Abnormal Vaginal pH and Mycoplasma genitalium infection

Abstract

Study Objective

Mycoplasma genitalium (MG) is a sexually transmitted pathogen linked to female morbidity, but testing for MG is not standardized. We aimed to determine which point-of-care (POC) vaginal tests could predict MG infection.

Design, Setting, Participants

A cross sectional study recruited sexually active adolescent women, aged 14–22 years (n=217) from an urban medical center.

Interventions and Main Outcome Measures

Vaginal swabs were POC tested for pH, amines, clue cells, sialidase and Trichomonas vaginalis (TV). MG was detected by research-use-only transcription mediated amplification (TMA) assay. Presence of Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) were confirmed using TMA. Three criteria were required for diagnosis of Clinical BV: pH >4.5, presence of amines, and > 20% clue cells. Associations were assessed using logistic regression (LR).

Results

TMA detected MG in 30 (14%), CT in 49 (23%), and NG in 21 (10%) of the samples tested. POC vaginal tests were positive for TV in 21%, amines in 52%, clue cells in 33%, sialidase in 22%, pH > 4.5 in 56%, and clinical BV in 19% of the samples tested. Using logistic regression, pH > 4.5 was a predictor of MG (odds ratio 4.4, p<.05). Of 131 women without clinical BV or TV, 25% of those with pH>4.5 had MG, compared to 9% of those with pH≤4.5 (p=.02).

Conclusions

Until standardized, approved testing for MG is available, pH may be a useful indicator to suspect MG, especially in the absence of BV and TV.

Introduction

Mycoplasma genitalium (MG) has been implicated as a causative agent in conditions with significant sequelae such as cervicitis,^1,2 endometritis,³ and pelvic inflammatory disease (PID). ^4,5 Additional studies have linked MG in women to post-abortion infections,⁶ tubal factor infertility,⁷ and preterm delivery.⁸ The prevalence of MG varies based on the population, but is similar to the prevalence of other STIs within that population.^9–12 Previously, we found MG to be as prevalent (22%) as Chlamydia trachomatis (CT, 24%) and Trichomonas vaginalis (TV, 18%) in adolescent women and as likely to be asymptomatic as CT and TV infections.¹¹

Although the prevalence of MG is similar to CT, and infection may confer future risks, most clinicians are unable to assess the potential risk of MG in their adolescent and young adult patients. MG is difficult to culture because it is slow growing and has strict nutrient requirements, and culture media is not widely available.¹³ Recently, nucleic acid amplification tests (NAAT) were developed that allowed further studies to investigate the links between MG and outcomes described above.^13–15 However, these sensitive NAATs are not approved by regulatory agencies, nor are there standardized guidelines for MG testing. Therefore, clinicians who suspect MG in their patients are at a disadvantage.

Clinicians often use simple point-of-care (POC) tests to diagnose vaginal conditions and STIs. For example, most clinicians rely upon Amsel’s clinical criteria to define Bacterial vaginosis (BV): elevated vaginal pH, amines, clue cells seen on wet mount, and homogeneous vaginal discharge.¹⁶ An elevated vaginal pH is also predictive of trichomoniasis.¹⁷ The clinician-performed amine or “whiff” test alone can indicate bacterial vaginosis.^17,18 Elevated white blood cells observed on wet mount can indicate candidiasis and STIs,^19,20 while an antigen test can detect TV. ^21,22

Given the potential clinical importance of MG and the lack of standardized, approved diagnostics, we evaluated whether any vaginal POC tests could predict the presence of MG in adolescent women.

Materials and Methods

Subjects

This study was nested within a larger cross-sectional study comparing the accuracy of self-collected specimens for trichomoniasis testing, as described previously.²³ The initial study and this analysis were approved by the hospital’s institutional review board. STROBE guidelines were utilized throughout.²⁴ Only the results of the clinician-collected vaginal swabs are reported for this study. Briefly, we recruited a convenience sample of women who were attending a visit at an urban, hospital-based Teen Health Center or Emergency Department between July 2006 and August 2008. Women aged 14–22 years who reported a history of vaginal intercourse within the previous three months were eligible. Subjects completed a questionnaire encompassing demographics, sexual behaviors (condom use, hormonal contraceptive use, number of sexual partners, etc.), and the presence or absence of vaginal symptoms. The clinician performed a pelvic exam and, using a speculum, collected four vaginal swabs.

Laboratory Methods

Vaginal swabs were used to perform POC tests (wet mount, pH, amine, sialidase and TV rapid antigen test) and other STI testing. The wet mount was read at 400× magnification by an experienced clinician or laboratory technician. White blood cells (WBC) were categorized as either low (≤5 per high powered field (hpf)) or moderate (≥6 per hpf).¹⁹ Clue cells were categorized as present if >20% of the epithelial cells per field were affected.¹⁶ Vaginal pH was recorded using pH paper (pHydrion®, Mikro Essentials Laboratories, Brooklyn, NY) and was classified as either >4.5 (elevated) or ≤4.5 (normal) based on Amsel’s criteria for BV.²⁵ Amines were positive if a strong fishy odor was elicited after applying a drop of potassium hydroxide to the vaginal swab.¹⁶ Sialidase and TV antigen tests were performed following manufacturer’s directions (BVBlue® Test and OSOM® TV Trichomonas Rapid Test, Genzyme Diagnostics, Cambridge, MA). Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) were detected utilizing NAAT (Aptima ®Combo2, Gen-Probe, Inc, San Diego, CA). MG was detected from the same swab using research-use-only (RUO) reagents as described previously.¹¹

Definitions of Other Outcomes

BV was defined using a modified Amsel’s criteria which required the presence of the following three factors: pH >4.5, clue cells >20%, and positive amine test.²⁶ Using these three objective criteria generates sensitivity and specificity similar to the full Amsel’s criteria, when using Nugent’s gram stain as the comparator.^18,26,27 Vaginal symptoms included elicited patient reports of vaginal discharge or itching on the day of the visit. Hormonal contraception in the last three months was marked as positive if the participant reported the use of contraceptive pills, ring, patch or the depot medroxyprogesterone acetate shot.

Statistical Analyses

Dichotomous variables were compared with MG status using chi squared test. Each variable statistically associated with MG was compared to other variables in the dataset to look for interactions. All variables associated with either the outcome (MG) or predictor variables (e.g., pH or hormonal contraception use) at p<0.1 were entered into the logistic regression model. We performed stepwise backward elimination to determine which factors were independently associated with MG, and present the final reduced model. Finally, we investigated the link between MG and pH in the subset of women without clinical BV or a positive POC TV test.

Results

Of the 249 females recruited for the original trichomoniasis study, 217 with MG testing comprise this study’s sample. The mean age of subjects was 18.3 years (range 14–22 years). Eighty-six percent self-identified as Black-Non Hispanic. Other subject history and laboratory variables are listed in Table 1. MG was detected in 30 (14%), CT in 49 (22.6%), NG in 21 (9.7%), and TV in 45 (20.7%) of our participants. For vaginal POC markers, 56% (118 of 211) had a vaginal pH >4.5, 33% (71 of 209) had >20% clue cells, 52% (109 of 211) tested positive for amines and 19% (39 of 209) had all three of the above thus meeting our study criteria for clinical BV. Use of hormonal contraception in the last three months was reported by 78 of 209 participants (38%) including estrogen containing oral contraceptive pills, patch or ring (n=33), or depot medroxyprogesterone acetate (n=45). There were no differences in MG infection by specific type of hormonal contraception, so these were analyzed together.

Table 1.

Subject Characteristics and predictor variables stratified by MG status. Totals differ due to missing data. Significance (P Value) assessed with chi-squared testing.

	Total n =217			MG Positive n = 30			MG Negative n = 187			P value
	Number with factor	Number tested	%	Number with factor	Number tested	%	Number with factor	Number tested	%
Age:18–21 years	115	217	53.0	15	30	50.0	100	187	53.5	0.72
Black race	189	217	87.1	27	30	90.0	162	187	86.6	0.61
Hormonal use	78	209	37.3	5	28	17.9	73	181	40.3	0.02
Multiple partners	189	202	93.6	24	26	92.3	165	176	93.8	0.78
Condom use	79	212	37.3	11	28	39.3	68	184	37.0	0.81
Vaginal Symptoms	160	217	73.7	25	30	83.3	135	187	72.2	0.20
NG positive	21	217	9.7	4	30	13.3	17	187	9.1	0.47
CT positive	49	217	22.6	9	30	30.0	40	187	21.4	0.30
POC TV positive	45	217	20.7	7	30	23.3	38	187	20.3	0.71
Wet Mount TV positive	21	217	9.7	5	30	16.7	16	187	8.6	0.16
Amine positive	109	210	51.9	15	29	51.7	94	181	51.9	0.98
Clue Cells >20%	71	215	33.0	7	30	23.3	64	185	34.6	0.20
pH >4.5	118	211	55.9	23	30	76.7	95	181	52.5	0.01
Clinical BV	39	209	18.7	3	30	10.0	36	179	20.1	0.19
Sialidase positive	22	105	21.0	3	15	20.0	19	90	21.1	0.92
WBC >6 / hpf	43	167	25.7	7	23	30.4	36	144	25.0	0.58

In Table 1, subjects with MG were significantly more likely to have a vaginal pH > 4.5 than those without MG (77% vs. 53%, p= 0.01). Hormonal contraceptives were reported by 40% of patients without MG compared with 18% of those with MG (p= 0.02). No other factors were found to be significantly associated with MG; therefore, we assessed possible confounders of vaginal pH and hormonal contraception. In Table 2, we show that variables positively associated with a vaginal pH > 4.5 were MG, a POC TV test, wet mount positive for TV, >20% clue cells, amines and sialidase, all at p<.05. Hormonal contraception users were less likely to be sialidase positive (11.6% versus 27.6%, p=.01) or infected with MG (6.4% vs. 17.6%, p=.02) than non-users. There was a trend toward an association between no hormone use and either amines or wet mount TV at p=.09 (Table 3).

Table 2.

Associations between vaginal pH and other variables for 211 subjects with pH testing results. Totals differ due to missing data. Data only shown for variables associated at p≤0.35. Significance (P Value) assessed with chi-squared testing.

	Total n=211			pH ≤4.5 Normal n=93			pH > 4.5 Abnormal n=118			P value
	Number with factor	Number tested	%	Number with factor	Number tested	%	Number with factor	Number tested	%
Hormonal use	78	204	38.2	39	93	41.9	39	111	35.1	0.32
MG positive	30	211	14.2	7	93	7.5	23	118	19.5	0.01
NG positive	21	211	10.0	6	93	6.5	15	118	12.7	0.13
CT positive	49	211	23.2	18	93	19.4	31	118	26.3	0.24
POC TV positive	45	211	21.3	14	93	15.1	31	118	26.3	0.05
Wet Mount TV positive	21	211	10.0	4	93	4.3	17	118	14.4	0.02
Amine positive	108	206	52.4	32	91	35.2	76	115	66.1	<.001
Clue Cells >20%	69	209	33.0	23	93	24.7	46	116	39.7	0.02
Sialidase positive	22	104	21.2	3	49	6.1	19	55	34.5	<.001
WBC >6 / hpf	42	165	25.5	17	81	21.0	25	84	29.8	0.20

Table 3.

Associations between hormonal contraception use and other variables for 209 subjects with Hormonal contraception information recorded. Totals vary due to missing data. Data only shown for variables associated at p≤0.35. Significance (P Value) assessed with chi-squared testing.

	Total N=209			Hormone Use N=78			No Hormone Use N=131			P value
	Number with factor	Number tested	%	Number with factor	Number tested	%	Number with factor	Number tested	%
Age, 18–21 years	110	209	52.6	36	78	46.2	74	131	56.5	0.15
Black race	183	209	87.6	65	78	83.3	118	131	90.1	0.15
MG positive	28	209	13.4	5	78	6.4	23	131	17.6	0.02
NG positive	19	209	9.1	4	78	5.1	15	131	11.5	0.12
CT positive	48	209	23.0	15	78	19.2	33	131	25.2	0.32
POC TV positive	44	209	21.1	13	78	16.7	31	131	23.7	0.23
Wet Mount TV positive	20	209	9.6	4	78	5.1	16	131	12.2	0.09
Amine positive	105	203	51.7	34	77	44.2	71	126	56.3	0.09
pH >4.5	111	204	54.4	39	78	50.0	72	126	57.1	0.32
Sialidase positive	21	101	20.8	5	43	11.6	16	58	27.6	0.05

Because the individual markers (pH, amines and clue cells) that comprised the variable “clinical BV” had independent associations with either hormonal contraception or MG, we used the individual markers rather than the composite variable “clinical BV” in the logistic regression model. This model demonstrated that an abnormal vaginal pH > 4.5 was significantly associated with an increased odds of MG [Odds Ratio (OR) 4.4, 95% confidence interval (CI) 1.1–17.5], while both hormonal contraceptive use and amines were protective against MG, even when controlling for sialidase and trichomoniasis (Table 4).

Table 4.

Multivariable logistic regression model for predictors of M. genitalium infection. N=99 subjects with complete data for each variable

	Odds Ratio	95% Conf. Interval	p
pH >4.5	4.37	1.08–17.7	0.04
Hormonal Contraception use	0.23	0.05–1.00	0.05
Amine positive	0.19	0.04–0.88	0.03
Sialidase positive	0.80	0.15–4.38	0.80
POC TV positive	2.46	0.50–12.2	0.27

Because both BV and TV were both associated with pH > 4.5, we analyzed the subset (n=131) of subjects who had no evidence of either clinical BV or TV. In these subjects, 13 of 52 (25%) with an elevated vaginal pH (> 4.5) tested positive for MG, whereas only 7 of 79 (9%) with a normal vaginal pH were positive for MG (p = 0.02, Fisher’s exact test).

Discussion

Vaginal POC tests help to indicate when MG is present. An elevated vaginal pH is associated with MG, even when controlling for confounding variables such as BV, TV, sialidase, amines, and hormonal contraception use. Women with amines appear to have a lower likelihood of MG infection than those without amines. In women without clinical BV or TV, one quarter of those with an elevated vaginal pH were infected with MG. Vaginal pH and amine tests are inexpensive, widely available tools that may potentially help clinicians estimate which patients are at risk for MG. In the absence of standardized, approved NAAT diagnostic tests for MG, information from vaginal pH and amine tests may help direct the care of women with vaginitis, cervicitis or PID who test negative for BV, TV, CT and NG.

This study reinforces some previous findings about Mycoplasma genitalium. First, MG is as common as other STIs in adolescents and young adult women, with prevalence falling between that of CT and NG. This is similar to prior cross sectional and longitudinal studies in young women.^11,28 Hormonal contraception use appears to be protective against MG.¹¹ Women on hormonal contraception may be older, and less likely to have any STI, amines or sialidase positive results (Table 2), but even when controlling for these variables in logistic regression, the association remains significant. Hormones may also change the vaginal environment by making it less hospitable for MG acquisition, or altering local immune factors.^29,30

In our study, we found a trend for women with clinical BV to have lower prevalence of MG than those without BV (10% vs. 20%). Similarly, Manhart et al. studied 719 adult women (age 16–45 years) and found that MG was lower in those with BV (defined by Amsel’s criteria) (4.2%) than without BV (8.2%). ¹ We hypothesize that because MG is positively associated with vaginal pH and negatively associated with amines, using both pH and amines to define BV may obscure the association, and this was confirmed in our logistic regression model. We could find no other published studies that evaluated MG with relationship to vaginal markers such as pH, amines, POC TV tests or sialidase.

A limitation to our study is that the cross-sectional design limits our ability to interpret the associations we found. We cannot determine whether MG causes an elevated vaginal pH or if women with high vaginal pH are more susceptible to MG infection. Elevated pH may represent an impaired local immune response or altered vaginal environment that allows MG to persist. The apparent protective effect of amines is also difficult to interpret. Finally, the lack of diversity in our population limits our ability to detect racial differences or to generalize our results to other populations.

Conclusion

Mycoplasma genitalium is as common in adolescent women as other STIs. Women with high vaginal pH may benefit from additional testing if standardized, approved diagnostic tests for MG are available. Longitudinal studies are needed to confirm our findings and to further evaluate whether the MG infection results in elevated vaginal pH, or if the presence of amines or a normal pH reduce susceptibility to MG.