Direct-from-Blood Detection of Pathogens: a Review of Technology and Challenges

ABSTRACT

Blood cultures have been the staple of clinical microbiology laboratories for well over half a century, but gaps remain in our ability to identify the causative agent in patients presenting with signs and symptoms of sepsis. Molecular technologies have revolutionized the clinical microbiology laboratory in many areas but have yet to present a viable alternative to blood cultures. There has been a recent surge of interest in utilizing novel approaches to address this challenge. In this minireview, I discuss whether molecular tools will finally give us the answers we need and the practical challenges of incorporating them into the diagnostic algorithm.

INTRODUCTION

Blood cultures have been the staple for detection of bloodstream pathogens beginning with manual systems that required daily and terminal subcultures and followed in the late 1960s by the development of automated blood culture systems (1, 2). Incremental improvements over time in the technology and our understanding of pathogen dynamics have improved blood culture yield. The advent of continuous monitoring blood culture systems allowed for the reduction of incubation time from 7 days to 5 days (1, 3). Blood culture bottles now typically accommodate a larger volume of blood (10 mL) than in the 1980s (5 mL), which allows us to better compensate for the low bacterial burden in patients with sepsis (1). Additives to culture media and use of advanced detection algorithms have further enhanced blood culture yield and reduced the time to a positive result. The use of closed incubation systems has offered not just better turnaround time (TAT) but also better yield compared to older automated systems (4).

A positive blood culture only informs us of the presence of bacterial growth. Patients with suspected sepsis should already be on broad-spectrum antibiotics based on symptoms/results at time of presentation (5). Actionable information comes from the organism identification and detection of resistance markers/antimicrobial susceptibility testing results, which guide treatment. The Gram stain allows the provider to potentially narrow therapy and de-escalate antibiotics. Bacteria, however, have been known to demonstrate uncharacteristic staining, and this along with user error could provide incorrect information with potentially severe consequences (6). Further information required biochemical and antimicrobial susceptibility testing, which would take anywhere from 24 to 72 h using automated systems. The use of rapid multiplexed PCR panels and matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS)-based identification of pathogens directly from positive blood cultures allows for the provider to use the local antibiogram to make informed choices for treatment (7). Multiplex PCR-based detection of resistance markers allows for even more confident antibiotic selection prior to availability of full susceptibility test results in 48 to 72h. Novel rapid susceptibility test methods could provide full susceptibility test results with MICs in under 12 h from the time of the positive blood culture result (8, 9). The presence of multiple pathogens in the same positive blood culture bottle may complicate the use of rapid susceptibility test methods (9).

All of this is contingent on the initial positive blood culture result, which requires at least 12 to 48 h of incubation before a positive result (1, 2). There has been a steady progressive reduction in time to actionable culture results for septic patients (10). This happened either by reduced time to blood culture positivity or incorporation of rapid identification methods directly with positive blood cultures (7, 11). Studies have shown significant risks to septic patients associated with delayed results and inappropriate therapy, with each hour of delay in administration of appropriate antibiotics associated with increased mortality (12–14). Some studies have suggested that the window of opportunity may be longer. In a large study examining over 9,000 patients with bloodstream infection (BSI), van Heurverswyn et al. demonstrated that the risk of death only increased at 12 h after blood culture collection (15). The authors speculate that this may be due to the variability in duration of time elapsed between onset of sepsis and arrival of the patient in the emergency department (15). Unfortunately, providers often do not have the luxury of waiting for a positive blood culture. Even worse, blood cultures may be negative in a significant proportion of patients with sepsis (16–18). The absence of reliable diagnostics in this setting is a significant hurdle to timely and appropriate management and emphasizes the need for molecular methods for the detection of sepsis pathogens.

SEPSIS MANAGEMENT AND THE NEED FOR BETTER DIAGNOSTICS

Any conversation about sepsis diagnostics requires an understanding of the current utilization of blood cultures in sepsis. In the United States, the Centers for Medicare and Medicaid Services (CMS) Severe Sepsis and Septic Shock Early Management Bundle (SEP-1) is a key driver of our approach to sepsis management (19). The 2016 Surviving Sepsis guidelines were published jointly by the Society for Critical Care Medicine (SCCM) and the European Society of Intensive Care Medicine (ESICM) (5). The guidelines recommended that blood cultures be collected prior to initiation of antimicrobial therapy within 1 h of presentation for patients with both suspected sepsis and septic shock. The 2016 guidelines were not endorsed by the Infectious Disease Society for America (IDSA) or the Society for Infectious Disease Pharmacists (SIDP) (20, 21). Concerns were raised about the lack of distinction between patients with suspected sepsis and those in septic shock. A significant proportion (>40%) of patients with suspected sepsis are subsequently found to have noninfectious conditions (20). This would lead to both unnecessary blood cultures, antimicrobial usage, and other interventions in the effort to meet the 1-h deadline. The IDSA recommended additional time for evaluation of patients with suspected sepsis (20). SIDP concurred with the IDSA on the need to draw a distinction between patients with suspected sepsis and those in septic shock (21).

The 2016 Surviving Sepsis guidelines put providers in the position of having to act first and think later and potentially promoted the indiscriminate use of blood cultures and antibiotics. The negative consequences include false-positive blood cultures due to contamination, which have been linked with extended length of hospital stay, unnecessary diagnostic procedures, removal of lines, and excessive antibiotic use with particular concern about the empirical use of vancomycin, which is associated with nephrotoxicity in up to 25% of patients (22–24). The costs of a contaminated blood culture episode can range from $8,000 to $25,000 to the hospital (25). The negative consequences of indiscriminate antibiotic use are well described but often disregarded. A review of in-hospital antibiotic administration patterns revealed that the primary drivers that impacted antibiotic use decision making among providers were fear of adverse outcome and uncertainty, with little consideration for antimicrobial resistance or patient harm caused by antibiotics (26). Implementation of diagnostic stewardship initiatives that advocate for judicious use of blood cultures have shown the advantages of an approach that does not cast a broad net with little consideration for collateral damage (24, 27, 28). The push to meet sepsis treatment goals could encourage unnecessary initiation of antimicrobials (20, 29). Some would argue that this aggressive approach has not demonstrated any clear impact on mortality, although this is disputed by others (29, 30).

The Surviving Sepsis guidelines were updated in 2021 and now draw a distinction between patients with septic shock/high likelihood of sepsis versus possible sepsis without shock (31). Antimicrobials (and blood cultures) are recommended within 1 h for patients with septic shock/high likelihood of sepsis and within 3 h for patients with possible sepsis without shock (31). This approach (now endorsed by the IDSA) allows for assessment of possible alternate diagnoses and better stewardship of antimicrobials and blood cultures. Even so, the authors of the 2021 guidance commented that over a third of patients with initial sepsis diagnosis were later found to have noninfectious conditions and that the absence of a “gold standard” test for sepsis pathogen detection remained an ongoing challenge (31). While this debate has primarily centered around blood cultures, any attempt to address the root cause of these problems (i.e., the inadequate sensitivity of blood cultures) using novel technologies will have to navigate the same challenges.

HOW DO WE ADDRESS THE LIMITATIONS OF BLOOD CULTURES?

Blood cultures remain an imperfect tool for a deadly condition—anywhere from 30 to 50% of positive blood cultures are contaminated, with the associated negative downstream impact (23, 32, 33). False-negative blood cultures occur in a significant proportion (40 to 60%) of cases with suspected sepsis (16, 34). Reasons include prior antibiotic therapy in 28 to 63% of patients, presence of fastidious/noncultivable organisms, inadequate volume of blood collected, or lack of sensitivity due to low organism load (35–38). In one study, 30% of central line-associated bloodstream infections (CLABSIs) were due to contaminated blood cultures, which in turn has significant quality and financial implications for health care institutions (39).

This leaves us with a diagnostic conundrum—novel approaches to the diagnosis of sepsis are needed, but the novel tools will need to straddle the fine line between optimal sensitivity and specificity. Challenges faced with inappropriate utilization of blood cultures will carry over and may even be exacerbated in nucleic acid amplification-based testing (NAAT), especially if used indiscriminately. In other areas of the laboratory, NAAT has dramatically changed our approach to diagnosis and management. Viral culture has been completely replaced by molecular diagnostics, with significant improvement in sensitivity and speed. On the other hand, the transition from culture-based to NAAT-based multiplex testing for gastrointestinal pathogens in stool has resulted in up to a 10-fold increase in yield, although the impact on patient outcomes remains a subject of debate (40). The implications of an increased false-positive rate in sepsis are likely to be far more significant.

There are two potential approaches to molecular detection of sepsis pathogens.

1. Rapid near-patient-based instruments that extract genomic material from whole blood and utilize NAAT-based testing directed against a specific set of target organisms to provide results in a relatively short time frame (1 to 3 h).

2. Next-generation sequencing (NGS)-based systems that are able to take an unbiased approach to the detection of bloodstream pathogens but due to the nature of the technology will require specialized equipment/expertise and 12 to 48 h to provide a result.

There are parallel developments in the use of host immune markers for the identification of sepsis patients that require aggressive management. While this technology has the potential to help target blood cultures and other interventions toward the patients most in need, it does not identify the pathogen involved. Further discussion on these methods can be found elsewhere (41).

CHALLENGES FACING NAAT/NGS-BASED DETECTION OF SEPSIS PATHOGENS

Blood tends to be a difficult sample type for molecular analysis, and a number of blood components, including immunoglobulins, hemoglobin, lactoferrin, etc., can interfere with NAAT-based amplification by directly interfering with polymerase activity, reducing reaction efficiency and fluorescence quenching (42, 43). The presence of large quantities of human genomic material in blood also poses a challenge necessitating either the separation of white blood cells prior to DNA extraction or degradation of human genomic material postextraction (36, 44). While sampling larger volumes of blood may improve assay sensitivity, it also increases the challenges associated with isolating pathogen genomic material. Alternative approaches attempt to sequester microbial cell-free DNA (cfDNA) from plasma to be identified by NGS (45, 46). Microbial cfDNA only accounts for small proportion (0.08% to 4.85%) of cfDNA in circulation, although elevated levels may be noted in a number of conditions, including sepsis (45, 47).

In order to improve sepsis diagnosis, NAAT-based testing needs to improve on the failings of blood cultures—e.g., through improved sensitivity and faster time to result—all while maintaining specificity. Blood cultures attempt to overcome their limitations in part by increasing the total volume of blood collected. By drawing two to three sets of blood cultures, we can sample 40 to 60 mL of blood (2, 48). NAAT-based testing is unable to sample such high volumes and will have to rely on increased analytical sensitivity to compensate. Blood cultures rely on viable organisms, whereas NAAT-based testing could detect nonviable cells or even fragments of bacterial genomes. Bacconi et al. estimated that the concentration of pathogen genomic material in the blood was 2 to 3 orders of magnitude higher than that of viable bacteria (17). These findings are correlated with studies using PCR assays to quantitate specific pathogens in blood. Peters et al. were able to quantitate the amount of Staphylococcus aureus and Enterococcus faecalis in 27/31 (87%) and 17/18 (94%) of patients with positive blood cultures, respectively (49). They calculated median bacterial DNA load (BDL) of Staphylococcus aureus at 10.6 × 10³ CFU equivalents/mL and Enterococcus faecalis at 2.7 × 10³ CFU equivalents/mL: i.e., significantly higher than expected levels of viable organism (49). Other studies have shown similarly higher loads of bacterial genomic material in blood relative to culture (50–52).

How does organism load impact assay performance?

Several studies have demonstrated that the quantitative bacterial loads as determined by PCR correlate with patient outcomes and disease severity. Kirkbright and colleagues showed that high 16S rRNA gene load was associated with development of septic shock in patients with positive blood cultures for S. aureus or Escherichia coli (51). A quantitative assay for mecA in patients with S. aureus bacteremia also found that higher levels of mecA were a predictor for mortality (53). In a single-center study of 353 patients, Rello et al. showed that elevated bacterial load (≥1,000 copies/mL) in blood of Streptococcus pneumoniae in patients presenting with community-acquired pneumonia was independently associated with likelihood of septic shock (54). The RADICAL study utilized PCR/electrospray ionization mass spectrometry (ESI-MS) for direct NAAT-based detection of pathogens from blood (55). The authors found that while there was no difference in 28-day mortality between patients with and without positive blood cultures, patients with positive PCR/ESI-MS result had significantly higher mortality (42% versus 26%; P = 0.001) (55). These findings appear to dramatically emphasize the potential for NAAT-based assays not just for detection of bloodstream pathogens but also as a means of identifying patients that might require aggressive intervention.

It needs to be noted that these findings are primarily in the setting of single pathogen assays often in patients with confirmed positive blood cultures. Unbiased analysis of microbial cfDNA from blood in asymptomatic individuals yielded positive results in 22.8% (38/167), with most cases only involving a single species (45). While these frequently included potential pathogens such as Klebsiella pneumoniae or Haemophilus influenzae, they were usually found at lower levels than in patients with culture-confirmed bacteremia (45). Bacterial 16S rRNA genes were detected in 23/89 (26%) of healthy individuals in another study (56). The cfDNA in these asymptomatic individuals likely originated via translocation across epithelia (46, 57, 58). Damage to tissue caused by trauma or infection could also facilitate this process (59). Elevated microbial cfDNA levels have been noted in conditions unrelated to sepsis (59). Grumaz et al. noted that plasma cfDNA levels were significantly elevated in patients with sepsis compared to healthy donors (average classified reads of 9.8% versus 3.5%) (47). This points to the need for quantitative cutoffs in the interpretation of the significance of detection of microbial cfDNA levels. Determination of what constitutes normal background signal is challenged by the variability not just between individuals but in the clinical status of the patient. Severely ill patients with compromised epithelia/trauma may demonstrate relatively higher levels of cfDNA. Blauwkamp et al. noted significantly higher levels of Pseudomonas aeruginosa cfDNA in plasma relative to other organisms, suggesting the need for organism-specific cutoffs to establish significance (45). Simner et al. commented on the challenges of the human contaminome in NGS-based testing, stating that the problem is particularly acute with samples like blood and CSF that have a relatively low ratio of microbial to host DNA, with known pathogens such as Staphylococcus and Pseudomonas often also appearing as common contaminants in significant amounts (60).

WHAT NAAT/NGS-BASED TESTING SYSTEMS ARE CURRENTLY AVAILABLE?

NAAT-based direct-from-blood testing.

Despite extraordinary efforts to unlock the holy grail of clinical microbiology (i.e., a reliable and cost-effective molecular assay for the detection of sepsis pathogens), testing options remain limited. This is not due to lack of effort but is a reflection of the challenges involved in developing such an assay. Significant investments were made in platforms such as the Roche Septifast (Roche Diagnostics, Basel, Switzerland) and the Iridica (Abbott Diagnostics, IL, USA), which were eventually withdrawn from the market (61). The story of the Septifast and the Iridica assays serve as a cautionary tale for assay developers who invested significant resources and time into bringing these assays to market in Europe. Despite being CE marked for over 2 years, the Iridica platform was only adopted by three hospitals and was eventually withdrawn from the market (62). The practical challenges of utilizing these assays along with regulatory rather than performance issues were the primary drivers of this decision (62). A number of other platforms are available outside the United States, including VYOO rapid pathogen identification system (Analytik Jena Gmbh, Jena, Germany), MagicPlex Sepsis (Seegene, Seoul, South Korea) and Sepsitest (Molzym Molecular Diagnostics, Bremen, Germany). These platforms demonstrated variable performance, with most being fairly time-consuming/labor-intensive and also having inadequate positive/negative predictive value to be considered a viable alternative to blood cultures (10). There are several platforms currently in various stages of development. Some of these offer sample extraction/concentration of whole blood by utilizing their systems, while they continue to develop NAAT-based detection of pathogens from blood. Detailed information on targets and assay performance is not discussed here because it is either not available or subject to change for most of these platforms as they go through development.

There is only one manufacturer currently offering an FDA-approved NAAT-based assay for the direct detection of bloodstream pathogens. The T2Bacteria (T2B) and T2Candida (T2C) (T2 Biosystems, Lexington, MA, USA) assays are intended for detection of bacterial and fungal pathogens, respectively, from whole blood. Each assay uses 4 mL of whole blood on an automated platform to provide results in 3 to 5 h. The T2B assay detects Enterococcus faecium, S. aureus, Acinetobacter baumannii, Klebsiella pneumoniae, Escherichia coli, and P. aeruginosa. The T2C assay can detect 5 Candida species—Candida albicans/Candida tropicalis, Candida glabrata/Candida krusei, and Candida parapsilosis. The assay relies on NAAT-based testing combined with T2MR magnetic resonance technology (i.e., superparamagnetic particles with target-specific probes) and claims a limit of detection (LOD) of 2 to 11 CFU/mL (63). The T2B assay covers targets representing the pathogens present in >50% of all bloodstream infections (BSIs).

How well does NAAT-based testing perform in sepsis?

In one analysis by Nguyen et al., the T2B assay showed 90% sensitivity and 90% specificity for proven BSIs. A total of 146/1,427 (10%) patients had positive T2B but negative blood cultures (63). The challenge with evaluation of NAAT-based testing in these settings is determining whether NAAT-positive/blood culture-negative results represent true positives missed by culture or NAAT false positives. By evaluating all culture results within 21 days of the positive T2B result, Nguyen et al. determined that at least 62/146 (42%) of the positive T2B results represented possible BSIs, but 58/146 (40%) were likely false positives (63). The positivity rate for T2B was 13% (181/1,427) versus only 3% (39/1,427) for blood cultures. T2B was false negative in 4/39 (10%) patients with positive blood cultures. Mean times for organism detection/identification were 3.6 h for T2B and 38.5 h for blood cultures (63). In another analysis of T2B performance by Di Angelis et al., the assay showed 83% and 97.6% sensitivity and specificity, respectively, compared to blood cultures (64). Over half of the (11/20) BSIs were caused by bacterial species that were not on the T2B panel. Ten isolates were detected on both platforms, but 20 were detected on T2B alone and 2/12 were missed by T2B. Of the 20 isolates detected by T2B alone, 7/20 (35%) were considered to be true-positive results based on chart review (64). Kalligeros et al. further examined patients with discordant T2B-positive/blood culture-negative results. A total of 11/21 (52.5%) represented probable BSIs, while 6/21 (28%) were likely contaminants (65).

Outside of the United States, there is a significant body of literature assessing the performance of the Septifast platform, which identifies 25 common sepsis pathogens (66, 67). Chang et al. evaluated 34 Septifast studies and found that the assay had high specificity but inadequate sensitivity (66). Similar findings were noted in another assessment of this technology, with 85.8% (95% confidence interval [CI], 83.3% to 88.1%) specificity and 50% (95% CI, 39.1% to 60.8%) sensitivity (67). Another technology review evaluated both the Septifast assay and Iridica system (which utilizes PCR /electrospray ionization-mass spectrometry [ESI-MS]) (68). The authors found in studies comparing the Septifast directly to blood cultures an estimated summary specificity of 0.86 (95%, credible interval [CrI], 0.78 to 0.92) and sensitivity of 0.48 (95% CrI, 0.21 to 0.74). Four studies comparing Iridica with blood culture determined estimated summary specificity of 0.84 (95% CrI, 0.71 to 0.92) and sensitivity of 0.81 (95% CrI, 0.69 to 0.90) (68). These studies all came to the same conclusion: that a positive result on these platforms offered value to the provider but a negative result did not (68).

Next-generation sequencing-based testing.

Next-generation sequencing (NGS) has the potential to change the landscape of diagnostic microbiology. There are still significant hurdles before we can harness the full potential of this technology. The unbiased approach to pathogen detection avoids many of the problems faced by NAAT-based testing alone. This may also prove to be a double-edged sword—the presence of microbial cfDNA in whole blood/plasma as previously described gives NGS an advantage but also challenges the specificity of the process. This does allow the use of NGS for detection of localized infections and does not limit its role to sepsis. NGS-based testing at this time requires expertise and instrumentation that are not routinely available in clinical microbiology laboratories. The testing process is relatively time-consuming relative to NAAT and cannot be used in the near-patient setting.

Performance of NGS-based testing.

Rossoff and colleagues evaluated the performance of NGS in immunocompromised pediatric patients where conventional testing had failed to identify a pathogen (69). This is a particularly challenging population because of the low volume of blood available for culture. The primary concern was fungal infection (55%), followed by sepsis (22%). NGS results were typically available within 48 h of sample collection. NGS was positive in 70/100 samples tested, with 33/70 detecting two or more pathogens. The majority of samples (52/70) had at least one bacterium identified, and 56/70 were determined to be clinically relevant pathogens. In 6 cases, NGS did not detect a pathogen that was subsequently identified by routine testing. It should be noted that most of the cases assessed here were chosen because conventional testing had failed to provide an answer.

Blauwkamp et al. examined the performance of an NGS-based approach for the detection of 1,250 human pathogens using microbial cfDNA in plasma from 350 patients presenting with sepsis (45). NGS identified 84.9% (112/132) of potential pathogens that were detected by routine methods and overall identified a potential pathogen in 169/348 versus 132/348 for routine diagnostic methods. Adjudication of results based on chart review gave the NGS platform a final sensitivity of 93.7%. In patients who had prior antimicrobial therapy within the 2 weeks prior, NGS identified potential pathogens in 46/96 (47.9%) compared to 19/96 (19.6%) for blood cultures alone. As mentioned previously, testing of plasma from asymptomatic individuals detected microbial cfDNA in 22.8% (38/167) of samples. Specificity was determined to be 62.7% in sepsis patients. The time to result on average was 53 h, including time required for shipping (45).

In other studies, NGS outperformed blood cultures by detecting a number of bacterial and viral pathogens that would have otherwise been missed (47, 70). In a study by Long et al., the number of positive patients increased from 12.82% (12/78) using routine blood cultures to 30.77% (24/78) using NGS (70). NGS was also able to detect resistance genes such as mecA and vanA/vanB, although this capability may vary between platforms as it requires high read coverage (47). Jing and colleagues utilized NGS to detect 53 pathogens in 209 samples (71). The overall positivity rate was 110/209 (52.6%). Conventional testing was positive in 86/209 samples, and NGS was positive in 30/123 samples that were negative by conventional testing. The overall sensitivity and specificity for NGS were 86.1% and 80.2%, respectively, using a composite reference standard. Compared to blood culture along, the sensitivity and specificity of NGS were 80.8% and 79.6%, respectively, for bacteria. In 12 samples, NGS failed to detect pathogens that were found by conventional testing (culture and PCR), including both bacterial and viral targets (71). Looking beyond sepsis, NGS may have an impact in identifying pathogens in patients with culture-negative endocarditis. Flurin et al. were able to identify a potential pathogen in 5/6 patients with blood culture-negative endocarditis (72).

Blauwkamp et al. also commented on the challenges of interpreting NGS results and establishing diagnostic sensitivity, especially given the detection of multiple organisms in each sample and the diversity of the microbiome across different patient populations (45). Microbial cfDNA was measured as molecules per microliter of plasma (MPM) and in cases of true infection was present in concentrations stretching across 20 to 450,000 MPM. For reference, microbial cfDNA in healthy donors was typically <300 MPM (45). This again speaks to the challenges of interpreting NGS results as these values will vary based on platform and patient population.

The common thread among these studies was that NGS offered incredible potential to identify pathogens that may be missed by conventional testing. The challenge, however, remains the difficulty in interpretation of culture-negative/NGS-positive results, false negatives, and the detection of multiple targets.

FUTURE PIPELINE

This minireview may not represent a complete list of platforms currently under development. Many of these platforms often have minimal information available, and these are subject to change.

Qvella FAST-ID BSI panel.

The Qvella FAST-ID BSI panel (Qvella Corporation, Ontario, Canada) is intended to detect targets covering 90% of pathogens causing BSIs from a single whole-blood sample in under 1 h. The built-in extraction selectively lyses blood cells, removes the debris, and isolates/concentrates bacteria. This is followed by lysis and PCR-based detection of rRNA targets covering a wide spectrum of Gram-positive, Gram-negative, and fungal targets.

LiDia Seq BSI/AMR test.

The LiDia Seq BSI/AMR test platform (DNAe, London, United Kingdom) uses semiconductor-based sequencing to detect pathogens and resistance markers from whole blood with results available in 3 to 4 h.

Biospectrix.

The Biospectrix platform (3iDx, Germantown, MD, USA) uses pressure-driven selective lysis within a cartridge to separate bacteria from whole blood while breaking down blood cells in under 30 min. The isolated bacteria could then be analyzed by a variety of molecular methods.

Helixbind.

The Helixbind platform (Helixbind, Inc., Boxborough, MA, USA) uses a novel electrostatic-based method in a fully automated cartridge-based system (RaPID) to detect pathogens in under 3 h from 6 mL of whole blood.

SepsiSTAT.

The automated SepsiSTAT system (Momentum Biosciences, Cardiff, United Kingdom) uses 12 mL of whole blood to extract bacterial organisms from both PCR-based detection and isolation of viable bacteria for other applications.

In addition, reference laboratories such as Dayzero Diagnostics, Karius, and PathoQuest, among others, offer NGS with relatively rapid turnaround times. These often incorporate overnight shipping and mobile app-based reporting for ease of use.

PROS AND CONS OF NOVEL APPROACHES TO SEPSIS DIAGNOSTICS

While both NAAT and NGS offer the promise of improved sensitivity over blood cultures, each approach brings unique advantages and challenges that will play into how they may be incorporated into the diagnostic algorithm.

Ease of use.

NAAT-based testing offers the potential of rapid, near-patient testing to rapidly detect sepsis pathogens that could address the requirements laid out in the Surviving Sepsis guidelines. NGS remains relatively more labor-intensive and expensive and requires significant expertise, with results not available in a timely manner. Hospitals would need specialized molecular laboratories and personnel on site or would have to utilize a reference laboratory for NGS. A fully automated NAAT platform could potentially be used in the point-of-care setting of a community hospital.

Sensitivity.

Both NAAT-based and NGS testing offer the potential of better sensitivity than blood cultures. The ability to detect bacterial cfDNA may offer both significant advantages and disadvantages. NAAT-based assays that sequester intact bacteria may sacrifice some of that potential for increased sensitivity in favor of increased specificity.

Specificity.

NGS-based approaches that utilize cfDNA will be challenged to develop quantitative cutoffs that will likely be platform/laboratory specific. It may be burdensome for individual clinical laboratories to independently develop such cutoffs. NAAT-based testing will likely provide qualitative results only, and institutions will have to develop guidelines on how to interpret culture-negative/NAAT-positive results. Assessment of actual assay specificity and interpretation of the significance of culture-negative and NAAT/NGS-positive results remain the most significant challenge to adoption of these technologies.

Sample type.

NAAT-based testing will likely have to rely on whole blood in order to maximize bacterial yield, whereas NGS has been primarily developed with plasma-based assays.

Cost of testing.

NGS-based testing of plasma is available primarily at reference labs, and costs are typically >$2,000 per sample. NAAT-based testing, while relatively cheaper, can easily add up to hundreds of thousands of dollars in annual expenditure if used in significant numbers. In both cases, costs are likely to drop significantly as the technology continues to develop. Studies that have evaluated the cost benefit of NAAT-based testing are limited in nature, and further work is needed in this area to justify the use of this technology (73).

Practical challenges.

NGS-based testing requires expensive instrumentation, trained personnel, and bioinformatics expertise. NGS also typically involves an extended testing time frame compared to NAAT-based testing. At the very least, overnight testing is required, but results may take 12 to 48 h. Many laboratories in the United States utilize reference laboratories for NGS-based testing, but this may extend the time frame to obtain results. NAAT testing will likely be easier to implement but will still require laboratory resources and personnel. The biggest challenge will be setting up a stewardship-based system to ensure that results are utilized in real time.

HOW WILL NAAT/NGS-BASED TESTING IMPACT THE CLINICAL MICROBIOLOGY LABORATORY

The utilization of NAAT-based testing involves a significant investment in laboratory resources and personnel. The cost of routine NAAT-based testing for BSI is likely to be significant and in light of the resource crisis facing clinical laboratories needs to be carefully considered. As with many novel technologies, NAAT-based testing does not replace existing diagnostic testing (blood cultures in this instance) and adds another layer of testing that would likely require support 24/7 to be effective. Implementation of novel technology in the clinical microbiology laboratory will not be successful unless clinical decision support is available for providers round the clock. This would involve developing and leveraging antimicrobial stewardship and electronic medical record (EMR)-based resources.

For laboratories that are consolidated and serve multiple institutions, this raises questions as to whether rapid NAAT-based testing platforms need to placed close to the point of care, such as emergency departments or local intensive care units (ICUs). In the absence of CLIA (Clinical Laboratory Improvement Amendments)-waived testing options, these will need to be manned by laboratory personnel 24/7. If utilizing consolidated laboratories, then the transport time involved may limit some of the benefits gained from rapid NAAT-based testing. While guidelines recommend that blood cultures be processed in a timely manner, there is no consensus or regulatory guidance on what constitutes “timely.” This is problematic because as improvements in blood cultures and associated processes reduce the time to detection, the time spent in transport now becomes a significant component of the time to result (74).

Limited data are available on the impact of rapid NAAT-based platforms on sepsis in the real-world setting. This is an area of concern because in the absence of data showing impact on mortality, length of stay, or utilization of other resources, it becomes increasingly difficult to cost justify these novel methods. The other challenge is the lack of clear guidance on the target population for rapid NAAT-based assays. A whole host of questions remain to be answered:

• Are NAAT-based assays best used by casting a wide net or better targeted against specific population groups?

• Are they better utilized in the tertiary care setting/ICU, and do they have a role in the emergency department (ED)/community hospitals?

• What is the role of repeat NAAT testing in patients who are initially NAAT negative but with persistent signs and symptoms of sepsis? What is the role of repeat testing of NAAT-positive patients to determine response to therapy?

• Will multiple NAAT tests need to be drawn from two different sites to delineate infection from contamination (similar to blood cultures)?

WILL THE INCREASED SENSITIVITY AND SPEED OF NAAT-BASED TESTING ALLOW FOR BETTER MANAGEMENT OF PATIENTS WITH SEPSIS AND BETTER OUTCOMES?

Blood cultures are clearly inadequate for the detection of pathogens in sepsis—even when positive studies have demonstrated that blood cultures do not always translate to significant difference in clinical outcomes, although they do offer the opportunity to perform antimicrobial susceptibility testing (18). The question then is whether NAAT-based testing can offer something different. Quirino et al. showed that in 6/17 T2B-positive patients, empirical antibiotic coverage was changed as soon as results were available (75). For the remaining 11 patients, the empirical coverage was considered adequate and no significant impact on 21-day mortality was noted when using T2B (75). Other studies have assessed the potential impact of T2B utilization without actual intervention and speculated that T2B would reduce time to appropriate therapy or enabled focused therapy in 16/137 ED patients tested (76). Drevinek and colleagues used T2B in 55 samples from 53 patients (77). A total of 36.4% (8/22) of cases of BSI were only detected by T2B, and 9/15 patients with positive T2B results were able to receive earlier/modified target therapy. T2B also reduced time to species identification by 55 h (77). A meta-analysis of studies evaluating T2B/T2C demonstrated no difference in mortality but some reduction in ICU/hospital length of stay (78). Studies using older technologies such as the Septifast have similar findings without assessing impact on outcomes beyond time to appropriate therapy (79). In a meta-analysis of a series of studies primarily using Septifast, D’Onofrio and colleagues commented on the paucity of reports on clinical outcomes. Only 2/25 studies demonstrated reduced hospital/ICU length of stay, and 4/25 showed decreased costs (80).

The studies evaluating T2B performance raise significant questions regarding utilization of NAAT-based rapid assays for BSI. Do the results generate actionable information? Since the targets only cover ~50% of potential BSI pathogens, even if the assay has adequate negative predictive value for the listed targets, the provider may not de-escalate empirical antibiotic coverage. In addition, even with positive T2B results, in the absence of information about the presence of resistance markers, it may be challenging to modify therapy until full culture and susceptibility results are available. For example, patients presenting with BSI with S. aureus will be empirically treated with vancomycin. A positive T2B result in this setting without a mecA result may not immediately change patient management. Interpreting the significance of T2B-positive/blood culture-negative results remains challenging, and the increased positivity rate for the T2B assay may create a number of challenges for health care institutions. It is possible that future panels with a wider spectrum of targets may be able to address some of these concerns.

When assessing the impact of NGS, Rossoff et al. noted that in their cohort of 100 cases, 34 invasive procedures might have potentially been avoided had the NGS result been available (69). A multicenter analysis of routine use of plasma NGS evaluated the impact of the assay in terms of treatment changes, establishing diagnosis or avoidance of invasive diagnostic procedures (81). In total, 50/82 samples had positive NGS results. A positive impact was noted in 6/82 (7.3%), a negative impact in 3/82 (3.7%), and no impact in 71/82 (86.6%) of cases (81). There were significant challenges in the interpretation of positive NGS results. The authors concluded that further work was needed to identify the patients that would benefit from NGS, especially considering the significant cost of testing.

The potential impact of increased sensitivity in sepsis diagnostics can be assessed in studies where sites have transitioned to systems that improved blood culture performance whether culture or NAAT based. Chavez et al. assessed the impact of transitioning their blood culture system from the VersaTrek (Thermo Fisher Scientific, Waltham, MA, USA) to the Virtuo system (bioMérieux, Durham, NC, USA) (82). Others had previously demonstrated that the Virtuo system had a better limit of detection (LOD) and time to detection (TTD) compared to its precursor, the BactT Alert (bioMérieux, Durham, NC, USA) (4). However, the actual extents of the differences in performance were only apparent during real world implementation. Blood culture positivity increased from 8.1% to 11.7% (P < 0.001) (82). This included a significant increase in yield of S. aureus as well as other bloodstream pathogens. There was a concomitant increase in coagulase-negative staphylococcus yield and blood culture contamination rates (1.5 versus 1.9%; P > 0.001) (82). Similar findings were observed by others making the same transition—significant increases in blood culture yield were accompanied by increased blood culture contamination rates (Linoj Samuel and Robert Tibbetts, Henry Ford Health, personal communication). The same group also found that the median duration of S. aureus bacteremia was significantly increased posttransition from 1 to 3 days (11). Patients with longer duration of S. aureus bacteremia were also more likely to receive fluorodeoxyglucose-positron emission tomography/computed tomography (FDG-PET/CT) scans, which were obtained more frequently during the postimplementation period (1% preimplementation versus 5% postimplementation; P = 0.004). Combination antimicrobial therapy was used more frequently in the postimplementation period (6% preimplementation versus 16% postimplementation; P = 0.003) (11). It would have been reasonable to speculate that increased blood culture sensitivity for S. aureus would translate to better patient outcomes, but despite the increased blood culture yield, there was no change in the 90-day all-cause mortality, 90-day all-cause readmissions, or length of stay when comparing preimplementation to postimplementation (11).

When investing in technologies that improve sensitivity and time to result, one cannot assume that this automatically translates to improvements in patient outcomes. Studies evaluating the impact of rapid multiplex detection of pathogens and resistance markers from positive blood cultures have demonstrated improved time to identification and effective/appropriate therapy (7). This comes at the cost of significant investments in capital, labor, and annual reagent costs in the range of hundreds of thousands of dollars. In many of these studies, these improvements have not always translated to improvements in patient outcomes such as mortality, length of stay, 30-day readmission, etc. (7, 83–86). While reduced antimicrobial use is a laudable goal, it remains debatable whether this metric is adequate to justify the investment required for NAAT/NGS. On the contrary, the increased positivity rates anticipated with NAAT/NGS usage may actually drive increased antimicrobial use. Increased positive NAAT results from blood will also lead to increased utilization of additional invasive and noninvasive diagnostic procedures such as CT scans and transesophageal echocardiography (TEE) (11). We know that overutilization of blood cultures is also an ongoing challenge—one that will likely carry over to NAAT-based assays (28). Careful consideration needs to be given to developing effective strategies to utilize novel technologies, especially in the setting of diminishing resources.

Shehadeh and colleagues determined that in order for NAAT-based testing to be cost-effective, it needs to reduce length of stay by 2 to 4 days (87). Many studies evaluating impact of NAAT-based testing are retrospective/speculative analyses and not adequately structured to find statistically significant and clinically relevant outcomes (80). The lack of data hampers our ability to move forward, and coordinated efforts are needed to show that NAAT/NGS can positively impact metrics reflecting improved outcomes and reduced costs of care such as mortality, length of stay, and 30-day readmission.

REGULATORY CONCERNS AND QUALITY METRICS

Efforts to incorporate novel NAAT/NGS-based testing into diagnostic algorithms may also be challenged by the current regulatory climate and quality monitoring systems. Since implementation of NAAT/NGS-based testing will ostensibly increase sensitivity for bloodstream pathogens, it may increase the number of central line-associated bloodstream infections (CLABSIs) reported from the institution (24). CDC guidance specifically mentions T2B/T2C and Karius, Inc., in this regard. CDC does make the allowance that results from non-culture-based methods such as T2B/T2C and Karius may not be counted toward CLABSI rates as long as corresponding blood cultures are drawn at least 2 days before or 1 day after the non-culture-based method (88). Moving forward, it is likely that hospital-onset bacteremia and fungemia that occur on hospital day 4 will also be added as a new reportable metric to the National Healthcare Safety Network (NHSN), similar to methicillin-resistant S. aureus (MRSA) bacteremia (89). CMS metrics need clarity to ensure that hospitals are not disincentivized to adopt these novel technologies.

In an effort to incentivize the adoption of novel technologies for the detection of sepsis pathogens, CMS approved T2B as the first sepsis in vitro diagnostic test that is eligible for New Technology Add-on Payment (NTAP). NTAPs are indicated for situations where the standard medical technology is inadequate and the new method provides substantial benefit that comes at a significant cost or is inadequately reimbursed (90). CMS stated that the “the T2Bacteria Test Panel represents a substantial clinical improvement over existing technologies because it reduces the proportion of patients on inappropriate therapy, thus reducing the rate of subsequent diagnostic or therapeutic intervention as well as length of stay and mortality rates caused by sepsis causing bacterial infections” (91). CMS determined that hospitals would receive an additional $97.50 in reimbursement for T2B usage. It is unclear at this time whether this benefit will be extended to any new NAAT-based sepsis platforms that obtain FDA approval. A product’s NTAP designation is only valid for 3 years (91).

THE WAY FORWARD

For decades now our ability to detect bloodstream pathogens has been hamstrung by the limitations of blood cultures to detect viable organisms. The advent of NAAT/NGS has the potential to significantly improve our diagnostic capability. There are still significant hurdles to the adoption of these technologies. It seems unlikely that either methodology will replace blood cultures in the near future. Laboratories struggling with workforce shortages and declining reimbursements may be slow to adopt these systems unless there are robust data showing the benefits and impact on patient outcomes.

These are some of the key questions that need to be addressed to determine the value proposition

• •Will the NAAT assays be fast and sensitive enough to allow for targeted intervention within the 3-h window specified by the Surviving Sepsis guidelines?
• •Will the NAAT/NGS assays have adequate negative predictive value to allow for de-escalation/cessation of antibiotics, or will the benefits be limited to patients with positive results?
• •What will be the impact of the increased positivity rate that is expected with NAAT/NGS-based testing? Will it lead to consumption of additional diagnostic resources and increased antimicrobial use? In addition, will these interventions translate to meaningful improvements in patient outcomes and reduced costs of care?

Sweeney et al. proposed the following model for adoption of novel sepsis detection methods (61).

1. Screen patients at the point of care using host inflammatory markers to identify those in need of aggressive intervention.

2. Utilize rapid samples to answer NAAT platforms in high-risk patients to identify the pathogen and resistance markers to direct therapy.

3. Use NGS for further investigation of patients with persistent illness refractory to empiric therapy in whom pathogen detection failed by routine or NAAT-based methods.

CONCLUSION

We are on the cusp of a revolution in the detection of pathogens causing BSIs. A pragmatic approach to adoption of NAAT/NGS-based testing is essential to ensure that there is a solid understanding of the performance characteristics of the test methods and the downstream impact on the patient and health care institution.