Rapid and accurate sepsis diagnostics via a novel probe-based multiplex real-time PCR system

TL;DR

A new rapid PCR system can identify sepsis-causing pathogens and antibiotic resistance genes in about an hour, improving diagnosis and treatment.

Contribution

A novel probe-based multiplex real-time PCR system for rapid sepsis diagnostics without nucleic acid extraction.

Findings

The SEPSI ID panel showed 96.88% sensitivity and 100% specificity in identifying pathogens.

The system detected 29 microorganisms and 23 resistance genes in about one hour.

It identified additional targets not found by conventional methods.

Abstract

Sepsis is a critical clinical emergency that requires prompt diagnosis and intervention. Its prevalence has increased due to the aging population and increased antibiotic resistance. Early identification and the use of innovative technologies are crucial for improving patient outcomes. Modern methodologies are needed to minimize the turnaround time for diagnosis and improve outcomes. Rapid diagnostic tests and multiplex PCR are effective but have limitations in identifying a range of pathogens and target genes. Our study evaluated two novel probe-based multiplex real-time PCR systems: the SEPSI ID and SEPSI DR panels. These systems can quickly identify bacterial and fungal pathogens, alongside antibiotic resistance genes. The assays cover 29 microorganisms (gram-negative bacteria, gram-positive bacteria, yeast, and mold species), alongside 23 resistance genes and four virulence factors. A streamlined workflow uses 2 µL of broth from positive blood cultures (BCs) without nucleic acid extraction and provides results in approximately 1 h. We present the results from an evaluation of 228 BCs and 22 isolates previously characterized by whole-genome sequencing. In comparison to the reference methods, the SEPSI ID panel demonstrated a sensitivity of 96.88%, a specificity of 100%, and a PPV of 100%, whereas the SEPSI DR panel showed a sensitivity of 97.8%, a PPV of 89.7%, and a specificity of 96.7%. Both panels also identified additional pathogens and resistance-related targets not detected by conventional methods. This assay shows promise for rapidly and accurately diagnosing sepsis. Future studies should validate its performance in various clinical settings to enhance sepsis management and improve patient outcomes.

We present a new diagnostic method that enables the quick and precise identification of pathogens and resistance genes from positive blood cultures, eliminating the need for nucleic acid extraction. This technique can also be used on fresh pathogen cultures. It has the potential to greatly improve treatment protocols, leading to better patient outcomes, more responsible antibiotic use, and more efficient management of healthcare resources.

Figures

Rapid detection of sepsis pathogens using the SEPSI ID/DR panel.

Tables

TABLE 1. SEPSI DR and SEPSI ID panel compositiona,b,c,d,e,f,g

Real-time PCR mixes included in the panels	Fluorophores
	FAM	HEX	ROX	Cy5	Cy5.5
SEPSI DR
Mix 1	bla OXA48	bla VIM	bla KPC	bla NDM	IC
Mix 2	bla TEM	bla CTX-M	bla SHV	bla IMP	IC
Mix 3	bla OXA23	mcr	bla GES	bla CMY	IC
Mix 4	bla FIM	AzolesRes	vanA	vanB	IC
Mix 5	mecC	OrfX	mecA	Panfungal	IC
Mix 6	bla ampC	bla DHA-1	mgrB	ompK36	IC
SEPSI ID
Mix 1	P. aeruginosa	K. pneumoniae	P. vulgaris/mirabilis	E. coli	IC
Mix 2	H. influenzae	K. oxytoca	E. cloacae	K. aerogenes	IC
Mix 3	L. pneumophila	S. marcescens	B. fragilis	A. baumannii	IC
Mix 4	S. malthophilia	E. faecalis	E. faecium	L. monocytogenes	IC
Mix 5	S. pyogenes	S. pneumoniae	S. agalactiae	N. meningitidis	IC
Mix 6	Staphylococcus spp.	A. fumigatus	S. aureus	C. freundii	IC
Mix 7	C. neoformans	PVL	C. albicans	TSST	IC
Mix 8	Candida spp.	EXT a/b	C. auris	K. pneumoniae HMV	IC

TABLE 2. Identification of additional pathogens and resistance genes through the SEPSI ID/DR panels versus standard reference methods

Identification achieved through reference methods		Identification using the SEPSI ID/DR panel	Reference
Microbial identification
E. faecalis		E. faecalis, Staphylococcus sppa.	See result no. 56 in File S1
E. faecalis, E. hormaechei		E. faecalis, E. cloacae/hormaecheib, S. aureus	See result no.58 in File S1
Mechanism of resistance detected
E. coli ESBLc		blaTEM, blaCTX-M, blaAmpC	For ASTg, see BC no. 18 in File S2
E. faecalis (susceptible to glycopeptides) + K. pneumoniae KPC		blaKPC, blaSHV, blaCTX-M	For ASTg, see BCs no. 53 in File S2
E. coli KPC		blaKPC, blaAmpC	For ASTg, see BC no. 56 in File S2
E. coli KPC		blaKPC, blaAmpC	For ASTg, see BC no. 61 in File S2
E. coli overall susceptible		blaAmpC, blaTEM	For ASTg, see BC no. 76 in File S2
E. coli ESBLc		blaAmpC, blaTEM	For ASTg, see BC no. 74 in File S2
S. aureus MRSAd		mecA	For ASTg, see BC no.8 in File S2
E. coli ESBLc		blaCTX-M, blaAmpC	For ASTg, see BC no. 22 in File S2
K. pneumoniae KPC		blaKPC, blaSHV	For ASTg, see BC no. 94 in File S2
E. coli ESBLc		blaAmpC, blaTEM	For ASTg, see BC no. 63 in File S2
K. pneumoniae KPC		blaKPC, blaTEM, blaSHV, blaCMY	For ASTg, see BC no. 19 in File S2
K. pneumoniae KPC		blaKPC, blaSHV, blaTEM	For ASTg, see BC no. 26 in File S2
K. pneumoniae resistant to carbapeneme		blaSHV, blaDHAf	For ASTg, see BC no. 14 in File S2
K. pneumoniae KPC		blaKPC, blaSHV, blaTEM, blaGES, blaCMY	For ASTg, see BC no. 16 in File S2
K. pneumoniae KPC		blaKPC, blaTEM blaSHV	For ASTg, see BC no. 13 in File S2
K. pneumoniae KPC, NDM, OXA-48		blaOXA48, blaNDM, blaKPC, blaSHV, blaCTX-M, blaTEM, blaCMY	For ASTg, see BC no. 44 in File S2
K. pneumoniae KPC		blaKPC, blaSHV, blaCTX-M, blaTEM	For ASTg, see BC no. 9 in File S2

TABLE 3. Comparison of the SEPSI ID/DR panel results with the reference method outcomes

Result	SEPSI ID blood culture result			SEPSI DR blood culture resistance result
	Positive	Negative	Total	Positive	Negative	Total
Positive	93	0	93	88	3	91
Negative	3	26	29	2	26	28
Total	96	26	122	90	29	119

TABLE 4. Diagnostic accuracy of SEPSI ID and SEPSI DR panelsa

Measure of diagnostic accuracy	% (95% CI)
	SEPSI ID blood culture	SEPSI DR blood culture resistance
Sensitivity	96.88 (95.58–99.36)	97.8 (92.2–99.7)
Specificity	100 (86.77–100)	89.7 (72.6–97.8)
PPV	100 (96.09–100)	96.7 (90.7–99.3)
NPV	89.66 (73.89–96.42)	92.9 (76.5–99.1)
LHR+	∞b	9.45 (3.24–27.6)
LHR-	0.03 (0.009–0.108)	0.02 (0.01–0.10)
Accuracy	97.54 (93.78–99.18)	95.8 (90.5–98.6)
F1 score	98.41	97.2

TABLE 5. Clinical isolate and ATCC characterization: resistance gene detectiona,c

ID isolateb	Genes found through WGS	ID WGS sequence	Sepsi DR results
E. faecium vanA	vanHAX	TRCIO_01_S5_L001_R1_001.fastq.gz TRCIO_01_S5_L001_R2_001.fastq.gz	vanA
E. faecium vanA	vanA, vanX	TRCIO_16_S20_L001_R1_001.fastq.gz TRCIO_16_S20_L001_R2_001.fastq.gz	vanA
E. faecalis	vanB defective	TRCIO_17_S21_L001_R1_001.fastq.gz TRCIO_17_S21_L001_R2_001.fastq.gz	none
E. faecium vanA	vanHAX	TRCIO_18_S22_L001_R1_001.fastq.gz TRCIO_18_S22_L001_R2_001.fastq.gz	vanA
E. faecium vanA	vanHAX	TRCIO_23_S27_L001_R1_001.fastq.gz TRCIO_23_S27_L001_R2_001.fastq.gz	vanA
E. faecium vanA	vanHAX	1-Emo_S20_L001_R1_001.fastq.gz 1-Emo_S20_L001_R2_001.fastq.gz	vanA
E. faecium vanA	vanHAX	2-Emo_S22_L001_R1_001.fastq.gz 2-Emo_S22_L001_R2_001.fastq.gz	vanA
E. faecium vanA	vanHAX	ent-7emo_S19_L001_R1_001.fastq.gz ent-7emo_S19_L001_R2_001.fastq.gz	vanA
E. faecium vanA	vanHAX	Entero-6emo_S7_L001_R1_001.fastq.gz Entero-6emo_S7_L001_R2_001.fastq.gz	vanA
E. faecium vanA	vanHAX	ent-3emo_S13_L001_R1_001.fastq.gz ent-3emo_S13_L001_R2_001.fastq.gz	vanA
E. faecium vanA	vanHAX	Entero-4emo_S5_L001_R1_001.fastq.gz Entero-4emo_S5_L001_R2_001.fastq.gz	vanA
K. pneumoniae NDM	blaNDM-5, blaCTX-M-15, blaSHV-11, blaCMY-2	NDM10_S10_L001_R1_001.fastq.gz NDM10_S10_L001_R2_001.fastq.gz	blaNDM, blaTEM, blaCTX-M, blaSHV, blaCMY
K. pneumoniae NDM	blaNDM-1, blaCTX-M-15, blaOXA-1, blaOXA-9, blaTEM-1D, blaSHV-11	NDM19_S19_L001_R1_001.fastq.gz NDM19_S19_L001_R2_001.fastq.gz	blaNDM, blaTEM, blaCTX-M, blaSHV
K. pneumoniae NDM	blaNDM-1, blaCTX-M-15, blaOXA-1, blaOXA-9, blaTEM-1D, blaSHV-11	NDM40_S1_L001_R1_001.fastq.gz NDM40_S1_L001_R2_001.fastq.gz	blaNDM, blaTEM, blaCTX-M, blaSHV
K. pneumoniae KPC	blaKPC-3, blaSHV-100	2023-KPC-Kpn-57_S1_L001_R1_001.fastq 2023-KPC-Kpn-57_S1_L001_R2_001.fastq	blaKPC, blaSHV
A. baumannii	blaOXA-23, blaOXA-66, blaTEM-1D, blaADC-25	2023-Acibau-1_S17_L001_R1_001.fastq.gz 2023-Acibau-1_S17_L001_R2_001.fastq.gz	blaTEM, blaOXA-23
A. baumannii	blaOXA-23, blaOXA-66, blaADC-25	2024-Acibau-3_S3_L001_R1_001.fastq.gz 2024-Acibau-3_S3_L001_R2_001.fastq.gz	bla OXA-23
P. aeruginosa	blaIMP-13, blaOXA-50, blaPAO	S20_L001_R1_001.fastq.gz S20_L001_R2_001.fastq.gz	bla IMP
P. aeruginosa	blaVIM-2, blaOXA-486, blaPAO	Pse-pro_S14_L001_R1_001.fastq.gz Pse-pro_S14_L001_R2_001.fastq.gz	bla VIM
E. faecalis VSE	na	na	none
E. faecium VSE	na	na	none
S. aureus MRSA	mecA	na	mecA, orfX
ATCC K.pneumoniae 700603	na	na	bla SHV
ATCC K. pneumoniae BAA-2814	na	na	blaTEM, blaSHV, blaKPC
ATCC E. coli 25922	na	na	bla AmpC
ATCC E. faecalis 29212	na	na	none
ATCC S. aureus 29213	na	na	orfX
ATCC P. aeruginosa 27853	na	na	none

Full text

INTRODUCTION

Sepsis and bloodstream infections (BSIs) are associated with significant morbidity and mortality rates (1). The early implementation of effective antimicrobial therapy, guided by rapid antimicrobial test results, has been shown to improve patient outcomes (1–3). Notably, in recent years, there has been a progressive increase in sepsis cases due to several factors, including aging, an increase in patients with complex clinical conditions and comorbidities, and the growing problem of antimicrobial resistance (AMR) (4). The COVID-19 pandemic has further exacerbated this situation (5). The impact of AMR on sepsis cannot be overstated (6–8). AMR, often termed the “silent pandemic” (9), frequently complicates the establishment of effective treatment, leading to increased morbidity and mortality, extended hospital stays, potential complications, and the occurrence of epidemic clusters. Additionally, AMR imposes substantial economic burdens due to the resulting need for more expensive drugs and procedures, prolonged hospitalization, and the potential to cause disability (10, 11). Addressing sepsis requires collaboration among clinicians, microbiologists, pharmacologists, and health directors to formulate an integrated strategy (12–14). Sepsis, as a time-dependent syndrome, represents a critical clinical and research emergency (15, 16). Consequently, laboratories must employ all available innovative technologies to enable rapid diagnosis (17, 18). In the diagnostic pathway for BSIs/sepsis, the appropriate and timely collection of blood cultures (BCs) is strongly recommended, as BC-based diagnostic methods remain the gold standard (19). Effective treatment of sepsis requires early recognition of clinical symptoms and access to a microbiology laboratory that uses advanced technologies for rapid testing. These rapid pathways provide timely diagnostic information, allowing clinicians to start targeted therapy quickly (18). Multiple studies have shown that technological advancements have focused primarily on rapid diagnostic tests (RDTs). However, syndromic panels and other multiplex PCR methods often fail to identify many pathogens and target resistance genes (20, 21). BC-free methods still lack sufficient sensitivity to be recommended as a replacement for BCs for detecting BSI pathogens (22). The ongoing implementation of next-generation sequencing (NGS) techniques, which are performed directly with blood or can be performed with BCs a few hours after incubation in a continuous monitoring incubation system, is promising (23). However, these techniques are currently the prerogative of a few large laboratories with access to advanced technologies that are generally expensive and require dedicated staff and great expertise, including bioinformatics (24). Therefore, at present, methods based on real-time PCR technologies still appear to be the simplest and most economically sustainable solution for most microbiological diagnostic laboratories. Hence, the development, validation, and evaluation of new RDTs are desirable.

The objective of our study was to evaluate a novel real-time PCR probe-based system for the rapid identification of bacterial and fungal pathogens, as well as the detection of target genes associated with resistance to major classes of antibiotics in patients with BSIs/sepsis.

RESULTS

An overview of the SEPSI ID and SEPSI DR panel compositions, including the specific targets and fluorophores used in each multiplex reaction, is provided in Table 1 to support the results that follow.

The SEPSI ID panel successfully identified a wide range of clinically significant pathogens commonly associated with sepsis. The most frequently detected bacteria were Escherichia coli, Klebsiella pneumoniae, Pseudomonas aeruginosa, Staphylococcus aureus, Enterococcus faecium, Acinetobacter baumannii, and Proteus mirabilis. The panel also detected the less common pathogens Stenotrophomonas maltophilia, Enterococcus faecalis, Staphylococcus spp. (the species are listed in File S1), and Listeria monocytogenes. These results were consistent with those of the reference methods (RFM), demonstrating the panel’s reliability in identifying these pathogens.

Specifically, the SEPSI ID panel was used to test 126 blood culture (BC) samples, including 100 positive and 26 negative cases. Four samples were excluded from the statistical analysis because the panel did not include their respective targets (namely, Corynebacterium spp., Streptococcus gallolyticus, and Enterobacter roggenkampii) (see File S1). Additionally, three samples were classified as false negatives: two Enterobacter hormaechei were not detected, and one Enterococcus faecium was missed in a mixed BC that tested positive for both E. coli and E. faecium by RFM.

Notably, the SEPSI ID panel identified two additional pathogens that were not detected by the RFM (Table 2).

None of the 26 negative BCs produced a false positive (FP). Therefore, the system exhibited a sensitivity of 96.88% (95% CI: 95.58%–99.36%). The specificity and the PPV were 100% (95% CI: 86.77–100 and 96.09–100, respectively), indicating that all positive results were true positives. The negative predictive value (NPV) reached 89.66% (95% CI: 73.89–96.42), reflecting strong reliability in ruling out false negatives (FNs) [Tables 3 and 4](#T3 T4)). The positive likelihood ratio (LHR+) was infinite due to the absence of false positives (i.e., 100% specificity), which makes the denominator of the formula equal to zero. This indicates an extremely high ability of the test to confirm the presence of the condition when the result is positive. The negative likelihood ratio (LHR−) was 0.03 (95% CI: 0.009–0.108), indicating that a negative test result is highly unlikely in a patient with this condition. The F1 score was 98.41%, reflecting high performance.

The SEPSI DR panel was used to test a total of 128 BC samples, including 102 positive and 26 negative samples. The SEPSI DR panel identified a broader range of resistance mechanisms compared with the reference method, including several resistance genes such as blaGES and blaDHA that were not detected by conventional testing. Nine positive samples were excluded from the statistical evaluation. These samples were positive for yeasts whose target was not included among those detected by the “panfungal target.” Among the remaining 93 positive BCs tested using the SEPSI DR panel, 88 produced results that were consistent with those of the RFM, whereas five presented discrepancies. Three of the five discordant results were classified as FPs, and two were classified as FNs. Among the FPs, two samples tested positive for E. coli due to the presence of the blaAmpC and blaTEM genes. Still, they were phenotypically susceptible according to RFM, likely due to a lack of gene expression. This discrepancy may be explained by promoter mutations in blaAmpC, which can suppress gene expression, as reported in the literature (26–28). The third FP involved a P. mirabilis strain that carried the blaCMY gene and appeared to be phenotypically unexpressed. FN was attributed to a culture that was positive for carbapenem-resistant P. aeruginosa but yielded only the blaSHV gene. This is likely because carbapenem resistance in P. aeruginosa is often mediated by mechanisms such as porin loss/modification or efflux pump overexpression, pathways that are not targeted by the SEPSI DR panel nor by most commercial molecular assays. The last FN was a mixed infection with E. coli and vancomycin-resistant E. faecium. The SEPSI DR panel failed to detect the vanA/B genes in this case. All the results are detailed in File S2. As expected, the 26 negative BCs yielded negative results.

The SEPSI DR panel had a sensitivity of 97.8% (95% CI: 92.2–99.7) and a specificity of 89.7% (95% CI: 72.6–97.8). The PPV was 96.7% (95% CI: 90.7–99.3), and the NPV was 92.9% (95% CI: 76.5–99.1) (Table 4). The LHR+ was 9.45, indicating that a positive result from the SEPSI DR panel provides strong evidence for the presence of resistance genes. The LHR− was 0.02 (95% CI: 0.01–0.10), indicating that a negative result from the SEPSI DR panel is highly reliable for ruling out the presence of resistance genes. This very low value suggests that the test is useful for confidently excluding antimicrobial resistance when no target is detected. The F1 score was 97.2%, reflecting a strong balance between PPV and sensitivity and confirming the panel’s overall diagnostic accuracy. Table 4 provides a comprehensive summary of the results obtained from both panels.

As shown in Table 1, the SEPSI DR panel revealed a wider range of resistance mechanisms, including genes such as blaGES and blaDHA, that were not identified by the RFM (see also File S2). Importantly, the SEPSI DR panel primer design for TEM and SHV allows for the detection of both extended-spectrum β-lactamases (ESBLs) and narrow-spectrum β-lactamases, including blaTEM-1/2 and blaSHV-1. These enzymes are commonly found in Enterobacterales and confer resistance to penicillins and first-generation cephalosporins (28). This explains the resistance to ampicillin observed in two samples: one positive for K. pneumoniae harboring blaSHV and the other positive for P. mirabilis harboring blaTEM (see File S2). The AST results for reference strains are provided in File S2.

Finally, Table 5 shows the SEPSI ID and SEPSI DR results for the ATCC reference strains and the isolates from our collection that were characterized using whole-genome sequencing (WGS) (all raw reads generated were submitted to the Sequence Read Archive (SRA) under the BioProjects ID PRJNA1243017 and PRJNA1259250). Although nearly all the expected targets were detected, WGS revealed additional resistance genes that were not covered by the SEPSI DR panel. Notably, the SEPSI DR assay successfully identified the blaAmpC gene in E. coli ATCC 25922, which exhibited reduced susceptibility to cefoxitin, a phenotype also reported by Tracz et al. (27). Importantly, the results from both the SEPSI ID and SEPSI DR panels can be obtained in approximately 1 h, including approximately 45 min for the amplification step, with additional time required for data export and software-based analysis, resulting in a final turnaround time of approximately 1 h (see Fig. 1).

Rapid detection of sepsis pathogens using the SEPSI ID/DR panel.

DISCUSSION

AMR, often referred to as the “silent pandemic,” is a growing concern, making the effective treatment of sepsis more challenging (9). AMR leads to increased morbidity, mortality, and healthcare costs (29). Rapid diagnostic technologies, such as real-time PCR, are crucial for identifying pathogens quickly and accurately, allowing timely and targeted therapy (30). RDTs lead to better patient outcomes, optimize the use of medications, reduce selective pressure on microbial isolates, minimize the emergence of AMR, and shorten hospital stays (18, 31, 32).

This study aimed to evaluate the effectiveness of a new system based on real-time PCR probes for the rapid identification of bacterial and fungal pathogens, as well as the detection of antibiotic resistance genes, in BCs from patients with BSIs. Both the SEPSI ID and SEPSI DR panels demonstrated excellent diagnostic performance. Specifically, the SEPSI ID panel showed high sensitivity, specificity, and PPV. The NPV was 89.66%, and the F1 score reached 98.41%. This confirms the panel’s high accuracy in identifying both true positives and true negatives. Additionally, the low LHR value (0.03) indicates that a negative test result substantially reduces the likelihood of infection that can be detected by the SEPSI ID panel. This supports the test’s strong exclusion capability and indicates that a negative result is highly reliable for ruling out infection. Nevertheless, the SEPSI ID panel did not identify some pathogens, such as Corynebacterium spp., S. gallolyticus, and E. roggenkampii. Importantly, these pathogens, although clinically significant, are relatively less common. Corynebacterium spp. are increasingly recognized as opportunistic pathogens, particularly in immunocompromised patients. S. gallolyticus is associated with serious infections and an increased risk of colorectal cancer (33, 34). E. roggenkampii is a multidrug-resistant pathogen that poses a significant threat in healthcare settings (35). With a sensitivity of 97.8%, specificity of 89.7%, and PPV of 96.7%, the SEPSI DR panel demonstrated relevant diagnostic capability. Furthermore, an F1 score of 97.2% indicates a strong balance between PPV and sensitivity. Additionally, the high LHR+ value of 9.45 means that a positive result from the SEPSI DR panel strongly suggests the presence of resistance genes. This value approaches the threshold typically considered indicative of high diagnostic utility, thereby reinforcing the panel’s effectiveness in confirming antimicrobial resistance.

Importantly, the SEPSI DR panel identified several resistance genes not detected by conventional methods, including blaGES and blaDHA, underscoring its broader detection capability.

Expanding diagnostic coverage improves the overall treatment approach by enabling earlier and more precise therapeutic decisions. Timely and targeted antimicrobial therapy has been shown to accelerate infection resolution, leading to shorter hospital stays and improved patient outcomes (36, 37). Furthermore, encouraging the appropriate use of antimicrobials helps reduce the selective pressure that drives the emergence of AMR (38). In this context, the SEPSI DR panel has the potential to contribute to antimicrobial stewardship by providing detailed resistance profiles that could support more informed antibiotic selection and potentially help preserve the long-term efficacy of antibiotics. It is possible that the panel might also help mitigate the development and spread of AMR by reducing reliance on broad-spectrum agents. Although our study did not directly compare the reporting times of the standard of care and the new RDT, notably, the novel system delivers results in approximately 1 h. In contrast, traditional AST typically requires 24−48 h (39), depending on the method used and the microorganism involved. This substantial reduction in turnaround time has the potential to significantly accelerate clinical decision-making in the management of sepsis. This is crucial in sepsis management, where every hour of delay in appropriate treatment increases mortality risk (36, 37).

Despite these promising results, this study has several limitations. First, the sample size was relatively small, which may limit the generalizability of the findings. Larger studies are needed to confirm these results and provide more robust data. Second, the study was conducted in a single laboratory setting; hence, the results may not reflect the performance of the diagnostic panels in different clinical environments with varying levels of expertise and resources. Third, the SEPSI ID panel did not identify certain pathogens, indicating that the panel coverage is not comprehensive. It would be beneficial to include these pathogens in the SEPSI ID panel to ensure comprehensive identification and effective treatment and management of infections. Additionally, the SEPSI DR panel exhibited lower specificity due to some FP results. This is a limitation of all molecular systems that can potentially detect resistance genes that may not be expressed, which can lead to unnecessary antibiotic treatments. To address this issue, it is important to share knowledge of the limitations of molecular systems with clinicians. Finally, the study did not evaluate the cost-effectiveness of implementing these diagnostic panels in routine clinical practice, which is an important consideration for widespread adoption. Future research should address these limitations by including larger, multicenter studies; expanding the range of detectable pathogens and resistance genes; and assessing the cost-effectiveness of these diagnostic tools in various healthcare settings.

In conclusion, despite some limitations, SEPSI ID/DR panels are highly effective at rapidly and accurately identifying pathogens and their resistance target genes, making them an invaluable asset in modern diagnostics. Their rapid performance accelerates decision-making, playing a pivotal role in improving patient care and outcomes. Adopting this technology can greatly increase clinical effectiveness and ensure timely treatment.

MATERIALS AND METHODS

SEPSI ID and SEPSI DR description

The system being evaluated is currently undergoing CE/IVDR certification and is intended for research use only. It comprises a qualitative, multiplex real-time PCR probe-based assay consisting of two diagnostic kits: SEPSI ID and SEPSI DR. These kits were developed in collaboration with the Clinical Microbiology Laboratory at the University of Rome “Tor Vergata” and Elettrobiochimica Srl. SEPSI ID and SEPSI DR are single-test systems that include a mixture of buffer, Taq DNA polymerase, primers, and probes in a prealiquoted liquid phase. The tests are divided into two strips, with six tubes for the DR test and eight tubes for the ID test, which are necessary for multiplex real-time PCR (Table 1). The strips included all the required liquid reagents and were stored at −20°C until use. Before testing, the reagents were thawed, and 2 µL of positive BC was added (nucleic acid extraction was not necessary). The primers and probes were sourced from Eurofins (Eurofins Genomics Europe; Ebensburg, Germany). In each tube, up to five targets (four targets plus an internal control) can be detected simultaneously. Five probes labeled with different fluorophores were used. The fluorophores used were FAM, ROX, Cy5, HEX, and Cy5-5, which were attached to the 5' end of the probe, whereas a nonfluorescent quencher, BHQ, was located at the 3' end. The system ensures the absence of amplification reaction inhibitors through the use of an internal control that amplifies and detects a human gene, specifically β-actin.

Specifically, the SEPSI ID/DR panel can be used to identify a total of 29 microorganisms, including 15 gram-negative and eight gram-positive bacteria, six species of yeast and mold (including a panfungal target), 23 resistance genes, and four virulence genes. Amplifications were conducted using the Bio-Rad CFX96 system (Bio-Rad, Hercules, California, USA). The assay was also validated on the CFX Opus 96 Real-Time System and the Bio-Rad CFX96 Touch system (Bio-Rad). The Taq polymerase used was Air Dryable 4X Direct DNA qPCR Blood (Meridian Bioscience, Cincinnati, Ohio, USA). Both panels provided the final results within 1 h.

The primer and probe sequences are not disclosed due to intellectual property rights protection by the University of Study of Rome “Tor Vergata” under Patent No. 102020000018400 and Patent No. 10200000018400.

The selection of target organisms and AMR/virulence genes for the panels was based on their clinical relevance in BSIs and their association with AMR. The panel includes ESKAPE pathogens and other key bacteria, which were selected on the basis of global and Italian epidemiological data (39, 40). Resistance gene targets were chosen to cover the most impactful mechanisms encountered in clinical settings. These genes include the following β-lactamase genes: blaKPC, blaNDM, blaVIM, blaIMP, blaOXA-48, blaOXA-23, blaTEM, blaSHV, blaCTX-M, blaGES, blaCMY, blaAmpC, and blaDHA-1. These genes are associated with resistance to carbapenems and extended-spectrum cephalosporins. Additional targets include mcr for colistin resistance; vanA and vanB for vancomycin resistance; and mecA, mecC, and OrfX for methicillin resistance (41). The mgrB gene was included because of its role in colistin resistance, and the ompK36 gene was included because of its contribution to membrane impermeability, particularly in K. pneumoniae. The panel also includes markers for azole resistance and a panfungal target to broaden diagnostic coverage. Virulence genes (Panton-Valentine leukocidin, exfoliative toxins A/B, toxic shock syndrome toxin, rmpA and magA) were selected for their association with increased pathogenicity and adverse clinical outcomes (42, 43). This ensures that the panel supports the rapid provision of actionable diagnostic information for managing sepsis. The reproducibility study for both panels is detailed in File S3. The results of cross-reactivity tests performed to detect targets individually or in combination (polymicrobial) are reported in File S4.

Samples tested

This study included patients admitted to our hospital with suspected sepsis or BSIs. All positive BC samples from adult patients aged 18 years and older were analyzed prospectively. These samples were processed in batches daily and stored at 4°C until the results were obtained.

The performance of the system was evaluated by including a total of 202 positive BCs: 100 were tested with the SEPSI ID kit, and 102 were tested with the SEPSI DR. To rule out the possibility of false-positive results, we tested 26 negative BCs using both panels. We assessed the performance of the new system alongside RFM. To assess the ability of the SEPSI DR panel to detect resistance genes, this study included 22 isolates from our collection and six ATCC strains. Of these 22 isolates, 19 had been previously characterized through WGS analysis (data not shown), whereas the remaining three had not been sequenced.

The ATCC strains included were E. faecalis ATCC 29212, K. pneumoniae ATCC 700603, K. pneumoniae BAA-2814, S. aureus ATCC 28213, E. coli ATCC 25922, and Pseudomonas aeruginosa ATCC 27853. The antimicrobial susceptibility profiles of these strains are well documented.

PCR assay conditions: protocol for processing positive BC broth

Broth from positive BC (2 µL) was added directly to each well of the strip, and a multichannel pipette was used to streamline the process. The strips were then sealed with the appropriate optical caps. Amplification was conducted using a CFX96 system (Bio-Rad) according to the established protocol: one initial step at 95°C for 3 min, followed by 35 cycles of 95°C for 10 s and 60°C for 25 s. After amplification, the results were exported using the “Export LIMS Folder” function, and data analysis was performed using EBC software (for Bio-Rad instruments only). The EBC software is designed to analyze real-time PCR data generated by the SEPSI ID and SEPSI DR panels, which are used either individually or in combination. Following the amplification step, the operator exports the data for processing by the software, which then generates a comprehensive report including both pathogen identification and interpretation of the associated resistance profile. Currently, the protocol has only been validated on the CFX96 system (Bio-Rad).

PCR assay conditions: protocol for processing bacterial isolates

Although the test was optimized for use with a positive BC sample, it can also be applied to fresh colonies of bacterial isolates. This can be accomplished by preparing a bacterial suspension in sterile distilled water corresponding to 0.5 on the McFarland scale. The PCR assay was conducted with 2 µL of the prepared bacterial suspension following the aforementioned BC protocols.

Reference methods

BC-based methods remain the gold standard for the isolation and detection of pathogens involved in BSIs (19). BCs were obtained from patients in whom a physician suspected infection (BSI/sepsis), and for adult patients, two sets of BCs (with each set composed of aerobic and anaerobic bottles) were routinely inoculated at the bedside and immediately delivered to our microbiology laboratory. Upon arrival at the laboratory, the BCs were incubated in a BACTEC FX system (Becton Dickinson; Sparks, MD, USA), and those identified as positive were subcultured and Gram-stained (smear microscopy). Specifically, aliquots (10 µL) of broth from the positive bottles were plated onto chocolate agar, blood agar, Columbia CNA agar, Sabouraud dextrose agar, and MacConkey agar (Thermo Fisher Scientific; Waltham, MA, USA) using the WASP automated system (Copan, Brescia, Italy). The plates were incubated under aerobic and anaerobic conditions at 37°C for 24 h. The growth of the bacterial colonies was monitored, and the bacterial species were identified using the MALDI-TOF assay (MALDI TOF Syrius; Bruker Daltonics, Bremen, Germany). Antimicrobial susceptibility testing (AST) was performed using the Phoenix system (Becton Dickinson). The AST results were interpreted according to EUCAST clinical breakpoints v14.0 (EUCAST available at https://www.eucast.org/clinical_breakpoints). The identification of carbapenemase-producing Enterobacterales (CPE) and P. aeruginosa was performed using immunochromatographic assays (NG-Test CARBA 5 (NG Biotech, Guipry, France). For some cases, particularly critically ill patients, upon detection of positivity, the BC was also processed using BioFire FilmArray BCID2 (bioMérieux, Las Balmas, France) (File S2), which allows the identification of pathogens as well as major resistance-related targets (44).

Statistical evaluation

Statistical analyses were conducted to evaluate the performance of the diagnostic test. Specifically, we computed sensitivity, specificity, PPV, NPV, LHR+, and LHR− to assess the probability of test results based on the presence or absence of the condition. Overall, the PPV was calculated as the proportion of correctly identified cases among the total number tested. The F1 score, which indicates a balance between PPV and sensitivity, was calculated using standard statistical methods (45).