Development of a Multiplexed qPCR Kit for the Detection of Bloodstream Infection

ABSTRACT

Background

Bloodstream infection (BSI) is associated with high morbidity and mortality worldwide. Currently, BSI diagnosis relies on a time‐consuming blood culture method, which usually takes 2 or more days to identify the causative pathogens. Cell‐free DNA (cfDNA) refers to those small nucleic acid fragments residing in plasma and other body fluids, which have been used to detect cancer, organ transplantation injury, and pathogenic infections. A new multiplexed fluorescent quantitative PCR kit aiming at plasma microbial cfDNA was developed in this study. The kit contains multiple panels, and each panel covers multiple pathogens, including E. coli , K. pneumoniae , A. baumannii , H. influenzae , P. aeruginosa , E. faecalis , E. faecium, S. aureus , and S. epidermidis.

Methods

PCR primers and probes were designed based on effective bacterial sequence segments, which were obtained from the analysis of next‐generation sequencing results of plasma samples of patients with bloodstream infections. Bioinformatics analyses and experimental evidence were used to test the conservativeness and specificity of the primers and probes. The lower detection limit of the kit was determined under optimal reaction conditions. Clinical samples were used to test the accuracy of the kit's detection.

Results

The lower detection limit of the kit has reached ≤ 12 copies per reaction. Clinical samples testing results showed a 90.48% consistency between the kit and blood culture.

Conclusion

The kit provides a rapid, accurate, and reliable method for diagnosing bloodstream infections. This can quickly provide an etiological basis for clinical diagnosis and further treatment, potentially improving patient outcomes.

The paper presents a novel multiplexed PCR kit designed for the rapid and accurate detection of pathogens causing bloodstream infection (BSI) through the microbial cell‐free DNA (cfDNA) of plasma. This innovative diagnostic tool, capable of identifying pathogens within approximately 1.5 h, shows a high degree of accuracy (90.48% consistency with traditional blood culture) and a low detection limit, significantly reducing the time required for BSI diagnosis and potentially improving patient outcomes.

1. Introduction

Bloodstream infection (BSI) is a prevalent outcome of invasive infections, which can rapidly deteriorate into sepsis and septic shock, and is a leading cause of high morbidity and mortality in the world [1, 2]. Globally, an estimated 48.9 million cases of sepsis were recorded in 2017, with 11 million sepsis‐related patient deaths, accounting for nearly 20% of all global deaths [3], which causes a serious economic and social burden throughout the world.

Bacteria, often Escherichia coli , Klebsiella spp., Pseudomonas aeruginosa , enterococci, streptococci, S. aureus, and coagulase‐negative staphylococci are the predominant causative pathogens of BSI [4, 5, 6, 7, 8]. Blood culture has been used as the gold standard for the diagnosis of BSI for long, despite being time‐consuming, low positive rate, and prone to contamination [9]. In fact, if not diagnosed and treated promptly, the mortality rate of septic patients would increase by 6% to 7% per hour [10]. Hence, a “blood culture test of two days or more” becomes unbearable for patients suffering from severe sepsis.

Detection systems that do not rely on blood culture have been reported. The MagicPlex Sepsis Test (Seegene, South Korea) and SepsiTest (Molzym, Germany) apply to whole blood samples directly. Their sensitivities are only 37% and 11% compared to blood cultures [11]. Reasons for the low detection sensitivity may be the following: First, suitable DNA extraction methods for genomes of gram‐positive bacteria, gram‐negative bacteria, and fungi are different in many aspects, making it challenging to determine one uniform method for an unknown sample; Second, abundant host genomic DNA extracted from whole blood introduces significant interference for pathogen detection. We have also developed a multiplexed PCR‐based BSI assay kit in conjunction with whole blood samples, which employed capillary electrophoresis (CE) technology as final signal detection [12]. The LOD of the kit was at 100 copies/μL, and each sample needs to be extracted using a bacterial genome extraction kit and a fungal genome extraction kit, respectively. Overall, the assay kit demonstrated relatively low LOD, with a complex and time‐consuming extraction process.

T2Bacteria (T2 Biosystems, America) claims a direct detection of pathogenic bacteria in whole blood by PCR combined with nuclear magnetic resonance (NMR) technology, which takes 3–5 h and achieves high sensitivity [13]. However, this method is not suitable for large‐scale applications due to its reliance on high‐priced equipment. Metagenomics next‐generation sequencing (mNGS) has been applied in the diagnosis of bloodstream infection because it is capable of acquiring abundant information on pathogenic bacteria [14, 15, 16]. Nevertheless, NGS is relatively expensive and requires support from a professional team. More disappointingly, it takes at least 1 or 2 days to get results, which possess very limited time‐wise advantage over blood culture.

Cell‐free DNA (cfDNA) present in plasma, serum, cerebrospinal fluid, urine, saliva, and other body fluids has become a diagnostic marker for cardiovascular diseases, tumors, prenatal diagnosis, and solid‐organ transplant [17, 18, 19, 20, 21]. In the case of bloodstream infection, a trend of parallel increasing plasma microbial cfDNA concentration and sepsis severity has been observed [22]. Moreover, metagenomic next‐generation sequencing of plasma microbial cfDNA has been successfully applied to identify potential pathogens in blood [23, 24]. Similarly, for the detection of a clinical sample, the long turnaround time for this method remains a persistent issue. Therefore, we designed and developed a kit for detecting microbial cfDNA in plasma by multiplexed quantitative PCR. This approach offers simple DNA extraction, shorter turnaround time, no expensive equipment, and effectively addresses issues presented in all aforementioned methods.

PCR primers and probes were designed based on effective bacterial sequence segments, which were obtained from the analysis of next‐generation sequencing results of plasma samples of patients with bloodstream infections. Combined bioinformatics analysis and experimental evidence helped demonstrate a good conservativeness and specificity of the chosen sequence segments.

In this study, primers and probes are based on the effective bacterial sequence segments of pathogens identified by bioinformatics analysis and next‐generation sequencing of cfDNA extracted from clinical plasma samples. To this point, primer and probe pairs for nine pathogens have been qualified. These primers and probes were arranged into three panels depending on their interaction potentials. Each panel demonstrated good conservativeness and specificity, with a minimum detection limit of 4–12 copies per reaction experimentally. The consistency between the test results by this kit and those of blood culture reached 90.48%, indicating its diagnostic power in BSI diagnosis clinically.

2. Materials and Methods

2.1. Technical Process of Sample and Sample Data

Blood samples from patients were handled through steps as illustrated briefly by a flowchart shown in Figure 1A. Commercial NGS is the last step in this process that provides DNA sequence raw data reads. Raw reads data processing steps are shown by the flowchart in Figure 1B that yield our designed primers and probes.

FIGURE 1.

Flow chart of technical procedures: (A) Blood sample collecting and handling from patient to DNA sequencing; (B) Data processing from sequencing raw reads to designed primers and probes.

2.2. Collection of Human Plasma Samples

Totally 258 patient plasma samples were collected in collaboration with the Affiliated People's Hospital of Ningbo University in Zhejiang Province, China. Ethics approval was granted by the Ethics Committee of the Hospital (Approval number: 2020 Research Ethics Review NO 68). Patient inclusion criteria are as follows: Age ≥ 16 years old; qualifying the diagnostic criteria of sepsis 3.0; and suspected bloodstream infection or concurrent severe pulmonary infection.

2.3. Extraction of Plasma cfDNA

cfDNA was extracted by using a commercial extraction kit (Vazyme, Nanjing, China), which adopted a popular magnetic bead‐based extraction method. Following the manufacturer's instructions, cfDNA can be isolated from 2 mL of plasma to reach an acceptable yield.

2.4. Next‐Generation Sequencing (NGS) of Sample DNA

Briefly, the process of sequencing includes DNA detection, library construction, and up‐sequencing. NGS was performed on an Illumina platform with PE150 strategy in Novogene Bioinformatics Technology Co. Ltd. (Beijing, China). Due to the low content of DNA in individual plasma samples, which are insufficient to meet the required amount for sequencing, five DNA samples were pooled together and submitted for sequencing. 2GB of data were collected from each sequencing sample from the Illumina platform by Novogene.

2.5. Raw DNA Sequencing Data Processing for cfDNA Reads

Raw DNA sequencing data collected by the Illumina platform first underwent quality control using the FastQC program v0.11.9 to remove low‐quality base calls from each read, followed by Cutadapt v2.6 to remove adaptor sequences that were added during the sequencing sample preparation. Then, such cleaned reads were searched against the human reference genome sequence (GRCh38) using the Burrows–Wheeler aligner (bwa v0.7.17) to identify their probable origin [25]. The mapped reads are identified as homo sapiens and were discarded; remaining reads with poor matching were kept for subsequent processing.

2.6. From Clean Reads to Effective Bacteria Sequences

Sequence reads not identified as human origin are subject to Kraken analysis [26]. For that process, the reference sequences (RefSeq) of bacteria, archaea, and viruses were downloaded from the Kraken website (https://ccb.jhu.edu/software/kraken2), and a standard Kraken database was constructed. The Kraken database stores the lowest common ancestor (LCA) values of taxonomic ID numbers for oligo k‐mers. When analyzing sequences, if the LCA value of a k‐mer has been previously set, it calculates the LCA of the stored value and the taxonomy information of the current sequence and stores this LCA as the value of that k‐mer. The taxonomic information is sourced from the NCBI Taxonomy database. Identification of a sequence read up to species level was accomplished using Kraken2 and the standard Kraken database. An effective sequence of a target pathogen was compiled from the consensus of its identified reads.

2.7. Primers and Probes Design for Bacteria DNA Targets

Primers and probes design were carried out following the method previously described but with one critical difference [12]. The design described in that publication was based on pathogen genome sequences deposited in public data repositories, while in this work, they were based on effective sequences obtained from NGS reads analysis. Identified Segments of effective sequences of target bacterium were randomly cut into numerous forward and reverse sequence groups, each 20–30 bp in length. They serve as PCR primer candidates. Candidates which do not meet basic requirement of primer design were disqualified. For a group of primer candidates sharing identical 3′ ends sequences, the shortest one was selected. Remaining members were reserved for specificity and conservation screening. Subsequently, candidates that met the criteria for conservativeness and specificity were paired on corresponding effective sequence to obtain primer pairs. Finally, a matching probe was determined following the previously described method against the amplicon sequence. Please refer to Figure 1B for a better understanding of this process.

2.8. Panel Construction and Interaction Checkup

Each DNA primer and probe set for a bacterial target designed through previous steps was checked in silico using AutoDimer software v1.0 [27] to reveal any potential for complementary interactions between them. Primer and probe combinations for their targets were grouped into separate panels to keep the overall predicted interactions a least.

2.9. Human DNA Reference and Internal Control Primers and Probes Design

Human DNA GAPDH was designated as an internal reference, and a set of primers and probes was also designed accordingly. For the design of internal reaction control (IC), 100,000 random sequences of length from 100 to 150 bp were generated first. For a sequence qualified to be a designing template, each of the four bases has to be roughly evenly represented within its whole length. In addition, there should be no more than five consecutive identical nucleotides repeating. With a qualified sequence template, the designing process of primers and the probe set for IC was the same.

2.10. Primers and Probes Preparation

Determined primers and probe sets are presented in Table 1 together with related information. All oligos were synthesized by Sangon Biotech (Shanghai, China), and were purified using polyacrylamide gel electrophoresis (PAGE). The oligos were diluted to 100 μM in 1 × TE buffer before use.

TABLE 1.

Primer and probe information for each target.

Target	Primer and probe sequence (5′‐3′)		Amplicon size (bp)
E. coli	Forward	TGTAGTTTCCCTGTACTGATAGG	87
	Reverse	ACGCATAGCTTCACATAATTCT
	Probe	FAM‐TATGACTTACCCTTATCGCAC‐MGB
E. faecium	Forward	TCTGTCTCCGTAAATACTGCAT	86
	Reverse	CGGGCTAATCAGTCAGGAATC
	Probe	Texas‐Red‐TCCACGTTCTGCACATATT‐MGB
S. epidermidis	Forward	GCTTCTACTATCAATCCTTATGGT	114
	Reverse	TTCTTTAATCGCACGTTCTGA
	Probe	CY5‐TCATTTGCGTCTTTCTCTTC‐MGB
A. baumannii	Forward	CTGCTGCGAGTAAAGAATCAA	131
	Reverse	AGTGTTCGGTTCAGTATTAGGT
	Probe	FAM‐ACGGTCCAATACACCAC‐MGB
H. influenzae	Forward	GGATACGACGCAATGCATTA	110
	Reverse	ACCACGTTTAGTAGATATTGAACA
	Probe	Texas‐Red‐ACCAAAACCACGCTCT‐MGB
K. pneumoniae	Forward	ACTCTCCTTCAGTAATCAACAGTA	128
	Reverse	CGGTAATAGGCGATAAGTGAGT
	Probe	CY5‐AATCGCATCCGTGGCTCATATTCA‐BHQ
S. aureus	Forward	TTTGCACACGTTCAGGACTT	109
	Reverse	CGATATCGCCTAAATCAGCAT
	Probe	FAM‐TCATCGCCAACTTGATCTTTA‐MGB
E. faecalis	Forward	AGTAGCGGCGTTAGGATTTC	104
	Reverse	AGTCTTCACCTTCTAGATACGTTAA
	Probe	Texas‐Red‐CATGTATCCTTGTTCACTTTC‐BHQ
P. aeruginosa	Forward	TAATGGGTAGGGCATCGGAAG	110
	Reverse	GGATATTCGTCCTCCATGACC
	Probe	CY5‐CATTCGATCGCTCCCTG‐MGB
GAPDH	Forward	AGATCTTTCAGCAGAGATGACA	100
	Reverse	GGAACAGACTTCCTAGGATCT
	Probe	VIC‐AAGACGAGACACCCAAGAAAACAACT‐BHQ
IC	Forward	CTGGATCTAGTACCGTTACGTG	115
	Reverse	CTCCGGTTATAAGGTAGTTCATTCTC
	Probe	VIC‐CTCCGGGCCCATTTTAGGTAGC‐BHQ

Note: FAM, Texas‐Red, CY5, VIC, fluorescent dye are labeled at the 5' ends of the probes; BHQ, MGB quencher are labeled at 3' ends.

2.11. Standard Bacteria Strains and Their Genomic DNA

All strains were obtained from ATCC and cultured in media under conditions as recommended by ATCC (Table S1). Bacterial genomes were extracted using the Bacterial DNA Kit (OMEGA bio‐tek) according to the manufacturer's instructions.

2.12. Plasmid Construction and Copy Number Quantitation

Standard plasmid pUC57 containing sequences of selected DNA target fragments was synthesized by Qingke Biotechnology Co, Nanjing, China. Absolute plasmid copy numbers of a dilution series were determined by Droplet Digital PCR (Sniper DQ24). Aliquots were stored at −80°C.

2.13. Checking Up Conservativeness and Specificity Experimentally

In order to check the conservativeness of selected DNA sequences, which is the PCR template, targeted bacterium genomes were used as positive controls. Genomes of other pathogens that may be present in BSI were used as a background in specificity checking. The reaction system of Panel 1 was prepared based on NuHi Ez Advanced U+ SNP Mix‐AP (Low DNA) kit (NuHigh Biotechnologies Co. Ltd., Suzhou, China): 0.8 μL DNA polymerase; 9.2 μL buffer; 0.6 μL primers and probes mix (final concentrations of the primers and probes were 200 nM and 100 nM); 5 μL template and 4.4 μL ddH₂O. The reaction system of Panel 2 or 3 was prepared based on NovoStart Probe qPCR SuperMix (UDG) kit (Novoprotein Scientific, Shanghai, China): 10 μL 2 × NovoStart Probe qPCR SuperMix (UDG); 0.6 μL primers and probes mix (final concentrations of the primers and probes were 200 nM and 100 nM); 5 μL template and 4.4 μL ddH₂O. All qPCR reactions were performed on ABI 7500 Real Time PCR System (Applied Biosystems, Foster, CA, USA), with the following running parameters: UNG reaction stage: 50°C, 2 min; predenaturing stage: 95°C, 5 min; PCR reaction stage: 40 cycles of 95°C for 5 s and 60°C for 30 s.

2.14. Determination of LOD

Determination of LOD was performed by using serial dilution of all the positive control plasmid, and repeated tests on each concentration 20 times. The PCR reaction system and running parameters were the same as described above. The LOD value was determined as the lowest concentration at which 95% of the 20 repeated test runs showed positive (Ct ≤ 38).

2.15. Testing of Clinical Patient Samples

Samples were collected by the Affiliated People's Hospital of Ningbo University. Plasma samples from patients undergoing blood culture tests were collected and stored at −80°C. Plasma from patients with positive blood cultures within the target detection range of the kit was used for testing. In total, 21 plasma samples were employed to verify the kit's performance versus blood culture method. Positive blood cultures included four cases of E. coli , three cases of K. pneumoniae , one case of S. aureus , three cases of S. epidermidis, two cases of P. aeruginosa , three cases of E. faecalis , two cases of E. faecium , two cases of A. baumannii , and one case of H. influenzae . Plasma cfDNA extraction was conducted using the aforementioned cfDNA extraction method, and then tested using the kit developed in this work.

3. Results

3.1. Distribution Pathogen DNA Fragments of BSI Revealed by NGS

A large amount of sequencing raw reads data was acquired from 258 patient plasma samples by NGS. Kraken2 software program helped to identify each read and classify them into the most probable taxon (Figure 2). A total of 2,716,512 reads were analyzed, and pathogen DNA sequences of BSI were identified.

FIGURE 2.

Taxonomic classification of NGS sequencing reads.

E. coli and K. pneumoniae were the most frequent in bloodstream infection [28]. In accordance with this fact, E. coli (25,804 reads) and K. pneumoniae (66,106 reads), which belong to the Enterobacteriaceae family, made up the largest portion (Figure 2). H. influenzae (7959 reads), P. aeruginosa (7509 reads), E. faecalis (2563 reads), E. faecium (1088 reads), S. aureus (1376 reads), and S. epidermidis (1216 reads) each occupied a minor portion, respectively (Figure 2). These identified bacterial reads were subsequently utilized as the sequence template basis for the designing of primers and probes, as previously described.

3.2. Design of Primers and Probe Sets

The reads classified into the most common pathogenic microbes of bloodstream infection were used to screen for suitable primers and probes. Finally, nine sets of primers and probes for target strains were obtained. Primer and probe sequence information is shown in Table 1.

3.3. Assignment of Pathogen Targets Into Testing Panels

Primers and probes are DNA oligos in chemical nature, which possess the potential to form dimerized molecules to negatively impact the reaction. Reverse complement sequence components of oligos in one solution system are the most significant cause of this potential pairwise interaction. This is especially harmful when such sequence components are located at the 3'‐end, where the chain elongation could happen. Potential pairwise interactions of oligos of the nine sets of primers and probes are analyzed and arranged into three panels in a way to keep the overall interactions minimum (Figure 3A).

FIGURE 3.

Heat map of pairwise potential interactions of oligos. (A) Interactions between all primers and probes; (B) Interactions between primers and probes with panel arrangement.

In order to express this potential interaction in a quantitative way, AutoDimer software v1.0 was brought in to calculate the interaction levels between each pair of oligo sequences. The returned number representing the level of each potential interaction is converted into color intensities to mark the corresponding cells in a heat map (Figure 3A). In this color‐coded heat map, bright red indicates the strongest interaction while dark blue indicates the weakest, with transitional levels in between. In Figure 3B, three marked boxes display the primer–probe interactions within the three panels. It can be observed that primer–probe pairs with stronger interactions have been avoided.

Be noticed, one primer and probe set for GAPDH was assigned to Panel 1 as the extraction efficiency control, and one primer and probe set for IC was assigned to Panels 2 and 3, respectively, as the amplification quality control.

3.4. Experimental Validation of PCR Primers and Probe Set Design

Designed primers and probes were validated using constructed plasmids as positive controls. Genomes of target bacteria on three panels were successfully detected by the designed PCR system. Designed primers and probe sets recognized their corresponding templates well without being disturbed by other added templates within each panel (Figure 4).

FIGURE 4.

PCR system validation and specificity test for three panels using standard strain sequence DNA. 1, E. coli ; 2, E. faecium ; 3, S. epidermidis ; 4, A. baumannii ; 5, H. influenzae ; 6, K. pneumoniae ; 7, S. aureus ; 8, E. faecalis ; 9, P. aeruginosa ; 10, E. cloacae ; 11, K. aerogenes ; 12, K. oxytoca ; 13, K. kristinae ; 14, S. maltophilia ; 15, S. acidaminiphila ; 16, P. fluorescens ; 17, M. catarrhalis ; 18, S. haemolyticus ; 19, S. hominis; 20, S. marcescens ; 21, B. cepacia ; 22, P. mirabilis ; 23, P. vulgaris ; 24, A. lwoffii ; 25, A. calcoaceticus ; 26, S. pneumoniae ; 27, S. pyogenes ; 28, S. salivarius ; 29, S. dysgalactiae; 30, S. sanguinis ; 31, NTC: No template control.

These results indicated that the designed system of all three panels is capable of dependent detection on their targets with trustful specificity among other closely related bacteria and common BSI pathogens.

3.5. Determination of LOD

The limit of detection (LOD) is the most critical quality of a testing kit. It was defined as the lowest target DNA concentration with which, as template, a 95% positive testing rate can be realized (at least 19 positives over 20 replicates). In order to determine the LOD, a series of positive plasmid DNA samples of concentration gradient were prepared as references. A cycle threshold value (Ct‐value) ≤ 38 is defined as the positive reading standard, while a Ct‐value of greater than 38 is regarded as negative. The results showed that the LOD of E. coli , E. faecium , and S. epidermidis is 12 copies per reaction. The LOD of A. baumannii , H. influenzae , K. pneumoniae , E. faecalis , and P. aeruginosa is eight copies per reaction. The LOD of S. aureus is four copies per reaction (Figure 5).

FIGURE 5.

Sensitivity of multiplexed qPCR detection assays in all three panels analyzed. Dotted lines at Ct = 38 serve as a threshold reference below which the resulting data are considered invalid. Ct, cycle threshold; ND, not detected within 40 cycles.

3.6. Performance of the Assay on Clinical Samples

CfDNA was extracted from plasma samples collected from 21 patients whose blood culture results are positive. Twenty out of these 21 cases (95.24%) were detected as positive by the developed assay (Table 2). In one case, E. faecium and A. baumannii were detected positive while the blood culture was positive for E. faecium only. In another case, the blood culture result showed positive for S. epidermidis while the developed assay yielded a negative result. Overall, the experimental results demonstrated that the developed assay has a relatively high sensitivity and specificity.

TABLE 2.

Clinical sample test results.

Sample ID	Blood culture result	Kit result (Ct value)	Consistency
S‐1	E. coli	E. coli	Consistent
S‐2	K. pneumoniae	K. pneumoniae	Consistent
S‐3	E. coli	E. coli	Consistent
S‐4	E. faecalis	E. faecalis	Consistent
S‐5	K. pneumoniae	K. pneumoniae	Consistent
S‐6	P. aeruginosa	P. aeruginosa	Consistent
S‐7	E. faecium	E. faecium	Consistent
S‐8	S. epidermidis	—	Inconsistent
S‐9	H. influenzae	H. influenzae	Consistent
S‐10	E. coli	E. coli	Consistent
S‐11	P. aeruginosa	P. aeruginosa	Consistent
S‐12	S. epidermidis	S. epidermidis	Consistent
S‐13	E. faecium	E. faecium A. baumannii	Inconsistent
S‐14	A. baumannii	A. baumannii	Consistent
S‐15	K. pneumoniae	K. pneumoniae	Consistent
S‐16	E. faecalis	E. faecalis	Consistent
S‐17	E. faecalis	E. faecalis	Consistent
S‐18	E. coli	E. coli	Consistent
S‐19	S. epidermidis	S. epidermidis	Consistent
S‐20	S. aureus	S. aureus	Consistent
S‐21	A. baumannii	A. baumannii	Consistent

4. Discussion

BSI has garnered global attention due to its high incidence and mortality rates [29]. Currently, the blood culture method is the gold standard for BSI diagnosis. Nevertheless, it has significant limitations, such as a low detection rate and a long turnaround time. Therefore, it is of great benefit to develop a rapid and accurate diagnostic kit for BSI detection. With the discovery of cell‐free DNA (cfDNA) and its widespread applications in oncology and infectious disease fields [30, 31], we envision its great potential in the realm of BSI detection.

Liquid biopsy approaches based on microbial cfDNA sequencing have been applied to pathogen detection [32, 33]. However, it involves high startup costs, requires a dedicated professional team for analysis, and is time‐consuming. To address these issues, we leveraged next‐generation sequencing technology in the analysis of blood microbial cfDNA sequences, with the goal of identifying distinctive and dependable target sequences for the development of qPCR assay. This approach directly provides significant advantages in cost and required turnaround time. Through the analysis of sequences of cfDNA of plasma samples from clinical patients, we obtained a total of 2,716,512 reads corresponding to a group of bacteria species. Notably, common pathogens associated with bloodstream infection, including E. coli , K. pneumoniae , A. baumannii , H. influenzae , P. aeruginosa , E. faecalis , E. faecium, S. aureus , and S. epidermidis, were among them (Figure 2). By algorithms and programs developed in‐house and software available from commercial sources and scientific/technology communities, we successfully designed primers and probe sets for nine selected strains. We did the design with conservativeness and specificity in mind and verified them experimentally. In the grouping of targets to separate panels, we did our best to avoid the inclusion of oligo pairs with pronounced potential interaction into the same panel, so as to minimize the possible interaction.

It is noteworthy that typical commercially supplied Taq enzymes are produced by using E. coli expression system, making them subject to non‐specific amplification interference due to E. coli nucleic acid residual contamination [34]. So, for Panel 1 in which E. coli is a target member, we adopted a low nucleic acid residue enzyme for PCR. Since Taq enzyme products on the market with host nucleic acid residues removed are priced high, we chose regular Taq enzymes for Panel 2 and Panel 3.

Testing on blood culture–positive clinical samples, results from the assay align well with that of blood culture overall. It indicated a high practical value of this assay. In one case, the assay failed while the blood culture method showed S. epidermidis positive. However, next‐generation sequencing method also failed to find any reads matching the sequence of S. epidermidis. We suspect that this discrepancy may relate to the degradation of cfDNA of the sample during storage. Alternatively, sample contamination during blood culture in the hospital might be another possible reason, because contamination with coagulase‐negative staphylococci is relatively common [35]. In another case, the assay showed positive results for both E. faecalis and A. baumannii for an E. faecalis blood culture positive sample. In this case, next‐generation sequencing result revealed corresponding reads to A. baumannii , supporting the assay's result over blood culture in sensitivity, accuracy, and versatility.

It is rather unfortunate that we are not quite confident to designate a positive cut‐off Ct value to the kit since the number of BSI samples acquired so far is insufficient. Therefore, more clinical samples are needed to drive this development work further. That includes the optimization of these three panels targeting nine common pathogens and extending to other BSI microbes.

In addition to the high sensitivity and rapid detection time demonstrated by the kit, a comprehensive evaluation of its performance against other existing detection methods has been conducted. In terms of cost‐effectiveness, the kit utilizes a well‐established quantitative real‐time PCR (qPCR) platform, which offers significant cost advantages due to its standardized reagents and streamlined workflow. In terms of operational convenience, this method simplifies the plasma cell‐free DNA (cfDNA) extraction process, requiring minimal hands‐on time. In contrast, whole blood–based detection kits typically involve additional steps such as bacterial cell wall disruption, which not only complicates the workflow but also reduces overall extraction efficiency. As for clinical adaptability, the kit is fully compatible with automated extraction systems, enabling high‐throughput processing of multiple samples simultaneously. Furthermore, its compatibility with all‐in‐one instruments makes it suitable for diverse clinical settings, including point‐of‐care testing, intensive care units (ICUs), and outpatient clinics.

In conclusion, we have developed a novel bloodstream infection diagnostic assay based on plasma cfDNA. The assay offers rapid and accurate detection, with the entire process, from sample to result, costing approximately only 1.5 h. This provides timely pathogenic evidence to doctors and patients for their medical needs, demonstrating a high value in clinical practice.

Author Contributions

Dijun Zhang, Jiang Cao, Yiping Wang, Yong Luo, and Yong Wu designed the experiments and edited the manuscript. Kai Niu, Zhiyu Pang, Jennie Luo, Chaoqun Xia, Zhi Xu, and Yanqiao Qian performed and analyzed the experiments. Miaomiao Niu, Zhitong Sun, Dijun Zhang, and Yong Luo performed the bioinformatics analysis and algorithm support. Kai Niu wrote the manuscript. All the authors have read and approved the final version of this manuscript.

Ethics Statement

Ethics approval was granted by the Ethics Committee of the Affiliated People's Hospital of Ningbo University (Approval number: 2020 Research Ethics Review NO 68).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1. Strains used in this study.

Data S1. The raw data of Figure 5.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.