Identification of Viridans Streptococci Isolated from Clinical Specimens

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, May 1982, p /82/ $02.00/0 Vol. 15, No. 5 Identification of Viridans Streptococci Isolated from Clinical Specimens KATHRYN L. RUOFF'* AND LAWRENCE J. KUNZ1' 2 Francis Blake Bacteriology Laboratories, The Massachusetts General Hospital, Boston, Massachusetts 02114,1 and Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, Massachusetts Received 19 October 1981/Accepted 5 January 1982 Of 532 strains of viridans streptococci isolated from clinical specimens, 517 were identified by using a scheme based on the work of Facklam (R. R. Facklam, J. Clin. Microbiol. 5: , 1977). The strains were distributed among nine of the species described by Facklam and an additional group of physiologically homogeneous streptococci not recognized in Facklam's scheme. The method for identification involves a battery of 10 tests that employ both conventional media and commercially prepared disks. The identification of most of the isolates was accomplished within 48 h. The characteristics of viridans streptococcal species have been described by a number of authors (2, 3, 7, 14). Identification of these socalled viridans strains cannot be accomplished with serological methods but requires the performance of many time-consuming physiological tests. Identification schemes that employ fewer tests have been proposed by Waitkins et al. (21) and Facklam (7). One of the objectives of the study reported here was to determine whether a shortened scheme based on that of Facklam (7) could be used successfully to identify viridans streptococci isolated in routine cultures of clinical specimens. Although fewer tests are employed in the shortened schemes (7, 21), long incubation times are nevertheless required, owing to the types of media utilized. Accordingly, a second objective of our study was to attempt to devise an identification scheme which uses relatively rapid methods. Recently, micromethods have been employed for the identification of viridans streptococci (11, 17, 22). In the present study, a combination of conventional media and Minitek (BBL Microbiology Systems, Cockeysville, Md.) disks was used to determine the physiological activities of viridans strains. These methods allowed for identification of most strains within 48 h. We were especially concerned with the identification of Streptococcus milleri, owing to the increasing awareness of the pathogenic role of this microorganism (4, 12, 14-16, 18). In Facklam's scheme of classification (7), this species is represented by Streptococcus MG-intermedius and Streptococcus anginosus-constellatus. This group of organisms is popularly referred to as S. milleri in clinical literature. We have therefore attempted to stress the identity of S. MG-intermedius, S. anginosus-constellatus, and a third group of strains described in this report and elsewhere (1) with S. milleri. MATERIALS AND METHODS Bacterial strains. The strains were alpha or nonhemolytic streptococci isolated from clinical specimens at the Massachusetts General Hospital. Each isolate was examined for the presence of Lancefield antigens A through 0 by the enzymatic extraction method of Watson et al. (23) and microprecipitin testing. Antisera were purchased from Wellcome Research Laboratories, Beckenham, England, or from Difco Laboratories, Detroit, Mich. Most of the strains could not be serologically identified, although a minority reacted with Group F, H, K, or 0 antisera. Cultures of streptococci were frozen in horse blood for prolonged storage. For use in this study, they were streaked onto horse blood agar plates which were incubated at 35 C in the presence of 3 to 5% CO2. Growth from the plates was suspended in sterile water (phenol red broth method) or Lombard-Dowell broth (Minitek and conventional methods), and the suspensions were used as inocula for the various biochemical tests. Stock strains of known organisms were obtained from the Massachusetts General Hospital Bacteriology Laboratory culture collection. Minitek procedure. Strains were tested for the ability to produce acid from mannitol, lactose, and raffinose with Minitek disks by the method of Setterstrom et al. (17), except 0.1 ml of cell suspension (0.5, McFarland standard) was added to each well. Since Minitek disks containing inulin were not available, 0.05 ml of a filter-sterilized solution containing 10.0% inulin and 0.005% phenol red (ph 7.0) was added to wells to provide for test substrate. Incubation of the plates was carried out at 37 C in an anaerobic GasPak atmosphere (BBL Microbiology Systems). Results were read after the addition of 2 drops of 0.025% phenol red (ph 7.2) to each well (17). A yellow color 920

2 VOL. 15, 1982 IDENTIFICATION OF VIRIDANS STREPTOCOCCI 921 TABLE 1. Percentages of isolates of different groups of viridans streptococci displaying positive results in various tests No. of % Positive by following physiological test: isolates Mannitol Lactose Raffinose Inulin Esculin Arginine S. mutans S. salivarius S. sanguis I S. mitis S. sanguis II S. milleri groups S. MG-intermedius S. anginosus-constellatus Unidentified urine isolates S. uberis S. morbillorum indicated a positive reaction, whereas any shade of orange or red was interpreted as a negative reaction. Conventional methods. Phenol red broth containing filter-sterilized carbohydrates at final concentrations of 1.0%o was used as the conventional method of determining the ability of the organisms to produce acid from mannitol, lactose, raffinose, and inulin. Setterstrom et al. (17) noted that comparable results with conventional methods were achieved when either phenol red broth or the bromocresol purple-containing broth recommended by Facklam (5) was employed. Since phenol red is the indicator used in the Minitek system, phenol red broth was chosen for the conventional method in this study. Phenol red broths were incubated for 2 weeks at 37 C, with reactions recorded at 48 h and 7 and 14 days. Esculin hydrolysis was determined as described by Vera and Dumoff (19). Arginine hydrolysis was assayed by the method of Niven et al. (13). Hippurate hydrolysis was assayed by the method of Facklam et al. (8). The presence of urease was determined by the method of Vera and Power (20). The esculin, arginine, hippurate, and urea broths were incubated at 37 C for 48 h, although positive urease reactions were always noted within less than 24 h. Mitis salivarius agar (GIBCO Diagnostics, Madison, Wis.) was used to determine levan production. Plates were incubated at 35 C in the presence of 3 to 5% CO2 for 48 h. Levan producers form large, domed colonies on this medium. The plates were not examined for dextran production, since this characteristic was not used as a differential trait in our identification scheme. Litmus milk (GIBCO Diagnostics) was incubated at 37 C for up to 1 week, although positive reactions usually occurred within 48 h. Conventional tests on certain unidentified urine isolates were performed by the methods of Facklam (5, 6). Carbohydrates were purchased from either Difco Laboratories or Sigma Chemical Co., St. Louis, Mo., and dehydrated media were obtained from Difco Laboratories. RESULTS High levels of agreement were observed between results obtained with the Minitek and conventional methods for determining acid production from carbohydrates. One hundred strains were tested in duplicate by both methods. The agreement between the two methods was 98% for lactose, raffinose, and inulin and 100% for mannitol. The data are comparable to those obtained by Setterstrom et al. (17). The Minitek results are available in 48 h, in contrast to incubation times of up to a week or longer required by the conventional methods. Accordingly, the Minitek procedure was used in conjunction with other tests to identify 532 viridans streptococci. Table 1 shows the percentages of each group of viridans strains which produced acid from mannitol, lactose, raffinose, and inulin and were capable of hydrolyzing arginine and esculin. The data are similar to those reported by Facklam (7), except for the group of organisms referred to as unidentified urine isolates. These strains cannot be placed in any of the species described by Facklam, but they resemble certain S. milleri strains described by Ball and Parker (1). Owing to the significant number of these isolates we encountered, a separate class under the S. milleri heading was established to accommodate these strains. The unidentified urine isolates and 15 strains that remained unclassified will be discussed in more detail subsequently. In addition to the tests shown in Table 1, other physiological activities of the isolates were assayed to identify them correctly at the species

3 922 RUOFF AND KUNZ level. Hippurate hydrolysis is a characteristic of Streptococcus uberis and Streptococcus acidominimus. Only the three isolates identified as S. uberis possessed the ability to hydrolyze hippurate. Litmus milk reduction is accomplished by all species except Streptococcus morbillorum, and only the two strains identified as this species were unable to reduce litmus milk. Since the occurrence of these organisms in human clinical specimens appears to be infrequent (reference 7 and Table 1), strains belonging to these species have been separated from the major groups of viridans streptococci in the tabulated data presented here. The mannitol-negative, inulin-positive species Streptococcus sanguis I and Streptococcus salivarius can often be differentiated on the basis of arginine hydrolysis, but additional tests are needed to differentiate between the relatively rare arginine hydrolysis-negative S. sanguis I strains and S. salivarius. In this study, the ability to produce levan on mitis salivarius agar and the production of urease were used as differentiating characteristics for these two species. Of the S. salivarius strains in Facklam's study, 55% were able to produce levan (7). About 50% of S. salivarius strains produce urease, and this activity has been reported to be unique for S. salivarius (9, 10). Seven of the strains identified as S. salivarius in this study produce both levan and urease, one produced only levan, and one produced only urease. None of the 108 strains identified as S. sanguis I gave positive results in either of these tests, suggesting that urease and levan production are useful characteristics for differentiating between S. sanguis I and S. salivarius. Of the other isolates possessing urease activity, two were identified as S. MG-intermedius and one as Streptococcus mitis. Either the production of urease is not unique to S. salivarius, or these three ureasepositive strains may be inulin-negative variants of S. salivarius. The three strains also failed to produce levan. Thus, only strains which pro- J. CLIN. MICROBIOL. duced urease or levan or both and fermented inulin were classified as S. salivarius in our study. Unidentified urine isolates. The group of organisms referred to as unidentified urine isolates were all mannitol, lactose, raffinose, and esculin positive and able to reduce litmus milk. All showed no activity with inulin, hippurate, and urea and failed to produce levan. Thus, these organisms do not fit into any of the species mentioned in Facklam's scheme. Fifty-three of these organisms were available for further study and were subjected to tests recommended by Facklam (5, 6). In heart infusion broths containing bromocresol purple and various carbohydrates, this group of strains displayed homogeneous reactions. All produced acid from mannitol, lactose, raffinose, melibiose, sucrose, salicin, and trehalose. None displayed activity on arabinose, glycerol, sorbitol, or inulin. All of the strains grew well in 2% NaCl broth, 11 strains grew slowly in 4% NaCl broth, and none tolerated 6.5% NaCl broth. None of the strains hydrolyzed starch or showed evidence of levan or dextran production on sucrose agar plates. All strains were nonhemolytic on horse blood agar when incubated anaerobically in a GasPak atmosphere. The growth of all strains was strongly stimulated by additional CO2 in the incubation atmosphere. Each of the isolates produced small, grayish, morphologically similar colonies on blood agar after a 24-h incubation period in the presence of added CO2. In the absence of added C02, the organisms grew slowly, with tiny colonies appearing only after 48 h of incubation. The majority of these organisms were isolated from urine cultures. Twenty (37%) were reactive with group F antiserum. Facklam (7) described five unidentified strains of viridans streptococci which appear to be identical to the 54 unidentified urine isolates observed in this study. Facklam's strains displayed a similar biochemical profile, and four of the isolates reacted with group F antiserum. Our unidentified urine isolates bear a striking resemblance to certain S. milleri strains described by Ball and Parker (1) which tended to be nonhemolytic, and stimulated by C02, and displayed a wide sugar fermentation pattern which included acidification of raffinose and either melibiose or mannitol or both. Unclassified strains. Fifteen of the strains examined in this study could not be placed into any of the species described by Facklam or the unidentified urine isolates group described here. Of nine strains which produced acid from mannitol, none displayed overall reaction patterns that were typical of any previously described mannitol-fermenting groups. Three strains which fermented only raffinose were negative in the litmus milk reduction test, thus excluding them from the S. anginosus-constellatus group. Two isolates produced acid only from lactose and inulin and were also positive in the arginine and urea hydrolysis tests. These strains thus possessed both the arginine hydrolysis activity of S. sanguis I and the urease activity of S. salivarius. Neither strain produced levan. One strain produced acid only from inulin and failed to reduce litmus milk; therefore, it could not be accommodated by Facklam's scheme of classification. Distribution of species among clinical specimens. Of the 532 strains examined in this study, 187 (35%) were isolated from blood cultures, 150 (28%) from urine cultures, 131 (25%) from wound cultures, 46 (9%) from body fluids, 10

4 VOL. 15, 1982 TABLE 2. Numbers of species of viridans streptococci displaying serological reactions No. in following Species serological group: F H K 0 S. mutans S. salivarius S. sanguis I S. mitis S. sanguis II S. milleri groups S. MG-intermedius S. anginosus-constellatus Unidentified urine isolates Unclassified S. uberis S. morbillorum (2%) from necropsy specimens, and 8 (2%) from miscellaneous cultures. The most common blood culture isolates were S. mitis, S. sanguis I, and S. sanguis II, which accounted, respectively, for 26, 23, and 23% of the blood culture isolates. The major groups isolated from urine cultures were S. MG-intermedius (accounting for 30% of the urine isolates) and the unidentified urine isolates (31%). Thirty-seven percent of the wound culture isolates were identified as S. MG-intermedius and 24% as S. sanguis I. Among isolates from body fluids, S. sanguis I (46%) and S. MG-intermedius (28%) were the predominant species. S. MG-intermedius was overall the most commonly isolated species, followed by S. sanguis I. Table 2 shows the distribution of groups F, H, K, and 0 antigens, the only ones detected among the streptococci examined. Most of the group F strains were classified as S. MG-intermedius or as unidentified urine isolates. S. sanguis I accounted for most of the group H strains, and S. sanguis II and S. mitis were the major species displaying the group K antigen. DISCUSSION The methods utilized in this study were designed to be a concise, rapid series of tests for the identification of viridans streptococci. Miniaturized methods (Minitek) were employed for four of the tests (acid production from mannitol, lactose, raffinose, and inulin), and conventional media were used for the remaining six tests IDENTIFICATION OF VIRIDANS STREPTOCOCCI 923 (esculin, arginine, and hippurate hydrolysis, levan and urease production, and reduction of litmus milk). An incubation time of 48 h seemed to be sufficient to identify the most frequently occurring species of viridans streptococci. Strains which were mannitol, lactose, raffinose, inulin, and esculin negative (3.0% of the isolates in this study) sometimes required more than 48 h of incubation for a positive reaction in litmus milk, leading to their classification as S. anginosus-constellatus. Although no strains of S. acidominimus were identified, the hippurate hydrolysis test would separate this species from S. morbillorum and S. anginosus-constellatus. The value of the hippurate hydrolysis test for identification of routinely isolated viridans streptococci of human origin is questionable in view of the absence of S. acidominimus and the low incidence of S. uberis in the present study and the extremely low frequencies of these two species in Facklam's work (7). Setterstrom et al. (17) investigated the use of Minitek disks for the determination of biochemical characteristics of viridans streptococci. They reported that an incubation time of 5 to 7 days was required for accurate determination of arginine and esculin hydrolysis. Although Minitek disks are more convenient to use and store than tubed media, the incubation time for these two tests far exceeds the recommended 48-h incubation time for arginine and esculin broth media. Setterstrom and co-workers were unable to perform miniaturized inulin tests owing to the unavailability of prepared inulin disks. The phenol red-inulin solution described in this paper afforded a rapid miniaturized method for the determination of acid production from inulin. Holloway et al. (11) reported successful use of the Minitek system for identification of viridans streptococci. Eleven different Minitek disks were used, along with 6 additional conventional tests. These authors reported that the identification of viridans strains could be accomplished in 18 h. Their scheme incorporated a relatively large number of tests, and some of the substrates used have little differential value for viridans streptococcal species. The scheme for identification of viridans isolates reported here is a relatively rapid and concise method for the identification of viridans streptococci in clinical laboratories. The percentages of positive test results observed in this study approximate those of Facklam (7), suggesting that our methods can be successfully substituted for techniques that require extended incubation periods. The identification of S. sanguis I and S. salivarius is complicated by the existence of strains of S. sanguis I that are arginine hydrolysis negative. Our inclusion of the urease and levan production tests was designed to increase

5 924 RUOFF AND KUNZ J. CLIN. MICROBIOL. TABLE 3. Modification of Facklam's (7) key for the identification of viridans streptococci Physiological reaction Species Acid production from mannitol and lactose Hippurate hydrolyzed... S. uberis" Hippurate not hydrolyzed Acid production from inulin... S. mutans No reaction on inulin, acid produced from raffinose, esculin hydrolyzed Unidentified urine isolates (S. milleri) Acid production from lactose, but not mannitol Acid production from inulin Arginine hydrolyzed, no levan or urease production... S. sanguis I Arginine not hydrolyzed, levan and/or urease produced... S. salivarius No reaction on inulin Esculin hydrolyzed... S. MG-intermedius (S. milleri) Esculin not hydrolyzed Acid produced from raffinose... S. sanguis II No reaction on raffinose... S. mitis No reaction on mannitol or lactose Hippurate hydrolyzed... S. acidominimusa Hippurate not hydrolyzed Esculin hydrolyzed and/or acid produced from raffinose and litmus milk reduced... S. anginosus-constellatus (S. milleri) Esculin not hydrolyzed, no acid produced from raffinose, litmus milk not reduced... S. morbillorum" a Species occurring infrequently in clinical bacteriology cultures. the probability of correctly identifying these two species. However, it would still be possible to misidentify a rare levan- and urease- negative S. salivarius strain as S. sanguis I with the scheme presented here. The situation is further complicated by the observation of the two unclassified strains which possessed both arginine hydrolysis and urease activities. In most cases, however, the arginine hydrolysis, urease, and levan production tests should serve to differentiate between S. sanguis I and S. salivarius strains. Our unidentified urine isolates provide a good example of the disagreement involved in classification of the viridans streptococci. Although Facklam attempted to divide the viridans streptococci into well-defined species, our unidentified urine isolates cannot be accommodated by his scheme. However, these organisms frequently occur and appear to be perfectly good counterparts of the less rigidly defined S. milleri species of workers other than Facklam. Table 3 presents a key for identification of viridans streptococci. The key is a modification of that of Facklam (7). Although the media and tests employed are somewhat different from those of Facklam, all of his species can be identified with the modified key. In addition, provision is made for the unidentified urine isolates, which are referred to as S. milleri. It should be remembered that the S. MG-intermedius and S. anginosus-constellatus species of Facklam correspond, respectively, to the lactose-positive and lactose-negative S. milleri of other authors (7). The hippurate hydrolysis test is included in the key, but it is our view that this test may be omitted in identifying routinely isolated viridans streptococci of human origin. The distribution of viridans species noted in this study is not comparable to that reported by Facklam (7). Facklam examined close to 2.5 times more strains, and many of these were obtained from sources (dental plaque, sputum, throat) not encountered in the work reported here. The strains examined in our study were collected between 1974 and They do not include all of the viridans streptococci isolated in our laboratory during that period. The organisms we studied were present in either abundant amounts or in pure culture in the specimens from which they were isolated and consequently were thought to be of clinical significance. It is interesting to note that, in both studies, S. MGintermedius was the most commonly isolated species, accounting for 19% of Facklam's strains and 25% of the strains examined in this work. S. MG-intermedius, along with S. anginosus-constellatus (S. milleri), accounted for 31% of our strains, by far the most commonly recognized and identified organisms isolated from the clinical specimens encountered in our study. If the

6 VOL. 15, 1982 IDENTIFICATION OF VIRIDANS STREPTOCOCCI 925 unidentified urine isolates are included in the S. milleri group, the frequency rises to 41%. The importance of S. milleri as a human pathogen has been noted previously (4, 12, 14-16, 18), and the results of this study support that view. ACKNOWLEDGMENTS We thank Jane Kissling for her helpful advice, Katherine S. Pigott for technical assistance, and Kathleen Sullivan for aid in preparing the manuscript. LITERATURE CITED 1. Bail, L. C., and M. T. Parker The cultural and biochemical characters of Streptococcus milleri strains isolated from human sources. J. Hyg. 82: Carlsson, J A numerical taxonomic study of human oral streptococci. Odontol. Revy. 19: Colman, G., and R. E. 0. Williams Taxonomy of some human viridans streptococci, p In L. W. Wannamaker and J. M. Matsen (ed.), Streptococci and streptococcal diseases; recognition, understanding, and management. Academic Press, Inc., New York. 4. DeLouvois, J Bacteriological examination of pus from abscesses of the central nervous system. J. Clin. Pathol. 33: Facklam, R. R Recognition of group D streptococcal species of human origin by biochemical and physiological tests. Appl. Microbiol. 23: Facklam, R. R Comparison of several laboratory media for presumptive identification of enterococci and group D streptococci. Appl. Microbiol. 26: Facklam, R. R Physiological differentiation of viridans streptococci. J. Clin. Microbiol. 5: Facklam, R. R., J. F. Padula, L. G. Thacker, E. C. Wortham, and B. J. Sconyers Presumptive identification of group A, B, and D streptococci. Appl. Microbiol. 27: Feltham, R. K. A A taxonomic study of the genus Streptococcus, p In M. T. Parker (ed.), Pathogenic streptococci. Reedbooks Ltd., Surrey, England. 10. Hardie, J. M., and G. H. Bowden Physiological classification of oral viridans streptococci. J. Dent. Res. 55(special issue A): Hoiloway, Y., M. Schaareman, and J. Dankert Identification of viridans streptococci on the Minitek miniaturised differentiation system. J. Clin. Pathol. 32: Murray, H. W., K. C. Gross, H. Masur, and R. B. Roberts Serious infections caused by Streptococcus milleri. Am. J. Med. 64: Niven, C. F., Jr., K. L. Smiley, and J. M. Sherman The hydrolysis of arginine by streptococci. J. Bacteriol. 43: Parker, M. T., and L. C. Ball Streptococci and aerococci associated with systemic infection in man. J. Med. Microbiol. 9: Parker, M. T., and L. C. Ball Streptococcus milleri as a pathogen for man, p In M. T. Parker (ed.), Pathogenic streptococci. Reedbooks Ltd., Surrey, England. 16. Poole, P. M., and G. Wilson Streptococcus milleri in the appendix. J. Clin. Pathol. 30: Setterstrom, J. A., A. Gross, and R. S. Stanko Comparison of Minitek and conventional methods for the biochemical characterization of oral streptococci. J. Clin. Microbiol. 10: Shlaes, D. M., P. I. Lerner, E. Wolinsky, and K. V. Gopalakrishna Infections due to Lancefield Group F and related streptococci (S. milleri, S. anginosus). Medicine 60: Vera, H. D., and M. Dumoff Culture media, p In J. E. Blair, E. H. Lennette, and J. P. Truant (ed.), Manual of clinical microbiology, 1st ed. American Society for Microbiology, Washington, D.C. 20. Vera, H. D., and D. A. Power Culture media, p In E. H. Lennette, A. Balows, W. J. Hausler, Jr., and J. P. Truant (ed.), Manual of clinical microbiology, 3rd ed. American Society for Microbiology, Washington, D.C. 21. Waitkins, S. A., L. C. Ball, and C. A. M. Fraser A shortened scheme for the identification of indifferent streptococci. J. Clin. Pathol. 33: Waitkins, S. A., L. C. Ball, and C. A. M. Fraser Use of the API-ZYM system in rapid identification of a and non-haemolytic streptococci. J. Clin. Pathol. 33: Watson, B. K., R. C. Moeflering, Jr., and L. J. Kunz Identification of streptococci: use of lysozyme and Streptomyces albus filtrate in the preparation of extracts for Lancefield grouping. J. Clin. Microbiol. 1: