Interpretation
Several studies have shown the presence of bacterial DNA within the
upper female reproductive tract (6–9), however a debate still exists
regarding the presence of bacteria in sites considered normally sterile
such as the human placenta (30,31). A previous study comparing vaginal
samples to FT samples found that the vagina contained four orders of
magnitude more bacteria than the FT (7). Another study found 20% of FT
samples harbored no bacterial signal after PCR amplification, and other
samples contained varying and sometimes very low amounts of bacterial
reads (8).
In our study, we found lower rates of positive FT samples in patients
with hydrosalpinx. As this condition is sometimes associated with an
infectious etiology, administration of extended antibiotic regimens
instead of surgical resection has been attempted. Pathogenic bacterial
presence has been hypothesized to be the cause of the deleterious effect
of the stagnant fluid on fertility (33). Our study results do not
support this hypothesis, since FT fluid of case-patients didn’t contain
bacteria, as opposed to controls, even in case-patients who underwent
surgery due to prior infection. One explanation for this phenomenon may
be the fact that case-patients were more exposed to antibiotic
treatments than controls, which altered the amount and composition of
bacterial communities, especially that of the vaginal ”reservoir”. When
we compared vaginal samples of patients with and without prior
antibiotic treatment, we did find a non-significant difference in read
counts. Nevertheless, when we compared vaginal samples of patients with
positive and negative FT samples, we did not find any differences in
average read counts, undermining the “quantitative” argument for this
difference in bacterial presence. We believe that the vagina is the most
important, if not the sole source of bacterial colonization of the FT.
Hence, we hypothesize that tubal occlusion resulting the hydrosalpinx,
is the cause of the paucity of bacterial presence within the fallopian
tubes, by blocking the anatomical pathway of bacterial migration. As
this study included only nine patients with hydrosalpinx, it was
underpowered to compare the results of women with or without prior
antibiotic exposure. Such a comparison would have provided important
data regarding the contribution of antibiotic treatment to FT bacterial
colonization in the setting of tubal occlusion and should be attempted
in future studies.
It is well accepted that different microbial profiles of the vagina are
associated with disease states (4). Looking into the differential
abundance of the vaginal microbiota, we found marked differences in the
bacterial composition of cases and controls. Figure 3 depicts the
differences in the presence of certain bacteria associated with a
dysbiotic vaginal environment between the two groups. Of these,Gardnerella, Atopobium, Prevotella, and Ureaplasma ,
bacteria known to be dominant in bacterial vaginosis (34), and therefore
infertility, and adverse pregnancy outcomes, were more prevalent in
samples from case-patients. Moreover, samples from case-patients lacked
the presence of lactobacilli, the most important genus associated with
the normal vaginal microbiota, compared with controls. As stated above,
case-patients had higher rates of antibiotic exposure, which cannot be
excluded as a contributing factor to the results. These findings, while
not providing an answer to whether they are the cause or effect of the
tubal occlusion, attest to the fact that women with hydrosalpinx are
more likely to have a dysbiotic vaginal environment.
As samples that were taken from occluded FT in case patients did not
contain enough bacterial DNA after the decontamination process, we
compared the microbiome of the vagina and fallopian tubes only of
matched control patients. As described in previously cited studies,
these two environments were different in both alpha and beta diversity.
The FT microbiome was found to be more diverse, and while Firmicutes was
the predominant phylum in both groups, FT samples were more abundant in
Proteobacteria, similar to findings in previous studies. This highlights
the fact that although bacteria in the FT may largely originate from the
vagina, both sites may still represent different microbial niches.
In our study, we used stringent criteria to filter out possible
contamination which led to the exclusion of 70% of FT samples in the
control and 90% of FT samples in the case groups. These figures are in
line with rates of culture-positive samples in other studies (7).
Exploring low-biomass samples is always subject to various contamination
mechanisms (32), and the lower the amount of bacterial DNA present, the
higher the relative weight any contamination bears on sample analysis.
This was previously described by Miles et al. (6) who compared vaginal
to FT samples, before and after exclusion of samples with containing low
numbers of sequence reads. The exclusion of these samples enabled
clustering of upper and lower genital tract samples more closely
together. This may imply that some of the less abundant samples did not
necessarily belong to actual anatomical microbiota. Furthermore, the
genera described as more abundant in FT than in vaginal samples in other
studies showed a clear trend towards environmental bacteria such asAcinetobacter, Pseudomonas, and Methylobacterium . That is
in contrast to culture-positive samples, which mostly contained bacteria
which are typical residents of the vaginal microbiota. We believe that
in the normal female reproductive tract, the major biological pathway of
bacterial presence in the FT is the translocation of the vaginal
microbiota through the upper genital pathway. This may account for the
very low amounts of bacteria that were described. Thus, decontamination
should be strongly considered when non-vaginal bacteria are found in FT
samples.