Immunosuppression in Patients Who Die of Sepsis and Multiple …€¦ · Adaptive T cells (CD4 and CD8) Immune supressor cells Regulatory T cells, myeloid-derived suppressor cells

CARING FOR THECRITICALLY ILL PATIENT

Immunosuppression in Patients Who Dieof Sepsis and Multiple Organ FailureJonathan S. Boomer, PhDKathleen To, MDKathy C. Chang, PhDOsamu Takasu, MDDale F. Osborne, BSAndrew H. Walton, MSTraci L. Bricker, BSStephen D. Jarman II, BSN, RNDaniel Kreisel, MD, PhDAlexander S. Krupnick, MDAnil Srivastava, MDPaul E. Swanson, MDJonathan M. Green, MDRichard S. Hotchkiss, MD

SEPSIS IS RESPONSIBLE FOR MORE

than 225 000 deaths annually inthe United States.1 Developingnew therapies for sepsis has been

particularly challenging, with more than25 unsuccessful drug trials.2-6 Charac-terized by an initial intense inflamma-tory response or “cytokine storm,” pa-tients with sepsis may present with fever,shock, altered mental status, and organdysfunction.5-8 Numerous investigativeagents have been directed at down-modulating this initial phase. Im-proved clinical management algo-rithms have led to survival of the majorityof patients in this early period.9-11 How-ever, those who survive early sepsis of-ten develop nosocomial infections withorganismsnot typicallypathogenic in im-munocompetent hosts and have reacti-vation of latent viruses.9,12,13 These ob-servations have led to the controversial hypothesis that the early hyperinflam-

matory state evolves to a subsequent hy-poinflammatory state with significant im-munosuppression.14-19 Although animalstudies demonstrate progression to animmunosuppressive phase, epidemio-

See also pp 2614 and 2618.

CME available online atwww.jamaarchivescme.comand questions on p 2625.

Context Severe sepsis is typically characterized by initial cytokine-mediated hyper-inflammation. Whether this hyperinflammatory phase is followed by immunosuppres-sion is controversial. Animal studies suggest that multiple immune defects occur in sep-sis, but data from humans remain conflicting.

Objectives To determine the association of sepsis with changes in host innate and adap-tive immunity and to examine potential mechanisms for putative immunosuppression.

Design, Setting, and Participants Rapid postmortem spleen and lung tissue har-vest was performed at the bedsides of 40 patients who died in intensive care units (ICUs)of academic medical centers with active severe sepsis to characterize their immune statusat the time of death (2009-2011). Control spleens (n=29) were obtained from patientswho were declared brain-dead or had emergent splenectomy due to trauma; control lungs(n=20) were obtained from transplant donors or from lung cancer resections.

Main Outcome Measures Cytokine secretion assays and immunophenotyping ofcell surface receptor-ligand expression profiles were performed to identify potentialmechanisms of immune dysfunction. Immunohistochemical staining was performedto evaluate the loss of immune effector cells.

Results The mean ages of patients with sepsis and controls were 71.7 (SD, 15.9) and52.7 (SD, 15.0) years, respectively. The median number of ICU days for patients withsepsis was 8 (range, 1-195 days), while control patients were in ICUs for 4 or fewer days.The median duration of sepsis was 4 days (range, 1-40 days). Compared with controls,anti-CD3/anti-CD28–stimulated splenocytes from sepsis patients had significant reduc-tions in cytokine secretion at 5 hours: tumor necrosis factor, 5361 (95% CI, 3327-7485)pg/mL vs 418 (95% CI, 98-738) pg/mL; interferon �, 1374 (95% CI, 550-2197) pg/mLvs 37.5 (95% CI, −5 to 80) pg/mL; interleukin 6, 3691 (95% CI, 2313-5070) vs 365(95% CI, 87-642) pg/mL; and interleukin 10, 633 (95% CI, −269 to 1534) vs 58 (95%CI, −39 to 156) pg/mL; (P� .001 for all). There were similar reductions in 5-hour lipo-polysaccharide-stimulated cytokine secretion. Cytokine secretion in sepsis patients wasgenerally less than 10% that in controls, independent of age, duration of sepsis, corti-costeroid use, and nutritional status. Although differences existed between spleen andlung, flow cytometric analysis showed increased expression of selected inhibitory recep-tors and ligands and expansion of suppressor cell populations in both organs. Unique dif-ferences in cellular inhibitory molecule expression existed in immune cells isolated fromlungs of sepsis patients vs cancer patients and vs transplant donors. Immunohistochemi-cal staining showed extensive depletion of splenic CD4, CD8, and HLA-DR cells and ex-pression of ligands for inhibitory receptors on lung epithelial cells.

Conclusions Patients who die in the ICU following sepsis compared with patientswho die of nonsepsis etiologies have biochemical, flow cytometric, and immunohis-tochemical findings consistent with immunosuppression. Targeted immune-enhancing therapy may be a valid approach in selected patients with sepsis.JAMA. 2011;306(23):2594-2605 www.jama.com

Author Affiliations are listed at the end of this article.Corresponding Author: Richard S. Hotchkiss, MD, De-partment of Anesthesiology, Washington UniversitySchool of Medicine, 660 S Euclid, St Louis, MO 63110([email protected]).Caring for the Critically Ill Patient Section Editor: DerekC. Angus, MD, MPH, Contributing Editor, JAMA([email protected]).

2594 JAMA, December 21, 2011—Vol 306, No. 23 ©2011 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by a Non-Human Traffic (NHT) User on 09/27/2020

logic studies in clinical sepsis arelacking.14-19 The purpose of this investi-gation was to assess evidence of immu-nosuppression in sepsis and to deter-mine mechanisms that might beresponsible for the presumed impairedimmunity. Cells from spleen and lungwere studied to compare and contrast thefunctional status and phenotype of cellsfrom a lymphoid organ and a periph-eral organ that is a frequent site of noso-comial infection.

METHODSAn overview of study design includ-ing the purpose of the various immu-nologic tests is shown in TABLE 1. Meth-ods are described in further detail in theeAppendix (available at http://www.jama.com).

Inclusion and Exclusion CriteriaPatients with sepsis who died whileundergoing treatment in surgical ormedical intensive care units (ICUs)were included. Sepsis was defined usinga consensus panel definition: presence

of microbiologically proven, clinicallyproven, or suspected infection andpresence of systemic inflammatory re-sponse syndrome (SIRS). The diagnosisof SIRS required at least 2 of the follow-ing:hypothermia(�36°C)orhyperther-mia (�38°C); tachycardia (�90/min);tachypnea(�20breaths/min)and/orar-terialPCO232mmHgorlowerand/orme-chanical ventilation; and leukocytosis(�12 000/µL)orleukopenia(�4000/µL)and/or left-shifted white blood cell dif-ferential count of 10% or higher.20

To try to limit potential confound-ing effects of other conditions affect-ing immunity, patients with cancer,chronic viral infections (human immu-nodeficiency virus, hepatitis B or C), orautoimmune diseases and patients tak-ing high-dose corticosteroids (hydro-cortisone, �200 mg/d) or immunosup-pressive medications were excluded.

Control Population

Controlspleenswereobtainedfromcriti-cally ill patientswithnosepsis including

those declared dead by neurological cri-teriaandtraumapatientswhohademer-gent splenectomy because of splenic in-jury (see eAppendix for inclusion andexclusion criteria). Patients who metbrain death criteria typically had beenmechanically ventilated for 48 to 72hours, required vasopressors to main-tain adequate organ perfusion, and hadassociated organ injuries. Patients re-quiring mechanical ventilation for morethan 4 days were excluded because of thehigh incidence of ventilator-associatedpneumonia. Control lung tissue was ob-tained from the excess tissue of trans-plant donor lungs and from the non–tumor-involved tissue of lobectomiesperformed for cancer (eAppendix).

All studies were approved by theWashington University or St John’s Hos-pital human research protection of-fices. For tissue obtained by rapid au-topsy of patients who died in the ICU,the studies were determined by theWashington University and St John’sHospital human research protection of-

Table 1. Immune System Analysis

Immune FunctionAnalysis of Cell

Populations Assay Methods Organ/Tissue ResultsImmune effector cells

Innate Dendritic cells,macrophages, naturalkiller cells, monocytes Flow cytometry,

immunohistochemistrySpleen and splenocytesLung and lung cells

Splenocytes: Figure 1,Figure 2, Figure 3Lung cells: Figure 4Spleen: Figure 5Lung: Figure 6, Figure 7

Adaptive T cells (CD4 and CD8)Immune supressor cells Regulatory T cells,

myeloid-derivedsuppressor cells

Flow cytometry Splenocytes: regulatoryT cellLung cells: regulatoryT cells and myeloid-derived suppressor cells

Results section of text

Immune cell receptor expressionExpression of moleculesassociated with antigenpresentation

HLA-DR expression Flow cytometry,immunohistochemistry

Splenocytes, spleen Splenocytes: Figure 3Spleen: Figure 5

Expression of molecules thatenhance immune responses

Receptors: CD28, CD69,IL-2�, IL-7R�Ligands: CD86

Flow cytometry,immunohistochemistry

SplenocytesLung cells

Splenocytes: Figure 2,Figure 3, Figure 5Lung cells: Figure 4

Expression of molecules thatsuppress immune responses

Receptors: BTLA, PD-1,CTLA-4Ligands: PD-L1, PD-L2,HVEM

Flow cytometry,immunohistochemistry

Splenocytes or lung cellsSpleen or lung

Splenocytes: Figure 2,Figure 3Lung cells: Figure 4Lung: Figure 6, Figure 7

Immune cell effector functionCytokine secretion

Innate immune response:lipopolysaccharide

Natural killer cells,dendritic cells,macrophages,monocytes

Enzyme-linkedimmunosorbentassay

Splenocytes Figure 1

Adaptive immune response:anti-CD3/anti-CD28

T cells

Abbreviations: BTLA, B- and T-lymphocyte attenuator; CTLA, cytotoxic T-lymphocyte antigen; HVEM, herpes virus entry mediator; IL, interleukin; PD, programmed cell death.

IMMUNOSUPPRESSION IN SEPSIS AND ORGAN FAILURE

©2011 American Medical Association. All rights reserved. JAMA, December 21, 2011—Vol 306, No. 23 2595


fices not to constitute human subjects re-search because there was no interactionwith the patient prior to death. After thepatient had died, written permission tocollect postmortem tissue for researchpurposes was obtained from the pa-tient’s next of kin. For lung resectionspecimens, all participants provided writ-ten informed consent prior to surgery.For lung tissue obtained from trans-plant donors, the tissue was collected af-ter death and provided to the labora-tory completely anonymized; therefore,it was also determined not to constitutehuman subjects research. However, thetransplant donation consent signed bynext of kin included a provision that tis-sue may be used for research purposes.Similarly, spleens were obtained fromMid-America Transplant, St Louis, Mis-souri, from brain-dead organ donorswhose next of kin gave written consentfor use of tissues for research purposes.

Splenic and Lung Harvestingand Cell Isolation

Spleen samples were obtained from 40patients with sepsis and 29 patientswithout sepsis within 30 to 180 min-utes of death. Splenocytes were disso-ciated and resuspended in sterile me-dia.21 Splenocytes either were studiedacutely or stored at 4°C for subse-quent analysis within 12 to 72 hours.Cell counting and viability were deter-mined as described previously.21

Lungs samples from patients with sep-sis (n=34)orwithout sepsis (n=20)wereimmediately processed. Lungs were fixedovernight for immunohistochemistry orsingle-cell suspensions were prepared,counted, viability determined, andstained for flow cytometry.

Cytokine Productionof Splenocytes

Mononuclear cells were prepared by den-sity gradient centrifugation.21 Cell viabil-ity was greater than 85% to 90%. Cellswere stimulated with lipopolysaccha-ride, anti-CD3/anti-CD28,orphorbol12-myristate 13-acetate (PMA)/ionomy-cin. Supernatants were harvested at 5 and22 hours and cytokines quantitated byenzyme-linked immunosorbent assay.

Flow Cytometric Analysis of Spleenand Lung CellsCell suspensions were incubated withisotype control or type-specific anti-bodies. To determine the percentage ofcells positive for each marker, the per-centage positive isotype control wassubtracted from that within the posi-tive gate of the type-specific marker.The mean fluorescence index (MFI)was determined by subtracting the geo-metric MFI of the isotype control fromthe geometric MFI of the type-specificmarker.

Immunohistochemistry of Tissues

For spleen tissue, formalin-fixed sec-tions underwent antigen retrieval fol-lowed by incubation with primary an-tibodies or isotype-matched controls.Slides were sequentially incubated withbiotinylated antibody and peroxidase-labeled streptavidin and immunoreac-tive cells were visualized with diamino-benzidine-chromogen. Lung tissue wasfixed in paraformaldehyde. Sections un-derwent endogenous peroxidasequenching and antigen retrieval fol-lowed by incubation with primary an-tibodies or isotype-matched con-trols.22 Slides were incubated inhorseradish peroxidase (HRP)–conjugated antibodies followed by vi-sualization with HRP-AB-C substrate.

Evaluation of Tissue Slides

Slides were evaluated in blinded fashionand scored as described previously.23

Statistical Analysis

Differences in sepsis vs nonsepsis cyto-kineproductionandphenotypicexpres-sion were analyzed by 2-tailed nonpara-metric t test(Mann-WhitneyU test)usingthe statisticalprogramGraphPadPrism,version5.0(GraphPadSoftware). Statis-tical significance was set at P� .05.

RESULTSPatient Characteristics

The most common etiologies of sepsiswere ventilator-associated pneumo-nia and peritonitis (TABLE 2 and eTable1). Other causes included necrotizingfasciitis, retroperitoneal abscess, in-

fected intravascular catheters, urinarytract infections, intrapelvic abscess, andosteomyelitis. The nonsepsis patientpopulation for control spleen tissuesconsisted of 20 patients who met braindeath criteria and 9 patients who hadsplenectomies because of traumatic in-jury (Table 2 and eTable 2). Controlpopulations for lung consisted of speci-mens from lung transplant donors(n=10) or lobectomy for lung cancer(n=10).

The mean ages of patients with sep-sis and controls were 71.7 (SD, 15.9)and 52.7 (SD, 15.0) years, respec-tively. The median number of ICU daysfor sepsis patients was 8 (range, 1-195),while the median duration of sepsis was4 days, with a range of 1 to more than40 days (Table 2). Control patients werein ICUs for 4 or fewer days. Patientswith sepsis had numerous comorbidi-ties; in contrast, comorbidities weremuch less frequent in organ donor con-trols. The mean serum albumin level forsepsis patients was 2.4 (SD, 0.62) g/dL,with a range of 1.0 to 3.9 g/dL (nor-mal hospital laboratory range for albu-min is 3.6-5.0 g/dL) (eAppendix andeFigure 4).

Cytokine Production in Splenocytes

Compared with controls, splenocytesfrom sepsis patients produced signifi-cantly fewer cytokines (tumor necro-sis factor [TNF], interferon [IFN-�],and interleukins [IL] 6 and 10) at 5 and22 hours (FIGURE 1, eFigure 1, eTable3, and eTable 4). Decreased cytokineproduction was profound at 5 hours,when selected sepsis tissues producedminimal amounts of cytokines regard-less of stimulus. For example, cyto-kine production stimulated with anti-CD3/anti-CD28 at 5 hours fromcontrols vs sepsis patients, respec-tively, resulted in TNF, 5361 (95% CI,3327-7485) pg/mL vs 418 (95% CI, 98-738) pg/mL; IFN-�, 1374 (95% CI, 550-2197) pg/mL vs 37.5 (95% CI, −5 to 80)pg/mL; IL-6, 3691 (95% CI, 2313-5070) pg/mL vs 365 (95% CI, 87-642)pg/mL; and IL-10, 633 (95% CI, −269to 1534) pg/mL vs 58 (95% CI, −39 to156) pg/mL (P� .001 for all). There




were similar reductions in 5-hour li-popolysaccharide- and PMA/ionomy-cin–stimulated cytokine secretion(Figure 1, eFigure 1, eTable 3, andeTable 4).

By 22 hours, some splenocytes fromsepsis patients exhibited partial recov-ery of cytokine production. Collec-tively, secretion of cytokines by stimu-lated sepsis splenocytes at 5 hours wasless than 10% of that of controls. How-ever, by 22 hours, secretion had in-creased to approximately one-third thatof controls and, for selected sepsis pa-tients, cytokine production was similarto that of controls (Figure 1). Cell vi-ability at 5 and 22 hours was 93% and86%, respectively, for controls vs 81%and 60%, respectively, for sepsis patients.

To determine if the age differentialbetween controls and sepsis patientswas associated with changes in stimu-lated cytokine production, compari-son of data from sepsis patients aged 52years or younger (n=5) vs older than52 years (n=21) was performed and re-vealed no statistical differences be-tween these 2 groups (eFigure 2). Fur-thermore, for virtually all cytokines,sepsis patients in both age groups werestatistically different from controls. Todetermine if duration of sepsis af-fected cytokine secretion, data from pa-tients who had sepsis for 4 or fewer dayswere compared with that of patientswho had sepsis for more than 4 days.No significant difference was seen be-tween these 2 groups (eFigure 3). Like-wise, because of potential confound-ing effects of low-dose corticosteroidson cytokine production, comparisonswere also made between sepsis pa-tients receiving corticosteroids (n=9)and not receiving corticosteroids(n=15). Data from both of these groupswere statistically different from that ofcontrols but were not different from thatof each other (eTable 4). To deter-mine if the patient nutritional state (asreflected by serum albumin) had an ef-fect on cytokine secretion in sepsis pa-tients, TNF and IL-6 secretion wereplotted against patient serum albumin(eFigure 4). No correlation was ob-served.

Analysis of Immune CellPopulations and Receptor ProfilesTo assess possible etiologies for themarkedly depressed cytokine secre-tion, we performed flow cytometricanalysis and examined expression of cellsurface receptors important in cellu-lar activation (FIGURE 2, FIGURE 3, andeTable 5). The splenic T cells of sepsispatients had increased expression of ac-tivation marker CD69 on both CD4 andCD8 subsets and of IL-2R� (CD25) onCD4 T cells, when analyzed both as per-centage of positive cells and as cellu-lar expression (as determined by geo-mean fluorescence intensity [MFI]).Expression of the percentage positivecells and MFI of the potent positivecostimulatory receptor CD28 was de-creased on CD4 T cells in sepsis; MFIfor CD28 was also decreased on CD8T cells. Inhibitory members of the CD28family programmed cell death 1 (PD-1)(CD279) and cytotoxic T-lymphocyteantigen 4 (CTLA-4) (CD152) were alsoexamined. Sepsis patients had signifi-cantly increased percentages of CD4 Tcells expressing PD-1 and increasedCTLA-4 positive CD8 T cells. Both MFIand the percentage of cells expressingthe IL-7 receptor � (IL-7R�, CD127),which is critical for cell survival and isdecreased in cell exhaustion, were re-duced on CD4 and CD8 T cells in sep-sis spleen.

CD28 and PD-1 are engaged by theirligands (CD80/CD86 for CD28, PD-L1/PD-L2 for PD-1), which are ex-pressed pr imar i ly on ant igen-presenting cells (APCs), includingdendritic cells and macrophages. Simi-lar to findings on T cells, the costimu-latory ligand CD86 was decreased andthe inhibitory ligand, PD-L1, was in-creased on both macrophages and otherAPCs (Figure 2 and Figure 3). HLA-DRexpression was highly significantly de-creased on these APCs.

Cells fromlungtissuewereextensivelycharacterizedforactivatingandinhibitoryreceptors (eAppendix), and data are pre-sented for markers in which we de-tected meaningful differences (FIGURE 4and eTable 6). The majority of T cellsfrom lungs of all groups expressed

Table 2. Clinical Characteristics

CharacteristicsNo. of

Patientsa

Patients with sepsis (n = 40)Age, mean (SD) [range], y 71.7 (15.9)

[29-94]

SexMale 20

Female 20

Site of infection

Intrapelvic abscess 1

Intravascular catheters 2

Necrotizing fasciitis 1

Osteomyelitis 1

Pneumonia 22

Peritonitis 21

Retroperitoneal abscess 1

Urinary tract infection 3

Days in hospital, median (range) 11 (1-195)

Days in intensive care unit,median (range)

8 (1-195)

Days of sepsis, median (range)b 4 (1-�40)

ComorbiditiesDiabetes 10

Heart disease 18

Morbid obesity 3

Neurologic 6

Renal disease 3

Respiratory 10

Liver 2

Organ failureCirculatory, vasopressors 36

Hepatic 12

Renal 28

Respiratory 35

MicrobeGram-positive 18

Gram-negative 25

Fungal 4

Control spleen donor patients (n = 29)Age, mean (SD) [range], y 52.7 (15.0)

[24-84]

Sex

Male 17

Female 12

Admission diagnosis

Motor vehicle collision, headtrauma

1

Motor vehicle collision,abdominal trauma

2

Gunshot wound, head 2

Gunshot wound, abdominal 2

Abdominal trauma rupturedspleen

5

Blunt head trauma 1

Anoxic brain injury 3

Cerebrovascular accident withinternal hemorrhage

13

aData are expressed as No. of patients unless otherwiseindicated.

bThe number of days that the patient was septic was fre-quently difficult to know with certainty; some patientsmay have been home with sepsis for several days priorto hospital admission.




PD-1. However, the MFI of PD-1 washigher on CD4 cells from sepsis pa-tients compared with both transplantand lung cancer controls. Analysis ofB- and T-lymphocyte attenuator (BTLA)(CD272), another T-cell inhibitory li-gand, revealed that BTLA was also ex-pressed on the majority of T cells butwas increased primarily on cells fromlung cancer patients. Interestingly, bothPD-L1 and PD-L2 expression weremarkedly increased on plasmacytoiddendritic cells of sepsis patients, pre-senting opportunity for potent inhibi-tory receptor-ligand interactions. In

contrast to plasmacytoid dendritic cells,PD-L2 was not detected on myeloiddendritic cells, and PD-L1 was ex-pressed at low levels that were not sta-tistically different between sepsis pa-tients and controls.

Expansion of suppressive cells, in-cluding regulatory T cells and myeloid-derived suppressor cells (MDSCs), hasbeen reported in sepsis and provides an-other plausible mechanism for immu-nosuppression.10 In spleen, regulatoryT cells were increased approximately2-fold in sepsis vs control patients(mean, 8.5% [SD, 0.9%] vs 4.6% [SD,

0.6%], respectively; P� .05). In con-trast, in lung, no increase in regula-tory T cells was detected, but there wereincreased cells consistent with anMDSC phenotype, defined by Lin1-/HLA-DR low/CD33/CD11b high stain-ing, isolated from sepsis vs control tis-sue (47.9% vs 15.7%; P� .01).

Spleen Immunohistochemistry

Evaluation of spleen tissue demon-stratedcellular loss inperiarteriolar lym-phoid sheath (PALS) and diminishednumber and size of splenic follicles insepsispatients, aspreviously reported.24

Figure 1. Cytokine Secretion in Stimulated Splenocytes

5-Hour splenocyte stimulation 22-Hour splenocyte stimulation

18 000

12 000

6000

0

pg/m

L

Lipopolysaccharidestimulation

30 000

20 000

10 000

0

Lipopolysaccharidestimulation

15 000

10 000

5000

0

Anti-CD3/CD28stimulation

60 000

45 000

30 000

15 000

0

Anti-CD3/CD28stimulation

TNFA

pg/m

L

900

600

300

0

30 000

20 000

10 000

0

7500

6000

1500

4500

3000

0

35 000

28 000

7000

21 000

14 000

0

IFN-γB

pg/m

L

35 000

28 000

21 000

14 000

7000

0

45 000

30 000

15 000

0

12 000

9000

6000

3000

0

45 000

30 000

15 000

0

IL-6C

pg/m

L

5425

5250

350

525

700

175

0

No Sepsis Sepsis

50 000

40 000

30 000

20 000

10 000

0

No Sepsis Sepsis

9000

87001200

900

600

300

0

No Sepsis Sepsis

40 000

30 000

20 000

10 000

0

No Sepsis Sepsis

IL-10D

Spleens were harvested from patients who died of sepsis (n=24-26) or nonsepsis etiologies (n=20-21). Cells were dissociated and washed and viability determined bytrypan blue exclusion. Viable splenocytes (1�107) were stimulated with lipopolysaccharide or anti-CD3/anti-CD28 antibody. Supernatants were harvested at 5 and 22hours and tumor necrosis factor (TNF), interferon � (IFN-�), and interleukins (IL) 6 and 10 were measured by enzyme-linked immunosorbent assay. There was a markeddecrease in cytokine secretion in sepsis patients vs nonsepsis controls. Data were analyzed by 2-tailed nonparametric t test (Mann-Whitney U test). Each data markerrepresents an individual patient. Horizontal lines represent mean values. P�.001 for all plots, except P�.01 for TNF with lipopolysaccharide stimulation at 22 hours.




Figure 2. Expression of Cell Surface Receptors on Splenic CD4 and CD8 T Cells

CD4 T cells CD8 T cells

100

75

50

25

0

Per

cent

age

Per

cent

age

Uni

ts

Uni

ts

Flow cytometry,% positive

P = .05 120

100

80

60

40

20

0


P = .001500

400

300

200

100

0

Geo-mean fluorescenceintensity

P = .007 500

400

300

200

100

0


P<.001

CD69A

75

50

25

0

Per

cent

age

Uni

ts

Uni

ts

Per

cent

age

P = .02 50

40

30

20

10

0

P = .1145

30

15

0

P = .69 18

12

6

0

P = .78

PD-1B

100

80

60

40

Per

cent

age

Per

cent

age

Uni

ts

Uni

ts

P<.02

100

75

50

25

0

P<.05400

300

200

100

0

P<.001 106

120

80

40

0

P<.001

IL-7RαC

30

20

10

0

Per

cent

age

Per

cent

age

Uni

ts

Uni

ts

P = .80 50

40

30

20

10

0

P = .0225

20

15

10

5

0

P<.001 15

10

5

0

P = .98

CTLA-4D

80

60

40

20

0

Per

cent

age

Per

cent

age

Uni

ts

Uni

ts

P<.001 30

20

10

0

P = .2180

60

40

20

0

P = .003 5

4

3

2

1

0

P = .25

IL-2RαE

120

90

60

30

No Sepsis Sepsis

Per

cent

age

Per

cent

age

Uni

ts

Uni

ts

P<.001 80

60

40

20

No Sepsis Sepsis

P = .07600

400

200

0

No Sepsis Sepsis

P = .003 100

75

50

25

0

No Sepsis Sepsis

P<.01

CD28F

See Figure 3 legend for explanation of geo-mean fluorescence intensity units, laboratory methods and statistical analysis. Compared with nonsepsis controls (n=24-26), sepsis patients (n=28-31) had activated T cells (increased CD69 in CD4 and CD8 T cells as well as increased CD25 [interleukin {IL} 2 receptor �] in CD4 T cells).Despite an activation phenotype, sepsis induced down-regulation of positive costimulatory receptors (CD28 in CD4 and CD8 T cells) as well as increased inhibitoryreceptors (programmed cell death 1 [PD-1] for CD4 T cells and cytotoxic T-lymphocyte antigen 4 [CTLA-4] for CD8 T cells). The IL-7 receptor � chain (CD127) wasdecreased in CD4 and CD8 T cells in sepsis.




Significantly reduced numbers of CD4,CD8,andHLA-DRcells (dendritic cells,macrophages, and B cells) typified sep-sis splenic tissue (FIGURE 5), though asmall subset of sepsis tissue retained a“normal”complementofthesecellpopu-lations (eTable 7). Nineteen of 22 sep-sis patients (but no controls [n=12])had either decreased HLA-DR positiv-ity, depletion of HLA-DR cells, or both.Conversely, in red pulp, increased ex-pression of HLA-DR on sinusoidal en-dothelial lining cells was noted in 16of 26 sepsis patients (Figure 5) but notin any controls. PD-1 was noted in fol-licular dendritic cells but generally notin T-cell zone dendritic elements. Incontrast, PD-L1 was more typically seenin T-cell zone dendritic cells. In sepsisspleen tissue with severe cellular deple-

tion, the dominant PD-L1–positivePALS population was capillary endo-thelium, and this was increased rela-tive to controls (eFigure 5). Macro-phages were consistently PD-L1 positivein all cell zones and were increased inPALS and red pulp. PD-L1 was uni-formly present in arteriolar and sinu-soidal endothelium in both sepsis andcontrol spleen. PD-1 reactivity was un-evenly expressed in arteriolar endothe-lium and absent on sinusoidal linings.Both B and T lymphocytes were vari-ably positive for PD-1 and PD-L1 inboth sepsis and control patients.

LigandsforT-Cell InhibitoryReceptorsin Lung Parenchymal Cells

Signaling through inhibitory receptorsonT cells requires engagement by their cog-

nate ligand. While typically expressed byAPCs, these may be induced on nonim-mune cells by inflammatory cytokines,perhaps serving to locally inhibit T cellsand dampen inflammation. Immunohis-tochemistry showed intense airway epi-thelial staining for herpes virus entry me-diator (HVEM), the receptor for BTLA,on lung isolated from the majority of sep-sis patients but not controls (FIGURE 6and eTable 8). This was true for pa-tients both with and without pneumo-nia. Herpes virus entry mediator was ex-pressed on macrophages from bothgroups. We also stained for PD-L1 andPD-L2, the ligands for PD-1 (FIGURE 7).PD-L1 and PD-L2 appeared to be ex-pressed in sepsis lung and lung resec-tions to a greater degree compared withtransplant donor lung.

Figure 3. Expression of Cell Surface Receptors on Splenic Antigen-Presenting Cells and Tissue Macrophages

Antigen-presenting cells Tissue macrophages

100

75

50

25

0

Per

cent

age

Per

cent

age

Uni

ts

Uni

ts


P = .06 100

75

50

25

0


P<.05200

150

100

50

0


P = .002 160

120

80

40

0


P<.001

CD86A

100

75

50

25

0

Per

cent

age

Per

cent

age

Uni

ts

Uni

ts

P<.001100

75

50

25

0

P = .00140

50

30

20

10

0

P = .0225

20

15

10

5

0

P = .08

PD-L1B

120

100

80

60

0

20

40

No Sepsis Sepsis

Per

cent

age

Per

cent

age

Uni

ts

Uni

ts

P = .003

120

90

60

30

0

No Sepsis Sepsis

P<.001

3500

2800

2100

1400

700

0

No Sepsis Sepsis

P<.0011500

1200

900

600

300

0

No Sepsis Sepsis

P<.001

HLA-DRC

Splenocytes (2�106) were stained with fluorescently conjugated antibodies or isotype-matched control antibodies and analyzed by flow cytometry (eAppendix). Apositive gate was established based on isotype control staining. The percentage positive for each marker was defined by subtracting the percentage within the positivegate in the isotype control from the percentage within the positive gate in the specific stain. The geo-mean fluorescence intensity was determined by subtraction of thenonspecific fluorescence of the isotype control antibody from the fluorescence of the specific conjugated antibody and is expressed in units, which are an average offluorescence intensity of the data collected within the selected gate after subtracting fluorescence intensity of isotope control. Antigen-presenting cells, ie, dendriticcells and macrophages/monocytes, as well as tissue-specific macrophages, showed an immunosuppressive phenotype in sepsis as evidenced by decreased expressionof CD86 and HLA-DR. In addition, antigen-presenting cells from sepsis patients had increased expression of PD-L1, the ligand for the inhibitory receptor programmedcell death 1 PD-1 on T cells. Each data marker represents an individual patient. Statistical analysis was performed by 2-tailed nonparametric t test (Mann-Whitney Utest). Horizontal lines represent mean values.




COMMENTEarly sepsis is characterized by exces-sive inflammation in what is often termeda “cytokine storm.”2-8 As sepsis persists,patients often have reactivation of en-dogenous viruses and develop nosoco-mial infections with opportunistic patho-gens.9,12,13 Investigators have argued thatthese findings suggest that patients withsepsis enter an immunosuppressive state,but this is highly controversial and mostpotential therapy for sepsis remains fo-

cused on blocking immune activa-tion.14,25-28 While several potential ab-normalities have been identified, acomprehensive analysis of the immunestatus of patients who die of sepsis hasnot been conducted and mechanistic ex-planations remain speculative.25-29 Thepresent study shows that splenocytesfrom sepsis patients had highly signifi-cant functional impairments as evi-denced by major reductions in cyto-kine secretion. Multiple inhibitory

mechanisms were identified, includingdominance of inhibitory over activat-ing receptors, expansion of suppressivecell types, and induction of inhibitoryligands on both APCs and tissue paren-chymal cells. These findings are presentin the setting of apoptosis-induceddepletion of immune cells (Figure 5).23,24

Both proinflammatory and anti-inflammatory cytokines were impaired.These spleen data are consistent withmultiple previous studies examining

Figure 4. Expression of Cell Surface Receptors on Cells Isolated from Lung Tissue

CD4 T cells

100

80

60

40

20

0

Per

cent

age

Uni

ts



6070

30

5040

20100



P = 0.06

P<.05

PD-1A

100

80

60

40

20

0

Per

cent

age

Uni

ts

50

60

40

30

20

10

0

P = .05 P = .08

B- and T-lymphocyte attenuatorB

CD8 T cells

100

80

60

40

20

0

70

3040

6050

2010

0

Per

cent

age

Uni

ts

PD-1C

Per

cent

age

Uni

ts

100

80

60

40

20

0

25

20

15

10

5

0

B- and T-lymphocyte attenuatorD

Plasmacytoid dendritic cells

Per

cent

age

Uni

ts

100

80

60

40

20

0

800

600

400

200

0

P<.001P<.005

P = .07PD-L1E

100

80

60

40

20

0

150

75

125

100

50

25

0

Lungcancer

resection

Lungcancer

resection

Transplantdonor

Sepsis Transplantdonor

Sepsis

P<.05

Per

cent

age

Uni

ts

PD-L2F

Lungcancer

resection

Lungcancer

resection

Transplantdonor

Sepsis Transplantdonor

Sepsis

Lung tissue was digested via collagenase followed by isolation of single cells. Lung cells (3�106) were stained with fluorescently conjugated antibodies or isotype-matchedcontrol antibodies and analyzed via flow cytometry (eAppendix). Lung CD4 and CD8 T cells from sepsis patients had increased programmed cell death 1 (PD-1), as de-termined by geo-mean fluorescence intensity (MFI), compared with nonsepsis controls. Data for MFI are expressed in units, which are an average of fluorescence intensityof the data collected within the selected gate after subtracting fluorescence intensity of isotope control. The majority of T cells from both sepsis and nonsepsis lung alsoexpressed PD-1 (�50%) and B- and T-lymphocyte attenuator (BTLA) (�60%). Lung CD4 T cells from resection controls had elevated BTLA expression (in percentage andMFI) compared with sepsis patients. Plasmacytoid dendritic cells, a subset of dendritic cells, isolated from sepsis lungs had increased PD-L1 (in percentage and MFI) andPD-L2 (in percentage), ligands for PD-1 expressed on lung T cells, compared with both nonsepsis control groups. For PD-1 analysis, n=22 (CD4) and n=20 (CD8) forsepsis patients, n=9 (CD4) and n=8 (CD8) for transplant donor controls, and n=8 (both CD4 and CD8) for lung resection controls. For BTLA analysis, n=21 (CD4) andn=19 (CD8) for sepsis patients, n=9 (CD4) and n=8 (CD8) for transplant donor controls, and n=8 (both CD4 and CD8) for lung resection controls. For plasmacytoiddendritic cell analysis, n=17 for sepsis patients, n=6 for transplant donor controls, and n=8 for lung resection controls. Statistical analysis was performed using a non-parametric 2-tailed Mann-Whitney U test. Each data marker represents an individual patient. Horizontal lines represent mean values.




blood that showed similar degrees ofcytokine suppression in sepsis.25,30-32

Thus, the severely reduced cytokine pro-duction observed in spleen is consis-tent with a systemic abnormality. In asubset of sepsis patients, cytokine pro-duction at 22 hours was comparable withcontrols without sepsis, suggesting thatin some patients, defective cytokine se-cretion may be reversible if cells are re-moved from the sepsis milieu.

Of particular interest is the expres-sion of ligands for T-cell inhibitory re-ceptors on tissue parenchymal cells.Compared with controls, splenic capil-lary endothelial cells of sepsis patientshad increased expression of the inhibi-tory ligand PD-L1 (eFigure 5). Simi-

larly, lung cells from sepsis patients, bothwith and without pneumonia, demon-strated intense airway epithelial stain-ing for HVEM, a key regulator of host in-flammatory response in autoimmunityand infection (Figure 6).33 Signifi-cantly, the majority of lung T cells ex-pressed BTLA, one of several ligands forHVEM and another regulator of host in-flammation.33 Thus, the required ele-ments for activation of this immunosup-pressive pathway are present in sepsis.Collectively, these data demonstrate thatparenchymal cells of spleen and lung ex-press important immunoregulatory pro-teins and that HVEM appears to be spe-cifically induced in lungs of sepsispatients.

The PD-1 pathway has emerged as animportant mechanism that inhibits T-cell function. Expression of PD-L1 andPD-L2 by dendritic cells promotes atolerogenic phenotype, resulting in T-cell suppression.34-36 In lung from sep-sis patients, both PD-L1 and PD-L2 weredetected on resident dendritic cells andairway epithelial cells. Thus, these path-ways might be engaged on T cells traf-ficking through lung, resulting in local-ized inhibition of T cells and therebypredisposing to infection at this site. Inspleen, PD-1 expression was detected onCD4 and CD8 T cells, and both PD-L1and PD-L2 are known to be expressedon endothelial cells (eFigure 5) and im-pair CD8 T-cell cytolytic properties.37

Figure 5. Immune Effector Cells in Spleen Tissue

B Immunohistochemical staining for CD4A Immunohistochemical staining for HLA-DR

C Immunohistochemical staining for CD8

Control patient Patient with sepsis



D T-cell counts

500

400

300

200

100

0

No Sepsis SepsisCel

l Cou

nt p

er H

igh-

Pow

er F

ield

, No.

CD4 T cells

P<.001 P<.005

200

150

100

50

0

No Sepsis

CD8 T cells

Sepsis

Spleen from sepsis patients (n=22) or nonsepsis controls (n=12) was stained for HLA-DR, CD4, or CD8 and examined by an investigator (P.E.S.) blinded to sample identity(eAppendix). 3,3-diaminobenzidine 4-HCl was used as a chromogen to stain the cells of interest (brown), and a hematoxylin counterstain (blue) was used for backgroundstaining. A (HLA-DR immunostain; 200�), In a representative control sample, HLA-DR immunoreactivity is robust in all periarteriolar T- and B-cell zones, consistent withmajor histocompatibility complex (MHC) II pattern 2A (see reference 42 and eTable 7). There is marked loss of HLA-DR reactivity in B and T lymphocytes typical of nearlysubtotal depletion of HLA-DR–reactive elements in sepsis (MHC II pattern 4). Note that sinusoidal endothelium staining is pronounced, a change seen only in sepsis. B(CD4 immunostain; 400�), Periarteriolar CD4 cells are quantitatively decreased in sepsis (right panel) relative to control (left panel). C (CD8 immunostain; 400�), Peri-arteriolar CD8 cells are also quantitatively decreased in sepsis (right panel) relative to controls (left panel). D, The dot plots are cell counts for CD4 and CD8 T cells, obtainedby counting number of cells per field in periarteriolar lymphoid sheaths. Two fields were counted per slide and averaged. Statistical analysis was performed using a non-parametric 2-tailed Mann-Whitney U test. Each data marker represents an individual patient. Horizontal bars represent mean values.




A long-standing question in sepsis iswhether immunologic changes are or-gan specific.38 Although differences ex-isted, both spleen and lung shared com-mon immunosuppressive mechanisms.CD86 and HLA-DR expression were de-creased in cells isolated from bothspleen and lung (Figure 1). In con-trast, regulatory T cells were in-creased in spleen but not in lung. T cellsisolated from both organs of sepsis pa-tients had increased expression of themajor inhibitory receptor PD-1. Rela-tive to spleen, there were twice as manyPD-1–expressing cells in lung, suggest-ing the possibility of selective recruit-ment of PD-1–expressing cells to lung.In both organs, expression of ligandsfor major inhibitory receptors was de-tected on APCs and parenchymal cells.

Functionalunresponsiveness ofTcellsduring chronic viral infections has longbeen recognized and recently mechanis-tically described as a state of cell “ex-haustion.”39-41 Driven by persistent an-tigen exposure, exhaustion is typified byprogressive loss of cell function.39-41 Phe-notypically, exhaustion is character-ized by persistent expression of mul-tiple inhibitory receptors, includingPD-1, TIM-3, and LAG-3, along with de-creased expression of CD127 (IL-7R�)and CD62L. The functional and pheno-typic characteristics described for cell ex-haustion are similar to the present find-ings in sepsis, a state that is highly likelyto result in protracted antigen exposureand cell stimulation.39 As noted in thepresent study, splenic T cells demon-strated profound reductions in cyto-kine production as well as increasedexpression of PD-1and CD69 and de-creased CD127. Collectively, these find-ings suggest that T-cell exhaustion maybe an important immunosuppressivemechanism in sepsis.

The present study has a number ofimportant therapeutic implications.Most investigative agents in sepsis havebeen directed at blocking inflamma-tion and immune activation. Al-though such therapies may be success-ful if applied early, they may be harmfulif applied later in the immunosuppres-sive phase. As supportive therapies of

sepsis have improved, early deaths havedecreased and most patients enter amore protracted phase, with evidenceof impaired immunity made manifestby infections with relatively avirulentorganisms.9,14,15 An important part ofimplementing more targeted thera-pies will be to accurately determine theimmune status of individual patients

during their disease. By using a com-bination of functional assays and flowcytometry to characterize immune cellphenotypic changes, it may be pos-sible to identify patients with more se-vere immune compromise, for whichtargeted immune-enhancing thera-pies may be beneficial. A recent studythat treated sepsis patients with granu-

Figure 6. Expression of HVEM in Lung Tissue

B Immunohistochemical staining for HVEM (ligand for B- and T-lymphocyte attenuator)

C

Patient with sepsis Transplant donor

Patients without sepsis

Patient with lung cancer(distal to cancer)

A Control (isotype matched)

100

75

50

25

0

Epithelium Endothelium Macrophage

HV

EM

-Pos

itive

, %

No sepsis 100

75

50

25

0

Epithelium Endothelium Macrophage

Sepsisa

Sepsis or nonsepsis lung tissue was fixed in 4% paraformaldehyde and paraffin-embedded sections prepared(eAppendix). Lung sections were incubated with isotype-matched controls (A) or primary antibodies to anti–herpes virus entry mediator (HVEM) (B) followed by visualization of brown staining. Sepsis lung tissue stainedpositive for HVEM, the ligand for B- and T-lymphocyte attenuator. Lung resections, in particular of the sameairway, were photographed using the 200� and 400� (inset) objectives for each antibody stain. C, Slidespresented in (B) were evaluated in a blinded fashion and scored (in percentage) based on their positive stain-ing for HVEM. Data were graphed as percentage positive HVEM staining in lung tissue (epithelium, endothe-lium, and macrophages) of sepsis patients and nonsepsis controls. Data presented in (C) are as follows: n=16sepsis patients, n=7 transplant donor controls, and n=5 lung resection controls. Each data marker representsan individual patient. Horizontal bars represent mean values. 3,3-diaminobenzidine 4-HCl was used as a chro-mogen to stain the cells of interest (brown), and a hematoxylin counterstain (blue) was used for backgroundstaining.aComparison of HVEM-positive epithelium greater in sepsis patients vs nonsepsis controls: P=.01.




locyte-macrophage colony-stimulat-ing factor based on monocyte HLA-DRexpression suggests that such thera-pies might be successful.42 Our find-ings suggest additional targets. Inter-leukin 7 has an excellent clinical safetyrecord and has been used to enhanceimmunity in patients with cancer, hu-man immunodeficiency virus type 1,and hepatitis C. Interleukin 7 restoredimmunity in patients with persistent vi-ral infections and has improved viralclearance and survival in animal mod-els of chronic viral disease and sep-sis.21,43-45 The decreased IL-7R� expres-sion documented in the present studycould be causative in T-cell loss in sep-sis, and IL-7 administration might ame-liorate this problem Another potential

therapeutic approach is to interfere withinhibitory receptor signaling on T cells.We have demonstrated up-regulationof multiple inhibitory receptors andtheir ligands, including PD-1 and BTLA.Blockade of PD-1/PD-L1 interactions inchronic viral infection has been shownto reverse the functional unresponsive-ness of T cells and to enhance viralclearance.40 Blockade of PD-1 im-proves survival in animal models of fun-gal infection and sepsis.46-48

This study has several importantlimitations of the ability to generalizethe findings to sepsis at large. Theselimitations include the relatively smallsample size and the heterogeneous na-ture of both sepsis and control pa-tients. In this regard, the control groups

were markedly different from the sep-sis patients in many aspects; for ex-ample, nutritional status, degree of co-morbidities, and length of illness.Although the hypoalbuminemia notedin the sepsis patients could be due to anumber of causes such as increased vas-cular permeability, poor nutritional sta-tus was undoubtedly a major contrib-uting factor, and malnutrition hasnumerous effects on host immunity thatcould be responsible for some of the ob-served immunologic findings.49 Also,the study was confined to sepsis pa-tients who died in the ICU, some aftera considerable duration of sepsis. Thus,the process may be different for sepsispatients who die sooner of sepsis, whodo not die in the ICU, or who make agood recovery, as well as for a broadercase mix in general. Finally, it is pos-sible that findings in sepsis patientswere prodromal events and not reflec-tive of patients with sepsis in generalbut of patients with sepsis who had un-successful responses to supportive mea-sures. It should be emphasized thatdeaths within the first 72 hours of sep-sis in previously healthy patients withinfections of highly virulent organ-isms are associated with extremely el-evated proinflammatory cytokines, andthese deaths are likely secondary to anoverexuberant immunoinflammatoryresponse.7,8,14 Thus, the present studyserves as a bridge between preclinicaland early clinical findings and must beviewed cautiously.

In conclusion, these data provide aunique insight into the status of the im-mune system during sepsis, not only ina lymphoid organ but in peripheral tis-sue. Identification of potential receptor-ligand interactions and signaling path-ways leading to immunosuppressionmay allow for targeted therapeutic in-terventions to restore host immunity.

Author Affiliations: Departments of Medicine (DrsBoomer, Green, and Hotchkiss and Ms Bricker), Sur-gery (Drs To, Kreisel, Krupnick, and Hotchkiss and MrJarman), and Anesthesiology (Drs Chang, Takasu, andHotchkiss and Messrs Osborne and Walton), Wash-ington University School of Medicine, and Depart-ment of Surgery, St John’s Mercy Medical Center (DrSrivastava), St Louis, Missouri; and Department of Pa-thology, University of Washington School of Medi-cine, Seattle (Dr Swanson).

Figure 7. Expression of PD-L1 and PD-L2 in Lung Tissue

A Immunohistochemical staining for PD-L1

B Immunohistochemical staining for PD-L2

Patient with sepsis Transplant donor

Patients without sepsis

Patient with lung cancer(distal to cancer)

Sepsis or nonsepsis lung tissue was fixed in 4% paraformaldehyde and paraffin-embedded sections prepared(eAppendix). Lung sections were incubated with isotype-matched controls (see Figure 6) or primary antibod-ies to anti–programmed cell death ligand 1 (PD-L1) (A), or anti–PD-L2 (B), followed by visualization of brownstaining. Lung resections, in particular of the same airway, were photographed using the 200� and 400�(inset) objectives for each antibody stain. 3,3-diaminobenzidine 4-HCl was used as a chromogen to stain thecells of interest (brown), and a hematoxylin counterstain (blue) was used for background staining. Sepsis lungtissue stained positive for PD-L1 and PD-L2, the ligands for PD-1, in lung epithelium compared with trans-plant donor lung tissue. Lung resection (normal-appearing lung tissue distal to cancerous tissue) stained posi-tive in airway epithelium for PD-L2, like sepsis lung tissue, compared with transplant donor lung tissue.




Author Contributions: Dr Hotchkiss had full access toall of the data in the study and takes responsibility forthe integrity of the data and the accuracy of the dataanalysis.Study concept and design: Boomer, Walton, Green,Hotchkiss.Acquisition of data: Boomer, To, Chang, Takasu,Osborne, Walton, Bricker, Jarman, Kreisel, Krupnick,Srivastava, Green, Hotchkiss.Analysis and interpretation of data: Boomer, Chang,Walton, Swanson, Green, Hotchkiss.Drafting of the manuscript: Jarman, Green, Hotchkiss.Critical revision of the manuscript for important in-tellectual content: Boomer, To, Chang, Takasu,Osborne, Walton, Bricker, Kreisel, Krupnick, Srivastava,Swanson, Green, Hotchkiss.Statistical analysis: Boomer, Osborne, Walton, Green,Hotchkiss.Obtained funding: Hotchkiss.Administrative, technical, or material support: To,Chang, Takasu, Osborne, Bricker, Jarman, Kreisel,Krupnick, Srivastava, Green, Hotchkiss.Study supervision: Boomer, Swanson, Green,Hotchkiss.Conflict of Interest Disclosures: All authors have com-pleted and submitted the ICMJE Form for Disclosureof Potential Conflicts of Interest. Dr Hotchkiss re-ports receiving grant support from Pfizer, Bristol-Meyers Squibb, and Aurigene. No other disclosureswere reported.Funding/Support: This research was supported by Na-tional Institutes of Health grants GM44118, GM55194,and HL104985.Role of the Sponsor: The National Institutes of Healthhad no role in the design and conduct of the study;collection, management, analysis, and interpretationof the data; or preparation, review, or approval of themanuscript.Online-Only Material: The eAppendix, eTables 1through 8, and eFigures 1 through 5 are available athttp://www.jama.com.Additional Contributions: We acknowledge the nursesand staff of the surgical and medical intensive care unitsat Barnes Jewish Hospital and St John’s Hospitaland of the staff at Mid-America Transplant, St Louis,Missouri.

REFERENCES

1. Angus DC, Linde-Zwirble WT, Lidicker J, ClermontG, Carcillo J, Pinsky MR. Epidemiology of severe sep-sis in the United States: analysis of incidence, out-come, and associated costs of care. Crit Care Med.2001;29(7):1303-1310.2. Warren HS. Strategies for the treatment of sepsis.N Engl J Med. 1997;336(13):952-953.3. Munford RS, Pugin J. Normal responses to injuryprevent systemic inf lammation and can beimmunosuppressive. Am J Respir Crit Care Med. 2001;163(2):316-321.4. Oberholzer A, Oberholzer C, Moldawer LL. Sep-sis syndromes: understanding the role of innate andacquired immunity. Shock. 2001;16(2):83-96.5. Rittirsch D, Flierl MA, Ward PA. Harmful molecu-lar mechanisms in sepsis. Nat Rev Immunol. 2008;8(10):776-787.6. Wheeler AP, Bernard GR. Treating patients with se-vere sepsis. N Engl J Med. 1999;340(3):207-214.7. Angus DC. Management of sepsis: a 47-year-oldwoman with an indwelling intravenous catheter andsepsis. JAMA. 2011;305(14):1469-1477.8. Abraham E, Singer M. Mechanisms of sepsis-induced organ dysfunction. Crit Care Med. 2007;35(10):2408-2416.9. Kollef KE, Schramm GE, Wills AR, Reichley RM,Micek ST, Kollef MH. Predictors of 30-day mortalityand hospital costs in patients with ventilator-associated pneumonia attributed to potentially anti-

biotic-resistant gram-negative bacteria. Chest. 2008;134(2):281-287.10. Monneret G, Venet F, Pachot A, Lepape A. Moni-toring immune dysfunctions in the septic patient: a newskin for the old ceremony. Mol Med. 2008;14(1-2):64-78.11. Levy MM, Dellinger RP, Townsend SR, et al. TheSurviving Sepsis Campaign: results of an interna-tional guideline-based performance improvement pro-gram targeting severe sepsis. Intensive Care Med.2010;36(2):222-231.12. Luyt CE, Combes A, Deback C, et al. Herpes sim-plex virus lung infection in patients undergoing pro-longed mechanical ventilation. Am J Respir Crit CareMed. 2007;175(9):935-942.13. Limaye AP, Kirby KA, Rubenfeld GD, et al. Cy-tomegalovirus reactivation in critically ill immunocom-petent patients. JAMA. 2008;300(4):413-422.14. Hotchkiss RS, Karl IE. The pathophysiology andtreatment of sepsis. N Engl J Med. 2003;348(2):138-150.15. Schefold JC, Hasper D, Reinke P, Monneret G, VolkHD. Consider delayed immunosuppression into theconcept of sepsis. Crit Care Med. 2008;36(11):3118.16. Adib-Conquy M, Cavaillon JM. Compensatoryanti-inflammatory response syndrome. ThrombHaemost. 2009;101(1):36-47.17. Remick DG. Pathophysiology of sepsis. Am JPathol. 2007;170(5):1435-1444.18. Murphy TJ, Paterson HM, Mannick JA, LedererJA. Injury, sepsis, and the regulation of Toll-like re-ceptor responses. J Leukoc Biol. 2004;75(3):400-407.19. Ward NS, Casserly B, Ayala A. The compensa-tory anti-inflammatory response syndrome (CARS) incritically ill patients. Clin Chest Med. 2008;29(4):617-625, viii.20. Bone RC, Balk RA, Cerra FB, et al; American Col-lege of Chest Physicians/Society of Critical Care Medi-cine Consensus Conference Committee. Definitions forsepsis and organ failure and guidelines for the use ofinnovative therapies in sepsis. Chest. 1992;101(6):1644-1655.21. Unsinger J, McGlynn M, Kasten KR, et al. IL-7 pro-motes T cell viability, trafficking, and functionality andimproves survival in sepsis. J Immunol. 2010;184(7):3768-3779.22. Shi SR, Chaiwun B, Young L, Cote RJ, Taylor CR.Antigen retrieval technique utilizing citrate buffer orurea solution for immunohistochemical demonstra-tion of androgen receptor in formalin-fixed paraffinsections. J Histochem Cytochem. 1993;41(11):1599-1604.23. Hotchkiss RS, Swanson PE, Freeman BD, et al.Apoptotic cell death in patients with sepsis, shock, andmultiple organ dysfunction. Crit Care Med. 1999;27(7):1230-1251.24. Hotchkiss RS, Tinsley KW, Swanson PE, et al. Sep-sis-induced apoptosis causes progressive profounddepletion of B and CD4 T lymphocytes in humans.J Immunol. 2001;166(11):6952-6963.25. Rigato O, Salomao R. Impaired production of in-terferon-gamma and tumor necrosis factor-� but notof interleukin 10 in whole blood of patients with sepsis.Shock. 2003;19(2):113-116.26. Docke WD, Randow F, Syrbe U, et al. Monocytedeactivation in septic patients: restoration by IFN-�treatment. Nat Med. 1997;3(6):678-681.27. Venet F, Chung CS, Monneret G, et al. Regula-tory T cell populations in sepsis and trauma. J LeukocBiol. 2008;83(3):523-535.28. Hotchkiss RS, Opal S. Immunotherapy for sep-sis—a new approach against an ancient foe. N Engl JMed. 2010;363(1):87-89.29. Delano MJ, Scumpia PO, Weinstein JS, et al.MyD88-dependent expansion of an immature GR-1()CD11b() population induces T cell suppression andTh2 polarization in sepsis. J Exp Med. 2007;204(6):1463-1474.30. Munoz C, Carlet J, Fitting C, Misset B, Bleriot JP,

Cavaillon JM. Dysregulation of in vitro cytokine pro-duction by monocytes during sepsis. J Clin Invest. 1991;88(5):1747-1754.31. Ertel W, Kremer JP, Kenney J, et al. Downregu-lation of proinflammatory cytokine release in wholeblood from septic patients. Blood. 1995;85(5):1341-1347.32. Sinistro A, Almerighi C, Ciaprini C, et al. Down-regulation of CD40 ligand response in monocytes fromsepsis patients. Clin Vaccine Immunol. 2008;15(12):1851-1858.33. Shui JW, Steinberg MW, Kronenberg M. Regu-lation of inflammation, autoimmunity, and infectionimmunity by HVEM-BTLA signaling. J Leukoc Biol.2011;89(4):517-523.34. Shiao SL, McNiff JM, Pober JS. Memory T cellsand their costimulators in human allograft injury.J Immunol. 2005;175(8):4886-4896.35. Bedke T, Pretsch L, Karakhanova S, Enk AH,Mahnke K. Endothelial cells augment the suppres-sive function of CD4 CD25 Foxp3 regulatory T cells:involvement of programmed death-1 and IL-10.J Immunol. 2010;184(10):5562-5570.36. Pober JS, Sessa WC. Evolving functions of endo-thelial cells in inflammation. Nat Rev Immunol. 2007;7(10):803-815.37. Rodig N, Ryan T, Allen JA, et al. Endothelial ex-pression of PD-L1 and PD-L2 down-regulates CD8

T cell activation and cytolysis. Eur J Immunol. 2003;33(11):3117-3126.38. Cavaillon JM, Annane D. Compartmentalizationof the inflammatory response in sepsis and SIRS. J En-dotoxin Res. 2006;12(3):151-170.39. Yi JS, Cox MA, Zajac AJ. T-cell exhaustion: char-acteristics, causes and conversion. Immunology. 2010;129(4):474-481.40. Barber DL, Wherry EJ, Masopust D, et al. Restor-ing function in exhausted CD8 T cells during chronicviral infection. Nature. 2006;439(7077):682-687.41. Zajac AJ, Blattman JN, Murali-Krishna K, et al. Vi-ral immune evasion due to persistence of activated Tcells without effector function. J Exp Med. 1998;188(12):2205-2213.42. Meisel C, Schefold JC, Pschowski R, et al. Granu-locyte-macrophage colony-stimulating factor to re-verse sepsis-associated immunosuppression: a double-blind, randomized, placebo-controlled multicenter trial.Am J Respir Crit Care Med. 2009;180(7):640-648.43. Levy Y, Lacabaratz C, Weiss L, et al. Enhanced Tcell recovery in HIV-1-infected adults through IL-7treatment. J Clin Invest. 2009;119(4):997-1007.44. Patel A, Patel J, Ikwuagwu J. A case of progres-sive multifocal leukoencephalopathy and idiopathicCD4 lymphocytopenia. J Antimicrob Chemother.2010;65(12):2697-2698.45. Pellegrini M, Calzascia T, Toe JG, et al. IL-7 en-gages multiple mechanisms to overcome chronic vi-ral infection and limit organ pathology. Cell. 2011;144(4):601-613.46. Lazar-Molnar E, Gacser A, Freeman GJ, Almo SC,Nathenson SG, Nosanchuk JD. The PD-1/PD-Lcostimulatory pathway critically affects host resis-tance to the pathogenic fungus Histoplasmacapsulatum. Proc Natl Acad Sci U S A. 2008;105(7):2658-2663.47. Huang X, Venet F, Wang YL, et al. PD-1 expres-sion by macrophages plays a pathologic role in alter-ing microbial clearance and the innate inflammatoryresponse to sepsis. Proc Natl Acad Sci U S A. 2009;106(15):6303-6308.48. Brahmamdam P, Inoue S, Unsinger J, Chang KC,McDunn JE, Hotchkiss RS. Delayed administration ofanti-PD-1 antibody reverses immune dysfunction andimproves survival during sepsis. J Leukoc Biol. 2010;88(2):233-240.49. Hughes SM, Amadi B, Mwiya M, Nkamba H,Tomkins A, Goldblatt D. Dendritic cell anergy resultsfrom endotoxemia in severe malnutrition. J Immunol.2009;183(4):2818-2826.




Documents

Immunosuppression in Patients Who Die of Sepsis and Multiple …€¦ · Adaptive T cells (CD4 and CD8) Immune supressor cells Regulatory T cells, myeloid-derived suppressor cells