Evaluation of the Qvella FAST System and the FAST-PBC cartridge for rapid species identification and antimicrobial resistance testing directly from positive blood cultures

ABSTRACT Blood culture diagnostics require rapid and accurate identification (ID) of pathogens and antimicrobial susceptibility testing (AST). Standard procedures, involving conventional cultivation on agar plates, may take up to 48 hours or more until AST completion. Recent approaches aim to shorten the processing time of positive blood cultures (PBC). The FAST System is a new technology, capable of purifying and concentrating bacterial/fungal pathogens from positive blood culture media and producing a bacterial suspension called “liquid colony” (LC), which can be further used in downstream analyses (e.g., ID and AST). Here, we evaluated the performance of the FAST System LC generated from PBC in comparison to our routine workflow including ID by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry using Sepsityper, AST by automatized MicroScan WalkAway plus and directly inoculated disk diffusion (DD), and MICRONAUT-AM for yeast/fungi. A total of 261 samples were analyzed, of which 86.6% (226/261) were eligible for the comparative ID and AST analyses. In comparison to the reference technique (culture-grown colonies), ID concordance of the FAST System LC and Sepsityper was 150/154 (97.4%) and 123/154 (79.9%), respectively, for Gram positive; 67/70 (95.7%) and 64/70 (91.4%), respectively, for Gram negative. For AST, categorical agreement (CA) of the FAST System LC in comparison to the routine workflow for Gram-positive bacteria was 96.1% and 98.7% for MicroScan and DD, respectively. Similar results were obtained for Gram-negative bacteria with 96.6% and 97.5% of CA for MicroScan and DD, respectively. Taken together, the FAST System LC allowed the laboratory to significantly reduce the time to obtain correct ID and AST (automated MicroScan) results 1 day earlier and represents a promising tool to expedite the processing of PBC.

B loodstream infections (BSIs) encompass a large variety of pathogens such as bacteria or fungi causing high rates of mortality and morbidity (1)(2)(3).Ineffective treatment of BSI due to delayed pathogen identification (ID) and late antimicrobial susceptibility testing (AST) can lead to adverse outcomes (4).Sepsis, defined as an organ dysfunction caused by a dysregulated host response to infection, is considered a major threat and one of the most important causes of health loss and deaths (4,5).Previous studies estimated around 49 million cases and 11 million deaths per year, which would correspond to 20% of all deaths in the world (5)(6)(7)(8).In this context, the rapid identification of pathogens followed by AST to enable appropriate antimicrobial therapy is crucial for effective management of septic patients.Since the standard method used in clinical laboratories is based on traditional culture methods and requires at least 24 hours for ID and 48 hours for final AST results, the development of more rapid methods for pathogen ID from positive blood culture (PBC) and subsequent resistance testing is of great interest.To this end, several alternative methods have been developed, which employ molecular detection methods on PBC, such as Biofire FilmArray (3,9,10), LightCycler SeptiFast (11), Genmark ePlex (12), Verigene Blood Culture panels (13), Accelerate Pheno system (14), and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) with specific kits (3,9,15).In this study, we evaluated the performance of the recently released FAST System (Qvella Corporation, Richmond Hill, Canada)-generated liquid colony (LC) in comparison to our routine workflow, including Sepsityper for rapid MALDI-TOF-based pathogen ID, AST using MicroScan WalkAway plus, and directly inoculated disk diffusion (DD).

Sample collection and study design
Between March and November 2022, a total of 261 prospective blood culture samples in different types of BD BACTEC blood culture bottles (Plus Aerobic, Plus Anaerobic, Lytic Anaerobic, Mycosis IC/F, and Peds Plus) (Becton Dickson, Heidelberg, Germany) were collected on different wards at Saarland University Medical Center in Homburg, southwest Germany.After being flagged positive in the BD BACTEC FX System (Becton Dickson), each sample was subjected to Gram staining and further processed using two procedures, which were performed in parallel: (i) the standard routine workflow and (ii) the FAST System workflow.

Inclusion/exclusion criteria
All positive blood culture samples obtained during routine diagnostics were included for the comparative assessment, except for the following exclusion criteria: post-mor tem blood culture, polymicrobial blood culture (i.e., if different morphologies were seen on Gram staining, samples were not subjected to Qvella FAST System; if cul ture-grown colonies after 18-24 hours of incubation showed previously unexpected polymicrobial growth, these samples were retrospectively excluded), processing of the blood culture ≥16 hours after being flagged positive, incorrect labeling, and failed runs (cartridge processing errors or system failures).

Routine workflow
Each PBC sample was subjected, using the Sepsityper Kit accordingly to the manufactur er's recommendations, to the Bruker BDAL database using MALDI Biotyper version 3.0 for MALDI-TOF MS ID (Bruker Daltonics, Bremen, Germany).Next, directly inoculated DD was performed using the interpretative recommendations put forth by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) protocol version 12.0, 2022 (16), in order to obtain a preliminary AST result within 18 hours.In brief, 125 ± 5 µL aliquots of positive blood were directly inoculated on Mueller-Hinton or Mueller-Hinton fastidious agar plates (BD Biosciences, Franklin Lakes, NJ, USA), followed by the addition of appropriate antibiotic disks or epsilometry (E-test) gradient strips according to the manufacturer's protocol (Liofilchem, Italy) and incubated at 35 ± 1°C.
In parallel, the reference method using agar plate subculture was conducted: colonies grown overnight on trypticase soy agar plates supplemented with 5% sheep blood, McConkey agar plates, Columbia agar plates (for anaerobic bacteria), and Sabouraud glucose agar plates (for fungi) (all plates produced by BD Biosciences) were analyzed by MALDI-TOF MS for species ID.Then, agar plate colonies were subjected to automatized AST employing broth microdilution method using the MicroScan WalkAway plus System (Beckman Coulter, Germany), with POS MIC33 panels (for Gram-positive bacteria; except for streptococci) and NEG MIC1 panels (for Gram-negative bacteria), according to the manufacturer's guidelines.If fungal species had been identified, antifungal susceptibility testing (AFST) using MICRONAUT-AM antifungal agent MIC plates (Bruker Daltonics, Germany) was performed.

FAST System workflow
The principle of the FAST System is based on the isolation and concentration of microbial cells directly from PBC using single-use FAST PBC Prep cartridges (Qvella Corporation, Canada).For sample preparation, 2-mL aliquots were drawn from PBC bottles.After a run of 24 (for one cartridge) or 38 (for two cartridges) minutes, the bacterial suspension called "liquid colony (LC)" was retrieved and further subjected to downstream analyses such as MALDI-TOF MS ID and AST.For the MALDI ID, 1 µL of the LC was spotted on the MALDI target plate in duplicate.After being dried, each spot was recovered with 1 µL of 70% (vol/vol) formic acid followed by the addition of 1 µL of saturated α-cyano-4-hydroxy-cinnamic acid matrix solution (Bruker Daltonics, Germany).For the AST, both manual (disk diffusion and/or E-tests) and automated ASTs were performed using a bacterial suspension of 0.5 McFarland prepared from the LC.Similar to the routine procedure, the automated AST method was performed using the MicroScan WalkAway plus System (Beckman Coulter, Germany), employing POS MIC33 panels (for Gram-positive bacteria; except for streptococci) and NEG MIC1 panels (for Gram-negative bacteria).

Comparative analysis
ID and AST (or AFST) results obtained with the two aforementioned workflows (routine workflow and FAST workflow) were compared for concordance.In addition, refer to the description of Inclusion/exclusion criteria, given above.

MALDI-TOF ID comparison
MALDI-ID obtained via Sepsityper or via the FAST System LC was compared to MALDI-ID obtained by culture-grown colonies as reference method.As two spots were meas ured for each sample, only the MALDI ID with the highest identification log score value was considered during the comparison (log score values ≥2 were considered as reliable species level ID, and score values ≥1.7, as probable species ID).Inconclusive IDs correspond to identification results with a score value below the threshold or "no peak found." Discrepant IDs correspond to identifications of different species for the same sample using two methods.

AST (or AFST) comparison
Results from DD (including E-tests) and automated AST (MicroScan WalkAway) (or AFST if fungal species were identified) obtained with the FAST System LC were compared to AST (or AFST) results obtained with the routine workflow.Categorical agreement (CA) was employed for the interpretation of the different AST results: "susceptible" (S), "resistant" (R), and "susceptible, increased exposure" (I) following the breakpoints put forth by EUCAST version 12.0 (16).Of note, AST results (i.e., MIC or diameter zone) in the area of technical uncertainty occurred in rare cases and were interpreted as given in Supplemen tal file S1-Table S1.Discrepancies were evaluated based on error types including minor errors (minE: results that are categorized increased [I] in one workflow but susceptible [S] or resistant [R] in the other), major errors (majE: results that are resistant [R] in the FAST System LC workflow but susceptible [S] in the routine workflow), and very major errors (VmajE: results that are susceptible [S] in the FAST System LC workflow, but resistant [R] in the routine workflow).Of note, as preliminary AST (i.e., directly inoculated DD) was compared with the DD results obtained from the FAST System workflow, isolates in which discrepancies (VmajE, majE, and minE) occurred were retested by performing standard DD from grown colonies obtained after reinoculation of cryo-cultures in order to clarify these discrepancies.The percentage of CA is calculated by dividing the number of categorical matches by the total number of antibiotics tested with the reference Full-Length Text Journal of Clinical Microbiology method × 100.The rate of VmajE was calculated by dividing the number of VmajE by the total number of resistant bacteria tested with the reference method × 100.The rate of majE was calculated by dividing the number of majE by the total number of susceptible bacteria tested with the reference method × 100.The rate of minE was calculated by dividing the number of minE by the total number of antibiotics tested with the reference method × 100.

Time to results
To estimate the rapidity of the FAST System workflow compared to the routine workflow, the time to ID and AST results was calculated by measuring the necessary time to obtain conclusive ID and AST results for both workflows.For the FAST System, the time was measured with a run of two cartridges simultaneously.Statistical tests were performed using analysis of variance (ANOVA), and a P-value <0.05 was considered as significant.

Study sample characteristics
Thirty-five out of 261 (13.4%) samples were excluded from the analysis, including 28 (10.7%)polymicrobial PBCs that were not detected by the initial Gram staining and 7 (2.7%) failed runs corresponding to cartridge processing errors or software failures.The remaining 226 (86.6%) samples were eligible for comparison.

AST (or AFST) comparison
Manual and automated AST (or AFST) results obtained with the two workflows (routine and FAST) were compared using categorical agreement based on EUCAST guidelines version 12. Automated AST results are presented in   4).
Furthermore, a complete overview of AST results is shown in Supplemental file S2-Table S4.Herein, for Gram-positive bacteria, the CA of the FAST LC in comparison to the routine procedure reached 96.1% and 98.4% for MicroScan and DD results, respectively.Similar results were obtained for Gram-negative bacteria, with CA of 96.6% and 97.5% for MicroScan and DD, respectively (Supplemental file S2-Table S4).However, a few discrepancies were also encountered.For MicroScan AST results of Gram-positive pathogens, a total of 50 errors were recorded.Most of the errors were minE (n = 22), detected mainly in S. epidermidis isolates with co-trimoxazole (13 discrepancies detected in 63 S. epidermidis tested) (Supplemental file S2-Table S1), followed by majE (n = 18) found mainly in S. epidermidis (three discrepancies with flucloxacillin, two with gentamicin, and two with co-trimoxazole detected in 63 S. epidermidis tested), and VmajE (n = 10) where gentamicin (n = 3) and flucloxacillin (n = 3) showed most discrepancies (Supplemental file S2-Table S1).When Gram-negative pathogens were tested, the majority of the 35 discrepancies were VmajE (n = 16) found mostly in coliform bacilli with fosfomycin (n = 5) being the most common discrepant substance, followed by minE (n = 14) and majE (n = 5) (Supplemental file S2-Table S1).Considering these results and to the best of our knowledge, our study is the first one combining the FAST LC with MicroScan WalkAway to assess automated AST.
For manual AST (DD), first, 53 discrepancies (21 for Gram positive and 32 for Gram negative) were found between the FAST System LC and our routine workflow based on direct inoculation DD of PBC, which was analyzed after 18 hours.Next, to clarify these errors, all isolates presenting these errors were re-analyzed by performing standard DD inoculated from grown colonies obtained after reinoculation of cryo-cultures on blood agar plates.It is important to highlight that 71.7% (38/53) of discrepant AST results were resolved, including 22 in Gram-negative bacteria and 16 in Gram-positive bacteria (Supplemental file S2-Table S2).Hence, the remaining discrepancies were for Gram-positive bacteria, seven VmajE mainly found in enterococci (n = 4) with gentamicin, tigecycline, and vancomycin, one majE in S. epidermidis with rifampicin, and one minE in S. mitis with co-trimoxazole (Supplemental file S2-Table S3).While for Gram-negative bacteria, 10 discrepancies were found: one VmajE in E. coli with piperacillin-tazobactam; three majE in E. coli (n = 2) with piperacillin-tazobactam and ceftazidime-avibactam and in K. variicola (n = 1) with cefotaxime; and 6 minE in E. coli, K. variicola, K. pneumoniae, and P. aeruginosa with ceftazidime and ciprofloxacin (Supplemental file S2-Table S3).
In addition, the two yeasts species (C.glabrata), which were correctly identified using the FAST System LC, showed 14/16 (87.5%)CA and 1/16 (6.3%) minE found with itraconazole.The only resistant profile found with itraconazole using the reference method showed a discrepancy (1/1 VmajE) when compared to the AST result obtained via the FAST System (Table 7; Supplemental file S2-Table S4).

Time to results
The time to obtain ID and AST results using the FAST System LC was estimated in comparison to our routine workflow.Here, the time required to obtain correct MALDI TABLE 4 Automated AST results from MicroScan WalkAway of Gram-negative bacteria (Enterobacterales and Pseudomonas spp.) following the FAST System workflow as compared to our routine workflow a ID and MicroScan AST results by means of FAST System LC was significantly reduced by 1 day as compared to the routine workflow using culture-grown colonies.Hence, the average time for the FAST System LC to obtain ID results was 1.11 (±0.16) hours.While for the workflow of the reference method using overnight culture in our labora tory, the average time was 27.4 (±7.30) hours.This difference is statistically significant (P < 0.05).In contrast, the difference between FAST and Sepsityper is not significant.For MicroScan AST, the average times for Gram-positive pathogens were 19.3 (±2.55) and 45.1 (±5.14) hours for FAST System LC and culture-grown colonies, respectively.Similarly, for Gram-negative pathogens, 22.2 (±2.6) and 44.5 (±3.5) hours were necessary to obtain MicroScan AST results, respectively, for FAST System workflow and culturegrown colonies.Both differences were statistically significant (P < 0.05).For DD AST, the turnaround time for both Gram positive and Gram negative was comparable, and there was no significant difference between FAST System and direct inoculation (Fig. 1).

DISCUSSION
One of the most important objectives in clinical microbiology is to provide a rapid and accurate microbiological diagnosis of BSI.Thus, tools to accelerate PBCs processing are urgently needed.Many of those tools are based on syndromic nucleic acid amplification such as multiplex PCR, which provides good sensitivity but are usually limited to a preselected panel of pathogens.Furthermore, such molecular techniques are limited by the types of organisms and resistance mechanisms included in the panel.In this study, we examined an alternative approach to accelerating culture-based processing.The FAST System generates a LC after an approximative run of 24 minutes by purifying the pathogens directly from the positive blood culture.Here, IDs (by MALDI-TOF MS) and AST (by MicroScan WalkAway plus and disk diffusion) results were analyzed and compared to those obtained via a culture-based routine workflow.By using the FAST System LC, correct and accurate ID results were achieved with a concordance of 96.9% as compared to the reference technique (culture-grown colonies).A concordance rate of 82.74% was achieved with Sepsityper.Similar results with Sepsityper were previously reported by Morgenthaler and Kostrzewa with 80% (n = 3320) species ID (17).Regarding the performance of the FAST System in terms of ID, Grinberg and colleagues reported an ID concordance of 94% (n = 201) using MALDI Biotyper as compared to colonies obtained by subculture (18).Similarly, Verroken and colleagues reported 89.5% (n = 266) concord ant ID results using the FAST System LC and MALDI Biotyper including 80 blood cultures bottles spiked with multidrug-resistant bacteria (19).Likewise, Kuo and colleagues demonstrated an accurate ID of 94.1% (272/289) using LC generated by the FAST System and MALDI-TOF MS Microflex (20).Fifty-four of the 272 samples corresponded to initially negative blood culture bottles which were seeded with key organisms (e.g., Streptococ cus agalactiae, Streptococcus pyogenes, Klebsiella spp., Proteus mirabilis, Enterobacter spp., Escherichia coli., etc.) in order to have a more diverse and larger set of bacteria with highly resistant patterns.All 54 samples (100%) were correctly identified with the LC suspen sion.Hence, by using combined prospective and spiked data set, they demonstrated that a wide array of clinically relevant Gram-positive and Gram-negative bacteria were identifiable using the LC suspension produced by the FAST System (19,20).
In our study, FAST-DD led to slightly better results (CA of 98.7% for Gram positive and 97.5% for Gram negative) when compared to directly inoculated DD from PBC, which are part of our routine workflow, than the automated method using FAST-MicroScan (CA of 96.1% for both Gram positive and Gram negative).Comparable CA values were previously reported for both Gram-positive [97.4% (18), 97.7%-100% (19), and 99.5% (20)] and Gram-negative bacteria [98.5% (18), 97.8%-99% (19), and 97.8% (20)] where automated AST systems other than MicroScan WalkAway (i.e., bioMérieux Vitek 2 or BD Phoenix) were utilized.Given these previous studies, our study though is the first to use MicroScan WalkAway demonstrating the compatibility of this automated AST technology with the FAST System.Nevertheless, a few VmajE and majE were reported, which could have a clinical impact on patient treatment, especially for VmajE, where a patient might be treated with antibiotics that are supposed to be sensitive when they are resistant to that treatment.Most VmajE and majE were detected for staphylococcal and Enterobacterales species, which are also the most frequently detected pathogen groups, without any noticeable error accumulation for specific substances tested among those groups (for details refer to Results: AST comparison or Supplemental material S2-Tables S1 to S3).
An important limitation of the FAST System is the lacking ability to process polymicro bial PBC.This has to be taken into account, since polymicrobial BSIs are considerably detected with a range of 6%-32% of all diagnosed BSI (21,22).Another limitation can be seen in the fact that a single instrument can only process two PBC cartridges simul taneously.Therefore, for potential implementation in a routine workflow, one would consider having multiple instruments or selecting urgent samples (e.g., an ICU patient) that should be prioritized for processing.Only one instrument was utilized during the period of this study.
In addition, several limitations of our study are offered for consideration.First, it was a monocentric analysis with a moderate sample size.Second, the number of some pathogens (e.g., yeast) obtained during the investigation was too low to draw meaning ful conclusions.Hence, we can simply rely on this limited number of samples, which limits the impact of the described findings.It is important to also mention that the FAST Prep cartridges utilized in this study were designed, primarily, for bacteria only.However, the results obtained with yeasts indicate a potential application to yeast samples, as well.Therefore, it would be interesting to perform additional investigations on a larger number of yeast samples to accurately evaluate the performance of the FAST System LC for yeast ID and AFST.
In summary, the FAST System LC enables a turnaround time reduction of approxi mately 24 hours for conclusive ID results.In addition, MicroScan AST results could be obtained with a reduction time of 24 hours from PBCs with similar accuracy compared to subculture-based reference techniques.The system connects well with Bruker Biotyper MALDI-TOF MS and MicroScan WalkAway plus, as it provides enough biomass to perform ID and inoculate DD and MicroScan AST plates.Our findings and those made in previous studies indicate that the FAST System LC approach is capable of delivering more timely results that can improve the outcomes of patients suffering from BSI.

FIG 1
FIG 1 Average time required to obtain ID and AST results (by MicroScan WalkAway plus and direct-inoculation DD) using different methods: FAST System, Sepsityper, and culture-grown colonies.
with MALDI score values of 1.86 and 1.93), while they were not successfully identified when the routine workflow (Sepsityper) was applied.Average MALDI scores of both workflows reached reliable values, comparable to those obtained for culture-grown colonies.For Gram-positive bacteria, we obtained scores of 2.00, 2.05, and 2.19, respectively, for FAST System, Sepsityper, and culture-grown colonies.While for Gram-negative bacteria, scores were 2.16, 2.12, and 2.24, respectively, for FAST System LC, Sepsityper, and culture-grown colonies.Likewise, log score values ≥2.00 were obtained for all groups (staphylococci, enterococci, Enterobacterales, etc.

TABLE 1
Identification of Gram-positive bacteria isolated from PBCs by MALDI-TOF MS using three different methods a NA, not applicable.b Average MALDI score.c Reference technique.

TABLE 2
Identification of Gram-negative bacteria isolated from PBCs by MALDI-TOF MS using three different methods a NA, not applicable.b Average MALDI score.c Reference technique.

TABLE 3
Automated AST results from MicroScan WalkAway of Gram-positive bacteria (staphylococci and enterococci) following the FAST System workflow as compared to our routine workflow a

TABLE 5
Manual AST results from DD of Gram-positive bacteria (staphylococci, enterococci, streptococci, and M. luteus) following the FAST System workflow as compared to our routine workflow (DD) a