AFTER THE INTRODUCTION OF Amsel criteria1 for the evaluation of bacterial vaginosis (BV), many treatment trials reported outcomes in terms of posttreatment detection of BV by Amsel criteria2–20 (Table 1). The Amsel criteria for BV require the presence of any three of the following four criteria: 1) homogeneous vaginal discharge, 2) vaginal fluid pH >4.5, 3) release of fishy amine odor from vaginal fluid when mixed with 10% KOH solution, and 4) presence of clue cells in vaginal fluid (representing at least 20% of cells in a commonly used modification21). A commonly reported successful posttreatment outcome has been absence of BV by Amsel criteria (i.e., <3 criteria present). However, individual studies have varied greatly in what formulation of the Amsel criteria and other measures are considered, including the Nugent criteria, consisting of a Gram stain score of 7 or more. The Gram stain score varies from 0 to 10 and is computed from concentrations observed on vaginal fluid Gram stain of Gardnerella, Prevotella and Mobiluncus, and Lactobacillus morphotypes.22
The current U. S. Food and Drug Administration (FDA) Draft Guidance for BV cure in a treatment trial, published for comment in 1998 but still considered a Draft Guidance, calls for one test of cure (TOC) visit at 21 to 30 days after starting treatment, and defines therapeutic cure as clinical cure (0 of the 4 Amsel criteria) and Nugent score of 0 to 3. Since then, some reports have provided outcomes based both on Nugent score <4 (normal) and <7 (not consistent with BV) along with outcomes based on Amsel criteria.8,12,13,23,24 However, few if any published evaluations of treatment for BV have defined outcomes based on cure defined by the FDA Draft Guidance criteria. Most recent publications8,12,13,16,23–25 continue to report treatment outcomes using less rigorous criteria involving Amsel criteria but not Nugent score, as in pre-1998 studies.
In our recent comparison of two intravaginal metronidazole-containing regimens for the treatment of BV, we observed a significant difference between the two regimens when we defined BV by the presence at follow up of at least three of four Amsel criteria, of a Nugent score consistent with BV, or both, but found no significant difference in rates of cure as defined by the 1998 FDA Guidance. As separately reported,26 at follow up 10 to 21 days after starting treatment, BV was detected by Amsel and Nugent criteria in 2% of subjects (one of 51) given ovules containing metronidazole and nystatin versus 23% (13 of 57) given gel containing metronidazole (P <0.01). However, rates of cure per FDA Draft Guidance criteria were 35% (18 of 51) in those given ovules versus 26% (15 of 57) in those given gel (P = 0.4). To determine the reason(s) for these disparate results, we analyzed the effects of these two treatment regimens on the various components of the Amsel and Nugent criteria.
Materials and Methods
Data were collected from a trial of two intravaginal metronidazole treatments carried out at Dos de Mayo Hospital in Lima, Peru, from February 19, 1999, to April 24, 2000. Recruitment and randomization are described elsewhere.26 Briefly, women 18 to 40 years old and reporting vaginal discharge or vaginal malodor were recruited for the study. Women were tested for pH ≥4.7 and amines in vaginal fluid using a point-of-care test,* appearance of vaginal discharge was noted, and vaginal fluid was collected for Gram stain and microscopy after mixing separately with saline and with 10% potassium hydroxide. Women with vaginal fluid positive for both pH ≥4.7 and for amines were randomly assigned to one of two intravaginal treatments: metronidazole 0.75% gel (an FDA- approved treatment for BV, delivering 37.5 mg metronidazole per nightly dose) or ovules containing 500 mg metronidazole plus 1,000 U nystatin (Flagystatin, intended for syndromic management of vaginal discharge and a commonly prescribed medication for BV symptoms in Peru26). Both regimens were prescribed nightly for 5 nights.
Women were asked to return at 14 days, 42 days, and 104 days after the baseline visit for speculum examination, pelvic examination, and examination of vaginal fluid for pH ≥4.7, amines, appearance, and saline/KOH microscopy and Gram stain. Women meeting Amsel criteria were given treatment for BV at that follow-up visit. Women whose Nugent score was subsequently read as consistent with BV7–10 were given treatment for BV at their next scheduled follow-up visit. The treatment code was broken only after all cases had completed follow up, outcomes at all visits had been classified according to Nugent score and Amsel criteria, and data analyses comparing the two arms for presence of BV (persistence or recurrence) were completed.
For purposes of analyses, the diagnosis of BV at baseline was based on a Nugent score of 7 to 10 plus at least three Amsel criteria: vaginal fluid pH ≥4.7 and positive amine test, plus either homogeneous vaginal discharge or clue cells representing more than 20% of cells in vaginal fluid.
Two definitions of “cure” for the two metronidazole-containing treatments were considered at 10 to 21 days after initiation of treatment: cure by “usual” criteria, defined to be absence of BV by widely reported criteria (<3 Amsel criteria or Nugent score <7); and cure by FDA guidance criteria, defined on the basis of finding none of the four Amsel criteria (clinical cure) and a Nugent score of <4. Because we asked women to return at 14 days after initiating treatment, we could not examine the outcomes at 21 to 30 days, as recommended by the FDA Draft Guidance, but we also analyzed data of all women returning for first follow up at any time after treatment. We then examined the effect of each treatment on each of the individual components of the Amsel criteria and on the components of Nugent’s score.
Comparisons between arms at the first follow-up visit were performed by Fisher’s exact tests.
We also examined whether any outcomes at the first follow up predicted subsequent recurrence of BV by both Amsel and Nugent criteria by analyzing data from second and third follow-up visits among women who lacked BV by both criteria at the first follow-up visit. Variables collected at first follow up, which were considered for prediction of subsequent recurrence of BV, included: 1) overall cure by FDA guidance and 2) individual components of these criteria for cure, including pH <4.7, negative amine test, absence of Gardnerella, Prevotella, or Mobiluncus morphotypes on Gram stain of vaginal fluid, presence of Lactobacillus morphotypes, and Nugent score <4. Fisher exact tests were used to assess whether these components of cure at the first follow-up visit were predictive of subsequent development of BV by both Amsel and Nugent criteria at the second follow-up visit. Survival analysis, which included both second and third follow-up visits, was used to confirm that results were similar when all visits were analyzed.
Analyses were performed in S-Plus 2000 professional release 2, 1999.
Of 184 women randomized, 151 had BV by both Amsel and Nugent criteria at baseline, of whom 138 (91%) returned for at least one follow-up visit (108 within 10–21 days), 120 (79%) for a second, and 75 (50%) for a third follow-up visit. The median (range) interval from baseline visit was 15 days (range, 10–132 days) to the first follow-up visit, 56 days (range, 38–218 days) to the second follow-up, and 127 days (range, 82–298 days) to the third follow-up visit.
Rates of cure per the FDA Guidance criteria were 35% (18 of 51) in those given metronidazole/nystatin ovules versus 26% (15 of 57) in those given metronidazole gel (P = 0.4) (Table 2).
Examining the specific components that make up the FDA Guidance criteria for cure revealed that metronidazole/nystatin ovules (delivering 500 mg metronidazole per night) were more effective than metronidazole gel (delivering 37.5 mg metronidazole per night) in reducing Gardnerella or Prevotella morphotypes to undetectable by Gram stain (P = 0.01), in eliminating clue cells to less than 20% of all vaginal epithelial cells (P <0.01), and in eliminating amines (P <0.01) at 10 to 21 days. Mobiluncus morphotypes were uncommon, but considering all first follow-up visits rather than only those ≤21 days, ovules were also more effective at reducing Mobiluncus morphotypes to undetectable (P = 0.06, data not shown).
However, ovules and gel treatments did not differ significantly in terms of effecting the reappearance of Lactobacillus morphotypes in Gram stain of vaginal fluid: at 10 to 21 days, Lactobacillus morphotypes were observed in 65% in the ovule arm versus 51% in the gel arm (P = 0.2), and 45% versus 30% had Lactobacillus morphotypes >30/100× microscopic field in these arms, respectively (P = 0.1). The two treatments also resulted in no significant difference in vaginal pH <4.7 seen in 59% of patients given ovules and 42% of patients given gel (P = 0.1). Absence of Lactobacillus morphotypes in Gram stain of vaginal fluid at 10 to 21 days was significantly correlated with pH ≥4.7 (P <0.001, Table 3). There was no difference between the two arms at 10 to 21 days in prevalence of homogeneous discharge consistent with BV, but this was uncommon in both arms.
In the absence of the criteria for BV at the first follow-up visit, we saw no evidence that failure of Lactobacillus morphotypes to appear on vaginal Gram stain predicted subsequent development of BV. Combining both arms of the study, among those without BV as defined by both Amsel and Nugent criteria at the first follow-up visit (or treated again with metronidazole at that visit), BV by both Amsel and Nugent criteria was detected at the second follow-up visit in 13 (22%) of 60 with Lactobacillus morphotypes versus six (22%)-of 30 without Lactobacillus morphotypes seen at the first follow-up (relative risk [RR] = 0.9, 95% confidence interval [CI] = 0.4–2.2); in 11 (22%) of 49 with pH <4.7 versus eight (20%) of 41 with pH ≥4.7 at the first follow up (RR = 0.9, 95% CI = 0.4–2.0); in 14 (19%) of 75 with no Gardnerella morphotypes versus 5 (33%) of 15 with such morphotypes at the first follow-up (RR = 0.6, 95% CI = 0.2–1.3); and in four (13%) of 30 who met FDA Guidance criteria for BV cure versus 15 (25%) of 60 who did not meet these criteria at the first follow up (RR = 0.5, 95% CI = 0.2–1.5). Similarly, no significant differences were noted for subanalyses within either arm, although the numbers of subjects within each arm were small. Survival analyses of both second and third follow-up visits yielded similar results to those of the second follow-up visits.
Follow-up at the interval recommended in FDA Guidance criteria did not increase the prevalence of Lactobacillus morphotypes on vaginal fluid Gram stain. Examining only first follow-up data, prevalence was 57% in 108 who returned for first follow up in 10 to 21 days (our follow-up interval) and 61% in 18 returning in 21 to 30 days (FDA Guidance follow-up interval). In addition, we saw no evidence that longer follow-up was associated with a higher likelihood of detecting Lactobacillus morphotypes. Of 64 women who did not meet both Amsel and Nugent criteria for BV or receive treatment for BV at 10 to 21 days, and who also returned for follow-up at 42 to 90 days, prevalence of Lactobacillus morphotypes was actually lower at the later visit, having fallen from 66% at 10 to 21 days to only 48% at 42 to 90 days.
In this study, intravaginal ovules containing 500 mg metronidazole and 100,000 U nystatin ovules were more effective than gel delivering 37.5 mg metronidazole in eliminating abnormal bacterial morphotypes and the related clue cells and amines associated with BV, but were not significantly more effective as measured by enhanced detection of vaginal lactobacilli as assessed by Gram stain. Absence of Lactobacillus morphotypes on Gram stain and pH ≥4.7 were highly correlated with one another and were commonly seen in both arms after treatment. The absence of a substantial difference between the two treatments in effecting an increase in vaginal lactobacilli as detected by Gram stain and in frequency of abnormal vaginal pH at 10 to 21 days explained why the ovules were not more effective than the gel in reducing the number of Amsel criteria present to 0 or the Nugent score to <4, even though they were significantly more effective in reducing the Amsel criteria to <3 and the Nugent score to <7.
The 1998 FDA Draft Guidelines for demonstrating cure for BV have not been widely adopted since they were published. We hypothesize that this may be because many researchers have been unaware of them and because cure rates using the guidelines are so poor. Whether the FDA Draft Guidelines should be widely adopted is important to consider. In our study, the FDA criteria for cure give dramatically lower rates than the “usual” criteria for cure. Agreement on what to use must be reached so that studies may be compared with one another. However, which is better? Our study demonstrates that the FDA criteria for cure could completely miss differences between regimens with markedly different efficacies against Gardnerella, clue cells, and amines when there is no substantial difference in promoting rapid vaginal recolonization with lactobacilli. In this example, we would have seen no difference between the two treatments if we had relied solely on cure defined by FDA Draft Guidance criteria. Our observation that combining both treatment arms, absence of Lactobacillus morphotypes at the first posttreatment follow up did not predict recurrence of BV between the first and second follow-up visits suggests that the early return of lactobacilli and resulting in lowering of vaginal pH may not be an essential component of a durable cure. If this proves to be the case, then cure as more usually defined based on Amsel criteria or Nugent score may give more useful information than does cure as defined by FDA Guidance. However, longer posttreatment follow-up of larger numbers of patients would be required to examine this question.
Other studies have reported higher prevalences of Lactobacillus morphotypes on Gram stain after treatment with standard doses of intravaginal or oral metronidazole.9,27 Most of our patients returned between 10 and 21 days, rather than in the 21- to 30-day window recommended in the 1998 FDA Draft Guidance for the test of cure visit. However, we saw no evidence that a later first follow-up visit would have increased the prevalence of Lactobacillus morphotypes seen. A lack of significant increase in prevalence of Lactobacillus morphotypes on Gram stain with posttreatment follow up longer than 21 days has also been seen in other studies. Agnew et al. found 94% and 91% with Lactobacillus morphotypes on Gram stain at >35 days after treatment with intravaginal metronidazole gel and oral metronidazole, respectively, but in the same study, prevalence after each treatment was already high (83%) at ≤21 days.27 In asymptomatic women diagnosed with BV by Nugent criteria, Schwebke found that 14 days after initiating treatment with metronidazole gel, 38% of 26 women had pH ≤4.5, and only 36% of 25 of the same women had pH ≤4.5, 2 or more weeks later.23 In fact, we have not found published evidence for a significant increase in Lactobacillus morphotypes among women followed for longer than 21 days. The prevalences of Lactobacillus morphotypes in baseline vaginal Gram-stained smears in BV treatment studies have varied in the literature from 5%28 to 68%.29 In our study of 151 subjects with BV by both Nugent and Amsel criteria, only 3% had Lactobacillus morphotypes detected in vaginal fluid at baseline.
Limitations of this study include the variability of elapsed time between visits, related in part to aggressive efforts to bring back participants who had missed earlier appointments. Although 91% of women returned for at least one follow-up visit, only 72% did so within 21 days after initiating treatment. Results of comparisons between arms were very similar whether all first follow-up visits were considered or only those within the first 21 days. Other limitations include the lack of Lactobacillus culture and inability to define whether hydrogen peroxide-producing lactobacilli were present. Although neither the FDA nor usual criteria for cure include culture for Lactobacillus or define whether lactobacilli seen are hydrogen peroxide-producing, vaginal cultures might have detected lower concentrations of lactobacilli not easily detected by Gram stain, and presence of H2O2-producing lactobacilli might predict subsequent recurrence even if presence of lactobacilli detected on Gram stain do not.
In conclusion, although the higher dosing of metronidazole delivered by the metronidazole plus nystatin ovules for topical therapy of BV as a 5-day course appears to have had a significantly greater impact than metronidazole gel in reducing the posttreatment persistence of Gardnerella morphotypes, amines, and clue cells, this regimen may not be sufficient to effect rapid vaginal recolonization by Lactobacillus in many women treated for BV. Future treatment trials should disaggregate outcomes that include impact on signs, abnormal bacterial flora, and recolonization by H2O2-producing lactobacilli. To choose between “usual” and FDA Guidance criteria for cure, researchers must consider whether criteria for cure of BV should include early restoration of Lactobacillus-dominant vaginal flora with resulting lowering of vaginal pH, like the FDA Guidance criteria do. Whether the FDA criteria should be widely adopted may depend on whether the longer-term prognosis for return of symptoms or complications of BV ultimately proves to be related to these indicators.
1. Amsel R, Totten PA, Spiegel CA, et al. Nonspecific vaginitis: Diagnostic criteria and microbial and epidemiologic associations. Am J Med 1983; 74:14–22.
2. Ahmed-Jushuf IH, Shahmanesh M, Arya OP. The treatment of bacterial vaginosis with a 3-day course of 2% clindamycin cream: Results of a multicentre, double blind, placebo controlled trial. Genitourin Med 1995; 71:254–256
3. Arrendondo JL, Higuero F, Hidalgo H, et al. Clindamycin vaginal cream versus oral metronidazole in the treatment of bacterial vaginosis. Arch AIDS STD Res 1992; 6:183.
4. Boeke AJ, Dekker JH, Van Eijk JT, et al. Effect of lactic acid suppositories compared with oral metronidazole and placebo in bacterial vaginosis: A randomized clinical trial. Genitourin Med 1993; 69:388–392.
5. Colli E, Landoni M, Parazzini F, et al. Treatment of male partners and recurrence of bacterial vaginosis: A randomized trial. Genitourin Med 1997; 73:267–270.
6. Ferris DG, Litaker MS, Woodward L, et al. Treatment of bacterial vaginosis: A comparison of oral metronidazole, metronidazole vaginal gel, and clindamycin vaginal cream. J Fam Pract 1995; 41:443–449.
7. Fischbach F, Petersen EE, Weissenbacher ER, et al. Efficacy of clindamycin vaginal cream versus oral metronidazole in the treatment of bacterial vaginosis. Obstet Gynecol 1993; 82:405–410.
8. Hanson JM, McGregor JA, Hillier SL, et al. Metronidazole for bacterial vaginosis. A comparison of vaginal gel vs oral therapy. J Reprod Med 2000; 45:889–896.
9. Hillier SL, Lipinski C, Briselden AM, et al. Efficacy of intravaginal 0.75% metronidazole gel for the treatment of bacterial vaginosis. Obstet Gynecol 1993;81:963–967
10. Livengood CH, Thomason JL, Hill GB. Bacterial vaginosis: Treatment with topical intravaginal clindamycin phosphate. Obstet Gynecol 1990; 76:118–123.
11. Livengood CH, McGregor JA, Soper DE, et al. Bacterial vaginosis: efficacy and safety of intravaginal metronidazole treatment. Am J Obstet Gynecol 1994; 170:759–764.
12. Livengood CH, Soper DE, Sheehan KL, et al. Comparison of once-daily and twice-daily dosing of 0.75% metronidazole gel in the treatment of bacterial vaginosis. Sex Transm Dis 1999; 26:137–142.
13. McCormack WM, Covino JM, Thomason JL, et al. Comparison of clindamycin phosphate vaginal cream with triple sulfonamide vaginal cream in the treatment of bacterial vaginosis. Sex Transm Dis 2001; 28:569–575.
14. Nayagam AT, Smith MD, Ridgway GL, et al. Comparison of ofloxacin and metronidazole for the treatment of bacterial vaginosis. Int J STD AIDS 1992; 3:204–207.
15. Parent D, Bossens M, Bayot D, et al. Therapy of bacterial vaginosis using exogenously-applied lactobacilli acidophilia and a low dose of estriol: A placebo-controlled multicentric trial. Arsneimittelforschung 1996; 46:68–73.
16. Paavonen J, Mangioni C, Martin M, et al. Vaginal clindamycin and oral metronidazole for bacterial vaginosis: A randomized trial. Obstet Gynecol 2000; 96:256–260.
17. Pulkkinen P, Saranen M, Kaaja R. Metronidazole combined with nystatin (vagitories) in the prevention of bacterial vaginosis after initial treatment with oral metronidazole. Gynecol Obstet Invest 1993; 36:181–184.
18. Schmitt C, Sobel JD, Meriwether C. Bacterial vaginosis: Treatment of clindamycin cream versus oral metronidazole. Obstet Gynecol 1992; 79:1020–1023.
19. Stein GE, Christensen SL, Mummaw NL, et al. Placebo-controlled trial of intravaginal clindamycin 2% cream for the treatment of bacterial vaginosis. Ann Pharmacother 1993; 27:1343–1345.
20. Vutyavanich T, Pongsuthirak P, Vannareumol P, et al. A randomized double-blind trial of tinidazole treatment of the sexual partners of females with bacterial vaginosis. Obstet Gynecol 1993; 82:550–554.
21. Eschenbach DA, Hillier S, Critchlow C, et al. Diagnosis and clinical manifestations of bacterial vaginosis. Am J Obstet Gynecol 1988; 158:819–828.
22. Nugent RP, Krohn MA, Hillier SL. Reliability of diagnosing bacterial vaginosis is improved by a standardized method of Gram stain interpretation. J Clin Microbiol 1991; 29:297–301.
23. Schwebke JR. Asymptomatic bacterial vaginosis: Response to therapy. Am J Obstet Gynecol 2000; 183:1434–1439.
24. Sobel J, Peipert JF, McGregor JA, et al. Efficacy of clindamycin vaginal ovule (3-day treatment) vs clindamycin vaginal cream (7-day treatment) in bacterial vaginosis. Infect Dis Obstet Gynecol 2001; 9:9–15.
25. Voorspoels J, Casteels M, Remon JP, et al. Local treatment of bacterial vaginosis with a bioadhesive metronidazole tablet. Eur J Obstet Gynecol Reprod Biol 2002; 105:64–66.
26. Sanchez S, Garcia P, Thomas KK, et al. Intravaginal metronidazole gel versus metronidazole plus nystatin ovules for bacterial vaginosis: A randomized controlled trial. Am J Obstet Gynecol 2004; 191(6): 1898–1906.
27. Agnew KJ, Hillier SL. The effect of treatment regimens for vaginitis and cervicitis on vaginal colonization by lactobacilli. Sex Transm Dis 1995; 22:269–273.
28. Bistoletti P, Fredricsson B, Hagstrom B, et al. Comparison of oral and vaginal metronidazole therapy for nonspecific bacterial vaginosis. Gynecol Obstet Invest 1986; 21:144–149.
29. Duff P, Lee ML, Hiller SL, et al. Amoxicillin treatment of bacterial vaginosis during pregnancy. Obstet Gynecol 1991; 77:431–435.