J Clin Pathol 1983;36:476-478

Minimal criteria for the identification of Gardnerella vaginalis isolated from the

JLS JOLLY From the Public Health Laboratory, Ipswich Hospital, Heath Road, Ipswich IP4 SPD

SUMMARY Vaginal swabs were examined for the presence of Gardnerella vaginalis. Of 294 iso- lates with appropriate colonial and cellular morphology subjected to an identification procedure, 203 (69%) were identified as G vaginalis. The 91 isolates not identified as G vaginalis were differentiated by their inability to ferment starch, cause diffuse /8 haemolysis on human agar or hydrolyse hippurate. Other tests, often used in the identification of G vaginalis, were found to be insufficiently specific. Failure to ferment starch coexisted with failure to cause /8 haemolysis and/or hydrolyse hippurate. The starch fermentation test may therefore be omitted. The tests for 8 haemolysis and hippurate hydrolysis, being relatively simple to perform and interpret, are considered indispensable for the accurate identification of G vaginalis in the service laboratory.

The pathogenic significance of Gardnerella vaginalis Material and methods is still in doubt. Since early work by Gardner and Dukes,' numerous conflicting reports have CULTURE appeared, with the result that clinicians and clinical High vaginal swabs were submitted to the laboratory laboratories alike hold differing opinions on the from General Practitioners, family planning and need to isolate and identify this organism. Difficul- gynaecological clinics. The swabs, transported to the ties in identifying G vaginalis, and in defining the laboratory in Stuart's transport medium, were cul- clinical condition associated with it, have hindered tured on (Columbia agar base the clarification of its role. Accurate identification (Oxoid) with 10% horse blood heated to 80°C), and and exact definition is obviously important when Schaedler agar, (Oxoid) (with horse blood added searching for an association between organism and (5%)). Chocolate plates were incubated in 7% CO2 disease. Methods of bacterial identification are of in air for 48 h and the Schaedler plates for 48 h in an little practical use if they are complex. Likewise anaerobic environment. Films from small, smooth, shortened identification schemes are of little use if entire colonies showing no haemolysis on chocolate they fail to differentiate similar . In the past or Schaedler agars were Gram-stained. All small the methods most widely used to identify G vaginalis Gram-positive, -negative and variable pleomorphic have been those suggested by Dunkelberg2 which bacilli were subjected to an identification procedure. emphasise colonial morphology and fermentation Only predominant or abundant growths were tests. A dissecting microscope is required to examined. The presence of other bacteria was also examine the colonies and the fermentation tests are noted. Presumed G vaginalis, found as part of a time-consuming and difficult to perform. More mixed bacterial flora, was subcultured on starch recent studies3 45 have suggested different serum agar.6 All specimens were also screened for identification schemes, in which the choice of Candida and . criteria and complexity vary considerably. The pur- pose of this study was to evaluate selected criteria to IDENTIFICATION ascertain the minimum number of tests needed by a Isolates were tested for starch fermentation by cul- routine laboratory for the identification of G vag- ture on starch serum agar. This plate was also used inalis. to detect inhibition by 3% hydrogen peroxide2 and catalase production; growth was scraped from the agar surface and touched onto the end of a capillary Accepted for publication 1 December 1982 tube filled with 3% hydrogen peroxide. The evolu- 476 Minimal criteria for the identification of Gardnerella vaginalis isolated from the vagina 477 tion of bubbles was noted. Culture on human blood GRAM FILM agar' incubated in 7% CO2 in air for 48 h, revealed Cellular morphology varied with the medium on diffuse ( haemolysis. The rapid method of Hwang8 which the organism was grown. Bacilli from choco- was used to test for hippurate hydrolysis. Disc tests late agar were slightly larger and more pleomorphic for sensitivity determined by the comparative than those from Schaedler agar. On this latter method,9 to sulphonamide (100 ,g), bacitracin (5 medium the morphology resembled the control IU) and (50 ,ug) were done on organism (NCTC 10287) in being predominantly Schaedler agar with lysed horse blood incubated coccobacillary. The Gram reaction varied consider- anaerobically for 24 h. Isolates of Gram-positive, ably with support medium and age of culture. -negative and variable small pleomorphic bacilli which showed starch fermentation, inhibition by 3% IDENTIFICATION hydrogen peroxide, negative catalase reaction,2 (3 The features of the 294 isolates are shown in Table haemolysis on human blood agar, hippurate hyd- 1. All selected isolates were catalase-negative and rolysis,'0 resistance to sulphonamide5 and sensitivity were inhibited by 3% hydrogen peroxide, although to bacitracin,'0 were identified as G vaginalis. the zone sizes varied. All isolates were resistant to Gram-positive, -negative and variable small sulphonamide and sensitive to bacitracin. The reac- pleomorphic bacilli failing to meet all of these tion patterns to starch fermentation, 8 haemolysis criteria were called G vaginalis-like organisms and hippurate hydrolysis are shown in Table 2. (GvLOs). Haemophilus vaginalis NCTC 10287 served as the Discussion control organism. In this study G vaginalis was isolated satisfactorily on chocolate and Schaedler agars. Using these Table 1 Collective features of294 isolates. media other potential bacterial pathogens could be detected, so that additional isolation media were not Characteristic No (%) necessary. The reaction of selected iso- Starch fermentation 263 (90) lates varied considerably with support media and Catalase-negative 294 (100 age of culture. Consequently cellular morphology Inhibition by 3% hydrogen peroxide 294 (100 /3 haemolysis on human blood agar 219 (75) and size were more helpful in identification than the Hippurate hydrolysis 251 (85) Gram reaction ifself. Indeed as the cell wall struc- Resistance to sulphonamide (100,g) 294 (100 Sensitivity to bacitracin (5 IU) 294 (100 ture is typical of neither Gram-positive nor Gram- Sensitivity to metronidazole (50 ,g) 259 (88) negative organisms," it seems inappropriate to adopt the Gram reaction as one of the identification criteria; that reaction was found to be of more use in excluding other bacteria than including possible G Table 2 Isolates (n = 294) reactions to three main vaginalis. criteria Gram-positive, -negative and variable small colonial Starch Hippurate 3 haemolysis No pleomorphic bacilli produced uniform fermentation hydrolysis (human blood) characteristics on the media used. Of 294 isolates examined all were found to be inhibited by 3% hyd- + + + 203 + + - 32 rogen peroxide, catalase-negative, resistant to sul- + - + 15 phonamides and sensitive to bacitracin. As only 203 _ _ 13 + 14 (69%) isolates met other criteria for identification as + _ 16 G vaginalis, these four tests seem insufficiently _ _ + 1 _ + + 0 specific for routine identification of that organism. Total - 30 - 43 -75 294 259 (88%) isolates were sensitive to metronidazole Total + 264 + 251 + 219 (50,ug), but as in vitro resistance is reported to be a common feature of this organism3 12 13 this test was excluded from the identification criteria. Separation of isolates into G vaginalis and GvLOs was based Results entirely upon starch fermentation, (3 haemolysis on human blood agar and hippurate hydrolysis. Exami- Two hundred and ninety-four isolates of Gram- nation of these characteristics amongst the 294 iso- positive, -negative and variable small pleomorphic lates reveals that 44 isolates were negative for two bacilli were examined; 203 (69%) were subse- or more characteristics and that no organism lysing quently identified as G vaginalis. human blood and hydrolysing hippurate failed to 478 Jolly ferment starch. All isolates failing to ferment starch specific . Ain J Obsiet Gynzaecol 1955;69:962-76. also failed to cause f8 haemolysis and/or hydrolyse Dunkelberg WE, Skaggs R, Kellogg DS. Method for isolationi and identification of Coryniebacierius'vagitnale (Haetiophilus hippurate. Consequently given ,B haemolysis and vaginialis). Appl Microbiol 197t); 19:47-52. hippurate hydrolysis, the importance of which is 3Wells JI, Goei SH. Rapid identification of Cor,tyebacteriumn vag- obvious,'" there seems little advantage in a starch imiale in non-purulent vaginitis. J Cliti Paihol 1981 ;34:917-2t). fermentation test. 4 Ison CA, Dawson SG, Hilton J, Csonka GW, Easmon CSF. Comparison of culture and microscopy in the diagnosis of Fermentation of maltose and dextrose, in addition Gardnerella vaginalis infection. J Clitn Paihol 1982;35:55(-4. to starch have figured prominently in many 'Bailey RK, Voss JL, Smith RF. Factors affecting isolation and identification procedures. 14-1 However, Dunkel- identification of Haemophilus vaginialis (Coryniebacieriutn vag- berg'8 has recently suggested that testing for acid inale). J Clin Microbiol 1979;9:65-71. hIslam AKMS. Starch serum agar-a differential medium for the production from dextrose and maltose in addition to isolation of Corytnebacterium vaginale (Haeonophilus vag- starch is superfluous, an organism which acidifies inalis). J Clitn Pathol 1977;30:291-2. starch being likely to acidify dextrose and maltose 'Greenwood JR, Pickett MJ, Martin WJ, Mack EG. Haernophilus an identification vaginalis ( vaginiale): method for isolation also. Ison et al4 have suggested and rapid biochemical identification. Health Lab Sci scheme that does not include fermentation tests. 1977;14:102-6. However it does not use the hippurate test, but 8Hwang M, Ederer GM. Rapid hippurate hydrolysis method for relies on the catalase and oxidase tests. This study presumptive identification of Grp B streptococci. J Clitn Mic- provides no evidence as to the usefulness of an oxid- robiol 1975;1: 114-5. 'Stokes EJ, Waterworth PM. sensitivity tests by diffu- ase test, but the catalase test was not found to be sion methods. ACP Broadsheet 55, 1966 (Revised 1972). helpful. Another identification scheme suggested by Greenwood JR, Pickett MJ. Salient features ofHaemophilus vag- Wells and Goei3 also dispenses with fermentation inialis. J Clin Microbiol 1979;9:200-4. and indeed with any subculture of isolates. Greenwood JR, Pickett MJ. Transfer of Haemophilus vagitnalis tests, Gardner & Dukes to a new Gardnerella: Gardnerella Unfortunately, as a consequence, it relies heavily vaginalis (Gardner & Dukes) comb nov Ini J Syst Bacteriol upon the recognition of characteristic colonial mor- 1980;30:170-8. phology, for which a dissecting microscope is neces- Balsdon MJ, Taylor GE, Pead L, Maskell R, Corynebacterium vaginale and vaginitis: a controlled trial of treatment. Lanicet sary. In addition to this disadvantage it has also been 1980;i:501-5. sugested that the so-called characteristic colonial 3Piot P, Van Dyck E, Goodfellow M, Falkow S, A taxonomic morphology is neither unique to, nor a feature of, all study of Gardnerella vaginalis (Haemophilus vaginalis), Gard- strains.'8 ner & Dukes 1955. J Gen Microbiol 1980;119:373-96. an identification 4 McCormack WM, Hayes CH, Rosner B, et al. Vaginal colonisa- The findings of this study suggest tion with Corynebacterium vaginale (Haemophilus vaginalis). J scheme that relies upon two tests, ,3 haemolysis on Infect Dis 1977;136 No 6:740-5. human blood agar and hippurate hydrolysis. Both ,s Pheifer TA, Forsyth PS, Durfee MA, Pollock HM, Holmes KK. tests are simple to perform and easy to interpret and Nonspecific vaginitis. Role of Haemophilus vaginalis and treatment with metronidazole. N Engl J Med 1978;298: are therefore ideal for the service laboratory. 1429-34. IBramley HM, Dixon RA. Jones BM. Haemophilus vaginalis I wish to express my sincere thanks to Dr HWK Fell (Corynebacterium vaginale, Gardnerella vaginalis) in a family for his advice on the preparation of the manuscript. I planning clinic population. Br J Vener Dis 1981;57:62-6. the critic- 7Kinghorn GR, Jones BM, Chowdhury FH, Geary 1. Balanopos- am particularly grateful for constructive thitis associated with Gardnerella vaginalis infection in men. isms made by Dr R Maskell and the persistent help Br J Vener Dis 1982;58:127-9. and encouragement given to me by Dr JVT 18 Dunkelberg WE. Isolation and identification of Corynebacterium Gostling. Finally thanks to Mrs E Smith for typing vaginale. J Am Med Technol 1978;40:9-11. the manuscript.

References Requests for reprints to: JLS Jolly, Public Health 'Gardner HL, Dukes CD. Haemophilus vaginalis vaginitis. A Laboratory, Ipswich Hospital, Heath Road, Ipswich IP4 newly defined specific infection previously classified "non- 5PD, England.