In our study of frequent sampling of the vaginal microbiota, we found that the bacterial community is dynamic and changes rapidly. We sought to study changes in the levels of bacteria in the human vagina and assess their relationship with the common condition, bacterial vaginosis. Our first goal was to determine how stable the microbial community is in healthy women. As noted in previous cultivation-based studies, healthy women tended to be colonized with several Lactobacillus
species, though G. vaginalis
was also frequently detected, including in 70% of women without BV using PCR methods 
. G. vaginalis
concentrations increased substantially with menses in 81% of monitored menstrual cycles, and levels decreased with the end of menstruation. During these surges of G. vaginalis
, L. iners
levels also trended upwards while levels of the other two lactobacilli decreased. Schwebke et al
have similarly reported increases in Gardnerella/Bacteroides
morphotypes by Gram stain and reduced quantities of lactobacilli during menses. Our study enhances the observations made by these authors by measuring levels of specific bacterial species and provides the ability to distinguish among Lactobacillus
species. In our study, all subjects (n
14) classified as negative for BV by the Amsel diagnostic criteria on Day 0 had at least a low level of G. vaginalis
on one of the days sampled. A longitudinal cultivation study reported the isolation of G. vaginalis
in at least one time point in the menstrual cycle in all subjects, though the number of subjects examined was small (n
. These findings raise the question of the role of the normal fluctuations of G. vaginalis
in the development of abnormal flora associated with BV.
The growth of G. vaginalis
may be tied to the availability of iron. Iron is an essential growth factor for most bacteria and the acquisition of iron enhances the replication of many pathogens 
. There is limited free iron in the human body as much of it is sequestered in compounds such as hemoglobin, the iron-containing metalloprotein in erythrocytes, and lactoferrin, present in mucosal tissues. One mechanism for acquiring iron is to lyse host cells such as erythrocytes with a cytolysin thereby liberating intracellular iron stores. G. vaginalis
produces a toxin, vaginolysin, a member of the cholesterol-dependent cytolysin family of toxins 
. Experiments examining the growth of G. vaginalis
have shown that this bacterium cannot grow in iron-limiting conditions, but can use iron sources such as hemoglobin for growth and can produce siderophores suggesting a well adapted ability to harvest iron from the environment 
. Our observation that surges in G. vaginalis
coincide with menses (and therefore vaginal blood) is consistent with this hypothesis. As to the role of G. vaginalis
in BV, we hypothesize that G. vaginalis
may function as a facilitator to enhance acquisition of other BV-associated bacteria that are also characteristic of this condition.
In healthy women, concentrations of L. iners, when present, tended to increase along with levels of G. vaginalis during menses ( & ). It is noteworthy that both G. vaginalis and L. iners are easily cultivated on blood agar medium. As noted with Participant G on multiple occasions, concentrations of L. iners tended to increase with antibiotic treatment for BV, suggesting that this bacterium may fill the niche vacated by the loss of BV-associated bacteria.
Our second goal was to determine if BV-associated bacteria were eradicated with antibiotic treatment in women with recurrent BV, and to assess the time to eradication. We found that quantities of BV-associated bacteria decreased rapidly with intravaginal metronidazole, evidencing multi-log declines on a daily basis, though the slope of this decline varied ( & ). This observation supports the hypothesis that recurrent BV is not a result of initial antibiotic failure, but rather is associated with the reappearance of BV-associated bacteria after completion of antibiotic therapy. We are able to detect as few as 2.5 gene copies per qPCR reaction translating to 375 16S rRNA copies per swab. Hence, if the bacteria were present below these concentrations, we could not have detected them. Understanding how women with recurrent BV reacquire the BV bacteria is critical for prevention efforts.
This study has some limitations. First, there were a small number of subjects in this study. Longitudinal studies collecting many samples require highly motivated participants. Some investigators have taken the approach of processing limited numbers of samples from many women 
. Our approach was to process large numbers of samples from a relatively small number of women in order to better explore vaginal bacterial dynamics. Second, we applied 11 bacterial assays; but this does not represent all bacterial taxa associated with the vagina. Complementary molecular technologies such as broad-range PCR, cloning and sequencing, or pyrosequencing may help overcome this limitation, though these approaches are not quantitative. Third, although daily qPCR data were obtained from the subjects, this was not correlated with daily clinical data such as exams since the swabs were mostly self-collected. Fourth, all subjects were from a single clinic and may not be broadly representative of reproductive age women with and without BV. Fifth, patterns of bacterial fluctuations were statistically examined in retrospect; hence p-values and 95% CIs arise from exploratory analyses and serve as a basis for future hypothesis-driven research.
In conclusion, the vagina is a dynamic microbial ecosystem supporting a changing and diverse bacterial population, both in healthy women and women with BV. Lactobacilli predominantly occur in healthy women, although their Lactobacillus species profiles vary. Subjects with BV have many fastidious BV-associated bacteria that respond well to the 5-day metronidazole treatment regimen, suggesting that these bacteria are susceptible to metronidazole or are dependent on other bacteria that are eradicated with antibiotic treatment. Recurrence of BV is associated with reappearance of BV-associated bacteria suggesting re-infection or resurgence from an endogenous reservoir. The rate of decrease of the BV bacteria with antibiotic treatment varies, suggesting that longer antibiotic treatments may be warranted in some women.