Monitoring T-cell function in septic shock: one-year experience with a fully automated assay Authors: Thomas Lafon, Morgane Gossez, Annabelle Schild, Martin Cour, Laurent Argaud, Franck Berthier, Pauline Perez, Fabienne Venet, Anne-Claire Lukaszewicz, Guillaume Monneret Published: 2026 ABSTRACT AbstractBackgroundIn sepsis, personalized immunotherapy is being evaluated as a strategy to restore immune function in the most severely affected patients. Biomarkers are critical in this process, as clear clinical indicators of immune status are lacking. Functional testing is considered the gold standard for assessing immune function, but its clinical implementation faces analytical and standardization challenges. The objective of the present prospective, observational, single-center cohort study was to evaluate a fully automated protocol for assessing T cell functionality in septic shock patients. MethodsIn 66 patients with septic shock, we assessed T lymphocyte functionality using an interferon-γ release assay (IGRA) in response to mitogen. The assay was performed via a fully automated protocol during a one-year period. Patients were monitored three times during the first week after intensive care unit (ICU) admission. Phenotypic immunological parameters, including T cell subpopulation counts and monocyte HLA-DR expression (mHLA-DR), were also assessed. Patient outcomes were followed for 28 days, and a composite clinical deterioration score was defined by the occurrence of 28-day mortality and/or ICU-acquired infection.ResultsCompared with reference values, we observed a significant reduction in IFN-γ release capacity, which correlated with characteristic alterations in cellular immunological parameters. By the end of the first week, reduced IFN-γ release combined with low mHLA-DR identified a severe immunological phenotype associated with an increased risk of clinical deterioration.ConclusionGiven the observational nature of this study, further well-designed investigations in larger patient cohorts are required to validate these findings and assess their potential clinical relevance. If confirmed, this fully automated assay (performed on whole blood, requiring no technician intervention, and providing results within four hours) may offer a practical tool for monitoring immune functional alterations in routine clinical practice.Supplementary InformationThe online version contains supplementary material available at 10.1186/s13054-026-06128-0.Keywords: Immunosuppression, Sepsis, IGRA, Interferon-γ, CD8, HLA-DR Introduction Most patients with sepsis admitted to intensive care units (ICU) exhibit a dysregulated immune response associating an initial hyper-inflammatory response and a compensatory anti-inflammatory response that can lead to profound ICU-acquired immunosuppression [1]. This immunosuppression is associated with an increased risk of nosocomial infections, viral reactivation, prolonged ICU stay, and higher mortality [2, 3]. Consequently, immunostimulatory strategies (e.g., IFN-γ, IL-7, anti–PD-1 therapies) have been proposed to restore immune competence in patients with the most severe alterations [4]. However, for these strategies to be effective in a personalized medicine approach, clinicians must be able to identify patients with the most profound and persistent immune alterations, who are most likely to benefit from immune adjuvant therapy. Since immunosuppression often lacks clear clinical manifestations, there is a pressing need to develop robust and standardized biomarkers of immune status following sepsis [5]. Among the immune alterations described in critically ill patients, decreased expression of HLA-DR on monocytes (mHLA-DR) has been the most extensively studied and is now considered a reliable biomarker for guiding immunotherapies [6], particularly when targeting myeloid cells (e.g., IFN-γ therapy). To this end, an actionable clinical threshold has been defined and applied in prospective randomized controlled trials. Regarding adaptive immune response, large cohort studies have shown that profound lymphopenia and unfavorable lymphocyte trajectories are associated with poor outcomes [7]. However, the substantial inter-individual variability of lymphocyte counts limits its predictive utility at the individual level, and results have not been consistently reproduced across studies. Increased PD-1 expression on T-lymphocytes has also been proposed as a marker of cellular exhaustion [8]. However, the lack of methodological standardization and insufficient validation in large-scale cohorts currently preclude use in clinical routine practice. Beyond phenotypic biomarkers and measurements of absolute cell counts, functional assays are often considered the gold standard for assessing immune status in clinical immunology [9]. Conceptually, these assays constitute the most straightforward method for diagnosing immunosuppression, as they directly assess the capacity of immune cells to respond to stimulation, either in vivo or in vitro. For example, in vivo skin testing using recall antigens provides highly informative results, as it reproduces the whole sequential steps of an immune response (antigen uptake and presentation, migration to lymph nodes, lymphocyte activation and proliferation, and local inflammatory infiltration). Pioneering studies conducted at McGill University in ICU patients demonstrated strong associations between impaired skin test responses and an increased incidence of ICU-acquired infections and mortality [10, 11]. However, these tests were abandoned in favor of laboratory-based assays. Among these, assessment of T-lymphocyte functionality is now feasible through various approaches [5]. In addition to reduced proliferative capacity, difficult to measure routinely in daily clinical practice, impaired cytokine production has been consistently reported by different methods including intracellular cytokine staining, ELISpot, and cytokine release assays [8, 12–15]. Although these techniques provide valuable insights into pathophysiology, they present substantial analytical and logistical barriers to widespread clinical implementation, including reliance on manual and time-consuming procedures using in-house protocols. Furthermore, these methods are generally not automated and are not consistently performed using whole blood but performed after cell purification. Together, these limitations represent a major barrier to the integration of functional immune monitoring into routine clinical practice. Recently, in a preliminary proof-of-concept study involving a small cohort of septic patients over a short period of time, we evaluated a novel fully automated whole-blood interferon-γ release assay prototype [16]. The objectives were to assess the practicability of the prototype with clinical samples and to determine whether this could reliably detect impaired lymphocyte function following sepsis. Encouraged by the initial, convincing results, we subsequently initiated the present prospective, observational, single-center cohort study. The main objective was laboratory-based, focusing on assessing the feasibility of the approach under routine-like conditions during a one-year follow-up study. The secondary objectives were to confirm our preliminary observations (i.e., altered interferon-γ release) and to examine their association with clinical outcomes, including mortality and ICU-acquired infections. Methods Patients In a prospective, observational, single-center cohort study, we consecutively included patients admitted to ICU in Anesthesiology and Intensive Care Department and Intensive Care Medicine department (Hôpital E. Herriot, Hospices Civils de Lyon) with septic shock, according to sepsis-3 definition [17]). These patients were enrolled in the IMMUNOSEPSIS study (ClinicalTrials.gov identifier: NCT04067674), which was designed to monitor immune parameters during the first week following the onset of septic shock. Inclusion criteria were: patient (over 18 years) admitted with diagnosis of septic shock within less than 48h at time of screening (defined by : presence of a microbiologically diagnosed or suspected infection, initiation of a vasopressive treatment to maintain mean arterial blood pressure ≥ 65 mm Hg initiated during the first 48h after ICU admission, presence of an hyperlactatemia > 2 mmol/L during the 24h before or after initiation of vasopressive treatment despite adequate volemic reanimation). Patients were excluded if they met any of the following: age < 18 years; pregnant or breastfeeding women; absence of social security coverage; restriction of liberty or placement under legal protection; language barrier. Heparinized blood samples were obtained at day (D): D 1/2, 3/4 and 5/8 (one sample per period), with D1 corresponding to the first 24 hours after sepsis onset. Because of the non interventional nature of the study, written informed consent was not required for inclusion. However, each clinician orally informed the patient or family members about the objectives, methodology, and conduct of the study, and provided them with a leaflet using plain language. Non opposition to participation to this study was systematically obtained from the patient or a third party before any blood sampling was performed and was recorded in patient’s clinical file. All procedures performed were in accordance with the ethical standards of the institutional and/or national research committee (Comité de Protection des Personnes Sud-Ouest II, n° RCB : 2019-A00210-57, n°CPP : 19.01.23, 71857) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Clinical and biological databases were reported to the National Commission for Information Technology and Freedom (CNIL, number 08-27). A written non-opposition to the use of donated blood for research purposes was obtained from healthy volunteers (n = 23, median age of 45 years, 64% women). Usual clinical data and biological data (age, sex, length of hospital and ICU stay, type of initial infection, severity scores at admission: SAPS II, SOFA, Charlson scores), exposure to invasive devices, living or deceased status after 28 days, presence or absence of ICU-acquired infection, mHLA-DR, absolute lymphocyte count, lymphocyte subsets) were collected. Exclusion criteria included pregnancy or breastfeeding, absence of social security coverage, legal guardianship or protection, and participation in an interventional drug study that could interfere with immunological results. Due to the relatively small sample size, and consequently the limited number of adverse events, we defined a composite “clinical deterioration” outcome, including day-28 mortality and/or the occurrence of ICU-acquired infections, in order to enable more robust statistical analyses. IGRA on VIDAS® 3 system Fully automated IGRA was performed on the VIDAS 3 system (bioMérieux, Marcy l’Etoile, F) using the supplied reagents: VIDAS STIMM T RUO kit and VIDAS IGRA 4 RUO kit (bioMérieux). At this stage, both kits are labeled “for research use only”. The test is based on ex‑vivo stimulation of leukocytes present in heparinized fresh whole blood sample during 3.5 hours in strips containing either non-specific stimulation T-cell mitogen (Phytohemagglutinin, PHA) or 1 nil condition (negative control, i.e., incubation without PHA). In practice, the primary heparinized tubes sampled from patients and strips containing stimulation reagents were loaded into the instrument. Then, blood transfers into reagent-containing strips were performed by the VIDAS 3 automatic pipetting unit (APU). The samples were homogenized and incubated with the individual stimulation reagents for 3.5 hours at 37°C in the VIDAS strips. Next, the system automatically performed an enzyme-linked immunofluorescent assay (ELFA) to quantify the sample IFN-γ concentrations in supernatants. According to the manufacturer’s instructions, IFN-γ results may be reported in international units (IU/mL), with a measuring range from 0.08 IU/mL to 8 IU/mL (values > 8 IU/mL are considered to reflect a normal response) or as continuous arbitrary reference fluorescence values (RFV). As continuous variables were required for statistical and correlation analyses, fluorescence values were used. Final IFN-γ release results were normalized based on baseline unstimulated condition (i.e. PHA stimulation results minus nil condition). Flow cytometry mHLA-DR expression expressed as total number of antibody bound per cells (AB/C) and lymphocyte subsets expressed as absolute count were obtained as previously described [6]. Cells were analyzed on AQUIOS and NAVIOS cytometer, Beckman Coulter (Hialeah, FL). Statistical analysis The statistical analysis and visualizations were performed using R.4.2.2 and Prism 8 softwares. Results are presented as medians and interquartile ranges. Mann-Whitney tests were used to compare quantitative variables between groups. Qualitative variables were compared using the Chi-square test. Spearman’s rho correlation test was employed to determine the correlation between immunological parameters. Survival analyses were performed using the Kaplan-Meier method, and differences between survival curves were evaluated with the log-rank test. As the study was observational and no preliminary data were available regarding its association with clinical outcomes, the required sample size could not be estimated. Therefore, the recruitment duration was arbitrarily set as a one-year inclusion period. Results Sixty-six patients were included between October 2025 and end of September 2026. Baseline clinical characteristics and severity scores are detailed in Table 1. The cohort was markedly severe, with median SAPS II and SOFA scores of 56 and 9, respectively. The 28-day mortality rate reached 32%, and 23% of patients experienced ICU-acquired infections. Cellular immunological results during monitoring are depicted in Table 2. Patients presented with features of severe ICU-acquired immunosuppression. As examples, upon admission, median mHLA-DR was 4328 AB/C (lab reference > 13 5000 AB/C) and median CD4+ T cell count was 187 (lab reference > 360 cells/microL). Overall, the clinical and immunological features of our cohort were comparable to that recently reported in more than 1000 septic shock patients [6]. Table 1Patients’ main characteristicsTotal cohortN=66ClinicaldeteriorationN=32No clinical deteriorationN=34p-valueDemographicsAge (y), median [Q1-Q3]71 [65-78]67 [62-77]74 [69-80]0.029Male, (%)49 (74%)25 (78%)24 (71%)0.578Charlson comorbidities index, median [Q1-Q3]3 [1-5]4 [1-6]z2 [1-4]0.178Body mass index, median [Q1-Q3]25 [22-31]26 [21-32]25 [22-30]0.643Immunosuppressed, (%)15 (23%)11 (34%)4 (12%)0.040Origin of septic shockCommunity-acquired43 (65%)20 (63%)23 (68%)0.797Hospital-acquired21 (32%)10 (31%)11 (32%)1.000ICU-acquired2 (3%)2 (6%)0 (0%)-Site of infectionPulmonary, (%)15 (23%)8 (25%)7 (21%)0.772Abdominal, (%)23 (35%)14 (44%)9 (26%)0.197Urinary, (%)11 (17%)5 (16%)6 (18%)1.000Others, (%)17 (26%)5 (16%)12 (35%)0.093Organ supportVasopressor use, (%)66 (100%)32 (100%)34 (100%)-Mechanical ventilation, (%)39 (59%)24 (75%)15 (44%)0.013Intermittent hemodialysis, (%)16 (24%)14 (44%)2 (6%)<0.001Continuous renal replacement therapy, (%)7 (11%)4 (12%)3 (9%)0.704Hydrocortisone hemosuccinate, (%)56 (84%)28 (88)28 (82)0.734Septic shock severity Day 1SAPS II, median [Q1-Q3]56 [44-70]63 [51-77]49 [42-57]0.004SOFA score, median [Q1-Q3]9 [7-12]11 [9-12]7 [6-9]<0.001Lactates (mmol/L), median [Q1-Q3]3.3 [2.5-5.4]4.2 [2.6-5.8]3.0 [2.3-4.6]0.129Mean arterial pressure (mmHg), median [Q1-Q3]56 [51-62]54 [44-60]59 [55-67]0.005Vasopressor dose (µg/kg/mn), median [Q1-Q3]0.86 [0.38-1.56]1.0 [0.46-2.16]0.66 [0.25-1.15]0.008PaO2/FiO2 ratio, median [Q1-Q3]259 [133-315]150 [109-278]302 [260-368]<0.001Diuresis (ml), median [Q1-Q3]1000 [475-1550]750 [240-1200]1150 [619-1700]0.017OutcomeICU length of stay, median [Q1-Q3]6 [3-14]8 [2-22]5 [3-13]0.269Hospital length of stay, median [Q1-Q3]16 [10-29]12 [6-29]20 [13-30]0.059D28 mortality, (%)21 (32)---ICU-acquired infection, (%)15 (23)---Results are expressed as % or median [Q1-Q3]. Variables were compared using Mann-Whitney and Chi2 tests Patients’ main characteristics Results are expressed as % or median [Q1-Q3]. Variables were compared using Mann-Whitney and Chi2 tests Table 2Patients’ immunological parameters throughout the monitoring periodTotal cohortN=66Clinical deterioration N=32No clinical deteriorationN=34p-valueD 1/D2 (n = 60)mHLA-DR AB/C [Q1-Q3]4328 [2419-5965]3599 [2278-7161]4743 [3018-5921]0.612NLR [Q1-Q3]19 [11-35]24 [13-45]17 [10-29]0.256Lymphocytes, cells/µL [Q1-Q3]514 [395-734]524 [316-813]514 [402-715]0.921CD3 T, cells/µL [Q1-Q3]305 [179-473]253 [127-505]313 [233-412]0.524CD4 T, cells/µL [Q1-Q3]187 [119-317]202 [78-376]178 [134-284]0.939CD8 T, cells/µL [Q1-Q3]95 [43-172]79 [36-163]100 [71-195]0.239D 3/D4 (n = 49)mHLA-DR AB/C [Q1-Q3]3900 [2415-9184]3524 [1896-8588]5310 [2944-10154]0.401NLR [Q1-Q3]9 [6-18]11 [6-33]9 [6-18]0.668Lymphocytes, cells/µL [Q1-Q3]687 [531-1027]640 [526-936]741 [530-1094]0.396CD3 T, cells/µL [Q1-Q3]474 [255-723]458 [282-563]507 [248-811]0.334CD4 T, cells/µL [Q1-Q3]346 [189-471]343 [169-452]380 [193-534]0.500CD8 T, cells/µL [Q1-Q3]125 [70-215]109 [44-195]152 [73-238]0.132D 5/D7 (n = 33)mHLA-DR AB/C [Q1-Q3]5378 [2607-9056]4574 [2547-7341]5722 [3184-10156]0.259NLR [Q1-Q3]14 [6-29]16 [8-44]7 [5-21]0.086Lymphocytes, cells/µL [Q1-Q3]849 [567-1386]780 [458-1228]974 [605-1599]0.421CD3 T, cells/µL [Q1-Q3]558 [302-1094]520 [283-1034]762 [325-1208]0.158CD4 T, cells/µL [Q1-Q3]388 [220-678]376 [180-598]552 [247-887]0.248CD8 T, cells/µL [Q1-Q3]165 [93-282]149 [78-260]218 [128-299]0.231mHLA-DR results are expressed as AB/C (antibody bound / cell), absolute lymphocyte count results are expressed as number of cells / µL, NLR: neutrophil-to-lymphocyte ratio. Variables were compared using Mann-Whitney test Patients’ immunological parameters throughout the monitoring period mHLA-DR results are expressed as AB/C (antibody bound / cell), absolute lymphocyte count results are expressed as number of cells / µL, NLR: neutrophil-to-lymphocyte ratio. Variables were compared using Mann-Whitney test Healthy volunteers (HV; n = 23; 66% women; median age 45 years) exhibited strong IFN-γ releases: all values except one were > 8 IU/mL, with median RFV of 7391 [6240 - 8498] (Fig. 1). Additional immunological parameters of HV are provided in Figure S1. IFN-γ release was markedly decreased in patients with septic shock (Fig. 1). The lowest values in patients were observed on D 1/2: 1372 RFV [711-3075]. Although a slight increase over time was observed, IFN-γ release responses remained low throughout the one-week follow-up (Fig. 1). At each time point, median values consistently remained below 4500 RFV, the threshold defining a normal response (corresponding to > 8 IU/mL). Details of IFN-γ release according to the different conditions (i.e., nil and PHA stimulation) are shown in Table 3. IFN-γ secretion results throughout the monitoring period Results are depicted as medians and individual values as delta IFN-γ production (value obtained after PHA stimulation minus the value from the nil condition). Dashed line represents reference values provided by the manufacturer and confirmed in 23 healthy donors in the present study. Number of samples at each time point as follows: D 1/2 (n = 60), D 3/4 (n = 49), D 5/7 (n = 33) Table 3IFN-γ production results throughout the monitoring periodsTotal cohortN=66clinical deterioration N=32No clinical deteriorationN=34p-valueD1/D2 (n=60)Baseline IFN-γ release [Q1-Q3]34 [18-111]41 [23-161]32 [16-58]0.169IFN-γ release after stimulation [Q1-Q3]1686 [869-3151]1316 [606-2587]1847 [1126-3274]0.228Delta IFN-γ production [Q1-Q3]1372 [711-3075]1224 [391-2283]1738 [1106-3258]0.086D3/D4 (n=49)Baseline IFN-γ release [Q1-Q3]36 [18-70]47 [25-71]29 [17-68]0.483IFN-γ release after stimulation [Q1-Q3]3294 [1080-5055]2953 [665-6136]3654 [1958-5026]0.518Delta IFN-γ production [Q1-Q3]3266 [871-5037]2906 [618-6044]3637 [1475-5013]0.585D5/D7 (n=33)Baseline IFN-γ release [Q1-Q3]28 [15-52]28 [14-62]29 [20-37]0.806IFN-γ release after stimulation [Q1-Q3]2679 [1529-6192]2809 [1270-5597]2617 [1576-7609]0.730Delta IFN-γ production [Q1-Q3]2652 [1511-6167]2791 [1254-5560]2584 [1558-7575]0.730Results are presented as fluorescence values under each condition: baseline (incubation without PHA, nil condition) and after PHA stimulation. Delta IFN-γ production corresponds to the value obtained after PHA stimulation minus the value from the nil condition. Variables were compared using Mann-Whitney test IFN-γ production results throughout the monitoring periods Results are presented as fluorescence values under each condition: baseline (incubation without PHA, nil condition) and after PHA stimulation. Delta IFN-γ production corresponds to the value obtained after PHA stimulation minus the value from the nil condition. Variables were compared using Mann-Whitney test We next examined correlations between IFN-γ release results and cellular parameters at each time point using a correlation matrix (Fig. 2A). The strongest correlations were observed with absolute CD8 counts. Correlation curves between IFN-γ release and CD8 are shown in Fig. 2B. Notably, the correlation was highest on D 1/2 and gradually decreased over time, from 0.72 on D 1/2 to 0.54 on D 7/8. Accordingly, alluvial plots depicting the concordance of normal results between parameters (categorized as below or above their respective normal thresholds: 4500 RFV for IFN-γ release, as provided by the manufacturer, and 150 cells/µL for CD8 counts, as defined by the immunology laboratory) demonstrated a progressive loss of concordance over time, decreasing from 80% on D 1/2 to 59% on D 5/7 (Fig. 3). Correlations between IFN-γ secretion and immunological parameters. A. Correlation matrix illustrating the relationships between delta IFN-γ production, mHLA-DR expression, and T-lymphocyte subsets at all time points B. Scatterplots showing the correlation between delta IFN-γ production and CD8 T cells at each time point across the monitoring period (Spearman’s rho correlation). Number of samples at each time point as follows: D 1/2 (n = 60), D 3/4 (n = 49), D 5/7 (n = 33) Concordance between IFN-γ secretion and CD8 throughout the monitoring period. Alluvial plots illustrate the concordance between results considered normal for both parameters: normal or decreased IFN-γ production stratified according to an RFV threshold of 4500 (reference value provided by the manufacturer), and normal or low CD8 T-cell counts stratified according to a threshold of 150 cells/µL (reference value established by the immunology laboratory). Number of samples at each time point as follows: D 1/2 (n = 60), D 3/4 (n = 49), D 5/7 (n = 33) We next investigated associations with clinical deterioration (either mortality and/or ICU-acquired infections). None of the immune parameters, including mHLA-DR, NLR, or any lymphocyte counts, showed significant differences with respect to clinical deterioration in this cohort (Table 2). Similarly, IFN-γ release did not differ between patients with favorable outcomes and those who deteriorated (Table 3). Of note, similar results were observed when mortality and infection occurrence were analyzed separately (Figure S2). Consistent with this, ROC analyses did not show strong predictive value of this marker in regard with clinical deterioration (Figure S3). The largest difference between groups was observed at D 1/2, but it did not reach statistical significance (p = 0.086); the AUC at D 1/2 was 0.63 [0.48–0.77] (p = 0.085). At this time point, patients who later presented with unfavorable outcomes tended to have lower IFN-γ release values. Considering the clinical importance of persistent immunosuppression after septic shock, we explored whether the combination of impaired lymphocyte function and decreased mHLA-DR could identify a more severe phenotype. Indeed, at the last time point (D 5/7), patients with both low IFN-γ releases and low mHLA-DR median values were at significantly higher risk of clinical deterioration (Fig. 4). IFN-γ secretion and mHLA-DR combined at D 5/7 are associated with clinical deterioration in septic shock patients. Kaplan Meier analysis for 28-day worsening (either ICU-acquired infection or mortality) between severe patients with low IFN-γ secretion (below median) and low mHLA-DR (below median) and the rest of the cohort As an additional analysis, we focused on values obtained on D1/2 under nil conditions (i.e., incubation without stimulation), identifying patients with IGRA values > 0.08 IU/mL, reflecting early elevated circulating IFN-γ levels independent of stimulation (see Figure S4), and assessed whether this was associated with clinical deterioration. Discussion The main finding of the current study is to demonstrate the practicality of a fully automated functional immunological test under routine-like conditions and to confirm that it accurately reflects impaired IFN-γ release, which is one of the characteristic features of sepsis-induced immunosuppression. The latter represents a major clinical challenge, contributing to secondary infections, prolonged ICU stays, and increased mortality [3, 18]. Identifying reliable biomarkers to monitor immune dysfunction is essential for guiding patient management and designing targeted immunomodulatory interventions [5, 19]. Ideal markers should combine several key characteristics: ease of use at the bedside, rapid and reproducible results, and widespread accessibility across clinical settings. So far, in the field, mHLA-DR is considered as a kind of reference marker [3, 6]. However, its measurement is based on flow cytometry which is not accessible in every hospital setting. Furthermore, the complexity of sepsis immunopathology likely necessitates approaches that capture not only cell numbers but also functional competence, highlighting the need for functional assays that can provide clinically actionable insights. Among functional immune testing these approaches, ELISpot assays, particularly those developed and extensively studied by Hotchkiss and colleagues, have provided some of the most compelling evidence of functional impairment in sepsis [12, 20–23]. By quantifying cytokine secretion at the single-cell level, ELISpot captures the actual functional capacity of lymphocytes, rather than relying solely on phenotypic markers. These studies have demonstrated that impaired T-cell function correlates with disease severity and may predict susceptibility to secondary infections. Similarly, in mixed ICU patients, stimulated whole blood assays assessing IFN-γ release upon TCR stimulation demonstrated associations with the occurrence of secondary infections and patient deterioration [24]. On the myeloid side, it is also worth noting that a test assessing TNF release upon LPS stimulation has been used to enroll patients in trials evaluating GM-CSF [13]. However, these approaches (ELISpot and cytokine release assay) are typically based on manual, home-made protocols, are technically demanding, time-consuming, and require specialized laboratory personnel, limiting their applicability in routine ICU practice. There is thus a strong incentive to develop functional assays that are automated, rapid, and feasible directly at the point of care. Our study confirms the expected profound decrease in IFN-γ release among patients with septic shock, demonstrating the robustness of this functional readout. As anticipated, IFN-γ release values showed a logical correlation with absolute CD8 counts [25], though the relationship was imperfect, underscoring that cellular quantity and functional competence may provide complementary information. Despite these clear functional alterations, the prognostic value of IFN-γ release was limited, as the results alone did not significantly differentiate patients who experienced clinical deterioration from those with favorable trajectories. This lack of predictive power may be attributable to the relatively small sample size and the overall high severity of illness in our cohort, which could homogenize the risk profile and mask subtle differences. Furthermore, it should be noted that, in this cohort, none of the immunological parameters were associated with clinical worsening. Nevertheless, combining IGRA (a reflection of adaptive immunity) with mHLA-DR (a reflection of myeloid immunity) at later time points (D5/7) identified a subgroup of patients at increased risk of unfavorable outcomes. This finding warrants further investigation to determine how IGRA may complement the well-established information provided by mHLA-DR. Although associations with clinical outcomes were limited and require further investigation before any immediate clinical utility can be established, these results nonetheless provide compelling evidence of profound lymphocyte dysfunction in septic shock. The assay is performed on whole blood, requires less than four hours to yield results, and is fully automated, with no technician intervention required. No technical issues, including alarms or machine malfunctions, were observed during the study, confirming the method’s robustness and reliability. To our knowledge, this represents the first fully automated functional lymphocyte assay that could be suitable for ICU implementation, offering the potential for real-time monitoring at the bedside. Pending clinical validation in additional cohorts, these characteristics make it a highly attractive tool for further investigation in broader clinical contexts, including less severely ill patients, emergency care, and trauma, and in larger cohorts. Future studies should also consider integrating this functional readout with additional immune biomarkers to develop composite scores capable of stratifying patients for personalized immunomodulatory strategies. Several limitations should be acknowledged. The sample size was relatively small, limiting statistical power, and the cohort consisted of patients with septic shock, which may have masked the discriminative ability of the assay. Additionally, we could not explore alternative stimulation conditions (other stimulant, dose-effect, various incubation times) which might provide further insight into immune competence. Conclusion After one year of implementation, the assay demonstrates suitability for routine clinical use, as it uses whole blood, requires no specialized technician, provides results within four hours, and could be implemented directly in the ICU, enabling 24/7 availability if required. These practical advantages, combined with its ability to detect profound lymphocyte dysfunction, support further investigation of the technique in broader clinical settings. Indeed, before any bedside use in the ICU, the next critical step is to evaluate its clinical utility in larger patient cohorts. This will primarily involve determining whether it is independently associated with adverse outcomes and assessing how it can be integrated with other markers of immunosuppression. Upon clinical validation, this approach, by enabling timely and functional immune monitoring, holds the potential to bridge the gap between mechanistic insights and actionable clinical decision-making in sepsis care. Supplementary Information Additional file 1. Acknowledgements Not applicable. Author contributions TL performed the statistical analyses, interpreted the data, and drafted the manuscript. GM designed the study and contributed to manuscript drafting. MG supervised the study. MC, LA, and PP enrolled patients. ACL designed the study and enrolled patients. AS contributed to data collection and interpretation of the results. FV designed and supervised the study. FB designed the study. All authors contributed to manuscript revision and approved the final version for publication. Funding This work was supported by the intramural program of EA7426 (Hospices Civils de Lyon, University Lyon-I, bioMérieux) and Hospices Civils de Lyon. VIDAS 3 and related reagents were provided by bioMérieux within the framework of the joint unit EA7426. The company had no role in the study design, data analysis, interpretation of the results, or manuscript preparation. Data availability All data supporting the findings of this study are available within the paper and its Supplementary Information. Declarations Ethical approval and consent to participate All patients belonged to the IMMUNOSEPSIS study registered at clinicaltrials.gov (NCT04067674). Because of the non interventional nature of the study, written informed consent was not required for inclusion. However, each clinician orally informed the patient or family members about the objectives, methodology, and conduct of the study, and provided them with a leaflet using plain language. Non opposition to participation to this study was systematically obtained from the patient or a third party before any blood sampling was performed and was recorded in patient’s clinical file. All procedures performed were in accordance with the ethical standards of the institutional and/or national research committee (Comité de Protection des Personnes Sud-Ouest II, n° RCB : 2019-A00210-57, n°CPP : 19.01.23, 71857) and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Clinical and biological databases were reported to the National Commission for Information Technology and Freedom (CNIL, number 08-27). Consent for publication Not applicable. Competing interests FB is an employee of bioMérieux. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Additional file 1. Data Availability Statement All data supporting the findings of this study are available within the paper and its Supplementary Information.