Evaluation of the Accelerate Pheno System for Rapid Identification and Antimicrobial Susceptibility Testing of Positive Blood Culture Bottles Inoculated with Primary Sterile Specimens from Patients with Suspected Severe Infections

The Accelerate Pheno system is approved for rapid identification and phenotypic antimicrobial susceptibility testing (AST) of microorganisms grown from positive blood cultures inoculated with blood from septic patients. We evaluated the performance of the system for identification and AST from positive blood culture bottles inoculated with primary sterile nonblood specimens from patients with suspected severe infections.

ABSTRACT

The Accelerate Pheno system is approved for rapid identification and phenotypic antimicrobial susceptibility testing (AST) of microorganisms grown from positive blood cultures inoculated with blood from septic patients. We evaluated the performance of the system for identification and AST from positive blood culture bottles inoculated with primary sterile nonblood specimens from patients with suspected severe infections. One hundred positive blood culture bottles with primary sterile specimens (63 cerebrospinal fluids, 16 ascites, 7 pleural fluids, 4 vitreous fluids, 5 joint aspirates, and 5 other aspirates) from 100 patients were included. Pathogen identification was in agreement with conventional methods for 72 of 100 cultures (72%) and for 81 of 112 (72%) pathogens when considering all pathogens and for 72 of 92 (78%) cultures and 81 of 104 (78%) pathogens when considering on-panel pathogens only. Eight of 31 isolates (26%) not identified by APS were pathogens not included in the APS panel. APS and conventional methods accordingly identified all pathogens from two of nine polymicrobial cultures (22%). APS generated antimicrobial resistance results for 57 pathogens of 57 cultures. The overall category agreement between APS and culture-based AST was 91.2%; and the rate for minor errors was 6.9%, for major was 1.7%, and for very major errors was 0.2%. APS may accelerate pathogen identification and phenotypic AST from positive blood culture bottles inoculated with primary sterile specimens from patients with serious infections, especially for hospitals without an on-site microbiology laboratory. However, the inclusion of nonblood specimens with a high likelihood of polymicrobial infections may result in an inferior performance.

INTRODUCTION

An inappropriate initial antibiotic therapy in septic patients is associated with increased morbidity, mortality, length of hospital stay, and costs of care (1, 2). The rapid identification (ID) and early availability of antibiotic resistance results of pathogens causing serious infections may improve the course of disease by early switch of an initial inadequate antibiotic therapy (3, 4). In addition, early knowledge of the causative pathogen and its antibiotic susceptibility enables rapid de-escalation of a calculated broad-spectrum antibiotic therapy to targeted antimicrobials. The Accelerate PhenoSystem (APS) is approved for the rapid pathogen identification and susceptibility testing of positive blood cultures from septic patients by using fluorescence in situ hybridization (FISH) technology and identifies pathogens from positive blood culture bottles within 2 hours. Phenotypic resistance results are available after an additional 4 to 5 h. Several studies have shown that APS is valuable for an accelerated identification and antimicrobial resistance testing (AST) of positive blood cultures from septic patients compared with conventional diagnostics (5–8). An early identification of the causative pathogen and knowledge of its antibiotic resistance can improve the course of disease or reduce hospital costs also for other serious infections (9–11). For acute bacterial meningitis, mortality and hospital costs have been positively correlated with the period until an adequate antibiotic therapy was administered (10, 11). In addition, inoculation of blood culture bottles with primary sterile specimens from patients with serious infections has been shown to increase pathogen recovery compared with conventional agar media and broths for serious infections, such as bacterial meningitis, endophthalmitis, and joint and pleural infections (12–15). To the best of our knowledge, APS use for identification and susceptibility testing from positive blood culture bottles inoculated with nonblood primary sterile specimens has not been evaluated before. We used APS for identification and resistance testing of 100 positive blood culture bottles inoculated with primary sterile specimens from diagnostic routine clinical specimens and compared them with the results of conventional diagnostics.

MATERIALS AND METHODS

Study population.

Between January 2018 and July 2019, 100 positive blood culture bottles (Bactec Plus aerobic/F medium, Bactec Plus anaerobic/F medium, or Bactec Peds Plus/F medium, all from Becton, Dickinson, Baltimore, MD) inoculated with primary sterile nonblood specimens from 100 unique patients of the University Hospital Essen were included in the study. The study was reviewed and approved by the Ethics Committee of the Medical Faculty of the University of Duisburg-Essen (no. 20-9590-BO). All samples were taken as a part of standard care procedures. The recommendations of the International Conference on Harmonization–Good Clinical Practices (ICH-GCP) guidelines have been followed, and the study was performed in accordance with the latest version of the Declaration of Helsinki.

Blood culture diagnostics.

Blood culture bottles were inoculated directly after sampling by the attending physicians. Vitreous fluid was the type of specimen with a minimum volume of 1 to a maximum of 3 ml inoculated into the blood culture bottles according to the current standard operation procedures. We incubated all specimens following the current German microbiology procedure quality standards (MiQ) of the German Society for Hygiene and Microbiology (DGHM). The maximum culture period for blood culture bottles inoculated with primary sterile specimens was 5 days for ascites and cerebrospinal and pleural fluid samples and 14 days for joint and vitreous fluid samples. Extended incubation periods have been reported to be valuable for detection of some bacterial and fungal pathogens from joint and vitreous fluid samples (16, 17). Cultivation was performed in a Bactec 9240 instrument (BD Diagnostic Systems, Franklin Lakes, NJ) with automatic detection of bacterial growth. Although the manufacturer recommends using APS within 8 hours after the blood culture is flagged positive, we extended the inclusion of blood culture bottles that were flagged positive up to 24 hours before the APS run was started. The time extension was due to the limited laboratory working hours that are weekdays from 7:30 a.m. to 4:00 p.m. and weekends from 8:00 a.m. to 12:00 p.m. Blood cultures flagged positive outside the working hours were processed the next morning. During the operating hours, blood cultures were processed as soon as they were flagged positive. Positive blood culture bottles stayed in the Bactec 9240 instrument until further processing. APS and conventional diagnostics were performed from the same positive blood culture bottle. We ran the Accelerate PhenoTest BC kit on the positive cultures from positive aerobic, anaerobic, or Peds Plus/F medium blood culture bottles after obtaining the result of the Gram stain. We included both monomicrobial and polymicrobial cultures for APS analysis and did not exclude any sample dependent on the result of the Gram stain. Identification of organisms on culture was primarily performed using the MicroScan Walkaway Plus system (Beckman Coulter, Brea, CA) by direct inoculation of 100-μl fluid of the positive blood culture bottle as described by Infante et al. (18). Identification was supplemented by the Vitek MS and Vitek II systems (both from bioMérieux SA, France) from colonies grown on subcultured Columbia blood agar, chocolate blood agar, MacConkey agar, Beerens-Schaedler anaerobic agar, kanamycin-vancomycin agar, or Brilliance Candida selective agar (all Oxoid, Wesel, Germany) from the positive blood culture bottles when necessary to achieve identification to the species level. Antimicrobial susceptibility testing (AST) was performed with the MicroScan Walkaway Plus instrument (Beckman Coulter) by direct inoculation of 100-μl fluid of positive blood culture bottles (18) or was performed with the Vitek II system (bioMérieux SA) or Etest from colonies grown on Columbia blood agar, chocolate blood agar, MacConkey agar, Beerens-Schaedler anaerobic agar, or Brilliance Candida selective agar (all Oxoid). Bacteria were subcultured for at least 16 h before AST was performed using Vitek II system or Etest in case of discrepancies or implausible results. In addition, the identification and AST results were confirmed by results of culture from available native specimens that were primarily plated on the abovementioned agar plates and identified by Vitek MS or Vitek II systems (bioMérieux). AST was primarily performed with the Vitek II system and supplemented by using the MicroScan Walkaway Plus instrument or Etest with isolates from conventional culture of native specimens.

Data analysis.

A result was assessed as true positive (TP) for pathogen identification when APS and conventional diagnostics both identified an organism. A true-negative result occurred when APS and conventional diagnostics both did not identify an organism that was included in the APS BC panel. A result was classified as false negative (FN) when APS failed to identify an organism that was identified by conventional diagnostics and reported the organism as “negative,” and a false-positive result occurred when APS incorrectly identified an organism that was not confirmed by conventional diagnostics. For a few isolates, individual organisms were reported as “indeterminate.” In these cases, APS reported that too few cells were present for a positive identification result for these organisms, while for the other organisms, valid results were reported for those samples. Sensitivity was calculated by the number of TP/(TP + FN), and specificity was calculated by number of TN/(TN + FP) for each organism and for entirety of Gram-positive and Gram-negative organisms that were identified by conventional diagnostics.

AST results were interpreted according to the guidelines of the European Committee on Antimicrobial Susceptibility Testing (EUCAST). Very major error (VME), major error (ME), and minor error (MiE) were calculated by comparing the results for AST by APS with the results of conventional methods. VME was defined as a susceptible APS result when conventional diagnostics reported a resistant result, ME as a false-resistant APS result when conventional diagnostics provided a susceptible result, and MiE was defined by a differentiation for one interpretation level of resistance results, e.g., as an intermediate result by APS when the result of conventional diagnostics was susceptible or resistant.

RESULTS

We evaluated the performance of APS for pathogen identification and antimicrobial susceptibility testing from 100 positive blood culture bottles inoculated with primary sterile specimens during a period of 18 months between January 2018 and July 2019. The most frequently included specimens were cerebrospinal fluid samples of 63 patients and ascites samples of 16 patients (Table 1). Pathogen identification by APS was in accordance with conventional diagnostics for 72 patients. According to the specimen type, APS identification was in agreement with conventional diagnostics for 76% of cerebrospinal fluid samples (48 of 63), for 56% of ascites samples (9 of 16), and for 71% of all other specimen samples (15 of 21) when considering all pathogens; and 79% (48 of 61) of cerebrospinal fluid samples and 82% (9 of 11) of ascites samples when considering on-panel pathogens only. Conventional diagnostics identified 91 mono- and 9 polymicrobial cultures. Polymicrobial cultures were detected from 100% (2 of 2) bile, 100% (1 of 1) liver abscess puncture fluid, 19% (3 of 16) ascites, and 4.5% (3 of 67) cerebrospinal fluid. For the other specimens, monomicrobial cultures were identified.

TABLE 1.

Sample specimens and performance of APS for organism identification

Specimen	Correct identification of all cultures (n [%])	Correct identification of cultures with on-panel organisms only (n [%])	No. (%) of off-panel organisms of all cultures
Cerebrospinal fluid	48 of 63 (76)	48 of 61 (79)	2 of 63 (3.2)
Ascites	9 of 16 (56)	9 of 11 (82)	5 of 16 (31.3)
Pleural fluid	6 of 7 (86)	6 of 7 (86)	0 of 7 (0)
Joint aspirate	4 of 5 (80)	4 of 5 (80)	0 of 5 (0)
Vitreous fluid	3 of 4 (75)	3 of 3 (100)	1 of 4 (25)
Bile	0 of 2 (0)	0 of 2 (0)	0 of 2 (0)
Liver abscess aspirate	0 of 1 (0)	0 of 1 (0)	0 of 1 (0)
Bone marrow aspirate	1 of 1 (100)	1 of 1 (100)	0 of 1 (0)
Peritoneal dialysate	1 of 1 (100)	1 of 1 (100)	0 of 1 (0)
Total	72 of 100 (72)	72 of 92 (78)	8 of 100 (8)

APS for pathogen identification was in agreement with conventional diagnostics for 70 of 91 monomicrobial cultures (77%) when considering all pathogens identified by conventional methods and 70 of 84 cultures (83%) when considering on-panel organisms from monomicrobial cultures only. Of nine polymicrobial cultures, pathogen identification between APS and conventional diagnostics was in agreement for two cultures (22%). For the seven other polymicrobial cultures, only one pathogen per patient was correctly identified by APS (Table 2). The other pathogens of the seven polymicrobial cultures were not detected by APS, and one Candida albicans isolate was misidentified as Enterobacter cloacae by APS. Of 112 pathogens detected by conventional diagnostics, 81 (72%) were correctly identified by APS (Table 3). Eight of 31 microorganisms (26%) not identified by APS were isolates of bacterial species that were not included in the APS panel (Stenotrophomonas maltophilia, Haemophilus influenzae [n = 2], Actinomyces neuii, Corynebacterium jeikeium, Bacillus cereus [n = 2], and Micrococcus luteus). In addition, three of four Klebsiella pneumoniae isolates were identified to the genus level by APS. Sixty-nine of 90 Gram-positive bacterial isolates (77%), 12 of 18 Gram-negative isolates (67%), and 0 of 4 fungal isolates detected by conventional methods were correctly identified by APS. Coagulase-negative staphylococci (CoNS) were the bacteria most frequently not detected by APS (14 isolates). None of four C. albicans isolates identified by conventional diagnostics was identified by APS, and three of them were cultured from polymicrobial cultures. For six monomicrobial cultures and one polymicrobial culture for which APS did not identify CoNS, indeterminate results were reported for some organisms, including CoNS for 6 cultures, S. aureus for 4, and C. albicans and C. glabrata for all 7 cultures. APS reported that there were too few cells present in these cultures for a positive identification result and recommended culture for identification and AST because of the possibility of one or more organisms present. For five of the seven polymicrobial cultures for which APS did not identify all organisms included in each culture, APS reported a monomicrobial call. Only for two of these cultures APS suggested identification from culture due to the possibility of more organisms present.

TABLE 2.

Discrepant identification results between APS and conventional diagnostics

Infection type	Culture IDa	APS ID (no. total not identified)	Comment
Monomicrobial	CoNS	No identification (13)
	C. albicans	No identification (1)
	Micrococcus luteus	No identification (1)	Off-panel
	Bacillus cereus	No identification (1)	Off-panel
	Corynebacterium jeikeium	No identification (1)	Off-panel
	Haemophilus influenzae	No identification (2)	Off-panel
	S. maltophilia	No identification (1)	Off-panel
	Actinomyces neuii	No identification (1)	Off-panel
Polymicrobial	Streptococcus parasanguinis, CoNS	CoNS
	E. faecalis, K. pneumoniae, E. coli, C. albicans	Klebsiella spp.
	E. coli, C. albicans	E. coli, E. cloacae	Incorrect identification
	CoNS, Bacillus cereus	CoNS	Off-panel (B. cereus)
	E. faecium, C. albicans	E. faecium
	Pseudomonas aeruginosa, E. faecium, CoNS	E. faecium
	Klebsiella oxytoca, CoNS	CoNS

^aID, identification.

TABLE 3.

Performance of APS for organism identification

Organism	No. identifieda				Sensitivity (%)	Specificity (%)	Comment
	TP	FNb	TN	FP
Gram positive
Coagulase-negative Staphylococcus spp.	54	14	26	0	79.4	100	6 INDc
Staphylococcus aureus	5	0	91	0	100	100	4 IND
Streptococcus spp.	5	1	94	0	83.3	100
Enterococcus faecalis	2	1	97	0	66.6	100
Enterococcus faecium	3	0	97	0	100	100
Micrococcus luteus	0	1	99	0	NDd	ND	Off-panel
Corynebacterium jeikeium	0	1	99	0	ND	ND	Off-panel
Bacillus spp.	0	2	98	0	ND	ND	Off-panel
Actinomyces neuii	0	1	99	0	ND	ND	Off-panel
Total Gram positive	69	21	800	0	76.7	100
Gram negative
Escherichia coli	5	1	93	0	83.3	100	1 IND
Klebsiella pneumoniaee	4	0	96	0	100	100
Klebsiella oxytoca	1	1	98	0	50	100
Enterobacter cloacae	2	0	97	1	100	98.9
Pseudomonas aeruginosa	0	1	99	0	0	100
Stenotrophomonas maltophilia	0	1	99	0	ND	ND	Off-panel
Haemophilus influenzae	0	2	98	0	ND	ND	Off-panel
Total Gram negative	12	6	680	1	66.7	99.85
Fungi	0	4	89	0	0	100
Candida albicans	0	4	89	0	0	100	7 IND

^aTP, true positive; FN, false negative; TN, true negative, FP, false positive.

^bFalse-negative APS results of on-panel pathogens from polymicrobial cultures are as follows: C. albicans, n = 3; Klebsiella oxytoca, 1; E. coli, 1; E. faecalis, 1; CoNS, 1; Streptococcus spp., 1; and P. aeruginosa, 1.

^cIND, indeterminate result.

^dND not determined.

^eThree of four K. pneumoniae isolates were identified only to the genus level by APS (Klebsiella spp.).

Our microbiology laboratory was not staffed 24/7, and we also included blood culture bottles that were flagged positive outside the working hours. We included positive cultures flagged positive for up to 24 hours before the APS run was started instead of 8 hours, as recommended by the manufacturer. Of 57 blood culture bottles with on-panel organisms only that were flagged positive within 8 hours before start of APS, APS did not correctly identify pathogens from 15 (26.3%) blood culture bottles. Twenty-four cultures with on-panel organisms were flagged positive between 8 and 16 hours, and of these cultures, 5 (20.8%) were not identified correctly by APS. A total of 11 cultures with on-panel organisms were positive between 16 and 24 hours before APS was started, and 1 (9.1%) was not identified correctly by APS.

APS performs antimicrobial susceptibility testing for bacterial organisms included in the panel except for Streptococcus spp., for which AST is not provided by the manufacturer. Of 81 organisms detected by APS, AST results were reported from 57 isolates (70%), including 42 coagulase-negative staphylococci, 5 S. aureus, 2 E. faecium, 1 Enterococcus faecalis, 5 Klebsiella spp., and 2 Escherichia coli isolates (Table 4). The overall category agreement between APS and conventional culture-based AST was 91.2% (385 of 422 reported drug microbiota combinations); rates for minor errors were 6.9% (29 of 422), major errors were 1.7% (7 of 422), and very major errors were 0.2% (1 of 422). Of 339 reported drug-microorganism combinations for Gram-positive bacteria, there were 26 errors (7.7%), including 1 very major error, 2 major errors, and 23 minor errors; 20 of the errors were related to trimethoprim-sulfamethoxazol for which the results of APS were more resistant in 17 cases (I instead of S or R instead of I). All of the Gram-positive discrepant susceptibility testing results between APS and conventional diagnostics were within coagulase-negative staphylococci. Of 83 Gram-negative drug microorganism combinations reported by both APS and conventional diagnostics, 11 discrepancies, comprising 5 major discrepancies and 6 minor discrepancies, occurred between APS and conventional drug resistance testing. Only 2 errors of the tested drug-microorganism combinations did not occur in a polymicrobial culture, and 7 discrepancies were found in 1 K. pneumoniae isolate of a polymicrobial culture.

TABLE 4.

Agreement of antimicrobial resistance results of APS compared with conventional diagnostics

Organism	Druga	Category agreement (n/total [%])	No. of errors (% of tests performed)			No. of isolates with result:b
			Very major	Major	Minor	R	I	S
S. aureus (n = 5)	Cefoxitin	5/5 (100)	0	0	0	0	0	5
	Daptomycin	5/5 (100)	0	0	0	0	0	5
	Doxycycline	5/5 (100)	0	0	0	0	0	5
	Erythromycin	5/5 (100)	0	0	0	1	0	4
	Linezolid	5/5 (100)	0	0	0	0	0	5
	TMP-SMX	5/5 (100)	0	0	0	0	0	5
	Vancomycin	5/5 (100)	0	0	0	0	0	5
CoNS (n = 42)	Cefoxitin	41/42 (98)	0	1 (2)	0	30	0	12
	Daptomycin	40/40 (100)	0	0	0	0	0	40
	Doxycycline	40/42 (95)	1 (2)	0	1 (2)	0	0	40
	Erythromycin	39/42 (93)	0	0	3 (7)	22	2	18
	Linezolid	42/42 (100)	0	0	0	0	0	42
	TMP-SMX	22/42 (52)	0	1 (2)	19 (45)	7	7	28
	Vancomycin	42/42 (100)	0	0	0	0	0	42
Enterococcus spp.	Ampicillin	3/3 (100)	0	0	0	2	0	1
	Daptomycin	3/3 (100)	0	0	0	0	0	3
	Linezolid	3/3 (100)	0	0	0	0	0	3
	Vancomycin	3/3 (100)	0	0	0	0	0	3
Sum errors GPc	1/339 (0.3)	2/339 (0.6)	23/339 (6.8)
GNc (E. coli, n = 2; Klebsiella spp., n = 5)	Amikacin	6/7 (86)	0	0	1 (14)	0	0	7
	Cefazolin	7/7 (100)	0	0	0	7	0	0
	Cefepime	5/6 (83)	0	0	1 (17)	0	0	0
	Ceftazidime	6/7 (86)	0	1 (14)	0	0	0	7
	Ceftriaxone	6/7 (86)	0	0	1 (14)	1	0	6
	Ciprofloxacin	7/7 (100)	0	0	0	0	0	7
	Colistin	6/7 (86)	0	1 (14)	0	0	0	7
	Ertapenem	6/7 (86)	0	0	1 (14)	0	0	7
	Gentamycin	5/6 (83)	0	0	1 (17)	0	0	6
	Meropenem	6/7 (86)	0	0	1 (14)	0	0	7
	PIP/TAZ	6/7 (86)	0	1 (14)	0	0	0	7
	Tobramycin	5/7 (71)	0	2 (29)	0	2	0	5
	SAM	1/1 (100)	0	0	0	0	0	1
Sum errors GN	0	5/83 (6.0)	6/83 (7.2)

^aTMP-SMX, trimethoprim-sulfamethoxazole; PIP/TAZ, piperacillin-tazobactam; SAM, ampicillin-sulbactam.

^bR, resistant; I, intermediate; S, susceptible.

^cGN, Gram negative; GP, Gram positive.

From 2 of 9 polymicrobial cultures, all isolates were correctly identified, comprising E. coli and E. cloacae isolates in one culture and E. faecalis and CoNS in the other culture. AST results were reported by APS for E. faecalis and not for the other three pathogens. The AST results for E. faecalis were in agreement with the results of conventional AST. For 6 of the 7 other polymicrobial cultures, AST by APS was provided for the one isolate of each culture that was identified by APS. AST results were in agreement with conventional culture for all tested antibiotics for 2 isolates. There were 10 discrepancies of 50 drug microbiota combinations in the 4 other isolates for which AST results were available.

DISCUSSION

To the best of our knowledge, this is the first study to analyze the performance of APS for rapid identification and drug susceptibility testing of organisms grown from positive blood culture bottles inoculated with primary sterile nonblood specimens. APS was in agreement for pathogen identification for 72 of 100 cultures and 72% (81 of 112) of the pathogens identified by conventional diagnostics. When considering on-panel organisms only, APS was in agreement with conventional diagnostics for 72 of 92 cultures (78%) and 81 of 104 (78%) pathogens.

Compared with the APS performance for pathogen identification from positive blood culture bottles inoculated with blood from septic patients that reported a correct identification between 73% and 95% (5–7), the performance from blood culture bottles inoculated with primary sterile specimens of patients with serious infections was at the lower range of these studies. Of note, various nonblood specimens from patients with suspected serious infections can include species not included in the APS panel that was specifically designed by the manufacturer for identification of pathogens from septic patients. In our study, 26% of the isolates not identified by APS (8 of 31) were not included in the APS panel. Cerebrospinal fluid was the specimen type with the highest number of samples in our study. The samples were mainly from neurosurgery patients with suspected bacterial meningitis caused by an infection of the extraventricular drain. This finding also explains the high number of staphylococci, including coagulase-negative staphylococci that are a common cause of extraventricular drain infection. Of nine isolates of species that can cause a rapid progression of acute bacterial meningitis (S. aureus, K. pneumonia, K. oxytoca, E. coli, and E. cloacae), APS correctly identified 8 isolates.; only 1 K. oxytoca isolate from a polymicrobial culture was not detected. However, the identification results cannot be translated to patients with community-acquired acute bacterial meningitis commonly caused by Neisseria meningitis, Streptococcus pneumoniae, Haemophilus influenzae, and Listeria monocytogenes, which are not included in the APS panel. It is likely that APS performed from cerebrospinal fluid of these patients results in a lower pathogen identification rate. Eight isolates from six different species identified by conventional diagnostics were not included in the APS panel. In contrast to Micrococcus luteus, Corynebacterium jeikeium and Bacillus cereus that are typically isolated as skin contaminates during sampling, Stenotrophomonas maltophilia and Haemophilus influenzae both cultured from ascites and Actinomyces neuii cultured from vitreous fluid were considered relevant pathogens causing infections.

The lower organism identification rate from polymicrobial cultures (22%, 2 of 9) than that from monomicrobial cultures (77%, 70 of 91) of our study indicates that incomplete pathogen identification should be considered for specimens with a high likelihood of a polymicrobial culture. In polymicrobial cultures, one microorganism may dominate so that the concentration of the other organisms may be below the limit of detection (LoD) for APS pathogen detection. The average LoD reported by the manufacturer for the Accelerate Pheno BC kit is 4 × 10⁸/ml for Gram-negative bacteria, 5 × 10⁸/ml for Gram-positive bacteria, and 2 × 10⁶/ml for Candida sp. (FDA DEN160032), which for most organisms is at or below the concentration at the time when blood cultures are detected to be positive by continuous blood culture monitoring systems. For the clinical study reported in the FDA evaluation of the automatic class III designation of the Accelerate Pheno BC kit that included 793 positive blood culture samples, 7 of 38 cultures that were determined as polymicrobial by the reference method were found to be polymicrobic by the APS BC kit.

We included cultures flagged positive for up to 24 hours before the start of APS, while the manufacturer recommends including positive cultures flagged positive for up to 8 hours only. The extension of the incubation period was due to the laboratory working hours weekdays from 7:30 a.m. to 4:00 p.m. and weekends from 8:00 a.m. to 12:00 p.m. This extension of the incubation period did not appear to negatively impact the identification rate in our study. The identification rate even tended to be increased (67% correct identification for cultures flagged positive within 8 hours [mean time, 3.32 hours] before the start of APS and 79% for cultures flagged positive between 8 and 24 hours before start of APS [mean time 12.9 hours]) for all cultures included in the study.

APS generated results of antimicrobial resistance testing for 70% of 81 identified organisms, which is lower than the findings reported from blood cultures of septic patients ranging between 77% and 91% (5–7). The overall categorical agreement for AST between APS and conventional culture was 91.2% for our study and ranged between 94 and 96% in studies evaluating the performance of APS for AST from blood cultures of septic patients (5–7). The majority of discrepancies with conventional diagnostics for Gram-positive bacteria were minor errors in coagulase-negative staphylococci related to trimethoprim-sulfamethoxazole. Seven of 11 discrepancies for AST between APS and conventional diagnostics in Gram-negative bacteria occurred in one K. pneumoniae isolate from a polymicrobial culture. Identification for Streptococcus sp. isolates by APS is performed to the genus level only, and no AST is performed for these isolates. Identification to the species level can be important, and the interpretation of the AST results for streptococci is dependent on the species.

APS has been shown to reduce the time to AST results from positive blood culture bottles in septic patients compared with conventional sepsis diagnostics. While rapid diagnostic tests in patients with bloodstream infections have only little effect without the support of antibiotic stewardship (19), the combination of APS with antibiotic stewardship programs has been shown to have the potential to improve antimicrobial treatment and to reduce costs, mainly driven by the reduction of hospitalization (20). Our study was not designed for and does not allow for studying cost effects of APS diagnostics in these patients, and it should be considered that processing primary sterile specimens differs from blood culture diagnostics in septic patients. In contrast to blood from septic patients, primary sterile nonblood specimens can be directly cultured onto agar plates due to the higher concentration of pathogens in native specimens than in blood from septic patients in parallel with the inoculation into blood culture bottles. Direct plating of sterile fluids may yield ID and AST results in comparable or even faster time periods as APS when microorganisms yield bacterial colonies the following day after plating and rapid AST is performed as described by EUCAST using disk diffusion from positive blood culture bottles. But often, automatic devices detect growth from blood culture bottles faster than colonies can be visible after agar plating. Of 14 samples that were processed in parallel with direct inoculation of sterile fluids onto agar plates and in liquid broths and were compared with incubation of blood culture bottles inoculated with primary sterile specimens, 3 isolates were not grown from conventional culture of native samples (agar plating and liquid broths), including 1 E. coli and 2 S. epidermidis isolates. Two more isolates (1 K. pneumoniae and 1 Actinomyces neuii) were not detected by agar plating but only from liquid broths. Except for Actinomyces neuii, which is not included in the APS BC panel, and S. mitis/S. oralis, for which APS does not perform AST testing, APS yielded earlier resistance results for all identified isolates than our conventional diagnostics. For 9 isolates, APS reported AST results at least 22 hours earlier than conventional diagnostics from culture of native samples (mean, 28 hours earlier; range, 22 to 46 hours earlier). This result was due mainly to early detection by the automated blood culture system and accelerated AST by APS. At the time the laboratory work of this study was performed, rapid AST performed by disk diffusion from positive blood culture bottles as recommended by EUCAST was not performed in our laboratory.

It should be considered that the APS BC kit used for the study was designed for identifying pathogens from blood of septic patients and not for identifying pathogens from primary sterile nonblood specimens that were included in our study.

However, the lower performance for correct identification and AST results, especially from polymicrobial cultures, and the higher costs of APS than those of conventional culture identification and AST should be taken into account for diagnostic and treatment decisions. Especially when considering de-escalating antimicrobials, these factors might be of relevance for microbiologists and clinicians.

The study was limited in that patients from only a single university hospital were included in the study. As our microbiology laboratory is not staffed 24/7 for routine diagnostic work, we also included blood culture bottles flagged positive outside the working hours and included them when they were flagged positive for up to 24 hours before the start of APS. The manufacturer recommends limiting the inclusion to bottles that turned positive for up to 8 hours. Nevertheless, the inclusion of cultures flagged positive between 8 and 24 hours before the start of APS showed a higher identification rate than cultures that were included within the first 8 hours after being flagged positive; so, this time extension did not appear to have a negative impact on identification results. Another limitation is the inclusion of a high number of cerebrospinal fluid samples and a low number of other nonblood specimens so that the results cannot be applied to all specimens. The number of yeast isolates cultured from the specimens was low, and three of four grown isolates were cultured from polymicrobial cultures. Polymicrobial cultures exhibited a lower performance so that the significance of the identification performance for yeasts is very limited. Furthermore, CoNS were overrepresented bacterial isolates, and the number of Gram-negative rods was limited in our study.

Our results indicate that APS may help for rapid identification and AST of organisms grown from positive blood culture bottles inoculated with primary sterile specimens from patients with suspected serious infections. Especially for hospitals without on-site microbiology laboratory and delays in sample processing, APS may accelerate identification and phenotypic AST from causative pathogens of serious infections from nonblood primary sterile specimens, when blood culture incubators and APS are available on-site. But, the inclusion of patient specimens with a high likelihood of infections caused by pathogens not included in the APS BC panel or inclusion of patients with polymicrobial infections may result in an inferior performance and may limit its value for de-escalating antimicrobial therapies for patients with a high probability of polymicrobial infections.