Site-specific prevalence and cell densities of selected microbes in the lower reproductive tract of menstruating tampon users.

OBJECTIVE
To assess differences in prevalence and cell densities of enterococci, Gram negative enterics (GNEs), yeast and Staphylococcus aureus among four genital sites and to examine whether the presence of organisms at one site affected the presence of organisms at other sites.


METHODS
Swab samples from the perineum, below and above the hymen, and the posterior fornix obtained from 52 tampon users on menstrual cycle day 3 were analyzed for site-specific prevalence and cell densities of microorganisms.


RESULTS
Enterococci and GNEs were the most prevalent study organisms at all sites and decreased in prevalence from the perineum to the posterior fornix. Cell densities similarly decreased from below the hymen to the posterior fornix. Yeast were detected at the hymen only; S. aureus frequency was similarly low at all sites. Yeast and S. aureus site-specific cell densities were similar. The above- and below-hymen sites were similar in prevalence and cell density of organisms. An above-chance association existed between the presence of any study organism below the hymen and above the hymen and was strongest for GNEs.


CONCLUSIONS
The pattern of genital colonization with enterococci and GNEs reflects their likely gastrointestinal source. The absence of significant differences in the prevalence and cell densities of study microflora above and below the hymen combined with an above-chance association of the presence of microorganisms at these regions suggests that the regions above and below the hymen are not different with respect to the presence of the organisms evaluated in this study.

The female genital microflora represents a dynamic ecosystem that fluctuates in response to a variety of influences [1][2][3][4][5][6][7][8][9][10][11] . Host factors, such as age, hormonal status, sexual behavior and parity all play a role. Bacterial vaginosis (BV) is associated with pronounced shifts in the endogenous flora, and agents such as systemic antibiotics, topical antifungal preparations, spermicides and douches may profoundly or selectively alter the balance of organisms present. Menstrual products do not cause broadscale qualitative or quantitative changes in the composition of vaginal flora [11][12][13][14] but effects on selected organisms have been observed 15,16 .
Studies that examined the prevalence and relative proportions of microorganisms in the lower genital tract of women have produced disparate results 4,15,17,18 . Variability in the methods of specimen collection and culture techniques, changes in vaginal microbiology over the course of the menstrual cycle, the absence of controls for conditions that alter the endogenous flora and non-uniformity in site of sampling are likely contributing factors.
Niche specialization may exist along the lower genital tract 5,19 . Colonization may be influenced by anatomical and epithelial-receptor differences, redox potential, pH, and proximity to the external genitalia and perineum. If so, specimens collected at a single genital site may not be representative of the entire microbial community.
Few studies have examined more than one site in the same individual and of these, several focused on a single organism [17][18][19][20][21][22][23] . Studies that have delineated microbial changes over the menstrual cycle 1,7,17,18,[24][25][26] or the influence of menstrual products on genital microflora 27 have typically focused on a single vaginal site, self-obtained swabs, or vaginal washes.
This study was an investigation of the prevalence and cell densities of selected microorganisms at multiple sites in the lower genital tract. We examined a single time point during menstruation in tampon users while excluding potential confounding factors such as BV, medications, and use of topical or personal cleansing products that may affect the genital microflora. The objective was to assess differences in isolation frequency and density of selected microbes between different genital sites and to examine the possibility that colonization with any particular organism at an individual site may have affected the presence of microorganisms at the other sites examined.

SUBJECTS AND METHODS
This was a single-center study to characterize selected microbes of the lower genital tract during menses in tampon users aged 18-45 from the population of Dallas, Texas. An Institutional Review Board reviewed the protocol and all subjects signed an informed consent statement. Seventy-nine subjects were screened for eligibility, 55 enrolled and completed the study, and 52 subjects were included in the dataset.
Ten days prior to the anticipated start of the menstrual period, subjects were screened for eligibility by means of medical history and habits and practices questionnaires. They were instructed to use their own tampons exclusively for menstrual protection and to wear a tampon on day 3 of the cycle for at least 2 hours but not more than 8 hours prior to their scheduled examination. On the day of their study examination (day 3 of the menstrual cycle), additional screening for BV and chlamydia, gonorrhea, and trichomoniasis was performed.
Eligible subjects were in good health (as evidenced by their medical and gynecologic history), had a gynecological exam with negative pap smear within the past 2 years, had regular (± 3 days) menstrual cycles of 21-35 days in length, typically used ten or more tampons per menstrual period, and had menstrual flow on day 3 of the cycle that allowed for the use of a tampon for at least 2 and not more than 8 hours. Women were excluded if they were pregnant or trying to conceive; had an active medical condition such as diabetes, hepatitis, AIDS or HIV positive status; were currently using, intended to use or had used immunosuppressive drugs, chemotherapeutic agents or antibiotics within the past 6 weeks; had a history of endometrial disease, fibroids, genital herpes or toxic shock syndrome (TSS); had abnormalities of the vulva or had genital warts or lesions; had used antifungal suppositories within the prior 6 weeks or vaginal spermicides within a week of study commencement; were unwilling to refrain from douching or the use of topical products (powders, perfumes, deodorants) in the genital area from time of enrollment through to the end of the study; were unwilling to refrain from sexual intercourse for 24 hours prior to their study examination; or were found to have BV, presence of Chlamydia trachomatis, Neisseria gonorrhoeae, Trichomonas vaginalis or a clinically-diagnosed active yeast infection at study commencement.
Microbiological sampling for identification of Staphylococcus aureus, enterococci, Gram negative enterics (GNEs) and yeast (speciated to Candida albicans, C. glabrata, C. tropicalis or other, if present) was performed on day 3 of a single menstrual cycle. After removal of the tampon, subjects were placed in a lithotomy position. Sterile, polyester applicator swabs (FALCON #220690 Becton Dickinson, Franklin Lakes, NJ) were used to obtain samples in sequence from four sites: a 2 cm 2 area of the perineum close to the anal area (not including the anus); the right side of the introitus (just below the hymen where the labia minora meet the vaginal opening); just above the hymeneal ring (approximately 2-3 cm past the introitus); and at the posterior fornix. Dry swabs were used to obtain the hymen and fornix samples; moistened swabs (sterile phosphate buffered saline, PBS) were used to obtain the perineum sample. Hymen and fornix swabs were preweighed (± 0.001 g) before samples were obtained. Each site was swabbed for approximately 10 seconds by rotating the swab so that the entire swab surface made contact with the site of sampling. Care was taken to avoid contamination from the external genitalia and crosscontamination from other sampling sites. The vaginal surface above the hymen was accessed by gentle, manual manipulation of the vaginal opening, avoiding contact of the sampling site by the fingers. A speculum was then inserted midway into the vagina for visualization and sampling of the posterior fornix, beyond the top of the speculum.
After sampling the four test sites, the pH was measured (Baxter pHIX, range 3.6-6.4, JT Baker Company, Phillipsburg, NJ) at the mid-right lateral wall. Individual swab samples from the right wall of the vagina were then obtained for assessment of BV, T. vaginalis, N. gonorrhoeae and C. trachomatis, respectively.
All samples were processed within 30 minutes of collection. Swabs from below the hymen, above the hymen and posterior fornix were re-weighed before processing. All swabs were diluted in 5 ml of PBS, vortexed for 10 seconds and plated on selective agar. Colonies growing on Mannitol Salt agar (BBL 4321173) were presumptively identified by typical colonial morphology and Gram stain as staphylococci; subsequently S. aureus was identified using the Remel BactiStaph System (Remel, Lenexa, KS). Colonies growing on Mycosel Agar (BBL 4321847) were identified as C. albicans if germ-tube positive. Germ-tube negative organisms growing on Mycosel agar were identified as yeast, not C. albicans. Non-albicans yeast, when present, were identified as C. glabrata, C. tropicalis or 'other', using API20C (bioMerieux Vitek, Hazlewood, MO). Colonies isolated on Enterococcosel agar (BBL 212205) were presumptively identified as enterococci by typical colony morphology and Gram stain, and biochemically confirmed with the Remel Bacticard Strep system (Remel, Lenexa, KS). GNEs were isolated on MacConkey II (BBL 212306) agar and identified by typical colony morphology, lactose fermentation and Gram stain. Microbial cell densities were determined using an automated dilution/spiral plating methodology (Autoplater 4000, Spiral Biotech, Norwood, MA), and based on the number of colonies on each plate whose identity was confirmed and reported as colony forming units (CFU) per cm 2 (perineum) or gram (all other sites).
Because the presence of menses precluded the use of pH and discharge as diagnostic criteria, the diagnosis of BV was made based on the presence of fishy amine odor upon addition of potassium hydroxide to the vaginal sample and the presence of clue cells. Trichomoniasis was diagnosed using the wet-mount diagnostic method; gonorrhea and chlamydia were diagnosed using the GEN-PROBE ® PACE ® 2 system (Gen-Probe Incorporated, San Diego, Ca.).
The Wilcoxon signed-rank test and the stratified pairwise Cochran-Mantel-Haenszel (CMH) test were used for comparative analysis of the data obtained. Significance was assigned at the two-sided p £ 0.05 level. The potential relationship between the presence of an organism at one site and its presence at another site was assessed using the kappa statistic. A kappa value of 0.4-0.75 indicated moderate agreement and a kappa value > 0.75 indicated strong agreement.

Population demographics
Of the 79 subjects screened, none had BV, trichomoniasis, chlamydia, or gonorrhea. Fifty-five met eligibility criteria and enrolled in and completed the study. Two were removed from the set of evaluable subjects because of unexplained discrepancies in the weight of samples for microbial analysis. A third subject was dropped for failing to comply with the protocol (wore a tampon for less than 2 hours prior to examination).
Demographic data are summarized in Table 1. Seventy-one percent of the subjects (n = 37) were Caucasian, 25% (n = 13) were African-American and 4% (n = 2) were Hispanic. Subjects ranged in age from 19-44 years with a mean of 33.2 ± 7.3 years. Mean weight was 75.15 ± 17.7 kg with a weight range of 49.5-113.6 kg. Mean height was 1.65 ± 0.07 meters, with a height range of 1.5-1.8 meters.
Oral contraceptives (OC) were used by 25% of subjects (see Table 1). Condoms, tubal ligation, and abstinence or withdrawal were the next most frequently used methods, each reported at 17% prevalence. The most common choices of tampon absorbency were Regular (6-9 g) and Super (9-12 g).

Vaginal pH and prevalence and quantitative assessment of microorganisms
Mean vaginal pH was 5.6 ± 0.08 (standard error of the mean, SEM) and ranged from 4.1-6.1. The prevalence of genital microflora detected at each anatomic site is summarized in Table 2. The enterococci and GNEs were the most prevalent organisms at all sites, but their frequency decreased along the genital tract from the perineum, across the hymen, to the posterior fornix of the vagina.
Colonization with yeast species was detected below the hymen in five subjects (two with C. albicans and three with other species); C. albicans was found at the posterior fornix in another subject. The frequency of S. aureus, when found, was similar at all sites. Two subjects colonized with this organism were colonized at more than one genital site: one subject was colonized at all four sites; the other was colonized above and below the hymen. The cell densities of genital microflora detected at each anatomic site are also summarized in Table 2.
Quantitative differences in the cell densities of flora between sites (other than the perineum) were 1.3 log10 units/g or less for all organisms studied. Pairwise comparisons of the prevalence of microorganisms at each site among all study subjects (Table 3) revealed a statistically higher prevalence of enterococci and GNEs at the perineum relative to all other sites (p £ 0.018, stratified CMH test). The prevalence of enterococci and GNEs was lower at the posterior fornix relative to the two hymeneal sites: statistical significance was reached when comparing the sampling sites below the hymen and at the fornix (p = 0.018 and p = 0.025 for enterococci and GNEs, respectively) and approached when comparing above the hymen to the posterior fornix (p = 0.052 and p = 0.059 for enterococci and GNEs, respectively). There were no significant differences in isolation frequency of any organism above or below the hymen and no significant differences in the prevalence of yeast and S. aureus among any of the evaluated sites. The study had at least 80% power to detect a minimum difference in prevalence of 15% between sites at the 0.05 level of statistical significance.
Pairwise comparisons of the cell density of microorganisms below the hymen, above the hymen, and the posterior fornix (mean of all subjects including those for whom the organism was absent, Table 4), revealed significantly higher cell densities of enterococci below the hymen compared to the posterior fornix (p = 0.005, Wilcoxon Signed Rank test) and significantly higher cell densities of GNEs below and above the hymen relative to the fornix (p = 0.001 and p = 0.007, respectively). Cell density of enterococci was not statistically significantly different above the hymen compared to the posterior fornix (p = 0.055). No significant differences in bacterial cell densities  between locations below and above the hymen were observed for any organism evaluated. Also, no significant differences in bacterial cell densities between hymeneal sites and the fornix were observed for yeast and S. aureus. The potential relationship between the presence of an organism at one site and its presence at any other site was assessed using the kappa statistic (Table 5). There was strong above-chance association (i.e., a predictive agreement) between the presence of GNEs below and above the hymen (weighted kappa = 0.805) and a moderate abovechance association in the presence of any individual organism at these two sites (weighted kappa = 0.574). There was no predictive agreement of the presence (or absence) of an organism among other sites.

DISCUSSION
This study examined the isolation frequency and cell density of selected microorganisms in the lower genital tract of menstruating tampon-users at progressively distal locations from the perianal area. The enterococci and GNEs were chosen as indicators of potential colonization derived from the intestinal tract; yeast species and S. aureus represent endogenous flora of potential pathological significance. We controlled for several variables known to influence the microbiology of the genital tract: a single time point during menstruation 9 (day 3 of the cycle) was examined in healthy women who had no current or recent history of medication or spermicide use 8 , who refrained from using douches 28  *Mean cell densities expressed as log 10 cfu/cm 2 (perineum) and log 10 cfu/g (other sites); SEM, standard error of the mean; GNE, Gram negative enterics Table 2 Prevalence and log 10 counts of microbes at selected genital sites genital area, and who were willing to refrain from intercourse 8 for 24 hours prior to evaluation. Subjects with chlamydia, gonorrhea, trichomoniasis, clinically diagnosed active yeast infection, or BV were excluded to increase the likelihood that any site-to-site variability occurred within a presumptively 'normal' endogenous microbiological environment 9,11 . The mean vaginal pH of 5.6 in this study was consistent with that found by other investigators during the first few days of the menstrual cycle 29 . The menstrual period is a time of increased  Table 3 Pairwise comparison of microorganism prevalence at selected genital sites (all study subjects, n = 52) microbiological variability in the lower genital tract 3,8,9,27 . A greater variety of organisms are recovered at this time 1,7,12,17 , irrespective of the type of menstrual product used 12,13,26 .
The enterococci and GNEs were the most frequently isolated organisms in the present study and the pattern of genital colonization was consistent with the intestinal tract as a source. Specifically, the isolation frequency of enterococci and GNEs was inversely related to the distance from the anus: frequencies were significantly higher at the perineum than all other sites, and lower at the posterior fornix than the hymenal sites.
Enterococci are recovered more often during menses than nonmenstrually 3,7,26 . In the present study, the 15% isolation frequency of enterococci at the posterior fornix contrasts with a reported rate of 38% in vaginal samples from 242 menstruating pad users evaluated on days 2-3 of the cycle, using the same sampling method 26 . Wide variability of enterococci prevalence has been found in studies with less well-characterized populations, menstrual product usage, and sampling times. For example, enterococci prevalence rates of 1.4% were found in the vaginal fornix of 145 women with normal discharge 15 , compared with 15% in swab samples of the posterior vagina taken from 22 women at any time during the menstrual cycle 24 .
Like the enterococci, GNEs (and Escherichia coli in particular) have been isolated more frequently during menses or in the first half of the menstrual cycle 3,6,14,15,26,30 . Colonization with GNEs was significantly more common in pad users than tampon users 14,15,30 ; because pads cover the perineal area, the higher frequency of recovery of vaginal enterococci and GNEs in pad users is consistent with the gastrointestinal tract as the source of colonization.
The cell densities of enterococci and GNEs isolated from the posterior fornix and above the hymen sites (log10 3.2-4.3/g) were comparable to those found vaginally during menstruation in pad users 26 but were 2-3 log10 units per gram lower than reported in earlier vaginal flora studies 18,24 . Direct comparison of cell density with historical data is complicated by the fact that previous studies did not always specify sampling times or sampling methodologies, and host-related variables were not as rigorously defined.
Yeast species were found in six of the subjects (11%). Although up to 20% of women may be asymptomatically colonized with Candida at any   ligation. Hormonally mediated increases in yeast are most frequently associated with highestrogen-content OCs that are no longer commonly prescribed and recent studies indicate OC use has a minimal effect on the endogenous flora 10 . Cell densities of C. albicans in the present study were lower than has been reported previously 32 .
The vulva is the preferred genital carriage site of S. aureus 33,34 , with reported prevalence rates as high as 67% 33 . There is a statistical association between vulvar and vaginal carriage of this organism 7,21 but, using traditional culturing techniques, S. aureus is recovered only sporadically from the vagina and is more consistently found there during menstruation 16,17,26,27,35 . Using current culturing techniques, vaginal colonization rates at any time during the cycle reportedly range between 2 and 30% 7,16,21,22,26,27,35,36 . Our results are consistent: detection of vaginal S. aureus during menses was approximately 4% and two of four subjects were colonized with S. aureus at more than one genital site. Molecular techniques that classify via genotype may reveal higher prevalence rates 37 . The cell densities of vaginal S. aureus were 2-3 log10 units lower than has been previously reported 18,24 .
The microbiology of the female genital tract is a dynamic ecosystem. By analogy to environmental ecosystems, local anatomical and physiological variables, as well as competition between microorganisms, may alter the qualitative or quantitative colonization of various regions of the genital tract. However, we found only two studies that examined more than one organism at multiple genital sites in the same subjects 18,19 . These studies suggested that the cervix might be a distinct ecological niche: substantial differences in cervical and vaginal colonization were seen in the same individuals 18,19 . Anatomical differences, pH, the presence of columnar epithelium on the ectocervix, the influence of age and oral contraceptives on cervical ectopy and changes induced by parturition may selectively influence microbial colonization of this site. For example, C. trachomatis and N. gonorrhoeae preferentially infect the cervix, while the group B streptococci 20 and Mycoplasma hominis 38 are isolated less frequently from the cervix than the vagina.
Despite the common assumption of niche specialization, few studies have systematically attempted to correlate the presence or absence of particular organisms at a single site with colonization of another genital site, and only single species have been investigated [20][21][22][23] . The present study examined enterococci, GNEs, yeast and S. aureus at multiple sites. We found the agreement between the presence or absence of selected microorganisms among the sites did not exceed chance alone, with one notable exception: when locations just below and just above the hymen were considered, there was a moderate above-chance association between the presence of any of the study organisms just below the hymen and the presence of any study organism directly above the hymen. This relationship (evaluated via the kappa analysis) was strongest for the GNEs. Furthermore, the isolation frequency of evaluated organisms at these two sites did not differ and there were no quantitative differences in the cell densities of any individual species between them. Taken together, these observations suggest that with respect to these organisms, the regions of the vagina just below and just above the hymen are not microbiologically distinct.
This investigation was conducted exclusively among tampon users. Studies indicate that tampon use has little effect on the overall qualitative or quantitative composition of the vaginal flora [11][12][13][14]27 . It might be speculated that the tampon removal cord that extends from the vagina could be a conduit for organisms from the external genitalia, but there is no evidence that this is the case. On the contrary, results from this and other studies suggest that vaginal colonization with externally derived enterococci and GNEs is more common when pads are used 14,15,30 .
In summary, there was a decreasing prevalence of enterococci and GNEs from the perineum to the posterior fornix of the vagina, with similar prevalence of these organisms across the hymen sites. Prevalence of yeast was restricted to the above-or below-hymen sites; S. aureus was present at similarly low prevalence rates at all four sites. Prevalence and cell density above the hymen was similar to prevalence and cell density below the hymen for enterococci, GNEs, yeast, and S. aureus. The results suggest that the regions just above and below the hymen are not different with respect to the presence of these organisms. Our results are consistent with previous studies showing that the enterococci, GNEs, yeast and S. aureus are part of the endogenous genital flora in a well-defined population of healthy women. The examined flora was recovered at lower levels in this defined population than reported previously among women drawn from the population at large.