Severe sepsis and septic shock are among the leading causes of death. It is estimated that almost 3 millions of cases appear annually in Northern America and in Europe; 35-50% of them die [1]. The abdomen is the second most common cause of sepsis in the European Intensive Care Units accounting for 26% of cases [2].

A general scheme of pathogenesis is well acceptable for all causes of sepsis. According to that scheme, sepsis is initiated when well-conserved microbial structures known as pathogen-associated molecular patterns (PAMPs) bind to receptors embedded either on the cell membranes or inside the cell cytoplasm of cells of the innate immune system, namely blood monocytes and tissue macrophages. These receptors are known as pattern recognition receptors (PRRs). Toll-like receptors (TLRs) are the best studied PRRs. Monomers of the peptidoglycan of the cell wall of Gram-positive cocci bind to TLR2 and lipopolysaccharides (LPS) of the outer membrane of Gram-negative bacteria bind to TLR4. The interaction of TLRs with PAMPs ends with the production of pro-inflammatory cytokines, like tumor necrosis factor-alpha (TNFα), interleukin (IL)-1β, IL-6 and IL-8. These pro-inflammatory mediators orchestrate septic reaction of the host. Soon after this first phase of hyper-production of pro-inflammatory mediators, a second phase ensues during which monocytes stimulated by PAMPs are no longer able to secrete a similar large amount of pro-inflammatory cytokines. Instead during this second phase a large amount of anti-inflammatory mediators like IL-10 are produced. This phase is considered a state of immunosuppression or immunoparalysis of the host when multiple organ dysfunctions take place [3]. Apoptosis of lymphocytes predominates during this stage and it is one of the main drivers leading to immunoparalysis [4].

However recent data of our group render questionable whether the above simplistic scheme may be generalized for severe sepsis/shock supervening in the field of all causes of sepsis. More precisely, flow cytometric analysis of monocytes and of lymphocytes was done within the first 24 hours upon diagnosis in 505 patients; 100 suffered from intra-abdominal infections [5]. Results revealed that advent of severe sepsis/shock in the event of abdominal sepsis differed considerably compared with sepsis originating from other sites. This was related with lower expression of HLA-DR on CD14-monocytes which is an index of immunoparalysis; greater counts of CD8-lymphocytes; and greater apoptosis of CD8-lymphocytes compared with other types of infections.

Based on the latter results it may also be hypothesized that all causes of abdominal sepsis are not similar in their ability to stimulate the host's immune response. Extensive work over the last two decades was done trying to simulate human abdominal sepsis with various animal models. The most widely applied models are that of intraperitoneal challenge with LPS or live bugs and that of cecal ligation and puncture (CLP) [6, 7]. The present study aimed to investigate the importance of the apoptosis of lymphocytes early in the course of experimental peritonitis for the final outcome. The study used a model of intra-abdominal infection after CLP which mimics acute polymicrobial infections occurring in humans.

Mean ± SE apoptosis of lymphocytes of the four sham-operated rabbits and of the 31 rabbits with induced peritonitis were 2.4 ± 0.9% and 25.3 ± 3.8% respectively at 4 hours (p: 0.007); 1.1 ± 0.6% and 19.5 ± 4.6% at 24 hours (p: 0.037); and 2.4 ± 0.7% and 21.2 ± 4.3% at 48 hours (p: 0.032). Respective apoptosis of monocytes were 36.9 ± 10.8% and 42.7 ± 4.3% at 4 hours (p-non significant); 13.7 ± 2.9% and 41.6 ± 5.4% at 24 hours (p: 0.042); and 27.8 ± 11.8% and 45.9 ± 6.1% at 48 hours (p: 0.039).

ROC analysis was performed within rabbits subject to the first set of experimental peritonitis for apoptosis of lymphocytes and of monocytes at 4 hours and at 24 hours as predictors of death. The only analysis yielding statistical significance was apoptosis of lymphocytes at 4 hours (AUC: 0.731, 95%CI: 0.544-0.918, p: 0.028). Apoptosis of lymphocytes lower than or equal to 32% had specificity 80% to predict death (Figure 2). This specificity cut-off has been considered of significance for biomarkers to define an event in sepsis studies [12].

Animals were divided into two groups according to that finding: i) those with lymphocyte apoptosis less than or equal to 32%. This was found in 18 animals; and ii) those with lymphocyte apoptosis greater than 32% (Figure 2). This was found in 13 animals. Thirteen (72.2%) and four (20.8%) of them died respectively (p: 0.017, Figure 3).

We hypothesized that at least two explanations may exist behind that phenomenon: a) apoptosis of lymphocytes of the first 4 hours may influence bacterial replication in tissues; and/or b) apoptosis of lymphocytes may determine death by an immune-related mechanism.

Blood cultures were sterile. Bacterial growth was measured in the liver, spleen, lung and kidneys of animals after death or sacrifice. No significant correlation was found between time to death and tissue bacterial growth in any tested organ. Bacterial species isolated from tissues of animals were Escherichia coli, Enterobacter cloacae and enterococci. No differences of tissue bacterial growth were found between animals with apoptosis of lymphocytes at 4 hours lower than or equal to 32% and animals with apoptosis of lymphocytes at 4 hours greater than 32% in the liver, in the spleen and in the kidney. However, bacterial growth in the lung was greater within animals with apoptosis of lymphocytes at 4 hours greater than 32% (Figure 4).

The impact of lymphocyte apoptosis on the immune response of the host involved a study on a) monocytes apoptosis; and b) the innate immune response. Apoptosis of lymphocytes and of monocytes over follow-up is shown in Figure 5. At that analysis animals are divided into two groups; those with lymphocyte apoptosis ≤32% at 4 hours and those with lymphocyte apoptosis > 32% at 4 hours. At baseline i.e. before induction of peritonitis, the two animal groups did not differ. However, during the entire follow-up after induction of peritonitis, apoptosis of lymphocytes and apoptosis of monocytes were significantly greater in the group of animals with lymphocyte apoptosis > 32% at 4 hours (Figure 5).

TNFα produced by PBMCs of these two groups after stimulation with LPS and Pam3Cys generally did not differ with the exception of 48 hours. At that time point, TNFα stimulated by Pam3Cys was greater within the group of animals with lymphocyte apoptosis ≤32% at 4 hours. Serum TNFα differed at 4 hours after induction of peritonitis; it was greater among the group of animals with lymphocyte apoptosis ≤32% at 4 hours (Figure 6).

One question needing clarification was the cause of animal death. Although post-mortem histology was not done, tissue bacterial growth of intraperitoneal organs i.e. liver and spleen was much greater among non-surviving animals at 7 days than among surviving at 7 days. More precisely, mean ± SE log₁₀ of bacteria in the liver of non-surviving animals at 7 days were 3.81 ± 0.75 cfu/g; among surviving animals at 7 days they were 1.98 ± 0.67 cfu/g (p: 0.045). Respective counts of bacteria in the spleen were 4.48 ± 0.96 cfu/g and 1.86 ± 0.61 (p: 0.031). However, tissue bacterial growth of extraperitoneal organs i.e. lung and kidney did not differ between non-surviving and surviving animals at 7 days. More precisely, mean ± SE log₁₀ of bacteria in the lung of non-surviving animals at 7 days were 2.78 ± 0.71 cfu/g; among surviving animals at 7 days they were 1.96 ± 0.56 cfu/g (p non-significant). Respective counts of bacteria in the kidney were 3.92 ± 0.88 cfu/g and 2.16 ± 0.63 (p non-significant). These findings allowed identify abdominal sepsis as the cause of animal death.

From the 20 untreated rabbits studied in the second set of experiments, apoptosis of lymphocytes at 4 hours was ≤32% in 14 rabbits and > 32% in six rabbits; 11 (78.6%) and two of them died (33.3%) respectively. This finding replicated results of the first set of experiments. From the 20 animals administered single shots of ciprofloxacin and metonidazole at 4 hours after induction of peritonitis, 11 animals had lymphocyte apoptosis at 4 hours before antibiotic infusion ≤32% and nine animals had lymphocyte apoptosis at 4 hours before antibiotic infusion > 32%. Four (27.4%) and three (23.3%) of them died respectively. Comparison of survival with the respective non-antibiotic treated animals is shown in Figure 7. It was evident that infusion of antibiotic prolonged survival of animals with lymphocyte apoptosis at 4 hours ≤32% but not of animals with lymphocyte apoptosis at 4 hours > 32%.

Sepsis is a syndrome characterized by great heterogeneity. Although all patients are ubiquitously characterized as septic, they greatly differ regarding gender, age, co-morbid conditions, underlying infections, and offending pathogens. On that sense, it may be hypothesized that intra-abdominal infections differ from other types of infections in the way they trigger sepsis cascade [13]. Some of them develop in the event of another intrabdominal sterile critical illness like acute pancreatitis. Some develop due to slow seeding of bacterial gut content in the circulation like an abscess or acute diverticulitis. Others develop due to the acute and abrupt leakage of large amounts of the bowel content like in acute peritonitis after gut rupture [6, 14, 15]. Although all these cases are polymicrobial in nature, they may differ in the type of stimulation of the immune system of the host.

A model of CLP is used in the present study attempting to evaluate the effect of bacteria load released from the gut in the event of intra-abdominal sepsis on the innate and adaptive immune responses of the host. The study of rabbits instead of rodents allows better reproducibility of the experimental conditions. Reproducibility is one major factor creating diversity in such type of experiments [7]. Rabbits are larger animals than mice. This limits a) diversities created by the surgical team within separate days of experiments; and b) the impact of sequential blood sampling on survival. Although recent evidence suggest that daily sampling of volumes as high as 35 μl in mice subject to CLP do not affect mortality [16], the usage of rabbits remains indispensable for the present study where daily sampling of volumes as high as 5 ml is necessary to collect the necessary amount of PBMCs.

Apoptosis of lymphocytes is conceived as sequelea of severe sepsis [17, 18]. In accordance with that an antibody against receptor PD-1 (programmed-death-1), which is a negative co-stimulatory molecule on T-lymphocytes, was administered in mice late after induction of CLP and sepsis. Treatment prevented apoptosis of T-lymphocytes and prolonged survival [19]. However, recent evidence shows that apoptosis of T-lymphocytes may take place early over the course of sepsis [20] and a new important role has been suggested for apoptosis of T lymphocytes to regulate the innate immune response [21].

Our findings corroborate with previous results showing that apoptosis of lymphocytes is taking place in sepsis since apoptosis of lymphocytes was greater in rabbits subject to CLP and sepsis than sham-operated rabbits at all times of sampling. However, the presented results indicate that the rate of lymphocyte apoptosis as early as 4 hours after induction of peritonitis may define final outcome. The importance of this novel concept is underscored by the efficacy of treatment outcome. More precisely, administration of antimicrobials improved survival of animals that preserved their lymphocytes at 4 hours after induction of peritonitis. The efficacy of antimicrobials was limited in those animals with a great level of apoptosis. These findings may suggest that factors other than proper care impact on outcome from sepsis. However, it is obvious that the current findings support like never before the need for a very early start of antimicrobials and for early debridement of an intra-abdominal infection, as already suggested in the current treatment guidelines [22].

The mechanism of regulation of final outcome by apoptosis of lymphocytes taking place early over the time course of sepsis is difficult to interpret. According to the present findings, apoptosis of monocytes is also affected and this may influence phagocytosis. Other authors suggest that the rate of apoptosis of T lymphocytes in sepsis may have a direct impact on the process of phagocytosis [21].

Two main limitations of the present study should be addressed: a) the lack of commercially-available specific antibodies allowing discrimination of blood lymphocytes from blood monocytes in the rabbit. However, their physical properties of forward scattering and side scattering allow safe discrimination of these two populations; and b) the lack of measurement of the precise lymphocyte count at every time of sampling. Although this may be of importance, it could hardly affect the presented results that are focusing on the apoptotic process per se.

The precise trigger of lymphocyte apoptosis early over the course of sepsis is unknown. In a recent publication of our group, one probable explanation is described. More precisely, serum was isolated from 48 patients with severe sepsis/shock with the first 12 hours from advent of organ failures. When CD4-lymphocytes isolated from healthy volunteers were incubated with patients' serum, apoptosis was induced after activation of both the extrinsic and the intrinsic apoptotic pathways. This was not the case whenever CD4-lymphocytes were incubated with healthy serum. These findings may be interpreted in two ways: either an inhibitor of apoptosis circulates normally which is down-regulated early in severe sepsis; or a circulatory factor stimulating apoptosis appears early in severe sepsis [23].

The presented results revealed that lymphocyte apoptosis at an early time point of experimental peritonitis was a major driver for death. A lower percentage of lymphocyte apoptosis within the first 4 hours from the induction of the septic process led earlier to death. Antimicrobials could be beneficial even at that state underscoring the significance of early administration of antimicrobial therapy for the septic patient with intra-abdominal infection. These findings create the need for similar studies in other models of sepsis.