No single blood test can confirm or rule out sepsis on its own, which is why clinicians rely on a combination of biomarkers, each capturing a different facet of the body’s response to infection. The most widely used markers today include procalcitonin, C-reactive protein, lactate, and interleukin-6, but the field is expanding rapidly toward gene-expression signatures, cell-surface markers, and machine-learning models that weigh several biomarkers at once. Understanding what each marker actually measures, where it performs well, and where it falls short is essential for making sense of a sepsis workup.
Procalcitonin and C-Reactive Protein
Procalcitonin (PCT) and C-reactive protein (CRP) are the two biomarkers you will encounter most often in any sepsis discussion. Both rise in response to infection, but they behave differently over time and across disease severity. CRP climbs more slowly and tends to plateau once sepsis worsens, whereas PCT rises in proportion to how severe the infection becomes, reaching its highest levels in septic shock. PCT also tends to be higher in patients who do not survive, giving it a closer link to both severity and outcome than CRP provides.1PubMed. Clinical laboratory differentiation of infectious versus non-infectious systemic inflammatory response syndrome That said, neither marker is perfect at distinguishing sepsis from other causes of systemic inflammation. A patient recovering from major surgery, for example, can have an elevated CRP without any infection present.
Where PCT really earns its clinical value is in guiding antibiotic decisions. PCT levels drop relatively quickly once a bacterial infection is under control, so tracking the trend over days can help clinicians decide when to stop antibiotics. Randomized trials have shown that PCT-guided protocols reduce antibiotic duration, and the recent ADAPT-Sepsis trial confirmed a reduction in antibiotic use, though the absolute difference was only about one day. That trial also compared PCT guidance head-to-head with CRP guidance and found CRP offered no measurable benefit. Notably, as shorter fixed-duration antibiotic courses become standard practice, the added value of PCT-guided de-escalation may shrink further.2PubMed. Reassessing Procalcitonin-Guided Antibiotic Therapy in Critically Ill Patients With Sepsis: Lessons From the ADAPT-Sepsis Trial Still, PCT remains the most studied biomarker for antibiotic stewardship in sepsis, and many hospitals have built their de-escalation protocols around it.3PubMed Central. Procalcitonin Guided Antibiotic Stewardship
Lactate as a Marker of Tissue Distress
Lactate measures something fundamentally different from PCT or CRP. Rather than reflecting the immune response to infection, lactate indicates whether organs and tissues are getting enough oxygen. When blood flow drops during sepsis, cells switch to less efficient energy production, and lactate accumulates. An elevated lactate level on arrival in the emergency department is one of the strongest red flags that a patient may be heading toward septic shock.
Just as important as the initial reading is how quickly lactate falls after treatment begins. A study of sepsis patients with initially high lactate found that the lactate level measured at six hours was a better predictor of 30-day survival than either the starting value or the rate of clearance alone.4PubMed Central. Prognostic value of lactate levels and lactate clearance in sepsis and septic shock with initial hyperlactatemia This is why many sepsis protocols call for repeat lactate measurements within the first several hours of treatment: a number that stays stubbornly high suggests the initial resuscitation is not working and the approach needs to change.
One wrinkle worth knowing is that lactate clearance does not predict outcomes equally across all infection types. A large retrospective study found that lower lactate clearance was strongly tied to higher in-hospital death rates among patients whose infection did not originate in the lungs. In patients with pneumonia, however, lactate clearance had no significant effect on mortality or ventilator-free days.5Scientific Reports. The impact of lactate clearance on outcomes according to infection sites in patients with sepsis: a retrospective observational study The takeaway is that even a well-validated marker like lactate can behave differently depending on the source of the infection.
Interleukin-6 and the Cytokine Response
Interleukin-6 (IL-6) is one of the main cytokines the immune system releases early in an infection. It rises fast, often within hours of infection onset, and that speed gives it an edge for early detection. In one prospective study, IL-6 distinguished sepsis patients from uninfected controls with high accuracy. More telling, IL-6 levels above roughly 349 pg/mL could separate patients in septic shock from those with sepsis alone, and the group with IL-6 at or above that threshold had significantly higher 28-day mortality.6PubMed Central. Diagnostic and prognostic value of interleukin-6, pentraxin 3, and procalcitonin levels among sepsis and septic shock patients: a prospective controlled study according to the Sepsis-3 definitions
Beyond predicting survival, IL-6 levels correlate strongly with how much organ damage is occurring. In critically ill sepsis patients, plasma IL-6 tracked closely with standard severity scoring systems and with the degree of organ dysfunction present.7PubMed Central. Impact of interleukin 6 levels on acute lung injury risk and disease severity in critically ill sepsis patients IL-6 is already used routinely in some European hospitals, and certain automated analyzers can return results within minutes, making it practical for emergency settings. The main limitation is that IL-6 rises in many inflammatory conditions beyond infection, so like CRP, an elevated level alone does not confirm sepsis.
Presepsin, a Marker from the Front Lines of Immune Defense
Presepsin (also called sCD14-ST) is a fragment of a receptor that sits on the surface of immune cells called monocytes and macrophages. When these cells encounter bacteria, they shed presepsin into the bloodstream, so it specifically reflects an active immune encounter with a pathogen rather than just general inflammation. A meta-analysis pooling data from multiple adult studies found presepsin had a pooled sensitivity of about 86% and specificity of about 78% for diagnosing sepsis, with an overall accuracy measure around 0.89.8PubMed Central. The accuracy of presepsin (sCD14-ST) for the diagnosis of sepsis in adults: a meta-analysis
Head-to-head comparisons with PCT have produced mixed results depending on the patient population and study design. One emergency department study found presepsin’s diagnostic performance comparable to PCT’s.9PubMed Central. Evaluation of Presepsin for Early Diagnosis of Sepsis in the Emergency Department Another study in critically ill patients found that when presepsin and PCT were used together in parallel, their combined sensitivity jumped to about 93%, substantially better than either marker alone.10PubMed Central. Comparison of Diagnostic Accuracy of Presepsin and Procalcitonin for Sepsis in Critically Ill Patients: A Prospective Observational Study This “panel” approach, using two or more biomarkers together to compensate for each one’s blind spots, is a recurring theme in sepsis research.
Monocyte Distribution Width From a Routine Blood Count
One of the more practical developments in recent years is monocyte distribution width (MDW), a measure of how variable in size the monocytes are in a standard complete blood count. When the immune system activates against an infection, monocytes change shape and size, and that variability shows up as a wider distribution. The appeal is that MDW can be generated automatically by hematology analyzers that hospitals already use for routine blood work, with no extra test ordered and no additional cost.
In a study of high-risk emergency department patients, an MDW value above 20.0 distinguished sepsis from other conditions with moderate accuracy. Where MDW really shines is as a rule-out tool: patients with MDW at or below 20.0 had a negative predictive value above 93%, meaning a normal MDW made sepsis very unlikely. Pairing MDW with a standard white blood cell count improved detection further. Patients who had both a normal white cell count and a normal MDW had roughly a six-fold lower probability of sepsis compared to those with abnormal values.11PubMed Central. Monocyte Distribution Width: A Novel Indicator of Sepsis-2 and Sepsis-3 in High-Risk Emergency Department Patients Because MDW comes from a test that is already being drawn on virtually every emergency patient, it adds almost no time or expense to the diagnostic workup.
Cardiac Markers in Sepsis
Sepsis can damage the heart even in patients with no prior cardiac disease, a phenomenon sometimes called septic cardiomyopathy. Cardiac troponin I (cTNI) and NT-proBNP, both familiar from cardiology, have found a role in flagging this complication. A study comparing the two found that troponin I had better diagnostic accuracy for identifying sepsis and septic shock, particularly by the third day of illness. Troponin I also showed moderate prognostic value for predicting 30-day mortality.12PubMed Central. Cardiac Troponin I Reveals Diagnostic and Prognostic Superiority to Aminoterminal Pro-B-Type Natriuretic Peptide in Sepsis and Septic Shock
Other work, however, has found NT-proBNP to be the stronger prognostic marker from the third day onward, with very high sensitivity by day four.13Anesteziologie a intenzivnà medicÃna. N-Terminal-Pro Brain Natriuretic Peptide (NT-pro BNP) and Troponin I as prognostic markers of septic cardiomyopathy The discrepancy likely reflects differences in patient populations and timing. In practice, rising cardiac markers in a sepsis patient should prompt clinicians to assess heart function directly, because septic cardiomyopathy changes fluid management and may warrant different hemodynamic support.
Coagulation and Endothelial Damage
Sepsis frequently triggers abnormal blood clotting throughout the small vessels, a process that in its most severe form becomes disseminated intravascular coagulation (DIC). D-dimer, a breakdown product of blood clots, is the most commonly measured marker of this process. Elevated D-dimer levels and DIC scores both show reasonable accuracy for distinguishing septic shock from less severe sepsis. Patients with very high D-dimer levels, above roughly 30 mg/L, faced about two and a half times the risk of dying within 30 days compared to those with lower values.14PubMed. D-Dimer Levels and the Disseminated Intravascular Coagulation Score to Predict Severity and Outcomes in Sepsis or Septic Shock
Beyond clotting factors, markers of damage to the blood vessel lining itself are drawing research interest. Angiopoietin-2, a protein released when endothelial cells are injured, was significantly higher in septic shock patients with organ failure involving the kidneys, liver, or cardiovascular system. In a small but detailed study, nonsurvivors had angiopoietin-2 levels nearly three times higher than survivors at ICU admission. A separate endothelial marker called endocan appeared to flag a different complication: low endocan at admission predicted the development of respiratory failure within three days.15PubMed. Evaluation of endothelial biomarkers as predictors of organ failures in septic shock patients These endothelial markers are not yet routine, but they illustrate how different biomarkers can point to specific organ threats.
Kidney Injury Markers
Acute kidney injury is one of the most common and dangerous organ complications of sepsis. The traditional marker, serum creatinine, rises only after the kidneys have already sustained substantial damage, creating a frustrating delay. A urine-based test measuring two proteins, TIMP-2 and IGFBP7, was cleared by the FDA specifically to assess which critically ill patients are at risk for developing moderate to severe kidney injury.16PubMed Central. Clinical Use of the Urine Biomarker [TIMP-2] × [IGFBP7] for Acute Kidney Injury Risk Assessment Both proteins are involved in cell-cycle arrest, a stress response that kidney cells undergo before outright damage sets in. By detecting this stress signal early, the test gives clinicians a window to pull back nephrotoxic drugs, adjust fluid management, or avoid contrast dye, potentially preventing the injury from progressing.
Sepsis Markers in Newborns and Children
Diagnosing sepsis in neonates is especially tricky because the usual biomarkers behave less reliably in the first days of life. CRP can take over 12 hours to rise after birth, and PCT naturally surges in the first 48 hours regardless of infection, muddying interpretation. Neutrophil CD64, a protein on the surface of white blood cells that increases when those cells are activated against bacteria, has emerged as a particularly useful marker in this population. A composite analysis across nine studies of neonates and children found sensitivity of about 83% and specificity of about 86% for detecting sepsis, placing CD64 among the most accurate biomarkers available for pediatric use.17Biochemia Medica. Neutrophil CD64 as a sepsis biomarker
For early-onset neonatal sepsis specifically, CD64 achieved 100% sensitivity in one study, meaning it caught every infected neonate in the sample. Its negative predictive value was also 100%, making it a strong tool for ruling sepsis out and avoiding unnecessary antibiotic courses in healthy newborns.18PubMed Central. Neutrophil CD64 as a diagnostic marker in neonatal sepsis A separate study in preterm neonates confirmed that CD64 outperformed white blood cell count, neutrophil count, CRP, and another surface marker called CD11b.19PLoS ONE. Diagnostic Utility of Neutrophil CD64 as a Marker for Early-Onset Sepsis in Preterm Neonates The practical barrier is that measuring CD64 requires flow cytometry equipment, which not every hospital nursery has on hand.
Gene Expression Signatures
Rather than measuring one or two proteins, gene expression profiling reads the activity of dozens or hundreds of genes at once from a blood sample, producing a “signature” of how the immune system is responding. This approach can potentially distinguish bacterial from viral infections and predict which patients will deteriorate. A systematic review and meta-analysis of host RNA biosignatures in adults found an overall accuracy around 0.86 for diagnosing sepsis, performing slightly better against healthy controls than against patients with non-infectious inflammation.20PubMed Central. The Diagnostic Utility of Host RNA Biosignatures in Adult Patients With Sepsis: A Systematic Review and Meta-Analysis
In children, the results are even more striking. A multicenter study derived a ten-gene signature that could distinguish bacterial from viral infections with over 94% accuracy in a validation cohort. A separate ten-gene severity signature predicted organ dysfunction within 24 hours. When both signatures were used together, accuracy for predicting organ failure climbed above 90% regardless of whether the infection was bacterial or viral.21PubMed. Host gene expression signatures to identify infection type and organ dysfunction in children evaluated for sepsis: a multicentre cohort study Transcriptomic approaches have also revealed that sepsis is not a single disease but contains distinct subclasses with different molecular profiles, which may explain why treatments that help one group of patients can fail or even harm another.22PubMed Central. Gene expression profiling in sepsis: timing, tissue, and translational considerations
Sepsis Phenotypes and Why They Matter for Treatment
The idea that sepsis contains hidden subgroups has moved beyond transcriptomics into clinical biomarker data. An analysis of patients from a major sepsis trial used a combination of protein biomarkers and clinical variables to identify two distinct phenotypes. Roughly one in eight patients fell into a hyperinflammatory phenotype characterized by higher IL-6, higher markers of endothelial activation, elevated bilirubin, and lower platelet counts. This group had a 60-day mortality of 41%, compared with 16% in the larger, less inflamed group. Perhaps most consequentially, the hyperinflammatory patients appeared to do worse when given early goal-directed therapy compared to usual care, while the other group showed no difference between treatments.23Scientific Reports. Identification of a hyperinflammatory sepsis phenotype using protein biomarker and clinical data in the ProCESS randomized trial Findings like these suggest that the failure of many sepsis trials may stem not from bad drugs but from treating a biologically mixed population as if it were one disease.
Machine Learning and Multi-Biomarker Panels
Given that no single marker captures the full picture, combining several biomarkers into a panel and feeding the data through an algorithm is an obvious next step. Multi-biomarker panels paired with machine learning models and clinical severity scores have shown improved ability to detect sepsis early, forecast organ failure, and guide treatment changes like antibiotic de-escalation.24International Journal of Pharmaceutical Sciences. Integrating Multi-Biomarker Panels and Emerging Technologies in A Systematic Meta Analysis of Biomarkers for Early Sepsis Detection, Prognosis and Therapeutic Monitoring in Critically Ill Patients
Some of this work is already reaching the bedside in prototype form. One research group built a device that rapidly measures seven biomarkers simultaneously and used a decision-tree model to predict patient mortality and recovery with over 92% accuracy across two time points in early sepsis.25Biosensors and Bioelectronics: X. Multiplexed host immune response biosensor for rapid sepsis stratification and endotyping at point-of-care Another team developed a portable nanobiosensor that measures multiple cytokines and uses machine learning to identify sepsis immunoparalysis, the dangerous phase in which the immune system becomes suppressed after its initial overreaction.26PubMed Central. A portable nanobiosensor with machine learning: enabling multi-cytokines point-of-care testing and early identification of sepsis immunoparalysis These point-of-care platforms aim to compress what currently requires multiple lab instruments and hours of turnaround time into a single bedside test delivering results in minutes.27PubMed Central. Point of care technologies for sepsis diagnosis and treatment
The Gut Microbiome and Metabolomics
A more exploratory frontier involves looking at the gut microbiome and the small molecules circulating in a patient’s blood. Sepsis disrupts the normal balance of gut bacteria, and these disruptions appear to feed back into disease severity. Patients with sepsis showed loss of beneficial bacteria and overgrowth of potentially harmful species like Enterococcus. Patients with a high burden of Bacteroides, particularly one species called B. vulgatus, had worse severity scores and longer ICU stays. Disturbances in tryptophan metabolism, an amino acid pathway heavily influenced by gut bacteria, tracked with both microbiome changes and sepsis severity.28PubMed Central. Altered intestinal microbiome and metabolome correspond to the clinical outcome of sepsis
Separately, researchers have built prediction models using microbiome and metabolite data from hospitalized sepsis patients. In one model, higher levels of the bacterium Klebsiella and the bile acid taurocholic acid, combined with lower levels of butyric acid (a short-chain fatty acid produced by healthy gut bacteria), predicted a higher probability of death.29PubMed Central. Global Signatures of the Microbiome and Metabolome During Hospitalization of Septic Patients None of this is ready for routine clinical use yet, but metabolomics and microbiome profiling may eventually enable more personalized approaches to sepsis management, identifying which patients need aggressive intervention and which are likely to recover with standard care.30PubMed. Metabolomics and the Microbiome as Biomarkers in Sepsis
Emerging MicroRNA Markers
MicroRNAs are tiny molecules that regulate gene activity, and their levels in the bloodstream shift during sepsis. Because they are relatively stable in blood and can be measured with high sensitivity, they are attractive candidates for early detection. One study identified a specific circulating microRNA, hsa-let-7d-3p, as a novel biomarker for the early detection of sepsis in human subjects.31PubMed Central. Evidence for the circulating microRNA hsa-let-7d-3p as a potential new biomarker for sepsis in human subjects MicroRNA research in sepsis is still in its early stages, with most studies involving relatively small cohorts. But the concept fits well with the broader shift toward molecular-level diagnostics, and microRNAs could eventually be incorporated into the multi-analyte panels that machine learning models are designed to interpret. For the moment, their clinical utility remains to be validated in larger, multicenter trials.