Early Circulating Lymphocyte Apoptosis in Human Septic Shock Is Associated with Poor Outcome

Abstract

The role of lymphocyte apoptosis in septic shock remains a controversial issue. Using Annexin V and flow cytometry analysis on freshly isolated cells, we evaluated circulating lymphocyte apoptosis in 23 septic shock, 25 sepsis without shock, 7 nonseptic critically ill, and 25 control patients. In patients with sepsis, we compared day 1 lymphocyte apoptosis (i.e., within 3 days of the onset of infection) with that observed 5–7 days after (day 6) according to shock state, mortality, and severity factors. At day 1, patients in septic shock exhibited higher lymphocyte apoptosis than that present in controls (16.5% ± 3.5% vs. 3% ± 0.5%, respectively, P = 0.0001). At day 6, patients with sepsis without shock restored undamaged CD4⁺T and CD8⁺T lymphocyte counts, whereas patients in septic shock increased only CD4⁺T cells. Similarly, survivors restored undamaged lymphocyte count at day 6 (+70%, P < 0.001), whereas nonsurvivors did not. Day 6 undamaged lymphocyte count negatively correlated with day 1 SAPS II, day 6 LOD score, mechanical ventilation, and ICU stay duration. We observed no apoptotic effect of septic shock plasma or septic shock circulating mononuclear cells on target lymphoid cell lines. We found no alteration in any death receptors Fas, TRAIL-R1, TRAIL-R2, or in their ligands on circulating blood cells. Catecholamines and interleukin 10 levels significantly increased in patients with septic shock, but did not correlate with apoptosis levels. We conclude that lymphocyte apoptosis is rapidly increased in blood of patients in septic shock and that lymphocyte apoptosis leads to a profound and persistent lymphopenia associated with poor outcome. These results suggest that lymphocyte apoptosis is one of the main components of human septic shock immune dysfunction and could be related more to microcirculatory disturbance than to circulating factors.

INTRODUCTION

Septic shock is a life-threatening disorder and represents the first cause of mortality in adult intensive care units, ranging from 40%–60%. Despite continuing progress in clinical management, a recent report showed only a slight improvement of the outcome. A current hypothesis is that initial supportive care and antibiotic administration remain insufficient in overriding the deep alteration in immune status leading to multiorgan failure and secondary infection. The first state of this syndrome is described as a systemic hyperinflammatory immune response (SIRS) (¹), contributing in some cases to organ failure and early death. However, if patients survive this initial aggression, they enter an immune deactivation state termed compensatory anti-inflammatory response syndrome (CARS). This state is marked by a severe immune suppression, including myelomonocytic cell deactivation, lymphopenia (²), and lymphocyte dysfunction (³).

Beside the loss of functional activity of immune cells, the role of lymphocyte apoptosis in the immune dysfunction of sepsis has been recently questioned. In septic animal models, apoptosis has been shown to be a mechanism of lymphocyte death (^4–6). Interestingly, recent reports have demonstrated that inhibition of lymphocyte apoptosis improved survival in septic animal models (^7,8). In humans, two recent studies exhibited diffuse apoptosis in lymphoid organs and spleen of patients with septic shock (^9,10), but these data are mainly post mortem findings and described late phenomena in septic shock course. Consequently, the actual impact of lymphocyte apoptosis on human septic shock evolution remains unknown. Cellular mechanisms involved in septic shock apoptosis are currently under investigation. The role of death receptors such as Fas/CD95 had been evoked in experimental or human sepsis (^11,12), but others such as TNF-R1 or TNF-related apoptosis-inducing ligand receptors (TRAIL-R1 and TRAIL-R2) have not been studied at present time. Downstream, caspase activation had been evidenced in different tissues (^9,10), and caspase inhibitors prevent lymphocyte apoptosis and protect mice from sepsis-induced death (⁷). Compensatory mechanisms such as mitochondrial antiapoptotic protein Bcl2 were reduced in lymphocytes from septic animals (¹³), and Bcl2 hyperexpression reduced mice sepsis-induced mortality (^8,14). However, upstream of these mechanisms, the initial triggering factors are not elucidated. Several candidates are suspected. In septic shock, catecholamines are released in high concentrations into the systemic circulation. Catecholamines have important regulatory functions on NF-κB activation (¹⁵), which plays a pivotal role in inflammatory process (¹⁶) and in cell survival. Furthermore, immune cell apoptosis could be modulated directly by adrenergic agents (¹⁷). In the same manner, cytokines such as interleukin (IL) 10 and IL-4 (¹⁸) may also act on lymphocyte apoptosis.

In this report, we first evaluated early circulating lymphocyte apoptosis and its consequences according to time, evolution, and severity of infectious process. We then investigated putative apoptosis triggering factors by determining death-related protein expression; by measuring catecholamines, IL-10, and IL-4 levels; and by performing functional studies with plasma and mononuclear cells from patients with sepsis.

MATERIALS AND METHODS

Patient selection

The protocol was approved by the Institutional Review Board of the Rennes University Hospital. Apoptosis was evaluated in patients with septic shock, sepsis without shock, nonseptic critically ill with high adrenergic stimulation, and control patients. Patients under 18 years of age with malignancies who received chemotherapy, immunosuppressive agents, steroids, or β-adrenergic blockers were not included. We used standard definitions for sepsis and septic shock (¹). Patients were considered to have recent infection when a blood culture or normally sterile body fluid obtained within 3 days before or 3 days after the day of inclusion in the study grew pathogenic microorganisms or when sepsis manifestations were accompanied by an infectious focus requiring antibiotics. The Simplified Acute Physiology Score (SAPS II) at day 1 (¹⁹) and the Logistic Organ Dysfunction (LOD) system at day 6 (²⁰) were used to assess severity. Nosocomial infections were collected according to CDC definitions. In addition, as we suspected time variations of apoptotic process, we compared immune cell apoptosis in patients with sepsis at day 1 of the study with that observed 5–7 days after (day 6).

Samples

Venous EDTA samples (5 mL) were immediately processed for cell counting, and flow cytometry analysis was performed on freshly isolated cells. These samples were collected on day 1 for all patients and on day 6 for patients with sepsis. Heparinized blood samples (10 mL) were also collected at day 1, and after centrifugation, plasma was stored at −80°C and cells were stored at −196°C until assayed.

Apoptosis flow cytometry detection

Venous blood samples obtained at day 1 and 6 were immediately processed as followed: to avoid the loss of apoptotic cells, we did not use Ficoll gradient preparation or wash before staining. Thus, 100 μL of total blood was incubated 30 min at 4°C with 10 μL of the following phycoerythrin (PE)- and PE-cyanin 5 (PE-Cy5)-conjugated monoclonal antibodies (MoAb): CD3 (PE-Cy5 conjugated, clone UCHT1; Immunotech/Beckman-Coulter, Marseille, France), CD4 (PE, clone 13B8.2; Immunotech), CD8 (PE, clone B9.11; Immunotech), and CD19 (PE-Cy5, clone J4.119; Immunotech). After red blood cell lysis (2 mL of ORTHO-LYSIS reagent; Orthodiagnostic Systems, Raritan, NJ), cells were centrifuged, incubated for 10 min at room temperature with 2 μL of annexin V (Annexin V FITC; Immunotech), 7-amino-actinomycin D (7AAD; Via-Probe BD Biosciences), 100 μL of calcium-enriched buffer, and were immediately analyzed using flow cytometry (Cytoron Absolute ORTHO). Annexin V is a phospholipid-binding protein with high, selective affinity for phosphatidyl serine exposed on the outer leaflet of the cell membranes early after cells have entered apoptosis (²¹). The simultaneous use of annexin V and 7AAD allows us to discriminate late apoptosis (7AAD⁺) among the Annexin V-positive cell populations. The same protocol was used for all the patients. Electronic compensation was required to exclude overlapping of the three emission spectra. Multivariate data of 50,000 events were collected in list mode, stored, and analyzed using the Immunocount II software. Annexin V-positive cell percentage represents only the immediate apoptotic process; consequently, we calculated the number of undamaged cells that represents the result of the entire apoptosis process by subtracting the Annexin V-positive cell count from the total cell count.

Expression of death receptors and their ligand

Detection of molecules involved in lymphocyte apoptosis was performed on cells without any fixative reagent using direct immunofluorescence with PE-conjugated anti-Fas/CD95 MoAb (clone UB2; Immunotech), PE-conjugated anti-CD95-L/CD178 MoAb (clone B-R17; Diaclone, Besançon, France), and PE-conjugated anti-TRAIL MoAb (clone B-S23; Diaclone). Indirect immunofluorescence was used for TRAIL-R1 and TRAIL-R2 detection (respectively, clone HW101 and HW102; Alexis Laüfelfingen, Switzerland). Stained samples were analyzed using flow cytometry. Data were analyzed using CellQuest software (Becton-Dickinson, Mountain View, CA).

Cytokine quantification assay

IL-10 and IL-4 levels in the plasma of patients with sepsis were quantified at day 1 using a flow cytometric multibead assay (CBA; BD Biosciences, Mountain View, CA).

Catecholamine quantification assay

Epinephrine and norepinephrine levels were assessed using high-performance liquid chromatography (HPLC; Chromsystems, München, Germany).

Functional assays

Ex vivo patient plasma effect on cell lines

Patient plasma effect on lymphocyte survival was evaluated using the T lymphoid Jurkat cell line (American Type Culture Collection, Rockville, MD) (²²) highly sensitive to Fas-ligand- and TRAIL-induced apoptosis. Plasma effect on target cells was assessed in patients with septic shock and in control patients by culturing in 12-well plates (BD, Franklin Lakes, NJ) 5 × 10⁵ Jurkat cells in a mix of 800 μL of RPMI-1640 (Gibco, Paisley, UK) properly supplemented and 200 μL of plasma. Cells were incubated for 24 h at 37°C in a humidified atmosphere containing 5% CO₂. We did not wash the cells before staining, and apoptosis was evaluated by FITC Annexin-V and flow cytometry analysis. The absolute apoptotic rate was noted, and, as previously described (²³), specific apoptosis was calculated as follows: the percentage of specific apoptosis = 100 × [(% of Annexin V + cells in assay) – (% of Annexin V + cells in control)]/[100 - (% of Annexin V+ cells in control)].

Ex vivo patient cell effect on Jurkat T lymphoid cell line

The effects of patient mononuclear cells on lymphocyte apoptosis were assessed by coculturing mononuclear cells of control patients and patients with sepsis with target-labeled Jurkat cells. Jurkat cells were stained using PKH 26 red fluorescent cell linker kit (Sigma, St Louis, MO) and were diluted at 10⁶ cells/mL. These target cells were cultured in RPMI-1640 complete medium with increasing amounts of patient mononuclear cells used as effectors, obtained at day 1 and previously stored at −196°C (the ratio of Jurkat cells to patient mononuclear cells: 1/1, 1/5, and 1/10). Cells were incubated for 24 h as previously described, harvested, and stained with Annexin V. Using flow cytometry gating, we selected cells exhibiting PKH 26 dye, and we analyzed their specific Annexin V binding.

Statistical analysis

Quantitative variables were expressed as means ± SEM. Differences between control, sepsis, septic shock, and nonseptic critically ill patients were analyzed using one-way analysis of variance (ANOVA). The post hoc test used was Scheffe's test. For values recorded in the subgroup of patients with sepsis on day 1 and day 6 (i.e., sepsis and septic shock), the Mann-Whitney U test and Wilcoxon matched-pairs signed rank test were used for comparisons when appropriate. The relationships between two continuous variables were analyzed using Spearman's rank correlation test. A P value <0.05 was considered significant.

RESULTS

Clinical characteristics

Fifty-five patients who fulfilled the inclusion criteria entered the study, including 23 patients with septic shock, 25 patients with sepsis without shock, and 7 critically ill patients without sepsis (four myocardial infarctions, one cardiac arrest, one pulmonary embolism, and one postoperative cardiac tamponade). They were compared with 25 control patients without any shock or sepsis (two chronic obstructive pulmonary diseases, four preoperative patients, two suicide attempts, and 17 healthy volunteers). General characteristics of patients are shown in Tables 1 and 2. Among the 48 patients with sepsis included in the study, 10 died after the first evaluation and eight were discharged in another institution before day 6. Consequently, only 30 patients with sepsis were evaluated at day 6; among them, 24 survived and 6 died between day 6 and day 57. Sepsis origins and microbiological data are shown in Table 3.

Table 1: Patient characteristics

Table 2: Severity characteristics of patients with sepsis

Table 3: Sources of infection and microorganisms isolated in population with sepsis

Early detection of lymphocyte apoptosis

The percentage of Annexin V-positive lymphocytes was increased in patients with septic shock when compared with controls. On the contrary, we did not observe any 7AAD-positive cells (i.e., late apoptosis) in circulating blood. Annexin V staining was low in the control group, a fact that confirms the specificity of this marker for apoptosis detection.

In patients with septic shock, the initial apoptotic lymphocyte percentage was increased by more than 5-fold when compared with the control group, and more than 2-fold when compared with the sepsis group or with the nonseptic critically ill group (Fig. 1). A wide range of apoptotic percentage values was observed in septic shock (1%–58%) contrary to that observed in control (1%–9%). One of the most extreme cases with a CD3⁺T lymphocyte apoptotic rate reaching 58% of the total CD3⁺T cells and initial lymphopenia (800/mm3) is represented in Figure 2. In contrast, in sepsis without shock and in nonseptic critically ill groups, value ranges were more limited (2%–18% and 1%–3%, respectively), and we observed a moderate and nonsignificant increase in lymphocyte apoptotic rate when compared with the control group. The level of undamaged circulating lymphocytes in septic shock was lower than that in control patients (749 ± 140 cells/mm³ vs. 1681 ± 131 cells/mm³, respectively, P < 0.0001), and 17 of them (73%) exhibited lymphopenia with an undamaged lymphocyte count below the lower limit of our laboratory (i.e., 1000/mm³). As the apoptotic process could have differently affected lymphocyte subsets, we evaluated the apoptotic cell percentage in lymphocyte subpopulations (i.e., CD3⁺T, CD4⁺T, CD8⁺T, and CD19⁺B cells). Results are shown in Figure 3. Whatever the clinical condition, the apoptotic rate was comparable in T lymphocyte subpopulations and B lymphocytes. Consequently, the initial drop in the undamaged lymphocyte count observed in septic shock was similar in all lymphocyte subpopulations, increasing between −50% and –70% when compared with control.

Fig. 1: Apoptotic lymphocyte percentages (means ± SEM) in control, septic, septic shock, and nonseptic critically ill patients. In patients with septic shock, initial apoptotic lymphocyte percentage (16.5% ± 3.5%) was increased by more than 5-fold when compared with the control group (3% ± 0.5%, * P = 0.0001) and more than 2-fold when compared with the sepsis group (7.5% ± 1%, ^† P < 0.05) or with the nonseptic critically ill group (7.5% ± 1.5%, ns).

Fig. 2: Characteristic cytograms of annexin V staining (horizontal axis) in CD⁺ T lymphocytes (vertical axis). Characteristic control, septic, and septic shock patients are presented. An increase in the percentages of early-stage apoptotic (upper right quadrant) CD3⁺T lymphocytes was observed in patient with sepsis and above all in patient with septic shock (control: 5%, patient with sepsis: 10%, and patient with septic shock: 58%).

Fig. 3: Apoptotic cell percentages (means ± SEM) in control, septic, septic shock, and nonseptic critically ill patients according to lymphocyte subsets. Comparison of apoptotic cell percentages in CD3⁺T, CD4⁺T, CD8⁺T, and CD19⁺B cells at day 1 evidences an increase in apoptotic rate in septic shock when compared with control patients (* P < 0.01) and when compared with patients who had sepsis without shock (^† P < 0.05).

Evolution of lymphocyte apoptosis

Thirty patients with sepsis were evaluated at day 6. In patients with septic shock (n = 12), the apoptotic rate tended to decrease at day 6 (15% ± 4% at day 1 vs. 8% ± 1.5% at day 6, P = 0.072), and the undamaged lymphocyte count increased slightly (576 ± 138 cells/mm³ to 825 ± 160 cells/mm³, +43%, P < 0.05), but remained below the lower limit of 1000 cells/mm³. In contrast, in patients with sepsis without shock (n = 18), the apoptotic lymphocyte percentage on day 1 and 6 remained low and comparable (8% ± 1% and 7% ± 1%, respectively) and there was a large rise in the undamaged lymphocyte count between day 1 and day 6 (from 685 ± 95 cells/mm³ to 1193 ± 167 cells/mm³, +74%, P < 0.01). In fact, as shown in Figure 4, the late consequences of apoptotic process were different according to the T lymphocyte subtypes. In septic shock, lymphocyte restoration was slight and was observed only in CD4⁺T cell subtype, whereas CD8⁺T lymphocytes did not increase and remained lower on day 6 than that observed in patients with sepsis without shock. In contrast, in patients with sepsis without shock, there was a large restoration in undamaged T cell count, including both CD4⁺T cell and CD8⁺T cell subtypes between day 1 and day 6. However, even in this group, the observed improvement reached only about 75% of the control CD4⁺T cell subtype and 50% of the CD8⁺T cell subtype.

Fig. 4: Lymphocyte count evolution at day (means ± SEM) in patients with sepsis without shock and patients with septic shock according to lymphocyte subsets. In patients with sepsis without shock (S), the day 6 undamaged cell count exhibited a large restoration in viable CD3⁺T cells when compared with the day 1 undamaged cell count (** indicates +80%, P < 0.01), including both CD4⁺T (** indicates +99%, P < 0.01) and CD8⁺T (** indicates +73%, P < 0.01) subtypes. In contrast, in patients with septic shock (Ss), the increase in undamaged CD3⁺ T cells (+50%) was not significant and was only due to a slight increase in the CD4⁺T cell count (* indicates +65%, P < 0.05), whereas undamaged CD8⁺T lymphocytes did not increase and remained lower on day 6 than that observed at day 6 in patients with sepsis without shock (^† P < 0.05).

Lymphocyte apoptosis is associated with poor outcome and death

Regarding the patients followed after day 6, the undamaged lymphocyte count noted at day 1 was similar in survivors and nonsurvivors, but the change between day 1 and day 6 showed a major difference in the ability to restore the initial lymphopenia. In survivors, the undamaged lymphocyte count increased from 655 ± 93 cells/mm³ at day 1 to 1122 ± 141 cells/mm³ at day 6 (P < 0.001). In contrast, in nonsurvivors, the undamaged lymphocyte count remained low at day 6 (740 ± 199 cells/mm³) comparable with value observed at day 1 (590 ± 130 cells/mm³). The ability to restore the lymphocyte population was analyzed according to lymphocyte subtype (Table 4). In survivors, there was a significant increase in the undamaged T cell count (+86%) between day 1 and day 6, including both CD4⁺T cell (+108%) and CD8⁺T cell (+64%) subtypes. In contrast, in nonsurvivors, this increase was not observed and, interestingly, the undamaged CD8⁺T lymphocyte count remained lower on day 6 than that observed in survivors (P < 0.05). Alterations observed in B cells were not significant in either group.

Table 4: Evolution of undamaged cell (annexin V negative cells) count (mean ± SEM) according to lymphocyte subsets in survivors and nonsurvivors patients with sepsis

To evaluate relationships between the effect of the apoptotic process and disease severity, we investigated correlations between the undamaged lymphocyte count at day 6 in patients with sepsis (n = 30) and several severity markers (Fig. 5). There was a negative correlation between the undamaged lymphocyte count at day 6 and SAPS II (ρ = −0.511, P < 0.01), LOD score at day 6 (ρ = −0.474, P < 0.05), ICU stay duration (ρ = −0.679, P < 0.001), and mechanical ventilation duration (ρ = −0.740, P < 0.0001). Moreover, 11 late nosocomial infections occurred secondarily in seven patients among the 48 patients with sepsis; in all of the cases, the initial undamaged lymphocyte count was deeply lowered (460 ± 63 cells/mm³), and six of them were evaluated at day 6 and exhibited a persistent, significantly low undamaged lymphocyte count when compared with patients without nosocomial infection (412 ± 152 cells/mm³ vs. 1233 ± 131 cells/mm³, respectively, P < 0.01)

Fig. 5: Relationship between lymphocyte loss at day and severity indicators. SAPS II (A), LOD score at day 6 (B), mechanical ventilation duration (C), and ICU stay duration (D) are negatively correlated with remaining undamaged lymphocyte count at day 6.

Role of cell factors on cell death

Cell expression of death receptors and their ligands

The Fas/CD95-positive T lymphocyte percentage at day 1 in patients with septic shock (43% ± 4%, n = 22) was not increased when compared with controls (40% ± 2%, n = 25), or with respect to patients with sepsis without shock (43% ± 2.5%, n = 25) and to nonseptic critically ill patients (48% ± 4%, n = 7). The Fas-positive T lymphocyte percentage expression was not related to mortality or to severity markers. We did not find TRAIL-R1 and TRAIL-R2 expression among patients with septic shock (n = 6; 3.2% ± 1.5% and 0.4% ± 0.1%, respectively). Lastly, the death-ligand membrane molecules Fas-L/CD178 and TRAIL expression on lymphocytes of septic shock (n = 6) remained not detectable and comparable with control patients (n = 3; data not shown).

Ex vivo patients mononuclear cell effect on target Jurkat cells

The effect of circulating mononuclear cells on Jurkat cell apoptosis is shown in Figure 6. The absolute apoptosis rate of Jurkat cells ranged between 23% (basal level) and 90% (control agonistic anti-Fas/CD95 MoAb CH11-induced apoptosis level). The absolute apoptosis rates of Jurkat cells cocultured with mononuclear cells of patients with sepsis (n = 4) or control patients (n = 3) were comparable irrespective of the ratio of mononuclear cells/Jurkat cells.

Fig. 6: In vitro effect of patient mononuclear cells on Jurkat cell apoptosis. Similar effects of control (C; n = 3) and septic shock (Ss; n = 4) patient circulating mononuclear cells on the apoptosis sensitive T Jurkat cell line at various effector:target (E/T) cell ratios are presented. Compare column 3 vs. 4 for E/T cell ratio 1/1, column 5 vs. 6 for E/T ratio 5/1, and column 7 vs. 8 for E/T ratio 10/1. An increase in apoptotic Jurkat cell percentage was related to an increase in the E/T ratio and was comparable in control and patients with septic shock. This apoptotic effect was similarly abrogated by the antagonist peptide DEVD in control and patients with septic shock (column 9 and 10). CH11-induced Jurkat cell apoptosis is shown as internal control (column 2).

Role of soluble factors on cell death

Circulating IL-10 and IL-4 levels

IL-4 levels were below the detection threshold (i.e., <6 pg/mL) in most of the patients with sepsis with and without shock (0.8 ± 0.3 pg/mL and 1.9 ± 0.5 pg/mL, respectively). IL-10 levels did not correlate with catecholamine levels or with the apoptotic lymphocyte percentage or undamaged cell count at day 1. Despite this fact, IL-10 levels were significantly increased in the group with septic shock when compared with the group with sepsis (414 ± 230 pg/mL vs. 40 ± 24 pg/mL, P = 0.0001).

Circulating catecholamine levels

As expected, initial catecholamine levels were significantly increased in septic shock and in nonseptic critically ill patients when compared with the control group or the group with sepsis (Fig. 7). Regarding lymphocyte apoptosis, catecholamine levels did not correlate with the apoptotic cell percentage or with the initial undamaged cell count (data not shown).

Fig. 7: Circulating catecholamine levels. Plasma catecholamine levels (means ± SEM) were significantly increased in septic shock and in nonseptic critically ill patients when compared with the control or with the sepsis without shock group (* P < 0.05 vs. control and sepsis without shock groups, ** P < 0.01 vs. control, and < 0.05 vs. sepsis without shock).

Effect of septic shock plasma on the Jurkat cells

The specific apoptotic percentage of Jurkat cells observed with patient plasma was significantly lower in patients with septic shock (n = 13) when compared with control patients (n = 4; –7% ± 3.5% vs. + 3% ± 5%, respectively, P < 0.05). In other words, septic shock plasma did not induce Jurkat cell apoptosis, and, on the contrary, weakly protected Jurkat cells from spontaneous apoptosis.

DISCUSSION

Two major points are evidenced by the present study: Exaggerated lymphocyte apoptosis is present in peripheral blood of patients with septic shock, contrary to those with simple sepsis or critically ill nonseptic patients, and lymphocyte apoptosis occurs rapidly, leads to a profound and persistent lymphocyte loss, and is associated with poor patient outcome.

A high level of peripheral lymphocyte apoptosis was observed in patients with septic shock. Annexin V staining at day 1 clearly demonstrated the rapid occurrence of this mechanism; in contrast, the negativity of later markers such as 7AAD could be explained by a rapid clearance of apoptotic cells by the immune system itself. Interaction between surface phosphatidylserine of damaged cells and macrophages could be one of the main mechanisms involved in the in vivo apoptotic cell clearance (²⁴). Lymphocyte apoptosis was previously described in animals and in human post mortem tissue examination (^4,9). The present report demonstrates the early occurrence of lymphocyte apoptosis in human septic shock and evidences the link between lymphocyte apoptosis, early lymphopenia, and patient outcome. Thus, more than 50% of patients with septic shock exhibited high initial lymphocyte apoptosis with values above the highest level observed in controls. As in other studies (⁹), a wide range of values was observed. Our data also evidences that the cell death process is associated with an early and persistent profound lymphopenia because the undamaged lymphocyte count was below 1000/mm3 in more than 70% of patients with septic shock at day 1 and day 6. To the contrary, the apoptosis rate observed in sepsis without shock was moderate and not significant when compared with control patients. A similar low apoptotic rate was observed in nonseptic critically ill patients. It is possible that the limited number of nonseptic critically ill patients might have lead us to underestimate lymphocyte apoptosis in this group, but the low range of the observed values (1%–13%) does not support this hypothesis. However, although sepsis without shock and critically ill nonseptic patients considered as groups did not demonstrate excessive lymphocyte apoptosis, it was clear that some of these patients had a moderate increase in apoptosis rate. Several factors such as ischemia/reperfusion injury, inflammatory processes, or coagulation disturbances encountered at low levels in these two situations could be extensively amplified when shock and sepsis occurred simultaneously and thus could lead to an excessive lymphocyte apoptosis. In the present study, initial apoptosis and lymphocyte loss equally affected the different T subpopulations and B-lymphocytes. These findings advocate more for the involvement of pathologically common signals such as reactive oxygen species or nitric oxide than for that of a well-regulated mechanism of lymphocyte population control. Moreover, because lymphocyte apoptosis was also described in spleen of patients with sepsis after their death (⁹), lymphocyte apoptosis appears not only as an early event, but also as a ubiquitous and prolonged phenomenon in septic shock. This persistent prolonged apoptosis explained not only the deep lymphopenia observed at day 6 in septic shock, but also the mild decrease in lymphocyte count observed in patients with sepsis without shock. In this case, one can hypothesize that the apoptotic cells were being cleared faster by the immune system and, as a consequence, were detectable in peripheral blood at a level barely superior to that observed in control patients.

The present study evidences the rapid occurrence of exaggerated lymphocyte apoptosis in human septic shock because apoptosis was already present at day 1, rapidly after the initial diagnosis of sepsis. Even if in the majority of patients the onset of sepsis was not truly known, the infectious diseases described in the present study were acute, and one can hypothesize that the excessive lymphocyte apoptosis observed at day 1 was not a late phenomenon in the disease. For example, one of the highest lymphocyte apoptotic rates (49%) exhibited by a patient with septic shock with values reaching 58% for the CD3⁺T cells (Fig. 1) was observed in a splenectomized woman only 2 h after the diagnosis and 8 h after the first symptoms of a fulminant Streptococcus pneumoniae septic shock. In this case, it was obvious that apoptotic signals were generated very early in the disease. In addition, the lymphopenia observed in this patient (800/mm³) suggests that apoptosis began before the evaluation and that apoptotic cell clearance process was already present. These data are in accordance with animal studies that showed that lymphocyte apoptosis was detectable as early as 4 h after inducing sepsis (²⁵). This point could be of importance in further clinical attempts to inhibit immune cell apoptosis and in treatment administration timing.

The question arises whether the unfavorable course of the septic state is related or not to lymphocyte apoptosis because the consequence of lymphocyte apoptosis remains a controversial issue (²⁶). Actually, the increase in lymphocyte apoptosis could be one of the components of the immune defect observed in septic shock or could be a useful mechanism in the control of the early proinflammatory response. The present study does not establish the direct responsibility of lymphocyte apoptosis in the poor outcome of patients with sepsis, but does give arguments in line with this hypothesis. A wide range in apoptotic rate (i.e., 1%–58% at day 1 and 1%–18% at day 6) was observed in the group with septic shock compared with the control group (1%–9%). Similar variations were found by others (⁹) in tissue specimens, with apoptotic cells rate ranging between 1% and 50%. These variations could reflect variations in septic injury intensity and could consequently suggest a relationship between lymphocyte apoptosis intensity and septic shock severity. In keeping with this hypothesis, the initial apoptosis percentage was higher in septic shock than in sepsis without shock and consequently was associated with a higher severity. Another argument is the fact that persistent lymphopenia was observed mainly in septic shock and in nonsurvivors. Moreover, we evidenced significant relationships between persistent lymphopenia and severity markers such as SAPS, LOD score, and consequences of initial severity such as ICU stay and mechanical ventilation duration. Perhaps the most significant finding observed in our study, which advocates for the deleterious role of lymphocyte apoptosis, is the association observed between late nosocomial infection and initial and persistent deep lymphocyte loss. To our knowledge, there are no data demonstrating direct causality between circulating lymphocyte apoptosis and worsening outcome in human septic shock. However, several experimental data are in line with this hypothesis: caspase inhibition or antiapoptotic factor Bcl-2 hyperexpression demonstrated the beneficial effect of lymphocyte apoptosis inhibition on survival (^7,8,14).

Two pathways have been suggested in septic lymphocyte apoptosis, i.e., the death receptor system (^11,12) and the mitochondrial pathway activated by various stimuli leading to caspase 9 activation. These two pathways subsequently act on final death cell program via caspase 3 cleavage (¹⁰). Furthermore, an attractive hypothesis could be a decrease in antiapoptotic factors expression such as Bcl2 (¹³) or Bcl-XL (²⁷), but whatever the mechanisms involved in the apoptotic cells, the initiating triggering signal was not evidenced. Here, we did not confirm in human the role of death receptor pathways or cell/cell interactions in circulating lymphocyte apoptosis, contrary to the study of Papathanassoglou (²⁸), which demonstrated a relationship between the high Fas and FasL expression on peripheral blood mononuclear cells and the high severity of multiorgan dysfunction in patients with SIRS. Other putative mediators such as catecholamines and IL-10 were not correlated to in vivo cell apoptosis and plasma did not increase spontaneous Jurkat cell death. These data tend to exclude the role of soluble factors such as FasL or of medications such as pressor or inotropic agents; in addition, we did not find relationship between such therapy and lymphocyte apoptosis. However, several limitations of our study have to be mentioned. We evaluated death receptor pathways only in peripheral T lymphocytes, but as it was previously described (^11,29), the Fas/FasL pathway could be involved in lymphoid organs; similarly, other death receptor pathways could be involved in lymphoid or hematopoietic tissues. Lastly, the apoptotic effect of peripheral mononuclear cells or plasma was evaluated using cell lines as target cells because they are highly sensitive to Fas- or Trail-induced apoptosis, but the results need to be confirmed using circulating lymphocytes of patients with sepsis. Nevertheless, our data can provide some insight into the mechanisms involved in this excessive lymphocyte apoptosis. The absence of apoptotic factor detectable in the peripheral plasma or on the mononuclear cells of the patients with septic shock is of importance and may direct future research toward the preeminent role of microcirculatory factors. We hypothesize that this microenvironment characterized by the presence of bacterial constituents, by the local release of immunosuppressive mediators, and by alteration in the endothelial cells may contribute in different ways to apoptosis induction. Very short half-time components linked to ischemia/reperfusion injury or to inflammatory processes such as reactive oxygen species (³⁰), nitric oxide (³¹), or PGE2 (³²) could act in the microcirculatory system or in tissues and could initiate the apoptotic process. Furthermore, endothelial factors activated by coagulation disturbances such as galectin expression (³³), cell adhesion molecule expression (³⁴), or decrease in protein C (³⁵) could be involved in this process.

Taken together, our findings could suggest that peripheral lymphocyte apoptosis is the consequence of an ubiquitous phenomenon related to the severity of septic shock, and therefore, of the microcirculatory disturbance, and that it is initiated in organs rather than in the peripheral bloodstream.

Lymphocyte apoptosis represents an early component of injury in septic shock that is observable not only in post mortem tissue examination, but also in peripheral blood and at the onset of human septic shock. Furthermore, our observations suggest that lymphocyte apoptosis leads to a deep lymphocyte loss and plays a detrimental role in the evolution of septic shock.

ACKNOWLEDGMENTS

The authors thank Prof. Alain Leguerrier and his team from Thoracic and Cardiovascular Surgery Unit, and Prof. Philippe Mabo and Dr. Gilles Rouault from Cardiovascular Diseases Department for their assistance. This work was supported by grants from Société de Réanimation de Langue Française and the Medical University of Rennes.