- Email: [email protected]
Etiology of cervical inflammation
Etiology of cervical inflammation Jonna Paavonen, M.D., Cathy W. Critchlow, M.S., Timothy DeRouen, Ph.D., Claire E. Stevens, M.A., P.A., Nancy Kiviat, M.D., Robert C. Brunham, M.D., Walter E. Stamm, M.D., Cho-Chou Kuo, M.D., Ph.D., Kathryn E. Hyde, B.A., Lawrence Corey, M.D., David A. Eschenbach, M.D., and King K. Holmes, M.D., Ph.D. Seattle, Washington We studied the relationships of selected microbial, clinical, demographic, and behavioral variables to mucopurulent cervicitis in two clinical settings, a sexually transmitted disease clinic and a student health clinic. From each clinic, we studied a group of women referred for suspected mucopurulent cervicitis and a representative sample of other women attending the clinic. After the women were stratified by patient group and summary odds ratios for all groups were obtained, mucopurulent cervicitis was most strongly associated with the isolation of Chlamydia trachomatis; other variables associated with mucopurulent cervicitis included the isolation of Ureap/asma urea/yticum, Gardnerella vagina/is, and Trichomonas vaginalis, the presence of serum antibody to C. trachomatis, the clinical diagnosis of bacterial vaginosis, and oral contraceptive use (positive associations) or isolation of yeast (negative association). After adjustment for cervical culture results for C. trachomatis, mucopurulent cervicitis was positively associated with oral contraceptive use (p = 0.02) and isolation of U. urea/yticum (p = 0.02) and negatively associated with isolation of yeast (p = 0.03). Among women with a positive cervical culture for C. trachoma tis, isolation of U. urea/yticum was significantly associated with mucopurulent cervicitis, while among the subgroup of women with a negative cervical culture for C. trachomatis and positive serum antibody to C. trachomatis, oral contraceptive use was strongly associated with mucopurulent cervicitis. These results confirm that in both clinical settings C. trachomatis is the major cause of mucopurulent cervicitis. The roles of U. urea/yticum, T. vagina/is, G. vagina/is, bacterial vaginosis, and oral contraceptive use in the etiology of mucopurulent cervicitis deserve further study. (AM J OBSTET GVNECOL 1986; 154:55664.)
Key words: Cervicitis, cervical inflammation, chlamydia, ureaplasma, bacterial vaginosis
Infection of the uterine cervix is often asymptomatic and represents a reservoir for sexual and perinatal transmission of pathogenic microorganisms. It might lead to at least three possible types of complications: (1) ascending intraluminal spread of pathogenic organisms from the cervix, producing endometritis and salpingitis; (2) ascending infection during pregnancy, resulting in chorioamnionitis, premature ru pture of the membranes, amniotic fluid infection, premature delivery, and puerperal and neonatal infections; and (3) the initiation or promotion of cervical neoplasia. In a previous study of randomly selected women attending a clinic for sexually transmitted diseases, we assessed objective criteria for the clinical diagnosis of mucopurulent cervicitis, that is, visualization of yellow
From the Departments of Medicine, Biostatistics, Obstetrics and Gynecology, and Pathobiology, University of Washington, School of Medicine, The Hall Health Center, University of Washington, and the Harborview Medical Center. Supported in part by National Institutes of Health Grants AI-12192 and AI-14180. Received for publication July 1, 1985; revised November 13, 1985; accepted November 25, 1985. Reprint requests: King K. Holmes, M.D., Ph.D., Department of Medicine ZA-92, HarborviewMedical Center, 325 Ninth Ave., Seattle, WA 98104.
556
mucopurulent endocervical discharge on a white swab or the presence of 10 or more polymorphonuclear leukocytes per microscopic field (at a magnification of 1000) in satisfactory Gram-stained endocervical smears. 1 These criteria are quite similar to those commonly used to make a presumptive diagnosis of nongonococcal urethritis in men." The major recognized infectious causes of cervicitis are thought to be the sexually transmitted pathogens Chlamydia trachomatis, Neisseria gonorrhoeae, and herpes simplex virus. However, previous studies of the etiology of cervicitis have focused on selected groups of patients seen in sexually transmitted disease clinics (contacts of men with nongonococcal urethritis) and have examined the role of only a limited number of potential pathogens. The present study of mucopurulent cervicitis was undertaken in two clinical settings to define the microbiologic etiology of cervical inflammation in consecutively referred women with suspected cervicitis and in representative samples of women from two clinical settings.
Material and methods Study population. Women attending the SeattleKing County sexually transmitted disease clinic at Har-
Volume 154 Number 3
borview Medical Center or the University of Washington student health center in Seattle were eligible for study. Four patient groups, two from each clinic, composed the study population: two groups of consecutive patients referred for suspected cervicitis (one at each clinic), one randomly selected group from women attending the sexually transmitted disease clinic for the first time or with a new problem, and a group of consecutive women registering for routine annual examinations at the student health clinic. Women referred for suspected cervicitis. Clinicians at the sexually transmitted disease clinic and student health clinic were asked to refer women? 16 years of age whom they believed to have cervicitis but in whom they did not suspect pelvic inflammatory disease. Women who had received antibiotics in the previous month, who were menstruating, who were pregnant, or who had an intrauterine contraceptive device were excluded. One hundred sixty-one women in the sexually transmitted disease clinic and 47 women in the student health clinic were thus referred and enrolled for suspected cervicitis. Representative samples of women in two clinical settings. A random sample of 110 women attending the sexually transmitted disease clinic were selected as previously described.' At the student health clinic, 49 consecutive women registering for routine annual examinations were asked to participate in the study and all agreed. Patient evaluation. A standardized history with detailed information concerning menstrual history, contraceptive method, sexual behavior, and prior history of sexually transmitted disease was obtained from all women. All women underwent a gynecologic examination by one of the investigators. The color of cervical secretions as viewed on a white cotton swab was classified as clear, white, cloudy, or yellow; yellow secretions were defined as mucopus. The severity of inflammation of the endocervical epithelium was assessed by separately scoring the degree of induced mucosal bleeding as well as erythema and edema of the zone of ectopy when observed with the unaided eye as 0 (none or normal), I + (mild), 2 + (moderate), or 3 + (severe) and summing the score. Bacterial vaginosis was diagnosed according to the criteria developed by Amsel et aI.' Definition of mucopurulent cervicitis. As previously described,' a diagnosis of mucopurulent cervicitis was established on the basis of the presence of either a yellow mucopurulent endocervical discharge or ? 10 polymorphonuclear leukocytes per 1000 X microscopic field in cervical mucus on a satisfactory Gram-stained smear. Overall, 28 smears were not evaluable because the slide contained> 100 squamous cells in association with inflammatory cells in cervical mucus. Quantification of leukocytes in endocervical secre· tions. After the ectocervix was wiped clean with a large
Etiology of cervical inflammation
557
cotton swab, endocervical mucus was collected on a white-tipped swab with care taken to avoid contamination by vaginal secretions. The swab was rolled onto a 2 cm' area of a glass microscope slide. The smear was air-dried and Gram-stained by the rapid method. Slides were then examined for the number of polymorphonuclear leukocytes per microscopic field at 1000 x magnification in five nonadjacent fields as previously described in detail. 1 Laboratory methods and collection of specimens. An unlubricated speculum was inserted in the vagina and the cervix was exposed. The lateral vaginal fornix was swabbed with a cotton-tipped applicator and vaginal pH was measured directly from this swab. A sterile paper strip (5 by 20 mm) was inserted into the endocervical canal and allowed to saturate with fluid from the cervical mucus. This was carefully done to avoid any bleeding from the endocervix during the procedure. The filter paper strip was placed in 0.2 ml of phosphate-buffered saline solution and frozen at - 20° C until assayed for antibody to C. trachoma tis. Calcium alginate swabs (Inolex, Glenwood, Illinois) were used to obtain urethral and cervical cultures for C. trachomatis. These specimens were separately placed in 2 ml of sucrose-phosphate-glutamate transport medium. Subsequent successive endocervical specimens were obtained with cotton-tipped swabs for cultures for N. gonorrhoeae, Mycoplasma hominil, Ureaplasma llrealyticum, Gardnerella vaginalis, Streptococcus agalactiae, and
herpes simplex virus. Cervicovaginal specimens were obtained for culture of Trichomonas vagina lis and yeast. Microbiologic studies. Isolation of C. trachomatis was performed in two cooperating laboratories with the use of either cycloheximide-treated McCoy cells in microtiter plates or diethylaminoethyl-dextran-treated HeLa 229 cells in tubes.' N. gonorrhoeae was isolated by direct inoculation of the swabs on Thayer-Martin agar with incubation and identification by standard methods. Specimens were inoculated onto A 7 agar medium and into broth media for isolation of M. hominis and U. llrealyticum. ' Broth showing evidence of growth was subcultured to A 7 agar for confirmation; colonies were identified by typical morphologic features. G. vaginalis was isolated on human blood Tween agar medium. Colonies showing l3-hemolysis were further identified as G. vaginalis as previously described.' Group B streptococci were isolated in Todd-Hewitt broth containing 5% sheep blood and 5 mg/L of gentamicin and 15 mg/L of nalidixic acid; broth was subcultured to sheep blood agar for confirmation and identification of group B streptococci on the basis of I)-hemolytic reaction and latex agglutination with group-specific antiserum (Streptex, Wellcome Diagnostics, Research Triangle Park, North Carolina). Cervical specimens were placed in viral transport medium and inoculated into a monolayer of fibroblasts from human foreskin for
558
Paavonen et al.
March, 1986
Am J Obstet Gyneco1
Table I. Associations of demographic and behavioral characteristics with mucopurulent cervicitis stratified by patient group STD clinic women Referred MPC Neg. (n = 9) No.
Age (yr) Race white Marital status single Education (yr) Gravidity;:" I Parity;:" I Method of contraception Oral contraceptives Spermicides Other None Age at first intercourse (yr) Median lifetime sex partners (No.) Past history of STDt
I
o/c
24.7 ± 4.1 7 78 6 67 13.8 ± 2.1 5 56 I II 2 22 3 33 I II 3 33 16.6 ± 1.4 20 5/8 63
Sampled MPC Pos. (n = /33)
No.
I
MPC Neg. (n = 51)
%
23.4 ± 5.2 1051132 80 1031130 79 13.2 ± 2.5 60/132 46 41 31 80 16 6 28
No.
I
MPC Pos. (n = 39)
%
23.7 ± 5.7 65 35/47 75 12.6 ± 1.9 24/50 48 12 24
33
62 12 5 22
9 13 3 26
17.2 ± 2.5 8 901132 68
7 32
18 25* 6 51 16.3 ± 2.5 63
No.
I
%
22.3 ± 4.7
72 76
28 29/38
12.5 ± 2.0 7
58 18
16 4 7 II
42 II 18 29
22/38
16.1 ± 2.1 10 61 23/38
STD = Sexually transmitted disease; MPC = mucopurulent cervicitis; Neg. = negative; Pos. = positive. *p oS 0.0 I. tGonorrhea, pelvic inflammatory disease, genital warts, genital herpes, or trichomoniasis. :j:p oS 0.05.
isolation of herpes simplex virus.' Cervicovaginal specimens were inoculated into Diamond's medium for isolation of T. vagina lis and on Sabaraud's agar for isolation of yeasts. Serum and secretion antibody determinations. Serum samples were tested by the microimmunoAuorescence technique for IgG antibodies for all C. trachomatis elementary body serotypes.' Cervical secretions were similarly tested for secretory IgA antibodies to C. trachomatis. Statistical methods. Initial comparisons of demographic, behavioral, and microbiologic characteristics of women with or without mucopurulent cervicitis were done in each of the four patient groups by standard X" or t-tests. In order to obtain an overall assessment of the strength of the associations between possible causative factors and mucopurulent cervicitis while minimizing the statistical problems of multiple comparisons, the associations were analyzed by first stratifying patients into the four categories (referred or not referred, sexually transmitted disease or student health clinic) and then combining these results by calculating an overall odds ratio according to the Mantel-Haenszel method.' This procedure assumes that the associations are the same in all strata, and this assumption was tested for each association by use of a X2 test for homogeneity, which is significant if there are large disparities among the odds ratios in the four strata. The p value reported with each odds ratio corresponds to the test of the hypothesis that the odds
ratio is equal to I (that is, no association). These analyses were conducted with the 1981 version of the BMPD statistical program P4F. Similar Mantel-Haenszel analyses were also conducted separately in groups of women with or without evidence of infection by C. trachomatis (as measured by positive cervical culture), and the results were combined into an overall statistic that adjusted for the presence of chlamydial infection. Since the purpose of this analysis was to compare the relative strengths of the associations of several variables with mucopurulent cervicitis, the p values reported for the summary odds ratios (see Tables III, IV, and V) have not been adjusted to account for the number of associations examined.
Results Incidence of mucopurulent cervicitis among referred and sampled study populations. Thirteen of the 367 women initially enrolled (10 from the sampled sexually transmitted disease clinic population and three from the referred sexually transmitted disease clinic patient group) were menstruating and thus excluded from further evaluation because cervical findings could not be adequately assessed. The diagnosis of mucopurulent cervicitis could not be made in an additional 31 women who did not have mucopus and in whom polymorphonuclear leukocytes could not be evaluated because either the Gram-stained smear was unsatisfactory or a smear was not obtained. Of the remaining
Etiology of cervical inflammation
Vulume 154 Kumber 3
559
Student health clinic women Referred
Sampled
II10.
I
%
No.
75 86
(I
0 E. (I
2
16.7 ± 2.0
No.
9 3
23 8
0 75 0 25
19 9 2 8
50 24 5 21
9 9 5 6
10:1:
4 11123
48
5 7
MPC Pas. (n = 15)
% 93 86 24 3 31 31 17 21
No.
323 patients, mucopurulent cervicitis was detected in 39 (43%) of the 90 randomly selected sexually transmitted disease clinic female patients, in 133 (94%) of the 142 sexually transmitted disease clinic patients referred for suspected cervicitis, in 15 (34%) of the 44 consecutive women seen for routine annual examination at the student health clinic, and in 39 (83%) of the 47 student health clinic patients referred for suspected cervicitis. Associations of demographic and behavioral characteristics with mucopurulent cervicitis in the four patient groups. As shown in Table I, there were differences in contraceptive usage patterns among women with and without mucopurulent cervicitis. Among sexually transmitted disease clinic women, oral contraceptives were used by 96 (57%) of 168 women with mucopurulent cervicitis versus only 11 (18%) of60 women without mucopurulent cervicitis (p < 0.001). A similar difference in oral contraceptive usage was seen among referred student health clinic women (19 of 38 versus zero of eight, p < 0.01) but not in the student health clinic sample. No significant differences were found in age, race, marital status, years of education, gravidity, parity, or sexual history between women with and without mucopurulent cervicitis in any of the study populations. In the referred student health clinic population, a greater proportion of women with mucopurulent cervicitis had a history of past sexually transmitted disease (11 or 48% of 23) compared with those without mucopurulent cervicitis (zero of seven, p < 0.05). Associations of specific microorganisms or clinical diagnoses with mucopurulent cervicitis. The prevalence of cervical and vaginal pathogens, clinical diag-
I
23.3 ± 3.8 131\4 131\4 15.0 ± 1.6 5 I
% 93 93 33 7 33 40 7 20
5 6 I
3 18.2 ± 1.9
18.6 ± 2.8
18.3 ± 3.0
19.8 ± 2.9
I
23.1 ± 4.5 26/28 25/29 15.6 ± 2.0 7
0 0
:"
017
% 85 100
29/34 21/21
15.9 ± 1.4 (I
I 23.4 ± 3.8
23.6 ± 2.0 6 617
MPC Neg. (n = 29)
MPC Pos. (n = 39)
MPC Neg. (n = 8)
24
4 7
47
noses, and chlamydial antibodies among those with and without mucopurulent cervicitis stratified by patient group is shown in Table II. In separate analyses done within each group, mucopurulent cervicitis was associated with several infectious agents or clinical conditions. In the sexually transmitted disease clinic groups, the presence of mucopurulent cervicitis was most highly associated with positive cultures for C. trachomatis, U. urealyticum, or T. vaginalis, with the presence of serum antibodies to C. trachorrwtis, and with the clinical diagnosis of bacterial vaginosis. In the student health clinic groups, mucopurulent cervicitis was associated with the isolation of C. trachomatis (p = 0.06), T. vaginalis, and group B streptococcus and with the clinical diagnosis of bacterial vaginosis; the isolation of yeast had a negative association with mucopurulent cervicitis. Although isolation of herpes simplex virus was not associated with mucopurulent cervicitis, characteristic ulcerative or necrotic lesions of the cervix were observed by colposcopy in eight (67%) of 12 women from whom herpes simplex virus was isolated versus 12 (4%) of 302 women from whom herpes simplex virus was not isolated (p < 0.00\). Association of selected variables with mucopurulent cervicitis by the Mantel-Haenszel test. Since may variables were associated with mucopurulent cervicitis in the separate group analyses, the comparisons of the strengths of these associations were made by determining Mantel-Haenszel summary odds ratios for each variable (Table III). Combining the odds ratios derived from all four groups was found justified by the nonsignificance of the test for homogeneity of the odds ratios from the patient groups (data not
560
Paavonen et al.
March, 1986 Am
J Obstet Gyneco1
Table II. Associations of cervical microorganisms, clinical diagnoses, and antichlamydial antibodies with mucopurulent cervicitis stratified by patient group STD clinic women Referred MPC Neg. (n = 9) No.
C. trachomatis Cervical culture Serum antibody Cervical secretory IgA antibody N. gonorrhoeae
Herpes simplex virus T. vaginalis M. hominis U. urea(vticum
Group B streptococcus G. vaginalis Yeast Bacterial vaginosis STD *p :s tp:s :/:p :s
I
I 4 0 017 0 0 3 4 2/8 5 118 3/8
MPC Pos. (n = 133)
%
No.
II 44t 0 0 0 0 33 44:/: 25 56 13 38
56 9111 18 2411 28 91132 71131 121132 551129 1081130 211130 95/126 1511 29 4411 21
I
MPC Neg. (n = 51)
%
No.
42 77 19 7 5 9 43 83 16 75 12 36
I 27/47 8/48 7/50 4/50 4/50 20 37 7 29/47 16/50 14/50
I
MPC Pos. (n = 39)
%
No.
2* 57:/: 17 14 8 8:/: 40 74t 14 62 32 28t
19 32/37 9/37 7 1138 11/38 20 35 7 29/38 6/38 19/38
I
% 49 86 24 18 3 29 51 90 18 76 16 50
= Sexually transmitted disease; MPC = mucopurulent cervicitis; Neg. = negative; Pos. = positive. 0.001. 0.05. 0.0 I.
Table III. Association of selected variables with mucopurulent cervicitis by the MantelHaenszel test
C. trachomatis Cervical culture Serum antibody Cervical secretory IgA antibody N. gonorrhoeae Herpes simplex virus T. vaginalis M. hominis U. urealyticum Group B streptococcus G. vaginalis Yeast Bacterial vaginosis Oral contraceptive use Spermicide use
*NS
Sampled
=
Summary odds ratio
p value
16.39 4.24 2.35 1.66 0.58 3. 17 1.32 2.70 1.14 2.39 0.36 I. 94 3.44 0.42
<0.001 <0.001 NS* NS NS 0.045 NS 0.004 NS 0.013 0.027 0.052 0.001 0.014
not significant; p > 0.10.
shown). As shown in Table III, several variables were significantly associated with mucopurulent cervicitis, including C. trachomatis (odds ratio = 16.39, P < 0.001), the presence of serum antibodies to C. trachomatis (odds ratio = 4.24, P < 0.001), oral contraceptive use (odds ratio = 3.44, P < 0.001), isolation ofT. vagi-naZis (odds ratio = 3.17, P = 0.04), isolation of U. urealyticum (odds ratio = 2.70, P < 0.01), isolation of C. vagi-naZis (odds ratio = 2.39. P = 0.(1),
bacterial vaginosis (odds ratio = 1.94, P = 0.(5), spermicide use (odds ratio = 0.42, P = 0.01, negative association), and the presence of yeast (odds ratio = 0.36, P = 0.03, negative association). Data concerning isolation of C. trachoma tis, N. gonorrhoeae, and herpes simplex virus from the randomly selected sexually transmitted disease clinic women were previously used in defining the objective criteria for diagnosis of mucopurulent cervicitis and its strong relationship to C. trachomatis. I If these data are excluded from analysis, isolation of C. trachomatis remains significantly associated with mucopurulent cervicitis (odds ratio = 7.97, P < 0.01), while isolation of N. gonorrhoeae and herpes simplex virus still demonstrates no such association. Summary odds ratios for mucopurulent cervicitis after adjusting for isolation of C. trachomatis. Since mucopurulent cervicitis was most strongly associated with positive C. trachomatis culture as measured by the largest odds ratio, we stratified the study population into those with and without cultural evidence of cervical C. trachomatis infection and repeated the analysis to clarify further the possible correlates of mucopurulent cervicitis. As shown in Table IV, oral contraceptive use (odds ratio = 2.55, P = 0.(2), isolation of U. ureaZyticum (odds ratio = 2.39, P = 0.(2), and isolation of yeast (odds ratio = 0.27, P = 0.03, negative association) were most strongly associated with mucopurulent cervicitis after adjustment for C. trachomatis. In addition, the presence of serum antibodies to C. tra-
Etiology of cervical inflammation
Volume 154 Number 3
561
Student health clinic women Referred
Sampled MPC Pos. (n = 39)
MPC Neg. (n = 8) No.
0 1/4
0 0 0 I ~!
:3 I 317 I :2
I
%
No.
0 25 0 0 0 13t 25 38 13t 43 8 25
12/38 20/38 3/38 2/38 0/37 0/38 10/36 23/37 0/38 22/37 2/37 7
I
MPC Neg. (n = 29)
%
No.
32 53 8 5 0 0 28 62 0 59 5 18
I 1115 1/28 0 0/26 0115 3125 11126 0116 5116 5116 3
chomatis (odds ratio = 2.20, P = 0.09), isolation of C. vaginalis (odds ratio = 1.95, P = 0.09), and bacterial vaginosis (odds ratio = 1.87, P = 0.09) demonstrated associations with mucopurulent cervicitis after adjustment for C. trachomatis, although these associations were not statistically significant. When the analysis was restricted to women with negative C. trachomatis cultures, odds ratios of 2.27, 2.24, 2.08, 2.03, and 1.95 were obtained for oral contraceptive use, the presence of serum antibodies to C. trachomatis, isolation of C. vaginalis, isolation of U. urealyticum, and the clinical diagnosis of bacterial vaginosis, respectively (0.05 < P < 0.1 for each), and a significant negative association was observed for the isolation of yeast (odds ratio = 0.24, P < 0.02). Among culturenegative women, oral contraceptive use was strongly associated with mucopurulent cervicitis in those who were seropositive (odds ratio = 6.61, P < 0.01) but not in those who were seronegative (odds ratio = 1.5). Among women with a postive C. trachomatis culture, only U. urealyticum (p < 0.01) was significantly associated with mucopurulent cervicitis. Although oral contraceptive use and serum antibody to C. trachomatis demonstrated the highest odds ratios for mucopurulent cervicitis among those with negative cultures for C. trachomatis, neither was associated with mucopurulent cervicitis among those with a positive C. trachomatis culture. Overall, 130 (82%) of 159 sexually transmitted disease clinic women with mucopurulent cervicitis and 31 (55%) of 56 sexually transmitted disease clinic women without mucopurulent cervicitis demonstrated cultural and/or serologic evidence of C. trachomatis infection. The corresponding figures among student health clinic
I
MPC Pos. (n = 15)
%
No.
3 7 4 0 0 0 12 42 0 31 31t lOt
2114 2111 1 0 0 0111 0114 7 2110 7/10 0/11
5
1
% 14 18 7 0 0 0 0 47 20 70 0 33
women were 25 (50%) of 50 and three (15%) of 20. Association of selected manifestations of cervical inflammation with C. trachomatis among women with mucopurulent cervicitis. Since mucopurulent cervicitis was associated with several variables after adjustment for C. trachomatis culture, we were interested in whether specific clinical manifestations of cervicitis differed for women with and without a positive C. trachomatis culture. As shown in Table V, among women with mucopurulent cervicitis, several findings were more common in those who had a positive cervical culture, including cervicitis severity score> 3 (odds ratio = 3.45, P < 0.001), induced mucosal bleeding (odds ratio = 2.37, P < 0.01), and edema of the area of ectopy (odds ratio = 2.05, P = 0.04).
Comment In our previous study of randomly selected sexually transmitted disease clinic women, we described objective criteria for the office diagnosis of mucopurulent cervicitis.! The presence of cervical mucopus (positive swab test) and ~ 10 polymorphonuclear leukocytes per 1000 X microscopic field in cervical mucus demonstrated the strongest associations with cervical infection by C. trachomatis, N. gonorrhoeae, or herpes simplex virus. In this extended study, we analyzed the associations of multiple pathogens, clinical diagnoses, and demographic and behavioral variables with mucopurulent cervicitis. The findings we obtained among the previously reported randomly selected sexually transmitted disease clinic women were examined to see if they extended to another group of sexually transmitted disease
562
Paavonen et al.
March, 1986
Am
J Obstet Gynecol
Table IV. Association of selected variables with mucopuprulent cervicitis by the Mantel-Haenszel test among all patients, adjusted for C. trachomatis, and among patients with a negative C. trachoma tis culture Among all patients, adjusted for isolation of C. trachomatis
Among patients with negative C. trachomatis culture
p Odds ratio*
I
Value
Odds ratio*
0.09 NSt NS NS NS NS 0.021 NS 0.09 0.026 0.09 0.021 NS
2.24 1.59 0.20 1.16 1.94 0.93 2.03 0.69 2.0S 0.24 1.95 2.27 0.57
I
P
Value
C. trachomatis
Serum antibody Cervical secretory IgA antibody N. gonorrhoeae Herpes simplex virus T. vaginalis M. hominis U. urealyticum Group B streptococcus C. vaginalis Yeast Bacterial vaginosis Oral contraceptive use Spermicide use
2.20 I.SS 0.37 I.IS 2.21 0.95 2.39 0.79 1.95 0.27 I.S7 2.55 0.53
0.09 NS NS NS NS NS 0.09 NS 0.07 O.OIS O.OS 0.06 NS
*Summary odds ratio. tNS = Not significant; p> 0.10.
clinic women (that is, those referred to us because of suspected cervicitis) and to groups of women seen in a student health clinic. We also attempted to analyze further the microbial and other correlates of mucopurulent cervicitis among women negative for C. trachomatis. In separate analyses in the four patient groups, multiple infectious agents (including C. trachomatis, U. urealyticum, and T. vaginalis), serum antibody to C. trachomatis, past history of sexually transmitted disease, oral contraceptive use, and bacterial vagi nos is were associated with mucopurulent cervicitis in one or more of the groups studied. On the other hand, isolation of yeast, usually Candida albicans, from the vagina had a negative association with mucopurulent cervicitis. To make overall comparisons among correlates of mucopurulent cervicitis, we next combined all four groups to obtain weighted average odds ratios for each variable. This summary analysis of the combined data demonstrated four major findings. First, C. trachoma tis detected by culture and serologic testing demonstrated the strongest association with mucopurulent cervicitis. Even after adjustment for C. trachoma tis culture and in particular among women with negative cultures for C. trachomatis, the presence of serum antibodies to C. trachomatis remained associated with mucopurulent cervicitis (odds ratio = 2.2, P = 0.09). This might suggest false negative cultures for C. trachomatis in many of the seropositive women or the persistence of abnormal cervical findings after chlamydial cervicitis among the culture-negative women who remain antibody-positive. The lack of association with secretery IgA antibodies to C. trachomatis, which may be a closer reflection of
current chlamydial infection, argues against the former explanation. Infections with C. trachomatis are characteristically chronic. Data from experimental infections suggest that the immune response to C. trachomatis may contribute to some of the manifestations of chronic infection." 7 The histopathologic features of chlamydial cervicitis are characterized by severe inflammatory infiltration of the stroma and prominent lymphoid follicles." The natural history of these histopathologic changes after eradication of C. trachomatis has not been extensively studied. We have observed that increased vascularity and erythema of the transformation zone in the cervix persists for prolonged periods of time after treatment (Paavonen J, unpublished results). Independent association of C. trachomatis with mucopurulent cervicitis on serologic testing suggests persistence of these histopathologic changes in the cervix. Second, both bacterial vaginosis and the isolation of C. vaginalis (a bacterial vagi nos is-associated pathogen) demonstrated associations with mucopurulent cervicitis, suggesting that changes in the cervicovaginal environment might foster the development of mucopurulent cervicitis or vice versa. The redox potential at the vaginal epithelial surface is greatly reduced in bacterial vaginosis." Oxygen consumption by polymorphonuclear leukocytes might decrease the redox potential and increase the pH of the vaginal environment to favor the growth of organisms associated with bacterial vaginosis. Mucopurulent endocervical discharge might similarly change the vaginal environment. Uniform clinical criteria for bacterial vaginosis and mucopurulent cervicitis should be used in future clinical studies
Etiology of cervical inflammation
Volume 154 Number 3
563
Table V. Association of cervical findings with results of C. trachomatis culture among sexually transmitted disease clinic women and student health clinic women with mucopurulent cervicitis C. trachomatis
Cervical finding
Cervicitis severity score 0-3 4-9 Ectopy Not present Present Erythema of ectopy:j: 0-1 2-3 Edema of ectopy:j: 0-1 2-3 Induced mucosal bleeding 0-1 2-3
Positive (n
= 88)
I
%
No.
23 63
27 73
10 78
Negative (n
= 133)
I
%
Odds ratio*
p Value
72 63
53 47
3.45
<0.001
II 89
21 114
16 84
1.58
NSt
22 56
28 72
42
37 63
1.49
NS
72
45 33
58 42
81 33
71 29
2.05
0.04
36 50
42
87 48
64 36
2.37
0.004
No.
58
*Summary odds ratio over patient groups. tNS = Not significant; p> 0.10. :j:Among women with ectopy.
of how the two conditions are interrelated. Does bacterial vaginosis resolve spontaneously after treatment of mucopurulent cervicitis, or should both conditions be treated concomitantly or successively? The negative association of yeast with mucopurulent cervicitis is also of interest, although it may simply be that our exclusion of women who had received antibiotics within the previous month did not exclude an effect of more remote antibiotic use, which could have promoted vulvovaginal candidiasis while eliminating mucopurulent cervicitis. The third major finding was the strong association of oral contraceptive use with mucopurulent cervicitis. Oral contraceptive-induced hyperplasia of the endocervical epithelium might render the cervix more susceptible to many sexually transmitted disease organisms. Most prior studies have demonstrated that oral contraceptive use is associated with an increased risk of C. trachomatis infection. 10 It is not clear, however, whether the higher risk of chlamydial infection in oral contraceptive users is secondary to differences in sexual behavior or to hormonal effects of oral contraceptives. Women using oral contraceptives have a higher incidence of cervical ectopy compared with that of women with natural cycles. 1O Cervical ectopy may render the cervix more susceptible to chlamydial infection, or detection of C. trachoma tis by culture might be easier among such women. Oral contraceptive use might also stimulate the growth of C. trachoma tis in cervical epithelial cells. However, among women with negative cervical cultures for C. trachomatis, oral contraceptive use was correlated with mucopurulent cervicitis only
among those with serum antibody to C. trachomatis (p < 0.01). This was true even after adjustment for variables reRecting sexual activity (such as age at first intercourse), suggesting that the relationship is not totally explained by increased sexual activity among women using oral contraceptives. Thus our results suggest that oral contraceptives increase the risk for mucopurulent cervicitis among women with current or past exposure to C. trachomatis. The fourth finding was that U. urealyticum was the only organism showing a significant positive association with mucopurulent cervicitis after adjustment for results of cervical cultures for C. trachomatis. This association is consistent with previous studies in which U. urealyticum has been associated, independent of C. trachomatis, with nongonococcal urethritis in men." Furthermore, several studies have associated U. urealyticum with puerperal infections,'2 low birth weight of infants,':l premature labor," subclinical endometritis,'" and infertility':l suggesting that U. urealyticum is a significant genital pathogen. In contrast, U. urealyticum has only occasionally been isolated from the fallopian tubes of women with salpingitis," and previous studies have not demonstrated a causative role for U. urealyticum in lower genital tract infections in women. Case-control studies have demonstrated a very high incidence of U. urealyticum in the lower genital tract of both symptomatic and asymptomatic women.'" However, a significant association of U. urealyticum was found in the present study with signs of a particular clinical entity, mucopurulent cervicitis. Quantitative differences in the con-
564
Paavonen et al.
centration of V. urealyticum might be sought in future comparisons of women with and without mucopurulent cervicitis. The lack of association of N. gonorrhoeae with mucopurulent cervicitis in sexually transmitted disease clinic patients in this study and in our previous studyI was unexpected and must be interpreted with caution. It is possible that the sample of patients referred to us with suspected mucopurulent cervicitis was biased in the direction of a falsely low incidence of gonorrhea if the referring clinicians tended to start treatment for gonorrhea on the basis of cervical Gram stain results (although we tried to avoid such bias) or if women with gonorrhea tended to present to the clinic with more acute, severe clinical symptoms suggesting pelvic inflammatory disease and thus were excluded from the cervicitis study. It may also be that women with gonorrhea develop acute symptoms, leading them to seek care in other clinical settings, or that the cervicitis produced by gonorrhea is transient whereas the inflammation produced by C. trachomatis is longer lasting, leading to more profound or more prolonged abnormalities of the cervix. None of the referred student health clinic women with mucopurulent cervicitis in our study had gonorrhea or genital herpes. In any case, manifestations of gonococcal nonchlamydial infection of the cervix require further study. Since our results suggest that two different types of mucopurulent cervicitis might exist, we compared chlamydial and nonchlamydial cases of mucopurulent cervicitis with respect to cervical findings on examination. Those with chlamydial cervicitis had higher cervicitis severity scores, more often had edema of the area of ectopy, and more often demonstrated induced mucosal bleeding, again suggesting that C. trachomatis is an invasive cervical pathogen. Subsequent studies are needed to assess our findings and to address the multifactorial etiology of cervical inflammation. Such studies should also elucidate the histopathologic characteristics of chlamydial and nonchlamydial cervicitis. We are indebted to Linda Cles and Judith Hale for performing microbiologic studies; to the staff of Seattle-King County STD Clinic (H. Hunter Handsfield,
March, 1986 Am J Obstet Gyneco1
M.D., Director); and to the staff of Hall Health Center, University of Washington (Gordon Bergy, M.D., Director), for their assistance; and to Ferne Beier for manuscript preparation. REFERENCES I. Brunham RC, Paavonen J, Stevens CE, et al. Mucopurulent cervicitis: the ignored counterpart in women of urethritis in men. N Engl J Med 1984;311: I. 2. Swartz SL, Kraus SJ, Herrmann KL, et al. Diagnosis and etiology of nongonoccal urethritis. Ann Intern Med 1978; 138:445. 3. Amsel R, Totten PA, Spiegel CA, et al. Nonspecific vaginitis: diagnostic criteria and microbial and epidemiologic associations. AmJ Med 1983;74:14. 4. Paavonen J, Kiviat N, Brunham RC, et al. Prevalence and manifestations of endometritis among women with cervicitis. AM J OBSTET GYNECOL 1985; 152:280. 5. Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New York: Wiley, 1981:173. 6. Bard J, Levitt D. Chlamydia trachomatis stimulates human peripheral blood B lymphocytes to proliferate and secrete polyclonal immunoglobulins in vitro. Infect Immun 1984;43:84. 7. Woodland RM, Johnson AP, Tuffray M. Animal models of chlamydial infection. Med Bull 1983;39: 175. 8. Paavonen J, Vesterinen E, Meyer B, Saksela E. Colposcopic and histologic findings in cervical chlamydial infections. Obstet Gynecol 1982;59:712. 9. Holmes KK, Chen KCS, Lipinsky CM, Eschenbach DA. Vaginal redox potential in bacterial vaginosis (nonspecific vaginitis). J Infect Dis 1985; 152:379. 10. Washington AE, Gove S, Schachter J, Sweet RL. Oral contraceptives, Chlamydia trachomatis infection, and pelvic inflammatory disease. JAMA 1985;253:2246. II. 130wie WR, Wang S-P, Alexander ER, Holmes KK. Etiology of nongonococcal urethritis: evidence for Chlamydia trachomatis and Ureaplmma urealyticum. J Clin Invest 1977;59:235. 12. Lamey JR, Eschenbach DA, Mitchell SH, et al. Isolation of mycoplasmas and bacteria from blood of postpartum women. AMJ OBSTET GYNECOL 1982;143:104. 13. McCormack WM, Taylor-Robinson D. The genital mycoplasmas. In: Holmes KK, Mardh P-A, Sparling PF, Wiesner PJ, eds. Sexually transmitted diseases. New York: McGraw-Hill, 1984:408. 14. Minkoff H, Grunebaum AN, Schwartz RH, et al. Risk factors for prematurity and premature rupture of membranes: a prospective study of the vaginal flora in pregnancy. AM J OBSTET GYNECOL 1984; 150:965. 15. Mardh P-A, Westrom L. Tubal and cervical cultures in acute salpingitis with special reference to Mycoplmma hominis and T strain mycoplasmas. Br J Vener Dis 1970;46: 179.
Key words: Cervicitis, cervical inflammation, chlamydia, ureaplasma, bacterial vaginosis
Infection of the uterine cervix is often asymptomatic and represents a reservoir for sexual and perinatal transmission of pathogenic microorganisms. It might lead to at least three possible types of complications: (1) ascending intraluminal spread of pathogenic organisms from the cervix, producing endometritis and salpingitis; (2) ascending infection during pregnancy, resulting in chorioamnionitis, premature ru pture of the membranes, amniotic fluid infection, premature delivery, and puerperal and neonatal infections; and (3) the initiation or promotion of cervical neoplasia. In a previous study of randomly selected women attending a clinic for sexually transmitted diseases, we assessed objective criteria for the clinical diagnosis of mucopurulent cervicitis, that is, visualization of yellow
From the Departments of Medicine, Biostatistics, Obstetrics and Gynecology, and Pathobiology, University of Washington, School of Medicine, The Hall Health Center, University of Washington, and the Harborview Medical Center. Supported in part by National Institutes of Health Grants AI-12192 and AI-14180. Received for publication July 1, 1985; revised November 13, 1985; accepted November 25, 1985. Reprint requests: King K. Holmes, M.D., Ph.D., Department of Medicine ZA-92, HarborviewMedical Center, 325 Ninth Ave., Seattle, WA 98104.
556
mucopurulent endocervical discharge on a white swab or the presence of 10 or more polymorphonuclear leukocytes per microscopic field (at a magnification of 1000) in satisfactory Gram-stained endocervical smears. 1 These criteria are quite similar to those commonly used to make a presumptive diagnosis of nongonococcal urethritis in men." The major recognized infectious causes of cervicitis are thought to be the sexually transmitted pathogens Chlamydia trachomatis, Neisseria gonorrhoeae, and herpes simplex virus. However, previous studies of the etiology of cervicitis have focused on selected groups of patients seen in sexually transmitted disease clinics (contacts of men with nongonococcal urethritis) and have examined the role of only a limited number of potential pathogens. The present study of mucopurulent cervicitis was undertaken in two clinical settings to define the microbiologic etiology of cervical inflammation in consecutively referred women with suspected cervicitis and in representative samples of women from two clinical settings.
Material and methods Study population. Women attending the SeattleKing County sexually transmitted disease clinic at Har-
Volume 154 Number 3
borview Medical Center or the University of Washington student health center in Seattle were eligible for study. Four patient groups, two from each clinic, composed the study population: two groups of consecutive patients referred for suspected cervicitis (one at each clinic), one randomly selected group from women attending the sexually transmitted disease clinic for the first time or with a new problem, and a group of consecutive women registering for routine annual examinations at the student health clinic. Women referred for suspected cervicitis. Clinicians at the sexually transmitted disease clinic and student health clinic were asked to refer women? 16 years of age whom they believed to have cervicitis but in whom they did not suspect pelvic inflammatory disease. Women who had received antibiotics in the previous month, who were menstruating, who were pregnant, or who had an intrauterine contraceptive device were excluded. One hundred sixty-one women in the sexually transmitted disease clinic and 47 women in the student health clinic were thus referred and enrolled for suspected cervicitis. Representative samples of women in two clinical settings. A random sample of 110 women attending the sexually transmitted disease clinic were selected as previously described.' At the student health clinic, 49 consecutive women registering for routine annual examinations were asked to participate in the study and all agreed. Patient evaluation. A standardized history with detailed information concerning menstrual history, contraceptive method, sexual behavior, and prior history of sexually transmitted disease was obtained from all women. All women underwent a gynecologic examination by one of the investigators. The color of cervical secretions as viewed on a white cotton swab was classified as clear, white, cloudy, or yellow; yellow secretions were defined as mucopus. The severity of inflammation of the endocervical epithelium was assessed by separately scoring the degree of induced mucosal bleeding as well as erythema and edema of the zone of ectopy when observed with the unaided eye as 0 (none or normal), I + (mild), 2 + (moderate), or 3 + (severe) and summing the score. Bacterial vaginosis was diagnosed according to the criteria developed by Amsel et aI.' Definition of mucopurulent cervicitis. As previously described,' a diagnosis of mucopurulent cervicitis was established on the basis of the presence of either a yellow mucopurulent endocervical discharge or ? 10 polymorphonuclear leukocytes per 1000 X microscopic field in cervical mucus on a satisfactory Gram-stained smear. Overall, 28 smears were not evaluable because the slide contained> 100 squamous cells in association with inflammatory cells in cervical mucus. Quantification of leukocytes in endocervical secre· tions. After the ectocervix was wiped clean with a large
Etiology of cervical inflammation
557
cotton swab, endocervical mucus was collected on a white-tipped swab with care taken to avoid contamination by vaginal secretions. The swab was rolled onto a 2 cm' area of a glass microscope slide. The smear was air-dried and Gram-stained by the rapid method. Slides were then examined for the number of polymorphonuclear leukocytes per microscopic field at 1000 x magnification in five nonadjacent fields as previously described in detail. 1 Laboratory methods and collection of specimens. An unlubricated speculum was inserted in the vagina and the cervix was exposed. The lateral vaginal fornix was swabbed with a cotton-tipped applicator and vaginal pH was measured directly from this swab. A sterile paper strip (5 by 20 mm) was inserted into the endocervical canal and allowed to saturate with fluid from the cervical mucus. This was carefully done to avoid any bleeding from the endocervix during the procedure. The filter paper strip was placed in 0.2 ml of phosphate-buffered saline solution and frozen at - 20° C until assayed for antibody to C. trachoma tis. Calcium alginate swabs (Inolex, Glenwood, Illinois) were used to obtain urethral and cervical cultures for C. trachomatis. These specimens were separately placed in 2 ml of sucrose-phosphate-glutamate transport medium. Subsequent successive endocervical specimens were obtained with cotton-tipped swabs for cultures for N. gonorrhoeae, Mycoplasma hominil, Ureaplasma llrealyticum, Gardnerella vaginalis, Streptococcus agalactiae, and
herpes simplex virus. Cervicovaginal specimens were obtained for culture of Trichomonas vagina lis and yeast. Microbiologic studies. Isolation of C. trachomatis was performed in two cooperating laboratories with the use of either cycloheximide-treated McCoy cells in microtiter plates or diethylaminoethyl-dextran-treated HeLa 229 cells in tubes.' N. gonorrhoeae was isolated by direct inoculation of the swabs on Thayer-Martin agar with incubation and identification by standard methods. Specimens were inoculated onto A 7 agar medium and into broth media for isolation of M. hominis and U. llrealyticum. ' Broth showing evidence of growth was subcultured to A 7 agar for confirmation; colonies were identified by typical morphologic features. G. vaginalis was isolated on human blood Tween agar medium. Colonies showing l3-hemolysis were further identified as G. vaginalis as previously described.' Group B streptococci were isolated in Todd-Hewitt broth containing 5% sheep blood and 5 mg/L of gentamicin and 15 mg/L of nalidixic acid; broth was subcultured to sheep blood agar for confirmation and identification of group B streptococci on the basis of I)-hemolytic reaction and latex agglutination with group-specific antiserum (Streptex, Wellcome Diagnostics, Research Triangle Park, North Carolina). Cervical specimens were placed in viral transport medium and inoculated into a monolayer of fibroblasts from human foreskin for
558
Paavonen et al.
March, 1986
Am J Obstet Gyneco1
Table I. Associations of demographic and behavioral characteristics with mucopurulent cervicitis stratified by patient group STD clinic women Referred MPC Neg. (n = 9) No.
Age (yr) Race white Marital status single Education (yr) Gravidity;:" I Parity;:" I Method of contraception Oral contraceptives Spermicides Other None Age at first intercourse (yr) Median lifetime sex partners (No.) Past history of STDt
I
o/c
24.7 ± 4.1 7 78 6 67 13.8 ± 2.1 5 56 I II 2 22 3 33 I II 3 33 16.6 ± 1.4 20 5/8 63
Sampled MPC Pos. (n = /33)
No.
I
MPC Neg. (n = 51)
%
23.4 ± 5.2 1051132 80 1031130 79 13.2 ± 2.5 60/132 46 41 31 80 16 6 28
No.
I
MPC Pos. (n = 39)
%
23.7 ± 5.7 65 35/47 75 12.6 ± 1.9 24/50 48 12 24
33
62 12 5 22
9 13 3 26
17.2 ± 2.5 8 901132 68
7 32
18 25* 6 51 16.3 ± 2.5 63
No.
I
%
22.3 ± 4.7
72 76
28 29/38
12.5 ± 2.0 7
58 18
16 4 7 II
42 II 18 29
22/38
16.1 ± 2.1 10 61 23/38
STD = Sexually transmitted disease; MPC = mucopurulent cervicitis; Neg. = negative; Pos. = positive. *p oS 0.0 I. tGonorrhea, pelvic inflammatory disease, genital warts, genital herpes, or trichomoniasis. :j:p oS 0.05.
isolation of herpes simplex virus.' Cervicovaginal specimens were inoculated into Diamond's medium for isolation of T. vagina lis and on Sabaraud's agar for isolation of yeasts. Serum and secretion antibody determinations. Serum samples were tested by the microimmunoAuorescence technique for IgG antibodies for all C. trachomatis elementary body serotypes.' Cervical secretions were similarly tested for secretory IgA antibodies to C. trachomatis. Statistical methods. Initial comparisons of demographic, behavioral, and microbiologic characteristics of women with or without mucopurulent cervicitis were done in each of the four patient groups by standard X" or t-tests. In order to obtain an overall assessment of the strength of the associations between possible causative factors and mucopurulent cervicitis while minimizing the statistical problems of multiple comparisons, the associations were analyzed by first stratifying patients into the four categories (referred or not referred, sexually transmitted disease or student health clinic) and then combining these results by calculating an overall odds ratio according to the Mantel-Haenszel method.' This procedure assumes that the associations are the same in all strata, and this assumption was tested for each association by use of a X2 test for homogeneity, which is significant if there are large disparities among the odds ratios in the four strata. The p value reported with each odds ratio corresponds to the test of the hypothesis that the odds
ratio is equal to I (that is, no association). These analyses were conducted with the 1981 version of the BMPD statistical program P4F. Similar Mantel-Haenszel analyses were also conducted separately in groups of women with or without evidence of infection by C. trachomatis (as measured by positive cervical culture), and the results were combined into an overall statistic that adjusted for the presence of chlamydial infection. Since the purpose of this analysis was to compare the relative strengths of the associations of several variables with mucopurulent cervicitis, the p values reported for the summary odds ratios (see Tables III, IV, and V) have not been adjusted to account for the number of associations examined.
Results Incidence of mucopurulent cervicitis among referred and sampled study populations. Thirteen of the 367 women initially enrolled (10 from the sampled sexually transmitted disease clinic population and three from the referred sexually transmitted disease clinic patient group) were menstruating and thus excluded from further evaluation because cervical findings could not be adequately assessed. The diagnosis of mucopurulent cervicitis could not be made in an additional 31 women who did not have mucopus and in whom polymorphonuclear leukocytes could not be evaluated because either the Gram-stained smear was unsatisfactory or a smear was not obtained. Of the remaining
Etiology of cervical inflammation
Vulume 154 Kumber 3
559
Student health clinic women Referred
Sampled
II10.
I
%
No.
75 86
(I
0 E. (I
2
16.7 ± 2.0
No.
9 3
23 8
0 75 0 25
19 9 2 8
50 24 5 21
9 9 5 6
10:1:
4 11123
48
5 7
MPC Pas. (n = 15)
% 93 86 24 3 31 31 17 21
No.
323 patients, mucopurulent cervicitis was detected in 39 (43%) of the 90 randomly selected sexually transmitted disease clinic female patients, in 133 (94%) of the 142 sexually transmitted disease clinic patients referred for suspected cervicitis, in 15 (34%) of the 44 consecutive women seen for routine annual examination at the student health clinic, and in 39 (83%) of the 47 student health clinic patients referred for suspected cervicitis. Associations of demographic and behavioral characteristics with mucopurulent cervicitis in the four patient groups. As shown in Table I, there were differences in contraceptive usage patterns among women with and without mucopurulent cervicitis. Among sexually transmitted disease clinic women, oral contraceptives were used by 96 (57%) of 168 women with mucopurulent cervicitis versus only 11 (18%) of60 women without mucopurulent cervicitis (p < 0.001). A similar difference in oral contraceptive usage was seen among referred student health clinic women (19 of 38 versus zero of eight, p < 0.01) but not in the student health clinic sample. No significant differences were found in age, race, marital status, years of education, gravidity, parity, or sexual history between women with and without mucopurulent cervicitis in any of the study populations. In the referred student health clinic population, a greater proportion of women with mucopurulent cervicitis had a history of past sexually transmitted disease (11 or 48% of 23) compared with those without mucopurulent cervicitis (zero of seven, p < 0.05). Associations of specific microorganisms or clinical diagnoses with mucopurulent cervicitis. The prevalence of cervical and vaginal pathogens, clinical diag-
I
23.3 ± 3.8 131\4 131\4 15.0 ± 1.6 5 I
% 93 93 33 7 33 40 7 20
5 6 I
3 18.2 ± 1.9
18.6 ± 2.8
18.3 ± 3.0
19.8 ± 2.9
I
23.1 ± 4.5 26/28 25/29 15.6 ± 2.0 7
0 0
:"
017
% 85 100
29/34 21/21
15.9 ± 1.4 (I
I 23.4 ± 3.8
23.6 ± 2.0 6 617
MPC Neg. (n = 29)
MPC Pos. (n = 39)
MPC Neg. (n = 8)
24
4 7
47
noses, and chlamydial antibodies among those with and without mucopurulent cervicitis stratified by patient group is shown in Table II. In separate analyses done within each group, mucopurulent cervicitis was associated with several infectious agents or clinical conditions. In the sexually transmitted disease clinic groups, the presence of mucopurulent cervicitis was most highly associated with positive cultures for C. trachomatis, U. urealyticum, or T. vaginalis, with the presence of serum antibodies to C. trachorrwtis, and with the clinical diagnosis of bacterial vaginosis. In the student health clinic groups, mucopurulent cervicitis was associated with the isolation of C. trachomatis (p = 0.06), T. vaginalis, and group B streptococcus and with the clinical diagnosis of bacterial vaginosis; the isolation of yeast had a negative association with mucopurulent cervicitis. Although isolation of herpes simplex virus was not associated with mucopurulent cervicitis, characteristic ulcerative or necrotic lesions of the cervix were observed by colposcopy in eight (67%) of 12 women from whom herpes simplex virus was isolated versus 12 (4%) of 302 women from whom herpes simplex virus was not isolated (p < 0.00\). Association of selected variables with mucopurulent cervicitis by the Mantel-Haenszel test. Since may variables were associated with mucopurulent cervicitis in the separate group analyses, the comparisons of the strengths of these associations were made by determining Mantel-Haenszel summary odds ratios for each variable (Table III). Combining the odds ratios derived from all four groups was found justified by the nonsignificance of the test for homogeneity of the odds ratios from the patient groups (data not
560
Paavonen et al.
March, 1986 Am
J Obstet Gyneco1
Table II. Associations of cervical microorganisms, clinical diagnoses, and antichlamydial antibodies with mucopurulent cervicitis stratified by patient group STD clinic women Referred MPC Neg. (n = 9) No.
C. trachomatis Cervical culture Serum antibody Cervical secretory IgA antibody N. gonorrhoeae
Herpes simplex virus T. vaginalis M. hominis U. urea(vticum
Group B streptococcus G. vaginalis Yeast Bacterial vaginosis STD *p :s tp:s :/:p :s
I
I 4 0 017 0 0 3 4 2/8 5 118 3/8
MPC Pos. (n = 133)
%
No.
II 44t 0 0 0 0 33 44:/: 25 56 13 38
56 9111 18 2411 28 91132 71131 121132 551129 1081130 211130 95/126 1511 29 4411 21
I
MPC Neg. (n = 51)
%
No.
42 77 19 7 5 9 43 83 16 75 12 36
I 27/47 8/48 7/50 4/50 4/50 20 37 7 29/47 16/50 14/50
I
MPC Pos. (n = 39)
%
No.
2* 57:/: 17 14 8 8:/: 40 74t 14 62 32 28t
19 32/37 9/37 7 1138 11/38 20 35 7 29/38 6/38 19/38
I
% 49 86 24 18 3 29 51 90 18 76 16 50
= Sexually transmitted disease; MPC = mucopurulent cervicitis; Neg. = negative; Pos. = positive. 0.001. 0.05. 0.0 I.
Table III. Association of selected variables with mucopurulent cervicitis by the MantelHaenszel test
C. trachomatis Cervical culture Serum antibody Cervical secretory IgA antibody N. gonorrhoeae Herpes simplex virus T. vaginalis M. hominis U. urealyticum Group B streptococcus G. vaginalis Yeast Bacterial vaginosis Oral contraceptive use Spermicide use
*NS
Sampled
=
Summary odds ratio
p value
16.39 4.24 2.35 1.66 0.58 3. 17 1.32 2.70 1.14 2.39 0.36 I. 94 3.44 0.42
<0.001 <0.001 NS* NS NS 0.045 NS 0.004 NS 0.013 0.027 0.052 0.001 0.014
not significant; p > 0.10.
shown). As shown in Table III, several variables were significantly associated with mucopurulent cervicitis, including C. trachomatis (odds ratio = 16.39, P < 0.001), the presence of serum antibodies to C. trachomatis (odds ratio = 4.24, P < 0.001), oral contraceptive use (odds ratio = 3.44, P < 0.001), isolation ofT. vagi-naZis (odds ratio = 3.17, P = 0.04), isolation of U. urealyticum (odds ratio = 2.70, P < 0.01), isolation of C. vagi-naZis (odds ratio = 2.39. P = 0.(1),
bacterial vaginosis (odds ratio = 1.94, P = 0.(5), spermicide use (odds ratio = 0.42, P = 0.01, negative association), and the presence of yeast (odds ratio = 0.36, P = 0.03, negative association). Data concerning isolation of C. trachoma tis, N. gonorrhoeae, and herpes simplex virus from the randomly selected sexually transmitted disease clinic women were previously used in defining the objective criteria for diagnosis of mucopurulent cervicitis and its strong relationship to C. trachomatis. I If these data are excluded from analysis, isolation of C. trachomatis remains significantly associated with mucopurulent cervicitis (odds ratio = 7.97, P < 0.01), while isolation of N. gonorrhoeae and herpes simplex virus still demonstrates no such association. Summary odds ratios for mucopurulent cervicitis after adjusting for isolation of C. trachomatis. Since mucopurulent cervicitis was most strongly associated with positive C. trachomatis culture as measured by the largest odds ratio, we stratified the study population into those with and without cultural evidence of cervical C. trachomatis infection and repeated the analysis to clarify further the possible correlates of mucopurulent cervicitis. As shown in Table IV, oral contraceptive use (odds ratio = 2.55, P = 0.(2), isolation of U. ureaZyticum (odds ratio = 2.39, P = 0.(2), and isolation of yeast (odds ratio = 0.27, P = 0.03, negative association) were most strongly associated with mucopurulent cervicitis after adjustment for C. trachomatis. In addition, the presence of serum antibodies to C. tra-
Etiology of cervical inflammation
Volume 154 Number 3
561
Student health clinic women Referred
Sampled MPC Pos. (n = 39)
MPC Neg. (n = 8) No.
0 1/4
0 0 0 I ~!
:3 I 317 I :2
I
%
No.
0 25 0 0 0 13t 25 38 13t 43 8 25
12/38 20/38 3/38 2/38 0/37 0/38 10/36 23/37 0/38 22/37 2/37 7
I
MPC Neg. (n = 29)
%
No.
32 53 8 5 0 0 28 62 0 59 5 18
I 1115 1/28 0 0/26 0115 3125 11126 0116 5116 5116 3
chomatis (odds ratio = 2.20, P = 0.09), isolation of C. vaginalis (odds ratio = 1.95, P = 0.09), and bacterial vaginosis (odds ratio = 1.87, P = 0.09) demonstrated associations with mucopurulent cervicitis after adjustment for C. trachomatis, although these associations were not statistically significant. When the analysis was restricted to women with negative C. trachomatis cultures, odds ratios of 2.27, 2.24, 2.08, 2.03, and 1.95 were obtained for oral contraceptive use, the presence of serum antibodies to C. trachomatis, isolation of C. vaginalis, isolation of U. urealyticum, and the clinical diagnosis of bacterial vaginosis, respectively (0.05 < P < 0.1 for each), and a significant negative association was observed for the isolation of yeast (odds ratio = 0.24, P < 0.02). Among culturenegative women, oral contraceptive use was strongly associated with mucopurulent cervicitis in those who were seropositive (odds ratio = 6.61, P < 0.01) but not in those who were seronegative (odds ratio = 1.5). Among women with a postive C. trachomatis culture, only U. urealyticum (p < 0.01) was significantly associated with mucopurulent cervicitis. Although oral contraceptive use and serum antibody to C. trachomatis demonstrated the highest odds ratios for mucopurulent cervicitis among those with negative cultures for C. trachomatis, neither was associated with mucopurulent cervicitis among those with a positive C. trachomatis culture. Overall, 130 (82%) of 159 sexually transmitted disease clinic women with mucopurulent cervicitis and 31 (55%) of 56 sexually transmitted disease clinic women without mucopurulent cervicitis demonstrated cultural and/or serologic evidence of C. trachomatis infection. The corresponding figures among student health clinic
I
MPC Pos. (n = 15)
%
No.
3 7 4 0 0 0 12 42 0 31 31t lOt
2114 2111 1 0 0 0111 0114 7 2110 7/10 0/11
5
1
% 14 18 7 0 0 0 0 47 20 70 0 33
women were 25 (50%) of 50 and three (15%) of 20. Association of selected manifestations of cervical inflammation with C. trachomatis among women with mucopurulent cervicitis. Since mucopurulent cervicitis was associated with several variables after adjustment for C. trachomatis culture, we were interested in whether specific clinical manifestations of cervicitis differed for women with and without a positive C. trachomatis culture. As shown in Table V, among women with mucopurulent cervicitis, several findings were more common in those who had a positive cervical culture, including cervicitis severity score> 3 (odds ratio = 3.45, P < 0.001), induced mucosal bleeding (odds ratio = 2.37, P < 0.01), and edema of the area of ectopy (odds ratio = 2.05, P = 0.04).
Comment In our previous study of randomly selected sexually transmitted disease clinic women, we described objective criteria for the office diagnosis of mucopurulent cervicitis.! The presence of cervical mucopus (positive swab test) and ~ 10 polymorphonuclear leukocytes per 1000 X microscopic field in cervical mucus demonstrated the strongest associations with cervical infection by C. trachomatis, N. gonorrhoeae, or herpes simplex virus. In this extended study, we analyzed the associations of multiple pathogens, clinical diagnoses, and demographic and behavioral variables with mucopurulent cervicitis. The findings we obtained among the previously reported randomly selected sexually transmitted disease clinic women were examined to see if they extended to another group of sexually transmitted disease
562
Paavonen et al.
March, 1986
Am
J Obstet Gynecol
Table IV. Association of selected variables with mucopuprulent cervicitis by the Mantel-Haenszel test among all patients, adjusted for C. trachomatis, and among patients with a negative C. trachoma tis culture Among all patients, adjusted for isolation of C. trachomatis
Among patients with negative C. trachomatis culture
p Odds ratio*
I
Value
Odds ratio*
0.09 NSt NS NS NS NS 0.021 NS 0.09 0.026 0.09 0.021 NS
2.24 1.59 0.20 1.16 1.94 0.93 2.03 0.69 2.0S 0.24 1.95 2.27 0.57
I
P
Value
C. trachomatis
Serum antibody Cervical secretory IgA antibody N. gonorrhoeae Herpes simplex virus T. vaginalis M. hominis U. urealyticum Group B streptococcus C. vaginalis Yeast Bacterial vaginosis Oral contraceptive use Spermicide use
2.20 I.SS 0.37 I.IS 2.21 0.95 2.39 0.79 1.95 0.27 I.S7 2.55 0.53
0.09 NS NS NS NS NS 0.09 NS 0.07 O.OIS O.OS 0.06 NS
*Summary odds ratio. tNS = Not significant; p> 0.10.
clinic women (that is, those referred to us because of suspected cervicitis) and to groups of women seen in a student health clinic. We also attempted to analyze further the microbial and other correlates of mucopurulent cervicitis among women negative for C. trachomatis. In separate analyses in the four patient groups, multiple infectious agents (including C. trachomatis, U. urealyticum, and T. vaginalis), serum antibody to C. trachomatis, past history of sexually transmitted disease, oral contraceptive use, and bacterial vagi nos is were associated with mucopurulent cervicitis in one or more of the groups studied. On the other hand, isolation of yeast, usually Candida albicans, from the vagina had a negative association with mucopurulent cervicitis. To make overall comparisons among correlates of mucopurulent cervicitis, we next combined all four groups to obtain weighted average odds ratios for each variable. This summary analysis of the combined data demonstrated four major findings. First, C. trachoma tis detected by culture and serologic testing demonstrated the strongest association with mucopurulent cervicitis. Even after adjustment for C. trachoma tis culture and in particular among women with negative cultures for C. trachomatis, the presence of serum antibodies to C. trachomatis remained associated with mucopurulent cervicitis (odds ratio = 2.2, P = 0.09). This might suggest false negative cultures for C. trachomatis in many of the seropositive women or the persistence of abnormal cervical findings after chlamydial cervicitis among the culture-negative women who remain antibody-positive. The lack of association with secretery IgA antibodies to C. trachomatis, which may be a closer reflection of
current chlamydial infection, argues against the former explanation. Infections with C. trachomatis are characteristically chronic. Data from experimental infections suggest that the immune response to C. trachomatis may contribute to some of the manifestations of chronic infection." 7 The histopathologic features of chlamydial cervicitis are characterized by severe inflammatory infiltration of the stroma and prominent lymphoid follicles." The natural history of these histopathologic changes after eradication of C. trachomatis has not been extensively studied. We have observed that increased vascularity and erythema of the transformation zone in the cervix persists for prolonged periods of time after treatment (Paavonen J, unpublished results). Independent association of C. trachomatis with mucopurulent cervicitis on serologic testing suggests persistence of these histopathologic changes in the cervix. Second, both bacterial vaginosis and the isolation of C. vaginalis (a bacterial vagi nos is-associated pathogen) demonstrated associations with mucopurulent cervicitis, suggesting that changes in the cervicovaginal environment might foster the development of mucopurulent cervicitis or vice versa. The redox potential at the vaginal epithelial surface is greatly reduced in bacterial vaginosis." Oxygen consumption by polymorphonuclear leukocytes might decrease the redox potential and increase the pH of the vaginal environment to favor the growth of organisms associated with bacterial vaginosis. Mucopurulent endocervical discharge might similarly change the vaginal environment. Uniform clinical criteria for bacterial vaginosis and mucopurulent cervicitis should be used in future clinical studies
Etiology of cervical inflammation
Volume 154 Number 3
563
Table V. Association of cervical findings with results of C. trachomatis culture among sexually transmitted disease clinic women and student health clinic women with mucopurulent cervicitis C. trachomatis
Cervical finding
Cervicitis severity score 0-3 4-9 Ectopy Not present Present Erythema of ectopy:j: 0-1 2-3 Edema of ectopy:j: 0-1 2-3 Induced mucosal bleeding 0-1 2-3
Positive (n
= 88)
I
%
No.
23 63
27 73
10 78
Negative (n
= 133)
I
%
Odds ratio*
p Value
72 63
53 47
3.45
<0.001
II 89
21 114
16 84
1.58
NSt
22 56
28 72
42
37 63
1.49
NS
72
45 33
58 42
81 33
71 29
2.05
0.04
36 50
42
87 48
64 36
2.37
0.004
No.
58
*Summary odds ratio over patient groups. tNS = Not significant; p> 0.10. :j:Among women with ectopy.
of how the two conditions are interrelated. Does bacterial vaginosis resolve spontaneously after treatment of mucopurulent cervicitis, or should both conditions be treated concomitantly or successively? The negative association of yeast with mucopurulent cervicitis is also of interest, although it may simply be that our exclusion of women who had received antibiotics within the previous month did not exclude an effect of more remote antibiotic use, which could have promoted vulvovaginal candidiasis while eliminating mucopurulent cervicitis. The third major finding was the strong association of oral contraceptive use with mucopurulent cervicitis. Oral contraceptive-induced hyperplasia of the endocervical epithelium might render the cervix more susceptible to many sexually transmitted disease organisms. Most prior studies have demonstrated that oral contraceptive use is associated with an increased risk of C. trachomatis infection. 10 It is not clear, however, whether the higher risk of chlamydial infection in oral contraceptive users is secondary to differences in sexual behavior or to hormonal effects of oral contraceptives. Women using oral contraceptives have a higher incidence of cervical ectopy compared with that of women with natural cycles. 1O Cervical ectopy may render the cervix more susceptible to chlamydial infection, or detection of C. trachoma tis by culture might be easier among such women. Oral contraceptive use might also stimulate the growth of C. trachoma tis in cervical epithelial cells. However, among women with negative cervical cultures for C. trachomatis, oral contraceptive use was correlated with mucopurulent cervicitis only
among those with serum antibody to C. trachomatis (p < 0.01). This was true even after adjustment for variables reRecting sexual activity (such as age at first intercourse), suggesting that the relationship is not totally explained by increased sexual activity among women using oral contraceptives. Thus our results suggest that oral contraceptives increase the risk for mucopurulent cervicitis among women with current or past exposure to C. trachomatis. The fourth finding was that U. urealyticum was the only organism showing a significant positive association with mucopurulent cervicitis after adjustment for results of cervical cultures for C. trachomatis. This association is consistent with previous studies in which U. urealyticum has been associated, independent of C. trachomatis, with nongonococcal urethritis in men." Furthermore, several studies have associated U. urealyticum with puerperal infections,'2 low birth weight of infants,':l premature labor," subclinical endometritis,'" and infertility':l suggesting that U. urealyticum is a significant genital pathogen. In contrast, U. urealyticum has only occasionally been isolated from the fallopian tubes of women with salpingitis," and previous studies have not demonstrated a causative role for U. urealyticum in lower genital tract infections in women. Case-control studies have demonstrated a very high incidence of U. urealyticum in the lower genital tract of both symptomatic and asymptomatic women.'" However, a significant association of U. urealyticum was found in the present study with signs of a particular clinical entity, mucopurulent cervicitis. Quantitative differences in the con-
564
Paavonen et al.
centration of V. urealyticum might be sought in future comparisons of women with and without mucopurulent cervicitis. The lack of association of N. gonorrhoeae with mucopurulent cervicitis in sexually transmitted disease clinic patients in this study and in our previous studyI was unexpected and must be interpreted with caution. It is possible that the sample of patients referred to us with suspected mucopurulent cervicitis was biased in the direction of a falsely low incidence of gonorrhea if the referring clinicians tended to start treatment for gonorrhea on the basis of cervical Gram stain results (although we tried to avoid such bias) or if women with gonorrhea tended to present to the clinic with more acute, severe clinical symptoms suggesting pelvic inflammatory disease and thus were excluded from the cervicitis study. It may also be that women with gonorrhea develop acute symptoms, leading them to seek care in other clinical settings, or that the cervicitis produced by gonorrhea is transient whereas the inflammation produced by C. trachomatis is longer lasting, leading to more profound or more prolonged abnormalities of the cervix. None of the referred student health clinic women with mucopurulent cervicitis in our study had gonorrhea or genital herpes. In any case, manifestations of gonococcal nonchlamydial infection of the cervix require further study. Since our results suggest that two different types of mucopurulent cervicitis might exist, we compared chlamydial and nonchlamydial cases of mucopurulent cervicitis with respect to cervical findings on examination. Those with chlamydial cervicitis had higher cervicitis severity scores, more often had edema of the area of ectopy, and more often demonstrated induced mucosal bleeding, again suggesting that C. trachomatis is an invasive cervical pathogen. Subsequent studies are needed to assess our findings and to address the multifactorial etiology of cervical inflammation. Such studies should also elucidate the histopathologic characteristics of chlamydial and nonchlamydial cervicitis. We are indebted to Linda Cles and Judith Hale for performing microbiologic studies; to the staff of Seattle-King County STD Clinic (H. Hunter Handsfield,
March, 1986 Am J Obstet Gyneco1
M.D., Director); and to the staff of Hall Health Center, University of Washington (Gordon Bergy, M.D., Director), for their assistance; and to Ferne Beier for manuscript preparation. REFERENCES I. Brunham RC, Paavonen J, Stevens CE, et al. Mucopurulent cervicitis: the ignored counterpart in women of urethritis in men. N Engl J Med 1984;311: I. 2. Swartz SL, Kraus SJ, Herrmann KL, et al. Diagnosis and etiology of nongonoccal urethritis. Ann Intern Med 1978; 138:445. 3. Amsel R, Totten PA, Spiegel CA, et al. Nonspecific vaginitis: diagnostic criteria and microbial and epidemiologic associations. AmJ Med 1983;74:14. 4. Paavonen J, Kiviat N, Brunham RC, et al. Prevalence and manifestations of endometritis among women with cervicitis. AM J OBSTET GYNECOL 1985; 152:280. 5. Fleiss JL. Statistical methods for rates and proportions. 2nd ed. New York: Wiley, 1981:173. 6. Bard J, Levitt D. Chlamydia trachomatis stimulates human peripheral blood B lymphocytes to proliferate and secrete polyclonal immunoglobulins in vitro. Infect Immun 1984;43:84. 7. Woodland RM, Johnson AP, Tuffray M. Animal models of chlamydial infection. Med Bull 1983;39: 175. 8. Paavonen J, Vesterinen E, Meyer B, Saksela E. Colposcopic and histologic findings in cervical chlamydial infections. Obstet Gynecol 1982;59:712. 9. Holmes KK, Chen KCS, Lipinsky CM, Eschenbach DA. Vaginal redox potential in bacterial vaginosis (nonspecific vaginitis). J Infect Dis 1985; 152:379. 10. Washington AE, Gove S, Schachter J, Sweet RL. Oral contraceptives, Chlamydia trachomatis infection, and pelvic inflammatory disease. JAMA 1985;253:2246. II. 130wie WR, Wang S-P, Alexander ER, Holmes KK. Etiology of nongonococcal urethritis: evidence for Chlamydia trachomatis and Ureaplmma urealyticum. J Clin Invest 1977;59:235. 12. Lamey JR, Eschenbach DA, Mitchell SH, et al. Isolation of mycoplasmas and bacteria from blood of postpartum women. AMJ OBSTET GYNECOL 1982;143:104. 13. McCormack WM, Taylor-Robinson D. The genital mycoplasmas. In: Holmes KK, Mardh P-A, Sparling PF, Wiesner PJ, eds. Sexually transmitted diseases. New York: McGraw-Hill, 1984:408. 14. Minkoff H, Grunebaum AN, Schwartz RH, et al. Risk factors for prematurity and premature rupture of membranes: a prospective study of the vaginal flora in pregnancy. AM J OBSTET GYNECOL 1984; 150:965. 15. Mardh P-A, Westrom L. Tubal and cervical cultures in acute salpingitis with special reference to Mycoplmma hominis and T strain mycoplasmas. Br J Vener Dis 1970;46: 179.