Acceleration of Antimicrobial Susceptibility Testing of Positive Blood Cultures by Inoculation of Vitek 2 Cards with Briefly Incubated Solid Medium Cultures

Abstract

Briefly incubated agar cultures from positive blood cultures were used for antimicrobial susceptibility testing (AST) by Vitek 2. The cultivation time until inoculation was 3.8 h for Gram-positive cocci and 2.4 h for Gram-negative rods. The error rates were low, providing early and reliable AST without additional time or cost expenditure.

TEXT

Since the outcome of sepsis is dependent on appropriate and timely treatment (1), the identification of pathogens and availability of antimicrobial susceptibility testing (AST) should occur as soon as possible (2). Routinely, positive blood culture (BC) broth is plated onto agar and incubated for 18 to 24 h before inoculation of Vitek 2 cards for combined identification and AST (3, 4). Recently, the rapid identification of bacteria from positive BCs by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) has been reported from cultures very briefly incubated on solid medium (5). Here, we evaluated the feasibility and accuracy of accelerated AST using the inoculation of Vitek 2 cards with biomass very briefly incubated on agar after subcultivation of positive BC broth.

(This work was presented in part at the 53rd Interscience Conference on Antimicrobial Agents and Chemotherapy, 10 to 13 September 2013, Denver, CO [6].)

A total of 138 blood samples flagged as positive by an automated BC system (BD Bactec 9240; BD Diagnostics, Heidelberg, Germany) and Gram stained as Gram-positive cocci (GPC) or Gram-negative rods (GNR) were processed as follows: two drops of BC broth were subcultured on Columbia blood agar. Starting from 2 h of incubation at 36°C in air with 5% CO₂, the plates were inspected hourly, and a 0.5 to 0.63 McFarland standard inoculum was prepared at the first time point at which the growth appeared sufficient. Subsequently, Vitek 2 (bioMérieux SA, Marcy l'Étoile, France) testing was performed. For comparison, a 0.5 to 0.63 McFarland standard inoculum from 24-h cultures was tested by Vitek 2, as recommended by the manufacturer (3). The choice of a definite Vitek 2 AST card was based on the Gram stain from BC broth or/and early MALDI-TOF identification, performed as previously described (5). AST was done in triplicate, i.e., three AST cards were used for each inoculum to minimize the impact of reproducibility errors on the comparison between rapid and standard testing. The median values were used for analysis. Additionally, the detection of particular resistance mechanisms was evaluated: cefoxitin screening for methicillin resistance in staphylococci, inducible clindamycin resistance in staphylococci and Streptococcus agalactiae, and the production of extended-spectrum β-lactamases (ESBL) in Enterobacteriaceae.

Of 138 BCs, 13 ambiguously identified samples, 8 samples with technical errors, 5 isolates that could not be tested by Vitek 2, and 8 mixed cultures were excluded. Thus, 104 isolates (68 GPC and 36 GNR) were available for analysis by rapid AST (Table 1).

TABLE 1.

Microorganisms available for AST results comparison^a

Organism	No. (%) available for AST
Gram-positive cocci	68 (100)
Staphylococcus aureus	8 (11.8)
Coagulase-negative staphylococci	40 (58.8)
Enterococcus spp.	17 (25.0)
Streptococcus pneumoniae	2 (2.9)
Streptococcus agalactiae	1 (1.5)
Gram-negative rods	36 (100)
Escherichia coli	25 (69.4)
Klebsiella pneumoniae	5 (13.9)
Enterobacter cloacae	4 (11.1)
Serratia marcescens	1 (2.8)
Pseudomonas aeruginosa	1 (2.8)

^an = 104.

The mean time necessary for cultivating plated positive BC broth on agar until successful inoculation of Vitek 2 cards was as short as 3.8 h for GPC and 2.4 h for GNR (Table 2). For GPC, inoculation was possible at 4 h in 85.3% of the isolates, with all specimens being inoculated within 7 h (Fig. 1). At 3 h, inoculation was possible with all GNR isolates, with the exception of a Pseudomonas aeruginosa isolate, which was inoculated at 5 h (Fig. 1). The mean cultivation times to inoculation of Vitek 2 cards and to an AST result for different bacterial species or groups are shown in Fig. S1 in the supplemental material. An AST result was available 20.2 h and 21.8 h earlier than with the standard method for GPC and GNR, respectively (Table 2).

TABLE 2.

Time to successful inoculation and AST result for Gram-positive cocci and Gram-negative rods in Vitek 2

Time period	Time (mean ± SD) (h) for:
	Gram-positive cocci (n = 68)			Gram-negative rods (n = 36)
	Brief culture	Control culture	P valuea	Brief culture	Control culture	P valuea
Cultivation time until Vitek 2 inoculation	3.8 ± 0.9	24.0	<0.0001	2.4 ± 0.6	24.0	<0.0001
Duration of Vitek 2 AST	9.8 ± 1.1	9.8 ± 1.8	NSb	8.8 ± 1.4	9.0 ± 2.8	NS
Total time from positive BC subculture to AST result	13.6 ± 1.4	33.8 ± 1.8	<0.0001	11.2 ± 1.7	33.0 ± 2.8	<0.0001

^aBy Wilcoxon signed-rank test.

^bNS, not significant.

FIG 1.

Cultivation time until Vitek 2 inoculation required for successful AST for Gram-positive cocci and Gram-negative rods.

Very major error (VME) (false-susceptible result of rapid AST), major error (ME) (false-resistant result of rapid AST), and minor error (mE) (false categorization involving intermediate result) rates, and categorical agreement (CA) (results within the same category) and essential agreement (EA) (MIC difference ≤ 1 double dilution step) were 1.6%, 0.3%, 0.1%, 99.2%, and 99.1% for GPC (Table 3) and 0%, 0.5%, 0.8%, 99.2%, and 99.3% for GNR (Table 4), respectively. In general, the performance of rapid AST in our study was better or similar to that reported in studies using differential centrifugation for the direct inoculation of Vitek 2 cards from positive BCs (7–15). We report error rates here, as required by both the International Organization for Standardization and the U.S. Food and Drug Administration (16, 17), which recommend that the rate of VME should be calculated using the number of isolates which have been determined as resistant by the reference method as the denominator. To obtain the rate of ME, number of ME should be divided by the number of susceptible isolates determined by the reference method. The mE rate is calculated based on the total number of isolates. Of note, only a minority of studies (7, 10, 15) investigating direct inoculation of Vitek 2 cards for AST presented their results in that way, while the most of studies on rapid Vitek 2 AST reported VME and ME error rates based on the total number of isolates (8, 9, 12–14, 18). If calculated based on the number of total isolates, the VME and ME rates are inherently lower and would amount in our study to 0.5% and 0.2% for GPC and 0 and 0.3% for GNR, respectively. The results of all three AST tests within the triplicate experiments were the same for 99.4% and 99.0% of the isolate-antibiotic combinations for rapid and standard testing, respectively (P = 0.246), demonstrating the high reproducibility of the rapid method.

TABLE 3.

Performance of rapid antimicrobial susceptibility testing of Gram-positive cocci (n = 68)

Antimicrobial agent (n)	No. of isolate-antibiotic combinations	No. of isolates that werea:		No. (%) of errors			Categorical agreement (%)	Essential agreement (%)
		Resistant	Susceptible	Very major	Major	Minor
Staphylococcus spp. (48)	872	275	586	4 (1.5)	2 (0.3)	0	99.3	99.0
Penicillin G	48	42	6				100	100
Oxacillin	48	30	18				100	97.9
Clindamycin	48	22	26				100	100
Erythromycin	48	30	18				100	100
Levofloxacin	48	25	23				100	100
Moxifloxacin	48	15	23				100	100
Gentamicin	48	19	29				100	100
Tobramycin	48	23	25				100	100
Vancomycin	48	0	48				100	100
Teicoplanin	48	7	41	2	1		93.8	89.6
Linezolid	48	0	48				100	97.9
Tetracycline	48	18	29	1			97.9	97.9
Tigecycline	48	0	48				100	100
Trimethoprim-sulfamethoxazole	48	17	31				100	100
Fusidic acid	48	9	39	1	1		95.8	97.9
Fosfomycin	48	14	34				100	100
Rifampin	48	4	44				100	100
Nitrofurantoin	48	0	48				100	100
Mupirocin	8	0	8				100	100
Enterococcus spp. (17)	238	91	144	1 (1.1)	0	1 (0.4)	99.2	98.7
Penicillin G	17	13	4				100	94.1
Ampicillin	17	10	7				100	100
Ampicillin-sulbactam	17	13	4				100	100
Imipenem	17	10	7				100	100
Erythromycin	17	14	0			1	94.1	94.1
Levofloxacin	17	10	7				100	100
Vancomycin	17	1	16				100	100
Teicoplanin	17	1	16				100	100
Linezolid	17	1	16				100	100
Tetracycline	17	8	9				100	100
Tigecycline	17	0	17				100	100
Trimethoprim-sulfamethoxazole	17	0	17				100	100
Quinupristin-dalfopristin	17	6	11				100	100
Nitrofurantoin	17	4	13	1			94.1	100
Total Gram-positive cocci (68)	1,163	377	767	6 (1.6)b,c	2 (0.3)b	1 (0.1)b	99.2b	99.1b,c

^aThe number of results within the intermediate category can be calculated by subtracting the resistant and susceptible results from the number of isolate-antibiotic combinations tested.

^bStreptococci (S. pneumoniae, n = 2 and S. agalactiae, n = 1) are included in the total number of Gram-positive cocci and in the calculation of the error and agreement rates for all Gram-positive cocci. No separate calculation of error and agreement rates was performed for streptococci because of the small number of isolates.

^cOne VME (tetracycline in a S. pneumoniae isolate) and no other errors were observed in streptococci. Essential agreement (EA) was observed for all streptococcal isolate-antibiotic combinations.

TABLE 4.

Performance of rapid antimicrobial susceptibility testing of Gram-negative rods (n = 36)

Antimicrobial agent	No. of isolate-antibiotic combinationsb	No. of isolates that werea:		No. (%) of errors			Categorical agreement (%)	Essential agreement (%)
		Resistant	Susceptible	Very major	Major	Minor
Ampicillin	35	29	6				100	100
Ampicillin-sulbactam	35	20	15		1		97.1	97.1
Piperacillin	1	0	0				100	100
Piperacillin-tazobactam	36	9	18			2	94.4	97.2
Cefuroxime	35	16	19				100	100
Cefuroxime axetil	35	15	20				100	100
Cefotaxime	36	10	26			1	97.2	97.2
Cefpodoxime	35	14	21				100	100
Ceftazidime	36	4	25				100	100
Cefepime	1	0	1				100	100
Ertapenem	35	0	35		1		97.1	97.1
Imipenem	36	0	35				100	100
Meropenem	36	0	35				100	100
Aztreonam	1	0	0				100	100
Ciprofloxacin	36	10	26				100	100
Moxifloxacin	35	10	25				100	100
Gentamicin	36	1	34				100	100
Amikacin	1	0	1				100	100
Tobramycin	1	0	1				100	100
Tetracycline	35	16	19			1	97.1	100
Tigecycline	35	1	29			1	97.1	100
Trimethoprim-sulfamethoxazole	36	14	22				100	100
Colistin	1	0	1				100	100
Total	609	169	414	0	2 (0.5)	5 (0.8)	99.2	99.3

^aThe number of results in the intermediate category can be calculated by subtracting the resistant and susceptible results from the number of isolate-antibiotic combinations tested.

^bOne as a number of isolate-antibiotic combinations for some antibiotics is due to the fact that those antibiotics are analyzed only in the card AST-N248, which was used for testing of the only isolate of P. aeruginosa.

Among 48 staphylococci tested, 30 were methicillin resistant as detected by positive cefoxitin screening (12.5% [1/8] of Staphylococcus aureus isolates and 72.5% [29/40] of coagulase-negative staphylococci). Thereby, 100% agreement was observed between the rapid and standard methods. Inducible clindamycin resistance was identified in three Staphylococcus epidermidis isolates. Again, this mechanism was detected concordantly by the two methods. Among 30 Escherichia coli and Klebsiella pneumoniae isolates, ESBL production was reported in eight isolates (7 E. coli and 1 K. pneumoniae) when short-term cultures were used. The standard method identified one additional ESBL producer (E. coli); however, this ESBL phenotype was not confirmed by an additional test.

The microbial counts of 0.5 McFarland standard suspensions from the briefly incubated cultures and 24-h cultures were 3.2 × 10⁷ and 9.4 × 10⁷ CFU/ml, respectively (P < 0.0001; for a comparison for a subgroup of isolates, see Table S1 in the supplemental material). This difference is not high, which might explain the accuracy of the rapid results.

In conclusion, the inoculation of brief solid medium cultures into Vitek 2 provided early and reliable AST without additional time or cost expenditure. However, further studies are necessary to confirm the reliability of this method with more pathogens with different resistance phenotypes. The thorough control of mature cultures is essential, as mixed cultures can occur. This method can be combined with the rapid MALDI-TOF identification from short-term agar cultures (5).