Keywords sepsis - septic shock - severe sepsis - diagnosis - biomarkers - screening - early
diagnosis - molecular diagnostic techniques - molecular testing
Sepsis represents a medical condition associated with preventable deaths, a high burden
of morbidity, and long-term sequelae. Global epidemiological data have shown that
48.9 million people develop sepsis yearly, and 11 million deaths are attributable
to septic shock worldwide, accounting for almost 20% of all global deaths.[1 ] Consequently, the World Health Organization has urged for actions to improve sepsis
prognosis. Delayed diagnosis and treatment of sepsis have consistently been considered
independent risk factors for the progression of organ dysfunction and death, particularly
in patients with septic shock.[2 ]
[3 ]
[4 ]
[5 ] According to the updated sepsis definition proposed by the last Surviving Sepsis
Campaign (SSC) guidelines in 2021,[6 ] which defines sepsis as life-threatening organ dysfunction caused by a dysregulated
host response to infection, an arsenal of theragnostic tools has been developed to
increase the specificity of sepsis detection.
Protocolized and accurate interventions are time-critical. These include early adequate
empirical antimicrobial therapy, infection source control, and optimal hemodynamic
resuscitation.[5 ]
[6 ] Current challenges in early detection of sepsis before clinical signs develop contribute
to delays in implementing standard-of-care SSC recommendations for the early approach
to sepsis and septic shock.[7 ]
[8 ] In some settings, evidence of the adverse outcomes of late-recognized cases has
been insufficient to perceive sepsis as a medical emergency that requires prompt treatment.
There is a wide variety of other contributing factors or barriers to improving early
diagnosis and treatment of sepsis.[9 ] Some studies have found barriers are often related to the lack of availability of
some resources, such as microbiology laboratory that processes blood cultures and
other microbiological detection tests. Still, advances in quality of care in sepsis
and a better understanding of underlying pathobiological processes leading to organ
dysfunction will aid in developing accurate, fast, and widely available point-of-care
tests. Bedside accurate tools help the development of future quality improvements
for the practical implementation of stand-of-care interventions, which have been consistently
demonstrated to decrease mortality when implemented on time.[5 ]
Over the last few years, sepsis biomarkers and rapid microbiological diagnostic tests
(RDTs) have been considered a paradigm for novel strategies to improve earlier sepsis
detection. Herein, we gathered the best available evidence on this topic. Biomarkers
used for phenotyping, prognosis, and stratification of patients already diagnosed
with sepsis, insights into machine-learning models, and other artificial intelligence
tools are out of the scope of this review.
For this narrative review, we performed a comprehensive literature search in the Cochrane,
PubMed, CINAHL, and Scopus databases from no start date to September 2023. The search
criteria included the following Medical Subject Headings (MeSH) terms: sepsis OR Septic shock OR Severe sepsis AND diagnosis OR biomarkers OR screening OR early diagnosis OR molecular diagnostic techniques OR molecular testing . We reviewed articles written in Spanish and English. We obtained all full-text versions
of the selected manuscripts. The first draft of the manuscript was reviewed and modified
by all authors. All authors approved the final manuscript.
Sepsis Suspicion: Reasoning on a Case-by-Case Basis Is Crucial
Sepsis Suspicion: Reasoning on a Case-by-Case Basis Is Crucial
When considering the pathophysiological events leading to sepsis, clinicians should
acknowledge that clinical signs of sepsis are the ultimate consequence of complex
underlying molecular and inflammatory derangements that culminate in measurable clinical
signs. The main challenges when trying to diagnose sepsis in its early stages using
clinical variables are the ability to differentiate sepsis from infection, the detection
of occult organ dysfunction in the presence of infection, differentiating sepsis from
local organ dysfunction as a consequence of specific infection (e.g., pneumonia),
attributing a new-onset organ dysfunction to sepsis, organ dysfunction as the consequence
of an unrecognized infection, and the variability of sepsis phenotypes (clinical and
biological), which are influenced by recent interventions, and other noninfectious
causes of inflammation with apparently close to similar clinical and biological host
response (e.g., trauma, burns, autoimmune disease, pancreatitis, major surgery, comorbidities,
age, gender, concurrent medications).[10 ]
The most appropriate workflow is the one that first rules out sepsis using objective
data in the context of any infection to manage the patient accordingly. However, this
is a real challenge with the available validated tools, and most cases are classified
as “suspicious of sepsis” after a clinical evaluation. Recognition of sepsis cases
before the occurrence of hypotension requires wise evaluations to prevent further
organ dysfunction, given that a significant proportion of sepsis patients present
subtle clinical signs and appear less sick at the time of presentation. The inadequate
recognition of these cases and delayed treatment are associated with high mortality
rates (up to 25% in some studies) due to the progression of illness to irreversible
organ dysfunction.[11 ] New organ dysfunction or overt inflammatory response in the context of infection
should prompt early evaluation to rule out sepsis. However, specific infectious conditions
may lead to local organ dysfunction without causing a dysregulated systemic host response.
Before discussing the potential biomarkers available for clinical purposes, we should
define “early” when discussing sepsis diagnosis. The literature has no valid and widely
accepted definition of early sepsis. Most studies have considered early sepsis before septic shock develops, for cases in which clinical signs are evident and the
infection is not confirmed, or during the early (reversible) stages of organ dysfunction.
In our view, those considerations are inaccurate and should be considered “sepsis
diagnosis” or, more precisely, late sepsis diagnosis. Some studies define early sepsis
when sepsis-3 criteria are present (infection + sequential organ failure assessment
[SOFA] score ≥ 2), but septic shock is not present yet.[12 ] All tools detecting sepsis after clinical data are present should be considered
diagnostic tools if organ dysfunction is already present, or there is a clinically
evident process possibly linked to infection progressing to organ dysfunction (e.g.,
systemic inflammatory response syndrome [SIRS], SOFA score 0–1). For the purpose of
this review, we will consider prediction of sepsis to any preclinical condition in
which there is an infection in a host in whom different pathophysiological pathways,
particularly immunological status, will irreversibly lead to organ dysfunction if
untreated. All screening tools detecting sepsis in this phase are predictors of sepsis (see [Fig. 1 ]).
Fig. 1 The potential usefulness of currently available biomarkers and rapid microbiological
tests for prediction, early diagnosis, and diagnosis of sepsis. MEWS, Modified Early
Warning Score; NEWS2, National Early Warning Score-2; PCR, polymerase chain reaction;
SIRS, systemic inflammatory response syndrome; SOFA score, Sequential Organ Failure
Assessment score.
Sepsis as a Clinical Syndrome: Delayed Recognition of a Time-Dependent Condition
Sepsis as a Clinical Syndrome: Delayed Recognition of a Time-Dependent Condition
Interestingly, clinical scores are currently recommended as the best widely available
tools in our arsenal for sepsis screening.[6 ] However, studies in prehospital settings have shown that up to one-third of patients
with documented infection who develop sepsis have normal vital signs.[13 ] Sepsis results from complex host interactions and dysregulated response, amplified
by endogenous factors, to a given pathogen. Therefore, recognizing sepsis from parameters
that reflect its clinical consequences can be considered as a delayed strategy for
detection. Clinical scores may not be ideal for sepsis prediction before organ dysfunction
is clinically overt. However, an accurate clinical assessment remains the core strategy
to detect potential sepsis cases in some low-resource settings.[14 ]
The lack of validity of SIRS as a tool for the early detection of sepsis has been
demonstrated. The classic systemic SIRS criteria for diagnosing sepsis focus only
on bedside clinical variables and laboratory parameters. The need for two or more
SIRS criteria excludes 12.5% of sepsis cases with the same organ dysfunction and mortality
risk as cases that fulfill SIRS criteria.[15 ] In the same study, the authors found SIRS criteria failed to define the transition
point in the overall risk of death.
Sepsis-3 criteria have not proven beneficial to decrease overall mortality or to improve
sepsis recognition and screening. Adding lactate, procalcitonin (PCT), or other clinical
variables improves its sensitivity. In the study of Machado et al,[16 ] the authors conducted a prospective study of two cohorts, with mortality as the
primary outcome. They included patients with suspected infection but without sepsis
and patients with sepsis. The predictive accuracy of quick sequential organ failure
assessment (qSOFA) score was assessed, considering the worst values prior to the suspicion
of infection or sepsis. One cohort had 5,460 patients, 78.3% had a qSOFA score ≤1,
and crude mortality was of 14%. The sensitivity of qSOFA score ≥2 for predicting mortality
was 53.9% (95% confidence interval [CI]: 50.3–57.5). The sensitivity was higher for
a qSOFA ≥1 (85%), a qSOFA score ≥1 or lactate ≥2 mmol/L, and SIRS plus organ dysfunction.
The second cohort included 4,711 patients, and 62.3% had a qSOFA score ≤1, and a mortality
rate of 17.3%. In public hospitals, the mortality rate was higher, 39.3%. In a previous
study, approximately one-quarter of infected patients had a qSOFA score ≥2, with 70%
of them having poor outcomes.[17 ] When using sepsis-3 criteria to detect sepsis, patients in the early phase of sepsis
are missed. The SOFA score performs better in diagnosing sepsis later in clinical
stages and predicting intensive care unit (ICU) admission.[12 ]
[16 ]
[18 ]
[19 ]
As previously mentioned, the qSOFA score is far from being a predictive tool, as clinical
repercussions of sepsis should be evident for a positive score. The frequency of patients
having hyperlactatemia who are still normotensive can be as common as 26% of sepsis
cases.[20 ] Different studies have demonstrated that qSOFA is less sensitive than SIRS to identify
organ dysfunction due to sepsis.[21 ]
[22 ]
[23 ] Despite the National Early Warning Score (NEWS; and the updated version NEWS2) and
the Modified Early Warning Score (MEWS) being better tools than qSOFA[24 ]
[25 ] and recommended by the current SSC guidelines,[6 ] clinicians still lack a specific bedside tool to differentiate sepsis from other
conditions in patients with unclear medical history or to predict sepsis in some subsets
of patients prone to develop sepsis in the following hours after infection. Clinical
scores are more useful for predicting mortality in sepsis than early predictors of
the risk for developing sepsis.
In-hospital quality-of-care programs often use automated sepsis screening tools in
electronic health records, which have been studied to detect sepsis early. However,
their accuracy is variable, given that some studies have shown low predictive values
while others show improvements in sepsis care processes.[26 ]
[27 ]
[28 ] There are studies showing no mortality benefits from sepsis screening tools.[29 ]
[30 ]
[31 ] In many settings, improvements in sepsis screening have been made by developing
and implementing performance improvement programs, which have been shown to standardize
and improve the standards of care for the management of sepsis patients. These programs
generally focus on sepsis screening, sepsis bundle performance metrics, health care
staff education and adherence to sepsis bundles, patient outcomes, and actions for
identified opportunities.[9 ]
[32 ]
[33 ]
[34 ]
[35 ]
[36 ] Parameters reflecting the underlying pathophysiology of sepsis are not included
in this type of clinical screening tools.
Improved Understanding of Sepsis Pathobiology for Earlier Detection
Improved Understanding of Sepsis Pathobiology for Earlier Detection
The early diagnosis of sepsis should be based on the early diagnosis of an infection,
along with the identification of a dysregulated response that may subsequently lead
to organ dysfunction.[6 ]
[37 ] Sepsis involves the early activation of pro- and anti-inflammatory molecular responses
and other nonimmunologic pathways triggered by a pathogen (e.g., neuronal, cardiovascular,
metabolic, bioenergetic, autonomic, hormonal, and coagulation) with outstanding prognostic
significance.[38 ] According to this framework, the ideal biomarker should have enough sensitivity
to rule out sepsis early during the triage of suspicious cases presenting to the emergency
department (ED) and enough specificity to differentiate sepsis from other conditions.
Accurate tools that improve clinical judgment are the game changer for improving sepsis
diagnosis, management, and prognosis. Host response biomarkers have been extensively
studied, as they play a critical role in diagnosis, early detection, phenotyping,
risk of organ dysfunction and death, personalized patient management, and antibiotic
stewardship.
Biomarkers for “Early” Sepsis Diagnosis
Biomarkers for “Early” Sepsis Diagnosis
Early diagnosis of sepsis based on biomarkers has evolved to enhance the accuracy
of our clinical assessments. Acceptable reliability of early diagnosis of sepsis using
only clinical scores is not feasible, as they have low sensitivity and specificity
for sepsis detection in the absence of a severe illness or organ dysfunction and have
important limitations to predicting the mortality risk.[39 ] Diagnostic biomarkers should add value and be able to change the pretest probability
and reclassify patients when there is diagnostic uncertainty, increasing specificity
and providing a high negative predictive value. Ideal biomarkers should be able to
detect sepsis even before clinical suspicion (predictive biomarkers), enabling presymptomatic
diagnosis. In real life, most clinicians use a combination of widely available laboratory
biomarkers to diagnose sepsis (e.g., white blood cell and neutrophil count, lactate,
C-reactive protein [CRP]), more than clinical scores; only 36% use the Sepsis-3 definition
alone, 34.2% still calculate the qSOFA, and 44.7% use the SOFA score.[40 ]
PCT has been extensively studied as a diagnostic tool for sepsis. Three meta-analyses
evaluating the diagnostic utility of PCT reported a pooled sensitivity and specificity
of 77 to 85% and 75 to 83%, respectively.[41 ]
[42 ]
[43 ] Of note, most studies reporting a lack of PCT usefulness for sepsis diagnosis have
included patients with a low pretest probability for sepsis of bacterial origin, and
international guidelines do not support the use of PCT to initiate antibiotics in
sepsis.[6 ]
[44 ]
[45 ] PCT is thought to be more accurate than CRP for detecting patients with suspected
sepsis; however, studies have shown PCT is not beneficial to early diagnose sepsis
cases with a less severe clinical condition.
Illness severity and pretest probability for sepsis influence the usefulness and cut-off
of PCT as a diagnostic tool.[46 ] A reliable cut-off value of 1.1 ng/mL with sensitivity and specificity of 77% and
79%, respectively (area under the receiver operating characteristic curve [AUROC]
of 0.85, 95% CI: 0.81–0.88) can be used to support sepsis diagnosis,[43 ] depending on pretest probability, the presence of clinical criteria, and severity
of illness.[12 ]
[18 ] In an interesting retrospective study by Kim et al, PCT was a useful biomarker for
sepsis and septic shock diagnosis in the ED when used in patients who fulfilled sepsis-3
criteria.[12 ] In other words, it was useful to enhance the diagnosis of sepsis when clinical repercussions
and organ dysfunction are already established, with a proposed cut-off of 0.41 ng/dL
for sepsis (sensitivity and specificity of 74.8% and 63.8%, respectively; AUROC: 0.745),
and 4.7 ng/dL for septic shock (sensitivity and specificity of 66.1% and 79.0%, respectively;
AUROC: 0.784). The lack of effectiveness of PCT to rule out or predict early sepsis has been recognized, and the current SSC guidelines do not recommend
its use to start antibiotics.[6 ]
Consequently, early diagnosis or prediction of sepsis using PCT is unreliable.[12 ] The less severe the condition, the less likely sepsis will be diagnosed early before
overt clinical signs or organ dysfunction develop. Previous studies on this matter
have assessed PCT usefulness compared to clinical criteria as the gold standard.[47 ] International guidelines do not recommend using PCT in ventilator-associated pneumonia,
a common condition related to sepsis in critically ill patients.[48 ]
[49 ]
[50 ] There is no agreed PCT cut-off value for diagnosis of infection regardless of the
presence of sepsis; some studies used PCT values from 0.5 to 2 μg/L, as previous studies
of community-acquired pneumonia (CAP).[51 ] A recent meta-analysis of patients with diverse etiologies of CAP showed that PCT
has low sensitivity during early CAP and cannot reliably distinguish viral from bacterial
pneumonia.[52 ] A previous study on PCT kinetics in patients with bacteriemia showed poor diagnostic
accuracy and a low PCT reliability to guide the initiation of therapy.[53 ] Moreover, PCT is not specific to sepsis; it increases in other conditions often
confused with sepsis, such as trauma, pancreatitis, or autoimmune disease.[46 ]
[54 ] The most widely accepted applicability of PCT in the context of sepsis is antimicrobial
stewardship and prognosis assessment.[46 ]
[55 ]
[56 ]
[57 ]
Various individual biomarkers are developed to enhance the clinical diagnosis of sepsis.
In a recent meta-analysis, soluble urokinase plasminogen activator receptor (suPAR)
was observed to have an AUROC of 0.83 for predicting sepsis (95% CI: 0.80–0.86).[58 ] In addition, AUROC for differentiating sepsis from non-sepsis SIRS was 0.81 (95%
CI: 0.77–0.84), and the sensitivity and specificity were 0.67 (95% CI: 0.58–0.76)
and 0.82 (95% CI: 0.73–0.88), respectively. Soluble triggering receptor expressed
on myeloid cells (sTREM-1) is expressed in innate immune cells (e.g., monocytes and
neutrophils). This protein reflects important processes of the inflammatory and cytotoxic
response to sepsis, such as the synergic activation of Toll-like receptors and the
augmented production of pro-inflammatory cytokines.[59 ] Serum levels of sTREM-1 have been studied as a biomarker for early sepsis.[60 ] Previous studies have shown an AUROC of 0.78 (95% CI: 0.69–0.86) to differentiate
sepsis from other causes of SIRS,[61 ] and an AUROC of 0.95 for septic shock diagnosis. In a study of 90 patients with
SIRS due to sepsis and other etiologies,[61 ] a PCT cut-off value of 1.57 ng/mL and sTREM-1 cut-off value ≥133 pg/mL yielded a
sensitivity of 71.1 and 67.33%, and specificity of 73.3 and 65.79%, respectively,
for the differentiation of sepsis from other causes of SIRS.
Biomarker-enhanced clinical scores may improve specificity of diagnosis, though sensitivity
remained low.[12 ] Other biomarkers have been more beneficial in predicting prognosis in sepsis, such
as pro-MR-adrenomedullin.[62 ]
[63 ]
[64 ]
Sepsis often presents a hyperinflammatory response pattern followed by an immunosuppressive
state, during which multiple organ dysfunction develops.[65 ]
[66 ] A biomarker or a combination of biomarkers could be a new alternative to predict,
identify, or provide new approaches to manage sepsis patients. In some settings, the
combination of biomarkers has been used as a strategy to increase the sensitivity
for early diagnosis and improved outcomes.[63 ]
[67 ] The combination of two or more biomarkers increases the diagnostic accuracy of sepsis
diagnosis in some studies.[68 ] Seeking more accurate therapeutic interventions and patient outcomes in this condition
should be the goal of any combination of biomarkers.
Still, association of different biomarkers reflecting the same pathophysiological
pathway may have no added value in terms of diagnostic accuracy. An important study
of ICU patients with SIRS showed no combination of biomarkers performed better than
CRP alone to diagnose sepsis.[69 ] Increased costs, complexities in interpreting results, lack of validation studies,
and inadequate training in the obtention and implementation in different settings
are other disadvantages of combining biomarkers. Standardization of sample collection,
analysis, and processing are needed for their reliability regardless of the laboratory
performing the tests. Combining point-of-care inflammatory biomarkers would solve
all those issues related to the usual measurement of biomarkers. This innovative strategy
needs to be further validated in clinical studies.[70 ]
Machine learning tools and biomarker-enhanced scores that involve the combination
of laboratory and clinical biomarkers have been overwhelming in recent years. Machine-learning
models using artificial intelligence have been studied over the last few years to
improve the usefulness of clinical and laboratory biomarkers by combining them for
early sepsis detection.[30 ]
[31 ]
[71 ] The performance of these models has been variable, and some limitations have been
identified due to the lack of availability of some biomarkers or clinical measurements.
Electronic alerts are more useful in emergency settings to reduce hospital length
of stay, improve time to treatment, and reduce mortality, though sometimes they are
poorly generalizable.[72 ]
Prediction of Sepsis: Detecting Occult Processes Leading to Sepsis and Organ Dysfunction
Prediction of Sepsis: Detecting Occult Processes Leading to Sepsis and Organ Dysfunction
As we have discussed before, even machine learning models that use clinical variables
and relevant host factors with characteristics that progress over time are not sufficiently
accurate to diagnose early sepsis, as they rely on clinical consequences and common
laboratory tests resulting from underlying molecular derangements leading to an aberrant
or dysregulated host response and organ dysfunction. The logical pathway would be
to find preclinical biomarkers of systems that accurately predict the risk of sepsis
once the infection is established (or before) and combine microbiological and inflammatory
panels. This review will not discuss biomarkers that have been studied as predictors
of organ dysfunction in sepsis and increased mortality.
Novel Molecular Biomarkers for Prediction or Early Diagnosis
Novel Molecular Biomarkers for Prediction or Early Diagnosis
Extensive research in the field of biomarkers is being performed to validate new molecules
detecting sepsis underlying processes at early stages, with the intention to facilitate
effective sepsis prediction at the time of infection, allowing for preventive rather
than early interventions and ultimately reducing the number of deaths. Interesting
studies on earlier biomarkers, including serial measurements of pancreatic stone protein,
demonstrated an increase of this marker 3 days preceding the onset of signs necessary
to diagnose sepsis clinically.[73 ] As discussed above, some studies propose using panels of biomarkers to predict or
diagnose sepsis early as the most pragmatic strategy, so far, to improve clinical
diagnosis of sepsis.[68 ]
[Table 1 ] gathers a summary of novel predictive biomarkers in sepsis; PCT was added to the
table as a comparator.
Table 1
Brief summary of potentially applicable biomarkers for sepsis prediction or early
diagnosis
Biomarker
Clinical applicability
PCT
Classic biomarker, not useful for sepsis prediction or early diagnosis of sepsis.
Diagnosis of bacterial sepsis or infection. More accurate in more severe illness.[61 ]
PCT cut-off for sepsis, 1.1 ng/mL[43 ]; 1.57 ng/mL[61 ]
• ↑ Concentrations in patients with sepsis and infection[77 ]
• Distinction between patients with sepsis and patients without sepsis in the ICU
↑ values in septic shock, sepsis, and controls (17.1, 1.8, and 0.04 ng/mL, respectively)[78 ]
sTREM-1
Sepsis indicator.[61 ]
[79 ]
[80 ] An early distinction between sepsis and SIRS predictive of septic shock.
Pancreatic stone protein (PSP)
C-type lectin protein that triggers polymorphonuclear cell activation. Serial measurements
are potentially useful to predict sepsis 3 days before clinical diagnosis.[73 ]
sPD-L1
Indicates sepsis-associated immunoparalysis (immunosuppression)[81 ]
[82 ]
Cut-off of 0.16 ng/mL, ↑ sPD-L1 immunosuppression phenotype.[82 ]
IL-10
Levels correlate with the hypoinflammatory phenotype.[82 ]
[83 ]
IL-1β and IL-6
Levels increase in the acute phase of sepsis.[84 ]
[85 ]
Pentraxin-3
Predicts the risk of sepsis in patients with suspected infection in the emergency
department.[86 ]
Sepsis versus SIRS.[87 ]
Calprotectin
Better distinction between sepsis versus nonsepsis patients in the ICU than PCT. Distinction
between sepsis and trauma patients.[88 ]
Bio-adrenomedullin
Useful to distinguish sepsis, septic shock, and nonsepsis patients (74, 107, and 29
pg/mL, respectively).[89 ]
Resistin (and eNamp)
Early sepsis biomarkers.[90 ]
[91 ]
suPAR
Risk of patients with suspected infection.[92 ]
LDL-C
Protective effect against sepsis.[93 ] Low values can reflect a risk of sepsis and admission to the ICU.
Risk of sepsis (OR: 0.86) and admission to the ICU (OR: 0.85). The lower quartile
had a greater risk of sepsis (OR: 1.48) and admission to the ICU (OR: 1.45) vs. the
highest quartile, considering other comorbidities.
Presepsin
Plasma levels are considered a biomarker of the activation of innate immune effector
cells in response to invasive organisms. Biomarker of phagocytosis.[94 ]
[95 ]
High accuracy (AUROC 0.954) for prediction of sepsis risk, an early diagnosis.[96 ]
[97 ] ↑ Presepsin in sepsis patients compared to nonsepsis SIRS.
CD64
High-affinity Fcγ receptor I in neutrophils upregulated in the early stages of activation
of the innate immune response. AUROC 0.879 of nCD64 for diagnosis of bacterial infection.[78 ]
↑CD68
Increased in the hippocampus, putamen, and cerebellum in patients with sepsis.[98 ]
VLA-3 (a3β1)
Indicative of sepsis.[99 ]
[100 ] Discrimination of sepsis and SIRS.
Increased α3β1 (VLA-3, CD49c/CD29) on neutrophils of septic patients. ↑ β1 (CD29),
on neutrophils of sepsis patients.[100 ]
↑ sTNFR-1
Distinguish sepsis from nonsepsis SIRS.[101 ]
↓ miR-125
Good predictive values for sepsis risk.[102 ]
↑lnc-ANRIL/miR-125a axis
Determine the risk for sepsis.[103 ]
miR-125a and miR-125b
Useful to distinguish sepsis from other SIRS states.[103 ]
↑ Lnc-MALAT1/miR-125a
Increased levels in sepsis and risk of sepsis.[76 ]
Lnc-MALAT1/miRNA-125a
Discriminates sepsis patients from healthy controls. Reflects inflammation level.[76 ]
lnc-MEG3
Increased values are predictive of sepsis risk. Lnc-MEG3 is a potential biomarker
for the prediction of sepsis via interacting with miR-21.[104 ]
Genetic polymorphisms
The expressions of inflammatory mediators, microRNA expression, and other mechanisms
have been described as a tool for predicting sepsis responses in infected patients.[74 ]
Abbreviations: ICU, intensive care unit; OR, odds ratio; PCT, procalcitonin; SIRS,
systemic inflammatory response syndrome.
Advances in the understanding of the genetic basis for sepsis activation of the innate
immune response,[74 ] the release of acute phase reactants, knowledge of biomarkers involved in the pathophysiology
of sepsis, and the serum levels of glycoproteins on cell membranes have allowed for
the study of different molecules and genes encoding those molecules (e.g., proteins,
cytokines, soluble receptors, chemokines) as sepsis-predictive biomarkers. Of note,
newer potentially predictive biomarkers have been validated in comparison with the
gold standard for screening in sepsis (clinical scores), though others have been studied
prospectively as predictors of sepsis risk, which correlate with mortality risk.[75 ]
More recent advances in gene expression and transcriptomics have led to the identification
of new classes of biomarkers, such as microRNAs, long-noncoding RNAs, or the human
microbiome. Noncoding RNAs have been studied as early predictive sepsis biomarkers.
The expression of the Lnc-MALAT1/miR-125a axis discriminates between sepsis patients
and healthy controls and is associated with an excellent diagnostic yield (AUROC of
0.931, 95% CI: 0.908–0.954).[76 ]
Given that a significant proportion of patients with early sepsis do not show clinical
signs but do develop an immunopathogenic phenotype leading to dysregulated organ dysfunction
and increased mortality, the most sophisticated prediction models have proposed the
use of clinical parameters with a panel of genes encoding inflammatory biomarkers.[105 ] The most important disadvantages of these models are the high cost and difficulties
in sample processing, laboratory testing, and lack of availability for all hospital
(or prehospital) settings. Predictive biomarkers have been studied compared to clinical
scores rather than in prospective cohorts of infected patients and their clinical
trajectories.
Novel technologies are poorly affordable in middle- or low-resource settings, which
account for 85% of sepsis cases.[1 ] Their lack of validity in prehospital settings or ED is outstanding. In such settings,
an objective and quick tool is highly needed for the early triage of patients. Important
studies exist on the potential immune response biomarkers for the prediction of sepsis.
However, they have been performed preferably in hospitalized patients or later in
the ICU.[82 ]
[88 ]
Heterogeneity in critically ill patients with sepsis involves a new paradigm with
clinical applications, as it has contributed to the challenging task of finding a
perfect combination of biomarkers. Novel genetic studies may enable better characterization
of different panels of biomarkers at the time of a specific infection to predict the
risk of sepsis. Identifying unique biological signatures in patients could enhance
selected enrollment in clinical trials and strengthen our diagnosis and early detection
approaches.[106 ] Most importantly, the clinical applicability of new discoveries is a sine qua non to revolutionize sepsis management and reduce deaths.
Rapid Microbiological Diagnosis as an Element for Sepsis Prediction
Rapid Microbiological Diagnosis as an Element for Sepsis Prediction
Early identification of causative microorganisms in suspected sepsis is needed to
optimize antimicrobial use and patient survival. However, current culture-based pathogen
identification often takes at least 24 to 48 hours to give meaningful results, weakening
their usefulness in decision-making to start antimicrobial treatment, thus, broad-spectrum
antibiotics are often used to ensure coverage of all potential organisms, implying
risks of overtreatment, toxicity, and selection of multidrug-resistant bacteria. Furthermore,
previous or current antimicrobial treatment decreases these tests' sensitivity. Empirical
broad-spectrum antimicrobial treatment leads to overtreatment in 60 to 70% of conditions
that mimic sepsis, such as other inflammatory states, or secondary to less severe
viral or susceptible bacterial infections.[107 ]
The clinical need for a faster microbiological approach to target treatments early
in the course of infections potentiated the therapeutic advantages of new microbiological
technologies, such as RDT.[108 ] Pathogen molecular diagnostic tests speed up the time to identification of pathogens
and their susceptibility to antibiotic and eventually targeted treatment.[109 ] There is a lack of evidence on the clinical impact of RDT in sepsis patients. Most
data have been extracted from studies performed in infections, such as bacteremia
or pneumonia, that could lead to sepsis.
Previous studies of matrix-associated laser desorption/ionization time-of-flight mass
spectrometry (MALDI-TOF MS) have been successful in the management of bloodstream
infections.[110 ]
[111 ]
[112 ] A previous meta-analysis and other studies showed that antimicrobial stewardship
programs are associated with reduced mortality, time to optimal treatment and length
of stay, and are cost-effective.[112 ]
[113 ]
[114 ] Patients with sepsis and gram-negative bacteremia may benefit from RDT due to the
wide range of possible infecting pathogens and the implications of inappropriate treatment
in the context of drug resistance. Previous studies of patients with drug-resistant
gram-negative bacteremia have shown earlier initiation of appropriate therapy, shortened
length of stay, and reduced 30-day mortality.[115 ] MALDI-TOF MS has been studied for rapid identification of antimicrobial susceptibility;
however, some misclassifications have been observed, and the accuracy of this method
needs to be improved.[108 ]
[116 ] There is a lack of studies evaluating MALDI-TOF MS in sepsis. In the study of Verroken
et al,[117 ] the authors assessed the impact of MALDI-TOF MS results in the management workflow
of antimicrobial stewardship in sepsis patients with positive blood cultures to Enterobacteriaceae , Pseudomonas aeruginosa, and Staphylococcus aureus . The mean time to pathogen identification was reduced by 61 to 65% (10.8 hours).
The mean time to optimal treatment was decreased significantly. The impact on mortality
was not assessed.
Multiplex polymerase chain reaction (PCR) has been previously studied for the rapid
identification of S. aureus and its resistance patterns. The FilmArray Blood Culture ID Panel (BCID), which can
identify 24 different bacteria, fungi, and common antimicrobial resistance genes (KPC,
mecA, and vanA/B) within 1 hour of organism growth in blood cultures, was evaluated
in the randomized Blood Culture Identification trial.[118 ] In this study, the molecular technique reduced the time to targeted treatment, decreased
the use of broad-spectrum antibiotics, and contributed to antimicrobial de-escalation.
There is a paucity of evidence on gram-negative pathogen identification in sepsis
using PCR.[118 ]
[119 ]
[120 ]
[121 ]
[122 ] In the study of Vincent et al,[123 ] the use of culture-independent PCR/electrospray ionization-mass spectrometry technology
resulted in rapid pathogen identification in critically ill patients. The authors
tested different sources of infection (e.g., pneumonia: 185 cases, blood stream: 616
cases, sterile fluid: 110 cases, and tissue infection: 529 cases) in critically ill
patients. The study reported the effectiveness of PCR to rule out infection within
6 hours compared with standard culture-based microbiological testing, with a sensitivity
of 81%, a specificity of 69%, and a negative predictive value of 97%. In a study of
617 patients with positive Gram stains in blood cultures, BCID resulted in faster
pathogen identification than standard blood cultures and usual susceptibility testing,
which improved antimicrobial de-escalation. The T2Bacteria Panel (including the identification
of Escherichia coli , Klebsiella pneumoniae , Pseudomonas aeruginosa , Enterococcus faecium, and S. aureus ), identified the causative pathogen in whole blood samples at a mean of 3.6 to 7.7 hours
compared with almost 72 hours with standard blood cultures.[122 ]
Ideally, RDT should provide pathogen species and data on antimicrobial susceptibility,
such as Accelerate Pheno system (APS; Accelerate Diagnostics, Denver, CO), an automated
system that reduces the time to pathogen identification and gives susceptibility data
(at 27 and 40 hours, respectively) compared with conventional cultures.[124 ] This system has been approved by the Food and Drug Administration in 2017 for testing
in blood.
There are some drawbacks regarding the use of RDT. These tests are not specific to
sepsis and are not useful for making a differential diagnosis between three conditions: colonization, infection,
and sepsis. Likewise, RDTs have led to overdiagnosis and overtreatment.[125 ] Data on clinical benefit and cost-effectiveness are still emerging. Costs and microbiology
lab expertise in molecular techniques are also seen as a barrier to their widespread
use, particularly in low-resource settings. None of these technologies have approached
the point of care, nor can they be described as genuinely culture-independent diagnostic
tests. Evidence on their effectiveness in improving mortality is conflicting and should
be further studied.[121 ] A recent systematic review of RDT in sepsis[126 ] reported improvements in appropriate antimicrobial therapy, nonsignificant change
in time to targeted therapy, decreased length of stay in two studies, and a significant
decrease in antimicrobial cost in six studies. The impact on mortality was unclear.
This study has important limitations on the number of studies included and high heterogeneity.
RDTs per se are not useful for diagnosing a specific immunopathogenic state that predisposes
patients to a significant risk of death, such as sepsis. More specific biomarkers
recently identified reflect the immunopathologic state leading to sepsis, which is
triggered by the interaction of infectious agents and the innate immune system.
Inflammatory biomarker-enhanced RDT will aid in very early diagnosis or prediction of sepsis before overt clinical consequences, differentiating sepsis from other acute
inflammatory conditions, identifying and quantifying the causative organism, determining
resistance patterns early to target treatments from time zero, improving antimicrobial
stewardship practices, and monitoring patient progression. Combining point-of-care
RDT tests and more specific inflammatory biomarkers is a novel strategy to enhance
biomarkers' availability and affordability for earlier sepsis diagnosis in ED. This
can improve time to diagnosis (up to 10 times faster when compared with the gold standard),[73 ] faster detection of pathogens,[127 ] quick resistance profiles, and detection and rapid monitoring of specific biomarkers.
Precision medicine has developed tools to identify new cases, predict prognosis, and
target treatments according to their clinical and molecular phenotypes.[128 ]
[129 ] Multiplex point-of-care devices and other theragnostic approaches are integrated
approaches that gather data for early diagnosis and classification of sepsis (e.g.,
inflammatory and organ dysfunction biomarkers and microbiological diagnosis).[130 ]
Monitoring different biomarkers gives a holistic view of patients' clinical status
and prognosis. Integrated point-of-care biomarkers are promising for democratizing
novel theragnostic tools and developing precision medicine elsewhere. To improve their
applicability in different settings, further clinical studies assessing the effectiveness
of these innovative techniques are needed. Widely available and affordable combinations
of RDT and predictive biomarkers (e.g., predictive biomarker-enhanced RDT point-of-care
tests) should be further studied and clinically validated and promise to be the game
changer in sepsis diagnosis.
Conclusion
Early sepsis prediction is still in its first stages, and it remains a complex field
for clinicians and researchers. In recent years, an increasing interest has evolved
in techniques to improve sepsis definition, prediction, early diagnosis, classification
of patients, defining prognosis, and personalizing treatment. Novel developments and
deep study of point-of-care biomarkers have been promising to enhance the accuracy
of near-patient diagnoses. The continuous developments of point-of-care tools using
widely applicable and affordable combinations of biomarkers and faster techniques
for accurate microbiological information have driven new insights for sepsis management.
