The influence of bacterial vaginosis on in-vitro fertilization and embryo implantation during assisted reproduction treatment

N.H. Liversedge1,4, A. Turner2, P.J. Horner3, S.D. Keay1, J.M. Jenkins1,5 and M.G.R. Hull1

1 Department of Obstetrics and Gynaecology, St Michael's Hospital, Southwell Street, Bristol BS2 8EG, 2 Public Health Laboratory, Bristol Royal Infirmary and 3 The Milne Centre, Bristol Royal Infirmary, Bristol BS2 8HW, UK


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
There is growing evidence that the pathogenic effects of bacterial vaginosis may not be confined to the lower genital tract. Possible associations with infertility and effects on fertilization and implantation were studied in patients undergoing in-vitro fertilization (IVF) treatment. High vaginal swabs taken at the time of oocyte collection were assessed by Gram staining. The prevalence of bacterial vaginosis and of intermediate and normal flora in 301 patients was 25.6, 14.0 and 60.4% respectively. Bacterial vaginosis was more prevalent in patients with tubal (31.5%, n = 149) compared with non-tubal (19.7%, n = 152) infertility (odds ratio (OR) 1.87, CI 1.11–3.18, P = 0.02). Bacterial vaginosis did not have an adverse effect on fertilization rate. Further, no significant difference in implantation rates was seen when comparing bacterial vaginosis (15.8%, OR 1.03, CI 0.66–1.61) and intermediate flora (13.1%, OR 0.82, CI 0.45–1.52) with normal flora (15.5%). Though confidence intervals around the observations were relatively wide, the findings suggest that routine screening for bacterial vaginosis in the hope of improving the success of IVF treatment is not justified. The prevention of complications in pregnancy associated with bacterial vaginosis might be a more relevant indication for screening at the time of IVF treatment, in particular patients with tubal disease, if treatment were shown to be effective for that particular purpose. However, antibiotic treatment before IVF has been shown to be positively disadvantageous for IVF by encouraging other organisms.

Key words: bacterial vaginosis/embryo implantation/Gram stain/in-vitro fertilization


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Bacterial vaginosis represents a common imbalance of the vaginal bacterial flora but there is growing evidence of associated pathogenic effects. Increased rates of late miscarriage when diagnosed in early pregnancy, premature rupture of the membranes, endometritis, preterm labour and delivery have been reported (Hay et al., 1994aGo; McCoy et al., 1995Go). Korn and co-workers showed that almost half the patients they studied with symptomatic bacterial vaginosis in the absence of other vaginal or cervical infections had histopathological evidence of plasma cell endometritis (Korn et al., 1995Go). Isolation from the vagina of Gardnerella vaginalis, a pathogen frequently associated with bacterial vaginosis, has been associated with irregular uterine bleeding and histological endometritis (Kristiansen et al., 1987Go; Moller et al., 1995Go). These observations suggest that the effects of bacterial vaginosis are not confined to the lower genital tract and may therefore interfere with fertility through deleterious effects on fertilization and implantation. A prospective observational study of patients undergoing in-vitro fertilization (IVF) treatment was designed to determine firstly the prevalence of bacterial vaginosis in such an infertile population, and secondly whether bacterial vaginosis detected at the time of oocyte collection is associated with any reduction in fertilization or the chance of resulting embryos to implant.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
With approval of the local Research Ethics Committee all patients having IVF treatment in the Bristol University IVF Service between March and December 1996 were invited to participate in the study for one cycle of treatment. Immediately prior to transvaginal oocyte collection, a sterile vaginal speculum was passed and sterile cotton swab used to obtain a sample of posterior vaginal fornix fluid. The swab was rolled on to a labelled, clean glass microscope slide and the bacterial smear was fixed with acetone and air-dried.

The slides were Gram-stained in weekly batches, examined microscopically under oil immersion at x1000 magnification, and reported according to the criteria described by Hay et al., namely; grade 1, normal flora (predominantly lactobacillus morphotypes); grade 2, intermediate flora (mixed lactobacillus and other morphotypes); and grade 3, bacterial vaginosis (few or absent lactobacillus morphotypes but greatly increased number of G. vaginalis and other bacterial morphotypes) (Hay et al., 1994bGo). The samples were examined by two microscopists (N.H.L. and A.T.) who were blind to each other's results and the clinical information. Discordant findings were reviewed to reach agreement.

Patients completed a short questionnaire on the day of oocyte collection regarding factors for which we could not control and which were thought likely to influence vaginal bacterial flora, namely antibiotic administration within the preceding 3 months, type of sanitary wear used, and whether or not vaginal douching was practised.

Standard methods of IVF treatment were used as described previously (Hull et al., 1992Go). Intracytoplasmic sperm injection (ICSI) was used for patients with infertility due to sperm dysfunction (Hull et al., 1992Go). All patients underwent initial pituitary desensitization by nasal buserelin (Suprefact, Hoechst, Hounslow, UK) in daily divided doses from the mid-luteal phase of the previous cycle two days after commencing norethisterone 5 mg twice daily for 7 days. Ovarian stimulation was achieved using mainly urinary follicle-stimulating hormone (Metrodin or Metrodin HP; Serono, Welwyn Garden City, UK). Ovarian follicular development was monitored by vaginal ultrasonography and serum oestradiol measurement. Vaginal ultrasound-guided oocyte collection was performed 36 h after administering 5000 IU human chorionic gonadotrophin (HCG; Profasi, Serono). Luteal support was provided by two further injections of 2000 IU HCG, 2 and 7 days following oocyte collection. Normal fertilization was defined by formation of two pronuclei and progressive cleavage up to the time of embryo transfer. After IVF up to three embryos (maximum) were transferred to the uterus. Pregnancy was diagnosed by transvaginal ultrasound evidence of a gestation sac after 4 weeks. The implantation rate in each patient group studied was defined as the total number of gestation sacs as a proportion of the total number of embryos transferred. Pregnancies were followed and miscarriage and birth rates, as well as babies as a proportion of individual embryos transferred, were calculated.

Statistical analysis
The results were compared statistically using Fisher's exact and {chi}2 tests. A weighted kappa coefficient test was employed to test the concordance of results for each grade of flora between assessors and to allow for disagreement. The statistical software employed was Arcus Pro-Stat Statistical Analysis System, version 3.23 (Medical Computing, Aughton, UK).

To determine study size, initially a power analysis was performed which assumed a 12.5% prevalence of bacterial vaginosis in the IVF population with an implantation rate of 13% in unaffected patients and 5% with bacterial vaginosis. To achieve 80% power with 5% type I error, uncorrected {chi}2 analysis required 123 embryos transferred to patients with bacterial vaginosis and 984 embryos to be transferred to patients without bacterial vaginosis. Assuming an average of 2.5 embryos per transfer, 443 patients were required. The power calculation was adjusted after 251 patients had been recruited, when the prevalence of bacterial vaginosis proved much higher than expected (25.8%) and more embryos were transferred each cycle (2.7), requiring 203 cycles with 553 embryos transferred.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Of 458 eligible patients who were approached, 302 (66%) consented to be included in the study. Of those, 287 patients had an interpretable bacterial smear and a cycle resulting in embryo transfer. One vaginal smear was inadequate for interpretation and 14 patients (4.7%) failed to achieve embryos but there was no case of fertilization failure due to infection of the embryo culture system.

In the 301 interpretable slides, bacterial vaginosis, intermediate flora, and normal flora were found in 77 (25.6%), 42 (14.0%), and 182 (60.4%) respectively. The weighted kappa coefficient for the two assessors was 0.880 (95% CI 0.65–1.11), representing good concordance of reported results.

The frequency distributions of the woman's age and method of insemination utilized related to the grades of vaginal bacterial flora were similar. There was, however, a significantly greater proportion of patients having standard IVF rather than ICSI in patients with tubal compared with non-tubal infertility (148/149 versus 69/152, P < 0.0001). The prevalence of the different grades of vaginal bacterial flora related to infertility category is shown in Table IGo. Bacterial vaginosis was significantly more frequent in patients with tubal disease (P = 0.02) and a normal flora was significantly more frequent in the non-tubal group (P = 0.004).


View this table:
[in this window]
[in a new window]
 
Table I. Grades of vaginal bacterial flora related to the woman's age, method of insemination, and infertility diagnosis (percentage frequencies in parenthesis except where specified)
 
The clinical outcomes for oocyte collection are shown in Table IIGo. There was a significantly higher fertilization rate in patients with bacterial vaginosis (58%, OR 1.29, CI 1.09–1.53, P = 0.004) but not intermediate flora (53%, OR 1.09, CI 0.89–1.34, NS) compared with normal flora (51%).


View this table:
[in this window]
[in a new window]
 
Table II. Clinical outcomes in the 301 patients having an oocyte collection for IVF, related to grades of vaginal bacterial flora
 
Implantation rates have been further analysed in Table IIIGo comparing tubal and non-tubal infertility groups, and tubal subgroups according to the presence or absence of a hydrosalpinx. No significant differences were observed.


View this table:
[in this window]
[in a new window]
 
Table III. Implantation, pregnancy and birth rates of 287 patients having an embryo transfer related to tubal state and grade of vaginal bacterial flora
 
Of 301 patients, 251 patients (83.4%) completed and returned the questionnaire relating to behavioural factors which possibly influenced vaginal bacterial flora, namely antibiotic administration within the preceding 3 months, type of sanitary wear used and vaginal douching. The results are shown in Table IVGo. No significant differences were observed between the different grades of vaginal bacterial flora.


View this table:
[in this window]
[in a new window]
 
Table IV. Prior behavioural factors related to grades of vaginal bacterial flora (responses to questionnaire) (percentages in parenthesis)
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
This study assessed possible associations of bacterial vaginosis with infertility in women undergoing IVF treatment. In this study the prevalence of bacterial vaginosis (25.6%) was twice that found in patients attending obstetric (12%) and gynaecology (11%) clinics in the UK, studied using similar methods and criteria to define the type of vaginal bacterial flora (Hay et al., 1992Go; Hay et al., 1994bGo). Further, the prevalence was nearly twice as high in the patients with tubal damage (31.5%) compared with non-tubal infertility (19.7%).

Tubal factor infertility is primarily the result of pelvic inflammatory disease (Sweet, 1995Go) and bacterial vaginosis has been associated with pelvic inflammatory disease (Soper et al., 1994Go). The natural history of bacterial vaginosis however, remains unclear. Whilst two observational studies suggest that bacterial vaginosis remits in the majority of cases within 12 months (Bump et al., 1984Go; Bump and Buesching, 1988Go), this is not supported by findings from placebo groups in randomized control trials (Boeke et al., 1993Go; Neri et al., 1993Go). Hay et al., in a longitudinal study of bacterial vaginosis during pregnancy, observed that in approximately 50% of cases a normal lactobacilli-dominated flora would regenerate (Hay et al., 1994bGo). Moreover, women with a past history of bacterial vaginosis following treatment are at increased risk of a further episode. The data from this study suggest that women with tubal factor infertility may have an inherent predisposition to abnormal genital tract bacterial flora associated with bacterial vaginosis. It is recognized that anaerobic bacteria are often present during acute pelvic inflammatory disease (Paavonen et al., 1987Go). It is therefore feasible that these organisms may be an aetiological factor in the development of tubal factor infertility following pelvic inflammatory disease. The results of this study are not inconsistent with this. However, this may merely be a reflection of the previously stated association between tubal factor infertility and pelvic inflammatory disease. The prevalence of bacterial vaginosis in non-tubal infertility patients was also higher than in reference populations (Hay et al., 1992Go, 1994bGo). Intermediate flora, a transitory state in progression from normal flora to bacterial vaginosis or resolution of bacterial vaginosis back to normal flora (Hillier et al., 1992Go; Hay et al., 1994bGo), in both tubal (16%) and non-tubal infertility (12%) was higher than reported in an antenatal population (5%: Hay et al., 1994bGo). These data show that in our study, predisposition to an abnormal vaginal bacterial flora was a common finding in infertile patients though with greater frequency in those with a tubal factor.

This study does not support the hypothesis that bacterial vaginosis adversely affects embryo implantation and early pregnancy outcome. Bacterial vaginosis has been associated with endometritis (Korn et al., 1995Go) and pre-term labour. The mechanism remains to be elucidated but is believed to involve the induction of upper genital tract inflammation. It is therefore perhaps surprising that bacterial vaginosis was not associated with a reduction in implantation rate. Further, the risk of first trimester spontaneous abortion was not significantly related to the grade of vaginal bacterial flora reported at oocyte collection, though numbers in this study were small. If there was an increased first trimester spontaneous abortion rate associated with the presence of bacterial vaginosis, one would expect the abortion rate to be higher in the tubal compared with non-tubal factor infertility patients in view of the significantly higher prevalence of bacterial vaginosis in tubal factor patients. Analysis of the Bristol University IVF Service database between January 1988 and June 1995 for all women <40 years of age revealed no significant difference in the spontaneous first trimester abortion rate per clinical pregnancy in women with tubal infertility (24/225 = 10.7%) compared with non-tubal (21/268 = 7.8%, OR 1.4, 95% CI 0.76–2.60, P = 0.28). This would suggest that the mechanisms involved in reduced implantation rates and first trimester abortion differ from those associated with pre-term delivery. Indeed, bacterial vaginosis is more prevalent in black afro-Caribbean and African women (Mead, 1993Go), yet there is no evidence to suggest that they as a group have either increased conceptual difficulties or first trimester abortion rates compared to Caucasians. Our current understanding of bacterial vaginosis would lead one to postulate that women with bacterial vaginosis have an increased incidence of endometritis. Whilst we do not test for endometritis directly, this is not an unreasonable assumption and would suggest that low-grade endometritis does not influence the early stages of pregnancy. However, there are other possible explanations as to why we found no association.

Bacterial vaginosis has been associated with Chlamydia trachomatis whose coincidental presence may have acted as a confounding variable. However, if there had been significant presence of Chlamydia trachomatis one would have expected it to bias the results in favour of an adverse effect upon IVF outcome, and this was not seen. Although patients in this study were not routinely screened for the presence of C. trachomatis in the lower genital tract, serology was performed using whole inclusion immunofluorescence (WIIF) testing. This is a good predictor of upper genital tract chlamydial inflammation (Conway et al., 1984Go; Chernesky et al., 1998Go). All patients who gave positive results in the WIIF test were treated with ofloxacin. Because bacterial vaginosis is associated with tubal factor infertility it is possible that prior treatment with ofloxacin was over-represented in women with bacterial vaginosis and this might have biased the results. An association has been found between bacterial vaginosis, prematurity, and vaginal fluid mucinase and sialidase activity (McGregor et al., 1994Go). They speculated that these enzymes may facilitate mucus penetration and epithelial attachment by both the enzyme-producing bacteria and co-existing bacteria. This hypothesis is supported by the work of Howe and colleagues, who have also found an association between such enzymatic activity in the vaginal fluid and bacterial vaginosis (Howe et al., in press). It is possible therefore that bacterial vaginosis might act as a co-factor in the development of endometritis by enhancing the pathogenicity of co-existing bacteria. Ofloxacin administration may have altered the natural history of the co-pathogens disease process and thus masked a possible link between bacterial vaginosis and impaired implantation.

Patients were tested for bacterial vaginosis at the time of oocyte collection when supraphysiological concentrations of oestrogen were circulating. Whether this affects the vaginal flora is unknown. It has been observed that in the hyperoestrogenic pregnant state the concentration of vaginal lactobacilli is higher than in the non-pregnant state (Larsen and Galask, 1982Go). However, in pregnancy there is an excess of progesterone, unlike in the pre-ovulatory phase. Therefore the two states are not comparable. If supraphysiological concentrations of oestrogen in the absence of progesterone predispose to bacterial vaginosis, this could explain the high incidence observed in this study at the time of sampling. With subsequent production of progesterone there might be a tendency to regeneration of the normal lactobacillus-dominated flora in women who had bacterial vaginosis at the time of oocyte collection. In addition, disturbance of the vagina may lead to alteration of the vaginal bacterial flora. Transvaginal ultrasound probing to monitor ovarian stimulation for IVF may have had such an effect. The study patients had a median of three vaginal ultrasound scans in the 10 days prior to oocyte collection, though only a small amount of neutral pH transducer gel was used and it was felt unlikely to affect the vaginal pH and hence bacterial flora. It also seems unlikely that the probing would disturb the flora any more than coitus, which in most couples is maintained during the treatment cycle. Vaginal douching was rarely practised by subjects and the frequency was not significantly different between the groups of different vaginal bacterial flora. Further, the grade of bacterial flora could not be related to the type of sanitary wear used, whether external or internal.

Fertilization rates were significantly higher in women with bacterial vaginosis. However, this may be due to the significantly greater proportion with tubal infertility, in which fertilization rates are more likely to be favourable compared with other infertility causes. Further, there was a significantly higher proportion of patients with tubal infertility having standard IVF (99%) rather than ICSI compared with patients with non-tubal infertility (45.4%).

These discussion points may explain the increased prevalence of bacterial vaginosis in this population compared with other studies and may potentially explain why we found no association between bacterial vaginosis at the time of oocyte collection and adverse outcome during IVF embryo transfer treatment. We believe that these points merit further investigation. However with regard to present clinical practice, since no detrimental effect on IVF outcome has been seen secondary to the presence of bacterial vaginosis, screening and treating women prior to IVF would appear to be unnecessary. Further, antibiotic therapy may increase the likelihood of inoculation of antibiotic-resistant pathogenic bacteria from the vagina into the embryo culture system during vaginal oocyte collection (Liversedge et al., 1996Go).

In summary, this study questions the assumption that bacterial vaginosis, if present at the time of oocyte recovery during IVF treatment, adversely affects fertilization and implantation. Although the prevalence of bacterial vaginosis was much higher in infertile patients having IVF treatment than found by others in antenatal and general gynaecological populations, no significant effect upon fertilization and implantation rates was found. Thus, although routine screening and treatment for bacterial vaginosis before IVF treatment would appear not to be helpful, there remains a possibility that treatment of bacterial vaginosis could reduce later complications in pregnancy. If such treatment was shown to be beneficial, then patients with tubal infertility having IVF treatment should be considered a target group for antenatal screening.


    Acknowledgments
 
The authors would like to thank the medical, nursing and clerical staff of the University of Bristol IVF Service, and Mrs Alison Headford and Miss Jackie Ramscar of the Department of Pathology, St Michael's Hospital Bristol for their assistance in the completion of this study.


    Notes
 
4 Present address: Obstetrics and Gynaecology, Derriford Hospital, Plymouth PL6 8DH, UK Back

5 To whom correspondence should be addressed Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Boeke, A.J., Dekker, J.H., van-Eijk, J.T. et al. (1993). Effect of lactic acid suppositories compared with oral metronidazole and placebo in bacterial vaginosis: a randomised clinical trial. Genitourin. Med., 69, 388–392.[ISI][Medline]

Bump, R.C. and Buesching, W.J. (1988) Bacterial vaginosis in virginal and sexually active adolescent females: evidence against exclusive sexual transmission. Am. J. Obstet. Gynecol., 158, 935–939.[ISI][Medline]

Bump, R.C., Zuspan, F.P., Buesching, W.J. et al. (1984) The prevalence, six-month persistence, and predictive values of laboratory indicators of bacterial vaginosis (non-specific vaginitis) in asymptomatic women. Am. J. Obstet. Gynecol., 150, 917–924.[ISI][Medline]

Chernesky, M., Luinstra, K., Sellors, J. et al. (1998) Can serology diagnose upper genital tract Chlamydia trachomatis infections? Studies on women with pelvic pain, with or without chlamydial plasmid DNA in endometrial biopsy tissue. Sex. Trans. Dis., 25, 14–19.[ISI][Medline]

Conway, D., Glazener, C.M.A., Caul, E.O. et al. (1984) Chlamydia serology in fertile and infertile women. Lancet, 1, 191–193.[Medline]

Hay, P.E., Taylor-Robinson, D. and Lamont, R.F. (1992) Diagnosis of bacterial vaginosis in a gynaecology clinic. Br. J. Obstet. Gynaecol., 99, 63–66.[ISI][Medline]

Hay, P.E., Lamont, R.F., Taylor-Robinson, D. et al. (1994a) Abnormal bacterial colonisation of the genital tract and subsequent preterm delivery and late miscarriage. Br. Med. J., 308, 295–298.[Abstract/Free Full Text]

Hay, P.E., Morgan, D.J., Ison, C.A. et al. (1994b) A longitudinal study of bacterial vaginosis during pregnancy. Br. J. Obstet. Gynaecol., 101, 1048–1053.[ISI][Medline]

Hillier, S.L., Krohn, M.A., Nugent, R.P. et al. (1992) Characteristics of three vaginal flora patterns assessed by Gram stain among pregnant women (for Vaginal Infections and Prematurity study Group). Am. J. Obstet. Gynecol., 166, 938–944.[ISI][Medline]

Howe, L., Wiggins, R. and Millar, M.R. (in press – Int. J. Sex. Trans. Dis. AIDS).

Hull, M.G.R., Eddowes, H.A., Fahy, U. et al. (1992) Expectations of assisted conception for infertility. Br. Med. J., 304, 1465–1469.[ISI][Medline]

Korn, A.P., Bolan, G., Padian, N. et al. (1995). Plasma cell endometritis in women with symptomatic bacterial vaginosis. Obstet. Gynecol., 85, 387–390.[Abstract/Free Full Text]

Kristiansen, F.V., Oster, S., Frost, L. et al. (1987) Isolation of Gardnerella vaginalis in pure culture from the uterine cavity of patients with irregular bleedings. Br. J. Obstet. Gynaecol., 94, 979–984.[ISI][Medline]

Larsen, B. and Galask, P. (1982) Vaginal microbial flora: Composition and influences of host pathology. Ann. Intern. Med., 96, 926–930.[ISI][Medline]

Liversedge, N.H., Jenkins, J.M., Keay, S.D. et al. (1996) Antibiotic treatment based on seminal cultures from asymptomatic male partners in in-vitro fertilization is unnecessary and may be detrimental. Hum. Reprod., 11, 1227–1231.[Abstract]

McCoy, M.C., Katz, V.L., Kuller, J.A. et al. (1995) Bacterial vaginosis in pregnancy: an approach for the 1990's. Obstet. Gynaecol. Surv., 50, 482–488.[Medline]

McGregor, J.A., French, J.I., Jones, W. et al. (1994) Bacterial vaginosis is associated with prematurity and vaginal fluid mucinase and sialidase: Results of a controlled trial of topical clindamycin cream. Am. J. Obstet. Gynecol., 170, 1048–1060.[ISI][Medline]

Mead, P.B. (1993) Epidemiology of bacterial vaginosis. Am. J. Obstet. Gynecol., 169, 446–449.[ISI][Medline]

Moller, B.R., Kristiansen, F.V., Thorsen, P. et al. (1995) Sterility of the uterine cavity. Acta Obstet. Gynecol. Scand., 74, 216–219.[ISI][Medline]

Neri, A., Sabah, G. and Samra, Z. (1993) Bacterial vaginosis in pregnancy treated with yoghurt. Acta Obstet. Gynecol. Scand., 72, 17–19.[ISI][Medline]

Paavonen, J., Teisala, K., Heinonen, P.K. et al. (1987) Microbiological and histopathological findings in acute pelvic inflammatory disease. Br. J. Obstet. Gynaecol., 94, 454–460.[ISI][Medline]

Soper, D.E., Brockwell, N.J., Dalton, H.P. et al. (1994) Observations concerning the microbial etiology of acute salpingitis. Am. J. Obstet. Gynecol., 170, 1008–1014.[ISI][Medline]

Sweet, R.L. (1995) Role of bacterial vaginosis in pelvic inflammatory disease. Clin. Infect. Dis., 20, S271–275.[ISI][Medline]

Submitted on January 15, 1999; accepted on June 1, 1999.