Another study highlighted the importance of menopausal status in FGT composition showing that postmenopausal women exhibit increased diversity and enrichment of Anaerococcus, Peptoniphilus and Porphyromonas species10. Interestingly, almost half (8/17) of the microbes enriched in postmenopausal women were also associated with endometrial cancer10. From a mechanistic standpoint, the interplay between estrogens and Lactobacilli is well-documented in the vaginal mucosa with oestrogens stimulating glycogen-dependent metabolism that increases colonisation with lactobacilli, which in turn metabolise glycogen to lactic acid maintaining an acidic vaginal environment75. In our study, no variable showed systematic bias with respect to the two comparison groups (benign vs endometrial cancer) apart from age; endometrial cancer patients were predominantly ≥ 65 years, while benign patients between 50–64 years of age. Importantly, postmenopausal women were equally distributed between the two cohorts.
To uncover a potential role of altered microbiota in endometrial carcinogenesis, we selected L. crispatus as a candidate health-promoting microbe, given that it is significantly depleted in endometrial cancer and substantial evidence suggests that commensal Lactobacillus species have anti-inflammatory and anti-cancer properties54–69. Endometrial cancer is thought to arise in an environment of chronic, low-grade inflammation characterised by an increase of pre-diagnostic circulating levels of CRP, IL-6, TNFα, sTNFR1 and sTNFR2 and IL-1Ra (EPIC cohort)76,77, which have also been linked to high BMI78,79, highlighting the interrelationship of the obesity- inflammation- endometrial carcinogenesis axis. Following establishment of endometrial cancer, cytokines are secreted by cancer cells, the tumour microenvironment and infiltrating immune cells but their role in oncogenic processes remains ambiguous. We interrogated the effect of L. crispatus on endometrial cell proliferation and its intersection with inflammatory pathways and found that L. crispatus-conditioned medium reduced viability of endometrial cancer organoids, whilst had minimal effect on studied cytokines/chemokines. It is not clear whether LCC medium is actually cytotoxic or restricts proliferation and/or induces lethality of endometrial organoids due to dilutional effects.
A limitation of this study involves the use of women with benign pathology as controls instead of healthy individuals that may display differential microbial structure. Women undergoing risk-reducing surgery can provide an ethically acceptable source of healthy surgical specimens in future studies. Longitudinal studies monitoring the bacterial composition intra-individually over time to capture transition from a healthy state to precancerous conditions (complex atypical hyperplasia) and invasive endometrial carcinoma are required to provide useful information on microbiota dynamics and temporal shifts within each patient. Finally, viral and fungal metagenomic studies could be useful to complement and expand our knowledge of the human-microbial and microbial-microbial cell interactions in health and disease.
In conclusion, only a subset of women with and without endometrial cancer harbours a microbiome in the upper GT that is quantitatively and qualitatively distinct from background contamination. Lactobacillus depletion in the lower and upper GT is characteristic in endometrial malignancy but whether depletion precedes or follows the development of cancer is still unclear. L. crispatus does not significantly affect cytokine production in benign and malignant endometrial organoids but this observation does not preclude that L. crispatus or other bacterial species could influence metabolic or oestrogen signalling pathways locally or could have an impact on infiltrating immune cell cytokine secretion, which could subsequently exert an action on endometrium. Transitioning from reductionist cell models to more comprehensive, physiologically relevant cell systems that capture the cellular diversity of organs, including stromal, mesenchymal and immune cells, is mandatory to reliably assess the functional role of resident microbiota and devise ways of manipulation that could restore eubiosis.