1 Dr Janet Wilson Leeds Teaching Hospitals NHS Trust, UKBacterial vaginosis uncovered: Aetiology, complications and practical implications Dr Janet Wilson Leeds Teaching Hospitals NHS Trust, UK

2 Conflicts of Interest Conflicts of InterestI have received unconditional research grants in the form of diagnostic tests from Hologic/GenProbe I have received a speaker honorarium from BD Diagnostics I have participated in BV research studies sponsored and funded by Starpharma

3 Prevalence of bacterial vaginosisCommonest cause of abnormal discharge in women of child bearing age In UK: General practice 9% Antenatal clinic 15% Pre TOP clinic % Attending STI clinic % 50% women are asymptomatic (complications still)

4 What is BV – aetiology 1950s Gardner and Dukes discovery of Haemophilis vaginalis in 1955 They hypothesised that Gardnerella vaginalis was the specific sexually transmitted aetiological agent of BV Original experiments by Gardner and Dukes: 1/13 women inoculated with G. vaginalis developed BV 11/15 inoculated with vaginal material from infected women developed BV Gardner et al. Am J Obstet Gynecol 1955

5 Aetiology – later 20th century20th Century polymicrobial theory from bacterial cultures Overgrowth of Gardnerella vaginalis, anaerobic bacteria, Mycoplasma hominis, Ureaplasma urealyticum, and Mobiluncus species, which replace the lactobacilli Bacteria increased times G. vaginalis present in many women with normal flora and in virgins - not considered main cause or to be sexually transmitted Treatments covering anaerobic bacteria such as metronidazole and clindamycin work best Suggests more central role than Mobiluncus and mycoplasmas Symbiotic relationship between P. bivia and G. vaginalis Exact trigger factor and role of bacteria unknown If necessary can remove 3rd and 4th bullet points

6 Fluoresence in-situ hybridisation (FISH)Aetiology – 21st century 21st Century polymicrobial +++ theory from molecular techniques Using bacterial 16S ribosomal DNA amplification newly identified organisms associated with BV: Atopobium vaginae, Megasphaera, Sneathia, Leptotrichia, Clostridiales order – BVAB 1, 2, 3 Fredricks et al. NEJM 2005 Detection of Megasphaera, A. vaginae, BVAB-1, BVAB-3, highly sensitive and specific for BV Fredricks et al. J Clin Microbiol 2007 But no new prime aetiological bacteria identified – cause still unknown Fluoresence in-situ hybridisation (FISH)

7 Aetiology – 21st century 21st Century polymicrobial +++ theoryUsing bacterial 16S ribosomal DNA amplification different patterns in vaginal microbial diversity identified Geographical variation reflecting different ancestries In US women five vaginal microbiota groupings (termed community state types, CST) identified Four dominated by different lactobacilli species, fifth depleted of LB and resembled BV CST associated with ethnicity – African-American and Hispanic women more likely to be LB depleted Ravel et al. Proc Natl Acad Sci USA 2011 Heatmap showing distribution of bacteria in vaginal microbial communities of 394 US women Questions what are normal flora? Aetiology still not explained and appearing more complex!

8 Aetiology – 21st century Discovery of biofilms on vaginal biopsies40 women without BV, 20 women with BV Loosely dispersed bacteria (Lactobacilli and aerobes) in those without BV Dense biofilm of G. vaginalis in women with BV A. vaginae also present in 80% of biofilms Only G. vaginalis developed a biofilm specific to BV Swidsinski et al. Obstet Gynecol 2005 G. vaginalis is the adherent Gram-negative bacteria on clue cells Cook et al. J Infect Dis 1989

9 Aetiology of BV G. vaginalis either: Dispersed = no BVCohesive = BV and biofilm Women with: Dispersed Gardneralla: all normal vaginal flora Cohesive Gardnerella: 10 BV on Gram stain, 2 intermediate No Gardnerella: 2 BV, 2 intermediate, 48 normal flora Cohesive Gardnerella had sensitivity 83%, specificity 97% Swidsinski et al. Gynecol Obstet Invest 2010 G. vaginalis colonises 50-70% of women with normal vaginal flora in small amounts Srinivasan et al. PLoS One 2012 This is probably colonisation by dispersed G. vaginalis

10 Aetiology of BV Are dispersed and cohesive G. vaginalis different strains? G. vaginalis from BV versus G. vaginalis from normal flora: Better adherent to epithelial cells and cytotoxic (non-BV was not) and more aggregative and produced a thick biofilm Harwich et al. BMC Genomics 2010 Only BV-associated isolates able to adhere in high density clusters to HeLa cells Castro et al. Sci Rep 2015 BV associated G. vaginalis produce mucinase Yeoman et al. Plos One 2011 Only two genotypes of G. vaginalis produce sialidase Lopes Dos Santos Santiago et al. Am J Obstet Gynecol 2011 Conclusion: Differences in metabolic and virulence potential

11 Aetiology of BV Are dispersed and cohesive G. vaginalis different strains? G. vaginalis phenotypic heterogeneity known for years Sequence-based analyses identified 4 distinct molecular subgroups with high level of diversity - clades but possibly different species? Ahmed et al. J Bacteriol 2012 60 clinical samples from chronic vaginitis clinic: 94-97% were positive for G. vaginalis but only 37% had BV BV associated with clades 1 and 3; intermediate with clade 2 No association with BV and clade 4 Balashov et al. J Med Micro 2014

12 Aetiology of BV Are dispersed and cohesive G. vaginalis different strains? They probably are – possibly even different species Back to single pathogen hypothesis of 1955 Could pathogenic cohesive biofilm-forming G. vaginalis be responsible for BV? Int J STD AIDS 2008

13 Aetiology of BV Unlike other anaerobes G. vaginalis possesses the three key virulence determinants to cause BV: Adherence to vaginal epithelial cells Capacity to produce biofilms Cytotoxic activity Patterson et al. Microbiology 2010 Alves et al. J Infect Dis 2014 Single species starts process of biofilm scaffolding in the mouth to which other species adhere – probably same mechanism in the vagina Does cohesive G. vaginalis start the vaginal biofilm to which the other BVAB attach? Is this the sexually transmitted agent of BV?

14 New model of BV pathogenicityNormal vaginal flora Lactobacilli

15 New model of BV pathogenicitySexual transmission of cohesive (biofilm forming) G. vaginalis Normal vaginal flora Lactobacilli

16 New model of BV pathogenicitySexual transmission of cohesive (biofilm forming) G. vaginalis Adherence to epithelial cells Displacement of lactobacilli Normal vaginal flora Lactobacilli

17 Adherence and displacement of LBL. crispatus produces bacteriocins and core proteins - reduce ability of G. vaginalis to adhere to epithelial cells Ojala et al BMC Genomics 2014 G. vaginalis produces bacteriocins active against LB Teixeira et al. J Med Microbiol 2010 Cohesive G. vaginalis displaced L. crispatus from epithelial cells Castro et al. Int J Med Sci 2013 G. vaginalis most pathogenic of BVAB as defined by highest initial adhesion Patterson et al. J Infect Dis 2014 G. vaginalis greatest capacity of BVAB to adhere in presence of L. crispatus Machado et al. Int J Mol Sci 2013

18 New model of BV pathogenicitySexual transmission of cohesive (biofilm forming) G. vaginalis Adherence to epithelial cells Displacement of lactobacilli Normal vaginal flora Lactobacilli Replication of G. vaginalis and other BVAB

19 Replication and biofilm formationSymbiotic relationship between P. bivia and G. vaginalis Pybus et al. J Infect Dis 1997 Growth benefit to G. vaginalis from addition of any second species Machado et al. Int J Mol Sci 2013 G. vaginalis most pathogenic of BVAB as defined by greatest propensity to form a biofilm Patterson et al. J Infect Dis 2014 G. vaginalis biofilm can withstand 4-fold higher concentrations of lactic acid and hydogen peroxide than dispersed G. vaginalis Patterson et al. Am J Obstet Gynecol 2007

20 New model of BV pathogenicitySexual transmission of cohesive (biofilm forming) G. vaginalis Adherence to epithelial cells Displacement of lactobacilli Normal vaginal flora Lactobacilli Replication of G. vaginalis and other BVAB Formation of biofilm along with other BVAB Biofilm protects against lactic acid and H2O2

21 New model of BV pathogenicitySexual transmission of cohesive (biofilm forming) G. vaginalis Adherence to epithelial cells Displacement of lactobacilli Normal vaginal flora Lactobacilli Replication of G. vaginalis and other BVAB Established BV Formation of biofilm along with other BVAB Biofilm protects against lactic acid and H2O2

22 New model of BV pathogenicitySexual transmission of cohesive (biofilm forming) G. vaginalis Adherence to epithelial cells Displacement of lactobacilli Normal vaginal flora Lactobacilli Replication of G. vaginalis and other BVAB Established BV Treatment Formation of biofilm along with other BVAB Biofilm protects against lactic acid and H2O2

23 New model of BV pathogenicitySexual transmission of cohesive (biofilm forming) G. vaginalis Adherence to epithelial cells Displacement of lactobacilli Normal vaginal flora Lactobacilli Replication of G. vaginalis and other BVAB Established BV Treatments have high failure and recurrence rates Formation of biofilm along with other BVAB Biofilm protects against lactic acid and H2O2

24 Complications of BV High rate of recurrence -is it sexually transmitted? STI and HIV acquisition/transmission Association with PID Pregnancy complications

25 Frequency of BV recurrence121 women successfully treated with oral metronidazole. Follow up 12 months Recurrence rates: 23% 1 month 43% 3 months 58% 12 months Bradshaw et al. J Inf Dis 2006

26 BV and sexual transmission?Recurrence 2-3 fold higher if same regular sex partner after treatment compared with change in partner Bradshaw et al. J Inf Dis 2006 Male contacts of women who developed BV had significantly more female partners in last 30 days than those without BV Schwebke et al. Sex Transm Dis 2005 Consistent condom use strongly protective against BV Ad OR 0.37 (95% CI ) Hutchinson et al. Epidemiology 2007 No prevalent or incident BV in young women without coital or non-coital sexual experience Fethers et al. J infect Dis 2009 50 premenarche girls: 10% had dispersed G. vaginalis; none had cohesive (biofilm associated) Swidsinki et al. Gyneco Obstet Invest 2010

27 Evidence of transmission in WSWOriginal experiments by Gardner and Dukes: BV occurred in 8% inoculated with G. vaginalis but in 73% inoculated with vaginal material from infected women In WSW significant concordance of vaginal flora in couples 43% Berger 1995 95% Marrazzo 2002 87% Evans 2008 BV associated with partner with BV and sharing sex toys in WSW ie practices that transmit vaginal fluid Marrazzo et al. Sex Transm Dis 2010

28 Concordance of G. vaginalis in partnersGardner and Dukes isolated G. vaginalis from urethra of 96% of male partners of women with BV Vaginal cultures from women with BV, urethral culture (within 24 hours) from male partners - G. vaginalis biotypes identical in 92% Piot et al, J Clin Microbiol 1984 Molecular typing of G. vaginalis showed male/female sexual partners shared the same strains Eren et al. PLoS One 2011 All male partners of women with cohesive G. vaginalis had cohesive G. vaginalis No concordance in male partners of women with dispersed G. vaginalis Swidsinski et al. Gynecol Obstet Invest 2010 Suggests cohesive G. vaginalis is the sexually transmitted agent of BV

29 BV and acquisition of STIsIncreased acquisition with BV: GC and CT fold Brotman et al. J Infect Dis 2010 TV 9-fold Rathod et al. Sex Transm Dis 2011 HSV fold Cherpes et al. Clin Inf Dis 2003 HSV-2 shedding 2.3 fold Cherpes et al. Clin Inf Dis 2005 Treatment then prophylaxis asymptomatic BV versus observation Prophylaxis 1.58 (95% CI ) ppys Observation 2.29 (95% CI ) ppys Schwebke et al. Am J Obstet Gynecol 2007 Treatment then RCT monthly metronidazole gel or placebo Almost 50% reduction of CT/NG/MG over 12 months with monthly presumptive treatment Balkus et al. JID 2016

30 BV and HIV acquisition Meta-analysis of 23 studies; 30 739 womenHIV incidence with BV RR 1.6 (95% CI ) - low HIV risk RR 2.3 (95% CI ) - high HIV risk RR 1.4 (95% CI ) BV prevalence 30% with RR 1.6; Population attributable risk 15% Atashili et al. AIDS 2008 HIV positive women: Prevalence of BV OR 1.29 ( ) Persistence of BV OR ( ) Immunocompromised women had more severe BV on Gram-stain OR ( ) Jamieson et al. Obstet Gynecol 2001

31 BV and HIV transmissionHeterosexual transmission Prospective study of 2236 HIV-1+ve women and uninfected male partners (part of large HIV/HSV trial) 50 incident infections (sequencing proven transmissions) HIV-1 incidence with BV /100 person years HIV-1 incidence normal flora 0.76/100 person years Adjusted Hazard ratio (95% CI ) Cohen et al. PLoS Medicine 2012 Vertical transmission 16S ribosomal DNA amplification of vaginal secretions from 10 transmitters and 54 non-transmitters MTCT of HIV associated with altered vaginal microbiota Gardnerella vaginalis significantly associated with HIV MTC transmission - Adjusted OR 1.7; P = 0.004 Frank et al. J Acquir Immune Defic Synr 2012

32 BV and pelvic inflammatory diseaseCross-sectional studies have reported significant association between PID and BV but cause or effect? Cervical and endometrial samples for BVAB from 545 women enrolled in PID Evaluation and Clinical Health (PEACH) study (no metronidazole given) A. vaginae, BVAB1, Sneathia sanguinegens, S. amnionii significantly associated with BV Endometritis at presentation: All BVAB Ad OR 2.0 ( ) Endometritis 30 days after treatment: All BVAB Ad OR 5.7 ( ) (Excluding women with CT and NG) All BVAB Ad OR 8.5 ( ) Recurrent PID: All BVAB Ad OR 3.9 ( ) Infertility: All BVAB Ad OR 3.8 ( ) Antibiotic sensitivities of many of BVAB not known but supports including cover against anaerobes in PID treatment Haggerty et al. Sex Transm Infect 2016

33 BV and pregnancy Cohort studies Meta-analysis of 18 studies; women Preterm birth Overall risk ( ) BV <16 weeks ( ) BV <20 weeks ( ) Spontaneous abortion Overall risk ( ) Postnatal infection Overall risk ( ) Leitich et al. Am J Obstet Gynecol 2003 BV common infection; population attributable risk of PTB due to BV estimated to be 30% in USA at cost $1 billion Koumans et al. Sex Transm Dis 2001

34 Practical Implications Treatment of BV 20th & 21st CenturiesNo new additions to treatment guidelines for years Metronidazole Oral 500mg bd for 7 days Vaginal gel 5gm for 5 days Clindamycin Vaginal cream 5gm for 7 days Clindamycin Oral 300mg bd for 7 days Tinidazole Oral 2g daily for 2 days Oral 1g daily for 5 days Clindamycin and metronidazole same efficacy irrespective of regimen Oduyebo et al. Cochrane Reviews 2009 14 days metronidazole same relapse rate as 7 days Schwebke et al. Clin Inf Dis 2007 Clindamycin + metronidazole cure/recurrence rates same as metronidazole alone Bradshaw et al. PLoS One 2012 Cure rates at one month 71-88%

35 Bacterial Response to TreatmentWith successful treatment 3-4 log reduction of Gardnerella and BVAB. No significant change in those not cured Fredricks et al. J Clin Microbiol 2009 Treatment RCT of clindamycin versus metronidazole Treatment failures ≤1 month retreated with same treatment - clinical cure rate equal to primary responder Suggests antimicrobial resistance is not the problem Bunge et al. Sex Transm Dis 2009

36 Treatment trials of male partners5/6 RCTs of treatment of male sexual partner - no improvement in cure rate, or reduction in recurrence rate 1/6 short term (<8 weeks) improvement with metronidazole Potter J. Br J Gen Pract 1999 All 6 had significant flaws 5/6 published before 1996 CONSORT statement, other in 1997 Eligibility criteria for ‘partners’ not given or duration of sex Power calculations inadequate; Insufficient randomisation methods Treatment adherence in females and males not reported 5/6 treatment suboptimal metronidazole 2gm stat - 1 study metronidazole 2gm stat day studies tinidazole 2gm stat - 1 study clindamycin 2% vag cream with oral clind for 7/7 Mehta, Sex Transm Dis 2012 Absence of evidence of effect is not the same as evidence of absence of effect

37 Bacterial Response to Treatment and BiofilmsBiofilms good mechanisms for avoiding Natural defences, antimicrobials and Antibiotics In vivo Metronidazole temporarily suppressed biofilm activity during treatment but activity quickly restarted after treatment LB gradually fill vacuum formed during treatment G. vaginalis cleared more slowly then pure anaerobes and levels often rebound during treatment BVAB can quickly re-emerge when treatment is stopped Mayer et al. J Infect Dis 2015 Biofilm persists on vaginal epithelium following lack of response to treatment Swidsinski et al. Am J Ob Gyn 2008

38 Attempts to disrupt the BV biofilmMany substances tried but either not effective or not suitable for use in humans so far In vivo Metronidazole Moxifloxacin Intravaginal lactic acid gel Intravaginal boric acid Octenidine (topical antiseptic used in oral biofilms) In vitro Synthetic retrocyclin (primate defensin not produced by humans) In mice DNase

39 Conclusions and where next?Bacterial vaginosis is a common condition Associated with serious sequelae – increased acquisition of STIs and HIV, associated with PID and adverse outcomes in pregnancy Cohesive G. vaginalis is probably the main aetiological agent and is possibly sexually transmitted It develops a biofilm to which the other BVAB attach The biofilm offers protection to the bacteria and standard antibiotic treatment is not always successful resulting in BV recurrence We need to find effective and safe BV biofilm-breaking treatment regimens to improve treatment cure rates We need further partner treatment trials to determine if partner treatment improves cure/reinfection rates