Detection of Trichomonas vaginalis Using the Polymerase Chain Reaction in Pregnant and Non-Pregnant Women

Objective: Trichomonas vaginalis vaginal infections are often both asymptomatic and difficult to detect by current methods. We evaluated the ability of a newly developed polymerase chain reaction (PCR) assay to identify T. vaginalis in vaginal samples from pregnant and non-pregnant women. Methods: In the 1st study, we compared the prevalence of T. vaginalis detection by PCR and culture using Diamond's medium in 52 women with symptoms of vaginal infection. In the 2nd study, T. vaginalis was detected using PCR and wet mount microscopy in 131 asymptomatic pregnant women. Results: Among the women with symptoms of vaginitis, 7 (13.5%) were PCR-positive for T. vaginalis. Six of the PCR-positive women, but none of the PCR-negative women, were culture-positive for this organism. All but 1 of the women with candidal vaginitis or bacterial vaginosis were PCR-negative for T. vaginalis. Among the asymptomatic pregnant women, all of whom were negative for T. vaginalis by wet mount, l0 (7.6%) were PCR-positive for T. vaginalis. Conclusions: PCR offers a rapid and sensitive alternative to culture and microscopy for the detection of T. vaginalis vaginal infections in both symptomatic and asymptomatic women.

zoan, is a common sexually transmitted infectious pathogen worldwide. Trichomoniasis is a frequent cause of vaginitis, exocervicitis, and urethritis. T. vaginalis infections are frequently asymptomatic and detectable only using culture techniques, z'3 Recent studies implicate both symptomatic and asymptomatic T. vaginalis infections in the pathogenesis of preterm birth, preterm rupture of membranes, 4 and post-hysterectomy cuff infections, s Despite the frequency of this infection, diagnosis oftrichomoniasis remains problematic. The most common clinical method, direct microscopic examination of vaginal fluid ("wet preparation"), has a sensitivity of only 35-80%. 6 Vaginal or urethral culture using Diamond's medium has a sensitivity >90% for detecting T. vaginalis. '7 Some isolates, however, grow poorly or not at all in this medium. 7 In addition, the use of Diamond's medium is labor intensive, requiring daily examination for up to 7 days before the presence or absence of T. In this study, we report on the further evaluation of this PCR assay to detect T. vaginalis in pregnant and non-pregnant women.

Subjects
The study groups consisted of 52 consecutive nonpregnant reproductive-age women with symptoms (discharge, burning, pruritis) of vaginitis evaluated at the University Hospital, Denver, CO, and 131 consecutive asymptomatic pregnant women seen at Cornell University Medical College, New York, NY. The pregnant women were between 15 and 40 years old with a singleton gestation <37 weeks and intact membranes.

Microbial Analysis
At both sites, samples were obtained from the posterior fornix with sterile cotton swabs and utilized to prepare a slide for a microscopic "wet preparation" saline examination. In Colorado, a 2nd swab was obtained and used to inoculate modified Diamond's medium (Remel, Lenexa, KS). Cultures were incubated at 37C in a 5% CO 2 atmosphere for 7 days. At daily intervals, beginning on day 2, aliquots were removed and examined microscopically for motile T. vaginalis. One swab was frozen at -70C.
The presence of Candida in the vaginal samples was assessed by examining wet mounts in 10% KOI-I for the presence 6f pseudohyphae and/or budding yeast in 10 high power (x 400) fields.
Bacterial vaginosis was diagnosed clinically when at least 3 of the following 4 criteria were met: pH > 4.5; thin homogeneous adherent discharge; presence of clue cells; amine odor on KOH wet mount.

PCR Analysis
The procedure of Riley et al. was utilized. Briefly, liquid was extracted from the previously frozen swab by squeezing with a sterile Pasteur pipette and transferred to a microcentrifuge tube. A pellet fraction was obtained and washed twice with phosphatebuffered saline (PBS). Lysis of T. vaginalis and exposure of its DNA were accomplished by resuspending the pellet in 0.25 ml Tris-HC1, pH 8.3, containing 50 mM KC1, 2.5 mM MgCI2, 1% Brij 35 detergent, and 200 Ixg/ml proteinase K and incubating in a thermal cycler for 60 min at 56C.
The proteinase K was then inactivated by raising the temperature to 95C for 10 min. Purified T. vaginalis from a symptomatic patient was always processed in parallel to the test samples. Buffer was processed as a negative control. Processed samples were stored at 4C until tested.
For the analysis, 1-p.l aliquots were mixed with 24 p.l H20 heated at 94C for 7 min to separate the DNA strands, and plunged into ice-H20 to prevent reannealing. An equal volume of reaction mixture was then added to yield 10 mM Tris-I--IC1, pH 8.3, 0.5 units Taq DNA polymerase, 1.5 mM MgC12, 50 mM KCI, 200 mM each dATP, dCTP, dGTP, and dTTP, and 2 oligonucleotide primer pairs specific for a 102 base pair region of the T. vaginalis genome. The samples were vortexed, microcentrifuged, and subjected to 40 cycles of sequential incubations at 94C for min, 47C for rain, and 67C for rain followed by a 7-rain extension cycle at 67C. A portion of the final reaction product was digested with Hinfl endonuclease as described. 7 The reaction products were subjected to electrophoresis on 6% polyacrylamide gels, stained with ethidium bromide, and visualized under ultraviolet light. A sample was consid- PCR results were not available to clinicians. All samples were assayed blindly without knowledge of clinical or microbiological findings. Since the PCR analyses were performed on frozen samples, after the women were no longer pregnant or being seen for vaginitis, the subjects were not treated based on PCR findings.

RESULTS
Among 52 non-pregnant women with symptoms of a vaginal infection, 7 (13.5%) were positive for vaginal T. vaginalis by PCR. Six of these 7 women were also positive for T. vaginalis by culture in Diamond's medium. The remaining women were all negative for T. vaginalis. Bacterial vaginosis was diagnosed in 12 (23.1%) of the women, only of whom was T. vaginalis positive by both PCR and culture. Similarly, of the 13 (25.0%)women with a Candida infection, only had evidence of T. vaginalis by PCR and culture. These results are summarized in Table 1.
None of the 131 pregnant women were positive for T. vaginalis by wet mount. However, PCR analysis revealed that 10 (7.6%) of these women harbored T. vaginalis.
The times of sample collection, gestational ages at delivery, and pregnancy outcomes of the PCRpositive pregnant women are shown in Table 2. Seven of the samples were not obtained until the 3rd trimester; the other 3 samples were obtained during the 13th, 18th, and 25th weeks of gestation. Eight (80%) of the pregnancies were uneventful and delivered at term (>36 weeks), including the woman whose trichomoniasis was detected at the 13th week. Two of the PCR-positive pregnant women delivered preterm at 36 weeks. One of these women delivered a small-for-gestational-age infant. DISCUSSION Among symptomatic women, PCR was at least as sensitive as culture in modified Diamond's medium in detecting T. vaginalis. The PCR assay was consistently negative in women whose symptoms were due to bacterial vaginosis or vaginal candidiasis. The single PCR-positive, culture-negative case most likely represented a T. vaginalis isolate that was unable to replicate in culture. 7 Alternatively, the concentration of T. vaginalis might have been too low to allow for visualization after a 7-day incubation period in Diamond's medium. PCR analysis can be completed in several hours, as opposed to 7 days for culture. This assay can be readily incorporated into the repertoire of any laboratory that performs PCR for detection of other microorganisms. Timely, definitive diagnosis of Trichomo-haS using PCR would eliminate the need for empiric metronidazole treatment.
Not surprisingly, PCR was superior to wet mount in identifying T. vaginalis in asymptomatic pregnant women. Our identification of T. vaginalis by PCR in 7.6% of these women indicates that this protozoan may be more prevalent than previously suspected in this population. Since preliminary results suggest that both symptomatic and asymptomatic women with T. vaginalis infection suffer in-creased risks of preterm birth and premature rupture of membranes, 4 methods to increase the detection rate of this organism, and subsequent treatment, may improve pregnancy outcomes in infected women. PCR analysis may be especially beneficial for those women who are negative for T. vaginalis by wet mount but whose symptomatic vaginitis remains unexplained. Further studies are required to confirm the utility of this important diagnostic test as well as to demonstrate the clinical benefits of treating pregnant and non-pregnant women whose trichomoniasis was identified only by PCR.

ACKNOWLEDGMENTS
These studies were supported in part by a grant from the Eleanor Naylor Dana Charitable Trust.
The technical assistant ofYvonne Schaeffer is gratefully acknowledged.