Journal of Antimicrobial Chemotherapy (988), -9 Haemophilus influenzae from four laboratories in one Canadian City Julie Righter" and Ingrid Luchstnger* 'Department of Microbiology, University of Toronto; b Mississauga Hospital, Toronto, Ontario, Canada Serotype, biotype and antimicrobial susceptibilities of 5 clinical isolates of Haemophilus influenzae from University, affiliated and community hospitals and a private laboratory were compared. For each drug, agar dilution susceptibility testing was compared to at least one other method (modified Kirby-Baucr and/or microdilution). Most isolates (86%) were non-typable, % were type b. Biotype II was most common (58%). The highest prevalence of serotype b (8%) was seen in the community hospital, which also had only 4% of all biotype III isolates. ^-Lactamase production ranged from % (private laboratory) to 5% (affiliated hospital); it was higher among type b (%), biotype II (7%), and from non-respiratory (6%) than respiratory sites (8%). 5% of 5 /Mactamase producers were found in the 4% of patients under age 6. Microdilution missed seven while agar dilution and disc diffusion detected all. All isolates were susceptible to cefamandole, cefuroxime, cotrimoxazole and chloramphenicol, 86%, 98%, 99% and 7% to ampicillin, cefaclor, tetracycline and erythromycin respectively. Microdilution is unreliable for detection of ampiciuin resistance mediated by /Mactamase production. Introduction Despite the fact that for over a decade increasing resistance to antibiotics has been reported in Haemophilus influenzae, only /Mactamase testing is routinely perforrned in the majority of Canadian microbiology laboratories. Even for ampicillin, this test yields incomplete information by itself since both enzymatic and non-enzymatic mechanisms of resistance occur (Albritton et ah, 978). Susceptibility testing of fastidious organisms poses problems of standardization, accuracy and reproducibility. The resulting uncertainty as to which method is optimal has discouraged diagnostic laboratories from introducing a routine procedure. A relationship between susceptibility pattern and serotype is well established with serotype b, where a high percentage of ampicillin resistance is found (Doern et al., 986). Evidence suggests an inverse relation between urease and ornithine decarboxylase production and /Mactamase production among non-encapsulated strains. This may indicate an inability of strains producing these enzymes to acquire a transposon to determine /Mactamase production (Kamme, 98). Biotyping H. influenzae strains according to Kilian's scheme has yielded information about the virulence of certain biotypes (Oberhofer & Back, 979). Correspondence to: Dr J. Righter, Microbiology Department, Toronto East General Hospital, 8 Coxwell Avenue, Toronto, Ontario M4C E7, Canada Downloaded from http://jac.oxfordjournals.org/ at Pennsylvania State University on September 8, 6 O5-745/88/WO +7 $./ 988 The British Society for Antimicrobial Chemotherapy
4 J. Righter and I. Locteiiiger We undertook a study of H. influenzae strains isolated concurrently from four different practice settings in the Toronto area. Our purpose was to compare susceptibility testing by disc diffusion, agar dilution and broth microdilution as performed in routine diagnostic laboratories, and to correlate sensitivity patterns with /Mactamase production, biotype, serotype, body site and practice setting. Methods The organisms were consecutive isolates from four diagnostic laboratories over a six month period. All were identified by standard bacteriological methods, including the porphyrin test. /?-Lactamase testing was performed by the rapid iodometric paper method (Gittens, 978). Both disc diffusion and agar dilution susceptibility tests were performed using GC agar base (Gibco) supplemented with mg/ haemin and mg/ nicotinamide adenine dinucleotide. All plates were held at 4 C and used within 48 h of preparation. Controls consisted of /Mactamase negative and positive strains of H. influenzae, Staph. aureus ATCC 9 and Escherichia coli ATCC 59. Concentration breakpoints were selected according to the National Committee for Clinical Laboratory Standards (985). Disc susceptibility testing was performed according to the modified Kirby-Bauer method as recommended by the National Committee for Clinical Laboratory Standards (984). All categories other than susceptible were classified as resistant. Agar dilution susceptibility testing was performed with a Steers replicator. The final inoculum was -5 xlo cfu/spot for co-trimoxazole, -5x* cfu/spot for all other antibiotics. Plates were incubated in a moist environment for 8 h at 5 C aerobically. Any visible growth was interpreted as resistant. All antibiotics were laboratory standards obtained from manufacturers: cefamandole and cefaclor from Eli Lilly Table I. Laboratory characteristics and /Mactamase production Laboratory characteristics WCH: 5 beds university hospital TEGH: 6 beds community hospital, affiliated with university MH: 65 beds community hospital BML: private medical laboratory ambulatory patients Number of isolates 9 75 6 49 Number (%) /Mactamase positive 6 (8) 4(5) 5(4) () Downloaded from http://jac.oxfordjournals.org/ at Pennsylvania State University on September 8, 6 Total 5 5 (4)
H. influenza* in Canada 5 Canada Ltd., ampicillin from Ayerst Laboratories, tetracycline from Pfizer Canada Inc., chloramphenicol from Parke-Davis Canada Inc., cefuroxime from Glaxo Canada Inc., erythromycin from Abbott Laboratories Ltd. and trimethoprim and sulphamethoxazole from Burroughs Wellcome Inc. Concentrations ranged from -5 to 6 mg/ in doubling dilutions, for co-trimoxazole -5 mg/ trimethoprim plus 9-5 mg/ sulphamethoxazole, to 6 mg/ trimethoprim plus 4 mg/ sulphamethoxazole. Microdilution MICs were determined with Microscan MIC panels with a final inoculum of 4 cfu/well in double-strength tryptic soy broth with % Fildes digest, incubated for 8 h at 5 C in 5% carbon dioxide. Each method of susceptibility testing was carried out for all isolates in a single laboratory, microdilution in the community hospital and the other two techniques in the affiliated hospital. Discrepancies in susceptibility results were defined as differences of two or more doubling dilutions. MIC so s and MIC 9O s were calculated according to the method of Reed & Muench (98): a mathematical formula is used to find endpoints between the two critical dilutions in each case. Biotyping was done by the Minitek system, as described by Back & Oberhofer (978). Serotyping was done by slide agglutination with Difco antisera (polyvalent, a, b, c, d, e and f). The significance of difference between observed and expected figures was assessed by chi-square testing. Results The 5 clinical isolates of H. influenzae were collected over the same time period from four laboratories in the same city reflecting different practice settings (Table I). Serotype and biotype distribution are shown in Table II. The community hospital showed the highest prevalence of serotype b (/6, 8%) and only /7 (4%) of all Serotype a b c d e f nt Biotype I II III IV V VI VII VIII Table II. ^-Lactamase and specimen source according to serotype and biotype Number (% of total) 6 (-4) 7 (-8) (-4) (-4) (-4) 4 (85-6) 4 (9-6) 44 (57-6) 7 (8-) (-) 5(-) (-8) (-4) (-4) /J-lactamase positive number (%} 6() (4-) 8(-) (4-) (-9) (7-) blood CSF Specimen source eye ear other 6 55 4 8 5 7 5 respiratory 4 6 47 6 98 45 4 Downloaded from http://jac.oxfordjournals.org/ at Pennsylvania State University on September 8, 6
6 J. Righter and L Lochsinger Table IIL Age distribution according to /Mactamase production Age distribution month- -5 6-5 >5 Newboms years years years years Total /f-lactamase positive 4 4 6 5 negative 7 5 67 5 biotype III isolates (i > <5%). There was no relationship between biotype distribution and antibiotic resistance. ^-Lactamase production was highest among type b (6/6, %) and biotype III (/7, 7%) but these differences were not statistically significant. Over half (8/5, 5%) were found in the 6 (4%) of patients under age 6 (Table III). Only 4/7 (8%) of isolates from respiratory sites were /Mactamase positive as compared with /8 (6%) from non-respiratory sites (i > <5) (Table IV). The highest incidence was from specimens from the eye (8/6, 9%). Cefamandole, cefuroxime, tetracycline, co-trimoxazole and chloramphenicol inhibited 9% of isolates at concentrations of -5 mg/l (Table V). The geometric mean MIC for cefaclor (calculated from agar dilution values) was higher for /Mactamase positive (4-5 mg/l) than negative (-5 mg/l) isolates. /7-Lactamase producers were identified as resistant to ampicillin in all cases by disc diffusion and agar dilution, but microdilution missed 7/5 (%). Five /Mactamase negative isolates were resistant by microdilution but susceptible by both other methods. The overall error rate by microdilution was /5 (4%). No ampicillin resistant, /Mactamase negative isolates were detected. All isolates were susceptible to cefamandole, cefaclor and tetracycline by disc. With cefamandole, there were 4 discrepancies between agar dilution and microdilution, the latter always producing a higher MIC, but only once did this value exceed 8 mg/l with a resulting change in interpretation. By agar dilution, six isolates were resistant to cefaclor; half of these were /Mactamase positive. With tetracycline, four discrepancies occurred between agar dilution and microdilution, two in each direction, all resulting in interpretive differences. Table IV. /Mactamase production according to specimen source Downloaded from http://jac.oxfordjournals.org/ at Pennsylvania State University on September 8, 6 Specimen source Blood CSF Eye Ear Other Respiratory Number of isolates 6 5 8 7 Number (%) /Mactamase positive () () 8(9) () () 4(8)
H. influenzae in Canada 7 Drug Ampicillin Cefamandole Cefuroxime Cefaclor Tetracycline Erythromycin Co-trimoxazole Chloramphenicol Table V. Antimicrobial susceptibilities disc diffusion 86 99 99 96 'Calculated from agar dilution values. % Susceptible agar dilution 85 98 99 7 MIC 5 5 5-8 5 7 5 5 Minimal inhibitory concentration i(mg/) 5 MIVDQ 5-5 5 4-5 -4 5 5 geometric mean" 76 5-54 -6 5-5 5 Erythromycin results by disc were all in the susceptible range except for one intermediate zone size. Only 68 (7%) would be classified as susceptible using the present NCCLS breakpoint of <-5 mg/. Eight isolates required 4 mg/ for inhibition; zone sizes for these ranged from 9 to 5 mm. Only one minor discrepancy occurred with cefuroxime: one isolate with MIC <-5 mg/ by agar dilution showed a zone size in the intermediate range. Results with co-trimoxazole were difficult to interpret: isolates were resistant by disc but nine of these were inhibited by ^-5 mg/ by agar dilution and microdilution. In three cases, microdilution MIC was significantly higher than agar dilution. Discussion A survey of 7 laboratories across Canada covering the years 979 to 98 revealed a national rate of /Mactamase production of % with no regional differences; the rate in 98 had climbed to 8-8% (Scheifele, 98). In Ontario (976-8) the rate was 4-8% (Bannatyne et al., 985). Both Canadian surveys included only blood and CSF isolates. A large collaborative study in the United States including 56 isolates from all clinical sites showed a rate of 5-%, similarly with no geographic resistance trends (Doern et al, 986). In our study, there were 5 isolates (4%) that produced /Mactamase with highest rates among type b and biotype III. A higher rate occurred in younger patients, as noted by others (Doern et al., 986). The rate in the affiliated hospital was much lower than all the others. Disc diffusion reliably detects ampicillin resistance due to /Mactamase production (Campos, Garcia-Tornel & Sanfeliu, 984) but misses non-enzymatic resistance (Mendelman et al., 986); our results confirmed the former but could not assess the latter. Agar dilution yields results comparable with disc diffusion (Woolfrey et al., 984) and microdilution (Campos et al., 984). In our study, agar dilution detected all /Mactamase producers while microdilution missed seven. Up to 7% of isolates are resistant to ampicillin without producing /Mactamase (Bell & Plowman, 98) and these strains are fully virulent (Markowitz, 98). Five of our 5 /Mactamase negative isolates were resistant to ampicillin by microdilution, even on Downloaded from http://jac.oxfordjournals.org/ at Pennsylvania State University on September 8, 6
8 J. Righter and I. Loctaringer repeat testing, but this could not be confirmed by either of the other methods. A ng ampicillin disc has been reported to be better than the standard /xg disc for detection of non-enzymatic ampicillin resistance (Mendelman et al., 986). Correlation between reduced susceptibility and /Mactamase production has been reported with cephalosporins (cephalothin, cefazolin, cefamandole, cefaclor and cefoperazone) and co-trimoxazole (Thirumoorthi, Kobos & Dajani, 98; Jones & Thornsberry, 985; Flournoy, 985; Lapointe & Beyeler, 985). We found only cefaclor to be less active. There were major discrepancies in crythromycin results between agar dilution and disc diffusion. This is largely an artefact resulting from inconsistencies between the approved NCCLS standards for the two methods: the zone size >8mm correlates with an MIC of <mg/l, but the breakpoint by dilution methods is -5mg/l. Conversely, co-trimoxazole disc diffusion results indicated higher resistance rates than MIC determinations by either of the other methods. Biotype I is clearly associated with invasive disease (Oberhofer & Back, 979; Retter & Bannatyne, 98); in our study, four of five blood/csf isolates were biotype I. Our rate of /Mactamase production was highest among biotype III isolates, at variance with a report of clustering of /Mactamase producers among biotype I and II strains (Albritton et al., 978). Our results agree with others in that biotype II predominated among eye isolates (Albritton et al., 978; Oberhofer & Back, 979). The great majority of isolates were non-typablc. All but of these were from eye or respiratory sites and the rate of /Mactamase production was %. Of the typable isolates, serotype b was most common and it included all blood and CSF isolates. The /Mactamase positive rate was higher than for any other serotype. Serotype b represented 8% of the isolates from the community hospital, nearly three times the overall prevalence. The chromogenic cephalosporin method of /Mactamase testing misses one type of /Mactamase (Rubin et al., 98). /?-Lactamase testing by the acidometric or iodometric method must be performed routinely; susceptibility testing by disc diffusion or agar dilution is recommended, at least for isolates causing invasive disease. Acknowledgements We thank Dr M. Vearncombe for her valuable contributions. References Albritton, W. L., Penncr, S., Slaney, L. & Brunton, J. (978). Biochemical characteristics of Haemophilus influemae in relationship to source of isolation and antibiotic resistance. Journal of Clinical Microbiology 7, 59-. Back, A. E. & Oberhofer, T. R. (978). Use of the Minitek system for biotyping Haemophilus species. Journal of Clinical Microbiology 7, -. Bannatyne, R. M., Toma, S., Cheung, R. & Hodge, D. (985). Antibiotic susceptibility of blood and cerebrospinal fluid isolates of Haemophilus inftuenzae. Journal of Antimicrobial Chemotherapy 5, 87-9. Barry, A. L., Jones, R. N. & Thornsberry, C. (985). Reliability of cefaclor, cefazolin, cefamandole, and cephalothin disks to predict susceptibility of Enterobacteriaceae, Staphylococcus species, and Haemophilus influenzae. American Journal of Clinical Pathology 84,64-8. Downloaded from http://jac.oxfordjournals.org/ at Pennsylvania State University on September 8, 6
Downloaded from http://jac.oxfordjournals.org/ at Pennsylvania State University on September 8, 6