An Intravaginal Live Candida Challenge in Humans Leads to New Hypotheses for the Immunopathogenesis of Vulvovaginal Candidiasis

Subjects, enrollment, and screening.

Thirty healthy women between the ages of 18 and 55 years were enrolled. All eligible participants were enrolled through the Obstetrics and Gynecology Clinic at the Louisiana State University Health Sciences Center, New Orleans, La. Informed consent was obtained from all participants/patients, and all procedures in the conduct of clinical research were performed in accordance with the Institutional Review Board at the Louisiana State University Health Sciences Center. Prior to enrollment, a detailed questionnaire (information about past episodes of VVC and their causes) was administered and an extensive prescreening procedure (Candida-specific immune responsiveness, human immunodeficiency virus [HIV] test, pregnancy test, and vaginal yeast colonization status) was conducted. Candida-specific immune responsiveness was determined by a delayed- and immediate-hypersensitivity skin test for Candida. To determine immediate hypersensitivity, a surface scratch test was performed as previously described (19) using Candida antigen (Candin; Allermed Laboratories, San Diego, Calif.). To determine delayed hypersensitivity, an intradermal test was performed as previously described (19) with the same Candida antigen. A standard blood-based HIV test with appropriate pre- and post-test counseling and a standard human chorionic gonadotropin urine-based pregnancy test were conducted for each participant. Additionally, a vaginal swab was taken from each participant and cultured on Candida Chromagar (CHROMagar, Paris, France) at 37°C for 48 h to evaluate the vaginal yeast colonization status. Candida hyphae were evaluated by placing a drop of potassium hydroxide (KOH) on a vaginal smear made from the swab.

Specimen collection and laboratory analyses.

Specimens collected from enrolled subjects included a vaginal lavage specimen, vaginal swab, vaginal smears, and blood (10 ml) by venipuncture into a red-top tube (without EDTA) (Becton Dickinson, Franklin Lakes, N.J.). The vaginal lavage fluid was collected after a 30- to 40-s aspiration with 5 ml of nonpyrogenic sterile saline. The lavage fluid was processed as previously described (12), and the sterile filtered soluble fraction was stored at −70°C until use. Prior to filtration and storage, the lavage fluid was serially diluted and plated on Sabouraud dextrose (SAB) agar to quantify the vaginal fungal burden. The lavage cell pellet was stored at −70°C in cryopreservative medium (50% fetal bovine serum, 30% RPMI 1640 medium, 20% dimethyl sulfoxide [all from GIBCO, Langley, Okla.]) until use. The vaginal swab was plated on Candida Chromagar and incubated at 37°C for 48 h. Green CFU were further evaluated for germ tube formation (evidence for C. albicans) in fetal bovine serum for 2 to 3 h in a 37°C water bath. Non-germ-tube-forming colonies were identified to species level using API 20 AUX (bioMérieux, Hazelwood, Mo.). One vaginal smear was spray fixed and stained using the Papanicolaou technique. Leukocytes were identified by morphology. If present, the only leukocytes identified were polymorphonuclear neutrophils (PMNs). PMN scores were determined from the stained vaginal smear and defined as follows: 1, 1 to 10 PMNs/five fields at ×400 magnification; 2, 11 to 20 PMNs/five fields; 3, 21 to 30 PMNs/five fields; and continuing with a similar pattern up to 11, >100 PMNs/five fields. The second vaginal smear was Gram stained (as described below). Serum was collected from clotted blood following centrifugation and stored at −70°C until use.

Preparation of inocula.

The inocula were prepared using C. albicans clinical isolate DB597.94 from an RVVC patient; the isolate was kindly provided by J. D. Sobel, Wayne State University School of Medicine, Detroit, Mich. The isolate was grown on SAB agar at 34°C, and one colony was used to inoculate 10 ml of phytone-peptone (PP) broth (BBL, Syracuse, N.Y.) supplemented with 0.1% glucose and grown to stationary phase for 18 to 22 h at 25°C in a shaking water bath. The blastoconidia were collected and washed with phosphate-buffered saline (PBS), enumerated on a hemocytometer using trypan blue dye exclusion, and resuspended at specific concentrations in a variety of vehicles including PBS, PP, PP-0.1% glucose, and PP-0.1% glucose-1% estrogen (the general concentration employed in mice to stimulate or sustain vaginal infection). In one case, the known number of blastoconidia were incubated in tissue culture medium for 9 h to form hyphae and used as the inoculum.

Clinical protocol.

Eligible participants (immediate hypersensitive negative, delayed hypersensitive positive, HIV negative, nonpregnant, C. albicans culture negative, KOH negative for hyphae, not on any form of hormonal contraception, not having had more than two explained episodes of VVC in the past year, and with no known immunological disorders) were placed into one of two groups based on their history of VVC. Nineteen women were defined as having no history of VVC (no known episodes of VVC), and 11 women were defined as having an infrequent history of VVC (no more than two episodes in any year) due to antibiotic usage, oral contraceptive usage, hormone replacement therapy, and/or pregnancy (in most cases physician diagnosed). Women were asked to come to the clinic at a time coinciding with a stage of the menstrual cycle (within 5 days after menses [follicular phase] or days 17 to 19 [luteal phase]), at which time specimens were collected, vaginal pH was determined, and a KOH test was performed. Each participant was then given 10³ to 10⁷ blastoconidia intravaginally in a volume of 0.1 or 0.5 ml in one of several vehicles (PBS, PP, PP-0.1% glucose, PP-0.1% glucose-1% estrogen). Immediately after inoculation, a feminine pad was placed over the vagina and the subject remained in the stirrups for 15 min to reduce leakage. The pad remained in place for 2 h, was stored refrigerated, and was brought into the clinic on the next visit. The pad was swabbed on Candida Chromagar to determine a measure of leakage.

The participants were asked to return to the clinic an average of three times during a 7- to 14-day period for collection of specimens (vaginal lavage fluid, smear, and swab). The participants were also asked to collect self-obtained vaginal swabs daily for a total of five times on the days they did not attend the clinic and return them to the clinic at their earliest visit. Pilot studies where swabs known to be positive for Candida and were stored for up to 5 days refrigerated before plating, were positive for Candida in all cases. The swabs and lavage fluid were processed as described above. C. albicans that was recovered was confirmed by green colonies on Chromagar and germ tube formation. A symptom report card was given to each subject, and she was asked to record the severity of any symptoms or signs every day during the length of her participation. Signs included discharge and vulvar or vaginal erythema, and symptoms included vaginal burning, itching, and soreness. Each was quantified using a grading system of 0 (absent), 1 (mild), 2 (moderate), and 3 (severe). A self-obtained vaginal swab was collected and shipped overnight once every 2 weeks for 6 weeks after the end of the formal observation period and cultured as above. Topical antifungal treatment was offered to anyone who requested it at any time and was required for anyone who had a symptomatic infection at the end of the observation period.

Criteria for symptomatic infection.

The presence of hyphae on the KOH smear and in the vaginal lavage fluid, a positive swab culture, and the presence of visible signs and symptoms served as evidence of symptomatic infection. A cumulative grade of 1 to 3 was defined as mildly symptomatic, 4 to 8 was defined as moderately symptomatic, and >8 was defined as severely symptomatic. In addition to the criteria for infection, the organisms recovered were confirmed to be the same as the organism used for inoculation. For this procedure, several isolated colonies from the swabbed plates were grown overnight in yeast peptone dextrose medium at 30°C and genomic DNA was isolated by standard procedures. Genomic DNA (3 μg) was digested with EcoRI, and fragments were separated on a 0.8% agarose gel, transferred, and probed with the Ca3 probe (kindly provided by D. R. Soll, University of Iowa), which recognizes a repeat sequence in the Candida genome (27). The inoculating strain DB597.94 and the Ca3 reference strain were included as controls. In all cases, the recovered organism was identical to the inoculating organism.

Identification of the stage of the menstrual cycle.

The stage of the menstrual cycle was verified by measuring the estradiol and progesterone concentrations in sera by radioimmunoassay at the Clinical Endocrinology Laboratory, Division of Reproductive Endocrinology, Detroit Medical Center, Detroit, Mich.

Bacterial flora and determination of BV.

A portion of the vaginal lavage fluid was added to BBL Port-a-Cul transport vials (Becton Dickinson) and then aseptically aspirated from the vial. Three drops of the lavage fluid were placed on the appropriate media for the cultivation of both aerobes and anaerobes and then streaked for isolation to estimate semiquantitatively the concentration of each organism in the specimen. A subgroup of specimens were selected for performance of 10-fold dilutions to obtain quantitative CFU. Aerobic cultures were inoculated on BBL Columbia agar-5% sheep blood, BBL chocolate agar, or BBL human blood Tween medium and incubated at 37°C under a 7% CO₂ atmosphere for 2 to 5 days. Anaerobic cultures were incubated in an anaerobic jar (O₂ concentration of less than 5 ppm) at 37°C for 2 to 5 days. Isolated colonies were identified using the RapID ANA II system (Remel, Inc., Lenexa, Kans.) for anaerobes; API 20E (bioMérieux) for Enterobacteriacae; Gram stain and growth on inhibitory mold agar for yeast; Gram stain and beta-hemolysis of organism on human blood Tween medium for Gardnerella; Gram stain-3% hydrogen peroxide, BBL Staphyloslide latex test, BBL Streptocard acid latex test, BBL 6.5% sodium chloride (NaCl) broth, BBL bile esculin slants, and hemolysis on BBL Columbia agar-5% sheep blood for aerobic gram-positive cocci; and Gram stain for lactobacilli. All bacteria media were purchased from Becton Dickinson. The bacterial flora was scored by Nugent's criteria from a Gram stain (0 to 3, normal; 4 to 6, intermediate or disturbed flora; 7 to 10, bacterial vaginosis [BV]) (23).

Chemotaxis assay. (i) Neutrophil isolation.

Venous blood (20 ml) from healthy male and female adult volunteers was collected in preservative-free heparin tubes, mixed with equal volumes of a dextran-saline solution, and incubated for ∼20 min at room temperature. The leukocyte-rich plasma was then aspirated and centrifuged for 10 min at 250 × g. The cell pellet was resuspended in 0.9% NaCl, layered over 10 ml of Ficoll-Hypaque (Amersham Pharmacia Biotech, Uppsala, Sweden), and centrifuged for 40 min at 400 × g. The resulting pellet was then resuspended in 20 ml of cold 0.2% NaCl for 30 s to lyse any remaining red blood cells. Isotonic conditions were then restored by the addition of ice-cold 1.6% NaCl, and the mixture was centrifuged for 6 min at 250 × g. The cells were resuspended in chemotaxis buffer (RPMI 1640 containing 2 nM glutamine, 25 mM HEPES, and 1% bovine serum albumin BSA) (Sigma-Aldrich Co., St. Louis, Mo.), enumerated using a hemacytometer, and adjusted to the proper concentration in chemotaxis buffer.

(ii) Chemotaxis assay.

A transwell system (Corning Inc., New York, N.Y.) using the Boyden chamber principle (6) was used to determine neutrophil migration in vaginal lavage fluid from women who were symptomatically infected, asymptomatically colonized, or not colonized. Recombinant interleukin-8 (5 ng/ml) and RANTES (10 ng/ml) (R&D Systems, Minneapolis, Minn.) were used as positive and negative controls, respectively. C. albicans culture filtrate antigen (CaCF) and the supernatant of a 24-h C. albicans blastoconidia culture were used as controls for the presence of C. albicans antigen in the lavage fluid, and PBS was used as the true-negative control. All control experiments were done in triplicate, and sample experiments were done in duplicate. Chemotaxis buffer (150 μl) was added to the upper chambers of a transwell plate, and the chemoattractant (vaginal lavage or control fluids) was added to the bottom chambers at dilutions of 1:5 and 1:50. The chambers were equilibrated for 5 min at room temperature. Chemotaxis buffer (10 μl) containing 2 × 10⁴ neutrophils was then added to the upper chambers, and the mixture was incubated for 30 min at 37°C under 5% CO₂. After incubation, the upper chambers were removed and the neutrophils that migrated into the bottom chamber were harvested and quantified using a hemocytometer. Data were expressed as a chemotactic index (cell count in experimental chamber/cell count in buffer control chamber). An additional control included placing 2 × 10⁴ neutrophils in the bottom chamber and confirming the cell number by using a hemocytometer and the procedure described above.

Statistics.

Student's t test was used for comparisons between vaginal fungal burden and PMN score and the chemotaxis assay. Fisher's exact test was conducted for any effects of BV, menstrual cycle, and inocula on the outcome of inoculation. Linear-regression analysis was performed to generate the correlation coefficient for vaginal fungal burden versus qualitative PMN score. Significance was defined as P < 0.05 using a two-tailed test.

Results from the intravaginal challenge with live C. albicans in women with no history of VVC.

The first study initiated was a small trial with six women who had no history of VVC. Those eligible from the prescreening criteria were tested during either the follicular or luteal phase of the menstrual cycle. The inoculum of 10³ to 10⁵ stationary-phase C. albicans blastoconidia was given in a volume of 0.1 ml of sterile PBS. The women showed no evidence of immediate hypersensitivity in the vulva or vaginal area during the 15-min observation period. Results of the culture of the preinoculum swab and lavage fluid were confirmed negative for all women when obtained, and the inoculum was serially cultured to identify the actual amount given. The feminine pad worn for 2 h and presented at the first return visit, although positive for yeast in each case, showed no evidence of significant leakage (determined by the number of CFU on the SAB plate). At each return visit (day 2 or 3, 4 or 5, and 7), a vaginal swab and lavage fluid were collected, signs were recorded, and the women submitted the symptom report card. Of these six subjects, only one was categorized as acquiring a symptomatic infection (she showed symptoms, had a vaginal smear positive for hyphae by KOH, and had a vaginal swab that was culture positive for C. albicans). The mean symptom rating was mild (3.9), with signs that included a light discharge. The other five women either were colonized asymptomatically (n = 2) or had no detectable colonization in swabs and lavage fluid specimens (n = 3). There was no effect of inoculum concentration or stage of the menstrual cycle (follicular or luteal) on acquiring a symptomatic infection. Interestingly, the woman who became symptomatic had a vaginal cellular infiltrate consisting exclusively of PMNs (Fig. 1A), whereas those asymptomatically colonized (Fig. 1B) or not colonized showed no cellular infiltration. The symptomatic woman was instructed to use the sealed over-the-counter topical antifungal therapy and return to the clinic after 5 days. The KOH stain of the vaginal smear and the vaginal swab culture results showed complete resolution of the infection upon return to the clinic. Any minor signs and/or symptoms had been eliminated as well. Self-obtained swabs collected every 2 weeks over the next 6 weeks were negative for yeast. The women who were colonized remained so for a variable length of time (depending on the subject) but eventually became detectably culture negative over the 6-week self-swab follow-up. Follow-up phone conversations with all women for up to 1 year detected no adverse effects or problems.

Following this low infection rate, the protocol for the next series of women tested included a higher inoculum (up to 10⁸), longer period of observation (14 days), inoculum given in a volume of 0.5 ml, vaginal pH measurement, and modification of the vehicle in which the inoculum was given, including growth medium with or without glucose and with or without estrogen. Thirteen new women, again with no history of VVC, were enrolled. Of these, two women became symptomatic and were diagnosed with a C. albicans infection (mean symptom ratings were mild [2.0] to moderate [7.3] with signs that included discharge and/or erythema). Again, the women who became symptomatically infected showed a vaginal cellular response by PMNs. The women were treated as described above, with no adverse effects. Of the remaining women, six became asymptomatically colonized and five were not detectably colonized. Again, inoculum concentrations, stage of the menstrual cycle, and vaginal pH (which ranged from 4 to 6.5) had no effect on acquisition of the symptomatic infection, and there was no measurable enhanced susceptibility to symptomatic infection when the vehicle included a growth medium for fungi with or without glucose or estrogen to promote Candida growth. One subject was given the inoculum in the hyphal form. She also failed to become symptomatic or even detectably colonized. Women who became asymptomatically colonized became culture negative over time (1 to 2 months). Long-term follow-up via phone conversations showed no problems for a period up to 1 year. The overall incidence of symptomatic infection in these women with no history of vaginitis was 15%. Table 1 contains the details of each subject enrolled. Fisher's exact tests for the effects of low versus high inocula (<10⁵ versus >10⁵) or the menstrual cycle (follicular versus luteal) on susceptibility to infection were not significant (P = 0.40 and 1.0, respectively).

Results of the intravaginal challenge with live C. albicans in women with an infrequent history of VVC.

Based on the above results suggesting that the incidence of infection was based primarily on the host and not on the organism or inoculation conditions, we were granted approval to expand the inclusion criteria to include subjects with documented and explained infrequent episodes of VVC (i.e., VVC due to antibiotic usage, pregnancy, or oral contraceptive usage). We termed this group the “level 2” recruits. Inocula for the first of these prescreened eligible women included an intermediate inoculum (10⁵ to 10⁷) and escalating incremental inclusion of vehicle supplements (PP, glucose, and estrogen). No adverse safety issues associated with the vehicle supplements or inoculum concentration were observed in the first several subjects. Therefore, a more generous inoculum of 10⁷ Candida cells in PP with glucose became the standard inoculum. The results from 11 women showed that 6 became symptomatically infected with C. albicans (mean symptom ratings were mild [2.0, 2.0, 2.0, 3.8, and 3.9] to moderate [7.3], with signs that included discharge and/or erythema), 4 became asymptomatically colonized, and 1 was not detectably colonized. One of the culture-positive women, although having mild symptoms, felt the need to request treatment after 1 day postinoculation, and so no additional specimens were collected. Like those tested previously, symptomatic culture-positive women had a vaginal cellular infiltrate consisting exclusively of PMNs (78% overall) and the stage of the menstrual cycle and the vaginal pH had no effect on acquisition of a symptomatic infection (data not shown). The infected women were treated as above with no adverse affects. All the level 2 recruits were treated with the antifungal agent as a precautionary measure at the end of the observation period. All the women cleared the organism, and long-term follow-up with self-obtained swabs showed no detectable fungal growth. Follow-up phone conversations detected no problems over the subsequent 6 months to 1 year. Figure 2 illustrates the difference in infection rates for level 1 and level 2 participants.

Effect of intravaginal challenge with live C. albicans on the vaginal bacterial flora.

For the majority of the experimental protocol, vaginal lavage fluid and swab specimens were evaluated for changes in the bacterial flora and incidence of BV. The results showed that over the observation period, women who became symptomatically infected, asymptomatically colonized, or not colonized had no demonstrable changes in the type or concentration of bacterial flora. The floras included several anaerobic and aerobic bacteria such as Lactobacillus species, Staphylococcus species, diphtheroids, Streptococcus species, enterics, Gardnerella vaginalis, Prevotella species, Porphyromonas asaccharolyticus, Bacteroides fragilis group, and Peptostreptococcus species (data not shown). The incidence of BV in the subjects enrolled was ∼45%. There was no association between the presence of BV and the outcome of the inocula. This is illustrated in Table 1 for women without a history of VVC. The results for all women are as follows: of those with a symptomatic infection (n = 9), 3 were BV positive and 6 were BV negative; of those with asymptomatic colonization (n = 12), 5 were BV positive, 6 were BV negative, and 1 was intermediate; of those not colonized (n = 9), 5 were BV positive, 3 were BV negative, and 1 was intermediate. Fisher's exact test between infection (infected or uninfected) and BV (BV positive and BV negative) showed no significant differences (P > 0.22). There was also no appreciable change in BV status over the observation period.

Relationship between vaginal fungal burden and PMN score.

The vaginal fungal burden in all participants who became symptomatically infected (with or without a history of VVC) was significantly higher than that in those who became asymptomatically colonized (P = 0.0011) (Fig. 3A). Similarly, the PMN score was significantly higher in those who became symptomatic than in those who became asymptomatically colonized (P = 0.0005) (Fig. 3B). In fact, there was a positive correlation between PMN score and vaginal fungal burden in those with symptomatic infection (representing all visits where vaginal fungal burden was detected) (r = 0.500, P = 0.039). In some cases, PMNs were present in the vaginal lavage fluid at the time of inoculation and persisted unchanged throughout the observation period. However, only under conditions of symptomatic infection did the PMN score change significantly.

Chemotaxis assays with cervicovaginal lavage fluid.

To determine whether the lavage fluids from those with symptomatic infection could cause the migration of PMNs from a healthy volunteer, we evaluated different concentrations of lavage fluid from women preinoculated and those symptomatically infected, asymptomatically colonized following inoculation, or not colonized in a chemotaxis assay. The results in Fig. 4 show that the lavage fluid (diluted 1:5) from women with a symptomatic infection promoted the migration of PMNs while lavage fluids from women preinoculated and women either colonized or not colonized postinoculation did not (P < 0.02). No chemotaxis was observed with any fluid at a 1:50 dilution (data not shown). Additionally, CaCF and a 24-h supernatant from a Candida blastospore culture that was similarly diluted also showed no chemotactic activity.