Unique microbial niches exist in different anatomical sites throughout the female reproductive tract with the sampling site influencing specific microbial species abundance and population representations [23]. As such, differing sampling methods—including sampling techniques and differing anatomical site selection—have demonstrated differing microbial population representation and microbial abundance results (Fig. 1) [24,25,26].

The most common sampling techniques and predominant bacteria found in each anatomical site of the female reproductive tract including the uterus, endometrium, cervix, and vagina for microbiome testing (image created in BioRender)

Previous studies investigating the vaginal microbiome used primarily vaginal swabs, with vaginal discharge and lavage samples used less commonly. Interestingly, although limited, data suggests differences in microbial composition in swabs, discharge, and lavage sampling for the vaginal microbiome [24]. In addition, comparison of swabbing types has also been compared, such as cytobrush and a swab, wherein the relative abundance of bacterial species was found to be very similar between both methods [24]. Additionally, several studies have recorded comparable bacterial abundances and diversity in self-collected versus clinically collected vaginal samples, supporting adequate sample collection without the requirement of a gynaecological examination [27, 28]. However, it is possible for microbes to be missed and variation to occur if the exact vaginal location is not disclosed following self-collection [29].

In cervical microbiome studies, swabs obtained from cervical and endocervical areas were the most represented, followed by specific devices used to obtain a transcervical sample. Interestingly, whilst there have been minimal comparisons to date examining cervical sampling techniques for microbiome composition, a cytobrush with an extended-tip spatula has been demonstrated as the best combination of collection for collecting endocervical cells for subsequent analysis [24]. This technique may also be suitable as the best practice for determining microbiome composition, but there was no mention of best practices concerning microbiome collection and minimisation of contamination from the vagina [24]. There must be significantly more studies investigating cervical microbiome collection techniques before a conclusion can be made on best practices.

Sample collections to investigate endometrial microbiome composition most commonly rely on endometrial fluid samples, followed by endometrial tissue from scratching and biopsies. Endometrial sampling poses significant challenges as contamination from the vagina and cervix is extremely difficult to avoid, with different sampling approaches returning significantly varied results regarding microbial content [24]. Endometrial biopsy samples considerably reduce the likelihood of contamination from the lower reproductive tract; however, they are highly invasive and usually obtained from patients with pre-existing conditions not representative of a healthy population [30]. Due to this, obtaining endometrial microbiome samples that lack contamination in a non-invasive way is extremely difficult. Reschini et al. suggest endometrial microbiome sampling via an embryo catheter in conjunction with rigorous aseptic methodologies to minimise contamination and invasiveness [25]. Determining any source of potential contamination is vital as this can significantly influence microbial community abundance, even for the same anatomical site.

Uterine microbiome samples collected via the distal section of a transfer catheter and the distal section of a transfer catheter are commonly used in embryo transfer or via aspiration. Until recently, the uterine microbiome was thought to be sterile, and little is still known about the microbiome of the uterus. Recent studies have identified microbes from uterine samples; however, caution is suggested for these findings with a high possibility of endometrial contamination likely influencing these results [26, 31, 32]. In patients not participating in ARTs, uterine samples are commonly obtained during hysterectomies and hysterectomy procedures [31]. However, for subsequent ART procedures, uterine microbiome samples are only obtainable through the distal section of a transfer catheter, posing significant potential contamination from vaginal, cervical, and endometrial microbes [26, 31]. Transcervical approaches are currently used, with further research and optimisation required to obtain pure uterine microbiome samples [26].

Delays in DNA extraction after sampling require samples to be stored effectively to ensure the composition of microbiota remains representative of the populations at the time of sampling. Frozen storage of samples (− 80 °C) is considered the ‘gold standard’ best practice to ensure preserved microbes and prevention of the microbiota from changing significantly [33, 34].

In terms of sample storage following collection across literature, the majority of microbiome composition samples were stored at either − 20 °C or − 80 °C. [13, 25, 35,36,37,38,39,40,41,42,43]. Whilst freezing bacterial samples has shown to alter the cellular structure of gram-positive bacteria, causing a higher ratio of Firmicutes to Bacteroidetes, the change was of no statistical significance [44]. To ensure the sample is the most representative population of the sample site, immediately freezing samples to − 80 °C is considered the ‘gold standard’ to ensure preservation of microbes and prevention of the microbiota changing significantly at room temperature (21–24 °C) [32, 45]. Prakash et al. (2020) identified that, generally, microbial samples stored at room temperature for an extended period resulted in changes in microbial communities and overall composition [46]. Choo et al. also found substantial microbe compositional changes in samples stored at room temperature when compared to samples stored at 4 °C or − 80 °C [43]. Whilst sample storage temperatures can significantly impact microbial composition, sample storage methods can also heavily influence alpha and beta diversity [47]. Alpha and beta diversity allow for the microbial differences within a sample (alpha) or across multiple samples (beta) to be quantified [48].

Additionally, some studies opt to add some form of DNA stabilising solution following sample collection prior to storage or bypassing freezing to maintain microbial composition of samples [40, 49,50,51,52,53]. The addition of DNA stabilising solutions before freezing have been found to preserve microbial DNA integrity, allowing for the transportation of microbiome samples [54]. However, different stabilising solutions can influence microbial alteration. As an example, Choo et al. found that samples stored in Tris–EDTA buffer showed the most significant bacterial change when compared to the non-commercial DNAgenotek produced OMNIgene.GUT that aims to optimise collection for nucleic acids [44, 55]. Samples stored in Tris–EDTA buffer demonstrated an increased operational taxonomic unit abundance of facultative anaerobes such as Proteobacteria and decreased abundance of Firmicutes and Actinobacteria [44]. The reduction in specifically Firmicute abundance would significantly impact female tract microbiome samples as the Lactobacillus genera results from the Firmicutes/Bacillota phylum and is not only the most abundant genera found but the biggest indicator of a healthy or dysbiotic environment [26, 56, 57]. It is also worth noting that the use of lysis buffers in sample storage has resulted in an overall reduction in DNA yield but not in DNA integrity [58]. Additionally, samples stored in an RNA preservation buffer such as RNA later have demonstrated significant changes in bacterial diversity when compared with storage alone at − 80 °C along with a reduced DNA extraction yield and purity [44].

Choosing to complete DNA extraction directly after sampling rather than DNA extraction directly following sample freezing has also shown different impacts on microbial composition. Vogtmann et al. found that compared to immediate DNA extraction, freezing microbial samples resulted in an increased abundance of the Firmicutes strain and a decreased abundance of Bacteriodetes microbes [59]. However, due to factors such as transport and time constraints, DNA extraction after sample storage is the most common method of choice [59].

Whilst sampling methods influence microbial composition, sampling methods are less influential over microbiome abundance and diversity than individual variation [60, 61]. Overall, it should be recommended that to enable the most accurate microbial representation and provide high DNA integrity and yield, it is best practise to immediately extract DNA from samples and store extracted DNA at − 80 °C. If DNA extraction is unable to be completed at that time, samples should be immediately frozen at samples at − 80 °C until DNA extraction can occur.

Just as sampling methods influence microbial composition, DNA extraction methods also impact the presence and abundance of certain microbes and vary significantly from study to study (Table 1). Choosing the correct DNA extraction method is vital as vaginal samples contain high proportions of host DNA (> 90%), which can impair species detection [62]. Specific DNA extraction techniques are believed to enhance the yield of gram-positive bacteria but may not be as effective for gram-negative bacteria, potentially misrepresenting the true population composition. Due to the increased rigidity and strength of gram-positive bacterial cell walls, more extensive lysis protocols are required to obtain their DNA. However, these rigorous methods can over-degrade gram-negative bacteria, thereby reducing their yields [63]. The inverse is also demonstrated when lysis techniques are unable to effectively lyse gram-positive bacteria, leading to decreased DNA yields and misrepresenting population diversity [63, 64]. However, efficient mycobiome lysis is dependent on repeated bead-beating and enzymatic lysis which are integral for fungal DNA extraction [65].

Interestingly, some results demonstrated the use of DNA extraction kits non-specific to microbiomes and including a pre-lysis step, on average, increased DNA yield and microbial diversity when compared to DNA extraction kits specific to microbiomes and lacking a pre-lysis step [66]. The use of optimised methodologies from previously published articles or custom protocols is the most common choice for DNA extraction [26, 35, 39, 42, 43, 71]. For studies using DNA extraction kits, the most commonly used kit used is the DNeasy PowerSoil kit, followed by other kits such as Agencourt Genfind v2 Blood & Serum isolation kit, PureLink Microbiota DNA extraction kit, and the QIAamp DNA Microbiome kit [26, 31, 35, 39, 42, 43, 71]. A comparative study by Mattei et al. recorded that the QIAGEN DNeasy method achieved a higher yield and quality than a modified MoBio PowerSoil kit. However, the MoBio PowerSoil kit demonstrated higher microbial diversity when compared to the DNeasy method and higher diversity than the standard protocol [72]. A summary of these comparisons is provided in Table 3 including the extraction technique, supplier, key steps used in template isolation, and the perceived advantages and limitations of the kits commonly used.

Additionally, some studies choose to complete a pre-digestion step before DNA extraction consisting of enzymatic and/or mechanical cell lysis to degrade bacterial cell walls for difficult-to-lyse bacteria in an effort to increase the overall yield of isolated DNA [51, 67,