Biomedicine and Chemical Sciences

Bloodstream infections include a variety of clinical syndromes from transient bacteremia to septic shock. The importance of their prognosis is related to the susceptibility of the host, the source of the infection, the nature of the pathogen and the time until the onset of effective treatment. The ability to communicate with the laboratory quickly is therefore an integral part of sepsis management and not just a technical goal [1]. The problem of antimicrobial resistance also makes empiric treatment more difficult, especially when multidrug-resistant Gram-negative bacteria and methicillin-resistant staphylococci are prevalent. The current global burden of resistance is a reminder of the need for early identification of susceptible and resistant organisms [2].

Conventional blood culture is still the backbone as it allows the recovery of viable organisms, allows for assessment of any unusual agents and allows for comprehensive antimicrobial susceptibility testing. However, Gram staining, subculture, identification of the colonies and susceptibility testing follow a positive instrument alarm. The processing schedule of these sequential steps often carries over into the next day or two, and interlaboratory variations in processing schedules add further delay [3]. Therefore, treatment with broad-spectrum antimicrobials can be continued until an organism identification is available.

Specific molecular assays can be used to amplify pathogen-specific sequences directly from the positive bottle and provide a preliminary result in a few hours. A randomized study showed that integration of stewardship advice, not just the test result, is relevant to translating rapid identification into prescribing decisions [4]. Multicenter testing has already demonstrated that multiplex identification is possible from blood cultures [5] and additional genes relevant to beta-lactam, glycopeptide and carbapenem resistance have been added to the panels [6]. However, molecular panels are limited by the selection of targets, low microbial concentration, mixed cultures, and the fact that the resistance genes do not always correspond to the expressed phenotype.

While there is a growing number of high performing commercial assays described, laboratories implementing a targeted real-time PCR workflow will still need to locally validate against their isolate spectrum and workflow. Systematic evidence also suggests that measurable clinical impact is related to the use of these test results [7]. The present study was thus designed to compare the organism-level diagnostic agreement, four classes of resistance markers and reporting intervals for a multiplex real-time PCR strategy carried out as soon as a blood-culture bottle was flagged as positive.

Materials and Methods

Design the study and determine eligibility.Plan the study and establish eligibility.

A prospective paired comparison design was planned for a clinical microbiology laboratory of a tertiary care teaching hospital for 12 months. The first episode of blood culture that was positive per adult patient (age ≥18 years) was included in the analytic unit. All bottles that were positive on an automated blood-culture instrument were included, regardless of Gram-stain results. Bottles that were collected or labeled incorrectly, bottles that were repeated from the same episode, and bottles that did not have enough broth remaining for both methods were excluded. Consecutive sampling was chosen to minimize discretionary specimen selection and clinical data were abstracted in a de-identified case-report format.

Uncommon on-panel organisms and off-panel samples were represented by the target of 240 evaluable episodes. Assuming on-panel sensitivity of about 95%, the number of reference-positive specimens would be at least 150, which would give an approximate half-width of the confidence interval of 3.5 percentage points. This was the minimum that was planned for, but estimates for rare species and resistance genes were still likely to be imprecise.

Conventional laboratory workflow

A direct Gram stain was done immediately after a positive instrument alert, and subculture onto blood agar, chocolate agar and MacConkey agar as appropriate. The material suspected of being yeast was also cultured on fungal medium. Routine biochemical/automated identification and corroborative tests were used to characterize pure colonies, as needed. Susceptibility was determined by a validated automated MIC method with confirmatory disk diffusion or gradient diffusion, as appropriate; interpretations were based on the current CLSI standards adopted by the laboratory. Results and confirmatory tests for methicillin resistance, glycopeptide resistance, extended-spectrum beta-lactamase phenotype and carbapenem nonsusceptibility were examined. Culture adjudication was completed without the use of the index PCR result.

The workflow for multiplex real-time PCR.

200 μL of mixed bottle broth was processed for nucleic-acid extraction within 1 hour of positivity alert with an internal amplification control and extraction-negative controls. Validated multiplex reactions for Escherichia coli, Klebsiella pneumoniae, Staphylococcus aureus, coagulase-negative staphylococci (as a group), Enterococcus spp., Pseudomonas aeruginosa and the Acinetobacter baumannii complex were performed using fluorescent hydrolysis probes. Equivocal signals were followed by repeat testing and prespecified species/group targets and analytical cutoffs (cycle threshold ≤35 with acceptable control amplification) were applied consistently. A multiplex reaction was used to detect mecA, vanA/vanB, blaCTX-M, blaNDM and blaKPC. The marker targets were used to identify the susceptibility of a plausible corresponding cultured pathogen and not to claim the susceptibility of an unidentified pathogen.

The PCR technologist was not aware of the results of the culture other than the initial direct gram stain. The discordant results were reviewed for extraction controls, replicate amplification, organism purity and identification records. If the signal was for two pathogens in a specimen that was otherwise monomicrobial, it was not confirmed until other microbiological results were obtained. Reporting time was defined as the period between instrument positivity and validated organism or resistance-marker result, not between venepuncture and instrument positivity, to ensure that the comparison was of the post-positivity diagnostic interval.

Statistical analysis

Numbers were summarized as n (%), continuous measures were summarized as mean ± standard deviation, and diagnostic performance was summarized as two-sided 95% binomial confidence intervals. Operational comparators for target-level sensitivity, specificity, positive predictive value, and negative predictive value were culture recovery of an organism included in the assay. Cohen's κ measured binary agreement. Within-pair differences were examined and paired identification times compared by a paired-samples t-test or a Wilcoxon signed-rank test was prespecified. Only taxonomically appropriate isolates with a defined phenotypic comparator were used in the calculation of marker performance. The criteria for statistical significance was p<0.05 and the analyses were considered exploratory and not a test of patient outcome.

RESULTS

A total of 240 distinct positive-alert episodes were eligible for comparison. Patients had a mean age of 57.4 ± 16.2 years and 142 (59.2%) were male. Intensive-care patients contributed 96 episodes (40.0%), reflecting an enriched population at risk of complicated infection. Culture established 210 organisms within the seven targeted categories; 20 specimens yielded organisms outside the PCR panel and 10 produced no confirmed viable isolate on subculture (Table 1).

Table 1. Clinical and microbiological characteristics of the paired blood-culture cohort (N=240)

Percentages are based on all 240 positive-alert episodes; the last four microbiological categories are mutually exclusive.

PCR correctly identified 203 of 210 on-panel isolates, giving a target-level sensitivity of 96.7% (203/210). Two apparent organism detections occurred among the 30 off-panel/no-growth specimens, yielding a specificity of 93.3% (28/30), positive predictive value of 99.0% (203/205), and negative predictive value of 80.0% (28/35). Binary agreement was 231/240 (96.3%; κ=0.84). The seven false-negative results occurred across different groups without a dominant organism-specific failure pattern (Table 2).

Table 2. Conventional culture versus multiplex real-time PCR for on-panel organism identification

The 30 culture-negative-for-panel episodes comprised 20 off-panel isolates and 10 specimens without reproducible growth. Of these, 2 gave an unconfirmed panel-positive PCR result.

The difference in diagnostic workflow was reflected in mean times of 2.3 ± 0.7 hours for valid PCR organism identification and 25.8 ± 5.4 hours for conventional culture identification after positivity (mean paired difference, 23.5 hours; p<0.001). PCR marker reporting required 2.9 ± 0.9 hours, while complete phenotypic susceptibility was available after 49.6 ± 8.7 hours. These endpoints are not interchangeable: a negative gene panel does not exclude every clinically important resistance mechanism.

Within appropriate taxonomic strata, 15 of 16 methicillin-resistant S. aureus isolates carried detectable mecA; all five vancomycin-resistant enterococcal isolates carried vanA or vanB. Among Enterobacterales, blaCTX-M was identified in 34 of 39 isolates with an extended-spectrum beta-lactamase phenotype and in two of 61 without that phenotype; blaNDM or blaKPC was detected in 12 of 14 carbapenem-nonsusceptible isolates and none of 86 carbapenem-susceptible isolates (Table 3). Resistance-gene detection therefore gave timely mechanistic evidence but remained incomplete as a phenotypic resistance screen.

Table 3. Resistance gene signals in organism groups with applicable phenotypic comparators

Resistance phenotypes are based on the relevant drug/class comparator. Enterobacterales strata overlap because ESBL and carbapenem nonsusceptibility can coexist. A negative gene finding does not establish phenotypic susceptibility.

DISCUSSION

The paired analysis indicates that multiplex real-time PCR could significantly shorten the time between automated blood-culture positivity and preliminary pathogen identification. The high overall sensitivity is due to the 7-group assay which covers the most common bloodstream organisms, and the 96.3% total agreement is due to the detection of organisms, as well as a relatively small but clinically relevant number of discordant cases. The outcome should not be interpreted as a 96.7% sensitivity for the diagnosis of bloodstream infection in all patients; it was tested in bottles that had already tested positive and the assay was tested against an organism panel, not against all causes of sepsis.

These observations are in line with the results of expanded multiplex blood-culture panels, which highlight the high performance for covered organisms, but a continued limitation due to organisms not included in the assay menu [8]. In rare cases, a negative result may be obtained even if the organism is growing, due to the molecular method. Likewise, the small Acinetobacter subgroup in the present material does not give sufficient precision for independent clinical validation. Unexpected pathogens, mixed-species signals, detailed characterization and epidemiologic investigation require routine culture.

The second contribution is the resistance-marker comparisons. There was a good phenotypic correlation between mecA and vanA/vanB in these very specific strata, which could be used in early antimicrobial review. However, for Enterobacterales, blaCTX-M does not detect all the extended-spectrum beta-lactamase (ESBL) genotypes and blaNDM/blaKPC is only a part of the families of carbapenemases; other mechanisms can also lead to reduced carbapenem susceptibility. The two blaCTX-M-positive isolates that were not an ESBL phenotype illustrate the fact that molecular detection and observed MIC categories represent different but related properties. Narrowing therapy should therefore be based on the isolate, full antibiogram (if available), the severity of the infection and local stewardship guidelines.

Earlier identification has been shown to be beneficial in terms of better opportunities for targeted therapy in published evaluations, although the extent of benefit to the patient will depend on the communication pathway between the laboratory and the prescriber [9]. Rapid identification has also been correlated with early optimization opportunities, particularly when alerts are sent to antimicrobial stewardship staff, in multicentre clinical observations [10]. A pre/post study implementation study found that more patients were receiving appropriate therapy at an earlier decision point after the introduction of BCID2 [11]. The shorter laboratory turnaround time seen here does not necessarily imply a decrease in mortality, length of stay or cost, as none of these were endpoints in these studies.

Preanalytical carryover, contamination, extraction efficiency, assay inhibition, and Ct thresholds need to be considered for practical deployment. The seven cases that were missed on the panel should be retested and audit of bottle broth handling should be conducted. On the other hand, a false positive target could be due to low level microbial DNA, contamination or suppressed viability following antibiotic exposure. The 93.3% estimate for specificity is very uncertain because positive blood-culture signal without growth was included in the denominator. Economic modelling indicates that the implementation costs and stewardship capacity influenced by workflow will influence the value of rapid testing [12].

There is more published evidence to support separating the analytic accuracy from the clinical utility of a result. A diagnostic-accuracy meta-analysis of the BCID2 platform combined the data from nine studies including 2,005 blood cultures, and revealed high specificity for most of the pathogen targets, with the caveat that per-target precision is dependent on the number of positive blood cultures [13]. The seven group panel used in this assessment is not representative of a larger commercial panel and should not be interpreted as being as sensitive as a larger commercial panel. A later network meta-analysis of 88 studies and 25,682 encounters confirmed that blood culture alone was not associated with better outcomes than rapid diagnostics and antimicrobial stewardship, but that there was an equivalent survival benefit from rapid testing and antimicrobial stewardship [14]. These findings support pathways for testing implementation, rather than turnaround time of the instrument.

There's context to resistance detection, too. El Sherif et al. evaluated the commercial multiplex blood-culture identification panel versus VITEK-2 for central-line-associated infections, and showed the potential utility of an early report of both the organism and the resistance genes [15]. In another prospective study, Peri et al. found good agreement with traditional identification for covered pathogens and identified missed off-panel organisms and limitations in polymicrobial samples [16]. These reports should all be viewed in favor of a tiered laboratory policy: molecular reporting should clearly list the analytes being tested, explain the limitations of a negative resistance-marker result, and should not replace culture-based identification and full susceptibility testing.

The advantages of the design are the same-bottle paired testing, the use of a prespecified diagnostic comparator, the separation of the detection of the organism from the agreement between the organism and the gene, and the measurement of reporting time from a reproducible laboratory event. Some of the main limitations are the single-centre case mix, small numbers of individual pathogens, only positive-alarm bottles are included, there may be mixed-specimen contamination, sequencing may not be able to resolve all discrepant molecular calls, and there are no patient-level treatment outcomes. The endpoint timing also does not include the frequently long time interval between blood sampling and a culture bottle becoming positive, so the assay is actually an accelerated post-positivity identification strategy, rather than a culture independent early sepsis test.

CONCLUSION

A high level of concordance was found between multiplex real-time PCR and conventional organism identification, and there was a significant decrease in the time to preliminary reporting after a blood-culture positivity alert. The results of targeted mecA, vanA/vanB, blaCTX-M and blaNDM/blaKPC were useful for providing early resistance clues but were not a substitute for full phenotypic susceptibility testing. Prospective clinical evaluation should be integrated with Gram stain, subculture and antimicrobial stewardship.

REFERENCES

1. Timsit JF, Ruppé E, Barbier F, Tabah A, Bassetti M. Bloodstream infections in critically ill patients: an expert statement. Intensive Care Med. 2020;46(2):266-284. doi:10.1007/s00134-020-05950-6.
2. GBD 2021 Antimicrobial Resistance Collaborators. Global burden of bacterial antimicrobial resistance 1990-2021: a systematic analysis with forecasts to 2050. Lancet. 2024;404(10459):1199-1226. doi:10.1016/S0140-6736(24)01867-1.
3. Tabak YP, Vankeepuram L, Ye G, Jeffers K, Gupta V, Murray PR. Blood culture turnaround time in U.S. acute care hospitals and implications for laboratory process optimization. J Clin Microbiol. 2018;56(12):e00500-18. doi:10.1128/JCM.00500-18.
4. Banerjee R, Teng CB, Cunningham SA, et al. Randomized trial of rapid multiplex polymerase chain reaction-based blood culture identification and susceptibility testing. Clin Infect Dis. 2015;61(7):1071-1080. doi:10.1093/cid/civ447.
5. Salimnia H, Fairfax MR, Lephart PR, et al. Evaluation of the FilmArray blood culture identification panel: results of a multicenter controlled trial. J Clin Microbiol. 2016;54(3):687-698. doi:10.1128/JCM.01679-15.
6. Rhoads DD, Pournaras S, Leber A, et al. Multicenter evaluation of the BIOFIRE Blood Culture Identification 2 Panel for detection of bacteria, yeasts, and antimicrobial resistance genes in positive blood culture samples. J Clin Microbiol. 2023;61(6):e01891-22. doi:10.1128/jcm.01891-22.
7. D'Onofrio V, Salimans L, Bedenić B, Cartuyvels R, Barišić I, Gyssens IC. The clinical impact of rapid molecular microbiological diagnostics for pathogen and resistance gene identification in patients with sepsis: a systematic review. Open Forum Infect Dis. 2020;7(10):ofaa352. doi:10.1093/ofid/ofaa352.
8. Caméléna F, Péan de Ponfilly G, Pailhoriès H, et al. Multicenter evaluation of the FilmArray Blood Culture Identification 2 Panel for pathogen detection in bloodstream infections. Microbiol Spectr. 2023;11(2):e02547-22. doi:10.1128/spectrum.02547-22.
9. Sparks R, Balgahom R, Janto C, Polkinghorne A, Branley J. Evaluation of the BioFire Blood Culture Identification 2 panel and impact on patient management and antimicrobial stewardship. Pathology. 2021;53(7):889-895. doi:10.1016/j.pathol.2021.02.016.
10. Senok A, Al Dabal L, Alfaresi M, et al. Clinical impact of the BIOFIRE Blood Culture Identification 2 Panel in adult patients with bloodstream infection: a multicentre observational study in the United Arab Emirates. Diagnostics (Basel). 2023;13(14):2433. doi:10.3390/diagnostics13142433.
11. Donnars A, Mahieu R, Declerck C, et al. BIOFIRE Blood Culture Identification 2 (BCID2) panel for early adaptation of antimicrobial therapy in adult patients with bloodstream infections: a real-life experience. Diagn Microbiol Infect Dis. 2023;105(2):115858. doi:10.1016/j.diagmicrobio.2022.115858.
12. Pliakos EE, Andreatos N, Shehadeh F, Ziakas PD, Mylonakis E. The cost-effectiveness of rapid diagnostic testing for the diagnosis of bloodstream infections with or without antimicrobial stewardship. Clin Microbiol Rev. 2018;31(3):e00095-17. doi:10.1128/CMR.00095-17.
13. Peri AM, Ling W, Furuya-Kanamori L, Harris PNA, Paterson DL. Performance of BioFire Blood Culture Identification 2 Panel (BCID2) for the detection of bloodstream pathogens and their associated resistance markers: a systematic review and meta-analysis of diagnostic test accuracy studies. BMC Infect Dis. 2022;22:794. doi:10.1186/s12879-022-07772-x.
14. Peri AM, Chatfield MD, Ling W, Furuya-Kanamori L, Harris PNA, Paterson DL. Rapid diagnostic tests and antimicrobial stewardship programs for the management of bloodstream infection: what is their relative contribution to improving clinical outcomes? A systematic review and network meta-analysis. Clin Infect Dis. 2024;79(2):502-515. doi:10.1093/cid/ciae234.
15. El Sherif HM, Elsayed M, El-Ansary MR, Aboshanab KM, El Borhamy MI, Elsayed KM. BioFire FilmArray BCID2 versus VITEK-2 system in determining microbial etiology and antibiotic-resistant genes of pathogens recovered from central line-associated bloodstream infections. Biology (Basel). 2022;11(11):1573. doi:10.3390/biology11111573.
16. Peri AM, Bauer MJ, Bergh H, Butkiewicz D, Paterson DL, Harris PNA. Performance of the BioFire Blood Culture Identification 2 panel for the diagnosis of bloodstream infections. Heliyon. 2022;8(7):e09983. doi:10.1016/j.heliyon.2022.e09983.