Introduction

The following review aims to provide clinicians working at the bedside of critically ill patients with practical and rational instructions for utilizing the technological advancements available in modern medicine. Our insights are based on personal experience, and we recognize that mastering the rational use of biomarkers in the diagnostic workup of sepsis requires years of practice. Historically, the earliest studies on procalcitonin (PCT) appeared in the 1980s, yet it was only after 2000 that limitations of this marker in routine clinical application became more clearly understood.

A precise definition of a biomarker is essential for discussing laboratory parameters. A biomarker is a measurable indicator of a patient’s clinical condition that can be accurately and reproducibly measured. Biomarkers are valuable for diagnosis, prognosis, early disease recognition, risk stratification, appropriate treatment (theranostics), and monitoring responses to treatments. Moreover, they are useful in clinical trials involving patients with suspected sepsis. A biomarker, or a combination of biomarkers within a panel, should offer both high specificity and sensitivity for diagnosing a disease or syndrome. However, even if a biomarker demonstrates only one of these attributes, it may still be sufficient for use as a “rule‑in” or “rule‑out” test (Table 1). Valid biomarkers are invaluable tools for clinicians, aiding in both diagnostic accuracy and therapeutic decision making. They can help identify patients who are suitable for specific interventions and adjust type or duration of treatment.¹

Sepsis is the leading cause of death in noncoronary intensive care units (ICUs). It is a serious disease with high mortality that places a substantial financial burden on health care systems. Over the past 2 decades, the incidence of sepsis has risen in low‑income countries as well as in the United States and across western and central Europe.^2,3 According to the latest available data from 2017, the global annual incidence of sepsis was approximately 48.9 million cases, resulting in 11 million deaths.⁴

In 2017, the largest contributors to sepsis incidence and mortality across all age groups were diarrheal diseases and lower respiratory infections, respectively. Complications of injuries and noncommunicable diseases accounted for nearly half of all sepsis‑related deaths. Alarmingly, 41% (20 million) of global sepsis cases occurred in children under the age of 5 years.

In 2017, member states of the World Health Organization recognized the urgent need for action and prioritized improving the prevention, diagnosis, and clinical management of sepsis. Subsequently, the 2021 Surviving Sepsis Campaign Guidelines introduced evidence‑based recommendations for managing adult patients with sepsis and septic shock.⁵ Early and accurate sepsis diagnosis is critical for improving treatment outcomes, reducing the financial burden of care, and enhancing patient long‑term quality of life.

Clinical diagnosis of sepsis

Sepsis is an acute, life‑threatening, immunological response to an infectious process that leads to multiorgan dysfunction and death. The Sepsis‑3 definition⁶ characterizes sepsis as a life‑threatening organ dysfunction caused by a dysregulated host response to infection.

Sepsis is a dynamic clinical syndrome with numerous, diverse, and variable symptoms and signs. Several screening tools based on clinical parameters have been shown to increase the rate of sepsis detection.⁷ Early recognition in the emergency department relies on key warning signs, including severe breathlessness, fever with chills and rigor, slurred speech, delayed responses, drowsiness, hallucinations, confusion or agitation, sudden muscle weakness, muscle pain or shivering, nausea and vomiting, absence of urine output, headache, and abdominal pain.⁸

Historically, the Systemic Inflammatory Response Syndrome (SIRS) criteria and the National Early Warning Score (NEWS) were the most widely used tools for identifying sepsis.⁹ However, the Quick Sequential Organ Failure Assessment (qSOFA) has not been positively validated as a standalone screening tool and has been excluded from the 2021 Surviving Sepsis Campaign guidelines.^5,10

Several other scoring systems have been evaluated for sepsis screening in prehospital settings and emergency medicine^10,11: 1) the BAS 90‑30‑90 score that includes systolic blood pressure below 90 mm Hg, respiratory rate over 30 breaths / minute, and pulse oxygen saturation (SpO₂) lower than 91%; 2) the Pre‑hospital Early Sepsis Detection (PRESEP) score, in which patients receive points for specific clinical signs, including fever greater than 38.3 ºC (4 points), chills and rigor (4 points), respiratory rate above 22 breaths/minute (2 points), severe hypotension (systolic blood pressure <⁠90 mm Hg) (2 points), SpO₂ below 92% (2 points), and heart rate greater than or equal to 90 bpm (2 points)⁷; 3) the Modified Early Warning Score (MEWS) assessing mental status, systemic blood pressure, heart rate / pulse / electrocardiography, core body temperature, respiratory rate, and SpO₂, with each parameter scored according to the severity of physiological deterioration¹¹; and 4) the SIRS criteria proposed by R. Bone in 1992 that evaluate 4 signs of a decline in physiological function: temperature (fever or hypothermia); tachypnea (respiratory rate >22 breaths/minute), tachycardia (>90 bpm); and leukocytosis (>12 000/µl) or leukopenia. Additional signs of hypoperfusion, such as hypotension (systolic blood pressure <⁠100 mm Hg), altered mental status, oliguria, and metabolic acidosis with hyperlactatemia are also included in this assessment.¹²

The sensitivity of these scoring systems ranges from 42% to 84%, and their specificity varies between 54% and 82%. Among these, the MEWS, the SIRS criteria, and the PRESEP score appear to be the most effective tools for identifying sepsis in adults, with sensitivity varying from 45% to 65% and specificity from 54% to 82%.^9-11 The efficacy of the existing scoring systems for identifying sepsis may be markedly improved by incorporating additional blood markers of inflammation (C‑reactive protein [CRP], neutrophil‑to‑lymphocyte ratio [NLR], PCT), shock (N‑terminal pro–B‑type natriuretic peptide [NT‑proBNP]), and metabolic acidosis (pH, lactate).¹³

Laboratory‑based diagnosis of sepsis

The accurate diagnosis of sepsis relies on 4 fundamental pillars: 1) clinical signs and symptoms (evaluated by qSOFA, SIRS criteria, and MEWS), 2) microbiological investigation (hemocultures), 3) imaging methods (X‑ray, ultrasonography, computed tomography, magnetic resonance imaging), and 4) laboratory biomarker testing (Figure 1). Currently, there is no “magic bullet” for diagnosing sepsis, and a comprehensive clinical evaluation remains the gold standard. However, numerous markers across various laboratory disciplines are used to varying degrees in the diagnosis of sepsis. An ideal biomarker should have a negative predictive value and sensitivity of 100% alongside a positive predictive value and specificity exceeding 85%.

Studies have shown that each hour of delay in diagnosing sepsis translates into higher rates of patient mortality. Furthermore, inappropriate antibiotic therapy, whether due to delays in administration or inadequacy of treatment, significantly worsens the prognosis in patients with septic shock.¹⁴

The laboratory diagnosis of sepsis involves assessing a comprehensive set of measures spanning clinical biochemistry, hematology, immunology, microbiology, and molecular biology. Effective diagnostic workup of an infection involves identifying the site of the infection, isolating the causative microorganism, and determining its antibiotic susceptibility. Additionally, it is crucial to assess the severity of the patient’s clinical condition, as indicated by their inflammatory response, and to evaluate their prognosis.

Although hundreds of biomarkers show diagnostic and prognostic potential in sepsis (Table 1), their routine clinical application remains limited. An ideal biomarker should demonstrate high sensitivity and specificity, be suitable for bedside use, and remain cost‑effective. However, most biomarkers in sepsis reflect the dysfunction in specific organs or systems rather than directly indicating sepsis itself. Consequently, the laboratory diagnosis of sepsis is a mosaic of diverse technological and methodological approaches that must be interpreted within the context of the patient’s clinical picture.

In patients with SIRS, we seek answers to the following questions: 1) Is the inflammation infectious or noninfectious in origin? If the etiology is infectious, then is it viral or bacterial? Both infectious and noninfectious causes can trigger a cascade of inflammatory responses, potentially leading to multiorgan failure. 2) Can the measured biomarker provide prognostic insights? 3) Will the biomarker’s measured level influence the therapy or predict treatment success?

The following overview describes the tests currently used in routine clinical practice in our hospitals. Numerous biomarkers are available for the diagnostic workup of sepsis.¹⁵ The selection of biomarkers listed below is based on our experiences over the past 30 years. Omics technologies represent the future of the diagnostic workup of sepsis. In current practice, generally, the most critical factor for these biomarkers is not their absolute value but rather the trend reflecting how values change over the course of sepsis (Figure 2).

Acute phase proteins

In patients with sepsis, acute phase proteins (APPs) are secreted in large quantities as part of an innate humoral immune response. The synthesis of positive APPs is induced and stimulated by proinflammatory cytokines, such as tumor necrosis factor α, interleukin (IL)-1, and IL‑6. A common limitation of APPs as biomarkers for sepsis is their insufficient specificity and sensitivity. However, CRP, PCT, and presepsin are well‑established markers routinely used in the diagnosis of sepsis.

Indexes and ratios of blood biomarkers for sepsis

Sepsis is associated with the activation and imbalance of multiple physiological systems and its presentation varies greatly due to numerous factors. These include the etiology of infection (eg, viruses, bacteria, fungi, or parasites), host genotype and phenotype, the physiologic reserve of organ systems, comorbidities, and other unpredictable factors. The dynamic course of sepsis induces characteristic changes in biomarker levels, receptor expression, the secretion of cytokines, hormones, and proteins, and the types of blood cell subpopulations—all of which can be quantified.

Each biomarker used in sepsis has its own validity, sensitivity, specificity, and odds ratio for specific cohorts of patients. However, each biomarker also has limitations associated with its biological properties. These limitations can be mitigated by combining biomarkers into indexes or ratios, which improve diagnostic accuracy, validity, and predictive value, providing clinicians with improved tools for decision making.

Ratios and indexes reflect the biological and physiological changes occurring throughout the course of sepsis. Notably, the biomarkers used to construct these ratios must have a clear biological relationship and display opposing dynamics (eg, peak versus nadir). For example, APPs can be positive or negative, and coagulation parameters in sepsis show increased D‑dimer levels along with decreased activities of antithrombin and protein C due to their consumption.

The complete blood count and white blood cell differential count provide valuable information for the diagnosis of sepsis.^35-37 Over 20 years ago, we suggested the NLR as a new inflammatory marker for sepsis and various severe clinical events.³⁸ The NLR is calculated as the ratio of the absolute or relative count (%) of neutrophils to lymphocytes in the peripheral blood. This ratio reflects characteristic changes in these cell subpopulations during inflammation, infection, cancer, and neuroendocrine stress, specifically neutrophilia and lymphocytopenia.

In healthy adults, NLR values range from 1.1 to 2.2. An NLR exceeding 3 indicates severe pathology in cardiovascular and cancer‑related inflammation, whereas values above 5 suggest bacterial infection. NLR values between 7 and 17 are indicative of bacteremia and sepsis. Initial and dynamic changes in NLR have predictive and partially prognostic value in patients with sepsis. Very high NLR values of at least 17 and, paradoxically, low values of less than 7 in the early phase are associated with poor outcomes and increased mortality.³⁹

The NLR is a valuable marker of systemic inflammation, aiding in accurate diagnosis in subsequent days. It can predict a patient’s response to therapy and clinical outcome. The utility of NLR in diagnosing sepsis was confirmed by Ljungström et al,¹³ who found that the area under the curve for NLR to predict positive blood cultures was 0.71. This aligns with a previous study by de Jager,⁴⁰ who reported an area under the curve of 0.73 to 0.77 for NLR.

Other biomarkers used in various indexes for sepsis include leukocyte, monocyte, eosinophil, lymphocyte, and platelet counts. These indexes are the platelet‑to‑lymphocyte ratio, the systemic inflammation response index (SIRI; calculated as NLR  ×  monocyte count⁴¹), and the systemic immune–inflammation index (SII; calculated as platelet count  ×  NLR⁴²). In the 2 latter indexes, the NLR is combined with the monocyte count or the platelet count.

The clinical relevance and utility of various biomarkers for diagnostics and prognostication in patients with sepsis have been extensively reviewed in recent reports.^36,37 Tracking the dynamic changes in blood parameters and ratios during the course of sepsis is crucial for predicting patient outcomes. Progressive increases in the NLR, platelet‑to‑lymphocyte ratio, and CRP‑to‑lymphocyte count ratio (CLR) are associated with poor outcomes.^43-45 Observations of biomarker dynamics show that changes in concentrations or counts often have greater predictive value than static measurements. The established cutoff values for commonly used biomarkers are summarized in Table 2.

Bone marrow is strongly activated during invasive infection and sepsis, leading to the release of large numbers of immature blood cells into circulation. In the early phase of sepsis, automated hematology analyzers are used to measure the width distributions of monocytes, platelets, and red blood cells.^35,43 These analyzers can also detect immature red blood cells (nucleated red blood cells and normoblasts), immature granulocytes, immature plasmablasts, and thromboblasts.^15,46-48 In patients with sepsis, platelets become activated, which results in changes in their volumes and numbers. The mean ratio of platelet volume to platelet count has been tested as a predictor of 1‑year mortality in critically ill patients.⁴⁹ An overview of the most commonly used ratios and indices for the diagnostic workup of sepsis is provided in Table 3.

Presepsin

Presepsin, a soluble N‑terminal fragment of the CD14 molecule, exists in 2 forms within the body: a membrane‑bound form and a soluble form. The soluble form is secreted by hepatocytes and is detectable in blood. The physiological role of CD14 includes recognizing bacterial, viral, and fungal antigens and activating signaling pathways mediated by toll‑like receptors. These receptors mainly recognize microbial components, such as lipopolysaccharides, lipoteichoic acids, and other lipoproteins. After the phagocytosis of microorganisms and activation of the toll‑like receptor signaling pathway, CD14 is cleaved from monocyte membranes and released into the circulation, where it can be measured.

Recent studies have found no significant differences in the sensitivity and specificity of presepsin, as compared with PCT or CRP. However, the point‑of‑care diagnostic tools, such as PATHFAST Presepsin, are widely available and allow for the rapid assessment of presepsin levels. The importance of presepsin in diagnosing infection and its use as a prognostic marker in adults with sepsis have been summarized in recent studies.^50,51

During the COVID‑19 pandemic, presepsin was applied in the diagnostic workup of infection.⁵² It has also been used for the early diagnosis of neonatal sepsis.^53,54 It is important to note that sensitivity and specificity values for presepsin vary across studies due to differences in the applied cutoff values.

CD64 and other surface markers on neutrophils

Flow cytometry is another diagnostic tool for sepsis, and it is used to measure CD64 expression on neutrophils. According to a meta‑analysis by Cong et al,⁵⁵ CD64 demonstrates higher sensitivity and specificity than PCT or IL‑6 in adult patients. CD64 is also commonly used in the diagnostic workup of neonatal sepsis.⁵⁶

CD64 is a membrane glycoprotein that binds to the IgG Fc receptor that serves as a marker of neutrophil activation. It is constitutively expressed by macrophages and monocytes, whereas its expression on neutrophils occurs after the activation of particular cytokines (specifically interferon γ and granulocyte colony‑stimulating factor).

In recent years, several studies have highlighted the diagnostic utility of CD64 in sepsis. Bhandari et al⁵⁷ found that CD64 had a sensitivity of 80% and a specificity of 79% for detecting neonatal sepsis. Icardi et al⁵⁸ reported that CD64 showed very good predictive value for the clinical and laboratory (microbiological) diagnosis of sepsis in adults, with a sensitivity of 94.6% and a specificity of 88.7%. However, conflicting evidence exists. A more recent study did not confirm the diagnostic usefulness of CD64 in adult sepsis.⁵⁹ This underscores the importance of identifying novel surface markers using new technologies, such as high‑dimensional mass cytometers.⁵⁹

The Intensive Care Infection Score

The Intensive Care Infection Score (ICIS) is calculated from blood count parameters that are routinely collected via hematology analyzers. It incorporates measurement of neutrophils (including count and fluorescence), the number of activated B‑lymphocytes with an antibody response, and the reticulocyte hemoglobin equivalent. The reticulocyte hemoglobin equivalent reflects the hemoglobin content in reticulocytes and is used to assess treatment efficacy in sideropenic anemia.

The ICIS has limitations, as it cannot be used in patients with hematologic malignancies or those undergoing immunosuppressive treatment. However, it correlates well with PCT and CRP levels. Currently, the ICIS is a readily available, inexpensive biomarker that can be used in diagnosing sepsis.^60-62

Omics technologies and artificial intelligence technology

Omics technologies—encompassing genomics, transcriptomics, proteomics, metabolomics, and pharmacogenomics—have evolved substantially since the final mapping of the human genome. Their integration into sepsis diagnostics is progressively transitioning from research to routine clinical practice.

This review provides an overview of the current applications of omics technologies in sepsis, with potential implications for the management of patients with sepsis. These applications include: 1) diagnosing sepsis by identifying biomarkers that differentiate between infectious and noninfectious inflammation; 2) identifying biomarkers that predict a patient’s clinical outcome; 3) finding biomarkers with the potential to be used in sepsis treatment; and 4) finding biomarkers that predict a patient’s response to therapy.

The assessment of genetic polymorphisms has not yet entered the mainstream in the routine diagnosis of sepsis, largely due to small individual effects of these genetic variations. However, in 2022, Engoren et al⁶³ published a study that maps polygenic risk in sepsis 2 and sepsis 3. The authors state that most genetic variants have small effect sizes, but cumulatively, the polygenic risk scores show good discriminatory power.⁶³

The role of transcriptomics has been explored in a report by Pelaia et al,⁶⁴ which highlights the powerful roles of both coding and noncoding RNAs in modulating the host’s septic response. Similarly, proteomic and metabolomic studies are introducing novel biomarkers with potential clinical utility.^65,66 Below, we present examples of diagnostic biomarkers that are already applied in clinical practice.

An example of applying omics technology in sepsis diagnostics is the TriVerity Test (Inflammatix, Sunnyvale, California, United States), which utilizes messenger RNA expression to distinguish between bacterial and viral infections. This test improves the prediction of 28‑day mortality and identifies the need for intensive patient care upon admission to the emergency department.⁶⁷ However, it is not yet available for routine clinical use.

In contrast, the SeptiCyte RAPID test (Immunexpress, Seattle, Washington, United States and Brisbane, Australia) is already approved in the United States, Australia, and Europe. This host‑response molecular test for sepsis uses the reverse transcription‑polymerase chain reaction with whole blood to quantify the relative expression levels of host‑response genes. It differentiates infection‑positive sepsis from infection‑negative SIRS in patients with escalating signs and symptoms of critical illness. The SeptiCyte RAPID test generates a score (SeptiScore) that falls within 4 discrete interpretation bands based on likelihood of sepsis.^68-71 The test relies on messenger RNA expression from 2 genes: phospholipase A2 group VII platelet activating factor (PLA2G7) and placenta‑specific 8 (PLAC8). Within this diagnostic approach, methods for detecting the presence of infectious agents are also applied; however, this topic exceeds the scope of the current discussion.

In recent years, we have witnessed a surge in the use of artificial intelligence (AI) in various fields of medicine. For sepsis, the unresolved issues of early diagnosis and selection of appropriate therapy—including antibiotics, intravenous treatment, and the potential possibility of personalized immunotherapy according to patient stratification—remain a challenge. The potential value of AI in addressing these issues is certainly great; however, with the implementation of AI in clinical practice, it is crucial to remember that sepsis remains fundamentally a clinical diagnosis and potential AI use should complement, not replace, clinical judgment.^72,73

Practical considerations of using biomarkers for sepsis diagnosis

Sepsis is a complex syndrome characterized by immense variability and heterogeneity, with each case presenting unique clinical features. This variability arises from the interplay of network pathophysiology, together with endogenous and exogenous factors that influence the immune, endocrine, and hemodynamic responses, as well as patient’s clinical course and prognosis.⁷⁴ The most important endogenous factors in this context are: age, sex, genotype, phenotype, organ reserve, immune and nutritional status, and comorbidities, with each factor modifying the symptoms and signs of sepsis. Important exogenous factors encompass: the site of infection, the quantity (load) and quality (virulence) of microorganisms, patient social background, and the quality and capacity of the health care system. Each type of bacteria, fungi, or parasite induces a specific cellular, cytokine, and humoral immune response during infection. All of these factors have effects on the clinical course of sepsis and patient outcomes.⁷⁴ For instance, Sweeney et al⁷⁵ used transcriptomics in a cohort of patients with sepsis to reveal 3 robust sepsis endotypes: inflammatory, coagulopathic, and adaptive endotypes, each with a different mortality rate. This shows why it is necessary to use biomarker panels to cover the broad spectrum of host responses to different microorganisms (Figures 2 and 3).

The initial values of biomarkers may be useful for the early diagnosis of sepsis in combination with scoring systems (PRESEP, NEWS, MEWS, qSOFA, SIRS criteria; Figure 1). In the early (hyper-) acute phase of sepsis, we recommend frequent blood collection for the assessment of multiple markers of inflammation: IL‑6, NLR, and PCT should be measured every 8 to 12 hours to track concentration trajectories and to check the intensity of the immune response.^{21-24,27,30,34}

In the ICU, we recommend daily assessment of sepsis markers, 24 hours a day and 7 days a week. Tracking the trajectories of NLR, PCT, CRP, and IL‑6 levels in the first 3 to 5 to 7 days in patients enables evaluation of their immune response, stratification and monitoring of sepsis, assessment of the response to the therapy, and prediction of outcomes.^{21,22,24,29-31,34}

For clinical use, every physician should know the biology, half‑lives, and kinetics of each biomarker, as well as the meaning of their absolute values (including different cutoff values for different groups of patients, eg, medical vs surgical patients). The appropriate interpretation of biomarkers should be performed only within the specific clinical context, including patient history, clinical course, and disease severity. The use a SOFA score in critically ill patients combined with biomarkers (IL‑6, NLR, PCT, CRP; Figure 3) is recommended.

The biomarker concentrations are measures of intensity of the immune–inflammatory response, which does not always correspond with a patient’s clinical status. Very high values confirm the presence of hyperinflammation, a hyperacute response, or a cytokine storm. Conversely, unexpectedly low or mildly elevated levels that do not correlate with the patient’s clinical status (eg, CRP of 50–60 mg/l, or NLR of 4–6 in septic shock) can signal a deteriorating clinical course and worse outcome.^33,34,39,74 For tips and tricks for the evaluation of the parameters for sepsis, see Table 4.

Conclusions

Hundreds of biomarkers with varying sensitivity and specificity are now available for the diagnostic workup of sepsis. Most of these are available in point‑of‑care testing modes and are financially viable for routine use. We believe that the optimal approach is to select a core set of reliable markers and effectively integrate them into clinical practice. In general, the trends of biomarkers changing over time are more important than their absolute values.

When using biomarkers, it is necessary to consider patient‑specific characteristics, such as underlying diseases, corticosteroid use, and other relevant factors. For example, the specificity and sensitivity of PCT at a cutoff of 0.5 µg/l are completely different from those at a cutoff of 2.5 µg/l. Similarly, the indicative value of CRP varies depending on whether the patient is undergoing corticosteroid therapy. Therefore, it is prudent to choose a few biomarkers and learn to work with them for each specific patient.