Abstract
Background
Bloodstream infections (BSIs) are a major cause of morbidity and mortality globally. Rapid and accurate pathogen identification and antimicrobial susceptibility testing (AST) are crucial for timely and effective treatment. Conventional blood culture (BC) workflows are time-consuming, typically requiring 18–48 h. Although integrating matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) with direct AST protocols is promising, certain limitations remain. Therefore, we aimed to establish a faster and more reliable method for the direct identification and AST of microorganisms from positive BC bottles using a membrane filtration-based protocol.
Results
The membrane filtration-based method achieved an overall identification success rate of 76.5%. Rates were highest for Gram-negative bacteria (88.1%), followed by anaerobes (80.0%), Gram-positive cocci (70.2%), and Gram-positive bacilli (43.8%). Yeast identification was unsuccessful. AST results showed strong concordance with those from the conventional method, with essential agreement (EA) exceeding 95% across all groups. For Gram-negative bacteria, EA was 98% and categorical agreement (CA) was 95.4%, with 3.6%, 0.5%, and 0.5% minor, major, and very major errors, respectively. EA and CA were 96.1% and 94.2% for Gram-positive cocci, and 95.5% and 93.4% for Streptococcus spp.
Conclusions
The membrane filtration-based method, effective for Gram-negative bacteria, reduced turnaround time by 10–12 h compared to conventional workflows. Further optimization can improve accuracy for Gram-positive bacteria and yeast, offering a promising, streamlined workflow to enhance BSI diagnostics and guide targeted antimicrobial therapy.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12866-025-04337-1.
Keywords: Bloodstream infection, MALDI-TOF MS, Antimicrobial susceptibility testing, Membrane filtration method, Rapid identification
Background
Bloodstream infections (BSIs) are a major global health concern and the leading cause of morbidity and mortality. In the United States, the estimated mortality rate ranges from 23.5 to 27.5 per 100,000 person-years [1, 2]. Rapid initiation of appropriate antimicrobial therapy is critical for improving clinical outcomes. For example, each hour of delay in treating septic shock is associated with a 7.6% decrease in survival [3]. Therefore, timely and accurate identification of causative pathogens and antimicrobial susceptibility testing (AST) are essential to guide effective and targeted therapies [4, 5].
The current diagnostic gold standard for BSIs is blood culture (BC), which involves incubation in liquid medium, subculturing onto solid agar plates, and subsequent identification and susceptibility testing of the isolates. Although used widely, this workflow is time-consuming, often requiring 18–48 h to produce actionable results [6]. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) has revolutionized microbial diagnostics by providing rapid and accurate identification of isolated colonies [7]. However, conventional MALDI-TOF MS protocols depend on isolated colonies, limiting their ability to accelerate the diagnostic timeline.
To overcome this limitation, direct identification of pathogens from positive BC bottles has emerged as a promising alternative. This approach eliminates the need for subculturing and reduces the time for identification and AST, enabling faster clinical decision-making. However, the direct analysis of BC samples is challenging due to the presence of host proteins, cellular debris, and blood-derived contaminants that may interfere with the spectral accuracy of MALDI-TOF MS [8].
Several sample preparation techniques have been explored to enhance direct identification, including chemical lysis with ethanol, formic acid, and ammonium chloride, as well as mechanical methods such as lysis-filtration and gel separation. Commercial kits have also been developed to facilitate this process [9–12]. Despite their effectiveness, many of these protocols are costly, labor-intensive, or require specialized handling.
Moreover, some studies have demonstrated success in integrating MALDI-TOF MS with direct AST protocols; however, significant limitations remain, including the lack of a comprehensive evaluation of direct AST results. Furthermore, current methods often achieve higher accuracy with Gram-negative bacteria but not Gram-positive cocci, such as Staphylococcus and Streptococcus, which are clinically significant pathogens [13, 14]. Additionally, the complexity and cost associated with existing protocols hinder their widespread adoption in routine clinical practice.
Therefore, we aimed to develop a simplified and cost-effective protocol using membrane filtration and centrifugation for direct identification and AST of microorganisms from BC bottles to provide a highly accurate, practical, and time-saving method suitable for routine use in clinical microbiology laboratories.
Materials and methods
Sample collection
BC bottles, including Bactec Plus Aerobic/F, Anaerobic/F, and Peds Plus/F Lytic bottles (Becton Dickinson, Sparks, MD, USA), were collected from patients with suspected BSIs at Tao-Yuan General Hospital, Taiwan, between March and November 2024. All bottles were incubated at 35 °C using the BACTEC™ FX system until flagged positive. Only BC bottles yielding a single microorganism confirmed by Gram staining were included in the study.
Conventional identification and AST
Upon a positive signal from a BC bottle, Gram staining was performed, followed by subculturing onto blood agar plates (BAP), chocolate agar plates (CAP), and CDC anaerobic agar plates (CDC-ANA). BAP and CAP were incubated at 37 °C in an atmosphere with 5% CO₂, whereas CDC-ANA plates were incubated in an anaerobic chamber at the same temperature. After incubation, colonies were analyzed using MALDI-TOF MS with Bruker Biotyper® 3.1 software (Bruker Daltonics, Bremen, Germany). Identification was considered species-level if the spectral score was ≥ 2.00, genus-level if between 1.700 and 1.999, and unreliable if < 1.70.
A standardized 0.5 McFarland suspension was prepared using single colonies from BAP. Based on the species identified, the appropriate BD Phoenix™ M50 AST panel was selected: NMIC-411 for Gram-negative bacteria, PMIC/ID-95 for Staphylococcus and Enterococcus spp., and SMIC/ID-8 for Streptococcus spp. AST was performed following the latest guidelines of the Clinical and Laboratory Standards Institute (CLSI) after 18–24 h of incubation.
In-House Lysis and microfiltration protocol
Three mL of culture fluid from each positive BC bottle was mixed with 1 mL of 1% Triton X-100 (Sigma-Aldrich, USA) using a syringe. The mixture was vortexed for 30 s and filtered through a sterile 10 μm Minisart® syringe filter (DTC SepsiFilt Kit, Taiwan) to remove blood cell debris. A total of 1.5 mL of the filtrate was centrifuged at 15,500 × g for 1 min at room temperature. The resulting pellet was used for MALDI-TOF MS identification and AST.
For MALDI-TOF MS analysis, a spectral score > 1.7 was considered sufficient for identification at the genus and species levels. After successful identification, the pellet was adjusted to a McFarland standard of 0.5, and the same AST was performed using the procedure described for the conventional method with the BD Phoenix™ M50 system.
Results
Comparison between in-house and conventional identification methods
We collected 578 monomicrobial-positive BC bottles from 500 patients at Tao-Yuan General Hospital. These included 336 Bactec Plus Aerobic/F, 228 Anaerobic/F, and 14 Peds Plus/F Lytic bottles. These samples were analyzed using the in-house method, which combined direct MALDI-TOF MS identification and BD Phoenix™ M50 panels for AST. The results were compared with those obtained using the conventional method. The identification success rates using the in-house membrane filtration method were 72.6% (244/336), 81.6% (186/228), and 85.7% (12/14) for Aerobic/F, Anaerobic/F, and Peds Plus/F Lytic bottles, respectively.
Of the 578 isolates, 280 (48.4%) were Gram-negative bacteria, 286 (49.5%) were Gram-positive bacteria, and 12 (2.1%) were yeast. Using the in-house method, the overall identification success rate was 76.5% (442/578). Success rates by microbial group were: Gram-negative bacteria (88.1%, 229/260; Table 1), anaerobic bacteria (80.0%, 20/25; Table 3), Gram-positive cocci (70.2%, 186/265; Table 2), and Gram-positive bacilli (43.8%, 7/16; Table 3). Yeast identification failed (0%, 0/12).
Table 1.
Gram-negative bacteria from monomicrobial blood cultures identified by in-house and conventional culture methods
| Group/Genus success rates |
Organisms | No. of isolates | success rates | |||
|---|---|---|---|---|---|---|
| Conventional ID(> 2.0) | In-house ID (> = 2) | In-house ID (1.7–1.999) | Mis-ID | |||
|
Acinetobacter sp 66.7% (26/39) |
Acinetobacter baumannii/complex | 28 | 18 | 2 | 0 | 71.4% |
| Acinetobacter baylyi | 2 | 0 | 1 | 0 | 50% | |
| Acinetobacter bereziniae | 2 | 2 | 0 | 0 | 100% | |
| Acinetobacter haemolyticus | 2 | 0 | 0 | 0 | 0% | |
| Acinetobacter junii | 2 | 0 | 2 | 0 | 100% | |
| Acinetobacter lowfii | 1 | 0 | 0 | 0 | 0% | |
| Acinetobacter ursingii | 2 | 1 | 0 | 0 | 50% | |
|
Aeromonas sp 80% (4/5) |
Aeromonas hydrophila | 1 | 1 | 0 | 0 | 100% |
| Aeromonas caviae | 1 | 0 | 0 | 0 | 0% | |
| Aeromonas sorbia/veronii | 3 | 3 | 0 | 0 | 100% | |
|
glucose non-fermentative GNB 88% (22/25) |
Alcaligenes faecalis | 1 | 0 | 0 | 0 | 0% |
| Achromobacter xylosoxidans | 4 | 2 | 2 | 0 | 100% | |
| Burkholderia cepacia | 2 | 1 | 1 | 0 | 100% | |
| Burkholderia cenocepacia | 1 | 1 | 0 | 0 | 100% | |
| Chryseobacterium arthrosphaerae | 1 | 0 | 1 | 0 | 100% | |
| Chryseobacterium indologenes | 2 | 2 | 0 | 0 | 100% | |
| Elizabethkingia anophelis | 2 | 2 | 0 | 0 | 100% | |
| Oligella urethralis | 1 | 1 | 0 | 0 | 100% | |
| Ralsotonia mannitolitylica | 1 | 0 | 0 | 0 | 0% | |
| Rosemonas mucosa | 1 | 1 | 0 | 0 | 100% | |
| Stenotrophornonas maltophilia | 7 | 5 | 2 | 0 | 100% | |
| Sphingomonas parapaucimobilis | 1 | 0 | 1 | 0 | 100% | |
| Sphingomonas paucimobilis | 1 | 0 | 0 | 0 | 0% | |
|
Enterobacterales 96.1% (149/155) |
Citrobacter koseri | 8 | 7 | 1 | 0 | 100% |
| Citrobacter braakii | 1 | 1 | 0 | 0 | 100% | |
| Citrobacter testosteroni | 1 | 0 | 0 | 0 | 0% | |
| Enterobacter asburiae | 1 | 1 | 0 | 0 | 100% | |
| Enterobacter cloacae | 13 | 11 | 2 | 0 | 100% | |
| Escherichia coli | 23 | 23 | 0 | 0 | 100% | |
| Enterobacter xiangfangensis | 1 | 1 | 0 | 0 | 100% | |
| Klebsiella aerogenes | 8 | 7 | 1 | 0 | 100% | |
| Klebsiella pneumoniae | 32 | 24 | 5 | 0 | 90.6% | |
| Klebsiella oxytoca | 2 | 2 | 0 | 0 | 100% | |
| Klebsiella veriicola | 2 | 2 | 0 | 0 | 100% | |
| Morganella morganii | 4 | 3 | 1 | 0 | 100% | |
| Proteus mirabilis | 16 | 15 | 1 | 0 | 100% | |
| Providencia rettgeri | 2 | 1 | 0 | 0 | 50% | |
| Prevetella denticola | 1 | 1 | 0 | 0 | 100% | |
| Serratia marcescens | 22 | 19 | 3 | 0 | 100% | |
| Serratia rubidaea | 2 | 1 | 1 | 0 | 100% | |
| Salmonella sp. | 16 | 13 | 2 | 0 | 93.8% | |
|
Fastidious GNB 40% (4/10) |
Sutterella wadsworthensis | 1 | 0 | 0 | 0 | 0% |
| Campylobacter coli | 1 | 0 | 1 | 0 | 100% | |
| Haemophilus influenzae | 4 | 1 | 1 | 0 | 50% | |
| Haemophilus parainfluenzae | 1 | 0 | 0 | 0 | 0% | |
| Moraxella sp. | 2 | 0 | 0 | 0 | 0% | |
| Neisseria subflava | 1 | 1 | 0 | 0 | 100% | |
|
P. aeruginosa 92.3% (24/26) |
Pseudomonas aeruginosa | 26 | 20 | 4 | 0 | 92.3% |
| Total | 260 | 194 | 35 | 0 | 88.1% | |
Table 3.
Gram-positive bacilli and anaerobic bacteria from monomicrobial blood cultures identified by in-house and conventional culture methods
| Group/Genus success rates |
Organisms | No. of isolates | success rates | |||
|---|---|---|---|---|---|---|
| Conventional ID(> 2.0) | In-house ID (> = 2) | In-house ID (1.7–1.999) | Mis-ID | |||
|
GPB 43.8%(7/16) |
Brevibacterium casei | 1 | 0 | 0 | 0 | 0% |
| Brevibacterium ravenspurgnese | 1 | 0 | 1 | 0 | 100% | |
| Bacillus cereus | 3 | 0 | 2 | 0 | 67% | |
| Bacillus sp | 2 | 0 | 0 | 0 | 0% | |
| Corynebacterium aurimucosum | 1 | 0 | 1 | 0 | 100% | |
| Corynebacterium afermentans | 1 | 0 | 0 | 0 | 0% | |
| Corynebacterium striatum | 1 | 0 | 0 | 0 | 0% | |
| Corynebacterium. matruchotii | 1 | 0 | 0 | 0 | 0% | |
| Corynebacterium jeikeium | 1 | 0 | 0 | 0 | 0% | |
| Eubacterium limosum | 1 | 0 | 1 | 0 | 100% | |
| Bifidobacterium longum | 1 | 0 | 0 | 0 | 0% | |
| Rothia amarae | 1 | 0 | 1 | 0 | 100% | |
| Rothia dentocariosa | 1 | 0 | 1 | 0 | 100% | |
| Total | 16 | 0 | 7 | 0 | 43.8% | |
|
Anaerobic bacteria 80% (20/25) |
Actinomyces sp | 2 | 0 | 0 | 0 | 0% |
| Bacteroides fragilis | 13 | 12 | 1 | 0 | 100% | |
| Bacteroides distasonis | 1 | 0 | 0 | 0 | 0% | |
| Bacteroides massiliensis | 1 | 1 | 0 | 0 | 100% | |
| Bacteroides thetaiotaomicron | 1 | 1 | 0 | 0 | 100% | |
| Bacteroides salyersiae | 1 | 0 | 1 | 0 | 100% | |
| Bacteroides splanchincus | 1 | 1 | 0 | 0 | 100% | |
| Fusobacterium mortiferum | 1 | 1 | 0 | 0 | 100% | |
| Clostridium ramosum | 1 | 1 | 0 | 0 | 100% | |
| Propionibacterium acnes | 1 | 0 | 0 | 0 | 0% | |
| Peptoniphilus harei | 1 | 1 | 0 | 0 | 100% | |
| Veionella sp. | 1 | 0 | 0 | 0 | 0% | |
| Total | 25 | 18 | 2 | 0 | 80% | |
Table 2.
Gram-positive Cocci from monomicrobial blood cultures identified by in-house and conventional culture methods
| Group/Genus success rates |
Organisms | No. of isolates | success rates | |||
|---|---|---|---|---|---|---|
| Conventional ID(> 2.0) | In-house ID (> = 2) | In-house ID (1.7–1.999) | Mis-ID | |||
|
S. aureus complex 65.3% (32/49) |
Staphylococcus aureus | 45 | 23 | 8 | 0 | 68.9% |
| Staphylococcus argenteus | 4 | 0 | 3 | 2 | 25% | |
|
CoNS and Micrococcus 73.6% (66/90) |
Staphylococcus carnosus | 1 | 0 | 0 | 0 | 0% |
| Staphylococcus capitis | 24 | 9 | 8 | 0 | 70.8% | |
| Staphylococcus caprae | 8 | 5 | 2 | 0 | 87.5% | |
| Staphylococcus cohnii | 4 | 1 | 1 | 0 | 50% | |
| Staphylococcus epidermidis | 18 | 5 | 8 | 0 | 72.2% | |
| Staphylococcus haemolyticus | 9 | 5 | 4 | 0 | 100% | |
| Staphylococcus hominis | 12 | 5 | 3 | 0 | 66.7% | |
| Staphylococcus lugdunensis | 3 | 2 | 1 | 0 | 100% | |
| Staphylococcus pettenkoferii | 1 | 0 | 1 | 0 | 100% | |
| Staphylococcus sprophyticus | 2 | 0 | 1 | 0 | 50% | |
| Staphylococcus sciuri | 2 | 0 | 0 | 0 | 0% | |
| Staphylococcus warneri | 3 | 2 | 1 | 0 | 100% | |
| Micrococcus luteus | 3 | 1 | 1 | 0 | 66.7% | |
|
Enterococcus sp: 92.1% (35/38) |
Enterococcus casseliflavus | 1 | 0 | 0 | 0 | 0% |
| Enterococcus faecalis | 22 | 15 | 6 | 0 | 95.5% | |
| Enterococcus faecium | 15 | 10 | 4 | 0 | 93.3% | |
|
Streptococcus sp: 60.2% (53/88) |
Streptococcus agalactiae | 21 | 13 | 5 | 0 | 85.7% |
| Streptococcus anginosis | 7 | 0 | 2 | 0 | 28.6% | |
| Streptococcus canis | 1 | 0 | 1 | 1 | 0% | |
| Streptococcus constellatus | 6 | 1 | 2 | 0 | 50% | |
| Streptococcus dysgalactiae | 16 | 1 | 8 | 0 | 56.3% | |
| Streptococcus equisimilis | 4 | 0 | 0 | 0 | 0% | |
| Streptococcus gallolyticus | 4 | 1 | 3 | 1 | 75% | |
| Streptococcus intermedius | 1 | 0 | 0 | 0 | 0% | |
| Streptococcus mutans | 1 | 0 | 1 | 0 | 100% | |
| Streptococcus oralis | 9 | 3 | 5 | 0 | 88.9% | |
| Streptococcus parasanguinis | 1 | 1 | 0 | 0 | 100% | |
| Streptococcus pneumoniae | 8 | 2 | 1 | 0 | 37.5% | |
| Streptococcus salivarius | 4 | 1 | 1 | 0 | 50% | |
| Streptococcus sanguinis | 1 | 0 | 1 | 0 | 100% | |
| Streptococcus pyogenes | 4 | 1 | 1 | 0 | 50% | |
| Total | 265 | 107 | 83 | 4 | 70.2% | |
Notably, the most frequent misidentifications occurred among Gram-positive cocci. For example, Staphylococcus argenteus was misidentified as Staphylococcus aureus or Staphylococcus schweitzeri, whereas Streptococcus canis was misidentified as Streptococcus dysgalactiae, and Streptococcus gallolyticus as Streptococcus alactolyticus.
Comparison of AST results
Bacterial isolates successfully identified using the in-house method were adjusted to a suspension concentration of 0.5–0.6 McFarland using bacterial pellets and subsequently subjected to AST. Of the 578 isolates, 254 were subjected to identification and AST using the in-house method. These included 134 Gram-negative bacilli, 52 Staphylococcus spp., 25 Enterococcus spp., and 43 Streptococcus spp.
For the 134 Gram-negative bacilli tested, involving 2,027 antimicrobial-agent combinations using the NMIC-411 panel, the in-house method showed strong concordance with the conventional method, achieving an essential agreement (EA) of 98%, a categorical agreement (CA) of 95.4%, a minor error (mE) rate of 3.6%, a major error (ME) rate of 0.5%, and a very major error (VME) rate of 0.5% (Table 4).
Table 4.
Comparison of AST results between in-house and conventional methods for Gram-negative bacilli using NMIC-411
| Antimicrobial agent | No. of test | EA(%) | CA(%) | mE(%) | ME(%) | VME(%) |
|---|---|---|---|---|---|---|
| Amikacin | 130 | 129 (99.2) | 128 (98.5) | 1 (0.8) | 1 (0.8) | 0 |
| Ampicillin | 104 | 103 (99) | 99 (95.2) | 4 (3.8) | 1 (1) | 0 |
| Ampicillin/sulbactam | 117 | 113 (96.6) | 103 (88) | 11 (9.4) | 1 (0.9) | 2 (1.7) |
| Cefazolin | 92 | 92 (100) | 90 (97.8) | 2 (2.2) | 0 | 0 |
| Cefepime | 130 | 125 (96.2) | 120 (92.3) | 8 (6.2) | 0 | 2 (1.5) |
| Cefmetazole | 104 | 101 (97.1) | 98 (94.2) | 4 (3.8) | 0 | 2 (1.9) |
| Cefotaxime | 121 | 119 (98.3) | 113 (93.4) | 8 (6.6) | 0 | 0 |
| Ceftazidime | 132 | 130 (98.5) | 128 (97) | 4 (3) | 0 | 0 |
| Ceftriaxone | 121 | 118 (97.5) | 114 (94.2) | 7 (5.8) | 0 | 0 |
| Ciprofloxacin | 119 | 119 (100) | 112 (94.1) | 7 (5.9) | 0 | 0 |
| Ertapenem | 106 | 102 (96.2) | 99 (93.4) | 3 (2.8) | 3 (2.8) | 1 (0.9) |
| Gentamicin | 130 | 130 (100) | 128 (98.5) | 2 (1.5) | 0 | 0 |
| Imipenem | 113 | 111 (98.2) | 107 (94.7) | 4 (3.5) | 2 (1.8) | 0 |
| Levofloxacin | 124 | 122 (98.4) | 119 (96) | 5 (4) | 0 | 0 |
| Meropenem | 128 | 126 (98.4) | 126 (98.4) | 1 (0.8) | 1 (0.8) | |
| Piperacillin/tazobactam | 130 | 124 (95.4) | 125 (96.2) | 2 (1.5) | 0 | 3 (2.3) |
| Sulfamethoxazole/trimethoprim | 126 | 123 (97.6) | 124 (98.4) | 2 (1.6) | 0 | |
| Total | 2027 | 1987(98%) | 1933(95.4%) | 72(3.6%) | 11(0.5%) | 11(0.5%) |
CA, categorical agreement; same categorical result (S, I, or R) as the reference method
EA, essential agreement; MIC within ± 1 twofold dilution of the reference MIC
VME, very major error; test method susceptible, reference method resistant
ME, major error; test method resistant, reference method susceptible
mE, minor error; discrepancy between intermediate and susceptible or resistant
Similarly, for 52 Staphylococcus and 25 Enterococcus isolates, 939 bacterial-antimicrobial combinations were analyzed using the PMIC/ID-95 panels. The in-house AST method achieved EA, CA, mE, ME, and VME rates of 96.1%, 94.2%, 3.4%, 0.7%, and 0.9%, respectively (Table 5).
Table 5.
Comparison of AST results between the in-house and conventional methods for Staphylococcus and Enterococcus using PMIC/ID-95
| Antimicrobial agent | No. of test | EA(%) | CA(%) | mE(%) | ME(%) | VME(%) |
|---|---|---|---|---|---|---|
| Ampicillin | 25 | 25 (100) | 25 (100) | 0 | 0 | 0 |
| Cefoxitin | 25 | 25 (100) | 25 (100) | 0 | 0 | 0 |
| Clindamycin | 52 | 50 (96.2) | 48 (92.3) | 2 (3.8) | 1 (1.9) | 1 (1.9) |
| Ciprofloxacin | 76 | 73 (96.1) | 68 (89.5) | 6 (7.9) | 1 (1.3) | 1 (1.3) |
| Daptomycin | 76 | 76 (100) | 76 (100) | 0 | 0 | 0 |
| Erythromycin | 74 | 68 (91.9) | 68 (91.9) | 3 (4.1) | 2 (2.7) | 1 (1.4) |
| Gentamicin | 52 | 49 (94.2) | 46 (88.5) | 5 (9.6) | 0 | 1 (1.9) |
| Gentamicin-Syn | 25 | 25 (100) | 25 (100) | 0 | 0 | 0 |
| Levofloxacin | 75 | 69 (92) | 64 (85.3) | 8 (10.7) | 2 (2.7) | 1 (1.3) |
| Linezolid | 77 | 75 (97.4) | 71 (92.2) | 6 (7.8) | 0 | 0 |
| Oxacillin | 51 | 50 (98) | 50 (98) | 0 | 1 (2) | 0 |
| Penicillin G | 50 | 49 (98) | 49 (98) | 0 | 0 | 1 (2) |
| Tetracycline | 77 | 76 (98.7) | 75 (97.4) | 2 (2.6) | 0 | 0 |
| Teicoplanin | 75 | 71 (94.7) | 74 (98.7) | 0 | 0 | 1 (1.3) |
| Sulfamethoxazole/trimethoprim | 52 | 51 (98.1) | 51 (98.1) | 0 | 0 | 1 (1.9) |
| Vancomycin | 77 | 77 (100) | 77 (100) | 0 | 0 | 0 |
| Total | 939 | 902(96.1) | 885(94.2) | 32(3.4) | 7(0.7) | 8(0.9) |
For the 43 Streptococcus isolates tested using the SMIC/ID-8 panel (469 combinations), the in-house method yielded an EA, CA, mE, ME, and VME of 95.5%, 93.4%, 4.5%, 0.4%, and 1.7%, respectively (Table 6).
Table 6.
Comparison of AST results between the in-house and conventional methods for Streptococci using SMIC/ID-8
| Antimicrobial agent | No. of test | EA(%) | CA(%) | mE(%) | ME(%) | VME(%) |
|---|---|---|---|---|---|---|
| Ampicillin | 39 | 38 (97.4) | 34 (87.2) | 5 (12.8) | 0 | 0 |
| Cefepime | 43 | 40 (93) | 39 (90.7) | 4 (9.3) | 0 | 0 |
| Cefotaxime | 43 | 43 (100) | 43 (100) | 0 | 0 | 0 |
| Ceftriaxone | 43 | 43 (100) | 43 (100) | 0 | 0 | 0 |
| Chloramphenicol | 43 | 42 (97.7) | 42 (97.7) | 0 | 0 | 1 (2.3) |
| Clindamycin | 43 | 38 (88.4) | 37 (86) | 2 (4.7) | 1 (2.3) | 3 (7) |
| Erythromycin | 43 | 41 (95.3) | 41 (95.3) | 0 | 0 | 2 (4.7) |
| Levofloxacin | 43 | 42 (97.7) | 41 (95.3) | 2 (4.7) | 0 | 0 |
| Penicillin G | 43 | 40 (93) | 36 (83.7) | 7 (16.3) | 0 | 0 |
| Tetracycline | 43 | 38 (88.4) | 39 (90.7) | 1 (2.3) | 1 (2.3) | 2 (4.7) |
| Vancomycin | 43 | 43 (100) | 43 (100) | 0 | 0 | 0 |
| Total | 469 | 448(95.5) | 438(93.4) | 21(4.5) | 2(0.4) | 8(1.7) |
Colony-forming unit (CFU) counts were measured before and after treatment to assess the potential impact of 1% Triton X-100 treatment on microbial viability. The results demonstrated that microbial viability was largely preserved, with post-treatment CFU counts ranging from 75 to 238% of the pre-treatment values, indicating no substantial loss of viable cells during the lysis step (Supplementary Table 1).
Discussion
Rapid and accurate identification of BSI pathogens, along with timely AST results, is crucial for optimizing antimicrobial therapy and improving patient outcomes. In this study, we evaluated an in-house method that combines direct MALDI-TOF MS identification with BD Phoenix™ M50 panels for AST, demonstrating high concordance with conventional methods and significantly reducing turnaround time.
Gram-negative bacteria had the highest identification success rate (88.1%). This is consistent with previous findings, where Gram-negative bacteria typically yielded more reliable MALDI-TOF MS spectra due to their thinner cell walls and higher protein release efficiency [15, 16]. For example, Enterobacterales and Pseudomonas aeruginosa had high identification rates (96.1% and 92.3%, respectively), supporting the clinical applicability of this method for common Gram-negative pathogens.
In contrast, Gram-positive bacteria, particularly Streptococcus spp., exhibited a lower identification rate (60.2%), likely due to thick peptidoglycan layers that hinder protein extraction and spectral clarity in MALDI-TOF MS analysis. The identification performance may be compromised by poor pellet quality, low biomass, clot formation, and the presence of residual blood proteins or polysaccharide capsules [16–18]. These biological and technical factors can hinder the efficient release and detection of ribosomal proteins, which are essential for accurate MALDI-TOF MS profiling. Moreover, the use of 1% Triton X-100 and the 10 μm filter for cell lysis and removal of host-derived impurities may require further optimization, as incomplete removal of blood cells, debris, or interfering substances can result in suboptimal lysis, ultimately affecting identification results. Previous meta-analyses and comparative studies have also reported these limitations [6, 15, 18]. While suboptimal blood volumes or prolonged time to positivity may affect microbial recovery and identification, these factors were not evaluated in this study. Assessing their potential impact is a future direction. The observed misidentification of Staphylococcus argenteus and Streptococcus canis further underscores the challenge of differentiating closely related species using MALDI-TOF MS. As both are rare or fastidious organisms, the errors likely reflect limitations in the MALDI-TOF MS database rather than sample preparation. This highlights the need for ongoing expansion and refinement of the reference library to improve identification accuracy for uncommon species. Enhanced lysis protocols (such as the inclusion of lysozyme or bead-beating steps) or more effective washing techniques may be explored in future studies to improve spectral quality for Gram-positive bacteria.
One methodological distinction in this study was the adoption of a lower MALDI-TOF MS score threshold (> 1.7) to define a successful identification. The standard criterion for species-level identification is typically ≥ 2.0; however, using > 1.7 increases sensitivity in direct blood culture identification when supported by appropriate preprocessing, as demonstrated by Tanner et al. [19]. This approach improved detection rates, but also elevated the risk of species-level misidentifications. In settings where rapid decisions are critical, such as intensive care units or emergency departments, accepting a lower threshold may be justified. We analyzed identification performance stratified by score ranges to evaluate the effect of the lower thresholds. Of the 446 isolates with scores > 1.7, 319 (71.5%) had scores ≥ 2.0, and all were correctly identified at the species level. The remaining 127 isolates (28.5%) had scores between 1.70 and 1.999, with 123 (96.9%) correctly identified and four instances of misidentification. These findings support the feasibility of using a threshold > 1.7 to enhance the identification rate while maintaining acceptable accuracy. However, clinicians must interpret such results with caution, especially when dealing with unusual pathogens or immunocompromised patients.
The AST results further validated the clinical utility of the in-house method. Across all tested bacterial groups, EA exceeded 95%, and CA similarly remained high. The low error rates observed indicate that using bacterial pellets directly obtained from blood cultures did not significantly compromise AST accuracy. Notably, Gram-negative bacilli, Staphylococcus spp., and Enterococcus spp. demonstrated high categorical and essential agreement, consistent with CLSI M52 performance criteria. Furthermore, several antimicrobials, such as vancomycin, cefotaxime, and ceftriaxone, showed 100% categorical and essential agreement for Streptococcus spp., supporting the potential for selective direct reporting of these agents used commonly in clinical practice. Conversely, the slightly higher VME rate observed for Streptococcus spp. (1.7%) requires attention, as it could lead to false susceptibility reports. This may reflect the inherent variability in minimum inhibitory concentration breakpoints among different streptococcal species or inconsistencies in their growth characteristics. Identifying such patterns of concordance and discordance may support the development of selective reporting strategies, whereby antimicrobials with high agreement are released directly, while others undergo confirmation against a reference.
A major advantage of the membrane filtration-based protocol is its significantly reduced diagnostic turnaround time. By eliminating the need for overnight subculturing, the workflow was shortened by 10–12 h. This time gain is particularly valuable for managing critical cases of sepsis and multidrug-resistant infections, in which early intervention is often associated with improved outcomes [3, 5]. Improvement in early identification of pathogens and susceptibility results supports the initiation of targeted antimicrobial therapy at an earlier stage, directly contributing to patient safety. By combining clinical and microbiological perspectives, this approach demonstrates clear multidisciplinary value in improving patient outcomes. From a laboratory operations perspective, such reductions may contribute to improved workflow efficiency and resource allocation.
Despite these advantages, this study has some limitations. Notably, the in-house method failed to identify yeast in any of the 12 positive samples. This limitation likely stems from the use of a membrane with a 10 μm pore size, which may have excluded yeast cells, commonly 5–10 μm in size or larger during budding or pseudohyphal phases [20]. Previous studies have emphasized that yeast identification requires larger pore sizes and more robust lysis and protein extraction protocols, including mechanical disruption and prolonged chemical treatment. To address this limitation, future protocol adaptations could involve the use of dual-filtration systems or yeast-specific preprocessing kits.
Although the method demonstrated good accuracy overall, its performance for Gram-positive bacteria was suboptimal. Further refinement of lysis conditions or improvements in pellet quality can improve protein yield and identification accuracy. As this was a single-center study, the generalizability of the findings may be limited. A multicenter validation would be ideal to assess reproducibility across different laboratory settings and enhance the external applicability of the method. Comparison with commercial kits may also help assess scalability and broader clinical applicability.
Conclusions
This study demonstrates that a simplified in-house protocol using membrane filtration and centrifugation can enable rapid and accurate identification and AST of microorganisms directly from positive BC bottles. This method significantly reduces the diagnostic turnaround time while maintaining a high level of agreement with conventional techniques, particularly for Gram-negative bacteria. However, the method needs to be refined further due to its limitations in yeast detection and Gram-positive bacterial identification. Importantly, this protocol was validated only in monomicrobial blood cultures; its performance for polymicrobial samples remains to be determined. Future enhancements in sample preparation, spectral database expansion, and multi-institutional validation could enhance its diagnostic performance. With continued optimization, this workflow holds strong potential for routine clinical implementation in microbiology laboratories.
Supplementary Information
Acknowledgements
Not applicable.
Clinical trial number
not applicable.
Authors’ contributions
All authors contributed to the conception and design of the study, data analysis and interpretation, drafting and revising of the manuscript, and approval the final version of the article.
Funding
This work was supported by grants from Tao-Yuan General Hospital, Ministry of Health and Welfare, Taoyuan, Taiwan (PTH113010).
Data availability
No datasets were generated or analysed during the current study.
Declarations
Ethics approval and consent to participate
This study was approved by the Human Research Ethics Committee of the Institutional Review Board (IRB) of Tao-Yuan General Hospital (IRB Number: TYGH112062). The requirement for informed consent was waived by the IRB, as blood cultures were not drawn specifically for this study. All blood samples were collected as part of routine clinical care, and only remnant specimens were used. This study was conducted in accordance with the principles of the Declaration of Helsinki.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Goto M, Al-Hasan MN. Overall burden of bloodstream infection and nosocomial bloodstream infection in North America and Europe. Clin Microbiol Infect. 2013;19:501–9. 10.1111/1469-0691.12195. [DOI] [PubMed] [Google Scholar]
- 2.Timsit JF, Ruppe E, Barbier F, Tabah A, Bassetti M. Bloodstream infections in critically ill patients: an expert statement. Intensive Care Med. 2020;46:266–84. 10.1007/s00134-020-05950-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Kuma A, Roberts D, Kenneth E, Wood D, Light B, Parrillo JE, Sharma S, et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit Care Med. 2006;34:1589–96. 10.1097/01.CCM.0000217961.75225.E9. [DOI] [PubMed] [Google Scholar]
- 4.Vlek ALM, Bonten MJM, Boel CHE. Direct matrix-assisted laser desorption ionization time-of-flight mass spectrometry improves appropriateness of antibiotic treatment of bacteremia. PLoS ONE. 2012;7:e32589. 10.1371/journal.pone.0032589. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kerremans JJ, Verboom P, Stijnen T, Hakkaart-van Roijen L, Goessens W, Verbrugh HA, et al. Rapid identification and antimicrobial susceptibility testing reduce antibiotic use and accelerate pathogen-directed antibiotic use. J Antimicrob Chemother. 2007;61:428–35. 10.1093/jac/dkm497. [DOI] [PubMed] [Google Scholar]
- 6.Jo SJ, Park KG, Han K, Park DJ, Park YJ. Direct identification and antimicrobial susceptibility testing of bacteria from positive blood culture bottles by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry and the Vitek 2 system. Ann Lab Med. 2016;36:117. 10.3343/alm.2016.36.2.117. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Stevenson LG, Drake SK, Murray PR. Rapid identification of bacteria in positive blood culture broths by matrix-assisted laser desorption ionization-time of flight mass spectrometry. J Clin Microbiol. 2010;48:444–7. 10.1128/JCM.01541-09. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Faron ML, Buchan BW, Ledeboer NA. Matrix-assisted laser desorption ionization–time of flight mass spectrometry for use with positive blood cultures: methodology, performance, and optimization. J Clin Microbiol. 2017;55:3328–38. 10.1128/JCM.00868-17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Tai-Fen L, Tsai-Wen W, Wei-Yu H, Xiang-Jun C, Hao-Chieh C, Yu-Tsung H. Comparison of a Sepsityper(R) kit and in-house membrane filtration methods for rapidly diagnosing positive blood cultures via MALDI–TOF MS. J Microbiol Immunol Infect. 2024;58:265–71. 10.1016/j.jmii.2024.11.007. [DOI] [PubMed] [Google Scholar]
- 10.Maddalena C, Giuseppe M, Perez M, Guido S, Stefania S, Correction. Evaluation of the liquid colony for identification and antimicrobial susceptibility testing directly from positive blood cultures. Ann Clin Microbiol Antimicrob. 2023;22:78. 10.1186/s12941-023-00629-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Tsuchida S, Murata S, Miyabe A, Satoh M, Takiwaki M, Matsushita K, et al. An in-house centrifugation and membrane filtration technique for identifying microorganisms from positive blood culture bottles with high identification rates using matrix-assisted laser desorption ionization-Time-of-flight mass spectrometry: A preliminary report. J Infect Chemother. 2020;26:266–71. 10.1016/j.jiac.2019.09.017. [DOI] [PubMed] [Google Scholar]
- 12.Lin J-F, Ge M-C, Liu T-P, Chang S-C, Lu J-J. A simple method for rapid microbial identification from positive monomicrobial blood culture bottles through matrix-assisted laser desorption ionization time-of-flight mass spectrometry. J Microbiol Immunol Infect. 2018;51:659–65. 10.1016/j.jmii.2017.03.005. [DOI] [PubMed] [Google Scholar]
- 13.Tsuchida S, Murata S, Miyabe A, Satoh M, Takiwaki M, Matsushita K, et al. An improved in-house lysis-filtration protocol for bacterial identification from positive blood culture bottles with high identification rates by MALDI-TOF MS. J Microbiol Methods. 2018;148:40–5. 10.1016/j.mimet.2018.03.014. [DOI] [PubMed] [Google Scholar]
- 14.Florio W, Morici P, Ghelardi E, Barnini S, Lupetti A. Recent advances in the Microbiological diagnosis of bloodstream infections. Crit Rev Microbiol. 2018;44:351–70. 10.1080/1040841X.2017.1407745. [DOI] [PubMed] [Google Scholar]
- 15.Morgenthaler NG, Kostrzewa M. Rapid identification of pathogens in positive blood culture of patients with sepsis: review and meta-analysis of the performance of the sepsityper kit. Int J Microbiol. 2015;827416. 10.1155/2015/827416. [DOI] [PMC free article] [PubMed]
- 16.Azrad M, Keness Y, Nitzan O, Pastukh N, Tkhawkho L, Freidus V, et al. Cheap and rapid in-house method for direct identification of positive blood cultures by MALDI-TOF MS technology. BMC Infect Dis. 2019;19:72. 10.1186/s12879-019-3709-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chen Jonathan HK, Ho P-L, Kwan Grace SW, She Kevin KK, Siu Gilman KH, Cheng Vincent CC, et al. Direct bacterial identification in positive blood cultures by use of two commercial matrix-assisted laser desorption ionization–time of flight mass spectrometry systems. J Clin Microbiol. 2020;51:1733–9. 10.1128/JCM.03259-12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Zengin Canalp H, Bayraktar B. Direct rapid identification from positive blood cultures by MALDI-TOF MS: specific focus on turnaround times. Microbiol Spectr. 2021;9:e01103–21. 10.1128/spectrum.01103-21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tanner H, Evans JT, Gossain S, Hussain A. Evaluation of three sample Preparation methods for the direct identification of bacteria in positive blood cultures by MALDI-TOF. BMC Res Notes. 2017;10:48. 10.1186/s13104-016-2366-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Dhiman N, Hall L, Wohlfiel SL, Buckwalter SP, Wengenack NL. Performance and costanalysis of matrix-assisted laser desorption ionization-time of flight mass spectrometryfor routine identification of yeast. J Clin Microbiol. 2011;49:1614–6. 10.1128/JCM.02381-10. [DOI] [PMC free article] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
No datasets were generated or analysed during the current study.