The objective of this study was to examine genital tissue, vaginal fluid, and vaginal microbial flora at 3 phases of the menstrual cycle in asymptomatic women. Vaginal examinations were performed 3 times in 74 women: at the menstrual phase (days 1–5), the preovulatory phase (days 7–12), and the postovulatory phase (days 19–24). Flora of 50 women without bacterial vaginosis (BV) was analyzed separately from flora of 24 women with BV. The volume of vaginal discharge increased and the amount of cervical mucus decreased over the menstrual cycle. Among subjects without BV, the rate of recovery of any Lactobacillus changed little (range, 82% to 98%; P = .2); however, a small increase occurred in the rate of recovery of heavy (3+ to 4+ semiquantitative) growth of Lactobacillus over the menstrual cycle ( P = .04). A linear decrease occurred in the rate of recovery of heavy growth of any non- Lactobacillus species, from 72% at days 1–5 to 40% at days 19–24 ( P = .002). A linear decrease also occurred in the rate of recovery of Prevotella species, from 56% on days 1–5 to 28% on days 19–24 ( P = .007), while a small linear increase occurred in the rate of recovery of Bacteroides fragilis ( P = .05). Among subjects with BV, the only significant change was an increase in the rate of recovery of Lactobacillus , from 33% at days 1–5 to 54% at days 19–24 ( P = .008). Among all subjects, the rate of recovery of heavy growth of Lactobacillus increased over the menstrual cycle and, in contrast, the concentration of non- Lactobacillus species tended to be higher at menses, which is evidence that the vaginal flora becomes less stable at this time.
Many sexually transmitted diseases are transmitted more efficiently from men to women than from women to men [ 1 , 2 ]. The vagina represents a critical portal of entry for most sexually transmitted diseases in women. Both a low pH in the vagina and the microbicidal effects from H 2 O 2 produced by vaginal Lactobacillus [ 3 , 4 ] appear to inhibit many genital infections.
Little is known, however, about normal physiological variation in the vagina over the menstrual cycle or how this vari-ation may alter the acquisition of genital or systemic infection. In this study, we examined vulvar and vaginal tissue specimens grossly and by colposcopy, measured the amount of cervical and vaginal fluid, and evaluated vaginal flora for 74 asymptomatic women free of gonorrhea, chlamydial infection, tricho-moniasis, and symptomatic candidiasis during 3 phases of the menstrual cycle. The findings constitute normal baseline observations and describe physiological variations expected over the menstrual cycle, against which changes induced by exogenous hormone use, intercourse, or intravaginal product use can be compared. As part of this study, we separately report findings related to cytology, histology, and in situ immuno-staining for specific immune cells of the vaginal epithelium [ 5 ].
The only other large study to examine vaginal flora over the menstrual cycle included women with bacterial vaginosis (BV) [ 6 ]. In this study, we compared the vaginal flora of the 50 subjects without BV separately from the flora of the 24 subjects with BV. These 2 groups were distinguished because it is now becoming apparent that BV is not a continuum of normal vaginal flora [ 7 ] and that even asymptomatic women with BV have an increased risk of infection after surgery [ 8 ] and during pregnancy [ 9 ], and perhaps an increased risk of acquiring HIV infection [ 10 , 11 ]. By contrast, women with Lactobacillus -dominant flora appear to be protected against infections that originate from the vagina [ 4 ].
Materials and Methods
Study population. Complete data from all 3 phases of the menstrual cycle were available for 74 of 85 women who enrolled in our study from June 1996 through June 1997. Six subjects missed 1 or 2 of the 3 visits, and 5 had missing or inadequate vaginal biopsies. Inclusion criteria included the following: age 18–40 years, regular monthly menstrual cycles, no or only 1 sexual partner, and no use of contraception other than condoms. Exclusion criteria included the following: complaint of increased or odorous vaginal discharge; recent history of vulvar pruritus or irritation; yellow-appearing cervical or vaginal discharge; chronic medical disease (diabetes or hypertension); use of hormonal contraception, spermicides, or a intrauterine contraceptive device; use of systemic antibiotics in the past 6 months; use of vaginal suppositories or douching in the last week; and positive cultures for Neisseria gonorrhoeae, Chlamydia trachomatis , or Trichomonas vaginalis or symptomatic candidiasis or BV. Women who had asymptomatic BV at any of the 3 visits, as determined by gram staining criteria [ 12 ], were included with women who had Lactobacillus -dominant flora for the clinical analyses, but the groups were separated for the microbiological analyses.
The first enrollment visit was at days 7–12 of the menstrual cycle. At enrollment, a detailed intake interview was conducted to assess the subjects' demographic characteristics, current symptoms, vaginal product use, contraceptive use, sexual history and habits, and past pregnancies. A standardized vulvar, vaginal, and cervical examination was performed. During this and subsequent examinations at days 19–24 of the same cycle and days 1–5 of the next menstrual cycle, observations were recorded of the visual appearance of the vaginal and cervical epithelium and any discharge. Results of the colposcopic examination of the vagina were recorded on standard forms based on criteria of the World Health Organization [ 13 ].
Three milliliters of PBS was injected into the vagina, all remaining vaginal discharge was mixed with PBS, and the discharge was removed with a sterile pipette. As the estimate of the additional volume returned in the aspirate, the amount of vaginal discharge was estimated as scant (<1 mL), normal (∼1–3 mL), or copious (>3 mL).
Cervical and vaginal swabs were used to obtain material for culture, gram staining, and determination of vaginal pH. The pH was determined by placing Color-pHast indicator strips (EM Science, Gibbstown, NJ) on the vaginal wall. A wooden stick for Papanicolaou smear was used to obtain cervical and vaginal cells for cytological analysis. Swabs rubbed on the vaginal wall were used to obtain vaginal cells for future bacterial attachment studies. A urine specimen for culture was obtained at each visit. At the second (days 19–24) and third (days 1–5) visits, an interval history was obtained of symptoms, vaginal product and antibiotic use, and sexual exposure. At these visits, the examination was performed and vaginal samples and culture specimens were obtained in the manner described for the first visit. Data are presented in tables 2–4 in the order of days in the menstrual cycle.
Microbiological studies. Cervical specimens were collected on Dacron swabs (Baxter Healthcare, Deerfield, IL) for identification of C. trachomatis by culture on McCoy cells [ 14 ] and of N. gonorrhoeae by culture on Thayer-Martin media. Cervical mucus was rolled onto a glass slide, and WBCs were counted described elsewhere [ 15 ].
Vaginal culture specimens were collected from the vaginal discharge with Dacron-tipped swabs with plastic shafts (Puritan brand, Hardwood Products, Guilford, ME), placed in Port-A-Cul transport media (Becton Dickinson Microbiology Systems, Cockeysville, MD), and transported to the laboratory within 6 hours of collection. Two swabs each containing ∼96 pg of vaginal fluid with a mean weight of 0.194 g [ 7 ] were removed from the transport tube, placed in 1.5 mL of PBS, and vigorously centrifuged. The mixture is a ∼1∶10 dilution of vaginal fluid [ 7 ]. One hundred microliters was inoculated onto each of the following media: sheep blood agar plates (Columbia base, PML Microbiologicals, Tualatin, OR), 2 human blood bilayer Tween agar plates (PML Microbiologicals), a chocolate agar plate (prepared in-house), a brucella agar plate with sheep blood (prepared in-house), a laked blood kanamycin agar plate (PML Microbiologicals), and a Rogosa selective lactobacillus agar plate (prepared in-house). A8 agar and selective broths were inoculated for recovery of Mycoplasma hominis and Ureaplasma urealyticum. The blood agar plates, 1 human blood bilayer Tween agar plate, the chocolate agar plate, the Rogosa selective lactobacillus agar plate, and the media for recovery of genital mycoplasmas were incubated at 37°C in 5%–10% CO 2 for 48 hours. The remaining media were incubated at 37°C within an anaerobic glove box (Sheldon Laboratories, Cornelius, OR) for a minimum of 5 days of uninterrupted anaerobiosis.
The anaerobic bacteria were identified by methods described elsewhere [ 7 ]. Mycoplasmas were identified by their characteristic morphology on the agar plate. Facultative bacteria were identified by standard microbiological methods [ 7 ]. As previously reported, 3+ semiquantitative growth was consistent with 10 6 cfu/mL of vaginal fluid, and 4+ semiquantitative growth was consistent with ≥10 7 cfu/mL of vaginal fluid [ 16 ].
Lactobacilli were identified by colony morphology and characteristic results of gram staining. No effort was made to identify lactobacilli to the species level. All lactobacilli were tested for production of H 2 O 2 in a qualitative assay on a tetramethylbenzidine agar plate [ 17 ]. After 2–3 days of incubation in an anaerobic glove box at 37°C, the agar plates were exposed to the ambient air for ≤30 min. The H 2 O 2 produced reacts with horseradish peroxidase in the agar and oxidizes tetramethylbenzidine, which causes the Lactobacillus colonies to turn blue. Midstream urine cultures were processed as described elsewhere [ 18 ].
Statistical analysis. SPSS (SPSS, Chicago) and Splus (Math Soft, Cambridge, MA) were used for statistical analyses. Trends over time for the same individual were examined on the categorical (dichotomous) variables in tables 2–4 using logistic regression for correlated data [ 19 ]. Linear regression for correlated data was also performed on the continuous variables. This model assessed a linear trend for the same subject over the 3 sampling times in the menstrual cycle using the generalized estimation equations function in Splus.
Demographics, history of pregnancy, current birth control, and sexual and douching history for the 74 women are shown in table 1 . Subjects in the study were predominantly young, single, white, nulliparous university students. Subjects used either no contraception or condoms (hormonal contraception, spermicides, and intrauterine contraceptive device use were exclusion criteria). Most subjects were currently sexually active, and contraception was used by 90% of 42 sexually active subjects and 45% of 32 subjects not sexually active. Douching was common, but no subject douched >3 times per month. Prior urinary tract infection had occurred in 38 (51%) of 74 subjects. The 24 subjects with BV were younger by about 3.6 years ( P = .02) and more often single (96% vs. 70%, respectively; P = .01) than the 50 subjects without BV, but other characteristics were similar.
Table 2 shows the results of clinical examinations for the 74 subjects at the 3 times of the menstrual cycle. There were no differences in results between the 50 women without BV and the 24 women with BV. Significantly more subjects had vulvar erythema on days 7–12 and 19–24 of the menstrual cycle than on days 1–5 ( P < .001). Vulvar erythema and pruritus were associated with the recovery of Candida albicans on days 19–24. On days 19–24, C. albicans was recovered from 5 of 16 subjects with vulvar erythema compared with 4 of 58 subjects without vulvar erythema ( P = .02) and from 4 of 6 subjects with pruritus compared with 5 of 67 subjects without pruritus ( P < .002). There was no association between pruritus and vulvar erythema in this small group. Pruritus was present in 3 of 15 subjects with erythema and 3 of 58 subjects without erythema at days 19–24 ( P = .1). Pruritus was present in 1 subject on days 1–5 and 0 on days 7–12.
A small vulvar fissure was present in 1 subject on days 7–12 and 3 different subjects on days 19–24. Two subjects had a small vulvar ulcer on days 19–24. Localized vulvar erythema was present in 3 subjects on days 1–5, 9 subjects on days 7–12, and 16 subjects on days 19–24. Ten of the subjects had erythema at 2 visits. No vulvar molluscum, excoriations, vesicles, pustules, or warts were visible. The only vaginal abnormality noticed grossly was mild erythema, identified in 2 subjects on days 1–5 and 3 different subjects on days 19–24. Colposcopy identified vaginal erythema in 1 subject, localized petechial hemorrhage in 2 subjects, and an abrasion on 2 subjects.
A significant increase in the estimated volume of vaginal discharge was observed over the menstrual cycle; a higher proportion of women had a normal amount (1–3 mL) later in the cycle ( P < .001). The subjective amount of vaginal discharge reported by the subjects did not change between days 1–5 and 19–24, and only 1 subject without BV had a vaginal odor on days 1–5 (data not shown). The color of vaginal discharge was white or clear in most subjects. The distribution of vaginal WBCs per high-power (×400) field was unchanged over the menstrual cycle. Vaginal and cervical WBC counts did not correlate (Spearman's rank correlation; P = .5). Vaginal discharge consistency (normal in 65 subjects, homogeneous in 8, and curdy in 1) and distribution (pooled in 48, diffuse in 24, and patchy in 2) at days 7–12 did not change over the cycle (data not shown).
The number of women with moderate or profuse amounts of cervical mucus observed in situ decreased over the menstrual cycle ( P < .001; table 2 ). The distribution of cervical WBCs remained relatively constant over the menstrual cycle, although later in the cycle a higher proportion of subjects had >50 WBCs per high-power field.
Selected vaginal microorganisms recovered from the 50 subjects without BV are listed in table 3 . Recovery of any Lactobacillus changed little during the menstrual cycle ( P = .2); Lactobacillus was recovered from 82%–98% of the subjects at 1 of the 3 visits over the menstrual cycle. There was a small significant change in the number of subjects with a high (3+ to 4+) semiquantitative level of Lactobacillus ( P = .04), but only 70% of subjects had high levels of Lactobacillus during menses. There were increases in high (3+ to 4+) levels of both H 2 O 2 -positive Lactobacillus and H 2 O 2 -negative Lactobacillus over the 3 points in the menstrual cycle ( table 3 ).
Non— Lactobacillus microorganisms were isolated from 90% to 96% of subjects at 1 of the 3 visits over the menstrual cycle ( table 3 ). The rate of recovery of heavy (3+ to 4+) growth of any non- Lactobacillus species was highest on days 1–5 and decreased thereafter ( P = .002; table 3 ). When the presence of U. urealyticum and M. hominis was excluded, there was still a significant decrease in high levels of non- Lactobacillus bacteria ( P = .004; table 3 ). The recovery of low (1+ to 2+) levels of non- Lactobacillus flora significantly increased over the cycle (data not shown; P = .002), indicating a tendency for a shift from high (3+ to 4+) to low (1+ to 2+) levels of non- Lactobacillus flora from menses to the later parts of the menstrual cycle.
There were no significant changes in aerobic flora over the menstrual cycle ( table 3 ). A slight nonstatistically significant downward trend occurred in the recovery of group B streptococci, Escherichia coli , and Gardnerella vaginalis , and a slight nonstatistically significant increase occurred in the recovery of C. albicans and U. urealyticum . The recovery of any Prevotella species, on the other hand, decreased significantly from days 1–5 to days 19–24 ( P = .007). In contrast, the recovery of Bacteroides fragilis increased significantly over the menstrual cycle ( P = .05).
Gram staining characteristics of BV and intermediate flora in the 74 subjects (including the 24 subjects with BV) decreased significantly over the menstrual cycle ( P = .05). BV was present in 20, 15, and 18 subjects and intermediate vaginal flora was present in 15, 9, and 12 subjects over the 3 times of the menstrual cycle. One subject had a urine pathogen (≥10 5 organisms/mL) at the days 1–5 and 19–24 visits.
Vaginal isolates from the 24 subjects with BV are compared over the 3 times of the menstrual cycle in table 4 . The number of subjects with any and with high levels of Lactobacillus increased significantly over the menstrual cycle, due mainly to a slight increase in the number of women with H 2 0 2 -positive Lactobacillus (which is in contrast to the group of 50 subjects without BV where there was no increase in the number of women with H 2 0 2 -positive Lactobacillus ). As expected, the total number of subjects with Lactobacillus at any of the 3 visits was lower in the group of subjects with BV than in the group of subjects without BV, since these groups were differentiated by gram staining criteria, including Lactobacillus morpho-types. A higher number (83%–96%) of subjects with BV, compared with those without BV, had high (3+ to 4+) levels of non- Lactobacillus species, and a higher number of subjects with BV had G. vaginalis, U. urealyticum, M. hominis, Prevotella species, and black anaerobic gram-negative rods (a heterogeneous group of anaerobes associated with BV) than did those with Lactobacillus -dominant flora. In the group of subjects with BV, there was no change in the number of women with these microbes over the menstrual cycle. In contrast, in the group of subjects with Lactobacillus -dominant flora, the level of Prevotella species increased significantly over the menstrual cycle. Thus there was a trend toward an increase in levels of H 2 O 2 -positive Lactobacillus in subjects with BV, but no change in the other flora (particularly anaerobic flora) associated with BV. Two subjects had a urine pathogen (≥10 5 organisms/mL) at the days 19–24 visit.
The mean vaginal pH ± SE at days 7–12 for subjects without BV (4.0 ± 0.04; table 3 ) was lower than that for those with BV (4.61 ± 0.08; P = .004; table 4 ). The vaginal pH at days 19–24 was also lower in subjects without BV than in those with BV ( P = .001).
This study comprehensively evaluated 74 asymptomatic subjects who did not have symptomatic lower genital infection and were not using systemic or vaginal contraception to determine changes in tissue, fluid, and microflora that normally occur in the vagina at 3 times in the menstrual cycle. Because genital infection [ 3 , 4 ] and both systemic contraception and vaginal contraception can influence vaginal flora, this study is the first to describe vaginal flora without these influences in a sexually active middle class group of young women with and without BV. The high rate of recovery of Lactobacillus among subjects without BV is consistent with other reports in which Lactobacillus species were recovered from 96% of pregnant women with normal vaginal gram stains [ 7 ]. Lactobacillus was even recovered from most menstruating women in other studies that did not take into account genital infection and use of contraception [ 20 ]. Lactobacillus was recovered from all but 1 subject at some point of the menstrual cycle, and most women without BV (70%–92%) from whom Lactobacillus was isolated had a high (3+ to 4+) semiquantitative level of Lactobacillus at all points during the menstrual cycle. The 3+ semiquantitative level found here corresponds to a mean of ≥10 6 cfu of Lactobacillus /mL, similar to the concentration found in pregnant women with a normal gram stain [ 7 ].
In this report, the number of subjects with high concentrations of Lactobacillus increased significantly over the menstrual cycle independent of the presence of BV. We did not determine the number of lactobacilli as compared with other bacteria, but in other reports, Lactobacillus species accounted for 90%–95% of the absolute number of vaginal bacteria present in women with Lactobacillus -dominant flora (and no BV) [ 7 ]. Therefore, Lactobacillus species were probably the dominant vaginal bacteria for >90% of these subjects without BV.
In other reports, Lactobacillus species appeared to inhibit microorganisms such as G. vaginalis and selected anaerobic bacteria associated with BV [ 3 , 4 ] and to inhibit genital infections, including those with C. trachomatis [ 3 ] and C. albicans [ 4 ]. H 2 O 2 -producing Lactobacillus also appears to inhibit E. coli colonization of the vagina [ 21 ] and catalase-negative bacteria [ 22 ] and HIV in vitro [ 23 ]. However, the relationship between vaginal microbiology, menses, and levels of estrogen is complex. In our study, the highest rate of recovery of heavy (3+ to 4+) growth of non- Lactobacillus species occurred at days 1–5. This heavy growth could occur because of an additional substrate from menstrual blood and might represent temporary instability of vaginal flora at menses or lower estrogen levels at menses. In most women, the flora appears to stabilize with Lactobacillus dominant and lower (3+ to 4+) levels of non- Lactobacillus microbes for the remainder of the menstrual cycle.
The administration of injectable depomedroxyprogesterone acetate over 6 months produced a large reduction in serum estradiol levels and a linear decrease in the level of H 2 O 2 -producing Lactobacillus , although the total number of subjects from whom Lactobacillus was isolated did not change, because of a slight increase in the level of H 2 O 2 -negative Lactobacillus (L. Miller, D. L. Patton, A. Meier, et al., unpublished data). The recovery of other non- Lactobacillus bacteria also did not change with depomedroxyprogesterone acetate use (L. Miller, D. L. Patton, A. Meier, et al., unpublished data). A different study design than the one used here would be required to differentiate the effects of menstrual blood from those of estrogen levels on vaginal flora.
A statistically significant linear decrease occurred in the recovery of Prevotella species from subjects without BV over the menstrual cycle. An earlier report also showed a decrease in the rate of recovery of Bacteroides species in the later part of the menstrual cycle [ 6 ]. With the exception of Prevotella (and to a small degree group B Streptococcus, E. coli , and G. vaginalis ), there was little decrease in the levels of other vaginal microorganisms over the menstrual cycle in women without BV. Thus, most vaginal flora microorganisms were maintained at a rather stable rate over these 3 points of the menstrual cycle. An exception was the B. fragilis group: there was a small linear increase in the number of subjects with B. fragilis in the later part of the menstrual cycle.
Lactobacillus species could modulate other vaginal microbes producing lactic acid, H 2 0 2 [ 7 , 17 ], or other bacteriocidins [ 24 ] or by competing for nutrients or bacterial adherence sites. Vaginal pH increases when menstrual blood is in the vagina, but as shown by our data, the vaginal pH at other times of the cycle is normally 4.0–4.5 ( table 2 ) [ 25 ]. This low pH is attributed to the production of lactic acid as a by-product of lactobacillus metabolism. However, lactic acid is also produced by the conversion of glycogen to lactic acid by vaginal epithelial cells when high levels of estrogen increase the glycogen concentration within the cells [ 26 ]. Most vaginal microbes grow poorly at the low pH normally present in the vagina of women without BV, which may explain some of the apparent inhibition of non- Lactobacillus species and HIV [ 27 ].
Additional inhibitory effects of Lactobacillus on other bacteria could result from H 2 O 2 production by lactobacilli. H 2 O 2 , together with a halide ion and the enzyme peroxidase, inhibits the in vitro growth of bacteria that do not contain catalase peroxidase, an enzyme that can detoxify H 2 O 2 [ 28 ]. Typical of this system, the in vitro inhibition of vaginal bacteria and HIV by H 2 O 2 -producing Lactobacillus is greatly increased by the addition of peroxidase and chloride [ 22 , 23 ]. This system appears to operate as a potent extracellular mechanism of bacterial killing in the vagina, which contains halide ions such as chloride, peroxidase, and, in women without BV, H 2 O 2 produced by many strains of Lactobacillus [ 22 ].
We found only 6 reports of studies that examined vaginal flora changes over the menstrual cycle [ 6 , 29–33 ]. Only 3 of these studies examined >10 subjects [ 30 , 31 , 33 ]. Although 2 studies excluded subjects who were using oral contraceptives or intrauterine contraceptive devices [ 6 , 30 ], none of the 6 previous studies excluded women who were using vaginal spermicide, excluded detection of cervical or vaginal infection by methods other than analysis of wet mounts (with 1 exception, [ 33 ]), or examined H 2 O 2 production by Lactobacillus . Most importantly, flora of women with the common condition of BV was not analyzed separately; as our data shows, grouping these subjects with subjects without BV tends to blunt results. The small number of subjects limits a comparison of results, but the 1 study with 34 subjects found no change in the level of Lactobacillus over the menstrual cycle, a finding perhaps attributed to inclusion of subjects with and without BV [ 30 ]. Consistent with our data, which demonstrate some degree of flora destabilization at menses, another study [ 31 ] has shown that the menstrual specimen contained the highest number of bacteria at the lowest concentration, and specimens from days 17–26 contained the fewest number of bacteria at the highest concentration.
Gross and colposcopic abnormalities of the vaginal epithelial surface were infrequent. It was of interest that C. albicans was associated with both vulvar erythema and pruritus on days 19–24 of the menstrual cycle. This finding was consistent with other studies that have found that symptoms of C. albicans infection tended to occur in the later part of the menstrual cycle [ 34 ]. This timing is of interest because the growth of C. albicans appears to be stimulated by estrogen (L. Miller, D. L. Patton, A. Meier, et al., unpublished data), and estrogen levels peak during the middle part of the cycle and decrease in the later part of the cycle [ 35 ]. Perhaps the increase in the estrogen level in the middle part of the cycle stimulates the growth of C. albicans , but 1–2 weeks pass before symptoms occur.
Only 8% of our subjects free of known bacterial infection of the cervix and vagina had a high number of WBCs (>11 WBCs per high-power field) in vaginal discharge. However, ∼20% of this population had a high number of WBCs (>31 WBCs per high-power field) in cervical mucus, a finding similar to that for women without known bacterial infection of the cervix who attended sexually transmitted diseases clinics [ 15 , 36 ]. Both the cervix and the vagina contribute to WBCs in the vagina, but about 20% of our subjects without evidence of bacterial infection of the cervix had a large number of WBCs in cervical mucus. The cause of this inflammation needs to be analyzed with new nonculture DNA technology.
The influence of intercourse, oral contraception, injectable progesterone, nonoxynol 9 products, and douching on the vaginal characteristics examined in this report is under study. Subjects in our study who did not have BV and did not use these products had few gross or colposcopic abnormalities of the vagina.