Evaluation of the PASCO Strep Plus Broth Microdilution Antimicrobial Susceptibility Panels for Testing Streptococcus pneumoniae and Other Streptococcal Species

Abstract

Antimicrobial resistance continues to increase worldwide among isolates of Streptococcus pneumoniae and other species of streptococci. Increasing rates of penicillin resistance, particularly in viridans group streptococci, and resistance to multiple classes of antimicrobial agents, including β-lactams, macrolides, and fluoroquinolones, in pneumococci have increased the importance of having accurate antimicrobial susceptibility testing results for guiding therapy. One commercial method of assessing resistance in streptococci is the PASCO Strep Plus panel. This broth microdilution-based method has recently been expanded to include a variety of newer antimicrobial agents. Therefore, we compared the results of the new PASCO Strep Plus panels for 26 antimicrobial agents against the results generated using the National Committee for Clinical Laboratory Standards (NCCLS) broth microdilution reference method for 75 pneumococci and 68 other streptococcal isolates. Only 4 (0.2%) very major errors (all with pneumococci and each with a different antimicrobial agent) were observed. There were 5 (0.3%) major errors observed with pneumococci (each with a different antimicrobial agent), but only 1 major error with nonpneumococcal streptococci. All of the very major and major errors resolved on retesting. Of the 65 (3.9%) and 17 (1.6%) minor errors observed with pneumococci and other streptococci, respectively, all were within 1 dilution of the broth microdilution reference MIC result. Thus, the PASCO Strep Plus panel has comparable accuracy to the NCCLS broth microdilution reference method.

Antimicrobial resistance continues to increase in Streptococcus pneumoniae (2, 3, 5, 7, 12, 13, 23, 26) and other species of streptococci worldwide (1, 4, 6, 11, 24, 28). Although there are abundant data documenting the growing problem of resistance in pneumococci, data on resistance patterns of viridans group streptococci and other streptococcal species are less common, since clinical laboratories often do not perform susceptibility testing unless the isolates are recovered from blood cultures or other normally sterile sites. Nonetheless, a number of reports document the growing trend of resistance, particularly among S. mitis, S. sanguis, and S. oralis isolates (4, 19, 28).

While the oxacillin disk screening test is still commonly used to determine the susceptibility of S. pneumoniae to penicillin and other β-lactam agents (9, 15), this test does not replace MIC methods. As noted in the National Committee for Clinical Laboratory Standards (NCCLS) guidelines (21), pneumococci producing zone diameters of ≤19 mm around a 1-μg oxacillin disk cannot be assumed to be resistant to penicillin. Rather, an MIC test is required, since many penicillin-susceptible pneumococci may produce zone diameters smaller than 20 mm (9). On the other hand, NCCLS guidelines (21) and studies at the Centers for Disease Control and Prevention (CDC) have indicated that neither ampicillin, oxacillin, nor penicillin disk tests work well for predicting β-lactam resistance with viridans group streptococci (unpublished observations). Thus, the emergence of penicillin resistance in viridans group streptococci can only be assessed using MIC methods. In fact, the emergence of resistance to penicillin, cephalosporins, macrolides, and, most recently, fluoroquinolones among S. pneumoniae (8) and other streptococcal species (4) has intensified the need to test not only penicillin but extended-spectrum cephalosporins and other non-β-lactam agents by MIC methods (3, 5, 18, 26, 27). Several commercial products have been developed to meet this need, including frozen and dried broth microdilution panels (14, 17, 22, 29) and agar gradient diffusion (16).

Recently, PASCO laboratories expanded their Strep Plus system for determining the susceptibility patterns of S. pneumoniae and nonpneumococcal streptococci to include additional antimicrobial agents. To evaluate the accuracy of the revised PASCO system (which is a frozen, broth microdilution plate containing a full range of antimicrobial agent dilutions), we compared the results of the new PASCO Strep Plus MIC panels to the results generated using the NCCLS reference broth microdilution method (20).

MATERIALS AND METHODS

Bacterial isolates.

A total of 143 streptococcal isolates (75 pneumococci and 68 other streptococcal isolates) were tested. This included 51 S. pneumoniae isolates from the strain collection of the CDC (29) and 69 organisms from the collection of Becton Dickinson Microbiology Services (BDMS) (Cockeysville, Md.). The organisms from BDMS included 12 S. pneumoniae isolates, 12 S. pyogenes isolates, 5 S. salivarius isolates, 5 S. sanguis isolates, 5 S. sobrinus isolates, 5 S. gordonii isolates, 5 S. mutans isolates, 3 S. mitis isolates, 3 S. oralis isolates, 3 S. agalactiae isolates, 3 S. constellatus isolates, 3 S. crista isolates, 2 S. sanginosus isolates, 1 S. downei isolate, 1 S. intermedius isolate, and 1 Streptococcus group G isolate. Organisms were identified using standard biochemical methods (25). Pneumococci were tested for optochin susceptibility and bile solubility, and unusual pneumococcal isolates that showed aberrant reactions were confirmed by serotyping. Streptococci were tested using Voges-Proskauer reagent, arginine, esculin, mannitol, sorbitol, and pyrrolidonyl arylamidase. Additional testing for identification of unusual species was undertaken at BDMS using an array of biochemical tests. Each isolate was subcultured twice on Trypticase soy agar containing 5% sheep blood (BDMS) prior to testing. Nine fresh clinical isolates (defined as organisms isolated from a clinical specimen that had been on an agar plate or slant for less than 7 days and never frozen) of pneumococci and 11 other streptococcal isolates from a variety of specimen types were provided by Robert Jerris, Grady Memorial Hospital, Atlanta, Ga. Three additional fresh pneumococcal isolates were sent to the CDC from other sources and were included in the study, increasing the total number of fresh isolates to 23. Two quality control strains, S. pneumoniae ATCC 49619 (20) and S. pneumoniae ATCC 51422 (formerly called CS101) (27), were tested daily with the isolates. The MIC results for these quality control organisms were within expected ranges.

Testing method.

Each bacterial isolate was tested using three unique commercial broth microdilution panels containing doubling dilutions of a total of 26 antimicrobial agents (provided by PASCO laboratories). The isolates were also tested using three reference broth microdilution panels containing dilutions of the same 26 antimicrobial agents prepared at the CDC. Panels were stored at −70°C and allowed to warm to room temperature before use. For the PASCO panels, organisms from growth on a Trypticase soy agar blood agar plate incubated at 35°C for 16 to 20 h were suspended in 6 ml of 0.85% saline to the turbidity of a 1.0 McFarland standard. A 1.5-ml aliquot of this suspension was transferred to 12.5 ml of SP Blood Supplement. The suspension was inverted 8 to 10 times, and the inoculum was poured into the inoculum tray. The three PASCO panels were inoculated with samples from this suspension using a new disposable inoculator for each panel. The final inoculum, as determined by colony counts from the growth control well, was approximately 10⁵ CFU/well (each well contains 100 μl). Panels were stacked no more than three high, covered with a plastic tray, and incubated in ambient air at 35°C for 20 to 24 h. The inoculum for the CDC panels was prepared using the same 1.0 McFarland suspension but was diluted to equal a 0.5 McFarland with 0.85% saline. Two milliliters was aseptically transferred into 38 ml of saline and vortexed. The 40-ml suspension was then poured into the inoculum tray. Three CDC panels were inoculated using a new disposable inoculator each time (each well contains 100 μl). CDC panels were stacked no more than three high and placed into a self-sealing plastic bag in ambient air at 35°C for 20 to 24 h. The final inoculum for the CDC panels, as determined by colony counts, was approximately 3 × 10⁴ CFU/well.

Reproducibility studies were also performed using 10 isolates tested in triplicate with three separate organism suspensions on three separate test days using PASCO panels only. Thus, each isolate was tested a total of nine times in the reproducibility studies. PASCO (test method) and CDC (reference method) panels were read visually and results were recorded on individual data collection sheets. The PASCO data management system was not evaluated during this study. Any organism with test results exhibiting very major or major errors was reread to control for reading errors. Data were initially entered into an EpiInfo Database, and the data set was later converted to a SAS (Cary, N.C.) version 6.12 data set for analysis.

Definitions.

A very major error occurs when the reference result is resistant and the test result is susceptible. A major error occurs when the reference result is susceptible and the test result is resistant. A minor error is defined as one in which the reference result is resistant or susceptible and the test result is intermediate or when the reference result is intermediate and the test result is susceptible or resistant. Organisms that exhibited very major or major errors were retested in triplicate on three different test dates using the same susceptibility testing procedure for each panel noted above. All results were recorded.

RESULTS AND DISCUSSION

The antimicrobial susceptibility patterns of 75 pneumococci and 68 other streptococci when tested against 26 antimicrobial agents were determined using PASCO Strep Plus panels and NCCLS broth microdilution reference panels. The overall percent agreement of the two methods within 1 log₂ dilution was 99.1% (Table 1) and ranged from 95.0% (rifampin) to 100% (ampicillin, cefdinir, cefepime, cefotaxime, ceftriaxone, cefuroxime, chloramphenicol, clinafloxacin, grepafloxacin, levofloxacin, linezolid, and ofloxacin) when all values were included. Deleting off-scale MICs increased the percent agreement within 1 log₂ dilution for clarithromycin, clindamycin, meropenem, penicillin, rifampin, and sparfloxacin (Table 1). Overall, 69.4% of the PASCO MIC results were identical to those of the reference method, while 23.9% were 1 dilution lower and 5.8% were 1 dilution higher. Thus, even though the inoculum in the PASCO panels tended to be slightly higher than that in the CDC panels (1× 10⁵ CFU/well for the PASCO panels versus 3 × 10⁴ CFU/well for the CDC panels), the PASCO MICs tended to be the same or slightly lower than those from the CDC panels, as was observed in our previous study (29). The reasons for this are unclear but may be due to the slight differences in growth observed between different lots of Mueller-Hinton broth. The Wilcoxon signed-rank test was not performed since there were no differences observed that were >2 dilutions from the reference value. Reproducibility testing of discrepancies using PASCO panels had 100% agreement at ±1 log₂ dilution.

TABLE 1.

Comparison of PASCO MIC results to broth microdilution MIC results for antimicrobial agents tested against 143 pneumococci and other streptococci

Antimicrobial agent	No. of isolates with the following difference in MICs compared to reference methodb					% Agreement (excluding off-scale valuesa)
	−2	−1	0	+1	+2
Amoxicillin	1	41	96	5	0	99.3
Amoxicillin-clavulanic	1	46	91	4	1	98.6
Ampicillin	0	25	110	8	0	100
Cefdinir	0	30	110	3	0	100
Cefepime	0	33	106	4	0	100
Cefotaxime	0	14	117	12	0	100
Ceftriaxone	0	25	111	7	0	100
Cefuroxime	0	21	120	2	0	100
Chloramphenicol	0	31	102	10	0	100
Clarithromycin	0	30	103	8	2	98.6 (100)
Clinafloxacin	0	27	111	5	0	100
Clindamycin	1	16	122	2	2	98 (98.6)
Erythromycin	1	4	123	15	0	99.3
Grepafloxacin	0	12	109	22	0	100
Imipenem	1	16	122	4	0	99.3
Levofloxacin	0	41	99	3	0	100
Linezolid	0	10	118	15	0	100
Meropenem	1	90	50	2	0	99.3 (100)
Moxifloxacin	1	40	96	6	0	99.3
Ofloxacin	0	6	106	31	0	100
Penicillin	0	24	108	10	1	99.3 (100)
Quinupristin-dalfopristin	5	87	47	4	0	96.5
Rifampin	5	80	54	2	2	95.0 (95.1)
Sparfloxacin	1	6	109	25	2	97.9 (97.9)
Trovafloxacin	2	66	70	5	0	98.6
Vancomycin	1	69	70	3	0	99.3

^aPercent agreement within 1 dilution of the reference MIC result. Off-scale values are those MICs below the lowest dilution on the test panel or above the highest dilution on the test panel. 

^bMIC results from PASCO Strep Plus panels that were 2 dilutions or 1 dilution below the MIC result from the reference panels (−2 and −1, respectively), that matched the MIC result from the reference panels (0), or that were 1 or 2 dilutions above the MIC result from the reference panels (+1 and +2, respectively). 

The very major, major, and minor errors encountered during testing are shown in Table 2 for pneumococci and Table 3 for other streptococci. There were four very major errors observed among the pneumococcal results (one each with amoxicillin-clavulanic acid, cefotaxime, chloramphenicol, and erythromycin), which reflects a rate of 0.2% (or 1.0% if only resistant pneumococcal strains are used as the denominator for calculations). All resolved on retesting. There were five major errors for pneumococci (one each with cefdinir, chloramphenicol, clindamycin, erythromycin, and meropenem), and one major error with erythromycin for an S. salivarius isolate observed in the study (Tables 2 and 3). Using only susceptible strains for the denominators yields major error rates of 0.4 and 0.1% for pneumococci and other streptococci, respectively. All major errors also resolved on retesting.

TABLE 2.

Interpretative errors for S. pneumoniae

Antimicrobial agentc	No. of organisms in categorya:			% Interpretative errorb
	S	I	R	Very major	Major	Minor
Amoxicillin	35	8	32	0	0	15 (20.0)
Amoxicillin-clavulanic	37	6	32	1 (1.3/3.1)	0	17 (22.7)
Cefdinir	34	5	36	0	1 (1.3/2.9)	1 (1.3)
Cefepime	36	13	26	0	0	4 (5.3)
Cefotaxime	41	10	24	1 (1.3/4.1)	0	10 (13.3)
Ceftriaxone	41	12	22	0	0	8 (10.7)
Cefuroxime	33	0	42	0	0	0
Chloramphenicol	50	—d	25	1 (1.3/4.0)	1 (1.3/2.0)	0
Clarithromycin	46	0	29	0	0	0
Clindamycin	58	0	17	0	1 (1.3/1.7)	0
Erythromycin	46	0	29	1 (1.3/3.4)	1 (1.3/2.2)	0
Grepafloxacin	74	0	1	0	0	0
Imipenem	40	10	25	0	0	4 (5.3)
Levofloxacin	74	0	1	0	0	0
Meropenem	41	0	0	0	1 (1.3/2.4)	4 (5.3)
Ofloxacin	73	1	1	0	0	0
Penicillin	29	16	30	0	0	1 (1.3)
Quinupristin-dalfopristin	74	0	1	0	0	1 (1.3)
Rifampin	75	0	0	0	0	0
Sparfloxacin	74	0	1	0	0	0
Trovafloxacin	74	0	1	0	0	0
Vancomycin	75	—d	—d	0	0	0
Total	1160	93	397	4 (0.2/1.0)	5 (0.3/0.4%)	65 (3.9%)

^aThese columns list the number of organisms tested in each interpretive category: S, susceptible; I, intermediate; and R, resistant. 

^bFor values in parentheses, the first percentage reflects the error rate using all organisms as the denominator and the second is the error rate calculated using only resistant isolates for very major errors and only susceptible isolates for major errors. Minor rates are calculated using the total number of organisms tested. 

^cOnly antimicrobial agents for which there are NCCLS interpretive criteria are shown. 

^dNo intermediate breakpoint is defined by NCCLS for chloramphenicol, and no intermediate or resistant breakpoints are defined for vancomycin for pneumococci. 

TABLE 3.

Interpretative errors for streptococci other than S. pneumoniae

Antimicrobial agentc	No. of organisms in categorya:			% Interpretative errorsb
	S	I	R	Very major	Major	Minor
Ampicillin	59	7	2	0	0	1 (1.5)
Cefepime	63	3	2	0	0	2 (2.9)
Cefotaxime	65	1	2	0	0	0
Ceftriaxone	64	2	2	0	0	2 (2.9)
Chloramphenicol	67	1	0	0	0	1 (1.5)
Clarithromycin	55	0	13	0	0	0
Clindamycin	62	0	6	0	0	0
Erythromycin	56	0	12	0	1 (1.5/1.8)	0
Grepafloxacin	65	2	1	0	0	1 (1.5)
Levofloxacin	68	0	0	0	0	0
Meropenem	65	—d	—d	0	0	0
Ofloxacin (β-hemolytic streptococci only)	16	0	0	0	0	0
Penicillin	57	9	2	0	0	2 (2.9)
Quinupristin-dalfopristin	58	9	1	0	0	8 (11.8)
Trovafloxacin	68	0	0	0	0	0
Vancomycin	68	—d	—d	0	0	0
Total	956	34	46	0	1 (0.1/0.1)	17 (1.6)

^aThese columns list the number of organisms tested in each interpretive category: S, susceptible; I, intermediate; and R, resistant. 

^bFor values in parentheses, the first percentage reflects the error rate using all organisms as the denominator and the second is the error rate calculated using only resistant isolates for very major errors and only susceptible isolates for major errors. Minor rates are calculated using the total number of organisms tested. 

^cOnly antimicrobial agents for which there are NCCLS interpretive criteria are shown. 

^dNo intermediate or resistant breakpoints are defined by NCCLS for meropenem or for vancomycin for nonpneumococcal streptococci. 

Overall, 82 (3.1%) minor errors were observed, which were all within 1 dilution of the reference MIC result. Of the 15 minor amoxicillin errors, 12 were resistant strains that were designated as intermediate by the results from PASCO panels. Similarly, 14 of 17 minor errors with amoxicillin-clavulanic acid were resistant by the reference method but intermediate by PASCO. Thus, the number of strains designated by PASCO panels as susceptible but designated as nonsusceptible (i.e., either intermediate or resistant) by the reference method was low. The cefotaxime minor errors were more varied, and in two cases the PASCO MICs were higher than those of the reference method. Although the error rates for amoxicillin-clavulanic acid, amoxicillin, cefotaxime, and ceftriaxone appear high (22.7, 20.0, 13.3, and 10.7%, respectively), these error rates reflect the organisms in the challenge set used in the study, many of which demonstrate borderline resistance to these antimicrobial agents (29). With clinical isolates, the error rates are likely to be much lower, particularly when considering that the actual MICs were within 1 dilution of the reference method. The minor error rates for the other antimicrobial agents were ≤6.0%. The only antimicrobial agent that posed a potential testing problem with nonpneumococcal streptococci was quinupristin-dalfopristin, which showed a minor error rate of 11.8%. Errors, however, were distributed among six different streptococcal species.

Susceptibility testing of pneumococci and other streptococci has increased in importance since therapeutic options are now more limited due to increasing resistance (3, 5, 8, 11, 18, 26). This also is true for species such as S. mitis, S. oralis, and S. sanguis, for which penicillin resistance is also emerging (6, 11, 19, 24). In response, PASCO has expanded the number of agents available for testing and included other streptococci in the latest panel, prompting this reevaluation of the system. Previously, Nolte et al. (22) reported that 10% of the results with the PASCO panels were outside the acceptable range of ±1 doubling dilution from the reference result for penicillin. This was in contrast to the results of a CDC study, which did not observe these same problems (29). Recent studies by Guthrie et al. (14) found a high correlation among the results produced by PASCO, MicroScan MICroSTREP MIC, and Sensititre panels, all of which are broth microdilution-based systems. We found results comparable to the NCCLS microdilution reference method for penicillin and the other 25 antimicrobial agents tested.

PASCO panels require a short setup time of approximately 3 min per panel and few additional supplies. We found the panels to be easy to read when using the PASCO MIC Manual Reader.

Recently Doern et al. reviewed the results from the College of American Pathologists on proficiency testing of S. pneumoniae (10). The results clearly show that susceptibility testing of pneumococci by laboratories in the United States is less than optimal. Many laboratories continue to perform disk diffusion testing for β-lactam drugs although there are neither breakpoints nor interpretive criteria for such testing in the NCCLS guidelines (21). In this era of increasing antimicrobial resistance, MIC testing is critical for providing accurate information on the susceptibility or resistance of pneumococci to β-lactam drugs. The new commercially prepared PASCO Strep Plus panels appear to produce results equivalent to the results of the NCCLS broth microdilution method and provide an accurate method for determining susceptibilities of pneumococci and other streptococci to a wide variety of antimicrobial agents.