Evaluation of three rapid antimicrobial susceptibility testing methods directly from positive blood cultures integrated with rapid pathogen identification

Abstract

Objective

Bloodstream infections (BSIs) is a major public health concern associated with high mortality. The Surviving Sepsis Guidelines recommend administering antibiotics within three hours after sepsis identified, thus, we aimed to optimize the procedure and reduce the turnaround time for pathogen identification and antimicrobial susceptibility testing.

Methods

We conducted a prospective study on positive blood culture flagged within 8 h by the BacT/Alert 3D system between November 2023 and July 2024. We enrolled single bacterial bloodstream infection by both the Gram staining and Liu’s staining and conducted a rapid pathogen identification by using Matrix-Assisted Laser Desorption Ionization Time-of-Flight Mass Spectrometry (MALDI-TOF MS) after a 3-hour incubation on agar plate. Appropriate antibiotics were chosen according to species identification for three RAST protocols, namely a novel short term agar plate or broth culture method and an integrated RAST protocol integrating CLSI and EUCAST criteria. First, the concordance rate between rapid and standard MALDI-TOF MS identification was evaluated. This was followed by an analysis of the categorical agreement (CA), along with minor discrepancy (mD), major discrepancy (MD), and very major discrepancy (VMD) rates, between rapid antimicrobial susceptibility testing (RAST) and the conventional methods (e.g., Vitek 2 automated system).

Results

A total of 112 monomicrobial positive blood culture samples were analyzed. The rapid pathogen identification method, reducing turnaround time by approximately 18 h, exhibited a concordance rate of 94.6% (95% CI, 88.8–97.5%). Gram-negative bacteria and Gram-positive cocci were accurately identified to the species level in 95.7% (95% CI, 85.8–98.8%) and 93.8% (95% CI, 85.2–97.6%) of cases, respectively. A total number of 595 inhibition zone diameters were generated in each method, including 374 Gram-positive cocci and 221 Gram-negative bacilli. Two novel RAST both achieved an overall CA of 95.5% (95% CI, 93.5–96.9%), with MD and mD below 3%. However, the hybrid RAST integrating CLSI and EUCAST standard demonstrated a MD of 3.7% (95% CI, 2.5–5.5%). When analyzed based on the antibiotic perspective, the below antibiotics all demonstrated 100% CA, including penicillin, cefoxitin, linezolid, and tetracycline against Gram-positive bacteria and ciprofloxacin, ampicillin and aztreonam against Gram-negative bacteria.

Conclusion

We have successfully developed and optimized a rapid identification and RAST workflow enabling swift and reliable detection of pathogens causing bloodstream infections.

Introduction

Bloodstream infection (BSI) is defined as the invasion of pathogenic bacteria into the bloodstream, followed by their proliferation and copious release of toxins and metabolites. This process triggers host cytokine responses, potentially leading to life-threatening conditions such as severe sepsis and septic shock [1, 2]. BSI exerts a substantial impact on morbidity and mortality, ranking within the top seven leading causes of death across North America and Europe according to recent epidemiological studies [3]. Effective antimicrobial therapy plays a pivotal role for patient survival. Several studies have shown a reduced length of stay and mortality associated with early and appropriate antimicrobials administration [4]. However, normal microbiological workflows involving pathogen isolation, identification, and antimicrobial susceptibility testing (AST) typically take up to 48 h to acquire antibiotic susceptibility profiles. This significant delay substantially increases the risk of inappropriate antibiotic use during this critical period, thereby escalating morbidity and mortality rates among bloodstream infection patients - particularly within the critically ill or septic population [5–7].

In response to the critical need for reducing turnaround time, the European Committee on Antimicrobial Susceptibility Testing (EUCAST) and the Clinical and Laboratory Standards Institute (CLSI) have developed rapid antimicrobial susceptibility testing (RAST) methodologies in 2020 and 2021, respetively. These initiatives introduced RAST protocols and established interpretive breakpoints for select antibiotics against common pathogens, thereby reducing the time required to report drug susceptibility results by a minimum of 24 h [8]. However, current guidelines exhibit notable limitations in coverage, the CLSI M100 document only provides some antibiotic breakpoints for Enterobacterals, Pseudomonas aeruginosa, and Acinetobacter baumannii, however, although the EUCAST guidelines partially address this gap by providing breakpoints for Gram-positive cocci like Staphylococcus aureus, Enterococcus, and Streptococcus pneumoniae, a critical standardization gap remains for other common pathogens in bacteremia, such as coagulase-negative staphylococci (CoNS) [9, 10].

Furthermore, accurate bacterial identification serves as an essential prerequisite for RAST protocols. By enabling rapid bacterial identification, we can preemptively select targeted antimicrobial agents and optimize agar plate preparation. Although literatures of the rapid pathogen identification by matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) has been published before, the main problem it encountered was the polymicrobial bloodstream infections, so how to identified these cases in the first step is very important [11, 12]. Conventionally, Gram staining was implemented to confirm the presence of bacteria/fungi, and differential bacteria in the blood culture, however, this method often encounters challenges in detecting Gram-negative bacteria or false positive samples in the clinical practice, frequently leading to misinterpretations.

And the incubation period was a key factor influencing the performance of the RAST either, it is reported that significant rates of errors could be observed after 6-hour incubation, and extended the culture time to 8 h could greatly improve the results [13]. Despite the short 8-hour incubation, 18.4% of the isolates remained in the zone of technical uncertainty [14]. In this study, we optimized the traditional workflow enabling precise identification of bloodstream infections, while implementing two innovative RAST methodologies to significantly reduce turnaround time and providing more species and antibiotics coverage in Chinese patients with bloodstream infections.

Materials and methods

Blood culture samples

This study was conducted at two medical centers of Guangzhou Women and Children’s Medical Center from November 2023 to July 2024. The Ethics approval was not required for this study, as it utilized only existing specimens and mainly focused on bacteria. Blood specimens were inoculated into blood culture bottles and incubated in the BacT/ALERT 3D Microbial Identification System (Vitek, bioMérieux, Marcyd’Etoille, France) for continuous bacterial growth monitoring. Upon being flagged within 8 h, the positive blood culture bottle was confirmed as monomicrobial bloodstream infections via Gram staining and Liu’s staining microscopy, following which they were included in the study cohort. Polymicrobial and/or fungal infections were excluded through routine morphological analysis. Only the first positive blood culture isolate from each patient was included in the final analysis.

Rapid pathogen identification through 3-hour cultivation using MALDI-TOF MS

For positive blood culture samples, four drops or 50µL culture broth for Gram-negative bacilli and eight drops or 100µL for gram-positive cocci were extracted using 20-gauge venting needle or sterile syringe. Specimens were inoculated onto Columbia blood agar and chocolate blood agar plates, which were incubated at 35 ± 2 °C with 5% CO₂ for 3 h. Bacterial patina was picked from blood agar plates with a 1 µl plastic loop and spotted onto the target plates for MALDI-TOF MS analysis. (VITEK MS, bioMérieux). 1 µl of the matrix solution (α-cyano-4-hydroxycinnamic acid, VITEK MS CHCA) was added onto smears, after air drying at room temperature, the target plates were then loaded into the mass spectrometer for rapid pathogen identification by MALDI-TOF MS (Vitek, bioMérieux, Marcyd’Etoille, France). Identification results obtained via the rapid MALDI-TOF MS method were compared against standard 18–24 h overnight culture-based identification protocols. The internal quality control for the MALDI-TOF MS is performed on a rotating weekly basis using the following standard strains: ATCC 19,433, ATCC 13,048, and ATCC MYA-2950, with strain 8739 serving as an internal control in every assay for verification.

RAST protocol using agar plate patina as bacterial suspension

Bacterial patina selected for rapid species identification was used to prepare a 0.5 McFarland bacterial suspension. Then the suspension was uniformly inoculated across the entire surface of the Mueller-Hinton agar (MHA) plates using sterile cotton swabs - with Streptococcus spp. requiring Mueller-Hinton blood agar. The antimicrobial discs were selected to dispense onto the surface of the inoculated MHA plate according to the MALDI-TOF MS identification results. This study employed 16 commonly used antimicrobials in China, including penicillin (P, 10 µg), erythromycin (E, 15 µg), clindamycin (DA, 2 µg), tetracycline (TE, 30 µg), levofloxacin (LEV, 5 µg), cefoxitin (FOX, 30 µg), linezolid (LZD, 30 µg), vancomycin (VA, 30 µg) for Gram-positive cocci, and imipenem (IPM, 10 µg), aztreonam (ATM, 30 µg), ceftazidime (CAZ, 30 µg), ceftriaxone (CRO, 30 µg), ampicillin (AMP, 10 µg), ciprofloxacin (CIP, 5 µg), amikacin (AK, 30 µg) and piperacillin/tazobactam (TZP, 30/6µg) for Gram-negative bacilli. The antimicrobial discs were ordered from Oxoid (Thermo Fisher Scientific, UK). The contents of these discs were conformed to CLSI and EUCAST standards. Plates were incubated at 35 ± 2 °C for 16–18 h, with Streptococcus spp. plates additionally supplemented with 5% CO₂, after which growth inhibition diameters were measured. Escherichia coli ATCC 25,922 and Staphylococcus aureus 25,923 were used as the quality control for the antimicrobial susceptibility test every week. We performed the inter-observer comparison quarterly for zone diameter measurement and staining morphology interpretation to ensure the reliability, which achieved perfect agreement (100%).

RAST protocol using short term broth culture as bacterial suspension

Four drop or 50 µl positive culture broth for gram-negative bacilli and eight drop or 100µL for Gram-positive cocci were extracted and dispensed into one sterile tube contain 2 ml tryptic soy broth (TSB). Incubated the mixture in a shaker at 37 °C with a speed of 250 rpm for 3 h. Subsequently, 1 ml of the bacterial suspension was transferred into an EP tube, centrifuged at 10,000 rpm for 1 min, with the supernatant carefully removed. The pellet was washed with 1 ml of sterile distilled water, centrifuged again under the same conditions, and the supernatant was discarded to yield a clean pellet. This pellet was then used to prepare a 0.5 McFarland standard bacterial suspension. The standardized suspension was uniformly inoculated across a MHA plate with Streptococcus spp. requiring Mueller-Hinton blood agar, onto which appropriate antibiotic discs were placed. The plates were then aerobically incubated at 35 °C ± 2 °C, with Streptococcus spp. plates additionally supplemented with 5% CO₂. and the diameters of growth inhibition zones were measured after an incubation period of 16–18 h. The quality control and the inter-observer comparisons for zone diameter measurement and staining morphology interpretation were performed as above.

Integrated RAST protocol integrating CLSI and EUCAST

According to CLSI and EUCAST guidelines, positive blood culture samples flagged by the BacT/Alert 3D automated system within 8 h were processed as follows: 50 µl (equivalent to four drops) of broth for Gram-negative bacilli and 100 µl (eight drops) for Gram-positive cocci were extracted from the positive blood culture bottles and dispensed onto MHA plates, with Streptococcus spp. requiring Mueller-Hinton blood agar. The inoculum was uniformly streaked across the entire agar surface using sterile swabs. Antimicrobial susceptibility discs were then applied, and plates were incubated at 35 ± 2 °C under ambient atmospheric conditions, with Streptococcus spp. plates additionally supplemented with 5% CO₂. Inhibition zone diameters were measured after 16–18 h of incubation. The quality control and the inter-observer comparisons for zone diameter measurement and staining morphology interpretation were performed as above.

RAST breakpoints used for interpretation of AST results

Antimicrobial susceptibility interpretive criteria were established according to the RAST 16–18 h breakpoints outlined in both CLSI and EUCAST. In instances where RAST-specific interpretive criteria were absent in the respective standards, conventional disk diffusion methodology outlined in the CLSI M100 (2024) document was employed as the reference methodology. Specific details were as follows: for Gram-negative bacilli, the CRO, CAZ, ATM, AMP and CIP breakpoints for Enterobacteriaceae, along with CAZ and CIP breakpoints for Pseudomonas aeruginosa, as well as CRO, CAZ and CIP breakpoints for Acinetobacter spp. were determined using RAST criteria outlined in CLSI 2024 M100. The IPM and AK breakpoints for Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, as well as IPM, AK for Acinetobacter baumannii adhered to the RAST criteria in EUCAST 2024. The AK and IPM breakpoints in Enterobacteriaceae other than E. coli and K. pneumoniae followed CLSI 2024 M100 standard zone diameter criteria. For Gram-positive cocci, FOX and DA breakpoints for Staphylococcus aureus, E and DA breakpoints for Streptococcus pneumoniae, LZD breakpoints for Enterococcus faecalis, LZD and VA breakpoints for Enterococcus faecium were defined using EUCAST 2024 RAST guidelines. Breakpoints for other Gram-positive cocci, such as CoNS isolates, were followed conventional standard CLSI 2024 M100 zone diameter criteria. The experimental procedure was illustrated in Fig. 1.

Fig. 1.

The rapid pathogen identification and three RAST methods for positive blood culture in this study

Analyse of RAST data

The antimicrobial susceptibility testing results were compared with those conventionally acquired by the Vitek 2 automated system. In addition, bacterial isolates exhibiting zone diameters within the Area of Technical Uncertainty in RAST system were categorized as indeterminate due to inconclusive categorization as either susceptible or resistant. The performance verification of antimicrobial susceptibility testing was conducted accorded to CLSI M52 guidelines for each antimicrobial agent tested, including categorical agreement (CA), minor discrepancy (mD), major discrepancy (MD), and very major discrepancy (VMD) [15]. CA was defined as the percentages of agreement between commercial and RAST methods in classifying isolates as either susceptible (S) or resistant (R), VMD (RAST = S and Vitek2 system = R), MD (RAST = S and Vitek2 system = R), and mD (RAST = S or R and Vitek2 system = I, RAST = I and Vitek2 system = S or R). In compliance with the CLSI M52 standard, which mandates minimum quality thresholds of ≥ 90% CA, with VMD and MD rates not exceeding 3.0% [15]. Statistical analyses were performed using Excel and SPSS, Version 25.0. Confidence intervals were determined using the online proportion confidence interval calculator from Statistics Kingdom [16].

Results

Rapid pathogen identification from positive blood culture

A total of 112 monomicrobial specimens were enrolled through Gram staining and Liu’s staining from November 2023 to July 2024, including 65 Gram-positive cocci and 47 Gram-negative bacilli. Five poly-microbial bloodstream infection samples were excluded base on this staining protocol. Subsequent identification revealed the following microbial combinations: two samples contained CoNS and Enterococcus faecalis; one sample was positive for Acinetobacter baumannii and Enterococcus faecalis; one sample showed growth of Escherichia coli and CoNS, and one sample was a combination of fungus and CoNS. We rapidly identified 112 positive monomicrobial blood cultures after agar cultivation for three hours. The most prevalent bacteria were Staphylococcus epidermidis, accounting for 24.1% (27/112), and followed by Escherichia coli accounting for 14.3% (16/112) and Staphylococcus hominis accounting for 12.5% (14/112). The distribution of bacteria was shown in Table 1. Compared with the conventional identification method using bacterial isolation by 18–24 h cultivation, the rapid pathogen identification exhibited an average concordance rate of 94.6% (95% CI, 88.8–97.5%). Specifically, it was 93.8% (95% CI, 85.2–97.6%) for Gram-positive cocci and 95.7% (95% CI, 85.8–98.8%) for Gram-negative bacilli, respectively. The species identification accuracy of Staphylococcus spp. reached 100%(95% CI, 92.9–100.0%). However, for Streptococcus, the genus identification accuracy could achieve 100%, while the species identification accuracy was only 66.7%, as shown in Table 1.

Table 1.

Comparison of rapid bacteria identification and traditional identification of 112 strains of positive blood culture specimens by MALDI-TOF MS technology

Microoganism	Conventional identification(n)	Rapid identification(n)	coincidence rate(%)	95% confidence intervals(%)
Gram-positive cocci	65	61	93.8%	(85.2, 97.6%)
Staphylococcus spp.	50	50	100.0%	(92.9, 100%)
Staphylococcus aureus	3	3	100.0%
Staphylococcus hominis	14	14	100.0%
Staphylococcus epidermidis	27	27	100.0%
Staphylococcus haemolyticus	1	1	100.0%
Staphylococcus capitis	4	4	100.0%
Staphylococcus cohnii	1	1	100.0%
Streptococcus spp.	9	6	66.7%
Streptococcus pneumoniae	2	2	100.0%
Streptococcus oralis	1	1	100.0%
Streptococcus sanguis	1	0	0.0%
Streptococcus Parasanguis	1	1	100.0%
Streptococcus dysgalactiae	2	2	100.0%
Streptococcus midis	1	0	0.0%
Streptococcus salivarius	1	0	0.0%
Enterococcus spp.	6	5	83.3%
Enterococcus faecalis	5	4	80.0%
Enterococcus faecium	1	1	100.0%
Gram-negative bacilli	47	45	95.7%	(85.8, 98.8%)
Enterobacterales	37	37	100.0%	(90.6, 100.0%)
Escherichia coli	16	16	100.0%
Klebsiella pneumoniae	13	13	100.0%
Klebsiella aerogenes	2	2	100.0%
Enterobacter cloacae	1	1	100.0%
Serratia marcescens	1	1	100.0%
Salmonellae spp.	1	1	100.0%
Pantoea	1	1	100.0%
Raoultella ornithinolytica	2	2	100.0%
Acinetobacter spp.	4	4	100.0%
Acinetobacter baumannii	3	3	100.0%
Acinetobacter nosocomial	1	1	100.0%
Pseudomonas aeruginosa	3	2	66.7%
Pluralibacter gergoviae	1	0	0.0%
Burkholderia cepacia	1	1	100.0%
Pseudomonas oryzihabitans	1	1	100.0%
Total	112	106	94.6%	(88.8, 97.5%)

Antimicrobial susceptibility testing results via conventional methods

AST results by the conventional method, mainly from Vitek 2 automated system, showed that the Staphylococcus spp. exhibited substantial resistance to antibiotics. The highest rate of resistance was recorded in relation to penicillin 91.7% (44/48), followed by macrolide antibiotics (above 50%), and levofloxacin 42% (21/50). Additionally, 35 strains demonstrated resistance to cefoxitin. Among them, 34 were MRSCN and 1 was MRSA. The penicillin resistance rate among Streptococcus spp. was 22.2% (2/9), with this resistance primarily attributed to Streptococcus oralis and Streptococcus mitis. Among the Enterococcus spp. isolates tested, 66.7% (4/6) demonstrated intermediate susceptibility to erythromycin. Notably, all Gram-positive cocci isolates remained susceptible to linezolid and vancomycin, with no resistant phenotypes detected. Antimicrobial susceptibility testing showed that among Enterobacterales isolates, resistance was highest to ceftriaxone (50.0%, 18/38), followed by ceftazidime (21.6%, 8/37). Additionally, one strain of Enterobacter cloacae exhibited resistance to imipenem with a rate of 2.7% (1/37). By contrast, Enterobacterales demonstrated higher susceptibility to amikacin and piperacillin/tazobactam, with susceptibility rates over 80%. The Acinetobacter spp. exhibited a 25% (1/4) resistance rate to ceftazidime, piperacillin/tazobactam, and ciprofloxacin, and one strain of Carbapenem-resistant Acinetobacter baumannii (CRAB) was identified. Pseudomonas aeruginosa demonstrated 100% susceptibility to imipenem, amikacin, ciprofloxacin, piperacillin/tazobactam, as detailed in the Table 2.

Table 2.

The antimicrobial susceptibility testing results was performed by conventional methods among 112 BSI bacteria

Antibiotics	Sample n	S n(%)	I n(%)	R n(%)
Staphylococcus spp.
Penicillin	48	4(8.3%)	0(0.0%)	44(91.7%)
Erythromycin	50	12(24.0%)	1(2.0%)	37(74.0%)
Clindamycin	50	23(46.0%)	1(2.0%)	26(52.0%)
Levofloxacin	50	28(56.0%)	1(2.0%)	21(42.0%)
Cefoxitin	50	15(30.0%)	0(0.0%)	35(70.0%)
Linezolid	50	50(100.0%)	0(0.0%)	0(0.0%)
Vancomycin	50	50(100.0%)	0(0.0%)	0(0.0%)
Streptococcus spp.
Penicillin	9	4(44.4%)	3(33.3%)	2(22.2%)
Erythromycin	8	3(37.5%)	1(12.5%)	4(50.0%)
Clindamycin	7	4(57.1%)	0(0.0%)	3(42.9%)
Levofloxacin	7	5(71.4%)	2(28.6%)	0(0.0%)
Cefoxitin	1	0(0.0%)	0(0.0%)	1(100.0%)
Linezolid	9	9(100.0%)	0(0.0%)	0(0.0%)
Vancomycin	9	9(100.0%)	0(0.0%)	0(0.0%)
Tetracycline	2	0(0.0%)	0(0.0%)	2(100.0%)
Enterococcus spp.
Penicillin	6	5(83.3%)	0(0.0%)	1(16.7%)
Erythromycin	6	1(16.7%)	4(66.7%)	1(16.7%)
Levofloxacin	6	5(83.3%)	0(0.0%)	1(16.7%)
Linezolid	6	6(100.0%)	0(0.0%)	0(0.0%)
Vancomycin	4	4(100.0%)	0(0.0%)	0(0.0%)
Enterobacterales
Ceftazidime	37	24(64.9%)	5(13.5%)	8(21.6%)
Ceftriaxone	36	18(50.0%)	0(0.0%)	18(50.0%)
Imipenem	37	36(97.3%)	0(0.0%)	1(2.7%)
Amikacin	37	35(94.6%)	0(0.0%)	2(5.4%)
TZP	36	31(86.1%)	1(2.8%)	4(11.1%)
Ciprfloxacin	3	1(33.3%)	0(0.0%)	2(66.7%)
Ampicillin	2	1(50.0%)	0(0.0%)	1(50.0%)
Aztreonam	1	0(0.0%)	0(0.0%)	1(100.0%)
ESBL test	37	23(62.2%) (-)	0(0.0%)	14(37.8%) (+)
Acinetobacter spp.
Ceftazidime	4	3(75.0%)	0(0.0%)	1(25.0%)
Imipenem	4	3(75.0%)	0(0.0%)	1(25.0%)
TZP	4	3(75.0%)	0(0.0%)	1(25.0%)
Ciprfloxacin	4	3(75.0%)	0(0.0%)	1(25.0%)
Pseudomonas aeruginosa
Ceftazidime	3	2(66.7%)	0(0.0%)	1(33.3%)
Imipenem	3	3(100.0%)	0(0.0%)	0(0.0%)
Amikacin	3	3(100.0%)	0(0.0%)	0(0.0%)
TZP	2	2(100.0%)	0(0.0%)	0(0.0%)
Ciprfloxacin	3	3(100.0%)	0(0.0%)	0(0.0%)
Salmonellae spp.
Ceftazidime	1	1(100.0%)	0(0.0%)	0(0.0%)
Ceftriaxone	1	1(100.0%)	0(0.0%)	0(0.0%)
Imipenem	1	1(100.0%)	0(0.0%)	0(0.0%)
Other Non-Enterobacterales
Ceftazidime	2	1(50.0%)	0(0.0%)	1(50.0%)
Imipenem	2	2(100.0%)	0(0.0%)	0(0.0%)
Amikacin	1	1(100.0%)	0(0.0%)	0(0.0%)
TZP	1	1(100.0%)	0(0.0%)	0(0.0%)
Ciprfloxacin	1	1(100.0%)	0(0.0%)	0(0.0%)

Note: S sensitive, I intermediary, R  resistant, ESBLs (+) bacteria product Extended Spectrum β-Lactamases, TZP Piperacillin/tazobactam

Performance of three rapid antimicrobial susceptibility testing

A total of 597 zone diameters generated by three RAST methods, including 374 Gram-positive cocci and 221 Gram-negative bacilli, none VMD was observed, the RAST result were shown in the Tables 3 and 4. RAST results using the agar plate patina demonstrated 95.5% (95% CI, 93.5–96.9%) CA with reference method, showing 2.2% (95% CI, 1.3–3.7%) MD, and 2.2% (95% CI, 1.3–3.7%) mD. Notably, for Gram-positive cocci, the CA rate reached 97.1% (95% CI, 94.8–98.4%), with both MD and mD rates below 3%. Complete CA (100.0%) was observed in Streptococcus spp., Staphylococcus haemolyticus, S. cohnii, and Enterococcus faecium. When analyzed based on the antibiotic perspective, 100.0% CA rates were achieved for penicillin, tetracycline, vancomycin, and linezolid. MD originated from two isolates: Staphylococcus hominis against levofloxacin and Staphylococcus epidermidis against clindamycin and levofloxacin. Meanwhile, mDs were most frequently observed in erythromycin susceptibility assays, accounting for 87.5% (7/8) among Gram-positive cocci. For Gram-negative bacilli, this method demonstrated a CA rate of 92.8% (95% CI, 88.6–95.5%) and a MD rate of 4.5% (95% CI, 2.5–8.1%). When categorized by antibiotic class, ceftriaxone, ciprofloxacin, and ampicillin all demonstrated 100% CA rates. Amikacin and piperacillin/tazobactam exhibited particularly high MD rates of 16.7% (7/35) and 6.7% (3/41), respectively, primarily caused by Enterobacterales other than Escherichia coli and Klebsiella pneumoniae. Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa each exhibited a mD for ceftazidime, and a single mD event (2.1%; 1/47) was observed with imipenem in Serratia marcescens. Pantoea spp. demonstrated high error rates, with imipenem and ciprofloxacin being the only two antibiotics showing essential agreement with reference method results.

Table 3.

Comparative performance of three RAST methods from the bacterial perspective [n (%)]

Microoganism	Abs	CA			VMD			MD			mD
		1	2	3	1	2	3	1	2	3	1	2	3
Gram-positive cocci	374	97.1% (94.8–98.4%)	97.1% (94.8–98.4%)	91.7% (88.9–94.5%)	0.0% (0.0–0.0%)	0.0% (0.0–0.0%)	0.0% (0.0–0.0%)	0.8% (0.3–2.3%)	0.8% (0.3–2.3%)	2.7% (1.5–4.9%)	2.1% (1.1–4.2%)	2.1% (1.1–4.2%)	5.6% (3.7–8.4%)
S. aureus	17	15	15	15	0	0	0	0	0	0	2	2	2
S. epidermidis	161	156	156	155	0	0	0	2	2	2	3	3	4
S. hominis	83	81	81	76	0	0	0	1	1	2	1	1	5
S. haemolyticus	6	6	6	6	0	0	0	0	0	0	0	0	0
S. capitis	24	23	23	21	0	0	0	0	0	0	1	1	3
S. cohnii	6	6	6	6	0	0	0	0	0	0	0	0	0
S. pneumoniae	12	12	12	12	0	0	0	0	0	0	0	0	0
S. oralis	4	4	4	4	0	0	0	0	0	0	0	0	0
S. sanguis	4	4	4	3	0	0	0	0	0	0	0	0	1
S. Parasanguis	5	5	5	5	0	0	0	0	0	0	0	0	0
S. dysgalactiae	12	12	12	11	0	0	0	0	0	1	0	0	0
S. midis	5	5	5	5	0	0	0	0	0	0	0	0	0
S. salivarius	5	5	5	5	0	0	0	0	0	0	0	0	0
E. faecalis	24	23	23	15	0	0	0	0	0	4	1	1	5
E. faecium	6	6	6	4	0	0	0	0	0	1	0	0	1
Gram-negative bacilli	221	92.8% (88.6–95.5%)	92.8% (88.6–95.5%)	88.7% (83.8–92.2%)	0.0% (0.0–0.0%)	0.0% (0.0–0.0%)	0.0% (0.0–0.0%)	4.5% (2.5–8.1%)	4.1% (2.2–7.6%)	5.4% (3.1–9.3%)	2.3% (1.0-5.2%)	3.2% (1.5–6.4%)	5.4% (3.1–9.3%)
E. coli	81	80	80	76	0	0	0	0	0	0	1	1	4
K. pneumoniae	68	66	66	65	0	0	0	0	0	0	1	2	3
K.aerogenes	10	8	8	8	0	0	0	2	2	2	0	0	0
E. cloacae	5	3	3	3	0	0	0	2	2	2	0	0	0
S. marcescens	4	2	2	1	0	0	0	1	1	1	1	1	2
P.gergoviae	6	6	6	6	0	0	0	0	0	0	0	0	0
Pantoea	5	2	2	2	0	0	0	2	2	3	1	1	0
R. ornithinolytica	10	6	6	4	0	0	0	3	2	4	0	1	2
A. baumannii	12	12	12	12	0	0	0	0	0	0	0	0	0
A. nosocomial	4	4	4	4	0	0	0	0	0	0	0	0	0
P. aeruginosa	14	13	13	13	0	0	0	0	0	0	1	1	1
Salmonellae	2	2	2	2	0	0	0	0	0	0	0	0	0
Total	595	95.5%	95.5%	90.6%	0.0%	0.0%	0.0%	2.2%	2.0%	3.7%	2.2%	2.5%	5.5%
		(93.5–96.9%)	(93.5–96.9%)	(88.0-92.7%)	(0.0–0.0%)	(0.0–0.0%)	(0.0–0.0%)	(1.3–3.7%)	(1.2–3.5%)	(2.5–5.5%)	(1.3–3.7%)	(1.5–4.1%)	(4.0-7.7%)

Note: Method 1: Agar plate culture method, Method 2: Tryptic Soy Broth culture method, Method 3: CLSI and EUCAST –integrated RAST method, CA: categorical agreement, mD: minor discrepancy, MD: major discrepancy, VMD: very major discrepancy, 95% Confidence intervals were presented in parentheses following the percentage

Table 4.

Analysis percentage of CA, mD, MD, and VMD for different antibiotics of RAST and AST by Vitek 2 of Gram-positive Cocci and Gram-negative bacilli after 16–18 h

	Antibiotics	CA n (%)			VMD n (%)			MD n (%)			mD n (%)			ATU n (%)
	Method	1	2	3	1	2	3	1	2	3	1	2	3	1	2	3
Gram-positive cocci	Penicillin	60(100)	60(100)	60(100)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)
	Erythromycin	57(89.1)	56(87.5)	53(82.8)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	7(10.9)	7(10.9)	11(17.2)	0(0.0)	0(0.0)	0(0.0)
	Clindamycin	55(96.5)	55(96.5)	50(87.7)	0(0.0)	0(0.0)	0(0.0)	1(1.7)	1(1.7)	2(3.5)	1(1.8)	1(1.8)	5(8.8)	0(0.0)	0(0.0)	0(0.0)
	Levofloxacin	61(96.8)	61(96.8)	55(87.3)	0(0.0)	0(0.0)	0(0.0)	2(3.2)	2(3.2)	3(4.8)	0(0.0)	0(0.0)	5(7.9)	0(0.0)	0(0.0)	0(0.0)
	Cefoxitin	51(100)	51(100)	51(100)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)
	Linezolid	65(100)	65(100)	65(100)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)
	Vancomycin	13(100)	13(100)	8(61.5)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	5(38.5)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)
	Tetracycline	2(100)	2(100)	2(100.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)
Gram-negative bacilli	Ceftazidime	43(91.5)	43(91.5)	42(89.4)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	1(2.1)	4(8.5)	4(8.5)	4(8.5)	0(0.0)	0(0.0)	0(0.0)
	Ceftriaxone	37(100)	37(100.0)	31(83.8)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	6(16.2)	0(0.0)	0(0.0)	0(0.0)
	Imipenem	46(97.9)	46(97.9)	46(97.9)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	1(2.1)	1(2.1)	1(2.1)	0(0.0)	0(0.0)	0(0.0)
	Amikacin	35(83.3)	35(83.3)	34(81.0)	0(0.0)	0(0.0)	0(0.0)	7(16.7)	7(16.7)	7(16.7)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	1(2.3)
	TZP	41(91.1)	41(91.1)	40(88.9)	0(0.0)	0(0.0)	0(0.0)	3(6.7)	2(4.4)	4(8.9)	0(0.0)	2(4.5)	1(2.2)	1(2.2)	0(0.0)	0(0.0)
	Ciprfloxacin	10(100.0)	10(100.0)	10(100)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)
	Ampicillin	2(100.0)	2(100.0)	2(100)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)
	Aztreonam	1(100.0)	1(100.0)	1(100)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)	0(0.0)

Note: Method 1: Agar plate culture method, Method 2: Tryptic Soy Broth culture method, Method 3: CLSI and EUCAST-integrated RAST method, CA: categorical agreement, mD: minor discrepancy, MD: major discrepancy, VMD: very major discrepancy, ATU: technical uncertainty, TZP: Piperacillin/tazobactam

RAST result using short term broth culture demonstrated a total of 95.5% CA (95% CI, 93.5–96.9%), with 2.0% MD (95% CI, 1.2–3.5%), and 2.5% mD (95% CI, 1.5–4.1%). The CA result demonstrated identical performance to the RAST using the agar plate patina method. And the MD and mD between two newly RAST methods were almost the same. And the discrepancies are as follows: Among Raoultella ornithinolytica isolates, the agar plate culture method exhibited two MDs - one each for amikacin and piperacillin/tazobactam. Comparatively, the tryptic soy broth method demonstrated one MD with amikacin and one mD with piperacillin/tazobactam. Another difference is noted with Klebsiella pneumoniae, which was classified as ATU for piperacillin/tazobactam by the agar plate culture method, however, it was a mD with this method.

Hybrid RAST integrating CLSI and EUCAST guidelines demonstrated 90.6% CA (95% CI, 88.0-92.7%), however, it exhibited with 3.7% MD (95% CI, 2.5–5.5%) and 5.5% mD (95% CI, 4.0-7.7%), exceeding the standard. For Gram-positive cocci, this method showed 91.7% (95% CI, 88.9–94.5%) CA rate, with 2.7% MD (95% CI, 1.5–4.9%) and 5.6% (95% CI, 3.7–8.4%) mD rates. The MDs predominantly originated from vancomycin susceptibility testing among Enterococcus spp., and were associated with levofloxacin and clindamycin among CoNS isolates either. The mD predominantly associated with erythromycin, clindamycin, and levofloxacin susceptibility testing among Staphylococcus spp. and Enterococcus spp., with erythromycin accounting for 52.4% (11/21) of the mDs. When stratified by antibiotic class, this method demonstrated 100% CA across penicillin, tetracycline, cefoxitin, and linezolid. For gram-negative bacilli, the overall CA rate across all tested isolates was only 88.7%, with both the MD and mD rates exceeded 5.0%. Within the Enterobacterales order (excluding Escherichia coli and Klebsiella pneumoniae), the discrepancies were primarily caused by amikacin and piperacillin/tazobactam, accounting for 72.4% (21/29) of MD and 63.8% (37/58) of mD. Among Escherichia coli isolates, mD were identified in two instances each for ceftazidime and ceftriaxone susceptibility results. Klebsiella pneumoniae isolates exhibited mD patterns involving one case each for ceftazidime, ceftriaxone, and piperacillin/tazobactam. Furthermore, a singular mD event was documented for ceftazidime in Pseudomonas aeruginosa. Consistent with the two novel RAST methods, this approach also detected a mD rate of 2.1% for imipenem among Serratia marcescens isolates, and Pantoea spp. demonstrated elevated error rates across tested antimicrobials except for imipenem.

Comparative evaluation of three RAST methods

Both the novel RAST methods met the internationally accepted threshold for reliable antimicrobial susceptibility testing, with CA being more than 95% with VMD and MD below 3% when using the antimicrobial breakpoints established by both EUCAST and CLSI guidelines for RAST analysis. The agar plate patina and tryptic soy broth culture methods exhibited significantly superior performance relative to the Integrated RAST integrating CLSI and EUCAST guidelines, (P = 0.001). Particularly among Gram-positive cocci, two novel RAST methods demonstrated significantly superior performance, with observed variance exceeding Bonferroni-corrected significance thresholds (P = 0.001), while no statistically significant difference was observed between the results of two novel RAST methods. as shown in Table 5.

Table 5.

Comparison of the drug susceptibility categorical agreement rates (%) among three RAST methods for Gram-positive Cocci and Gram-negative bacilli

		method	CA	discrepancy	total	χ2	P
Total bacteria	method1		568	26	594	0.018	0.893
	method2		568	27	595
	method1		568	26	594	11.142	0.001
	method3		539	55	594
	method2		568	27	595	10.320	0.001
	method3		539	55	594
Gram-positive cocci	method1		363	11	374	0.001	1.000
	method2		363	11	374
	method1		363	11	374	10.090	0.001
	method3		343	31	374
	method2		363	11	374	10.090	0.001
	method3		343	31	374
Gram-negative bacilli	method1		205	15	220	0.030	0.863
	method2		205	16	221
	method1		205	15	220	2.279	0.131
	method3		196	24	220
	method2		205	16	221	1.800	0.180
	method3		196	24	220

Note: method1: Ager plate culture method; method2: Tryptic Soy Broth culture method;

method 3: CLSI and EUCAST RAST; discrepancy: minor discrepancy plus major discrepancy plus very major discrepancy; The number of CA and error refer to Table 3

Discussion

Bloodstream infections, a major global health threat, demand rapid pathogen detection and antibiotic resistance profiling for timely treatment. However, current conventional diagnostic pipelines - encompassing Gram staining, culture-based identification, and phenotypic AST - require approximately 48 h to obtain complete result [13]. Although CLSI (2021) and EUCAST (2020) issued RAST guidelines, challenges remain. In this study, we optimized the workflow to address these issues and significantly reduce turnaround time.

Firstly, despite the relatively low clinical prevalence of polymicrobial bloodstream infections [17, 18], MALDI-TOF MS demonstrates limited efficacy in resolving such mixed microbial specimens, consequently, morphological staining methods remain essential prior to rapid identification procedures. In clinical practice, Gram staining is routinely employed as a first-line diagnostic tool in morphology, however, it exhibited low sensitivity in detecting Gram-negative bacteria in flagged positive blood culture, which compromises detection accuracy in polymicrobial bloodstream infections containing predominantly Gram-positive and Gram-negative organisms [19]. To address this limitation, we propose a dual-staining protocol combined with Gram staining and Liu’s staining before rapid identification in this study. We excluded five specimens exhibiting multiple pathogens by this method, which ultimately enhanced the accuracy of bacteria identification by 4.2%. This may be because Liu’s staining does not require an alcohol decolorization step, which achieving enhanced structural preservation of bacterial morphology in specimen slides. This is particularly critical for Gram-negative bacilli, which exhibit significantly weaker adhesion to glass slide surfaces.

Secondly, RAST must be integrated with rapid bacterial identification techniques to achieve its full clinical potential. This synergy is critical because rapid bacterial identification could guide appropriate antimicrobial panel selection, agar plate and incubation environment selection, thereby improving RAST interpretive accuracy and greatly reduce the turnaround time. In this study, we collected 112 positive mono-microbial blood culture specimens, bacterial patina was prepared after a 3-hour incubation period, followed by rapid bacterial identification using MALDI-TOF MS. At the species level, the rapid pathogen identification method achieved 94.6% (106/112) concordance with the overnight culture method. This concordance rate is markedly higher than the 85.6% reported in a previous high-throughput identification study of clinical isolates in a routine microbiology setting [20]. The accuracy rate of rapid species identification for Enterobacteriaceae and Staphylococcus spp. was 100%, which is higher than that reported in other studies either [21–23]. However, for viridans streptococci, the accuracy at species level was significantly lower (66.7%). These findings aligned with previous reports highlighting the limitations of MALDI-TOF MS in discriminating closely related species within the viridans streptococci group, particularly when using standard database references [24]. Due to the slow growth of viridans streptococci, preparing a direct smear after only three hours of incubation may transfer an excessive amount of agar. This leads to increased background noise in the mass spectrum and reduces identification efficiency. This issue can be addressed by using broth culture followed by protein extraction, a protocol that increases bacterial protein yield and effectively suppresses background interference. Furthermore, since genus-level identification accuracy remains intact, the selection of antibiotics for rapid susceptibility testing is unaffected, even when species-level concordance varies.

While CLSI M100 lacks standardized RAST antibiotic breakpoints for comprehensive Gram-positive bacterial coverage, the EUCAST RAST guideline notably omits RAST breakpoints for CoNS and critical breakpoints for anti-MRSA agents (vancomycin, linezolid), as well as key Enterococcus spp. including E. faecalis and E. faecium. Such limitations necessitate the urgent integration of composite RAST interpretive criteria from multiple reference systems and supplemented with standard overnight CLSI breakpoints. In this study, we simultaneously used three different RAST methods to determine the antimicrobial susceptibility of BSI bacteria, including the CLSI and EUCAST RAST method and two novel culture-optimized RAST workflows. For the hybrid RAST integrating CLSI and EUCAST guidelines, the overall CA rate was 90.6%, VMD rate was 0%, MDs was 3.7%, and mDs was 5.5%. The CA rate marginally met the minimum requirement, however, the MD rate exceeded the recommended quality threshold by 1%. The CA rate for Gram-negative bacilli was 88.7%, with a MDs rate of 5.4%, both of which fell below the acceptable threshold. MDs were primarily associated with TZP and AK against Enterobacteriaceae excluding E. coli and K. pneumoniae. The mDs were predominantly linked to CRO and CAZ within the Enterobacteriaceae. The CA rate of Gram-positive cocci was 91.7%, VMD was 0%, MDs was 2.7%, and the mDs was 5.6%, which complied with the CLSI M52 standards. The MDs were predominantly observed in VA, LEV and DA, while the mDs were primarily associated with LEV, DA and E. The CA rates observed between Gram-negative bacilli and Gram-positive cocci were different from those of literature reports with better performance demonstrated for the former group [25]. This discrepancy may be attributed to four factors: ① We selected antimicrobial agents and agar plate via rapid pathogen identification. ② We combined CLSI and EUCAST breakpoints to determine antibiotic susceptibility. ③ We implemented 16–18 h interpretive criteria to ensure optimal bacterial growth. ④ This study included a broader range of Enterobacteriaceae species, like Pantoea compared to previous investigations. Following optimization, the CA rates achieved clinical acceptability and the bacterial coverage was excellent, with great improvement observed among Gram-positive cocci. Additionally, this method requires no additional investment. As for labor, the protocol is designed to integrate seamlessly without increasing workload, making it more cost-effective.

Two novel RAST methodologies developed in this study both demonstrated superior performance metrics, with comparable efficacy observed between them, both achieving an overall CA of 95.5% (Gram-positive cocci: 97.5%; Gram-negative bacilli: 92.8%) with zero VMD and both MD and mD rates below 3.0%, fully compliant with CLSI standards. The CA rates for all tested antibiotics were above 90% except erythromycin and amikacin, and the vancomycin of which reached 100% despite lacking established breakpoints in the current EUCAST and CLSI RAST guideline. For Gram-positive cocci, both MDs and mDs were observed at rates of 0.8% and 2.1%, respectively. These discrepancies were predominantly associated with LEV and E. The error rates of both novel methods are lower than those in other studies investigated [26, 27]. Surprisingly, the performance of two novel methods for Gram-negative bacteria demonstrated inferior performance compared to that of Gram-positive organisms. Among Gram-negative bacteria, the agar plate-based methodology demonstrated a MD rate of 4.5%, while the tryptic soy broth-based method showed a MD rate of 4.1%, just slightly below that of the CLSI and EUCAST RAST method. The origin of MDs was almost the same.

A comparative evaluation of three RAST methods by the chi-square test revealed that both novel RAST methods demonstrated significantly superior performance to the integrated CLSI and EUCAST RAST method (P = 0.001). Particularly among Gram-positive cocci, two novel RAST methods performed much better compared to the hybrid EUCAST/CLSI RAST method, with statistically significant differences observed (P = 0.001). Following methodological optimization, the CA rates observed among Gram-positive cocci exhibited progressive enhancement, this may attribute to the increase of bacterial inoculum in the novel methods. In contrast, no statistically significant differences were detected among Gram-negative bacilli, this finding suggests that when the incubation time was extended to 16–18 h in our study, the inoculum of bacteria exerted a negligible influence among Gram-negative bacilli.

Both novel RAST methods demonstrate comparable performance with no adverse events or specimen contamination observed. However, the experimental procedure using the agar plate method is simpler and, unlike the short-term broth culture, does not require additional equipment such as an incubator shaker or a centrifuge. Therefore, from the perspectives of cost and workflow efficiency, the agar plate method holds a distinct practical advantage. This method reduces the total turnaround time by approximately 18 h. This overall saving is achieved primarily by accelerating the identification step, while the duration of the RAST procedure itself remains comparable to that of existing RAST methods.

Limitations

Although a wide range of species and a large overall number of strains were included, the sample size for some species (e.g., viridans streptococci, Acinetobacter spp., and Pseudomonas aeruginosa) was relatively small. This limitation stems from the lower incidence of these bacteria during the study period. This may affect the robustness of findings for these specific groups. Future work will aim to expand the collection of such strains to strengthen the conclusions.

In conclusion, our method enabled rapid bacterial identification before RAST, significantly improved the targeted selection of both antibiotics and specialized agar plates for fastidious pathogens. The CLSI and EUCAST RAST could combine together to provide a broader spectrum of clinically relevant bacterial species coverage. Notably, two novel RAST methods performed better, particularly among Gram-positive cocci, bringing significant improvement in clinical and economic benefits for patients with bloodstream infections.

Abbreviations

BSIs

Bloodstream Infections

MALDI-TOF MS

Matrix-Assisted Laser Desorption Ionization Time-of-Flight Mass Spectrometry

RAST

Rapid Antimicrobial Susceptibility Testing

CLSI

Clinical and Laboratory Standards Institute

EUCAST

European Committee on Antimicrobial Susceptibility Testing

CA

Categorical Agreement

mD

Minor Discrepancy

MD

Major Discrepancy

VMD

Very Major Discrepancy

P

Penicillin

E

Erythromycin

DA

Clindamycin

TE

Tetracycline

LEV

Levofloxacin

FOX

Cefoxitin

LZD

Linezolid

VA

Vancomycin

IPM

Imipenem

ATM

Aztreonam

CAZ

Ceftazidime

CRO

Ceftriaxone

AMP

Ampicillin

CIP

Ciprofloxacin

AK

Amikacin

TZP

Piperacillin/tazobactam

Authors’ contributions

Bingshao Liang: Funding acquisition, Writing – original draft. Xueying Li: Methodology, Writing – original draft. Qiyin Zheng: Methodology. Hao Cai: Methodology. Sufei Zhu: Methodology. Meng Zhang: Methodology. Zhenting Huang: Methodology. Qiulian Deng: Methodology and Data curation. Huamin Zhong: Methodology and Formal Analysis. Fei Gao: Methodology and Data curation. Qian Xu: Funding acquisition. Yan Long: Funding acquisition, Writing – review & editing. Jielin Wang: Conceptualization, Investigation.

Funding

This work was supported by the National Natural Science Foundation of China (No. 82002202), Guangdong Provincial Clinical Research Center for Laboratory Medicine (2023B110008), Guangzhou Municipal Science and Technology Bureau (202201020654), and Traditional Chinese Medicine Bureau of Guangdong Province (No. 20251132).

Data availability

All data generated or analysed during this study are included in this published article.

Declarations

Ethics approval

The Ethics Committee of the Guangzhou Women and Children’s Medical Center approved the study protocol (No. 421B00, 2022). The study was conducted in accordance with the Declaration of Helsinki. Written informed consent was exempted, as it utilized only existing specimens and mainly focused on bacteria.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.