Reassessing the promise of NGS in sepsis diagnosis: overcoming scalability barriers

Keywords: Sepsis diagnosis, Next-Generation Sequencing (NGS), rapid diagnostics, pathogen detection, Point-of-Care Testing (POCT)

Sepsis remains a leading cause of mortality worldwide, necessitating rapid and accurate diagnosis to improve patient outcomes [1]. Traditional diagnostic methods, such as blood cultures and molecular techniques like polymerase chain reaction (PCR), are routinely employed for identifying sepsis-causing pathogens. However, these conventional diagnostic approaches have significant limitations in terms of both speed and comprehensiveness. As a result, the search for new methods enabling rapid identification of sepsis pathogens has become a focal point of research.

Over the past 15 years, next-generation sequencing (NGS) technologies have emerged as potentially powerful diagnostic tools, offering comprehensive pathogen detection and rapid identification capabilities that surpass those of traditional methods. Numerous studies underscore the significant potential of NGS, emphasizing its ability to identify a wide range of pathogens in a single test [2–5]. However, despite nearly two decades of being heralded as a promising technology for rapid pathogen detection, NGS has yet to be routinely adopted for this purpose. Several barriers, including high costs, infrastructure requirements, complex data interpretation, and the need for standardized protocols, have been identified as contributing factors [6–9]. While these challenges are significant, they do not appear insurmountable; this suggests that there may be other overlooked fundamental issues.

Examining the specific requirements of NGS and the nature of sepsis, as well as the need for a rapid pathogen detection system, might help to uncover overlooked underlying issues. The critical importance of speed in sepsis diagnosis is underscored by the fact that delays in initiating treatment are associated with a high mortality risk [10–12]. While NGS technology enables rapid pathogen identification, it typically requires the collection of multiple patient samples to be cost-effective and operationally efficient. This raises two primary issues, which are closely related to each other. The first issue is the reliance on higher-capacity machines to reduce costs. The transition from Sanger sequencing to next-generation sequencing technologies (second-generation sequencing) in the mid-2000s [13,14] has led to significant cost reductions [15–17] that far exceed Moore’s Law expectations [18]. However, these reductions have not continued at the expected pace, and the observed cost decreases have primarily resulted from economies of scale [19].

Although cost reductions may seem like a positive development for the widespread use of NGS, new high-capacity devices fail to meet the immediate diagnostic needs in clinical settings. This leads to the second primary issue: scalability. Most NGS platforms are designed for batch processing rather than efficiently handling individual samples. The batching process leads to unacceptable delays in the context of sepsis where every hour of delayed treatment increases the risk of mortality. Most current high-throughput NGS platforms are optimized for batch processing, where multiple patient samples (typically 8, 16, 24, 96, or more) are pooled for simultaneous analysis. The substantial fixed costs associated with initiating an NGS run – including reagents, consumables, and operator time – necessitate distributing these expenses across a larger number of samples to achieve a more economically viable per-sample cost (see Table 1). Laboratories generally accumulate a certain number of samples before commencing a sequencing run, with this threshold determined by factors like workload, instrument capacity, and financial objectives [21,34]. Furthermore, batch processing streamlines laboratory workflows, minimizes manual intervention, and reduces overall processing time for a given set of samples during nucleic acid extraction, library preparation, and sequencing. High-throughput NGS instruments are designed to maximize the utilization of their sequencing capacities, and filling these capacities with a full batch of samples ensures the most efficient and cost-effective output [35].

Table 1.

Indicative consumable cost and sample to report time for selected molecular platforms used in sepsis diagnostics.

Platform* (Example kit)	Minimal batch size**	Turnaround time***	Consumables cost US$****	References
Multiplex realtime PCR (Magicplex Sepsis, TaqMan)	1 sample	3–6 h	$3–20	[20]
Illumina NextSeq 550 midoutput	≥24 samples	15–24 h	≈ $90 (48plex)	[21,22]
Illumina NovaSeq 6000 (S4)	≥96 samples (ideal 384+)	24–36 h	≈ $25 (384 plex)	[23]
Ion Torrent GeneStudio S5 (Ion 530 chip)	≥16 samples	10–14 h	≈ $80 (16plex)	[24,25]
ONT MinION	1–6 samples	7–9 h	≈ $130 (6plex)	[26,27]
ONT PromethION (adaptive sampling)	1–96 samples	6–24 h	≈ $60 (96plex)	[28,29]
MGI DNBSEQG400	≥24 samples	16–24 h	≈ $60 (48plex)	[30,31]
MGI DNBSEQ-T7	≥96 samples (ideal 300–600)	18–24 h	≈ $20 (384 plex)	[32]
MGI DNBSEQ-T10×4 RS	≥384 samples (ideal 1000+)	24–36 h	≈ $10–50 (high-throughput)	[33]

*Diagnostic breadth: Multiplex real-time PCR detects a fixed panel of ≈90 ± 10 predefined pathogens, whereas all other listed platforms perform an unbiased survey of pathogens plus antimicrobial-resistance (AMR) genes.

**“Minimal” reflects practical economics; instruments can technically run smaller pools but at higher cost per sample.

***Sample collection to preliminary species and AMR call.

****Consumables only (library kit, flow cell/chip, mastermix); excludes instrument amortization and personnel costs.

This reliance on batch processing inherently contributes to diagnostic delays, which are often overlooked in discussions of NGS’s speed. Many studies extolling the speed of NGS fail to account for the time required for sample collection and batching prior to analysis. In real-world clinical settings, waiting to collect samples for NGS can lead to delays that undermine its benefits. Therefore, despite its lauded speed, NGS has not yet had a significant impact on routine rapid diagnosis due to these challenges.

However, in terms of both total turnaround time and scalability, approaches like those developed by Oxford Nanopore Technologies (ONT), which require fewer samples and provide real-time data, have introduced groundbreaking sequencing capabilities [36]. These technologies make sequencing more accessible through portable and cost-effective devices, but they still face several challenges such as accuracy. With the latest reagents and chemistries, single-molecule sequencing can yield high-quality reads with accuracies up to Q30 (≥99.9%; ≤1 error per kilobase) [37,38]. Prospective metagenomic studies of bloodstream infections using these chemistries have demonstrated sufficient sensitivity for the rapid “rule-in/rule-out” of common sepsis pathogens, enabling clinically actionable decisions within 3.5 to 7 hours of venepuncture [39,40]. In contrast, applications such as single-nucleotide variant (SNV) calling for antimicrobial resistance genes (ARGs) or outbreak strain typing require higher consensus accuracy, which can be achieved by polishing nanopore assemblies with short-read data [41].

In addition to cost and scalability challenges, the advanced infrastructure and specialized personnel required for NGS further complicate its adoption. Many healthcare facilities lack the necessary components to implement NGS effectively [42]. Addressing these infrastructure and expertise gaps is essential for realizing the full potential of NGS in routine clinical practice for sepsis diagnosis.

Accordingly, to enhance the practical utility of NGS in sepsis diagnosis and help address existing challenges, two potential approaches may be considered: one focusing on technological innovation and the other on implementation strategy.

1. Point-of-Care Testing: The development of platforms that can be used directly at the patient’s bedside would significantly reduce the time between sample collection and pathogen identification.

2. Centralized Daily Sequencing: A centralized system that pools and analyzes patient samples each day, ensuring faster results.

The first of these critical advancements – POC testing – requires further technological maturation. There is ongoing work to create portable and more affordable devices that are easy to use and offer high reading accuracy. The industry’s drive to create economically viable NGS devices capable of efficient single-sample processing – moving beyond reliance on batching for cost-effectiveness – is crucial for widespread POC implementation. These innovations aim to bring sequencing directly to the bedside, significantly reducing diagnostic delays and enhancing the scalability of NGS in high-demand clinical settings. Achieving this will require continued advances in miniaturization and portability, as well as the development of simplified, automated “sample-to-answer” workflows. It would be highly valuable for these systems to reduce manual handling, minimize the need for advanced expertise, and thereby offer the capability to accelerate early steps that often contribute to diagnostic delays, such as microbial nucleic acid isolation and library preparation. The development of automated “sample-to-answer” systems that accelerate these initial steps can enable rapid results even with single or very few samples [43]. Furthermore, ongoing research is essential to improve sequencing accuracy and reduce error rates in portable devices, ensuring diagnostic reliability. Finally, the integration of user-friendly interfaces and rapid, AI-driven data interpretation tools will ensure immediate, actionable results directly at the patient’s bedside. These cumulative innovations are pivotal for transforming sepsis management by enabling rapid pathogen identification where and when it is most needed.

However, no significant technological barriers prevent the medical field from pursuing the second pathway – particularly considering that three chokepoints influencing how rapidly NGS results can be returned as clinically actionable have already been addressed. First, automated library preparation has significantly improved efficiency: robotic workstations have reduced hands-on time by 80–90% and lowered error rates, enabling a single technologist to prepare 96- or even 384-sample runs [44]. Second, bioinformatics analysis and interpretation have been markedly accelerated by the use of cloud-compatible, containerized pipelines, which can now complete taxonomic classification and antimicrobial resistance (AMR) calling for a 5 Gb dataset in under 60 minutes using off-the-shelf GPU instances. This eliminates the need for local high-performance computing infrastructure and dedicated personnel [45,46]. Third, report generation and delivery have been streamlined through modern laboratory information management systems (LIMS), which auto-populate standardized PDF or HL7 reports and transmit them directly to electronic health record (EHR) systems – reducing turnaround time by an additional 4–24 hours [47]. Given the economic and logistical constraints, it is unrealistic for every healthcare institution to operate a full-scale NGS laboratory. Instead, the establishment of regional or national centralized sequencing laboratories – equipped with rapid processing instruments and pipelines optimized for urgent sepsis samples – could serve as a practical and scalable solution. This centralized model would reduce the need for technical personnel, NGS devices and computational infrastructure at each center, leading to cost reductions.

Building on this foundation, centralized sequencing has shown promise in improving efficiency and reducing costs under controlled conditions. About 15 years ago, improvements made by large genome centers to the Illumina sequencing system demonstrated the potential for increased efficiency and reduced costs through centralized operations [48]. Today, a notable commercial example of this approach is an NGS plasma test, where patient samples are collected at various locations and sent to a central laboratory for processing, ensuring consistency and accuracy in pathogen detection (e.g., Karius) [49]. While such examples illustrate the feasibility of centralized models, they remain limited in scope and do not yet reflect widespread clinical integration, particularly for acute, time-sensitive conditions like sepsis.

In conclusion, while NGS undoubtedly has the potential to revolutionize sepsis diagnosis, its current implementation falls short in scenarios requiring immediate results due to logistical delays in sample processing. Unlike chip technologies – which continually become cheaper in line with Moore’s Law, originally formulated for the semiconductor industry [50]—NGS technology does not follow a similar cost-reduction trajectory. Thus, addressing this challenge through the development of single- or few-sample processing, economically viable NGS devices, or centralized rapid processing hubs is critical. For centralized hubs, automated bioinformatics workflows – supported by artificial intelligence – have further minimized reliance on specialized personnel, enhancing both speed and diagnostic reliability [51]. Additionally, standardized practices in sample collection, sequencing, and analysis have strengthened inter-laboratory reproducibility [52]. The integration of cloud-based computing resources for data storage and analysis has provided scalable and flexible infrastructure capable of handling the large volumes of data generated by NGS. This approach facilitates collaboration and data sharing among researchers and clinicians, enhancing the overall utility of sequencing data [53].

While continued research is necessary to develop new devices, centralized sequencing can be implemented using currently available technologies to overcome these obstacles. The future of sepsis diagnosis lies in the ability to rapidly and accurately identify pathogens, and the strategic use of available technology will play a crucial role in this transformation

Funding Statement

This paper was not funded.

Disclosure statement

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties

Writing assistance

No writing assistance was utilized in the production of this manuscript

Reviewer disclosure

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.