Perioperative lymphopenia has been linked with an increased risk of postoperative infectious complications, but the mechanisms remain unclear. We tested the hypothesis that bioenergetic dysfunction is an important mechanism underlying lymphopenia, impaired functionality and infectious complications. In two cohorts of patients (61–82 years old) undergoing orthopaedic joint replacement (n=417 and 328, respectively), we confirmed prospectively that preoperative lymphopenia (≤1.3 x 109·l−1; <20% white cell count; prevalence 15–18%) was associated with infectious complications (relative risk 1.5 (95% confidence interval 1.1–2.0); P=0.008) and prolonged hospital stay. Lymphocyte respirometry, mitochondrial bioenergetics and function were assessed (n=93 patients). Postoperative lymphocytes showed a median 43% fall (range: 26–65%; P=0.029; n=13 patients) in spare respiratory capacity, the extra capacity available to produce energy in response to stress. This was accompanied by reduced glycolytic capacity. A similar hypometabolic phenotype was observed in lymphocytes sampled preoperatively from chronically lymphopenic patients (n=21). This hypometabolic phenotype was associated with functional lymphocyte impairment including reduced T-cell proliferation, lower intracellular cytokine production and excess apoptosis induced by a range of common stressors. Glucocorticoids, which are ubiquitously elevated for a prolonged period postoperatively, generated increased levels of mitochondrial reactive oxygen species, activated caspase-1 and mature interleukin (IL)-1β in human lymphocytes, suggesting inflammasome activation. mRNA transcription of the NLRP1 inflammasome was increased in lymphocytes postoperatively. Genetic ablation of the murine NLRP3 inflammasome failed to prevent glucocorticoid-induced lymphocyte apoptosis and caspase-1 activity, but increased NLRP1 protein expression. Our findings suggest that the hypometabolic phenotype observed in chronically lymphopenic patients and/or acquired postoperatively increases the risk of postoperative infection through glucocorticoid activation of caspase-1 via the NLRP1 inflammasome.

CLINICAL PERSPECTIVES

  • The mechanism underlying a potential role for perioperative lymphopenia with increased risk for postoperative infectious complications is unclear.

  • We confirmed that preoperative lymphopenia is associated with infectious complications, and a hypometabolic phenotype driven by neurohormonal activation of the NLRP1 inflammasome.

  • These data suggest that bioenergetic insufficiency is a common, yet hitherto under-recognized, pathological mechanism that offers a new paradigm to reduce morbidity and enhance recovery from various acute inflammatory challenges.

INTRODUCTION

Immune dysfunction plays an integral role in the morbidity that frequently follows major trauma [1,2], yet the underlying mechanisms are poorly understood. Lymphocytes are essential for both effective innate and adaptive responses to trauma and inflammation [3,4]. Relative lymphopenia is observed commonly following tissue trauma and sepsis [5,6]. This occurs, at least in part, from redistribution of lymphocytes from the circulation to lymphoid tissue and sites of inflammation, and from loss of lymphocytes through apoptosis [5,7].

Activation of lymphocytes is critically dependent upon rapid increases in energy generation from both glycolysis and oxidative phosphorylation [8,9]. Environmental changes alter the bioenergetic capacity of lymphocytes, particularly during acute stress [10]. Inflammation is actively terminated by CD4+ T-cells, which restrain macrophage inflammasome-mediated caspase-1 dependent release of the pro-inflammatory cytokine IL-1β in a cognate manner [11]. Changes in the perioperative metabolic phenotype of lymphocytes may contribute to their critical role in immunomodulation at wound (and other) sites [12]. However, clinical translational data are sparse, and it remains unclear how the neurohormonal ‘stress’ response to surgery and trauma alters metabolic reprogramming in lymphocytes.

We therefore hypothesized that postoperative lymphopenia and impaired lymphocyte function are associated with a bioenergetically dysfunctional state, and that the resulting immunosuppression and increased rate of programmed cell death (apoptosis) would be associated with an increased risk of postoperative infections. To explore this hypothesis, we first characterized serial bioenergetic changes in lymphocytes sampled from patients undergoing elective joint replacement surgery, as this represents a standardized tissue trauma insult [1315]. On the basis of these findings, we then proceeded to explore the impact of chronic lymphopenia on clinical outcomes following surgery, and the interaction at the molecular level between lymphocyte mitochondrial dysfunction and a ubiquitous mechanism of stress that contributes to the postoperative metabolic phenotype.

MATERIALS AND METHODS

Patient samples

Written informed consent was obtained from all patients undergoing elective arthroplasty for degenerative arthropathy who were recruited over two separate periods (March 2004 to April 2005; October 2009 to March 2012) following ethical committee approval (MREC: 09/H0805/59). Adherence to STROBE guidelines is provided in the Supplementary Online Data. Lymphocyte counts were measured pre- and post-operatively by the local haematology laboratory (University College London NHS Trust) using a Sysmex XE2100 Analyzer. Established preopera-tive lymphopenia was defined as a lymphocyte count ≤1.3×109 cells·l−1 or <20% of the total leucocyte count. Length of hospital stay and postoperative morbidity were recorded using the Post-Operative Morbidity Survey [16] by study nurses/data collectors blinded to the lymphocyte and laboratory data. Wound morbidity was defined clinically by microbiologically proven, or suspected, wound infection (which includes the failure to achieve wound dryness, a risk factor for developing wound infection) [17,18]. Patients were cared for by their attending clinical staff blinded to the results of the additional laboratory tests performed. Full clinical details concerning surgery, anaesthesia, postoperative care and clinical outcome measures are provided in the Supplementary Appendix.

Mice

All experiments were performed according to the U.K. Home Office legislation and ARRIVE guidelines. Splenocytes were obtained from 14 to 18-week-old male C57BL/6 wild-type (WT) and NLRP3−/− mice (kindly provided by Professor Clare Bryant, Veterinary Medicine, University of Cambridge, U.K.). Ethical approval was granted by the University of Cambridge Ethics Committee and carried out under U.K. Home Office project licence number 80/2135. Splenocytes were harvested from the spleens of WT and NLRP3−/− mice by passing the spleen contents through a cell strainer. Cells were washed with PBS and red blood cells were depleted using red blood cell lysis buffer.

Primary lymphocyte isolation

In 93 patients from the second cohort, patient lymphocytes were isolated using the density gradient technique (Ficoll Plus, BD Biosciences) within 1.5 h of blood sampling. Purity >97% was achieved, after removal of monocytes by the adherence technique [19]. The yield of lymphocytes for concurrent quantification and functionality experiments was limited by the 15 ml volume of blood taken, particularly in lymphopenic patients. A summary of laboratory assays conducted on the samples is provided in Supplementary Table S1. Laboratory assays were undertaken masked to lymphocyte count.

Lymphocyte respirometry

Mitochondrial oxygen consumption was measured (Seahorse XF System) in 106 lymphocytes sampled from matched lymphopenic and non-lymphopenic patients in the same assay plate and under identical experimental conditions; ≥5 measures were made per patient sample. Metabolic measurements were made in a subset of patients preoperatively and 72 h after surgery. Total lymphocyte oxygen consumption was measured at baseline and after addition of the uncoupling agent carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP; 2 μM) to induce maximal oxygen consumption. Non-mitochondrial respiration was quantified by inhibiting mitochondrial respiration through addition of rotenone (1 μM) and antimycin A (1 μM). Mitochondrial respiration coupled towards ATP production was measured by the fall in oxygen consumption following addition of oligomycin (0.2 μg·ml−1), an inhibitor of ATP synthase (see Supplementary Figure S2). Glycolytic capacity was assessed by inhibiting glycolysis with 2-deoxyglucose. Intracellular ATP content was measured using a luciferase luminescence assay (ATPlite, PerkinElmer), corrected to total protein (BCA protein assay, Pierce Biotechnology). ATP was then calculated as nanomoles per milligram of protein.

Lymphocyte apoptosis and functional assays

Using lymphocytes from lymphopenic and non-lymphopenic patients, apoptosis was quantified using three separate flow cytometry-based assays. We used annexin, caspase-3 and mitochondrial transmembrane potential (JC-1) assays, after incubating isolated lymphocytes with concanavalin A (5 μg·ml−1) lipopolysaccharide (100 ng·ml−1; Salmonella typhi) and 10−7 M dexamethasone for 72 h to mimic the perioperative period over which metabolic profiles were attained. We used lipopolysaccharide (LPS) as this has been implicated in postoperative morbidity and recent data have shown that TLR4 signalling in CD4+ T-cells acts to enhance activation in the absence of T-cell receptor (TCR) stimulation and promote survival [20]. Briefly, 5×105 isolated lymphocytes were washed in PBS and resuspended in 100 μl of annexin buffer containing annexin V–FITC/propidium iodide (PI). Early apoptosis was defined by annexin V–FITC+PI-staining of cells. Intracellular staining of caspase-3 was achieved using FITC-conjugated Asp-Glu-Val-Asp-fluoromethylketone (DEVD-FMK) (Abcam), a cell-permeable caspase inhibitor which irreversibly binds active caspase-3 in cells undergoing apoptosis. Loss of lymphocyte mitochondrial transmembrane potential (ΔΨm) was quantified using the cationic dye JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3-tetraethylbenzimidazolylcarbocyanine iodide). All patient samples were compared with their own positive control prepared in parallel, by incubating with 5 μM FCCP for 45 min at 37°C prior to JC-1 staining.

T-lymphocyte proliferation

T-lymphocyte proliferation was quantified by labelling 106 lymphocytes·ml−1 with carboxyfluorescein succinimidyl ester (Invitrogen), prior to activating the cells with concanavalin A (5 μg·ml−1; Sigma) for 72 h. Intracellular cytokine production in lymphocytes was measured using diluted whole blood to both minimize cellular disturbance and preserve physiological concentrations of factors that influence T-cell function. After fixation for CD4/8 surface staining, permeabilization (Cytofix/Cytoperm, BD Biosciences) enabled intracellular staining for IL-2, tumour necrosis factor-α (TNFα) and interferon-γ (Miltenyi Biotec). Brefeldin A (10 μg·ml−1) was added to cells to inhibit transport of cytokines that consequently accumulate intracellularly. All flow cytometry experiments were performed on a calibrated cytometer (Cyan ADP, Beckman Coulter). Further, and additional, experimental details are provided in the Supplementary Appendix.

Caspase-1 activity

Caspase-1 activity was quantified using flow cytometry in human lymphocytes and mouse splenocytes using 6-carboxyfluorescein-Tyr-Val-Ala-Asp-fluoromethylketone (FAM-YVAD-FMK) (FLICA; ImmunoChemistry Technologies, ABD Serotec).

Western blot analysis

Immunoblots were performed using protein from 2.5×106 lymphocytes loaded per well of a 10% acrylamide gel. Immunoblots were performed with polyclonal goat anti-human IL-1β/IL-1F2 antibody (1:1000 dilution; AF-201-NA; R & D Systems) and rabbit anti-human NLRP1 antibody (1:1000 dilution; NBP1-54899; Novus Biologicals). Equal protein loading was confirmed by re-probing the membranes with anti-β-actin antibodies (Santa Cruz Biotechnology).

Real-time PCR

RNA from primary human lymphocytes was extracted using the RNAeasy Kit according to the manufacturer's protocol (Qiagen). RNA integrity and concentration was tested on an ND-1000 spectrophotometer (Nanodrop). Real-time PCR was executed in two steps. Reverse transcription was performed using the standard protocol of the TaqMan kit (Roche) from 2 μg of RNA using random hexamers in a final reaction volume of 50 μl. For the second step, 0.5 μl of cDNA reaction product was used per well and in triplicate. The amplification product was monitored by incorporation of SYBR Green (Roche). Real-time PCR was carried out using Eppendorf Realplex apparatus. The following primers (Sigma–Aldrich) were used, as reported previously [21]: NLRP1 forward primer 5′-ACCTGATCCCAAGTGACTGC-3′, and reverse primer 5′-TCTTCTCCAGGGCTTCGATA-3′; succinate dehydrogenase (SDH) forward primer 5′-CA-AACAGGAACCCGAGGTTTT-3′ and SDH reverse primer 5′-CAGCTTGGTAACACATGCTGTAT-3′. Expression values of NLRP1 were normalized to SDH and absolute lymphocyte count and are reported in units of 2ΔCT related to expression in lymphocytes obtained from a healthy volunteer, where DCt is the difference in CT values between NLRP1 and SDH in the same sample.

Statistical analyses

Data were analysed by descriptive and comparative approaches. Power calculations are described in the Supplementary Appendix. Results are expressed as absolute numbers and percentages (means ± S.D., or medians (interquartile range) as indicated). Fisher's exact test was used to compare categorical data distributions. Parametric methods and tests were used as indicated to analyse normally distributed data, while two-group non-parametric Wilcoxon signed-rank and Mann–Whitney U tests were used to analyse data deviating from a normal distribution. The non-parametric Kruskal–Wallis test was used for comparison of more than two independent groups. Time to discharge from hospital and resolution of postoperative morbidity were analysed using Kaplan–Meier survival plots (log-rank test). Statistical analyses were undertaken using NCSS 2007 software. Reported P values are two-sided; a P value ≤0.05 was considered to indicate statistical significance.

RESULTS

Postoperative morbidity and mortality in patients with and without lymphopenia

Preoperative lymphopenia was recorded in 15–18% of the two cohorts (Tables 1 and 2). Where available, long-term laboratory data demonstrated that lymphopenia was a constant chronic feature, with a mean coefficient of variation for absolute lymphocyte count of 12.7% (Supplementary Figure S1). This finding was associated with clinically insignificant increases with age and male gender. Preoperative co-morbidities and perioperative clinical management were similar (Tables 1 and 2). Established preoperative lymphopenia predisposed to an excess of postoperative complications with hazard ratios (HRs) of 1.28 ((95% confidence interval (CI) 1.02–1.60); P=0.03) and 1.31 ((95% CI 1.01–1.70); P=0.035) in the first (n=417) and second (n=328) patient cohorts respectively (Figures 1A and 1B). Preoperative lymphopenia was associated with an increased risk of morbidity on postoperative day 5 (relative risk: 1.8 (95% CI 1.3–3.2), P=0.004), including wound morbidity (relative risk 1.8 (95% CI 1.2–2.7); P=0.004) and infection (relative risk 1.5 (95% CI 1.1–2.0); P=0.008). Hospital discharge was also delayed for both cohorts (HR 1.56 (95% CI 1.26–1.96) for the first cohort and 1.31 (1.01–1.69) for the second cohort, both P<0.01; Figures 1C and 1D). In both cohorts of elective orthopaedic patients, absolute lymphocyte count consistently fell postoperatively (Figures 2A and 2B), and failed to return to preoperative levels by postoperative day 5 (P<0.001).

Table 1
Clinical details and demographics of laboratory patient cohorts

Values shown are means (95% confidence intervals) unless stated. *Abnormal ECG (electrocardiogram) defined as any of: left ventricular hypertrophy, left bundle branch block, ST/T-wave abnormalities. †Renal insufficiency defined according to the Revised Cardiac Risk Index (RCRI) as serum creatinine >177 μmol·l−1. eGFR, estimated glomerular filtration rate using the Modification of Diet in Renal Disease calculation. hsCRP, high-sensitivity C-reactive protein. ‡VSAQ METS, estimated metabolic equivalents on the Veterans’ Specific Activity Questionnaire. ARB, angiotensin receptor blocker. hsCRP measured in 35 patients with normal compared with 12 patients with low lymphocyte counts. ACE, angiotensin-converting enzyme; CCF, congestive cardiac failure; COPD, chronic obstructive pulmonary disease; GA, general anaesthesia; IHD, ischaemic heart disease; IQR, inter-quartile range; RA, regional anaesthesia; THR, total hip replacement; TKR, total knee replacement.

NormalLymphopenicP-value
Number 59 21  
Age, years (median, IQR) 71 (65–76) 77 (73–81) 0.004 
Male (n, [%]) 13 [22] 12 [57] 0.003 
Body mass index (kg·m−231.6 (25.7–36.3) 28.1 (26.1–30.1) 0.018 
Smoking (n, [%]) 5 [8] 0 [0] 0.375 
Preoperative co-morbidities    
 Hypertension (n, [%]) 41 [69] 15 [71] 1.0 
 IHD (n, [%]) 6 [10] 3 [14] 0.691 
 CCF (n, [%]) 3 [5] 1 [5] 1.0 
 Abnormal ECG* (n, [%]) 9 [15] 6 [29] 0.202 
 Cerebrovascular disease (n, [%]) 6 [10] 2 [10] 1.0 
 Renal insufficiency† (n, [%]) 1 [2] 0 [0] 1.0 
 Diabetes mellitus (n, [%]) 9 [15] 5 [24] 0.504 
 COPD (n, [%]) 1 [2] 3 [14] 0.0503 
 Rheumatoid arthritis (n, [%]) 3 [5] 0 [0] 1.0 
Preoperative blood biomarkers    
 hsCRP (mg·l−1) (median, IQR)2 1.4 (0.6–3.4) 1.7 (0.6–14.3) 0.384 
 eGFR (ml·min−1·1.73 m−277 (66–90) 78 (57–90) 0.554 
Preoperative medication    
 Aspirin (n, [%]) 8 [14] 5 [24] 0.310 
 Clopidogrel (n, [%]) 1 [2] 1 [5] 0.459 
 Warfarin (n, [%]) 2 [3] 3 [14] 0.110 
 Beta-blocker (n, [%]) 13 [22] 7 [33] 0.381 
 Statin (n, [%]) 14 [24] 4 [19] 0.768 
 ACE-inhibitor (n, [%]) 13 [22] 4 [19] 1.0 
 ARB (n, [%]) 4 [7] 2 [10] 0.650 
 Diuretic (n, [%]) 14 [24] 3 [14] 0.537 
 Steroids (n, [%]) 0 [0] 1 [5] 0.262 
Preoperative risk    
 Duke Activity Status Index (median, IQR) 15.5 (10.0–23.5) 10.3 (7.2–19.0) 0.270 
 VSAQ METS‡ (median, IQR) 5 (3–5.25) 3 (3–4) 0.188 
 POSSUM physiology score (median, IQR) 18 (16–22) 22 (18–27) 0.005 
 POSSUM predicted morbidity (median, IQR) 24 (19–36) 38 (24–58) 0.004 
Perioperative variables    
 THR (n, [%]) 23 [40] 8 [38] 0.741 
 TKR (n, [%]) 36 [60] 13 [52] 0.741 
 Revision procedures (n, [%], n(hip):n(knee) 3 [5], 2:1 3 [14], 1:2 0.274 
 General anaesthesia (GA) alone (n, [%]) 30 [51] 11 [56] 0.834 
 Regional anaesthesia alone (RA) (n, [%]) 15 [25] 7 [33] 0.834 
 Combined GA and RA (n, [%]) 14 [23] 2 [10] 0.834 
 Operative time (min) 95 (84–107) 109 (80–137) 0.320 
 Intraoperative crystalloid (ml) 1920 (1700–2140) 1820 (1370–2260) 0.660 
 Intraoperative colloid (ml) 330 (160–500) 360 (96–630) 0.866 
 PRC transfusion at any time (n, [%]) 15 [25] 7 [33] 0.572 
 Postopearative temperature (°C) (median, IQR) 36.4 (36.2–36.5) 36.4 (36.3–36.8) 0.600 
 Preoperative haematocrit 0.41 (0.40–0.42) 0.39 (0.38–0.41) 0.127 
 Postoperative haematocrit 0.31 (0.30–0.32) 0.29 (0.27–0.31) 0.07 
 Change in haematocrit −0.10 (−0.11 to −0.09) −0.11 (−0.13 to −0.09) 0.486 
 Postoperative ICU (n, [%]) 5 [8] 6 [29] 0.032 
NormalLymphopenicP-value
Number 59 21  
Age, years (median, IQR) 71 (65–76) 77 (73–81) 0.004 
Male (n, [%]) 13 [22] 12 [57] 0.003 
Body mass index (kg·m−231.6 (25.7–36.3) 28.1 (26.1–30.1) 0.018 
Smoking (n, [%]) 5 [8] 0 [0] 0.375 
Preoperative co-morbidities    
 Hypertension (n, [%]) 41 [69] 15 [71] 1.0 
 IHD (n, [%]) 6 [10] 3 [14] 0.691 
 CCF (n, [%]) 3 [5] 1 [5] 1.0 
 Abnormal ECG* (n, [%]) 9 [15] 6 [29] 0.202 
 Cerebrovascular disease (n, [%]) 6 [10] 2 [10] 1.0 
 Renal insufficiency† (n, [%]) 1 [2] 0 [0] 1.0 
 Diabetes mellitus (n, [%]) 9 [15] 5 [24] 0.504 
 COPD (n, [%]) 1 [2] 3 [14] 0.0503 
 Rheumatoid arthritis (n, [%]) 3 [5] 0 [0] 1.0 
Preoperative blood biomarkers    
 hsCRP (mg·l−1) (median, IQR)2 1.4 (0.6–3.4) 1.7 (0.6–14.3) 0.384 
 eGFR (ml·min−1·1.73 m−277 (66–90) 78 (57–90) 0.554 
Preoperative medication    
 Aspirin (n, [%]) 8 [14] 5 [24] 0.310 
 Clopidogrel (n, [%]) 1 [2] 1 [5] 0.459 
 Warfarin (n, [%]) 2 [3] 3 [14] 0.110 
 Beta-blocker (n, [%]) 13 [22] 7 [33] 0.381 
 Statin (n, [%]) 14 [24] 4 [19] 0.768 
 ACE-inhibitor (n, [%]) 13 [22] 4 [19] 1.0 
 ARB (n, [%]) 4 [7] 2 [10] 0.650 
 Diuretic (n, [%]) 14 [24] 3 [14] 0.537 
 Steroids (n, [%]) 0 [0] 1 [5] 0.262 
Preoperative risk    
 Duke Activity Status Index (median, IQR) 15.5 (10.0–23.5) 10.3 (7.2–19.0) 0.270 
 VSAQ METS‡ (median, IQR) 5 (3–5.25) 3 (3–4) 0.188 
 POSSUM physiology score (median, IQR) 18 (16–22) 22 (18–27) 0.005 
 POSSUM predicted morbidity (median, IQR) 24 (19–36) 38 (24–58) 0.004 
Perioperative variables    
 THR (n, [%]) 23 [40] 8 [38] 0.741 
 TKR (n, [%]) 36 [60] 13 [52] 0.741 
 Revision procedures (n, [%], n(hip):n(knee) 3 [5], 2:1 3 [14], 1:2 0.274 
 General anaesthesia (GA) alone (n, [%]) 30 [51] 11 [56] 0.834 
 Regional anaesthesia alone (RA) (n, [%]) 15 [25] 7 [33] 0.834 
 Combined GA and RA (n, [%]) 14 [23] 2 [10] 0.834 
 Operative time (min) 95 (84–107) 109 (80–137) 0.320 
 Intraoperative crystalloid (ml) 1920 (1700–2140) 1820 (1370–2260) 0.660 
 Intraoperative colloid (ml) 330 (160–500) 360 (96–630) 0.866 
 PRC transfusion at any time (n, [%]) 15 [25] 7 [33] 0.572 
 Postopearative temperature (°C) (median, IQR) 36.4 (36.2–36.5) 36.4 (36.3–36.8) 0.600 
 Preoperative haematocrit 0.41 (0.40–0.42) 0.39 (0.38–0.41) 0.127 
 Postoperative haematocrit 0.31 (0.30–0.32) 0.29 (0.27–0.31) 0.07 
 Change in haematocrit −0.10 (−0.11 to −0.09) −0.11 (−0.13 to −0.09) 0.486 
 Postoperative ICU (n, [%]) 5 [8] 6 [29] 0.032 
Table 2
Demographics of lymphopenic and non-lymphopenic patients in both cohorts

Values shown are means (95% confidence intervals) unless stated. ASA, American Society of Anaesthesiology grade; WCC, white cell count; POSSUM, Physiologic and Operative Severity Score for the enUmeration of Mortality and Morbidity; THR, total hip replacement; TKR, total knee replacement.

NormalLymphopenicP-value
Cohort 1 
 Number [%] 343 [82] 74 [18]  
 Age, years (median, IQR) 69 (62–77) 76 (68–82) <0.001 
 Male (n, [%]) 117 [34] 36 [49] 0.019 
 ASA ≥3 [%] 63 [18] 18 [24] 0.240 
 POSSUM score (median, IQR) 16 (15–18) 18 (16–20) <0.001 
 POSSUM predicted morbidity (median, IQR) 16.7 (14.2–21.1) 21.1 (16.2–30.3) <0.001 
 WCC (×109/l) 7.15 (6.95–7.35) 6.90 (6.50–7.30) 0.322 
 Neutrophil count (×109/l) 4.30 (4.14–4.46) 5.07 (4.70–5.44) <0.001 
 Neutrophils,% 59.3 (58.5–60.0) 72.6 (71.3–73.9) <0.001 
 Lymphocyte count (×109/l) 2.10 (2.04–2.17) 1.13 (1.08–1.19) <0.001 
 Lymphocytes,% 30.2 (29.4–30.9) 17.0 (16.1–17.8) <0.001 
 THR (n, [%]) 139 [41] 36 [49] 0.199 
 TKR (n, [%]) 204 [59] 38 [51] 0.199 
 Operative time (min) 146 (142–150) 148 (138–157) 0.818 
 Postoperative temperature (°C) (median, IQR) 36.0 (35.5–36.3) 36.1 (35.7–36.5) 0.289 
 Preoperative haematocrit 0.40 (0.39–0.41) 0.40 (0.37–0.42) 0.847 
 Postoperative haematocrit 0.32 (0.31–0.33) 0.34 (0.31–0.36) 0.267 
Cohort 2 
 Number [%] 279 [85] 49 [15]  
 Age (years) (median, IQR) 69 (61–76) 73 (66–79) 0.016 
 Male (n, [%]) 88 [32] 24 [49] 0.01 
 ASA ≥3 [%] 99 [35] 32 [65] <0.001 
 POSSUM physiology score (median, IQR) 17 (15–20) 18 (16–25) 0.006 
 POSSUM predicted morbidity (median, IQR) 24.4 (16.7–36.4) 31.4 (18.9–50.9) 0.026 
 WCC (×109/l) 6.92 (6.67–7.18) 7.03 (6.60–7.46) 0.597 
 Neutrophil count (×109/l) 4.09 (3.90–4.28) 5.11 (4.70–5.51) <0.001 
 Neutrophils (%) 58.1 (57.2–59.1) 71.9 (70.1–73.7) <0.001 
 Lymphocyte count (×109/l) 2.07 (1.98–2.17) 1.15 (1.08–1.22) <0.001 
 Lymphocytes (%) 30.6 (29.7–31.5) 16.9 (15.8–17.9) <0.001 
 THR (n, [%]) 122 [44] 23 [47] 0.979 
 TKR (n, [%]) 157 [56] 26 [53] 0.979 
 Operative time (min) 139 (132–146) 132 (112–151) 0.490 
 Postoperative temperature (°C) (median, IQR) 36.4 (36.2–36.5) 36.4 (36.3–36.8) 0.490 
 Preoperative haematocrit 0.40 (0.39–0.41) 0.39 (0.38–0.41) 0.334 
 Postoperative haematocrit 0.31 (0.30–0.32) 0.31 (0.29–0.32) 0.512 
NormalLymphopenicP-value
Cohort 1 
 Number [%] 343 [82] 74 [18]  
 Age, years (median, IQR) 69 (62–77) 76 (68–82) <0.001 
 Male (n, [%]) 117 [34] 36 [49] 0.019 
 ASA ≥3 [%] 63 [18] 18 [24] 0.240 
 POSSUM score (median, IQR) 16 (15–18) 18 (16–20) <0.001 
 POSSUM predicted morbidity (median, IQR) 16.7 (14.2–21.1) 21.1 (16.2–30.3) <0.001 
 WCC (×109/l) 7.15 (6.95–7.35) 6.90 (6.50–7.30) 0.322 
 Neutrophil count (×109/l) 4.30 (4.14–4.46) 5.07 (4.70–5.44) <0.001 
 Neutrophils,% 59.3 (58.5–60.0) 72.6 (71.3–73.9) <0.001 
 Lymphocyte count (×109/l) 2.10 (2.04–2.17) 1.13 (1.08–1.19) <0.001 
 Lymphocytes,% 30.2 (29.4–30.9) 17.0 (16.1–17.8) <0.001 
 THR (n, [%]) 139 [41] 36 [49] 0.199 
 TKR (n, [%]) 204 [59] 38 [51] 0.199 
 Operative time (min) 146 (142–150) 148 (138–157) 0.818 
 Postoperative temperature (°C) (median, IQR) 36.0 (35.5–36.3) 36.1 (35.7–36.5) 0.289 
 Preoperative haematocrit 0.40 (0.39–0.41) 0.40 (0.37–0.42) 0.847 
 Postoperative haematocrit 0.32 (0.31–0.33) 0.34 (0.31–0.36) 0.267 
Cohort 2 
 Number [%] 279 [85] 49 [15]  
 Age (years) (median, IQR) 69 (61–76) 73 (66–79) 0.016 
 Male (n, [%]) 88 [32] 24 [49] 0.01 
 ASA ≥3 [%] 99 [35] 32 [65] <0.001 
 POSSUM physiology score (median, IQR) 17 (15–20) 18 (16–25) 0.006 
 POSSUM predicted morbidity (median, IQR) 24.4 (16.7–36.4) 31.4 (18.9–50.9) 0.026 
 WCC (×109/l) 6.92 (6.67–7.18) 7.03 (6.60–7.46) 0.597 
 Neutrophil count (×109/l) 4.09 (3.90–4.28) 5.11 (4.70–5.51) <0.001 
 Neutrophils (%) 58.1 (57.2–59.1) 71.9 (70.1–73.7) <0.001 
 Lymphocyte count (×109/l) 2.07 (1.98–2.17) 1.15 (1.08–1.22) <0.001 
 Lymphocytes (%) 30.6 (29.7–31.5) 16.9 (15.8–17.9) <0.001 
 THR (n, [%]) 122 [44] 23 [47] 0.979 
 TKR (n, [%]) 157 [56] 26 [53] 0.979 
 Operative time (min) 139 (132–146) 132 (112–151) 0.490 
 Postoperative temperature (°C) (median, IQR) 36.4 (36.2–36.5) 36.4 (36.3–36.8) 0.490 
 Preoperative haematocrit 0.40 (0.39–0.41) 0.39 (0.38–0.41) 0.334 
 Postoperative haematocrit 0.31 (0.30–0.32) 0.31 (0.29–0.32) 0.512 

Lymphopenia and postoperative outcomes

Figure 1
Lymphopenia and postoperative outcomes

(A and B) Kaplan–Meier plots illustrate the time taken for patients to become morbidity-free in both cohorts, as defined by the Postoperative Morbidity Score (POMS) assessed at the time points indicated. P-values are shown for log-rank test. (C and D) Kaplan–Meier plots illustrate the time taken for patients to be discharged from hospital, for both cohorts. P-values are shown for log-rank test.

Figure 1
Lymphopenia and postoperative outcomes

(A and B) Kaplan–Meier plots illustrate the time taken for patients to become morbidity-free in both cohorts, as defined by the Postoperative Morbidity Score (POMS) assessed at the time points indicated. P-values are shown for log-rank test. (C and D) Kaplan–Meier plots illustrate the time taken for patients to be discharged from hospital, for both cohorts. P-values are shown for log-rank test.

Postoperative changes in lymphocyte count

Figure 2
Postoperative changes in lymphocyte count

(A) Cohort 1: pre- and post-operative lymphocyte count (n=417 patients). (B) Cohort 2: pre- and post-operative lymphocyte count (n=328 patients). Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.001). $ denotes difference between preoperative and all postoperative time points (P<0.001).

Figure 2
Postoperative changes in lymphocyte count

(A) Cohort 1: pre- and post-operative lymphocyte count (n=417 patients). (B) Cohort 2: pre- and post-operative lymphocyte count (n=328 patients). Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.001). $ denotes difference between preoperative and all postoperative time points (P<0.001).

Perioperative metabolic profile of lymphocytes

Bioenergetic changes were serially analysed in freshly isolated lymphocytes obtained from 13 patients preoperatively (Supplementary Figure S2), and again 72 h after surgery when lymphopenia was evident (Figure 3A). Mitochondrial oxygen consumption, both at baseline and following activation with the T-cell mitogen concanavalin A, was substantially lower in the postoperative period (P=0.001; Figure 3B). FCCP-induced maximal respiration declined by 46±9% postoperatively compared with lymphocytes obtained from the same patients preoperatively (P<0.001; Figure 3A). Spare respiratory capacity, defined as the difference between maximal and basal rates of oxygen consumption, represents the extra capacity available within cells to respond promptly to increased metabolic demands, thereby supporting cellular survival. Absolute spare respiratory capacity declined by 43% postoperatively (range: 26–65%; P=0.029; Figure 3C). Once activated, lymphocytes switch their metabolic phenotype towards glycolysis-derived ATP [22], which is particularly important for cytokine production [23,24]. However, lymphocytes sampled postoperatively displayed a substantially lower glycolytic activity (Figure 3D); furthermore, the anticipated increase in glycolytic activity following inhibition of mitochondrial respiration was reduced by ~70% compared with individualized patients preoperative values (Figure 3E).

Perioperative changes in lymphocyte metabolism

Figure 3
Perioperative changes in lymphocyte metabolism

(A) Pre- and post-operative lymphocyte counts for serial analyses reported in (BE). (B) Mitochondrial oxygen consumption rate (OCR) at baseline, after ATP synthase inhibition with oligomycin to remove the contribution of non-mitochondrial respiration, and maximal oxygen consumption following uncoupling with FCCP, in isolated lymphocytes taken serially from 13 patients. (C) Spare capacity represents the difference between maximal oxygen consumption and basal respiration. (D) Extracellular acidification (ECAR) measured at baseline, following ATP synthase inhibition and maximal oxygen consumption, in isolated lymphocytes taken from the same patients. (E) Inhibition of glycoylsis with 2-deoxyglucose reveals glycolytic capacity. Asterisk denotes P≤0.01, by paired Student's t-test.

Figure 3
Perioperative changes in lymphocyte metabolism

(A) Pre- and post-operative lymphocyte counts for serial analyses reported in (BE). (B) Mitochondrial oxygen consumption rate (OCR) at baseline, after ATP synthase inhibition with oligomycin to remove the contribution of non-mitochondrial respiration, and maximal oxygen consumption following uncoupling with FCCP, in isolated lymphocytes taken serially from 13 patients. (C) Spare capacity represents the difference between maximal oxygen consumption and basal respiration. (D) Extracellular acidification (ECAR) measured at baseline, following ATP synthase inhibition and maximal oxygen consumption, in isolated lymphocytes taken from the same patients. (E) Inhibition of glycoylsis with 2-deoxyglucose reveals glycolytic capacity. Asterisk denotes P≤0.01, by paired Student's t-test.

Since the relative contributions of redistribution and apoptosis of lymphocytes following surgery are difficult to quantify in humans, the bioenergetic phenotype of postoperative lymphocytes was compared with that measured preoperatively in patients with chronic lymphopenia. The demographics and clinical characteristics were well-matched between these patient groups (Table 2). Of note, preoperative lymphocytes taken from chronically lymphopenic patients exhibited a similar metabolic phenotype to postoperative samples from non-lymphopenic patients. Intracellular ATP (Figure 4A) and mitochondrial oxygen consumption (Figure 4B) were lower in lymphocytes taken from patients with established lymphopenia, both before and after ex vivo activation. The proportion (~75%) of oxygen consumption utilized for ATP production was similar for both groups. Cells from lymphopenic patients had a spare respiratory capacity ~30% of that measured in lymphocytes from non-lymphopenic patients (P<0.001; Figure 4C). This value was similar to the absolute postoperative value measured in elective patients with normal preoperative lymphocyte counts (Figure 3C). Lymphocytes taken from lymphopenic patients also displayed lower basal glycolytic activity following inhibition of mitochondrial respiration (Figure 4D). Glycolytic capacity was markedly lower in lymphopenic patients, as assessed by inhibiting glycolysis with 2-deoxyglucose (Figure 4E). This suggests that the failure to maintain ATP levels is, at least in part, due to a lack of a compensatory increase in glycolysis.

Lymphocyte metabolism in patients with preoperative lymphopenia

Figure 4
Lymphocyte metabolism in patients with preoperative lymphopenia

(A) Intracellular ATP content was measured before and after stimulation with concanavalin A (CON-A), in isolated lymphocytes taken from normal (n=19) and lymphopenic (n=6) patients, preoperatively. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (B) Mitochondrial oxygen consumption is shown at baseline, after ATP synthase inhibition with oligomycin to remove the contribution of non-mitochondrial respiration, and maximal oxygen consumption following uncoupling with FCCP, in isolated lymphocytes taken preoperatively from normal (n=24) and lymphopenic (n=7) patients. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each timepoint (P<0.05; two-way ANOVA). (C) Spare capacity represents the difference between maximal oxygen consumption and basal respiration. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (D) Extracellular acidification (ECAR) measured at baseline, following ATP synthase inhibition and maximal oxygen consumption, in isolated lymphocytes taken preoperatively from normal (n=24) and lymphopenic (n=7) patients. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.001; two-way ANOVA (drug × group interaction). (E) In separate experiments, the addition of 2-deoxyglucose (2-DG) confirmed that most of the ECAR is related to glycolytic activity in patients with normal lymphocyte count (n=20) and lymphopenia (n=3). All values shown are means ± S.E.M.

Figure 4
Lymphocyte metabolism in patients with preoperative lymphopenia

(A) Intracellular ATP content was measured before and after stimulation with concanavalin A (CON-A), in isolated lymphocytes taken from normal (n=19) and lymphopenic (n=6) patients, preoperatively. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (B) Mitochondrial oxygen consumption is shown at baseline, after ATP synthase inhibition with oligomycin to remove the contribution of non-mitochondrial respiration, and maximal oxygen consumption following uncoupling with FCCP, in isolated lymphocytes taken preoperatively from normal (n=24) and lymphopenic (n=7) patients. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each timepoint (P<0.05; two-way ANOVA). (C) Spare capacity represents the difference between maximal oxygen consumption and basal respiration. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (D) Extracellular acidification (ECAR) measured at baseline, following ATP synthase inhibition and maximal oxygen consumption, in isolated lymphocytes taken preoperatively from normal (n=24) and lymphopenic (n=7) patients. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.001; two-way ANOVA (drug × group interaction). (E) In separate experiments, the addition of 2-deoxyglucose (2-DG) confirmed that most of the ECAR is related to glycolytic activity in patients with normal lymphocyte count (n=20) and lymphopenia (n=3). All values shown are means ± S.E.M.

Apoptotic and functional phenotype of preoperative lymphopenia

The hypometabolic phenotype of lymphopenic patients, characterized by mitochondrial dysfunction leading to increased apoptosis through the intrinsic pathway, can plausibly explain the susceptibility of these patients to postoperative infection. We tested this hypothesis by incubating freshly isolated lymphocytes obtained from preoperative patients with common perioperative stressors at physiologically relevant doses, i.e. free of potential confounding influences such as anaesthetic agents. Lymphocytes obtained from lymphopenic patients expressed higher levels of the early apoptosis marker annexin V (Figures 5A and 5B), increased expression of the executioner caspase-3 (Figure 5C) and a decrease in mitochondrial membrane potential (Figure 5D). Lymphocyte subsets were affected similarly between lymphopenic and non-lymphopenic patients (results not shown). Since neither FAS (CD95) cell-surface death receptor expression (a key trigger of lymphocyte apoptosis [25], Supplementary Figure S3), preoperative monocyte HLA-DR expression (Supplementary Figure S4), nor high-sensitivity C-reactive protein (Supplementary Table S2) differed between lymphopenic and non-lymphopenic patients, systemic inflammation did not appear to account for the differences observed in their propensity for apoptosis.

Lymphocyte count and apoptosis

Figure 5
Lymphocyte count and apoptosis

(A) Flow cytometry plots of apoptosis in lymphocytes incubated with dexamethasone, taken from patients with normal, or low (lymphopenia) counts. Annexin V-positive cells indicate early apoptosis (lower right panel); joint annexin staining of cells with PI-positive lymphocytes indicates later apoptosis, prior to death (upper right panel). (B) Group data are shown for early apoptotic changes (annexin V-positive, PI-negative) in isolated lymphocytes taken preoperatively from normal (n=39) and lymphopenic (n=10) patients, untreated (CTRL) or treated with LPS (100 ng·ml−1; Salmonella typhi), dexamethasone (10−7 M) or concanavalin A (5 μg·ml−1). Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (C) Confirmation of the increased propensity for apoptosis in lymphopenic patients is illustrated by isolated lymphocytes taken preoperatively from normal (n=9) and lymphopenic (n=5) patients which stain positively for the executioner caspase, activated caspase-3, following treatment with LPS (lipopolysaccharide), (dexamethasone concanavalin A. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (D) Lymphopenic patients’ cells demonstrate greater loss of mitochondrial membrane potential in isolated lymphocytes, as measured by JC-1 using flow cytometry (n=4 patients/group; median and interquartile range (IQR) values are shown). Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (E) Proliferation of isolated lymphocytes following 3 days’ stimulation by concanavalin A, taken preoperatively from normal and lymphopenic patients. The histogram shows summary data: the proportion of isolated lymphocytes undergoing at least one division (measured using carboxyfluorescein succinimidyl ester (CSFE) incorporation) following 3 days’ stimulation by concanavalin A is shown, in lymphocytes taken preoperatively from normal (n=17) and lymphopenic (n=8) patients. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (F) Higher proportions of CD4+ lymphocytes from patients with normal counts express more intracellular IL-2, and interferon (IFN)-γ cytokines (P=0.001) after 24 h of incub-ation of whole blood (1:1 dilution in RPMI 1640 medium) in the presence/absence of PMA and ionomycin (iono), which activate T-cells directly. All data are shown as medians with interquartile ranges (IQR). Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.05; two-way ANOVA). CTRL, control; LPS, lipopolysaccharide; DEX, dexamethasone; CON-A, concanavalin A.

Figure 5
Lymphocyte count and apoptosis

(A) Flow cytometry plots of apoptosis in lymphocytes incubated with dexamethasone, taken from patients with normal, or low (lymphopenia) counts. Annexin V-positive cells indicate early apoptosis (lower right panel); joint annexin staining of cells with PI-positive lymphocytes indicates later apoptosis, prior to death (upper right panel). (B) Group data are shown for early apoptotic changes (annexin V-positive, PI-negative) in isolated lymphocytes taken preoperatively from normal (n=39) and lymphopenic (n=10) patients, untreated (CTRL) or treated with LPS (100 ng·ml−1; Salmonella typhi), dexamethasone (10−7 M) or concanavalin A (5 μg·ml−1). Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (C) Confirmation of the increased propensity for apoptosis in lymphopenic patients is illustrated by isolated lymphocytes taken preoperatively from normal (n=9) and lymphopenic (n=5) patients which stain positively for the executioner caspase, activated caspase-3, following treatment with LPS (lipopolysaccharide), (dexamethasone concanavalin A. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (D) Lymphopenic patients’ cells demonstrate greater loss of mitochondrial membrane potential in isolated lymphocytes, as measured by JC-1 using flow cytometry (n=4 patients/group; median and interquartile range (IQR) values are shown). Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (E) Proliferation of isolated lymphocytes following 3 days’ stimulation by concanavalin A, taken preoperatively from normal and lymphopenic patients. The histogram shows summary data: the proportion of isolated lymphocytes undergoing at least one division (measured using carboxyfluorescein succinimidyl ester (CSFE) incorporation) following 3 days’ stimulation by concanavalin A is shown, in lymphocytes taken preoperatively from normal (n=17) and lymphopenic (n=8) patients. Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.01; one-way ANOVA). (F) Higher proportions of CD4+ lymphocytes from patients with normal counts express more intracellular IL-2, and interferon (IFN)-γ cytokines (P=0.001) after 24 h of incub-ation of whole blood (1:1 dilution in RPMI 1640 medium) in the presence/absence of PMA and ionomycin (iono), which activate T-cells directly. All data are shown as medians with interquartile ranges (IQR). Asterisk denotes difference between patients with lymphopenic and normal lymphocyte counts at each time point (P<0.05; two-way ANOVA). CTRL, control; LPS, lipopolysaccharide; DEX, dexamethasone; CON-A, concanavalin A.

Lymphocytes undergo proliferation when challenged by a range of mitogens and/or specific antigens. A reduced ability to mount a proliferative response denotes cell-mediated immune dysfunction [26]. Lymphocytes from lymphopenic patients were less likely to proliferate (Figure 5E) following 72 h of activation by the mitogen concanavalin A. Consistent with this phenotype, both CD4+ (P<0.001) and CD8+ (P=0.018) lymphocytes from lymphopenic patients produced fewer intracellular Th1 cytokines following activation, including interferon-γ and IL-2, compared with patients with normal lymphocyte counts (Figure 5F).

Stress-induced hypometabolism in lymphocytes via activation of the NLRP1 inflammasome

We next sought a unifying mechanism that could explain similar bioenergetic phenotypes in both chronic and postoperative lymphocytes, and–critically–also account for a failure to increase glycolysis in the face of lower oxidative phosphorylation. As there were no clear clinical correlates in the population with lymphopenia, we surmised that a ubiquitous stress-glucocorticoid-response mechanism may promote lymphocyte apoptosis and bioenergetic impairment [27]. This hypothesis would fit with the prototypical neurohormonal response to surgery and trauma characterized by substantial and prolonged elevations in circulating glucocorticoids [28]. Caspase-1 inhibits glycolysis at several glycolytic checkpoints [29], and also regulates lymphocyte homoeostasis through pyroptosis as observed in HIV-associated CD4 depletion [30,31]. Caspase-1 activation is chiefly dependent on the inflammasome, a multiprotein complex that induces assembly of active caspase-1 through autocatalytic cleavage and promotes secretion of the pro-inflammatory cytokines IL-1β and IL-18. Both NLRP1 and NLRP3 inflammasomes are present in human T-cells [32]. Mitochondrial dysfunction is strongly implicated, although not obligatory [33], for the activation of the NLRP3 inflammasome through generation of mitochondrial reactive oxygen species (ROS) [34]. We therefore hypothesized that physiological concentrations of dexamethasone would trigger caspase-1 activation via the inflammasome in primary human lymphocytes. Indeed, incubation of primary lymphocytes with dexamethasone for 60 min generated mitochondrial ROS (Figures 6A and 6B), an increase in caspase-1 activity (Figures 6C and 6D), and mature IL-1β (Figure 6E). This short period of exposure to dexamethasone (compared with the longer duration of the surgical ‘stress’ response) increased caspase-1 activity substantially, equating to approximately 30% of the caspase-1 activity generated by the prototypical NLRP3 trigger nigericin (10 μM; Figure 6D). Similarly, inhibition of mitochondrial electron transport chain complex III with myxothiazol produced similar results (not shown). We then next assessed whether NLRP3 deficiency in murine splenocytes treated with dexamethasone prevented apoptosis (Figure 6F). Both WT and NLRP3−/− splenocytes showed similar degrees of apoptosis (Figure 6G). However, protein expression of NLRP1 increased following exposure to dexamethasone in both wild-type and NLRP3−/− splenocytes (Figure 6H), suggesting that glucocorticoids may primarily inhibit glycolysis through NLRP1-mediated increases in caspase-1 (Figure 5I). In patients with a postoperative fall (0.50±0.22×109 litre−1) in lymphocyte count, a 3.7-fold (interquartile range, IQR: 1.1–28) increase in NLRP1 mRNA transcription occurred (P=0.028; n=4 patients).

Glucocorticoid activation of caspase-1 through the NLRP1 inflammasome

Figure 6
Glucocorticoid activation of caspase-1 through the NLRP1 inflammasome

(A) Release of mitochondrial ROS in primary human lymphocytes, following incubation for 24 h with dexamethasone (1 μM), as assessed by Mitosox Red (n=5). (B) Group data for experiment shown in (A) (mean (S.D.); n=5). (C) Dexamethasone triggers caspase-1 activity in primary human lymphocytes (n=5). (D) Group data: caspase-1 activity (mean (S.D.); n=5). Summary data also shown for positive control (NLRP3 inflammasome activator, nigericin; treatment for 40 min). (E) Representative immunoblot showing dexamethasone-induced maturation of IL-1β in primary human lymphocytes (n=3 experiments). (F) Similar degree of apoptosis in splenocytes isolated from WT or NLRP3-deficient mice (n=5) following 16 h incubation with dexamethasone (1 μM) or PBS (as assessed by surface expression of annexin V–FITC and PI). (G) Summary data showing proportions of annexin V- and PI-positive splenocytes from WT (n=5) or NLRP3-deficient mice (n=5) following 16 h of incubation with dexamethasone (1–10 μM) or PBS. (H) Murine splenocytes deficient in NLRP3 generate caspase-1 activity following exposure to dexamethasone, with concomitant increase in NLRP1 inflammasome protein expression. (I) Summary data showing proportions of caspase-1-positive splenocytes from WT (n=5) or NLRP3-deficient (n=5) mice following 16 h of incubation with dexamethasone (1–10 μM) or PBS control. Asterisk throughout denotes difference between genotype/treatment (P≤0.05; two-way ANOVA). Dexa, dexamethasone.

Figure 6
Glucocorticoid activation of caspase-1 through the NLRP1 inflammasome

(A) Release of mitochondrial ROS in primary human lymphocytes, following incubation for 24 h with dexamethasone (1 μM), as assessed by Mitosox Red (n=5). (B) Group data for experiment shown in (A) (mean (S.D.); n=5). (C) Dexamethasone triggers caspase-1 activity in primary human lymphocytes (n=5). (D) Group data: caspase-1 activity (mean (S.D.); n=5). Summary data also shown for positive control (NLRP3 inflammasome activator, nigericin; treatment for 40 min). (E) Representative immunoblot showing dexamethasone-induced maturation of IL-1β in primary human lymphocytes (n=3 experiments). (F) Similar degree of apoptosis in splenocytes isolated from WT or NLRP3-deficient mice (n=5) following 16 h incubation with dexamethasone (1 μM) or PBS (as assessed by surface expression of annexin V–FITC and PI). (G) Summary data showing proportions of annexin V- and PI-positive splenocytes from WT (n=5) or NLRP3-deficient mice (n=5) following 16 h of incubation with dexamethasone (1–10 μM) or PBS. (H) Murine splenocytes deficient in NLRP3 generate caspase-1 activity following exposure to dexamethasone, with concomitant increase in NLRP1 inflammasome protein expression. (I) Summary data showing proportions of caspase-1-positive splenocytes from WT (n=5) or NLRP3-deficient (n=5) mice following 16 h of incubation with dexamethasone (1–10 μM) or PBS control. Asterisk throughout denotes difference between genotype/treatment (P≤0.05; two-way ANOVA). Dexa, dexamethasone.

DISCUSSION

Previous work from our laboratory has demonstrated that bioenergetic dysfunction is a core feature of multi-organ failure [35]. Our current data extend these findings by demonstrating that a similar association exists in lymphocytes taken from patients following tissue trauma and those with longstanding lymphopenia, characterized by low metabolic capacity. The mechanism underlying this global decrease in metabolic status is consistent with stressor responses activating caspase-1, via the NLRP1 inflammasome. These metabolic data explain both the acute and chronic relative paucity of lymphocytes through an increased propensity to apoptosis, and their impaired immune functionality associated with postoperative morbidity and prolonged hospitalization after a standardized surgical insult in an elective population. Furthermore, they provide translational data consistent with the concept that bioenergetic abnormalities contribute towards both the severity and progression of complex and multifactorial pathology [36]. Using high-throughput bioenergetic assays, we have also shown that cellular energetic interrogation from cells isolated rapidly from human blood can help elucidate biologic phenotypes that enable clinical and immune function to be mapped to metabolic dysfunction.

The bioenergetic profile and propensity to apoptosis of both acute and established lymphopenia is consistent with activation of caspase-1, a pivotal executioner caspase that mediates the inflammatory response to a plethora of pathogen-associated molecular patterns (PAMPs) and danger-associated molecular patterns (DAMPs). Activation of a ubiquitous sensor of danger molecules in T-cells is consistent with the lack of clear association with any particular pathology as identified within the clinical data presented here. Chronic infection and cancer result in a T-cell exhaustion phenotype [37], while lymphopenia is associated with worse outcomes in cardiac failure [38]. However, within our orthopaedic surgical cohorts, the lymphopenic patients were no different in terms of background levels of inflammation (as reflected by high-sensitivity C-reactive protein), or medications and co-morbidities associated with acquired mitochondrial abnormalities. Where laboratory data were available, lymphopenia had been present for several years prior to surgery.

We studied orthopaedic patients as they represent the majority of elective surgical procedures across different healthcare systems [39] (National Joint Registry at http:://www.njrcentre.org.uk/njrcentre/), and involve an increasingly aged population with multiple co-morbidities. Secondly, the surgical procedures they underwent represent standardized models of tissue injury that are free of confounding factors such as sepsis and malignancy. Thirdly, a significant burden of morbidity occurs in a high proportion of these patients, as demonstrated using different outcome measures, including hospital readmission rates [41]. Fourthly, previous reports have separately associated anergy, the failure of a delayed hypersensitivity response, and relative lymphopenia [42] with adverse outcomes following surgery [43], trauma [44,45] and sepsis [46]. However, the paucity and heterogeneity of patient studies, plus the complexity of the surgical procedures (chiefly in patients undergoing cancer surgery), limit robust conclusions from being drawn. Putative pathophysiological mechanisms have also remained obscure and underexplored. Small initial studies reporting that the anaesthetic technique may reduce the extent of lymphopenia [47] have remained controversial. Crucially, the finding that a substantial proportion of patients exhibit relative lymphopenia even preceding any traumatic and/or infectious insult indicates a population of patients at increased risk of developing critical illness. This is consistent with recent data documenting excess mortality in patients in whom post-trauma lymphopenia failed to reverse [48]. Importantly, our study confirms that the detrimental role of lymphocyte dysfunction reported in murine [49] and human [48] models of injury/critical illness appears to be independent of other components of the immune system.

Experimental murine models demonstrate that genetic ablation designed to prevent lymphocyte apoptosis following sepsis reduces mortality [50]. Caution is nevertheless warranted over direct comparison between murine and human immunity [51]. Lymphopenic patients sustained higher rates of wound morbidity. This is particularly relevant given epidemiological data demonstrating that even apparently minor surgical complications such as wound infection increase the risk of postoperative mortality [52]. Transgenic models of wound infection demonstrate that CD4+ lymphocytes are pivotal in modulating the function of polymorphonuclear cells at infected wound sites [12]. These data suggest therapeutic potential for the bioenergetic manipulation of lymphocyte function, as has been demonstrated experimentally with selective apoptosis in alloreactive T-cells preventing graft-versus-host disease [53].

Nigericin, an acute activator of the inflammasome, is an established inhibitor of lymphocyte proliferation [54]. However, endogenous molecules that similarly inhibit function have remained obscure. Despite their anti-inflammatory therapeutic role, long-term glucocorticoid treatment cause significant adverse side effects, which–perhaps counterintuitively–may be accounted for by mechanisms that reveal the potential for the inflammasome. Glucocorticoids increase both NLRP3 mRNA and protein in THP-1 monocyte-like and primary monocytes [55]. Glucocorticoid-induced up-regulation of NLRP3 sensitizes monocytes to extracellular ATP, thereby increasing release of pro-inflammatory molecules including mature IL-1β. Microarray studies in human lung A549 cells have revealed the pro-inflammatory potential for glucocorticoids, which synergistically up-regulate various inflammatory genes with TNFα [56]. Our findings also recapitulate several features observed in glucocorticoid resistance in malignant lymphocytes (chronic lymphocytic leukaemia; CLL), where glucocorticoids decrease metabolic glycolytic activity of CLL cells and promote the loss of mitochondrial membrane potential [57]. A potential link between resistance to therapeutic glucocorticoid use in inflammatory bowel disease and a mutation in the NLRP1 gene (L155H mutant variant) has also been suggested [58]. Elevated glucocorticoid levels are unlikely to be the sole mechanism underlying our observation of metabolic dysfunction in lymphocytes following surgical trauma. A role for programmed cell death 1 and its inhibitory receptor has been implicated in the phenomenon of T-cell exhaustion [7]. Receptor-mediated negative regulation of lymphocyte metabolism by other immune cells is also plausible; binding of programmed death ligand 1 (PD-L1) on the surface of monocytes, neutrophils and dendritic cells to PD-1 on T-cells is described across several different inflammatory pathologies [59,60]. Similarly, following human endotoxaemia, mature CD16hiCD62Llow neutrophils suppress T-cell proliferation [61].

Strengths of the present study include the detailed phenotyping in a stable well-defined population of patients free of pathologies associated with lymphopenia, who underwent scheduled standardized tissue trauma. We strove to ensure that lymphopenic samples were analysed concurrently with cells obtained from patients with a normal lymphocyte count. The serial analysis of perioperative samples, enabling each patient to act as their own control, is a robust model that circumnavigates the significant challenges of appropriate control samples. Limitations include the low numbers of cells that could be obtained from postoperative and lymphopenic patients, which prevented transfection studies in primary T-cells. Further work on more detailed phenotyping of these cells may also provide further information on specific subtypes associated with senescence, such as CD8+ effector memory T-cells that re-express the naive T-cell marker CD45RA [62].

In conclusion, our findings suggest that lymphocyte bioenergetic abnormalities represent a common, yet hitherto under-recognized, pathological mechanism that offers a new paradigm to reduce morbidity and enhance recovery from various acute inflammatory challenges. Neurohormonal activation of the NLRP1 inflammasome results in a hypometabolic phenotype, likely to be driven in part by caspase-1-mediated inhibition of glycolysis. Taken together, these findings suggest that manipulation of the hypometabolic lymphocytic phenotype offers a novel strategy to reduce postoperative infections and sepsis.

AUTHOR CONTRIBUTION

Gareth Ackland was responsible for the study hypothesis and design. Most patients were under the care of Fares Haddad. Mark Edwards undertook all experiments in conjunction with other authors as follows: Gareth Ackland and Ana Gutierrez del Arroyo performed apoptosis, proliferation and flow cytometry experiments; Gareth Ackland and Pervez Sultan performed the mitochondrial flux analyses; Mark Edwards, Shamir Karmali, Ramani Moonesinghe and Michael Mythen collected and/or analysed postoperative outcome data; Gareth Ackland and Pervez Sultan performed the cytokine experiments; John Whittle and Ana Gutierrez del Arroyo undertook immunoblot experiments; Gareth Ackland and Mervyn Singer analysed/interpreted the mitochondrial study data; and Gareth Ackland, Ana Gutierrez del Arroyo and, Pervez Sultan analysed/interpreted flow cytometry experiments. Gareth Ackland, Mark Edwards and Pervez Sultan had full access to all of the data in the study and had final responsibility for the decision to submit for publication. Gareth Ackland, Mark Edwards and Mervyn Singer drafted the paper. All authors participated in critical revision of the paper.

We thank Sadaf Iqbal, Laura Gallego Paredes, Anna Reyes for assistance with sample collection; Mr Rahul Patel, Mr Jig Patel, Mr James Youngman for clinical assistance; and Maj Mutch, Clare Majetowsky, Ernesto Bettini and Denise Wyndham for clinical data collection assistance.

FUNDING

This work was supported by the Academy of Medical Sciences/Health Foundation Clinician Scientist scheme (to G.L.A.); the National Institute for Health Research (NIHR), and the HCA International perioperative fellowship (to M.R.E., P.S., J.W.). M.S. is an NIHR Senior Investigator. This work was undertaken at University College London Hospitals NHS Trust/University College London who received a proportion of funding from the Department of Health UK NIHR Biomedical Research Centre funding scheme.

Abbreviations

     
  • CI

    confidence interval

  •  
  • CLL

    chronic lymphocytic leukaemia

  •  
  • DAMP

    danger-associated molecular pattern

  •  
  • FCCP

    carbonyl cyanide-p-trifluoromethoxyphenylhydrazone

  •  
  • HR

    hazard ratio

  •  
  • IL

    interleukin

  •  
  • LPS

    lipopolysaccharide

  •  
  • PAMP

    pathogen-associated molecular pattern

  •  
  • PD-L1

    programmed death ligand 1

  •  
  • PI

    propidium iodide

  •  
  • ROS

    reactive oxygen species

  •  
  • SDH

    succinate dehydrogenase

  •  
  • TNFα

    tumour necrosis factor α

  •  
  • WT

    wild-type

References

References
1
Hotchkiss
 
R.S.
Opal
 
S.
 
Immunotherapy for sepsis–a new approach against an ancient foe
N. Engl. J. Med.
2010
, vol. 
363
 (pg. 
87
-
89
)
[PubMed]
2
Hotchkiss
 
R.S.
Karl
 
I.E.
 
The pathophysiology and treatment of sepsis
N. Engl. J. Med.
2003
, vol. 
348
 (pg. 
138
-
150
)
[PubMed]
3
Kim
 
K.D.
Zhao
 
J.
Auh
 
S.
Yang
 
X.
Du
 
P.
Tang
 
H.
Fu
 
Y.X.
 
Adaptive immune cells temper initial innate responses
Nat. Med.
2007
, vol. 
13
 (pg. 
1248
-
1252
)
[PubMed]
4
Palm
 
N.W.
Medzhitov
 
R.
 
Not so fast: adaptive suppression of innate immunity
Nat. Med.
2007
, vol. 
13
 (pg. 
1142
-
1144
)
[PubMed]
5
Cabrera-Perez
 
J.
Condotta
 
S.A.
Badovinac
 
V.P.
Griffith
 
T.S.
 
Impact of sepsis on CD4 T cell immunity
J. Leukoc. Biol.
2014
, vol. 
96
 (pg. 
767
-
777
)
[PubMed]
6
Drewry
 
A.M.
Samra
 
N.
Skrupky
 
L.P.
Fuller
 
B.M.
Compton
 
S.M.
Hotchkiss
 
R.S.
 
Persistent lymphopenia after diagnosis of sepsis predicts mortality
Shock
2014
, vol. 
42
 (pg. 
383
-
391
)
[PubMed]
7
Boomer
 
J.
To
 
K.
Chang
 
K.
Takasu
 
O.
Osborne
 
D.
Walton
 
A.
Bricker
 
T.
Jarman
 
S.
Kreisel
 
D.
, et al 
Immunosuppression in patients who die of sepsis and multiple organ failure
JAMA
2011
, vol. 
306
 (pg. 
2594
-
2605
)
[PubMed]
8
Fox
 
C.J.
Hammerman
 
P.S.
Thompson
 
C.B.
 
Fuel feeds function: energy metabolism and the T-cell response
Nat. Rev. Immunol.
2005
, vol. 
5
 (pg. 
844
-
852
)
[PubMed]
9
Ron-Harel
 
N.
Sharpe
 
A.H.
Haigis
 
M.C.
 
Mitochondrial metabolism in T cell activation and senescence: a mini-review
Gerontology.
2015
, vol. 
61
 (pg. 
131
-
138
)
[PubMed]
10
Pearce
 
E.L.
Poffenberger
 
M.C.
Chang
 
C.H.
Jones
 
R.G.
 
Fueling immunity: insights into metabolism and lymphocyte function
Science
2013
, vol. 
342
 pg. 
1242454
 
[PubMed]
11
Guarda
 
G.
Dostert
 
C.
Staehli
 
F.
Cabalzar
 
K.
Castillo
 
R.
Tardivel
 
A.
Schneider
 
P.
Tschopp
 
J.
 
T cells dampen innate immune responses through inhibition of NLRP1 and NLRP3 inflammasomes
Nature
2009
, vol. 
460
 (pg. 
269
-
273
)
[PubMed]
12
McLoughlin
 
R.M.
Solinga
 
R.M.
Rich
 
J.
Zaleski
 
K.J.
Cocchiaro
 
J.L.
Risley
 
A.
Tzianabos
 
A.O.
Lee
 
J.C.
 
CD4+ T cells and CXC chemokines modulate the pathogenesis of Staphylococcus aureus wound infections
Proc. Natl. Acad. Sci. U.S.A.
2006
, vol. 
103
 (pg. 
10408
-
10413
)
[PubMed]
13
Ghaferi
 
A.A.
Birkmeyer
 
J.D.
Dimick
 
J.B.
 
Complications, failure to rescue, and mortality with major inpatient surgery in medicare patients
Ann. Surg.
2009
, vol. 
250
 (pg. 
1029
-
1034
)
[PubMed]
14
Bennett-Guerrero
 
E.
Welsby
 
I.
Dunn
 
T.J.
Young
 
L.R.
Wahl
 
T.A.
Diers
 
T.L.
Phillips-Bute
 
B.G.
Newman
 
M.F.
Mythen
 
M.G.
 
The use of a postoperative morbidity survey to evaluate patients with prolonged hospitalization after routine, moderate-risk, elective surgery
Anesth. Analg.
1999
, vol. 
89
 (pg. 
514
-
519
)
[PubMed]
15
Kirksey
 
M.
Lin Chiu
 
Y.
Ma
 
Y.
Gonzalez Della Valle
 
A.
Poultsides
 
L.
Gerner
 
P.
Memtsoudis
 
S.G.
 
Trends in in-hospital major morbidity and mortality after total joint arthroplasty: United States 1998–2008
Anesth. Analg.
2012
, vol. 
115
 (pg. 
321
-
327
)
[PubMed]
16
Ackland
 
G.L.
Moran
 
N.
Cone
 
S.
Grocott
 
M.P.
Mythen
 
M.G.
 
Chronic kidney disease and postoperative morbidity after elective orthopedic surgery
Anesth. Analg.
2011
, vol. 
112
 (pg. 
1375
-
1381
)
[PubMed]
17
Ahmed
 
A.A.
Mooar
 
P.A.
Kleiner
 
M.
Torg
 
J.S.
Miyamoto
 
C.T.
 
Hypertensive patients show delayed wound healing following total hip arthroplasty
PLoS One
2011
, vol. 
6
 pg. 
e23224
 
[PubMed]
18
Saleh
 
K.
Olson
 
M.
Resig
 
S.
Bershadsky
 
B.
Kuskowski
 
M.
Gioe
 
T.
Robinson
 
H.
Schmidt
 
R.
McElfresh
 
E.
 
Predictors of wound infection in hip and knee joint replacement: results from a 20 year surveillance program
J. Orthop. Res.
2002
, vol. 
20
 (pg. 
506
-
515
)
[PubMed]
19
Levis
 
W.R.
Robbins
 
J.H.
 
Methods for obtaining purified lymphocytes, glass-adherent mononuclear cells, and a population containing both cell types from human peripheral blood
Blood
1972
, vol. 
40
 (pg. 
77
-
89
)
[PubMed]
20
Reynolds
 
J.M.
Martinez
 
G.J.
Chung
 
Y.
Dong
 
C.
 
Toll-like receptor 4 signaling in T cells promotes autoimmune inflammation
Proc. Natl. Acad. Sci. U.S.A.
2012
, vol. 
109
 (pg. 
13064
-
13069
)
[PubMed]
21
Dracheva
 
S.
Marras
 
S.A.
Elhakem
 
S.L.
Kramer
 
F.R.
Davis
 
K.L.
Haroutunian
 
V.
 
N-methyl-D-aspartic acid receptor expression in the dorsolateral prefrontal cortex of elderly patients with schizophrenia
Am. J. Psychiatry
2001
, vol. 
158
 (pg. 
1400
-
1410
)
[PubMed]
22
Jones
 
R.G.
Thompson
 
C.B.
 
Revving the engine: signal transduction fuels T cell activation
Immunity
2007
, vol. 
27
 (pg. 
173
-
178
)
[PubMed]
23
van der Windt
 
G.J.
O’Sullivan
 
D.
Everts
 
B.
Huang
 
S.C.
Buck
 
M.D.
Curtis
 
J.D.
Chang
 
C.H.
Smith
 
A.M.
Ai
 
T.
Faubert
 
B.
, et al 
CD8 memory T cells have a bioenergetic advantage that underlies their rapid recall ability
Proc. Natl. Acad. Sci. U.S.A.
2013
, vol. 
110
 (pg. 
14336
-
14341
)
[PubMed]
24
Chang
 
C.H.
Curtis
 
J.D.
Maggi
 
L.B.
Faubert
 
B.
Villarino
 
A.V.
O’Sullivan
 
D.
Huang
 
S.C.
van der Windt
 
G.J.
Blagih
 
J.
Qiu
 
J.
, et al 
Posttranscriptional control of T cell effector function by aerobic glycolysis
Cell
2013
, vol. 
153
 (pg. 
1239
-
1251
)
[PubMed]
25
Nagata
 
S.
 
Apoptosis by death factor
Cell
1997
, vol. 
88
 (pg. 
355
-
365
)
[PubMed]
26
Arch
 
R.H.
Thompson
 
C.B.
 
Lymphocyte survival–the struggle against death
Annu. Rev. Cell Dev. Biol.
1999
, vol. 
15
 (pg. 
113
-
140
)
[PubMed]
27
Stahn
 
C.
Buttgereit
 
F.
 
Genomic and nongenomic effects of glucocorticoids
Nat. Clin. Pract. Rheumatol.
2008
, vol. 
4
 (pg. 
525
-
533
)
[PubMed]
28
Naito
 
Y.
Tamai
 
S.
Shingu
 
K.
Shindo
 
K.
Matsui
 
T.
Segawa
 
H.
Nakai
 
Y.
Mori
 
K.
 
Responses of plasma adrenocorticotropic hormone, cortisol, and cytokines during and after upper abdominal surgery
Anesthesiology
1992
, vol. 
77
 (pg. 
426
-
431
)
[PubMed]
29
Shao
 
W.
Yeretssian
 
G.
Doiron
 
K.
Hussain
 
S.N.
Saleh
 
M.
 
The caspase-1 digestome identifies the glycolysis pathway as a target during infection and septic shock
J. Biol. Chem.
2007
, vol. 
282
 (pg. 
36321
-
36329
)
[PubMed]
30
Doitsh
 
G.
Galloway
 
N.L.
Geng
 
X.
Yang
 
Z.
Monroe
 
K.M.
Zepeda
 
O.
Hunt
 
P.W.
Hatano
 
H.
Sowinski
 
S.
Munoz-Arias
 
I.
Greene
 
W.C.
 
Cell death by pyroptosis drives CD4 T-cell depletion in HIV-1 infection
Nature.
2014
, vol. 
505
 (pg. 
509
-
514
)
[PubMed]
31
Gibbison
 
B.
Angelini
 
G.D.
Lightman
 
S.L.
 
Dynamic output and control of the hypothalamic-pituitary-adrenal axis in critical illness and major surgery
Br. J. Anaesth.
2013
, vol. 
111
 (pg. 
347
-
360
)
[PubMed]
32
Kummer
 
J.A.
Broekhuizen
 
R.
Everett
 
H.
Agostini
 
L.
Kuijk
 
L.
Martinon
 
F.
van Bruggen
 
R.
Tschopp
 
J.
 
Inflammasome components NALP 1 and 3 show distinct but separate expression profiles in human tissues suggesting a site-specific role in the inflammatory response
J. Histochem. Cytochem.
2007
, vol. 
55
 (pg. 
443
-
452
)
[PubMed]
33
Munoz-Planillo
 
R.
Kuffa
 
P.
Martinez-Colon
 
G.
Smith
 
B.L.
Rajendiran
 
T.M.
Nunez
 
G.
 
K(+) efflux is the common trigger of NLRP3 inflammasome activation by bacterial toxins and particulate matter
Immunity
2013
, vol. 
38
 (pg. 
1142
-
1153
)
[PubMed]
34
Tschopp
 
J.
Schroder
 
K.
 
NLRP3 inflammasome activation: the convergence of multiple signalling pathways on ROS production?
Nat. Rev. Immunol.
2010
, vol. 
10
 (pg. 
210
-
215
)
[PubMed]
35
Brealey
 
D.
Brand
 
M.
Hargreaves
 
I.
Heales
 
S.
Land
 
J.
Smolenski
 
R.
Davies
 
N.A.
Cooper
 
C.E.
Singer
 
M.
 
Association between mitochondrial dysfunction and severity and outcome of septic shock
Lancet
2002
, vol. 
360
 (pg. 
219
-
223
)
[PubMed]
36
Chacko
 
B.K.
Kramer
 
P.A.
Ravi
 
S.
Benavides
 
G.A.
Mitchell
 
T.
Dranka
 
B.P.
Ferrick
 
D.
Singal
 
A.K.
Ballinger
 
S.W.
Bailey
 
S.M.
, et al 
The Bioenergetic Health Index: a new concept in mitochondrial translational research
Clin. Sci.
2014
, vol. 
127
 (pg. 
367
-
373
)
[PubMed]
37
Wherry
 
E.J.
 
T cell exhaustion
Nat. Immunol.
2011
, vol. 
12
 (pg. 
492
-
499
)
[PubMed]
38
Vaduganathan
 
M.
Ambrosy
 
A.P.
Greene
 
S.J.
Mentz
 
R.J.
Subacius
 
H.P.
Maggioni
 
A.P.
Swedberg
 
K.
Nodari
 
S.
Zannad
 
F.
Konstam
 
M.A.
, et al 
Predictive value of low relative lymphocyte count in patients hospitalized for heart failure with reduced ejection fraction: insights from the EVEREST trial
Circ. Heart Fail.
2012
, vol. 
5
 (pg. 
750
-
758
)
[PubMed]
39
Hagen
 
T.P.
Vaughan-Sarrazin
 
M.S.
Cram
 
P.
 
Relation between hospital orthopaedic specialisation and outcomes in patients aged 65 and older: retrospective analysis of US Medicare data
BMJ
2010
, vol. 
340
 pg. 
c165
 
[PubMed]
40
Reference deleted
[PubMed]
41
Jencks
 
S.F.
Williams
 
M.V.
Coleman
 
E.A.
 
Rehospitalizations among patients in the Medicare fee-for-service program
N. Engl. J. Med.
2009
, vol. 
360
 (pg. 
1418
-
1428
)
[PubMed]
42
Gibson
 
P.H.
Croal
 
B.L.
Cuthbertson
 
B.H.
Small
 
G.R.
Ifezulike
 
A.I.
Gibson
 
G.
Jeffrey
 
R.R.
Buchan
 
K.G.
El-Shafei
 
H.
Hillis
 
G.S.
 
Preoperative neutrophil-lymphocyte ratio and outcome from coronary artery bypass grafting
Am. Heart J.
2007
, vol. 
154
 (pg. 
995
-
1002
)
[PubMed]
43
Christou
 
N.V.
Meakins
 
J.L.
Gordon
 
J.
Yee
 
J.
Hassan-Zahraee
 
M.
Nohr
 
C.W.
Shizgal
 
H.M.
MacLean
 
L.D.
 
The delayed hypersensitivity response and host resistance in surgical patients. 20 years later
Ann. Surg.
1995
, vol. 
222
 (pg. 
534
-
546
)
[PubMed]
44
Puyana
 
J.C.
Pellegrini
 
J.D.
De
 
A.K.
Kodys
 
K.
Silva
 
W.E.
Miller
 
C.L.
 
Both T-helper-1- and T-helper-2-type lymphokines are depressed in posttrauma anergy
J. Trauma
1998
, vol. 
44
 (pg. 
1037
-
1045
)
[PubMed]
45
Bandyopadhyay
 
G.
De
 
A.
Laudanski
 
K.
Li
 
F.
Lentz
 
C.
Bankey
 
P.
Miller-Graziano
 
C.
 
Negative signaling contributes to T-cell anergy in trauma patients
Crit. Care Med.
2007
, vol. 
35
 (pg. 
794
-
801
)
[PubMed]
46
Venet
 
F.
Chung
 
C.S.
Kherouf
 
H.
Geeraert
 
A.
Malcus
 
C.
Poitevin
 
F.
Bohe
 
J.
Lepape
 
A.
Ayala
 
A.
Monneret
 
G.
 
Increased circulating regulatory T cells (CD4(+)CD25 (+)CD127 (-)) contribute to lymphocyte anergy in septic shock patients
Intensive Care Med.
2009
, vol. 
35
 (pg. 
678
-
686
)
[PubMed]
47
Rem
 
J.
Brandt
 
M.R.
Kehlet
 
H.
 
Prevention of postoperative lymphopenia and granulocytosis by epidural analgesia
Lancet
1980
, vol. 
1
 (pg. 
283
-
284
)
[PubMed]
48
Heffernan
 
D.S.
Monaghan
 
S.F.
Thakkar
 
R.K.
Machan
 
J.T.
Cioffi
 
W.G.
Ayala
 
A.
 
Failure to normalize lymphopenia following trauma is associated with increased mortality, independent of the leukocytosis pattern
Crit. Care
2012
, vol. 
16
 pg. 
R12
 
[PubMed]
49
Hotchkiss
 
R.S.
Tinsley
 
K.W.
Swanson
 
P.E.
Chang
 
K.C.
Cobb
 
J.P.
Buchman
 
T.G.
Korsmeyer
 
S.J.
Karl
 
I.E.
 
Prevention of lymphocyte cell death in sepsis improves survival in mice
Proc. Natl. Acad. Sci. U.S.A.
1999
, vol. 
96
 (pg. 
14541
-
14546
)
[PubMed]
50
Hotchkiss
 
R.S.
Tinsley
 
K.W.
Swanson
 
P.E.
Chang
 
K.C.
Cobb
 
J.P.
Buchman
 
T.G.
Korsmeyer
 
S.J.
Karl
 
I.E.
 
Prevention of lymphocyte cell death in sepsis improves survival in mice
Proc. Natl. Acad. Sci. U.S.A.
1999
, vol. 
96
 (pg. 
14541
-
14546
)
[PubMed]
51
Mestas
 
J.
Hughes
 
C.C.
 
Of mice and not men: differences between mouse and human immunology
J. Immunol.
2004
, vol. 
172
 (pg. 
2731
-
2738
)
[PubMed]
52
Khuri
 
S.F.
Henderson
 
W.G.
DePalma
 
R.G.
Mosca
 
C.
Healey
 
N.A.
Kumbhani
 
D.J.
 
Determinants of long-term survival after major surgery and the adverse effect of postoperative complications
Ann. Surg.
2005
, vol. 
242
 (pg. 
326
-
341
)
[PubMed]
53
Gatza
 
E.
Wahl
 
D.R.
Opipari
 
A.W.
Sundberg
 
T.B.
Reddy
 
P.
Liu
 
C.
Glick
 
G.D.
Ferrara
 
J.L.
 
Manipulating the bioenergetics of alloreactive T cells causes their selective apoptosis and arrests graft-versus-host disease
Sci. Transl. Med.
2011
, vol. 
3
 pg. 
67ra68
 
54
Daniele
 
R.P.
Holian
 
S.K.
Nowell
 
P.C.
 
A potassium ionophore (Nigericin) inhibits stimulation of human lymphocytes by mitogens
J. Exp. Med.
1978
, vol. 
147
 (pg. 
571
-
581
)
[PubMed]
55
Busillo
 
J.M.
Azzam
 
K.M.
Cidlowski
 
J.A.
 
Glucocorticoids sensitize the innate immune system through regulation of the NLRP3 inflammasome
J. Biol. Chem.
2011
, vol. 
286
 (pg. 
38703
-
38713
)
[PubMed]
56
Lannan
 
E.A.
Galliher-Beckley
 
A.J.
Scoltock
 
A.B.
Cidlowski
 
J.A.
 
Proinflammatory actions of glucocorticoids: glucocorticoids and TNFalpha coregulate gene expression in vitro and in vivo
Endocrinology
2012
, vol. 
153
 (pg. 
3701
-
3712
)
[PubMed]
57
Tung
 
S.
Shi
 
Y.
Wong
 
K.
Zhu
 
F.
Gorczynski
 
R.
Laister
 
R.C.
Minden
 
M.
Blechert
 
A.K.
Genzel
 
Y.
Reichl
 
U.
Spaner
 
D.E.
 
PPARalpha and fatty acid oxidation mediate glucocorticoid resistance in chronic lymphocytic leukemia
Blood
2013
, vol. 
122
 (pg. 
969
-
980
)
[PubMed]
58
De Iudicibus
 
S.
Stocco
 
G.
Martelossi
 
S.
Londero
 
M.
Ebner
 
E.
Pontillo
 
A.
Lionetti
 
P.
Barabino
 
A.
Bartoli
 
F.
Ventura
 
A.
Decorti
 
G.
 
Genetic predictors of glucocorticoid response in pediatric patients with inflammatory bowel diseases
J. Clin. Gastroenterol.
2011
, vol. 
45
 (pg. 
e1
-
e7
)
[PubMed]
59
Bowers
 
N.L.
Helton
 
E.S.
Huijbregts
 
R.P.
Goepfert
 
P.A.
Heath
 
S.L.
Hel
 
Z.
 
Immune suppression by neutrophils in HIV-1 infection: role of PD-L1/PD-1 pathway
PLoS Pathog.
2014
, vol. 
10
 pg. 
e1003993
 
[PubMed]
60
Hotchkiss
 
R.S.
Monneret
 
G.
Payen
 
D.
 
Immunosuppression in sepsis: a novel understanding of the disorder and a new therapeutic approach
Lancet Infect. Dis.
2013
, vol. 
13
 (pg. 
260
-
268
)
[PubMed]
61
Pillay
 
J.
Kamp
 
V.M.
van Hoffen
 
E.
Visser
 
T.
Tak
 
T.
Lammers
 
J.W.
Ulfman
 
L.H.
Leenen
 
L.P.
Pickkers
 
P.
Koenderman
 
L.
 
A subset of neutrophils in human systemic inflammation inhibits T cell responses through Mac-1
J. Clin. Invest.
2012
, vol. 
122
 (pg. 
327
-
336
)
[PubMed]
62
Henson
 
S.M.
Lanna
 
A.
Riddell
 
N.E.
Franzese
 
O.
Macaulay
 
R.
Griffiths
 
S.J.
Puleston
 
D.J.
Watson
 
A.S.
Simon
 
A.K.
Tooze
 
S.A.
Akbar
 
A.N.
 
p38 signaling inhibits mTORC1-independent autophagy in senescent human CD8(+) T cells
J. Clin. Invest.
2014
, vol. 
124
 (pg. 
4004
-
4016
)
[PubMed]

Supplementary data