Abstract
A prospective study of erythromycin and clindamycin resistance was performed with 304 consecutive group B streptococci (GBS) isolates. According to two automated susceptibility testing systems, Vitek-1 and Vitek-2, and double-disk agar diffusion, 79.9% were susceptible to both erythromycin and clindamycin. However, for macrolide-lincosamide-streptogramin B-inducible isolates, the accuracies of the Vitek-1 and Vitek-2 systems were 5.6 and 94.4%. In light of these results, we recommend that GBS be routinely tested using the double-disk diffusion method.
Group B streptococci (GBS) are a frequent cause of peripartum maternal and neonatal bacteremia (2, 9). In Canada, women are screened for GBS carriage at 35 to 37 weeks of gestation (10). If they are found to be carrying GBS during the intrapartum period, they are treated with penicillin. Penicillin-allergic women are typically treated with a macrolide or clindamycin (1, 8).
Erythromycin resistance rates in GBS have been increasing globally. In Ontario, the rate was 8% in the period from 1995 to 1998, 18% in 1999, and 20% in this study (3). Studies in the United States reported rates of 12% in 1990 to 1995, increasing to 20% in 1995 to 2000 (6). Even higher rates have been seen in Korea (40% in 1998) and Taiwan (46% in 1998 to 2000) (4, 12). Consequently, the Centers for Disease Control and Prevention now recommends laboratory testing for macrolide resistance in GBS for penicillin-allergic patients (2). Many clinical laboratories in North America use automated susceptibility testing systems, such as the Vitek or MicroScan system, to test for macrolide resistance in GBS.
The genetic basis for this resistance is either the modification of the ribosome encoded by the erm genes, as in the macrolide-lincosamide-streptogramin B (MLSB) phenotype, or the efflux of macrolide antibiotics by the mefA-encoded gene pump, as in strains with the M phenotype (5, 13). The MLSB phenotype confers resistance to the macrolides, lincosamides, and streptogramin B antibiotics.
We conducted a prospective blinded study to determine the accuracy of the Vitek-1 and Vitek-2 systems for detection of erythromycin resistance, as well as inducible and constitutive clindamycin resistance, with GBS. The standardized double-disk diffusion test was used as the “gold standard” method.
A total of 304 consecutive GBS isolated from vaginal swabs from the greater Toronto area were identified using the Prolex Strep Grouping Latex kit (Pro-Lab, Richmond Hill, Ontario, Canada). The Vitek-1 GPS-105 card (bioMérieux, Hazelwood, Mo.), Vitek-2 AST-P526 card (bioMérieux), and double-disk tests were set up by separate technologists involved in this study.
Standardized double-disk agar diffusion tests were performed as described previously (11). The method uses Mueller-Hinton agar containing 5% sheep blood with erythromycin and clindamycin disks placed exactly 15 mm edge to edge and incubated in 5% CO2 (antimicrobial disks and plates were obtained from Oxoid (Nepean, Ontario, Canada). Zone diameters were measured, and the GBS were categorized as susceptible or resistant as per the NCCLS supplement M100-S13 (7).
Macrolide resistance phenotypes were interpreted in the erythromycin-resistant GBS isolates as follows: (i) the M phenotype was erythromycin resistant and clindamycin susceptible, with no evidence of induced clindamycin resistance visible in the overlap zone; (ii) the MLSB-inducible phenotype was erythromycin resistant, with clindamycin resistance visible in the overlap zone as evidenced by a blunted D-shaped zone of inhibition; and (iii) the MLSB-constitutive phenotype was clearly resistant to both erythromycin and clindamycin (11).
According to standardized double-disk diffusion, 243 (79.9%) of the 304 GBS isolates were susceptible to both erythromycin and clindamycin (Table 1). The remaining 61 isolates (20.1%) showed different patterns of resistance: four isolates (1.3%) had the M phenotype, 18 isolates (5.9%) had the MLSB-inducible phenotype, and 39 isolates (12.8%) had the MLSB-constitutive phenotype. The overall rates of erythromycin and clindamycin resistance were 20.1 and 18.8%, respectively.
TABLE 1.
Accuracy of Vitek-1 and Vitek-2 susceptibility results versus standardized double-disk diffusion test results with 304 GBS isolates
| Double-disk test result (no. of isolates, %) | Vitek-1 accuracy (%) | Vitek-2 accuracy (%) |
|---|---|---|
| Erythromycin and clindamycin susceptible (243, 79.9) | 100 | 100 |
| M phenotype (4, 1.3) | 100 | 0 |
| MLSB-inducible phenotype (18, 5.9) | 5.6 | 94.4 |
| MLSB-constitutive phenotype (39, 12.8) | 71.8 | 87.2 |
| Total (304, 100) | 90.8 | 96.7 |
When the susceptibility results for the Vitek-1 and Vitek-2 systems were compared to the double-disk results, there was 100% agreement for the 243 susceptible isolates (Table 1). Both the Vitek-1 and Vitek-2 systems were able to detect the four M-phenotype strains, but the expert rules in the Vitek-2 system overrode these results and incorrectly reported all four strains as clindamycin resistant, resulting in four major errors (1.3%).
For the 18 MLSB-inducible GBS isolates, the Vitek-1 and Vitek-2 systems had accuracies of 5.6 and 94.4%, respectively (Table 1). While both systems were only able to correctly identify one isolate as resistant to clindamycin (Table 2), the Vitek-2 expert rules overrode these results and correctly reported 15 isolates as clindamycin resistant. For the 39 MLSB-constitutive strains, the accuracies of the Vitek-1 and Vitek-2 systems were 71.8 and 87.2%, respectively (Table 1). Both systems were able to correctly identify most isolates as being resistant to both erythromycin and clindamycin (Table 2). In addition, the Vitek-2 expert rules overrode the results for two isolates, correctly reporting them as clindamycin resistant. Still, a significant number of very major errors were accumulated by the Vitek-1 system in the detection of strains with the MLSB resistance phenotype. There were fewer errors with the Vitek-2 system due to the application of expert rules.
TABLE 2.
Vitek-1 and Vitek-2 results for 18 MLSB-inducible and 39 MLSB-constitutive isolatesa
| Type of isolate and system | No. of isolates with indicated Vitek erythromycin/clindamycin susceptibility
|
|||||
|---|---|---|---|---|---|---|
| S/S | S/R | I/S | I/R | R/S | R/R | |
| Inducible | ||||||
| Vitek-1 | 6† | 0 | 3† | 0 | 8† | 1 |
| Vitek-2 | 1† | 0 | 4* | 1 | 11* | 1 |
| Constitutive | ||||||
| Vitek-1 | 4† | 1† | 2† | 0 | 4† | 28 |
| Vitek-2 | 5† | 0 | 0 | 0 | 2* | 32 |
*, Vitek-2 expert rule applied; †, Very major error; S, susceptible; I, intermediate; R, resistant.
The sensitivities, specificities, and positive and negative predictive values for detecting erythromycin and clindamycin resistance are summarized in Table 3. Isolates identified as erythromycin intermediate by the Vitek systems were considered to be resistant for the purpose of these analyses, since most clinicians would not choose to treat an erythromycin-intermediate GBS with erythromycin. The negative predictive values of the Vitek-1 and Vitek-2 systems for detecting erythromycin resistance are 95.7 and 97.6%, respectively (Table 3). The negative predictive values for detection of clindamycin resistance are 90.1 and 97.6% for Vitek-1 and Vitek-2, respectively.
TABLE 3.
Sensitivities, specificities, and predictive values of Vitek-1 and Vitek-2 for detection of erythromycin and clindamycin resistance
| System and drug | % Sensitivity | % Specificity | Negative predictive value (%) | Positive predictive value (%) |
|---|---|---|---|---|
| Vitek-1 | ||||
| Erythromycin | 82.0 | 100 | 95.7 | 100 |
| Clindamycin | 52.6 | 100 | 90.1 | 100 |
| Vitek-2 | ||||
| Erythromycin | 90.2 | 100 | 97.6 | 100 |
| Clindamycin | 89.5 | 98.4 | 97.6 | 92.7 |
The overall accuracies of Vitek-1 and Vitek-2 were 90.8 and 96.7%, respectively. Both systems generated very major errors when categorizing MLSB-resistant GBS isolates: respectively, the Vitek-1 and Vitek-2 systems misidentified 28 (9.2%) and six (2.0%) erythromycin and clindamycin-resistant GBS as susceptible. The Vitek-2 also had four major (1.3%) and five minor (1.6%) errors. In particular, Vitek-1 demonstrated poor detection accuracies of 5.6 and 71.8% among MLSB-inducible and MLSB-constitutive GBS, respectively. The Vitek-2 system performed better than the Vitek-1 system, correctly identifying 94.4 and 87.2% of MLSB-resistant strains due to the use of expert rules for reporting final susceptibility results.
An erythromycin resistance rate of 20.1% (from this study) results in negative predictive values of 95.7% for Vitek-1 and 97.6% for Vitek-2. With clindamycin resistance at 18.8% (from this study), the negative predictive values for detection of clindamycin resistance are 90.1 and 97.6% for Vitek-1 and Vitek-2, respectively. Although the Vitek-2 system performed reasonably well, the Vitek-1 system performed poorly in detecting MLSB-resistant GBS. Such values would be unacceptable for sterile-site isolates with use of the Vitek-1 system. As resistance rates continue to increase, these negative predictive values will decrease. Thus, we recommend that GBS be routinely tested using the standardized double-disk diffusion methodology.
Acknowledgments
We thank MDS Laboratories for supplying some of the GBS strains and Sylvia Pong-Porter, Yana Rzayev, and all the unmentioned laboratory technologists at TML/MSH Department of Microbiology for their assistance in identifying the GBS.
REFERENCES
- 1.American College of Obstetricians and Gynecologists. 2003. Prevention of early-onset group B streptococcal disease in newborns. Int. J. Gynaecol. Obstet. 81:115-122. [DOI] [PubMed] [Google Scholar]
- 2.Centers for Disease Control and Prevention. 2000. Early-onset group B streptococcal disease—United States, 1998-1999. JAMA 284:1508-1510. [PubMed] [Google Scholar]
- 3.de Azavedo, J. C., M. McGavin, C. Duncan, D. E. Low, and A. McGeer. 2001. Prevalence and mechanisms of macrolide resistance in invasive and noninvasive group B streptococcus isolates from Ontario, Canada. Antimicrob. Agents Chemother. 45:3504-3508. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hsueh, P. R., L. J. Teng, L. N. Lee, S. W. Ho, P. C. Yang, and K. T. Luh. 2001. High incidence of erythromycin resistance among clinical isolates of Streptococcus agalactiae in Taiwan. Antimicrob. Agents Chemother. 45:3205-3208. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Luna, V. A., P. Coates, E. A. Eady, J. H. Cove, T. T. Nguyen, and M. C. Roberts. 1999. A variety of gram-positive bacteria carry mobile mef genes. J. Antimicrob. Chemother. 44:19-25. [DOI] [PubMed] [Google Scholar]
- 6.Murdoch, D. R., and L. B. Reller. 2001. Antimicrobial susceptibilities of group B streptococci isolated from patients with invasive disease: 10-year perspective. Antimicrob. Agents Chemother. 45:3623-3624. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.National Committee for Clinical Laboratory Standards. 2003. Performance standards for antimicrobial susceptibility testing. Document M100-S13 (M2). National Committee for Clinical Laboratory Standards, Wayne, Pa.
- 8.Schrag, S., R. Gorwitz, K. Fultz-Butts, and A. Schuchat. 2002. Prevention of perinatal group B streptococcal disease. Revised guidelines from CDC. Morb. Mortal. Wkly. Rep. Recomm. Rep. 51:1-22. [PubMed] [Google Scholar]
- 9.Schrag, S. J., S. Zywicki, M. M. Farley, A. L. Reingold, L. H. Harrison, L. B. Lefkowitz, J. L. Hadler, R. Danila, P. R. Cieslak, and A. Schuchat. 2000. Group B streptococcal disease in the era of intrapartum antibiotic prophylaxis. N. Engl. J. Med. 342:15-20. [DOI] [PubMed] [Google Scholar]
- 10.Society of Obstetricians and Gynaecologists of Canada and Canadian Paediatrics Society. 1997. Statement on the prevention of early-onset group B streptococcal infections in the newborn. J. Obstet. Gynaecol. Can. 19:751-758. [Google Scholar]
- 11.Tang, P., D. E. Low, S. Atkinson, K. Pike, A. Ashi-Sulaiman, A. Simor, S. Richardson, and B. M. Willey. 2003. Investigation of Staphylococcus aureus isolates identified as erythromycin intermediate by the Vitek-1 system: comparison with results obtained with the Vitek-2 and Phoenix systems. J. Clin. Microbiol. 41:4823-4825. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Uh, Y., I. H. Jang, G. Y. Hwang, K. J. Yoon, and W. Song. 2001. Emerging erythromycin resistance among group B streptococci in Korea. Eur. J. Clin. Microbiol. Infect. Dis. 20:52-54. [DOI] [PubMed] [Google Scholar]
- 13.Weisblum, B. 1985. Inducible resistance to macrolides, lincosamides and streptogramin type B antibiotics: the resistance phenotype, its biological diversity, and structural elements that regulate expression—a review. J. Antimicrob. Chemother. 16(Suppl. A):63-90. [DOI] [PubMed] [Google Scholar]