An analysis of the Global Burden of Disease Study estimated 48.9 million incident sepsis cases and 11.0 million sepsis-related deaths worldwide in 2017, roughly 19.7% of all deaths that year. Despite advances in antimicrobials, haemodynamic resuscitation and organ support, mortality varies widely across populations and case definitions, motivating the search for biomarkers that identify high-risk patients earlier in the course of illness. The biological rationale rests on the immune paralysis seen in critically ill patients.

Sepsis can involve overlapping inflammatory and immunosuppressive responses rather than a neat sequence of two discrete phases, and T-cell dysfunction, reduced monocyte antigen presentation and lymphopenia have all been described. Apoptotic depletion of lymphocytes, including B cells and CD4-positive T cells, has been documented in human sepsis, and immune-cell counts and the neutrophil-to-lymphocyte ratio have previously been investigated as prognostic indicators in intra-abdominal sepsis. Apoptosis itself can proceed through intrinsic mitochondrial signalling or extrinsic death-receptor signalling; binding of Fas ligand (CD95L) to cell-surface Fas (CD95) can initiate caspase-8 signalling.

Annexin V binds externalised phosphatidylserine on the cell membrane and, together with propidium iodide exclusion, identifies an early apoptotic phenotype, but this readout is not specific to Fas-mediated signalling. Circulating sFas and sFasL are related to the Fas pathway, yet their net activity depends on molecular form and biological context, so measured concentrations alone do not establish pathway activation or prove a causal mechanism. The study team, from the clinical laboratory departments of the First Affiliated Hospital of Jinzhou Medical University and Liaoyang Second People's Hospital in China, enrolled adults of 18 years or older admitted to a medical-surgical intensive care unit who met Sepsis-3 criteria, meaning a SOFA score increase of at least two points from baseline attributable to suspected or confirmed infection.

Patients were excluded if they were younger than 18, had malignancy, haematological disease or recent immunosuppressive therapy, carried an autoimmune diagnosis such as systemic lupus erythematosus, stayed in the ICU for less than 48 hours, died within 24 hours of admission, were pregnant, or lacked a Day-1 biomarker sample. Peripheral blood was drawn within 24 hours of ICU admission, with additional measurements planned on Days 3 and 7 when clinically feasible. Peripheral blood mononuclear cells were separated by Ficoll centrifugation and analysed with an Annexin V-FITC/propidium iodide flow-cytometric assay processed within four hours, alongside CD45 and CD3 staining; serum sFas and sFasL were quantified by sandwich ELISA.

The headline result is stark. Of the 218 patients, 75 died within 28 days, a mortality rate of 34.4%. At admission, non-survivors had an average of 28.7% ± 5.9% Annexin V-positive lymphocytes compared with 14.3% ± 4.2% in survivors, sFas concentrations of 4218 ± 863 pg/mL versus 2104 ± 612 pg/mL, and sFasL concentrations of 312 ± 78 pg/mL versus 164 ± 43 pg/mL, with reported between-group p values below 0.001 for all three comparisons.

Non-survivors were also older (median 67 versus 59 years), carried higher baseline SOFA scores (10.8 ± 2.9 versus 6.4 ± 2.1) and APACHE II scores (26.3 ± 6.1 versus 18.5 ± 5.2), and more frequently required mechanical ventilation (69.3% versus 30.8%) and vasopressors (62.7% versus 26.6%). Pulmonary infections were the most common sepsis source at 41.7%, followed by abdominal, bloodstream and urinary tract infections. The biomarkers also tracked the severity of organ dysfunction.

All three admission measurements correlated positively and significantly with Day-1 SOFA score (Spearman r = 0.61 for Annexin V positivity, 0.58 for sFas and 0.49 for sFasL) and with APACHE II score (r = 0.54, 0.52 and 0.43 respectively), and moderately with serum lactate, procalcitonin and C-reactive protein. Notably, the Annexin V-positive percentage correlated inversely with the absolute lymphocyte count (r = −0.48), a pattern compatible with, but not proof of, a link between an early apoptotic phenotype and lower circulating lymphocyte numbers. These correlation analyses are explicitly exploratory, and interpretation must account for the 21 reported tests and the possibility of false positives arising from multiple comparisons.

In regression modelling, the reported adjusted associations remained significant after accounting for clinical covariates. The reported adjusted odds ratios correspond, after unit conversion for interpretability, to approximately 1.22 per 100 pg/mL for sFas and 1.15 per 10 pg/mL for sFasL, alongside an adjusted odds ratio of 1.18 per percentage point for Annexin V positivity; SOFA score, APACHE II score, serum lactate and vasopressor use also retained independent-looking associations. With only 75 deaths available for numerous correlated candidate predictors, however, overfitting is a real concern, and the stepwise selection procedure, model coefficients, multicollinearity diagnostics and internal validation were not provided.

The adjusted estimates must therefore be regarded as unverified exploratory figures rather than validated independent predictors. Discrimination analysis added to the apparent promise. Reported Day-1 area-under-the-curve values were 0.843 for Annexin V positivity, 0.827 for sFas, 0.791 for sFasL, 0.768 for SOFA and 0.779 for APACHE II, while a combined logistic-regression model incorporating all three biomarkers plus SOFA reached 0.891, with a reported DeLong comparison against SOFA alone yielding p = 0.003.

Data-derived cut-offs, such as 20.5% Annexin V positivity or 3180 pg/mL sFas, produced apparent sensitivities between 72% and 80% and specificities between 81% and 85%. Crucially, all of these figures were calculated in the same development cohort, are not optimism-corrected, and have not been externally validated, so the cut-offs and predictive values are likely optimistic and cannot yet guide care. The longitudinal picture is even more provisional.

Reported Day-3 and Day-7 measurements suggested widening gaps, with non-survivors showing rising Annexin V positivity and sFas while survivors' values declined, and patients classified as having ICU-acquired secondary infection showed higher Day-3 Annexin V positivity and sFas. But follow-up samples were obtained only when clinically feasible, so the baseline denominators of 143 survivors and 75 non-survivors cannot be assumed for later time points; death, discharge and incomplete sampling create informative missingness that makes complete-case trajectories unreliable. The secondary-infection association is additionally vulnerable to reverse temporality and confounding, and must not be read as evidence that apoptosis causes infection.

The authors also note that a COVID-19 study reported lower, not higher, sFasL in more severe disease, showing that the direction of association may differ across conditions and cannot be assumed to generalise. The study's limitations are laid out candidly. Excluding patients who died within 24 hours or stayed fewer than 48 hours may introduce survivor-selection bias and restrict applicability to the sickest patients, whose characteristics were unavailable.

Only 75 deaths supported the modelling, with no calibration statistics, Brier score or external validation. The flow-cytometric parent population, whether all CD45-positive lymphocytes or only CD3-positive T cells, remains unclear, and pre-analytical consistency, assay quality control and the timing of sampling relative to sepsis onset were insufficiently documented. Shorter ICU stays among non-survivors are mechanically truncated by death and cannot be equated with faster recovery.

The conclusion drawn by the study is measured: admission Annexin V positivity, sFas and sFasL were associated with 28-day mortality in this cohort, and the combined panel's apparent discrimination is hypothesis-generating, but prospective, multicentre external validation with prespecified analysis and documented sample availability is required before these markers can be recommended for routine risk stratification or therapeutic monitoring in sepsis.

Subject of Research: Prognostic value of lymphocyte apoptosis biomarkers for 28-day mortality in sepsis

Article Title: Prognostic significance of lymphocyte apoptosis markers (Annexin V, sFas, and sFasL) in sepsis: a retrospective clinical study

Article References: Liu, Y., Wu, J., & Li, J. (2026). Prognostic significance of lymphocyte apoptosis markers (Annexin V, sFas, and sFasL) in sepsis: a retrospective clinical study. Frontiers in Cell and Developmental Biology, 14, 1983746. https://doi.org/10.3389/fcell.2026.1983746

Image Credits: AI Generated

DOI: 10.3389/fcell.2026.1983746

Keywords: sepsis, lymphocyte apoptosis, Annexin V, soluble Fas, soluble Fas ligand, biomarkers, 28-day mortality, SOFA score, intensive care, flow cytometry, prognosis, immune dysfunction