The new england journal of medicine

Review Article

Dan L. Longo, M.D., Editor and Desquamative Inflammatory

Jorma Paavonen, M.D., Ph.D., and Robert C. Brunham, M.D.​​

From the Department of Obstetrics and aginal symptoms are remarkably common. In the United States, Gynecology, Helsinki University Hospital, vaginal are among the 25 most common medical reasons for Helsinki (J.P.); and the Department of Medicine, University of British Columbia, which women consult a physician, resulting in 5 million to 10 million office Vancouver, Canada (R.C.B.). Address re­ V 1-3 visits per year. Vaginal infections affect a woman’s quality of life by causing print requests to Dr. Paavonen at the De­ frustration, anxiety, , and vulvovaginal discomfort. In addition partment of Obstetrics and Gynecology, Helsinki University Hospital, Haartman­ to direct health care costs associated with the management of vaginal infections, in­katu 2, 00290 Helsinki, or at ­jorma​ there are indirect costs related to adverse reproductive health consequences.3,4 An .­paavonen@​­hus​.­fi. abnormal vaginal microbiome, or vaginal dysbiosis, which characterizes bacterial N Engl J Med 2018;379:2246-54. vaginosis and desquamative inflammatory vaginitis, has been linked to adverse DOI: 10.1056/NEJMra1808418 outcomes, pelvic inflammatory disease, an increased risk of sexually Copyright © 2018 Massachusetts Medical Society. transmitted infections, and other reproductive health problems, such as a poor outcome of in vitro fertilization (IVF).5-7 This review focuses on bacterial vaginosis and desquamative inflammatory vaginitis because both are common, underrecognized disorders, and important new data about them have emerged. Not discussed in this review are and vulvovaginal , two other common causes of vaginal symptoms.

Vaginal Microbiome Natural fluctuations in the vaginal microbiome occur during the reproductive cycle and throughout a woman’s life. During a woman’s reproductive years, the vaginal microbiome appears to be principally influenced by the effects of on vaginal epithelial cells, the predominance of lactobacilli, and low pH. The vaginal microbiome can also be transiently influenced by several other factors, such as use of antimicrobial agents, sexual activity, and menses, all of which challenge our understanding of the dynamic patterns of .4 On the basis of genomic investigations, the vaginal microbiome has been classified into at least five com- munity state types (CSTs).4,8 Four CSTs are dominated by a (lactic acid– producing) species: Lactobacillus crispatus, L. gasseri, L. iners, or L. jensenii (Table 1). One type (CST IV) is characterized by low concentrations or an absence of lacto- bacilli and high concentrations of obligate or facultative anaerobic flora. This CST is associated with both bacterial vaginosis and desquamative inflammatory vagi- nitis. L. crispatus, L. gasseri, and L. jensenii usually occur as a single or predominant microorganism in the vaginal microbiome, whereas L. iners commonly occurs as a component of a polymicrobial vaginal flora, often transitioning to bacterial vagi- nosis.9 L. crispatus excludes other organisms through low pH due to robust lactic acid production together with hydrogen peroxide and specific host antimicrobial proteins called defensins.8 The presence of hydrogen peroxide–producing lactobacilli is associated with reduced levels of vaginal proinflammatory cytokines.10 Low pH associated with

2246 n engl j med 379;23 nejm.org December 6, 2018 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved. Bacterial Vaginosis and Desquamative Inflammatory Vaginitis

Table 1. Diagnostic Findings in Vaginal Secretions from Women with Healthy Vaginal Flora, Women with Bacterial Vaginosis, and Women with Desquamative Inflammatory Vaginitis.

Desquamative Inflammatory Variable Healthy Vaginal Flora Bacterial Vaginosis Vaginitis pH <4.7 ≥4.7 ≥4.7 Amine odor Negative Positive Negative Clue cells Absent Present Absent Epithelial cells Mature squamous cells Mature squamous cells Immature parabasal cells Neutrophils Absent Absent Present Flora Sparse monomorphic bacilli Abundant polymorphic coccobacilli Abundant polymorphic cocci and bacilli Microbiome Lactobacilli , Atopobium Escherichia coli, group B strepto- ­vaginae, others cocci, others CST* I, II, V III, IV IV

* Community state type (CST) I is dominated by Lactobacillus crispatus, CST II by L. gasseri, CST III by L. iners, and CST V by L. jensenii; CST IV is composed of a polymicrobial mixture of strict and facultative anaerobes.8 lactobacilli may be an evolutionarily selected trait disorder of the vaginal microbiome that is char- to defend against sexually transmitted and other acterized by the absence of vaginal lactobacilli. infections,11 since a low-pH environment markedly Bacterial vaginosis is one of the most com- inhibits bacterial growth. Hydrogen peroxide– mon vaginal ecosystem–related microbiologic syn- producing lactobacilli predominate in normal dromes among women of childbearing age. An vaginal flora, typically accounting for 70 to 90% estimated 7.4 million cases of bacterial vaginosis of the total microbiome in a healthy .12,13 occur each year in the United States. The preva- Figure 1A and 1B show, respectively, the physical lence rates are in the range of 15% among preg- appearance of normal vaginal secretions on spec- nant women, 20 to 25% among young women ulum examination and rodlike on a mi- seen at student health clinics, and up to 30 to croscopic examination of a wet-mount prepara- 40% among women seen at sexually transmitted tion of normal vaginal fluid. Table 1 compares the disease clinics. The prevalence rates for bacterial characteristics of vaginal secretions among healthy vaginosis vary strikingly among ethnic groups women, women with bacterial vaginosis, and those and countries.22 Rates are generally higher in with desquamative inflammatory vaginitis. black and Hispanic populations and lower in white and Asian populations. The reasons for Bacterial Vaginosis the large differences in prevalence rates accord- ing to race or ethnic group and geographic re- A link between Haemophilus vaginalis and abnormal gion are unknown. Two reviews provide excel- was first described in 1955.4,14 lent overall summaries of bacterial vaginosis and Subsequently, H. vaginalis was renamed Gardnerella the literature on the disorder.22,23 vaginalis, and the syndrome was renamed non- Despite advances in our understanding of bac- specific vaginitis14 or anaerobic vaginosis because terial vaginosis, it remains an enigmatic condi- anaerobic organisms, in addition to G. vaginalis, tion.24 A recent natural history study showed were observed. Currently, the condition is called that incident bacterial vaginosis is associated with bacterial vaginosis.4 The search for a single etio- an initial decrease in the abundance of L. crispa- logic agent has continued, with investigations of tus and a subsequent increase in the abundance mobiluncus species in the 1980s and 1990s15-17 of bivia, G. vaginalis, A. vaginae, and mega- and, in the 2000s, investigations of Atopobium sphaera type 1 (anaerobes commonly found in vaginae18-20 and the discovery of Clostridiales-type bacterial vaginosis).9 Dramatic differences in me- bacteria.21 However, most investigators have con- tabolite compositions and concentrations of mi- cluded that bacterial vaginosis is a polymicrobial crobial origin in bacterial vaginosis have also

n engl j med 379;23 nejm.org December 6, 2018 2247 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

Figure 1. Features of Healthy Vaginal Flora, Bacterial Vaginosis, and Desquamative Inflammatory Vaginitis. Panel A shows healthy cervicovaginal mucosa and a small amount of vaginal discharge, findings that are consistent with a predominance of lactobacilli. Physiological cervical ectopy and clear cervical mucus are evident. In Panel B, microscopic examination of a wet­mount preparation shows rodlike bacteria, which are consistent with lactobacilli. No leukocytes are present. Panels C and D show the features of bacterial vaginosis: heavy, milky, homogeneous vaginal discharge with bubbles (Panel C), which are consistent with gaseous by­products of anaerobic bacteria, and vaginal epithelial cells covered by coccobacilli on microscopic examination (Panel D), a feature of clue cells. No leuko­ cytes are present. Panels E and F show the features of desquamative inflammatory vaginitis: heavy, yellowish vaginal discharge and inflamed cervicovaginal mucosa (Panel E), with microscopic examination showing a high number of leukocytes (with a predominance of mononuclear leukocytes) and round parabasal cells (Panel F), findings that are consistent with .

been identified by means of a global metabolo- (Fig. 1C) that causes vulvovaginal discomfort mics approach.25 and vulvar . The disorder is also char- Bacterial vaginosis is characterized by a milky, acterized by the absence of clinically significant homogeneous, malodorous vaginal discharge vaginal inflammation as indicated by an absence

2248 n engl j med 379;23 nejm.org December 6, 2018 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved. Bacterial Vaginosis and Desquamative Inflammatory Vaginitis of neutrophils. Several studies have shown that absence of rods (Fig. 1D); an absence of rods inflammatory cytokines are increased in the indicates an absence of lactobacilli. These tests vaginal discharge of patients with bacterial vagi- have been introduced into clinical practice and nosis, suggesting leukocyte inhibition despite a are widely used to determine whether bacterial proinflammatory milieu.8,22,26 Bacterial vaginosis– vaginosis is present. A vaginal pH of less than associated odor is typically a fishy smell (i.e., a 4.7 provides an easy-to-read cutoff value for dis- positive whiff test after the addition of one drop tinguishing between normal flora and bacterial of potassium hydroxide to vaginal discharge on vaginosis and is used to rule out bacterial vagi- a glass slide). This smell is caused by the release nosis (Table 1). of organic acids or polyamines on alkalinization A recent study validated the use of an investi- of vaginal fluid, which are by-products of an- gational molecular nucleic acid amplification test aerobic bacterial metabolism (e.g., putrescine and that has been approved by the Food and Drug cadaverine).27 The polymicrobial load is increased Administration for the diagnosis of bacterial by a factor of up to 1000, as compared with vaginosis and other vaginitis syndromes.33 Quan- normal, lactobacilli-dominated vaginal flora. Thus, titative polymerase chain-reaction assays for the bacterial vaginosis represents an abnormal vag- diagnosis of bacterial vaginosis are based on inal ecosystem, both qualitatively and quantita- detection of the predominant bacterial vaginosis– tively. associated organisms, such as G. vaginalis, A. vagi- The absence of a clear disease counterpart in nae, and mobiluncus species. In the study, involv- males has made it difficult to determine whether ing 1740 symptomatic patients, the performance bacterial vaginosis is sexually transmitted. A sys- of the nucleic acid amplification test for detect- tematic review of randomized trials of treatment ing bacterial vaginosis, as compared with the for male sexual partners to prevent recurrent reference method (the combined results of vagi- bacterial vaginosis in women showed that none nal Gram’s staining and wet-mount ), of the trials had sufficient power to determine was acceptable (sensitivity, 90.5%; specificity, the role of the male partner in the recurrence of 85.8%).33 However, the test requires additional bacterial vaginosis.28 Another review concluded validation. that, as compared with placebo, treat- ment for the sexual partners of women treated Pathogenesis for bacterial vaginosis had no effect on rates of Bacterial vaginosis can be considered a biofilm clinical or symptomatic improvement among the , with a dense polymicrobial biofilm con- women or on the rate of recurrence of bacterial sisting primarily of G. vaginalis adhering to the vaginosis for up to 12 weeks after treatment.29 .34 An A. vaginae biofilm is However, bacterial vaginosis and sexually trans- always present with a G. vaginalis biofilm,22 and mitted infections have many characteristics in higher bacterial loads of G. vaginalis and A. vagi- common, and several findings are consistent nae increase the probability of biofilm formation. with a strong association between incident bac- The vaginal biofilm appears to create a favorable terial vaginosis and sexual activity.29,30 Thus, there anaerobic environment for other obligate anaer- may be either unmeasured confounders in these obic bacteria.22 An important finding related to studies or a transmissible microbial component upper genital tract complications is that half of of bacterial vaginosis that has not yet been iden- women with bacterial vaginosis also have a bac- tified. terial vaginosis–associated biofilm covering the .35 That this biofilm ascends to the Diagnosis endometrium may explain the links among ad- The validation of two standardized, reproducible verse pregnancy outcomes, pelvic inflammatory diagnostic tests for bacterial vaginosis that are disease, and bacterial vaginosis. However, the based on the use of vaginal swabs has been an exact role of biofilm in relation to infectious important development. One test is laboratory- diseases of the upper genital tract remains un- based Gram’s staining for vaginal flora31; the certain.36 For instance, the endometrial cavity is other is a bedside, wet-mount microscopic test not sterile in most women, and the presence of for vaginal clue cells.32 Clue cells are epithelial low levels of bacteria in the is not associ- squamous cells covered by coccobacilli in the ated with clinically significant inflammation.37

n engl j med 379;23 nejm.org December 6, 2018 2249 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

The striking increase, by a factor of 1000, in vaginal-cuff cellulitis after hysterectomy,48 and potentially virulent bacteria in women with bac- postpartum endometritis49 are increased by up terial vaginosis, as compared with women who to a factor of 6 among women with bacterial have healthy vaginal flora, may explain the asso- vaginosis. ciation of bacterial vaginosis with upper genital Several studies have assessed the value of tract infection. screening for and treating bacterial vaginosis in the prevention of . The results have Bacterial Vaginosis and Other Sexually been highly variable, and antimicrobial treatment Transmitted Infections of bacterial vaginosis in pregnancy does not uni- Bacterial vaginosis is associated with not only versally reduce adverse pregnancy outcomes.50,51 the acquisition but also the transmission of other Treatment of bacterial vaginosis in early preg- sexually transmitted infections, especially human nancy (at <20 weeks of gestation) may be more immunodeficiency (HIV) infection.38,39 In effective in preventing preterm birth than treat- women with bacterial vaginosis, CD4 T cells are ment in later pregnancy.52 Since microorganisms recruited to the lower genital tract mucosa.40,41 associated with bacterial vaginosis can ascend Among HIV-infected women, the quantity of HIV into the endometrium before pregnancy, they may in vaginal secretions from women with bacterial infect the chorioamnion during pregnancy.53,54 vaginosis is increased substantially, as compared Genetic factors may be an important compo- with HIV in vaginal secretions from women nent in the pathogenesis of preterm birth associ- without bacterial vaginosis.8 The bacterial vagi- ated with bacterial vaginosis. In one study, the nosis–associated vaginal microbiome also inac- risk of preterm birth was increased by a factor tivates the topical microbicide tenofovir, which of 6 among women with both bacterial vaginosis is used for the prevention of HIV transmission.42 and a single-nucleotide polymorphism (SNP) for trachomatis infection is strongly associat- tumor necrosis factor α but was increased by a ed with bacterial vaginosis.43,44 Chlamydia-asso- factor of only 2 among women with either fea- ciated increases the amount of cervical ture alone.55 Among women with other inflam- secretions. This increase, in turn, may change the matory SNPs, preterm birth rates were increased vaginal ecosystem, favoring the growth of anaero- by a factor of 2 to 5 for women who had bacte- bic microorganisms. Thus, controlling C. tracho- rial vaginosis as compared with those who did matis infection rates may prevent bacterial vagi- not have bacterial vaginosis.56 nosis, perhaps explaining why efforts to control The link between bacterial vaginosis and pel- C. trachomatis have had a disproportionately posi- vic inflammatory disease has been more consis- tive effect on reducing rates of pelvic inflamma- tently replicated than the association of bacterial tory disease.45 vaginosis with adverse pregnancy outcomes. Lapa- roscopic studies have shown that microorgan- Overall Disease Burden isms that are prevalent in high concentrations in Bacterial vaginosis has a large variety of sequelae the vagina in women with bacterial vaginosis are in the upper genital tract, including increased also observed in the endometrium and fallopian risks of preterm birth, first-trimester miscarriage tubes in women with proven pelvic inflammatory in women undergoing IVF, amniotic-fluid infec- disease.57 tion, chorioamnionitis, after child- birth or abortion, and infections after hysterec- Treatment tomy, as well as pelvic inflammatory disease, Table 2 summarizes the guidelines from the both in general and after abortion.23,46 The attrib- Centers for Disease Control and Prevention for utable proportion of these sequelae has not been the treatment of bacterial vaginosis.58 The guide- universally quantified. Overall, bacterial vagino- lines consist of various regimens of oral or vagi- sis is associated with only a modest increase, by a nally applied or clindamycin. Oral factor of 2, in the risk of preterm birth. Although metronidazole, topical metronidazole, and topi- this risk has been consistently observed in mul- cal clindamycin are equally effective, although tiple populations, the vast majority of women oral metronidazole has more side effects.59 The with bacterial vaginosis do not deliver preterm. presence of A. vaginae, which is often resistant to The risks of endometritis after cesarean section,47 metronidazole, predicts a high risk of recurrence,

2250 n engl j med 379;23 nejm.org December 6, 2018 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved. Bacterial Vaginosis and Desquamative Inflammatory Vaginitis

suggesting that metronidazole is not an ideal Table 2. Treatment Guidelines for Bacterial Vaginosis.* empirical agent.60 The exact relationship of bac- terial vaginosis–associated biofilm with treatment Treatment Regimen failure is not known. It is plausible, however, Recommended treatments that biofilm infection is difficult to eradicate by Metronidazole 500 mg orally twice a day for 7 days means of antimicrobial therapy. The role of pro- Metronidazole 0.75% gel One applicator (5 g) intravaginally once a day biotics as supplementary agents in the treatment for 5 days 22 of bacterial vaginosis is under study. In one Clindamycin 2% cream One applicator (5 g) intravaginally at bedtime trial, oral lactobacilli combined with metronida- for 7 days zole was more effective than metronidazole alone Alternative treatments in resolving bacterial vaginosis.59 2 g orally once a day for 2 days Tinidazole 1 g orally once a day for 5 days Desquamative Inflammatory Clindamycin 300 mg orally twice a day for 7 days Vaginitis Clindamycin ovules 100 mg intravaginally at bedtime for 3 days Desquamative inflammatory vaginitis is a newly * The guidelines are from the Centers for Disease Control and Prevention.58 recognized clinical syndrome characterized by persistent purulent vaginal discharge and vagi- nal erythema, often with submucosal cervico- in abnormal vaginal flora. As with bacterial vagi- vaginal petechiae (Fig. 1E).61,62 Inflammation is the nosis, understanding the mechanism underlying cardinal feature of this disorder, which has also the loss of vaginal lactobacilli should shed light been called idiopathic inflammatory vaginitis. on the pathogenesis of desquamative inflamma- Donders and colleagues have recently reviewed tory vaginitis. the literature on this inflammatory vaginitis, which they call “.”63 However, the Symptoms and Signs term “desquamative inflammatory vaginitis” holds Clinical manifestations of desquamative inflam- priority and was first introduced in 1965 by Gray matory vaginitis include purulent vaginal dis- and Barnes.62 The term “aerobic vaginitis” was charge and a strong inflammatory reaction. The introduced in 2002 in reference to a disease en- vaginal discharge is homogeneous and yellow- tity caused by an abnormal vaginal microbiome ish, with no fishy smell (Table 1). Vulvar irrita- genomically defined as CST IV.63 The published tion and vaginal mucosal erythema with ecchy- literature on desquamative inflammatory vagini- motic lesions or erosions are present in severe tis is still surprisingly limited, consisting mainly cases (Fig. 1E). Symptoms may last for a long of retrospective case series or short reviews.61,63,64 time and fluctuate, suggesting a chronic or re- current natural history. Cause The exact cause of desquamative inflammatory Epidemiology vaginitis is unknown but appears to be a dysbio- In the few studies that have systematically ana- sis of the normal vaginal microbiome associated lyzed the prevalence of desquamative inflamma- with inflammation. In desquamative inflamma- tory vaginitis among pregnant or nonpregnant tory vaginitis, the vagina is colonized with facul- women, the rates have ranged from 2 to 20%.63 tative bacteria, not the obligate anaerobic bacte- One important limitation of epidemiologic stud- ria that colonize the vagina in bacterial vaginosis. ies has been the lack of standardized biomarkers The microflora in desquamative inflammatory for the diagnosis of desquamative inflammatory vaginitis typically consist of Escherichia coli, Staphy- vaginitis. Lack of diagnostic precision is com- lococcus aureus, group B streptococcus, or Entero- pounded by the fact that the existence of this coccus faecalis.63 The microbiome associated with condition has not been accepted universally by desquamative inflammatory vaginitis is less well clinicians. In our experience, highly symptom- understood than the bacterial vaginosis micro- atic desquamative inflammatory vaginitis is rel- biome. Desquamative inflammatory vaginitis may atively rare, whereas a less symptomatic form of also represent a systemic inflammatory syndrome vaginal dysbiosis, characterized by reduced num- that produces vaginal inflammation, resulting bers of lactobacilli, increased numbers of facul-

n engl j med 379;23 nejm.org December 6, 2018 2251 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved. The new england journal of medicine

Table 3. Treatment Recommendations for Desquamative Inflammatory branes, chorioamnionitis, and other adverse 63 Vaginitis.* pregnancy outcomes, such as miscarriage. Dys- biosis in women with desquamative inflamma- Treatment Regimen tory vaginitis might increase the risk of neonatal Recommended treatments group B streptococcal infection or urinary tract Clindamycin 2% cream Intravaginally daily at bedtime for 1 to infection caused by E. coli. Desquamative inflam- 3 wk; consider maintenance therapy matory vaginitis is also likely to be important in once or twice a week for 2–6 mo upper genital tract infection such as pelvic in- Topical glucocorticoid flammatory disease, although this has not been Hydrocortisone, 300–500 mg Intravaginally daily at bedtime for 3 wk; definitively proved. consider maintenance therapy once or twice a week for 2–6 mo Treatment Clobetasol propionate Intravaginally daily at bedtime for 1 wk (duration not evidence-based) Recommended treatment approaches for desqua- mative inflammatory vaginitis are presented in Additional recommended 65 ­treatments† Table 3. These treatment options have not been properly tested in randomized clinical trials. Fluconazole 150 mg orally once a week as maintenance therapy Metronidazole is not effective in desquamative Topical vaginal estrogen Twice a week inflammatory vaginitis, and treatment failure with metronidazole in women with bacterial * The recommendations are from Reichman and Sobel.65 Official treatment guide­ vaginosis may suggest desquamative inflamma- lines for desquamative inflammatory vaginitis have not been developed. tory vaginitis. Clindamycin is active against the † Additional recommended treatments are for use in combination with clindamy­ cin or one of the glucocorticoids. broad spectrum of facultative bacteria linked to desquamative inflammatory vaginitis and also has an antiinflammatory effect. In clinical prac- tice, topical clindamycin, often used as prolonged tative bacteria, and inflammation, is much more maintenance therapy, seems to be an effective common. To what extent this dysbiosis trans- treatment approach for severe forms of desqua- lates into symptomatic disease remains to be mative inflammatory vaginitis. Maintenance ther- determined. apy once weekly is commonly used to reduce the risk of recurrences or flare-ups. Diagnosis An observational study suggested that topical Microscopic examination of wet-mount prepara- application of 2% clindamycin, with or without tions of vaginal secretions reveals an increase in 10% hydrocortisone, is useful in the treatment inflammatory cells and parabasal epithelial cells of severe desquamative inflammatory vaginitis.66 (Table 1 and Fig. 1F), and vaginal flora are usu- Women with desquamative inflammatory vagini- ally abnormal, with an elevated pH.63 The point- tis characterized by a heavy parabasal-cell com- of-care diagnosis is based on the presence of an ponent may benefit from intravaginal application increased number of leukocytes and parabasal of as maintenance therapy.67 Official epithelial cells. Microscopic examination of wet- treatment guidelines for desquamative inflamma- mount preparations is the preferred diagnostic tory vaginitis have not been developed or imple- method for desquamative inflammatory vagini- mented. tis, since Gram’s staining of vaginal flora does not discriminate between bacterial vaginosis and Conclusions and Future desquamative inflammatory vaginitis. The use of Research Directions routine vaginal cultures is not recommended. The human vaginal ecosystem is highly dynamic. Disease Burden The vaginal microbiome can affect host physiol- The disease burden caused by desquamative in- ogy, and host physiology can affect the vaginal flammatory vaginitis has not been well studied. microbiome. Research is needed for a better The disorder has been linked to an increased understanding of the interactions among the vagi- risk of preterm birth, premature rupture of mem- nal microbiome, host physiology, reproduction,

2252 n engl j med 379;23 nejm.org December 6, 2018 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved. Bacterial Vaginosis and Desquamative Inflammatory Vaginitis and host defense. Recent genomic research has of immune barriers enhance susceptibility to increased our knowledge of the vaginal micro- sexually transmitted infections,68 and this, in turn, biome. Future research based on genomic, pro- increases the disease burden caused by an ab- teomic, and metabolomic techniques may ultimate­ normal vaginal microbiome. Understanding the ly have a major effect on women’s reproductive mechanisms that initiate and maintain a healthy health. The mechanisms that initiate and main- vaginal microbiome will be essential for the de- tain colonization with vaginal lactobacilli, espe- velopment of improved treatments for bacterial cially L. crispatus, in women of reproductive age vaginosis and desquamative inflammatory vagi- need to be elucidated. New biomarkers for an nitis, as well as effective topical microbicides for abnormal vaginal microbiome are needed for the prevention of HIV infection and other sexu- clinical practice. ally transmitted infections. Microorganisms of the vaginal microbiome alter the innate immune response and barrier No potential conflict of interest relevant to this article was reported. properties of the human vaginal epithelium. Sub- Disclosure forms provided by the authors are available with stantial increases in vaginal pH and disruption the full text of this article at NEJM.org.

References 1. Marrazzo J, Sobel J, Hillier S. Vaginal of vaginal interleukin-1beta, independent of by cultivation-independent methods war- infections. In: Morse SA, Ballard RC, bacterial vaginosis. Sex Transm Dis 2015;​ rants further investigation. J Clin Micro- Holmes KK, Moreland AA, eds. Atlas of 42:358-63.​ biol 2004;​42:​1829-31. sexually transmitted diseases and AIDS. 11. Godha K, Tucker KM, Biehl C, Archer 19. Ferris MJ, Masztal A, Aldridge KE, 4th ed. Philadelphia: ​Saunders, 2010:​76- DF, Mirkin S. Human vaginal pH and mi- Fortenberry JD, Fidel PL Jr, Martin DH. 93. crobiota: an update. Gynecol Endocrinol Association of Atopobium vaginae, a re- 2. Muzny CA, Schwebke JR. Vaginal 2018;​34:​451-5. cently described metronidazole resistant ­infections. In:​ Goldman M, Troisi R, 12. Gajer P, Brotman RM, Bai G, et al. anaerobe, with bacterial vaginosis. BMC Rexrode K, eds. Women and Health. 2nd Temporal dynamics of the human vagi- Infect Dis 2004;4:​ 5.​ ed. London:​ Academic Press, 2013:​473-83. nal microbiota. Sci Transl Med 2012;​4:​ 20. Marconi C, Cruciani F, Vitali B, 3. Carr PL, Rothberg MB, Friedman RH, 132ra52. Donders GG. Correlation of atopobium Felsenstein D, Pliskin JS. “Shotgun” versus 13. Verstraelen H, Verhelst R, Claeys G, vaginae amount with bacterial vaginosis sequential testing: cost-effectiveness of De Backer E, Temmerman M, Vanee­ markers. J Low Genit Tract Dis 2012;​16:​ diagnostic strategies for vaginitis. J Gen choutte M. Longitudinal analysis of the 127-32. Intern Med 2005;20:​ 793-9.​ vaginal microflora in pregnancy suggests 21. Fredricks DN, Fiedler TL, Marrazzo 4. Hillier SL. The complexity of micro- that L. crispatus promotes the stability of JM. Molecular identification of bacteria bial diversity in bacterial vaginosis. N Engl the normal vaginal microflora and that associated with bacterial vaginosis. N Engl J Med 2005;353:​ 1886-7.​ L. gasseri and/or L. iners are more conducive J Med 2005;​353:​1899-911. 5. Cohen CR, Lingappa JR, Baeten JM, to the occurrence of abnormal vaginal mi- 22. Kenyon C, Colebunders R, Crucitti T. et al. Bacterial vaginosis associated with croflora. BMC Microbiol 2009;​9:​116. The global epidemiology of bacterial vagi- increased risk of female-to-male HIV-1 14. Gardner HL, Dukes CD. Haemophilus nosis: a systematic review. Am J Obstet transmission: a prospective cohort analy- vaginalis vaginitis: a newly defined spe- Gynecol 2013;​209:​505-23. sis among African couples. PLoS Med cific infection previously classified non- 23. Onderdonk AB, Delaney ML, Fichorova 2012;9(6):​ e1001251.​ specific vaginitis. Am J Obstet Gynecol RN. The human microbiome during bac- 6. Hillier SL, Nugent RP, Eschenbach DA, 1955;69:​ 962-76.​ terial vaginosis. Clin Microbiol Rev 2016;​ et al. Association between bacterial vagi- 15. Roberts MC, Hillier SL, Schoenknecht 29:223-38.​ nosis and preterm delivery of a low-birth- FD, Holmes KK. Comparison of gram 24. Marrazzo JM, Martin DH, Watts DH, weight infant. N Engl J Med 1995;​333:​ stain, DNA probe, and culture for the et al. Bacterial vaginosis: identifying re- 1737-42. identification of species of Mobiluncus search gaps proceedings of a workshop 7. Eschenbach DA. Bacterial vaginosis in female genital specimens. J Infect Dis sponsored by DHHS/NIH/NIAID. Sex and anaerobes in obstetric-gynecologic 1985;​152:​74-7. Transm Dis 2010;37:​ 732-44.​ infection. Clin Infect Dis 1993;​16:​Suppl 4:​ 16. Hillier SL, Critchlow CW, Stevens CE, 25. Srinivasan S, Morgan MT, Fiedler TL, S282-S287. et al. Microbiological, epidemiological and et al. Metabolic signatures of bacterial 8. Smith SB, Ravel J. The vaginal micro- clinical correlates of vaginal colonisation vaginosis. MBio 2015;​6(2):​e00204-e00215. biota, host defence and reproductive physi- by Mobiluncus species. Genitourin Med 26. Mitchell C, Marrazzo J. Bacterial vagi- ology. J Physiol 2017;​595:​451-63. 1991;67:​ 26-31.​ nosis and the cervicovaginal immune re- 9. Muzny CA, Blanchard E, Taylor CM, 17. Schwebke JR, Lawing LF. Prevalence sponse. Am J Reprod Immunol 2014;​71:​ et al. Identification of key bacteria in- of Mobiluncus spp among women with 555-63. volved in the induction of incident bacte- and without bacterial vaginosis as detected 27. Chen KC, Amsel R, Eschenbach DA, rial vaginosis: a prospective study. J Infect by polymerase chain reaction. Sex Transm Holmes KK. Biochemical diagnosis of Dis 2018;218:​ 966-78.​ Dis 2001;28:​ 195-9.​ vaginitis: determination of diamines in 10. Mitchell C, Fredricks D, Agnew K, 18. Burton JP, Devillard E, Cadieux PA, vaginal fluid. J Infect Dis 1982;​145:​337-45. Hitti J. Hydrogen peroxide-producing lac- Hammond JA, Reid G. Detection of Atopo- 28. Mehta SD. Systematic review of ran- tobacilli are associated with lower levels bium vaginae in postmenopausal women domized trials of treatment of male sexual

n engl j med 379;23 nejm.org December 6, 2018 2253 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved. Bacterial Vaginosis and Desquamative Inflammatory Vaginitis

partners for improved bacteria vaginosis modulator of host inflammatory responses morphism in the promoter region of TNF outcomes in women. Sex Transm Dis 2012;​ in the female genital tract. Immunity 2015;​ and bacterial vaginosis: preliminary evi- 39:822-30.​ 42:965-76.​ dence of gene-environment interaction in 29. Amaya-Guio J, Viveros-Carreño DA, 42. Klatt NR, Cheu R, Birse K, et al. Vagi- the etiology of spontaneous preterm birth. Sierra-Barrios EM, Martinez-Velasquez MY, nal bacteria modify HIV tenofovir micro- Am J Obstet Gynecol 2004;​190:​1504-1508, Grillo-Ardila CF. Antibiotic treatment for bicide efficacy in African women. Science 3A. the sexual partners of women with bac- 2017;​356:​938-45. 56. Gómez LM, Sammel MD, Appleby DH, terial vaginosis. Cochrane Database Syst 43. Bautista CT, Wurapa E, Sateren WB, et al. Evidence of a gene-environment in- Rev 2016;​10:​CD011701. Morris S, Hollingsworth B, Sanchez JL. teraction that predisposes to spontaneous 30. Marrazzo JM, Thomas KK, Agnew K, Bacterial vaginosis: a synthesis of the litera- preterm birth: a role for asymptomatic Ringwood K. Prevalence and risks for ture on etiology, prevalence, risk factors, bacterial vaginosis and DNA variants in bacterial vaginosis in women who have and relationship with chlamydia and gon- genes that control the inflammatory re- sex with women. Sex Transm Dis 2010;​37:​ orrhea infections. Mil Med Res 2016;​3:​4. sponse. Am J Obstet Gynecol 2010;​202(4):​ 335-9. 44. Bautista CT, Wurapa EK, Sateren WB, 386.e1-386.e6. 31. Nugent RP, Krohn MA, Hillier SL. Re- Morris SM, Hollingsworth BP, Sanchez JL. 57. Paavonen J, Teisala K, Heinonen PK, liability of diagnosing bacterial vaginosis Association of bacterial vaginosis with et al. Microbiological and histopathologi- is improved by a standardized method of chlamydia and among women cal findings in acute pelvic inflammatory gram stain interpretation. J Clin Microbiol in the U.S. army. Am J Prev Med 2017;​52:​ disease. Br J Obstet Gynaecol 1987;​94:​454- 1991;​29:​297-301. 632-9. 60. 32. Amsel R, Totten PA, Spiegel CA, Chen 45. Scholes D, Satterwhite CL, Yu O, Fine D, 58. Workowski KA, Bolan GA. Sexually KC, Eschenbach D, Holmes KK. Nonspe- Weinstock H, Berman S. Long-term trends transmitted diseases treatment guidelines, cific vaginitis.: diagnostic criteria and mi- in infections and 2015. MMWR Recomm Rep 2015;​64(RR-3):​ crobial and epidemiologic associations. related outcomes in a U.S. managed care 1-137. Am J Med 1983;​74:​14-22. population. Sex Transm Dis 2012;​39:​81-8. 59. Oduyebo OO, Anorlu RI, Ogunsola 33. Gaydos CA, Beqaj S, Schwebke JR, 46. Brown RG, Marchesi JR, Lee YS, et al. FT. The effects of antimicrobial therapy et al. Clinical validation of a test for the Vaginal dysbiosis increases risk of pre- on bacterial vaginosis in non-pregnant diagnosis of vaginitis. Obstet Gynecol term fetal membrane rupture, neonatal women. Cochrane Database Syst Rev 2009;​ 2017;​130:​181-9. sepsis and is exacerbated by erythromycin. 3:CD006055.​ 34. Hardy L, Jespers V, Dahchour N, et al. BMC Med 2018;​16:​9. 60. Bradshaw CS, Tabrizi SN, Fairley CK, Unravelling the bacterial vaginosis-asso- 47. Watts DH, Eschenbach DA, Kenny GE. Morton AN, Rudland E, Garland SM. The ciated biofilm: a multiplex Gardnerella Early postpartum endometritis: the role association of Atopobium vaginae and vaginalis and Atopobium vaginae fluores- of bacteria, genital , and Gardnerella vaginalis with bacterial vagi- cence in situ hybridization assay using Chlamydia trachomatis. Obstet Gynecol nosis and recurrence after oral metronida- peptide nucleic acid probes. PLoS One 1989;73:​ 52-60.​ zole therapy. J Infect Dis 2006;​194:​828-36. 2015;​10(8):​e0136658. 48. Soper DE, Bump RC, Hurt WG. Bacte- 61. Oates JK, Rowen D. Desquamative in- 35. Swidsinski A, Verstraelen H, Loening- rial vaginosis and trichomoniasis vagini- flammatory vaginitis: a review. Genitourin Baucke V, Swidsinski S, Mendling W, Hal- tis are risk factors for cuff cellulitis after Med 1990;66:​ 275-9.​ wani Z. Presence of a polymicrobial endo- abdominal hysterectomy. Am J Obstet 62. Gray LA, Barnes ML. Vaginitis in metrial biofilm in patients with bacterial Gynecol 1990;163:​ 1016-23.​ women, diagnosis and treatment. Am J vaginosis. PLoS One 2013;​8(1):​e53997. 49. Watts DH, Krohn MA, Hillier SL, Obstet Gynecol 1965;​92:​125-36. 36. Bradshaw CS, Sobel JD. Current treat- Eschenbach DA. Bacterial vaginosis as a 63. Donders GGG, Bellen G, Grinceviciene ment of bacterial vaginosis-limitations and risk factor for post-cesarean endometritis. S, Ruban K, Vieira-Baptista P. Aerobic vagi- need for innovation. J Infect Dis 2016;​214:​ Obstet Gynecol 1990;75:​ 52-8.​ nitis: no longer a stranger. Res Microbiol Suppl 1:​S14-S20. 50. Brocklehurst P, Gordon A, Heatley E, 2017;​168:​845-58. 37. Mitchell CM, Haick A, Nkwopara E, Milan SJ. for treating bacterial 64. Nyirjesy P, Peyton C, Weitz MV, Mathew et al. Colonization of the upper genital vaginosis in pregnancy. Cochrane Data- L, Culhane JF. Causes of chronic vaginitis: tract by vaginal bacterial species in non- base Syst Rev 2013;​1:​CD000262. analysis of a prospective database of af- pregnant women. Am J Obstet Gynecol 51. Thinkhamrop J, Hofmeyr GJ, Adetoro fected women. Obstet Gynecol 2006;​108:​ 2015;212(5):​ 611.e1-9.​ O, Lumbiganon P, Ota E. Antibiotic pro- 1185-91. 38. Atashili J, Poole C, Ndumbe PM, Adi- phylaxis during the second and third tri- 65. Reichman O, Sobel J. Desquamative mora AA, Smith JS. Bacterial vaginosis mester to reduce adverse pregnancy out- inflammatory vaginitis. Best Pract Res and HIV acquisition: a meta-analysis of comes and morbidity. Cochrane Database Clin Obstet Gynaecol 2014;​28:​1042-50. published studies. AIDS 2008;​22:​1493-501. Syst Rev 2015;​6:​CD002250. 66. Sobel JD, Reichman O, Misra D, Yoo 39. Martin HL, Richardson BA, Nyange 52. Sangkomkamhang US, Lumbiganon P, W. Prognosis and treatment of desquama- PM, et al. Vaginal lactobacilli, microbial Prasertcharoensuk W, Laopaiboon M. Ante- tive inflammatory vaginitis. Obstet Gyne- flora, and risk of human immunodefi- natal lower genital tract infection screen- col 2011;​117:​850-5. ciency virus type 1 and sexually transmit- ing and treatment programs for prevent- 67. Shen J, Song N, Williams CJ, et al. Ef- ted disease acquisition. J Infect Dis 1999;​ ing preterm delivery. Cochrane Database fects of low dose estrogen therapy on the 180:​1863-8. Syst Rev 2015;​2:​CD006178. vaginal microbiomes of women with atro- 40. Gosmann C, Anahtar MN, Handley 53. Goldenberg RL, Culhane JF, Iams JD, phic vaginitis. Sci Rep 2016;​6:​24380. SA, et al. Lactobacillus-deficient cervico- Romero R. Epidemiology and causes of 68. Doerflinger SY, Throop AL, Herbst- vaginal bacterial communities are associ- preterm birth. Lancet 2008;​371:​75-84. Kralovetz MM. Bacteria in the vaginal ated with increased HIV acquisition in 54. Goldenberg RL, Hauth JC, Andrews microbiome alter the innate immune re- young South African women. Immunity WW. Intrauterine infection and preterm sponse and barrier properties of the hu- 2017;​46:​29-37. delivery. N Engl J Med 2000;​342:​1500-7. man vaginal epithelia in a species-specific 41. Anahtar MN, Byrne EH, Doherty KE, 55. Macones GA, Parry S, Elkousy M, manner. J Infect Dis 2014;​209:​1989-99. et al. Cervicovaginal bacteria are a major Clothier B, Ural SH, Strauss JF III. A poly- Copyright © 2018 Massachusetts Medical Society.

2254 n engl j med 379;23 nejm.org December 6, 2018 The New England Journal of Medicine Downloaded from nejm.org by ROBERT BRUNHAM on December 5, 2018. For personal use only. No other uses without permission. Copyright © 2018 Massachusetts Medical Society. All rights reserved.