JCM Accepts, published online ahead of print on 4 June 2014 J. Clin. Microbiol. doi:10.1128/JCM.00795-14 Copyright © 2014, American Society for Microbiology. All Rights Reserved.
1
Cervical and vaginal flora is highly concordant with respect to bacterial vaginosis-associated
2
organisms and commensal lactobacillus species in reproductive-aged women
3 4
William L. Smith B.S.a, Spencer R. Hedges Ph.D. a, Eli Mordechai Ph.D.
a,b
5
Ph.D. a,b Jason P. Trama Ph.D. b, Scott E. Gygax Ph.D. a, Andrew M. Kaunitz M.D. c and David W.
6
Hilbert Ph.D. a#
, Martin E. Adelson
7 8
a
9
Diagnostic Laboratories, A Member of Genesis Biotechnology Group, 2439 Kuser Rd., Hamilton,
Femeris Women’s Health Research Center and
10
New Jersey, USA
11
c
12
Jacksonville, Florida, USA
Running head: Cervical and vaginal microbial flora
15 16
Oncoveda Cancer Research Center, Medical
Department of Obstetrics and Gynecology, University of Florida College of Medicine,
13 14
b
#Address correspondence to:
[email protected] 17 18 19 20 21 22 1
23
ABSTRACT
24
Matched vaginal and cervical specimens from 96 subjects were analyzed by quantitative polymerase
25
chain reaction for the presence and concentration of bacterial vaginosis-associated microbes and
26
commensal Lactobacillus spp. Detection of these microbes was 92% concordant, indicating that
27
microbial flora at these body sites is generally similar.
28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 2
45
The vaginal microbiome plays an important role in female reproductive tract health. The vaginal
46
microbiome of healthy women generally falls into five categories, four of which are dominated by a
47
single lactobacillus species (L. crispatus, L. gasseri, L. iners or L. jensenii), and a fifth that is
48
characterized by diverse anaerobic and facultative species (1). This final group has traditionally been
49
associated with bacterial vaginosis (BV), a disease characterized by vaginal discharge and odor (2),
50
as well as Human Immunodeficiency Virus (HIV) transmission (3), Trichomonas vaginalis infection
51
(4) and pre-term labor (5). Although less well-studied, the cervical microbiome may arguably be of
52
greater relevance, as the cervix is the site of infection by numerous pathogens such as HIV, human
53
papilloma virus (HPV), Chlamydia trachomatis and Neisseria gonorrhea. A recent study found that
54
a lactobacillus-dominated cervical microbiome is inversely associated with the prevalence of HIV,
55
herpes simplex virus type 2 and high-risk human papilloma virus (6).
56
cervico-vaginal flora have variously sampled the cervix, vagina or both sties simultaneously.
57
Therefore, it is of general interest to comparatively analyze the flora at these sites to determine if
58
results generated from sampling the vagina can be generalized to the cervix and vice-versa. For
59
example, the cervcal transformation zone is enriched in T-cells and antigen presenting cells
60
compared to the vagina (7), and thus differential immune at these sites could impact the composition
61
of their respective microbiomes. To our knowledge only two prior studies have addressed this issue
62
(8, 9), and although informative, these studies analyzed small subject cohorts and utilized semi-
63
quantitative methodology. Our goal in this study was to quantitatively analyze the cervical and
64
vaginal flora with respect to lactobacillus species as well as microbes associated with abnormal
65
flora, such as BV, from a substantial subject cohort.
66 3
Studies evaluating the
67
We performed a cross-sectional study from a prospective cohort, enrolling subjects who were
68
receiving ambulatory gynecologic care at either the Care Center for Women at University of Florida
69
Health (a teaching hospital with related ambulatory clinics) or the University of Florida Southside
70
Women’s Health Specialists (a freestanding ambulatory care center). The study took place from May
71
2006 to June 2009. Written informed consent was obtained from all subjects and the study was
72
carried out with the approval of Western IRB. The purpose of the study was to evaluate the
73
microbiology and immunology of BV. Therefore, subjects were recruited who were either
74
complaining of vaginitis symptoms (e.g. itching, discharge or odor) or were attending for well-visits
75
or routine care. Subject information was collected by completing a questionnaire. Exclusion criteria
76
included pregnancy and menopausal or post-menopausal status, as these conditions are known to
77
influence the composition of the vaginal flora (10, 11). In addition, patients were excluded if they
78
were positive for Chlamydia trachomatis, Neisseria gonorrhea or Trichomonas vaginalis as these
79
infections can provoke cervical and/or vaginal inflammation which could confound immunological
80
analysis of BV. In this study we analyzed separate vaginal and cervical specimens from 96 subjects.
81
Subject ages ranged from 18-54 years, with an average of 31.4 years. Race was self-reported; 44
82
subjects (46%) were Caucasian, 43 (45%) were African American, 9 (9.4%) were Hispanic, 1 was
83
Asian (1.0%) and 1 was Other (1.0%). Patients were diagnosed for BV using the following signs and
84
symptoms (i) vaginal discharge, (ii) odor, (iii) elevated vaginal pH and (iv) the presence of clue cells
85
(exfoliated vaginal epithelial cells with attached bacteria) (i.e. Amsel’s Criteria) (12). Twenty-four
86
subjects (25%) were diagnosed as having BV (Amsel’s Criteria=3-4), 21 subjects (22%) neither had
87
BV nor were completely healthy (Amsel’s Criteria=1-2) and 51 subjects (53%) were healthy
88
(Amsel’s Criteria=0). Specimens were collected with OneSwab collection devices (Medical 4
89
Diagnostic Laboratories). After placing a vaginal speculum and visualizing the cervix, the os was
90
swabbed. Immediately thereafter, a second swab was taken from the left or right mid-vaginal
91
sidewall. These samples were stored at 4C overnight and shipped to Medical Diagnostic
92
Laboratories where they were stored at -20C until analyzed. DNA extracted from OneSwab
93
transport medium by the X-tractor Gene automated nucleic acid extraction system (Corbett Robotics,
94
San Francisco, CA) according to the manufacturer’s instructions. A panel of quantitative real time-
95
PCR assays was used to quantify DNA from BV-associated microbes and commensal lactobacillus
96
species (Supplemental Table 1). The targets of these reactions are as follows: A. vaginae (tuf
97
encoding Elongation Factor Tu), BVAB2 (16S rDNA) (13), G. vaginalis (hisC gene encoding
98
Histidinol-Phosphate Aminotransferase), Megasphaera spp. (16S rDNA) (13) and lactobacillus
99
species (tuf encoding Elongation Factor Tu) (14). All assays have a limit of detection of ≤ 10 copies
100
of target per reaction, do not cross-react with DNA from a diverse panel of microbes commonly
101
found in the female reproductive tract, and exhibit linear amplification from 101 to 108 target copies
102
per reaction (14) (data not shown). All reaction plates included 4 separate no template control wells.
103
The quantity of PCR target copies detected is expressed per reaction, each one using 0.5 l of DNA
104
as template in a 4l reaction. The final primer concentrations were 200-800 nM. All of the reactions
105
were run using the following conditions: 2 minutes at 50C, 2 minutes at 95C, followed by 40
106
cycles of 10 seconds at 95C (melting) and 45 seconds at 60C (annealing and extension). The
107
Megasphaera spp. reaction is run as a duplex PCR using both Type 1 and Type 2 probes, and the
108
Lactobacillus spp. PCR is run as a multiplex, using probes for L. crispatus, L. gasseri, L. iners and
109
L. jensenii. All reactions were performed using a CFX384 Real-Time Thermocycler (Bio-Rad). For
5
110
each reaction, a standard curve was generated using reactions with 102, 104 or 106 copies of a
111
plasmid encoding the amplification target. The distribution of PCR target copy concentration in
112
patient specimens was non-normal (D’Agonstino and Pearson omnibus normality test). Therefore, to
113
detect correlations we determined Spearman’s rank correlation coefficient. Only comparisons that
114
remained significant after controlling for multiple comparisons using the Bonferroni correction are
115
reported. Statistical analysis was performed using Prism 5.02 (GraphPad). For significance testing,
116
an alpha of 0.05 was used.
117 118
We compared cervical and vaginal specimens from the same patients to determine concordance for
119
detection for these microbes (Table 1). We found high concordance rates: 95% for A. vaginae,
120
Megasphaera spp and L. jensenii; 94% for BVAB2 and L. crispatus; 93% for L. gasseri; 85% for G.
121
vaginalis and 81% for L. iners. In total, 705 out of 768 test results were concordant between
122
specimens for each patient, for an overall concordance rate of 92%. Of these concordant results, 188
123
were matched positive specimens and 517 were matched negative specimens. Of the 63 instances of
124
discordant test results, 29 were cervical-positive and vaginal-negative and 34 were the converse. We
125
next analyzed swab pairs where at least one was positive for the quantity of PCR target in each
126
sample (Fig. 1). We found that concentrations at the two sites were correlated for G. vaginalis
127
(P=0.006, =0.41) and BVAB2 (P < 0.001, =0.67). No significant correlations were observed for
128
the other microbes detected.. In addition, there were no significant differences in the concentration
129
of genomic copies of individual microbes when cervical and vaginal specimens were compared (data
130
not shown).
131 6
132
The frequent co-colonization of the cervix and vagina with the same microbes is not surprising in
133
light of the essentially continuous nature of the lower female reproductive tract. Interestingly, only
134
two of the eight microbes assayed—G. vaginalis and BVAB2—were correlated with respect to
135
quantity at the two sites.. Studies of G. vaginalis have found that virulent strains are superior to
136
commensal strains in their ability to adhere to both vaginal (15) and cervical (16) epithelium, which
137
may contribute to the correlated concentrations we observed at these sites. In addition, G. vaginalis
138
has a highly diverse population structure (17) and co-colonization with multiple strains is common
139
(18); possibly we are observing one strain predominating at the cervix and the other the vagina.
140
BVAB2 and vaginal Megasphaera spp. have not yet been cultured in the laboratory so their
141
adherence properties are unknown; although A. vaginae has been cultured the relevant studies are
142
also lacking. With respect to lactobacillus, a study of human vaginal isolates found that they varied
143
greatly in their ability to adhere to vaginal epithelial cells (19). In addition, an adhesin has been
144
identified in L. crispatus that specifically mediates adherence to stratified squamous epithelium
145
(including vaginal epithelium) but not to columnar epithelium (which constitutes the cervical
146
epitihelium) and its expression is strain-specific (20). Therefore, we speculate that some strains of
147
lactobacillus may be differentially able to colonize the two sites, resulting in the poor correlation we
148
observed.
149 150
One major possible confounding factor in this study is contamination. One trivial explanation for
151
similarity between cervical and vaginal flora is that cross-contamination occurred during sampling.
152
Based upon physical anatomy it is highly unlikely that a vaginal specimen would be contaminated
153
with cervical flora; in contrast, it is more plausible than a cervical specimen could come in contact 7
154
with the vaginal wall after sampling, resulting in contamination with vaginal flora. There are several
155
observations that lead us to reject cross-contamination as the most likely explanation for our
156
observations. First, of the 63 instances of discordance, 29 of them (46%) were positive cervical
157
specimens and negative vaginal specimens, which cannot be explained by contamination of cervical
158
specimens with vaginal flora. Second, we chose to analyze specimens which were concordant for
159
multiple organisms and compared the concentration, as we would expect to observe a consistently
160
higher concentration in vaginal specimens if systematic contamination occurred. However, among
161
52 such subjects, we found that for 24 of them (46%) at least one microbe was present at a higher
162
concentration in the cervical specimen than in the vaginal specimen. Again, this pattern argues
163
against cross-contamination as an explanation for the cervical-vaginal concordance. However, future
164
studies could consider the utilization of sham cervical samples as a control for cross-contamination.
165 166
With respect to prior studies examining cervical and vaginal flora, Kim et al. analyzed a cohort of
167
eight subjects, sampled from the cervix and several vaginal sites using lavage, swabbing and
168
scraping, and PCR amplified 16S rDNA, generated amplicon libraries and sequenced them to
169
evaluate microbial community composition (8). Their results were broadly similar to ours, in that the
170
same microbes, such as Lactobacillus spp., Pseudomonas spp. and G. vaginalis, were detected at the
171
cervix and vagina in the same subjects but their concentration varied greatly. The methodology used
172
in this study is advantageous in that it is unbiased in detection; however, it is only semi-quantiative
173
and failed to resolve the lactobacillus detected to the species level. Another study, using DNA
174
hybridization, analyzed cervical and vaginal specimens from 12 patients and also found that
175
microbes such as G. vaginalis, A. vaginae and L. iners were present in both specimens from the 8
176
same paients (9). Again, quantitative PCR is a superior method to DNA hybridization with respect
177
to sensitivity and quantitation. In summary, utilizing quantitative PCR methods to analyze a series of
178
BV-associated and commensal lactobacillus species we found a high overall degree of concordance
179
with respect to colonization, although the quantity of a particular microbe detected at both sites
180
varied substantially. With respect to diagnostic testing for the microbes in question, cervical and
181
vaginal specimens are generalizable with respect to qualitative detection; however, quantitative
182
methodology may require independent validation for cervical and vaginal specimens.
183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 9
198
References
199
1.
Ravel J, Gajer P, Abdo Z, Schneider GM, Koenig SS, McCulle SL, Karlebach S, Gorle
200
R, Russell J, Tacket CO, Brotman RM, Davis CC, Ault K, Peralta L, Forney LJ. 2011.
201
Vaginal microbiome of reproductive-age women. Proc Natl Acad Sci U S A 108 Suppl
202
1:4680-4687.
203
2.
Sobel JD. 1997. Vaginitis. N Engl J Med 337:1896-1903.
204
3.
Mirmonsef P, Krass L, Landay A, Spear GT. 2012. The role of bacterial vaginosis and
205 206
trichomonas in HIV transmission across the female genital tract. Curr HIV Res 10:202-210. 4.
Martin DH, Zozaya M, Lillis RA, Myers L, Nsuami MJ, Ferris MJ. 2013. Unique
207
vaginal microbiota that includes an unknown Mycoplasma-like organism is associated with
208
Trichomonas vaginalis infection. J Infect Dis 207:1922-1931.
209
5.
Hillier SL, Nugent RP, Eschenbach DA, Krohn MA, Gibbs RS, Martin DH, Cotch MF,
210
Edelman R, Pastorek JG, 2nd, Rao AV, McNellis D, Regan JA, Carey JC, Klebanoff
211
MA. 1995. Association between bacterial vaginosis and preterm delivery of a low-birth-
212
weight infant. The Vaginal Infections and Prematurity Study Group. N Engl J Med
213
333:1737-1742.
214
6.
Borgdorff H, Tsivtsivadze E, Verhelst R, Marzorati M, Jurriaans S, Ndayisaba GF,
215
Schuren FH, van de Wijgert JH. 2014. Lactobacillus-dominated cervicovaginal microbiota
216
associated with reduced HIV/STI prevalence and genital HIV viral load in African women.
217
The ISME journal.
10
218
7.
Pudney J, Quayle AJ, Anderson DJ. 2005. Immunological microenvironments in the
219
human vagina and cervix: mediators of cellular immunity are concentrated in the cervical
220
transformation zone. Biol Reprod 73:1253-1263.
221
8.
Kim TK, Thomas SM, Ho M, Sharma S, Reich CI, Frank JA, Yeater KM, Biggs DR,
222
Nakamura N, Stumpf R, Leigh SR, Tapping RI, Blanke SR, Slauch JM, Gaskins HR,
223
Weisbaum JS, Olsen GJ, Hoyer LL, Wilson BA. 2009. Heterogeneity of vaginal microbial
224
communities within individuals. J Clin Microbiol 47:1181-1189.
225
9.
Nikolaitchouk N, Andersch B, Falsen E, Strombeck L, Mattsby-Baltzer I. 2008. The
226
lower genital tract microbiota in relation to cytokine-, SLPI- and endotoxin levels:
227
application of checkerboard DNA-DNA hybridization (CDH). APMIS 116:263-277.
228
10.
Romero R, Hassan SS, Gajer P, Tarca AL, Fadrosh DW, Nikita L, Galuppi M, Lamont
229
RF, Chaemsaithong P, Miranda J, Chaiworapongsa T, Ravel J. 2014. The composition
230
and stability of the vaginal microbiota of normal pregnant women is different from that of
231
non-pregnant women. Microbiome 2:4.
232
11.
Brotman RM, Shardell MD, Gajer P, Fadrosh D, Chang K, Silver MI, Viscidi RP,
233
Burke AE, Ravel J, Gravitt PE. 2014. Association between the vaginal microbiota,
234
menopause status, and signs of vulvovaginal atrophy. Menopause 21:450-458.
235
12.
Amsel R, Totten PA, Spiegel CA, Chen KC, Eschenbach D, Holmes KK. 1983.
236
Nonspecific vaginitis. Diagnostic criteria and microbial and epidemiologic associations. Am
237
J Med 74:14-22.
11
238
13.
Fredricks DN, Fiedler TL, Thomas KK, Oakley BB, Marrazzo JM. 2007. Targeted PCR
239
for detection of vaginal bacteria associated with bacterial vaginosis. J Clin Microbiol
240
45:3270-3276.
241
14.
Balashov SV, Mordechai E, Adelson ME, Sobel JD, Gygax SE. 2014. Multiplex
242
quantitative polymerase chain reaction assay for the identification and quantitation of major
243
vaginal lactobacilli. Diagn Microbiol Infect Dis 78:321-327.
244
15.
Jr HM, JM A, GA B, JL P, AT O, PH G, KK J. 2010. Drawing the line between
245
commensal and pathogenic Gardnerella vaginalis through genome analysis and virulence
246
studies. BMC Genomics 11:375.
247
16.
Castro J, Henriques A, Machado A, Henriques M, Jefferson K, Cerca N. 2013.
248
Reciprocal interference between Lactobacillus spp. and Gardnerella vaginalis on initial
249
adherence to epithelial cells. Int J Med Sci 10:1193-1198.
250
17.
Ahmed A, Earl J, Retchless A, Hillier SL, Rabe LK, Cherpes TL, Powell E, Janto B,
251
Eutsey R, Hiller NL, Boissy R, Dahlgren ME, Hall BG, Costerton JW, Post JC, Hu FZ,
252
Ehrlich GD. 2012. Comparative genomic analyses of 17 clinical isolates of Gardnerella
253
vaginalis provide evidence of multiple genetically isolated clades consistent with
254
subspeciation into genovars. J Bacteriol 194:3922-3937.
255
18.
Balashov SV, Mordechai E, Adelson ME, Gygax SE. 2014. Identification, quantification
256
and subtyping of Gardnerella vaginalis in noncultured clinical vaginal samples by
257
quantitative PCR. J Med Microbiol 63:162-175.
12
258
19.
Andreu A, Stapleton AE, Fennell CL, Hillier SL, Stamm WE. 1995. Hemagglutination,
259
adherence, and surface properties of vaginal Lactobacillus species. J Infect Dis 171:1237-
260
1243.
261
20.
Edelman SM, Lehti TA, Kainulainen V, Antikainen J, Kylvaja R, Baumann M,
262
Westerlund-Wikstrom B, Korhonen TK. 2012. Identification of a high-molecular-mass
263
Lactobacillus epithelium adhesin (LEA) of Lactobacillus crispatus ST1 that binds to
264
stratified squamous epithelium. Microbiology 158:1713-1722.
265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 13
280
Figure Legends
281
Figure 1. Quantity of PCR target copies measured in matched cervical and vaginal specimens. P
282
values are displayed for significant correlations. Due to the log scale, discordant samples (where one
283
sample as a value of zero) are not displayed.
14
Figure 1
G. vaginalis
BVAB2 5
Vaginal (log g copies)
Vaginal (log copies)
5 4 3 2 1
P=0.006 1
2
3
4
5
4 3 2 1
Cervical (log copies)
P