Probe-capture targeted next-generation sequencing: A novel approach for pathogen and antimicrobial resistance detection in sepsis

Abstract

Sepsis is a syndrome of life-threatening organ dysfunction caused by a dysregulated immune response to infection.^{1, 2} It is a global health priority as recognised by the World Health Organisation³ and according to the Global Burden of Disease Study 2017, 11 million sepsis related fatalities and 48.9 million sepsis episodes occurred globally in 2017.⁴ Bloodstream infections (BSIs), defined by the presence of viable bacteria (or fungi) in the bloodstream, are an important cause of sepsis.^{5, 6} Early identification of a causative pathogen and its antimicrobial resistance (AMR) profile is essential for appropriate and timely treatment.^{7, 8} However, using the current gold standard of microbial identification through blood cultures (BCs), a causative pathogen is only detected in around 30% of cases.⁹ Although culture-based methods are inexpensive and simple, their turnaround time (TAT) can stretch to several days¹⁰ (Figure 1), delaying diagnoses and leading to inappropriate antimicrobial treatment.¹¹ Furthermore, BCs may fail due to slow-growing, fastidious or non-culturable microbes, low microbial load or prior antibiotic treatment.^12-14 Clinical metagenomic next-generation sequencing (mNGS) uses untargeted shotgun sequencing of all DNA or RNA in a sample to identify microbial genomes present in the sample^15-18 (Figure 1). mNGS provides a high-throughput pathogen detection method and is increasingly employed in various infectious syndromes, including BSI,^{19, 20} central nervous system,²¹ bone and joint,²² respiratory infections.²³ Although mNGS is pathogen agnostic and could be quicker than BC (Figure 1), sensitivity can be compromised due to high human DNA background (>99%) in blood samples and low microbial loads during BSI (1–10 colony-forming units [CFU]/mL).^24-27 Thus, a more sensitive approach for diagnosing BSI using NGS is urgently required.

Targeted NGS (tNGS) is an approach that selectively amplifies specific genomic regions or gene sequences, such as AMR determinants. This targeted enrichment enhances sensitivity and minimises host nucleic acid background, offering improved performance over untargeted shotgun mNGS.³² An example of tNGS is hybridisation-based probe-capture metagenomics. Short DNA/RNA oligonucleotide probes (or “baits”) are designed to be complementary to various pathogen sequences and can be designed to identify over 3,000 species.^33-35 This method enables greater genome coverage by using overlapping probes, ensuring more comprehensive target capture and reduced background or host DNA^{32, 36} (Figure 2). Although RNA probes offer greater sensitivity and hybridisation stability than DNA probes, thereby improving hybrid capture efficiency, their inherent instability and careful handling requirements have led to the preferred use of DNA probes (Figure 2).³⁷ Probe-capture was initially developed for human genomic studies to identify disease-associated mutations and rare variants in human genomes³⁸ and has also been developed for NGS platforms like Illumina and Agilent genomics.^{39, 40} This capture approach has shown effectiveness in investigating mitochondrial disorders,^{41, 42} and screening of potential genetic variants in patients with cancer.^{43, 44} The effectiveness of probe-capture has been demonstrated for infectious disease surveillance^{45, 46} and detection of an array of pathogens, including viruses such as HIV,⁴⁷ HCV⁴⁸ and severe acute respiratory syndrome coronavirus 2,⁴⁹ bacteria such as Mycobacterium tuberculosis⁵⁰ and fungi such as Candida albicans.⁵¹ The design of these oligonucleotide panels can be either customised from synthetic DNA manufacturing platforms (e.g. Integrated DNA Technologies⁵² and Twist Bioscience⁵³) or bought as pre-designed commercial panels (e.g. Comprehensive Viral Research Panel, Twist Bioscience).⁵⁴ Furthermore, comprehensive AMR panels capable of sequencing a multitude of AMR genes are available from various sequencing platforms, including Illumina (e.g. Illumina Respiratory Pathogen ID/AMR Enrichment Panel [RPIP] Kit)^{55, 56} and Ion Torrent GeneStudio (DARTE-QM).⁵⁷

Probe-capture tNGS has successfully been utilised to identify BSI pathogens from whole blood and plasma samples.^{33, 59, 60} Sun and colleagues demonstrated that a probe-capture technique had a higher pathogen detection rate compared to BCs (51.6% vs. 17.4%, p < 0.001), and when combined with both BCs and real-time polymerase chain reaction (PCR), probe-capture had a concordance rate of 91.8% with a sensitivity and specificity of 100% and 87.1%, respectively. Based on the probe-capture results, antibiotic therapy was successfully adjusted for 64 (34.8%) patients and 41 (22.3%) showed a more than 2-point reduction in the Sequential Organ Failure Assessment (SOFA) score post-antibiotic adjustments.⁵⁹ This study also demonstrated the possibility of obtaining final reports within 24 h, significantly shorter compared to BCs (median TAT of 19.1 vs. 111 h, p < 0.001). Although specific PCR can provide rapid pathogen identification in clinical diagnosis within 12 h⁶¹ and has been shown to be more sensitive than mNGS, it is not as agnostic as probe-capture.⁶² Commercial PCR panels, such as the BioFire Blood Culture Identification (BCID) 2, require positive BCs, delaying TAT.⁶³ In a prospective study evaluating 80 patients with suspected BSI, probe-capture demonstrated faster TAT as well as higher sensitivity compared to both BC (91.3% vs. 23.2%, p < 0.001) and mNGS (91.3% vs. 63.9%, p = 0.001). Additionally, 22/80 (31.9%) of patients showed clinical improvement post-therapy adjustments based on probe-capture results.³³

While probe-capture tNGS offers significant advantages over conventional and shotgun-based approaches, several practical and technical limitations must be considered for its broader clinical implementation. These include a need for a known reference genome or transcriptome for probe design, a hurdle when it comes to identifying novel pathogens where reference data are limited.⁶⁴ Utilising a defined panel can limit detection of novel pathogens, as these probes can tolerate up to 20% difference with over 50% coverage to the reference genome and cannot identify more divergent organisms.⁶⁵ Probe-capture approach is challenging for hypervariable sequences, such as in pathogens with high mutation rates and potential for decreased capture performance with increasing divergence between probe and target.^66-68 Moreover, probe-capture can be less effective for highly repetitive regions or those with high guanine-cytosine content, as it presents complexity with probe design and hybridisation.⁶⁹ Furthermore, this approach could have additional bioinformatics challenges, including hybridisation bias and complications in downstream analysis, such as off-target capture and chimeric reads that complicate processing.^{46, 70, 71} Finally, this approach could incur substantial cost and significant time investments with complex workflows.^{54, 72}

Despite these challenges, probe-capture tNGS remains a highly promising and potentially transformative approach for infectious disease diagnostics, including sepsis diagnostics. Its ability to deliver robust and precise pathogen and AMR genes even in samples with low microbial loads and high host background addresses key limitations of conventional and shotgun-based mNGS methods.^{18, 59} This positions probe-capture as a valuable tool not only for diagnostics but also for AMR stewardship and surveillance initiatives.⁵⁹ Moreover, its potential for integration into centralised clinical laboratories paves the way for rapid, actionable diagnostics in critical care settings where time-sensitive decisions are critical. The clinical utility of probe-capture tNGS must now be rigorously evaluated, ideally through randomised clinical trials focusing on patient-centred outcomes to ensure meaningful translational impact.^{33, 59}

All authors contributed equally to this work.

The authors declare no conflict of interest.

This work was supported by the National Health and Medical Research Council, a Clinical Research Fellowship from Queensland Health and Queensland Children's Hospital Foundation.

Ethical clearances not required.