Share this article on:

A Multiplex Polymerase Chain Reaction–Based Diagnostic Method for Bacterial Vaginosis

Obata-Yasuoka, Mana MD; Ba-Thein, William MD, PhD; Hamada, Hiromi MD, PhD; Hayashi, Hideo MD, PhD

Original Research

OBJECTIVE To develop a polymerase chain reaction (PCR)-based diagnostic method for bacterial vaginosis using bacterial vaginosis–associated anaerobes.

METHODS A multiple PCR assay was developed using primers specific to 16S ribosomal deoxyribonucleic acid (DNA) (Mobiluncus mulieris and Mobiluncus curtisii), nanH (Bacteroides fragilis), and an internal spacer region of ribosomal DNA (Gardnerella vaginalis). The vaginal swabs from pregnant and nonpregnant women were examined by Gram stain–based Nugent scoring system. One hundred seventy-two samples of 853 Gram stain–interpretable samples were randomly selected and subjected to multiplex PCR assay.

RESULTS The sensitivity of the PCR assay ranged from 103 to 104 colony-forming units per vaginal swab. The prevalence of the bacterial vaginosis, intermediate, and normal categories was found by Nugent scoring system to be 21.6% (184/853), 26.0% (222/853), and 52.4% (447/853), respectively. By the multiplex PCR–based diagnostic method, 20.3% (35/172) of the samples were identified as bacterial vaginosis. The diagnostic sensitivity, specificity, positive predictive value, and negative predictive value of multiplex PCR in comparison with Gram stain examination were 78.4% (95% confidence interval [CI] 65.1%, 91.6%), 95.6% (95% CI 92.1%, 99.0%), 82.9% (95% CI 70.4%, 95.4%), and 94.2% (95% CI 90.3%, 98.1%), respectively.

CONCLUSION This multiplex PCR can be used as a diagnostic or screening test for bacterial vaginosis.

Positive amplification of one of the anaerobes, Mobiluncus species, Bacteroides fragilis, and Gardnerella vaginalis, is indicative of bacterial vaginosis.

Department of Obstetrics and Gynecology, Institute of Clinical Medicine, and Department of Infection Biology, Institute of Basic Medical Sciences, University of Tsukuba, Tsukuba, Japan.

Address reprint requests to: Mana Obata-Yasuoka, MD, Institute of Clinical Medicine, University of Tsukuba, Department of Obstetrics and Gynecology, Tsukuba, 305–8575, Japan; E-mail:

This work was supported by a grant-in-aid (No. 12877045) from the Ministry of Education, Culture, Sports, Science and Technology of Japan.

Received January 30, 2002; Received in revised form April 9, 2002; Accepted April 25, 2002.

Bacterial vaginosis is the most common lower genital tract infection among women of reproductive age. It is a clinical syndrome resulting from an altered normal vaginal microflora, characterized by the replacement of the predominant Lactobacillus species with anaerobic bacteria such as Gardnerella vaginalis, Mobiluncus species, Bacteroides fragilis, Prevotella species, Mycoplasma species, Ureaplasma urealyticum, Fusobacterium nucleatum, and Peptostreptococcus species.1 Women with bacterial vaginosis are at increased risk for adverse pregnancy outcomes and genital infections, such as chorioamnionitis, premature rupture of membranes, premature delivery,2,3 miscarriage in the first trimester after in vitro fertilization, 4 and postoperative upper genital tract infections.5–7 Furthermore, bacterial vaginosis is associated with an increased risk for human immunodeficiency virus (HIV)-1 infection8 and urinary tract infections.9

Clinical diagnosis of bacterial vaginosis is made when three of the following Amsel criteria are present: 1) an adherent grayish-white discharge, 2) an elevated vaginal pH (≥ 4.5), 3) a positive sniff/whiff test, and 4) the presence of clue cells.10 The laboratory methods available for diagnosing bacterial vaginosis include culture, direct Gram staining of vaginal secretions, Papanicolaou smears, oligonucleotide probe–based hybridization for G vaginalis, and biochemical tests for metabolic byproducts of vaginal bacteria.11 None of these methods are in common practice except Gram stain examination. Gram stain examination of vaginal smear is simple, having a sensitivity of 62% to 100%, and a positive predictive value of 76% to 100%.11 With Gram stain–based Nugent diagnostic criteria,12 vaginal flora can be categorized as normal (Lactobacillus predominant), intermediate (mixed flora), and bacterial vaginosis.

No standard management protocol for bacterial vaginosis has been established, but it is recommended that women at high risk for preterm labor be screened, and treatment should be given to those with symptoms of or positive test results for bacterial vaginosis.13 Because 50% of patients diagnosed with bacterial vaginosis show no symptoms,10 Gram stain–based diagnosis is used as the standard approach for the management of bacterial vaginosis. Patients in the intermediate and the normal categories according to Nugent criteria are generally not considered for treatment. Therefore, any laboratory test that can differentiate bacterial vaginosis from nonbacterial vaginosis should be adequate for management purposes. In this study, we describe a multiplex polymerase chain reaction (PCR)-based diagnostic method that differentiates bacterial vaginosis from nonbacterial vaginosis using multiple anaerobes that are implicated in bacterial vaginosis.

Back to Top | Article Outline


Reference bacterial strains, Mobiluncus mulieris (ATCC35243), Mobiluncus curtisii (ATCC35241), B fragilis (GAI10675), and G vaginalis (ATCC14018) were obtained from the American Type Culture Collection, Manassas, VA, and Department of Microbiology, Gifu University School of Medicine, Gifu, Japan. We used brain heart infusion (Difco Laboratories Inc., Detroit, MI) agar with 7% sheep blood to grow M mulieris, M curtisii and G vaginalis, and modified Gifu anaerobic medium agar (Nissui, Tokyo, Japan) for B fragilis. We investigated three different methods of template deoxyribonucleic acid (DNA) preparation for multiplex PCR: 1) complete lysis of bacteria by the boiling method, in which about 103 bacterial cells were suspended in 100 μL of distilled water and placed in a boiling water bath for 10 minutes followed by centrifugation at 20,000 g to separate the supernatant containing template DNA, 2) the treatment of the pellet with proteinase K (1 mg/mL) solution in addition to the above method, and 3) direct addition of 2.5 μL of bacterial suspension into a 25-μL PCR mixture.

Vaginal swabs were taken from Japanese patients (pregnant and nonpregnant women) who visited the Obstetrics and Gynecology Department in the Mito Saiseikai General Hospital, Ibaraki prefecture, Japan from May 1999 to August 2001. Women who were younger than 15 years or older than 76 years and who had had antimicrobial therapy within 2 weeks of the sampling were excluded from this study. Clinical samples were collected and processed after approval from the ethical committee of the hospital concerned and after informed consent was obtained from the patients. Vaginal swabs (one per patient) were taken by experienced interns and used immediately for microbial identification by Gram staining and aerobic cultures. Gram-stained slides were examined by a single trained person following the scoring system proposed by Nugent et al.12 The same vaginal swab that had been used for Gram staining and cultures was placed into 2 mL of sterile physiological saline and vortexed to suspend the bacteria attached to the swab. The cell suspension was then centrifuged at 20,000g for 5 minutes. The cell pellet was resuspended in 50 μL of distilled water, boiled for 10 minutes, and centrifuged as above to obtain the supernatant containing PCR template. About 2.5 μL of each sample was used in multiplex or simplex PCR as described below.

We designed three primer sets to amplify 16S ribosomal DNA from M mulieris and M curtisii, a gene encoding neuraminidase (nanH) from B fragilis, and an internal spacer region of ribosomal DNA from G vaginalis (Table 1). The multiplex PCR was designed to yield products that can be resolved on 1.8% agarose gel. The multiplex amplification mixture (25 μL) contained the following: deoxyribonucleotide triphosphates (200 μmol/L each), Mg2+ (2 mmol/L), 1 × PCR buffer, Ex Taq DNA polymerase (TaKaRa Shuzo Co., Ltd., Kyoto, Japan) (0.04 U/μL) and primers, Mob2A/Mob2B (300 nmol/L each), BFnA/BFnB (75 nmol/L each), and GarSA/GarSB (300 nmol/L each). A higher concentration of the same primers (400 nmol/L) was used in the simplex PCR. A thermocycler (PTC-100, MJ Research Inc., Waltham, MA) was programmed for the first 10 cycles of the touchdown routine, followed by an additional 30 routine amplification cycles. The touchdown routine comprised the preliminary denaturation for 1 minute at 94C, denaturation for 1 minute at 94C, ramping at 1.5C per second to the 60C–touchdown annealing temperature (which was 5C above the final annealing temperature for the first cycle, and then was decreased by 0.5C per cycle in subsequent cycles until the final annealing temperature was reached), annealing for 1 minute, and extension for 1 minute at 72C. The routine amplification was as follows: denaturation for 1 minute at 94C, ramping at 1.5C per second to the annealing temperature, annealing for 1 minute at 60C, and extension for 1 minute at 72C, with a final extension step of 7 minutes at 72C.

Table 1

Table 1

The specificity of the PCR assay was confirmed by sequencing the products derived from multiplex and simplex PCRs of the reference strains M mulieris (ATCC35243), M curtisii (ATCC35241), B fragilis (GAI10675), and G vaginalis (ATCC14018). The specificity of the PCR assay was further examined by using clinical isolates, such as Escherichia coli, Klebsiella oxytoca, Klebsiella pneumoniae, Proteus mirabilis, Proteus vulgalis, Pseudomonas aeruginosa, Acinetobacter calcoaceticus, Staphylococcus aureus, Staphylococcus haemolyticus, Staphylococcus epidermidis, Streptococcus agalactiae, Streptococcus pyogenes, Enterococcus faecalis, and Candida albicans. To determine the sensitivity of the amplification reactions, the colony-forming unit count of the step dilutions from the culture of each reference strain was compared with the result of simplex or multiplex PCR of the same sample. The number of colony-forming units per swab was calculated as the number of colony-forming units present in 2 mL of step-diluted culture.

Diagnostic sensitivity, specificity, positive predictive value, and negative predictive value were calculated as described by Galen.14

Back to Top | Article Outline


A total of 1000 vaginal swabs taken from pregnant and nonpregnant women were examined by Gram staining following Nugent criteria. Due to the presence of cell debris and artifacts, 147 samples were not appropriate for Gram stain interpretation. The prevalence of the bacterial vaginosis, intermediate, and normal categories among the Gram stain–interpretable samples was 21.6% (184/853), 26.0% (222/853), and 52.4% (447/853), respectively. From the Gram stain–interpretable samples, 172 including the bacterial vaginosis (n = 37), intermediate (n = 45), and normal (n = 90) categories, which had been randomly selected to represent the above-mentioned prevalence, were subjected to multiplex PCR assays. Demographic data for those 172 subjects, including age, gravidity, parity, and gestational weeks are shown in Table 2.

Table 2

Table 2

As shown in Figure 1, PCR with multiplex primers (Mob2A/Mob2B, BFnA/BFnB, and GarSA/GarSB) produced 1015, 842, and 570 base-pair fragments specific to the Mobiluncus species, B fragilis, and G vaginalis, respectively, whereas no amplified products were detected in other unrelated bacteria, such as E coli, K oxytoca, K pneumoniae, P mirabilis, P vulgalis, P aeruginosa, A calcoaceticus, S aureus, S haemolyticus, S epidermidis, S agalactiae, S pyogenes, E faecalis, and C albicans. The sensitivity of simplex and multiplex PCR assays was evaluated using the reference strains shown in Table 3. PCR was positive when 103 to 104 colony-forming units of any tested bacterial species were present in a vaginal swab (ie, equivalent to 2 mL of sample solution). An equal number of Mobiluncus species or B fragilis was required for both the multiplex and simplex PCRs to show a positive result. In the case of G vaginalis, however, two orders of fewer bacteria were needed for simplex PCR to be positive.



Table 3

Table 3

Whereas 21.6% (8/37), 8.1% (3/37) and 67.6% (25/37) of the samples in the bacterial vaginosis category were positive for the Mobiluncus species, B fragilis, and G vaginalis, respectively, a corresponding 0% (0/135), 0.7% (1/135), and 3.7% (5/135) were observed in the nonbacterial vaginosis category (Table 4). Alternatively, 78.4% (29/37) of the samples were found to be positive for one or more anaerobes by PCR in the bacterial vaginosis category, whereas only 4.4% (6/135) were positive in the nonbacterial vaginosis category (ie, a combination of 8.9% (4/45) in the intermediate category and 2.2% (2/90) in the normal category) (Table 4). We assigned bacterial vaginosis to 20.3% (35/172) of the samples that were positive for one or more species of targeted bacteria.

Table 4

Table 4

Taken together, the diagnostic sensitivity, specificity, and positive and negative predictive values of the multiplex PCR in comparison with Gram stain examination were 78.4% (95% confidence interval [CI] 65.1%, 91.6%), 95.6% (95% CI 92.1%, 99.0%), 82.9% (95% CI 70.4%, 95.4%), and 94.2% (95% CI 90.3%, 98.1%), respectively. About 11% (6/53) in the inappropriate group were PCR-positive for G vaginalis (data not shown).

Back to Top | Article Outline


We have developed a multiplex PCR–based diagnostic test for bacterial vaginosis involving the detection of multiple anaerobes in vaginal swab samples. The test design is based on the fact that M mulieris, M curtisii, B fragilis, and G vaginalis are the most common anaerobes associated with bacterial vaginosis, and that their predominance in the vaginal flora is indicative of bacterial vaginosis.

We tested different PCR template preparation methods in conjunction with the use of different primer sets under various PCR conditions. Among the template preparation methods we tested, the boiling method was found to be a simple method that was applicable to the laboratory-grown bacteria as well as to the clinical samples, with the exception of G vaginalis. The additional treatment of samples with proteinase K (1 mg/mL) improved the PCR sensitivity for the detection of G vaginalis. Because of the inherent variability that could arise in samples collected from patients of diverse clinical status, reproducibility and nonspecific amplification are the major concerns for multiplex PCR in this study. We found that the touchdown PCR method, which takes a little longer than the routine PCR protocols, produced consistent results with the clinical samples.

Whether the status of vaginal flora is normal or bacterial vaginosis depends on the ratio between lactobacilli and bacterial vaginosis–associated anaerobes. The multiplex PCR is most sensitive when more than one anaerobe is detectable in the vaginal swab. This fact was reflected in the samples with Nugent scores of 9 and 10, where the correlation of the PCR assay with the Nugent scores was 100% (Table 4). On the other hand, as the complexity of the vaginal flora decreased, the positivity of the PCR assay deteriorated in parallel. The sensitivity of PCR (103 to 104 colony-forming units per swab) in this study is considered optimal for detecting anaerobes in the clinical samples because a considerable number of Mobiluncus species, B fragilis, or G vaginalis can be present in the normal vaginal flora,15–17 and their presence can lead to overdiagnosis by PCR. Other unrelated organisms, which are generally present in the vagina, were not detected by PCR (Figure 1). Collectively, these results demonstrated the appropriateness of using these bacterial vaginosis–associated anaerobes in PCR-based diagnosis of bacterial vaginosis.

We found only three of the samples (8.1%) in the bacterial vaginosis category positive for B fragilis. The correlation of PCR with the Nugent criteria observed here therefore mainly reflects the presence of Mobiluncus species and G vaginalis. The overall correlation and thus the diagnostic sensitivity of the multiplex PCR method could be improved if some species belonging to the genus Prevotella, previously included in the genus Bacteroides, are included in the assay.

Gram stain examination of vaginal specimens, which is rarely used by practitioners, requires a trained microbiologist to be accurate and for the interpretation of results to be consistent. Some Gram-stained smears in this study appeared to contain an abundance of gram-negative rods, Bacteroides morphotypes, but their PCR results were negative for B fragilis. The cultures also revealed other gram-negative rods, such as E coli and K pneumoniae, indicating the subjective nature of the method and the possibility of false positives arising from inordinate interpretation. It was also noted that a considerable number of samples that were not interpretable in the Gram stain examination were found to be positive in PCR.

The detection by simplex PCR of bacterial vaginosis–associated anaerobes, such as M mulieris, M curtisii, and G vaginalis from the vaginal samples has been described previously,18,19 but these studies did not relate to the diagnosis of bacterial vaginosis. With its 78.4% diagnostic sensitivity and 95.6% specificity, the method described here is indicative of bacterial vaginosis when one or more anaerobes is positive in the vaginal sample. The high negative predictive value (94.2%) implies that this method is applicable for screening, as in prenatal management; however, considering that 18.1% (6/35) of the patients with positive results were overdiagnosed (for a positive predictive value of 81.9%), the use of our method would be most efficacious when used in combination with the clinical diagnostic criteria that were not applied in this study.

Although Gram staining remains an indispensable tool for the diagnosis of bacterial vaginosis, the method described here, in combination with clinical diagnosis, can be used as an alternative bacterial vaginosis screening test, especially for women who are simultaneously undergoing other screening tests (eg, Papanicolaou smear, PCR-based detection of Neisseria gonorrhoeae and Chlamydia trachomatis). Finally, the increasing use of PCR in clinical laboratories suggests that this PCR-based bacterial vaginosis diagnostic method could be considered for routine use.

Back to Top | Article Outline


1. Hill GB. The microbiology of bacterial vaginosis. Am J Obstet Gynecol 1993;169:450–4.
2. Gravett MG, Nelson HP, DeRouen T, Critchlow C, Eschenbach DA, Holmes KK. Independent associations of bacterial vaginosis and Chlamydia trachomatis infection with adverse pregnancy outcome. JAMA 1986;256:1899–903.
3. McGregor JA, French JI, Richter R, Franco-Buff A, Johnson A, Hillier S, et al. Antenatal microbiologic and maternal risk factors associated with prematurity. Am J Obstet Gynecol 1990;163:1465–73.
4. Ralph SG, Rutherford AJ, Wilson JD. Influence of bacterial vaginosis on conception and miscarriage in the first trimester: Cohort study. Bmj 1999;319:220–3.
5. Larsson PG, Platz-Christensen JJ, Thejls H, Forsum U, Pahlson C. Incidence of pelvic inflammatory disease after first-trimester legal abortion in women with bacterial vaginosis after treatment with metronidazole: A double-blind, randomized study. Am J Obstet Gynecol 1992;166:100–3.
6. Watts DH, Krohn MA, Hillier SL, Eschenbach DA. Bacterial vaginosis as a risk factor for post-cesarean endometritis. Obstet Gynecol 1990;75:52–8.
7. Soper DE, Bump RC, Hurt WG. Bacterial vaginosis and trichomoniasis vaginitis are risk factors for cuff cellulitis after abdominal hysterectomy. Am J Obstet Gynecol 1990; 163:1016–21; discussion 1021–3.
8. Martin HL, Richardson BA, Nyange PM, Lavreys L, Mandaliya K, Jackson DJ, et al. Vaginal lactobacilli, microbial flora, and risk of human immunodeficiency virus type 1 and sexually transmitted disease acquisition. J Infect Dis 1999;180:1863–8.
9. Harmanli OH, Cheng GY, Nyirjesy P, Chatwani A, Gaughan JP. Urinary tract infections in women with bacterial vaginosis. Obstet Gynecol 2000;95:710–2.
10. Amsel R, Totten PA, Spiegel CA, Chen KC, Eschenbach D, Holmes KK. Nonspecific vaginitis. Diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983;74:14–22.
11. Hillier SL. Diagnostic microbiology of bacterial vaginosis. Am J Obstet Gynecol 1993;169:455–9.
12. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of gram stain interpretation. J Clin Microbiol 1991;29:297–301.
13. American College of Obstetricians and Gynecologists. Bacterial vaginosis screening for prevention of preterm delivery. ACOG committee opinion no. 198. Washington, DC: American College of Obstetricians and Gynecologists, 1998.
14. Galen RS. Predictive value and efficiency of laboratory testing. Pediatr Clin North Am 1980;27:861–9.
15. Hillier SL, Krohn MA, Klebanoff SJ, Eschenbach DA. The relationship of hydrogen peroxide-producing lactobacilli to bacterial vaginosis and genital microflora in pregnant women. Obstet Gynecol 1992;79;369–73.
16. Hillier SL, Krohn MA, Rabe LK, Klebanoff SJ, Eschenbach DA. The normal vaginal flora, H2O2-producing lactobacilli, and bacterial vaginosis in pregnant women. Clin Infect Dis 1993;16(suppl 4):S273–81.
17. Delaney ML, Onderdonk AB. Nugent score related to vaginal culture in pregnant women. Obstet Gynecol 2001; 98:79–84.
18. van Belkum A, Koeken A, Vandamme P, van Esbroeck M, Goossens H, Koopmans J, et al. Development of a species-specific polymerase chain reaction assay for Gardnerella vaginalis. Mol Cell Probes 1995;9:167–74.
19. Schwebke JR, Lawing LF. Prevalence of Mobiluncus spp among women with and without bacterial vaginosis as detected by polymerase chain reaction. Sex Transm Dis 2001;28:195–9.
© 2002 by The American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.