BR

Severe Traumatic Head Injury Affects SystemicCytokine Expression

Damien J LaPar, MD, MSC, Laura H Rosenberger, MD, MS, Dustin M Walters, MD, Traci L Hedrick, MD,rian R Swenson, MD, MS, Jeffrey S Young, MD, FACS, Lesly A Dossett, MD, Addison K May, MD, FACS,obert G Sawyer, MD, FACS

BACKGROUND: The neuroimmunologic effect of traumatic head injury remains ill-defined. This study aimed tocharacterize systemic cytokine profiles among traumatically injured patients to assess the effectof traumatic head injury on the systemic inflammatory response.

STUDY DESIGN: For 5 years, 1,022 patients were evaluated from a multi-institutional Trauma Immunomodu-latory Database. Patients were stratified by presence of severe head injury (SHI; head InjurySeverity Score �4, n � 335) vs nonsevere head injury (NHI; head Injury Severity Score �3,n � 687). Systemic cytokine expression was quantified by ELISA within 72 hours of admission.Patient factors, outcomes, and cytokine profiles were compared by univariate analyses.

RESULTS: SHI patients were more severely injured with higher mortality, despite similar ICU infectionand ventilator-associated pneumonia rates. Expression of early proinflammatory cytokines,interleukin-6 (p � 0.001) and tumor necrosis factor�� (p � 0.02), was higher among NHIpatients, and expression of immunomodulatory cytokines, interferon-� (p � 0.01) andinterleukin-12 (p � 0.003), was higher in SHI patients. High tumor necrosis factor�� levels inNHI patients were associated with mortality (p � 0.01), increased mechanical ventilation (p �0.02), and development of ventilator-associated pneumonia (p � 0.01). Alternatively, amongSHI patients, high interleukin-2 levels were associated with survival, decreased mechanicalventilation, and absence of ventilator-associated pneumonia.

CONCLUSIONS: The presence of severe traumatic head injury significantly alters systemic cytokine expressionand exerts an immunomodulatory effect. Early recognition of these profiles can allow fortargeted intervention to reduce patient morbidity and mortality. ( J Am Coll Surg 2012;214:

478–488. © 2012 by the American College of Surgeons)

aiet

Trauma-related injuries remain the fifth leading cause ofdeath in the United States, accounting for approximately125,000 deaths annually.1 Of traumatic deaths, motor ve-hicle accidents serve as the mechanism of injury in 23% ofall injury-related fatalities and totaled nearly 42,000 deathsin 2007.1 In addition, among those age 45 years or younger,trauma remains the leading cause of morbidity and mor-

Disclosure Information: Nothing to disclose.Presented at the Southern Surgical Association 123rd Annual Meeting, HotSprings, VA, December 2011.

Received December 15, 2011; Accepted December 15, 2011.From the Division of Trauma and Surgical Critical Care, Department ofSurgery, University of Virginia, Charlottesville, VA (LaPar, Rosenberger, Wal-ters, Hedrick, Swenson, Young, Sawyer) and Division of Trauma and SurgicalCritical Care, Department of Surgery, Vanderbilt University Medical Center,Nashville, TN (Dossett, May).Correspondence address: Damien J LaPar, MD, MSc, Department of Sur-

gery, University of Virginia, PO Box 800681, Charlottesville, VA 22903.email: djl8w@virginia.edu

478© 2012 by the American College of SurgeonsPublished by Elsevier Inc.

tality, and many trauma-related deaths are associated withtraumatic head injuries.1

Traumatic head injury (THI)–related deaths encompassa broad range of injuries and insults. The relationship be-tween death and THI has been associated with individualGlasgow Coma Scale (GCS) scores, as mild (GCS 13 to15), moderate (9 to 13), and severe (3 to 8) THI havesubstantial differences in neurologic recovery and disabil-ity.2 THI typically occurs in 2 stages, primary and second-ry brain injury, based on mechanism and timing. Primarynjury refers to the initial damage caused by the incitingvent, resulting in either focal or diffuse damage, and isypically diagnosed radiographically.3,4 Secondary brain in-

jury is the result of neuroinflammation, occurring days toweeks after the initial traumatic event. This secondary in-jury results from cerebral inflammation, blood�brain bar-rier dysfunction, and an upregulation of immunoinflam-matory processes, including cerebral and systemic cytokine

cascades.5,6

ISSN 1072-7515/12/$36.00doi:10.1016/j.jamcollsurg.2011.12.015

479Vol. 214, No. 4, April 2012 LaPar et al Traumatic Head Injury and Cytokine Expression

Cytokines are cell-signaling peptides secreted by sys-temic immune cells and CNS cells that act to provide in-tercellular communication. After traumatic injuries, bothpro- and anti-inflammatory cytokine cascades are acti-vated, including the interleukin (IL) family, tumor necrosisfactor (TNF), interferons, and various growth factors.7 Cy-tokine release is a necessary, important, and highly complexprocess that helps recruit neutrophils, B cells, and T cells,platelets, and other factors to areas of injury and damage.8,9

However, it is often an unrecognized imbalance in pro- andanti-inflammatory cytokine expression that ultimately re-sults in the initiation of an unrecoverable physiologic state.Understanding systemic cytokine expression profiles forvarious patient populations holds tremendous clinical po-tential in aiding clinicians with early goal-directed therapy.Although select patient series have highlighted systemiccytokine profiles in critically ill and injured patients,10 thespecific impact of severe THI on the human neuroinflam-matory response and resultant cytokine expression profilesremains ill-defined.

The purpose of this study was to characterize systemiccytokine profiles among traumatically injured patients toassess the effect of traumatic head injury on the systemicinflammatory response. These data were then used to cor-relate systemic cytokine expression with patient morbidityand mortality to identify clinically significant associationsto assist with early goal-directed therapy.

METHODSStudy designThis study was approved by the Institutional Review Boardat both the University of Virginia, including a waiver forinformed patient consent, and Vanderbilt University,which required an assent from a surrogate before datacollection and informed patient consent for all cases ofresolved critical illness. A retrospective review of themulti-institutional Trauma Immunomodulatory Database(TIMD) was performed to extract patient records for those

Abbreviations and Acronyms

GCS � Glasgow Coma ScaleIL � interleukinISS � Injury Severity ScoreNHI � nonsevere head injurySHI � severe head injuryTHI � traumatic head injuryTIMD � Trauma Immunomodulatory DatabaseTNF � tumor necrosis factorVAP � ventilator-associated pneumonia

sustaining traumatic injuries for the study period October

2001 to May 2006. TIMD is a prospectively collected da-tabase developed collaboratively between the Departmentsof Surgery at the University of Virginia Medical Center andVanderbilt University Medical Center.

Patients and subject enrollmentPatient data for those 18 years of age and older of both sexesand all races who were admitted to either surgical or traumaICU were entered into TIMD. Patients with traumaticinjuries met the principal inclusion criteria for this study.Trauma patients were stratified according to presence orabsence of a severe traumatic head injury. Severe head in-jury (SHI) was defined as any head injury eliciting a headInjury Severity Score (ISS) �4. Patient characteristics, in-cluding demographics, traumatic injury–related details,and comorbid disease, systemic cytokine profiles, and out-comes were analyzed to determine the influence of severetraumatic head injuries.

Data collection and variable definitionsTIMD data collection was performed by trained, full-timeresearch personnel consisting of clinically trained nurses,nurse practitioners, and physicians assistants. Collecteddata were gathered through review of patient medicalrecords as well as personal interviews of patients, fami-lies, and patient health care providers. Baseline patientdemographics, comorbid disease states, hospital andICU discharge data, and in-hospital complication anddeath-related details were entered into the secure,password-protected TIMD computer database. Data col-lected were used to calculate the ISS11 and GCS score at thetime of admission or initial presentation. In-hospital mor-tality reflects all patient deaths from any cause before hos-pital discharge.

Patient infections, excluding catheter-related infections,were defined according to standard US Centers for DiseaseControl and Prevention definitions and guidelines.12 Ac-cordingly, pneumonia was diagnosed with the presence ofsystemic evidence of infection, production of sputum, iso-lation of a predominant organism, development of a newor changing infiltrate or effusion on chest roentography, orgrowth of �100,000 colony-forming units on quantitativeculture through endotracheal aspiration or �10,000colony-forming units through bronchoalveolar lavage.Bacteremia or bloodstream infections were diagnosed withisolation of organisms for any single blood culture usingaspectic sample collection techniques, except for coagulase-negative Staphylococcus epidermidis, which required po-sitive growth from 2 separate positive blood cultures.Urinary tract infections occurred in the presence of either�100,000 organisms/mL urine or �10,000 organisms/mL

with symptoms. Catheter-related infections included those

uikgatt

s(mp(phts

0iaeid

480 LaPar et al Traumatic Head Injury and Cytokine Expression J Am Coll Surg

with �15 colonies with the semi-quantitative roll platetechnique in the setting of clinical infection with possiblebut not necessary positive blood culture with the sameorganism. Catheter-tip cultures were obtained after re-moval from patients with either persistently elevated orrising white blood cell counts or in those with body tem-perature �38.5°C. Skin, soft-tissue, wound, and perito-neal infections occurred in those with culture-positive re-sults when available or were diagnosed clinically.

Cytokine quantification and analysisSystemic cytokine levels were obtained from patient serumsamples collected within 72 hours of inpatient admission,with nearly half collected within the first 24 hours of ad-mission. For each patient, a 10-mL blood sample was col-lected through peripheral venopuncture or access of a cen-tral venous line. Collected blood samples were immediatelycentrifuged and the plasma was separated and stored at�70°C before analysis. Subsequent ELISA techniqueswere used to determine systemic cytokine concentrations.The ELISA-based Luminex100 System (Miraibio, Inc) was

sed to analyze all patient samples. Cytokine assessmentncluded serum concentrations of proinflammatory cyto-ines (IL-1, IL-2, IL-6, IL-12, interferon-�, TNF-�,ranulocyte-macrophage colony-stimulating factor) andnti-inflammatory cytokines (IL-4 and IL-10), as well ashe chemokines IL-8 and IL-5. Minimum cytokine de-ectibility was defined as 2.7 pg/mL.

Statistical analysisAll statistical methodology was designed to test the nullhypothesis that outcomes and serum cytokine concentra-tions after traumatic injuries are not significantly different,despite the presence of severe traumatic head injury. Stan-dard statistical significance was set to � � 0.05. All studyoutcomes and data comparisons were established a prioribefore data collection. Missing data for all variables of in-terest underwent sequential case-wise deletion to obtain acomplete dataset for subsequent analysis.

Descriptive statistics for all variable comparisons werecalculated using appropriate univariate hypothesis tests.Categorical variables are expressed as within-group per-centages and were compared for independent samples us-ing either Pearson’s chi-square or Fisher’s exact test. Con-tinuous variables are expressed as either mean � SD ormedian (interquartile range), depending on overall variabledistribution. Independent-sample single-factor ANOVAwas used for parametric data comparisons, and Mann-Whitney U-test was used for all nonparametric data com-parisons. Calculated test statistics were used to derive allreported 2-tailed p values. Predictive Analytics Software

with complex sampling module software, version 18.0.0 A

(IBM Corporation) was used for all data manipulation andstatistical analyses.

RESULTSPatient characteristicsA total of 1,022 patients were included in this study. Allpatients had complete cytokine data. On patient stratifica-tion, 335 (32.8%) trauma patients had the presence of anSHI, and 687 (67.2%) had a nonsevere head injury (NHI).Patient demographics, characteristics, trauma profiles, andcritical care interventions are detailed in Table 1. SHI pa-tients were younger and more commonly male. No signif-icant differences were noted between study groups withrespect to race and ethnicity. For patient comorbid disease,NHI patients more commonly presented with underlyingpulmonary disease (p � 0.01) and autoimmune disease(p � 0.01), and incremental clinical differences existedbetween groups for other comorbid disease states. Overall,SHI patients were more severely injured with higher me-dian ISS scores (34 vs 25; p � 0.001) and, as expected, hadignificantly higher head and neck ISS component scoresp � 0.001), as well as lower median GCS scores on ad-ission (6 vs 9; p � 0.001). Alternatively, NHI patients

resented with a higher prevalence of penetrating trauma14.2% vs 6.3%; p � 0.001). As a result, nearly all SHIatients were treated with mechanical ventilation (97.9%);owever, duration of ventilation, incidence of tracheos-omy placement, and provision of blood-product transfu-ions were higher for NHI patients.

Patient outcomes associated with presence of SHIAmong patient outcomes (Table 2), hospital mortality wassignificantly higher (13.4% vs 9.3%; p � 0.04) for SHIpatients. However, no significant differences were observedbetween study groups with respect to composite ICU in-fection rates, median number of ICU infections, incidenceof ventilator-associated pneumonia (VAP), or median ICUor total hospital lengths of stay.

Systemic cytokine profiles and cytokine differencesbetween survivors and decedentsTable 3 displays baseline cytokine profiles for trauma pa-tients with and without SHI. Expression of early proin-flammatory cytokines, IL-6 (p � 0.001) and TNF-� (p �.02), was higher among NHI patients; and expression ofmmunomodulatory cytokines, interferon-� (p � 0.01)nd IL-12 (p � 0.003), was higher in SHI patients. Toxamine the impact of cytokine profiles on patient mortal-ty, univariate analyses were performed for the outcomes ofeath among both SHI and NHI patients (Table 4).

mong both SHI and NHI patients, high systemic IL-4

w

C

481Vol. 214, No. 4, April 2012 LaPar et al Traumatic Head Injury and Cytokine Expression

levels were observed for decedents. In addition, higher ex-pression of IL-6, IL-8, and TNF-� levels was associated

ith mortality for NHI patients.

Cytokine profiles associated with duration ofmechanical ventilation and VAPAdditional analyses were performed to identify suspecteddifferences in systemic cytokine expression profiles as afunction of duration of mechanical ventilation and the on-set of VAP (Tables 5 and 6, respectively). These analyses

Table 1. Trauma Population Characteristics for Patients wi

VariableSevere head

(n � 335

Patient age, y, mean � SD 35.0 � 29Sex, female, % 19.4Race/ethnicity, %

White 84.2Black 10.4Hispanic 3.9Native American 0.3Other 0.3

omorbidities, %Coronary artery disease 10.4Cerebrovascular disease 3.0Chronic renal disease 0.9Corticosteroids 0.6Liver disease 2.7Malignancy 3.3Peripheral vascular disease 1.2Pulmonary disease 6.0Autoimmune disorders 0.0Diabetes mellitus 6.0

InterventionsMechanical ventilation, % 97.9Total ventilator days, mean � SD 7.0 � 6.Tracheostomy, % 12.7Transfusions, % 62.7

Penetrating trauma, % 6.3Scoring, mean � SD

ISS 34.0 � 13Abdomen 0.0 � 2.Chest 3.0 � 4.Extremities 2.0 � 3.Face 0.0 � 2.Head or neck 5.0 � 1.External 0.0 � 1.GCS (Admit) 6.0 � 5.

GCS, Glasgow Coma Scale; ISS, Injury Severity Score.

demonstrated that significant differences in select cyto-

kines existed among both SHI and NHI patients. AmongNHI patients, overall IL-8 and TNF-� levels varied acrosscategories of mechanical ventilation duration (�3 daysvs 4 to 6 days vs �7 days). In addition, high TNF-�levels in NHI patients were associated with increasedmechanical ventilation (p � 0.02) and development ofVAP (p � 0.01). Alternatively, among SHI patients,high IL-2 levels were associated with survival, decreasedmechanical ventilation, and absence of VAP. In addition,higher IL-8 and IL-10 levels were noted in SHI patientsin whom VAP developed compared with those without

d without Severe Head Injuriesy No severe head injury

(n � 687) p Value

43.0 � 30.0 �0.00130.7 �0.001

0.5681.413.44.50.10.3

13.6 0.192.8 0.840.9 �0.991.3 0.523.4 0.701.5 0.062.0 0.45

11.4 0.011.9 0.017.9 0.31

90.4 �0.0017.5 � 8.0 0.01

8.9 0.0777.1 �0.00114.2 �0.001

25.0 � 17.0 �0.0012.0 � 3.0 �0.0013.0 � 2.0 �0.0012.0 � 3.0 �0.0010.0 � 1.0 �0.0010.0 � 2.0 �0.0010.0 � 1.0 0.839.0 � 5.0 �0.001

th aninjur)

.0

0

.00000000

VAP development.

hinsrtitipu3

t

INBVUCWACHI

IIII

Go

482 LaPar et al Traumatic Head Injury and Cytokine Expression J Am Coll Surg

DISCUSSIONIn this study, we presented novel data related to patientoutcomes and systemic cytokine profiles among traumati-cally injured patients to comment on the neuroimmu-nologic effect of traumatic head injuries on the systemicinflammatory response. Using data collected from a multi-institutional cohort of patients, these results corroborateearlier reports of the negative effect that SHIs have ontrauma patient outcomes and extend the discussion ofthese outcomes to demonstrate significant associationswith varied cytokine responses. Specifically, these data sug-gest that significant differences exist in cytokine levelsamong both SHI and NHI patients, and that among bothpro- and anti-inflammatory cytokines, IL-4, IL-6, andTNF-� levels are associated with hospital mortality. Addi-tional differences were noted for IL-2 and TNF-� levels

Table 2. Outcomes for Trauma Population with and withou

VariableSevere hea

(n � 33

Hospital mortality, % 13.4CU Infection, % 50.1o. of ICU infections, mean � SD 1.0 �

lood stream infection, % 24.2entilator-associated pneumonia, % 29.6rinary tract infection, % 15.2atheter-related infection, % 13.1ound infection, % 4.0

bdominal infection, % 2.0NS/CSF infection, % —ospital length of stay, d, mean � SD 18.0 �

CU Length of stay, d, mean � SD 9.0 �

CSF, cerebrospinal fluid.

Table 3. Cytokine Profile for Trauma Population with andwithout Severe Head Injuries

VariableSevere head injury*

(n � 335)No severe

head injury* p Value

IL-1 2.7 � 4.8 2.7 � 2.5 0.02L-2 3.9 � 11.6 2.8 � 9.9 0.11L-4 37.7 � 195.5 38.0 � 214.9 0.90L-5 2.7 � 1.6 2.7 � 3.1 0.35L-6 126.2 � 236.7 171.9 � 359.7 �0.001

IL-8 23.9 � 48.5 35.0 � 55.3 �0.001IL-10 54.3 � 111.0 48.2 � 94.4 0.09IL-12 3.4 � 10.2 2.7 � 5.3 0.003IFN 4.5 � 13.1 2.9 � 7.4 0.01GM-CSF 5.1 � 9.0 3.8 � 6.3 0.09TNF-� 4.5 � 8.5 5.4 � 7.6 0.02

*Values are mean � SD.

rM-CSF, granulocyte-macrophage colony-stimulating factor; IFN, interfer-

n; IL, interleukin; TNF, tumor necrosis factor.

with respect to duration of mechanical ventilation and de-velopment of VAP. Overall, these results provide usefulclinical insight to be incorporated into existing clinical al-gorithms for the treatment of trauma patients with andwithout SHI.

Patient outcomes in the present study are in agreementwith other trauma series. For patient mortality, the 13.4%mortality observed for SHI patients in this series favorscomparatively with other series with documented mortal-ity rates of 30% for severe traumatic brain injury.13 Themortality rate observed in this series reflects the exclusionof patients not surviving to 48 hours. The most commoninpatient complication after severe THI is infection, owingto invasive monitoring, prolonged ICU length of stay, me-chanical ventilation, and post-traumatic immunosuppres-sion.14 Rates of systemic infection after isolated SHI inreported series is as high as 50% to 65%,15-17 which is

igher than the 48% composite incidence rate of infectionsn the current series. Although rates of VAP were not sig-ificantly different between SHI and NHI patients in thiseries, several other studies have found significantly higherates of respiratory infections compared with other ana-omic locations after SHI. Consistent with previous find-ngs of increased infections with increased severity of headrauma, Hranjec and colleagues revealed 74% of infectionsn their THI patients were respiratory infections amongatients with mean GCS 6.8,10 and other series have doc-mented respiratory infections comprising approximately0% to 45% of all infectious complications.15-17

Historically, the brain was considered an “immunologi-cally privileged organ,” lacking the ability to elicit an in-flammatory immune response.3,4,18 This was attributed tohe presence of the blood–brain barrier; however, it is now

ere Head Injuriesry No severe head injury

(n � 687) p Value

9.3 0.0446.7 0.32

0.0 � 1.0 0.6035.4 0.0625.9 0.2313.5 0.7210.7 0.568.4 0.225.1 0.34— NA

17.0 � 15.0 0.629.0 � 9.0 0.34

t Sevd inju5)

1.0

15.08.0

ecognized that endothelial permeability after THI results

miuaarttnfccm

tsacsToaarnbci

IIIIIIIIGT

G n; IL,

483Vol. 214, No. 4, April 2012 LaPar et al Traumatic Head Injury and Cytokine Expression

in transmigration of systemic immune cells, including neu-trophils, monocytes, and lymphocytes.5 Concurrent with

igrant immune cells into the CNS, resident cells, includ-ng astrocytes, microglia, and neurons, are activated andpregulated and subsequently secrete inflammatory medi-tors, such as cytokines, chemokines, cytotoxic proteases,nd oxygen radicals.3,5,6,18 This neuroinflammatory envi-onment results in secondary brain injury from cell death,issue damage, and neurodegeneration. In addition, despitehe clear neurotoxic character of certain cytokines, recog-ition has recently been given to select neuroprotectiveunctions.7,19 The dual role of cytokines highlights theomplexity and multifaceted interactions that exist amongytokine cascades, and indicates a potential for neuro-odulatory therapies in the future.

Table 4. Cytokine Profiles for Trauma Patients with and wi

Cytokine

Severe head injury (n � 335)Survivor*(n � 290)

Mortality*(n � 45) p

IL-1 2.7 � 4.4 2.7 � 5.1L-2 4.2 � 11.4 3.1 � 13.2L-4 30.3 � 152.8 108 � 439.2L-5 2.7 � 2.8 2.7 � 6.9L-6 116.1 � 227.4 162.5 � 307.5L-8 22.3 � 45.5 28.7 � 56.8L-10 48.0 � 91.5 50.8 � 139.4L-12 3.4 � 10.1 3.2 � 12.6FN 4.5 � 11.5 4.1 � 15.1M-CSF 5.1 � 8.8 4.9 � 9.2NF-� 4.5 � 6.0 4.8 � 9.0

*Values are mean � SD.M-CSF, granulocyte-macrophage colony-stimulating factor; IFN, interfero

Table 5. Cytokine Profiles for Trauma Patients with and wi

Cytokines

Severe head injury (n � 335)<3 Days*(n � 35)

4 to 6 Days*(n � 93)

> Days*(n � 133)

IL-1 2.7 � 3.1 2.7 � 11.3 3.0 � 12.4IL-2 11.0 � 16.9† 6.3 � 15.3 5.6 � 11.4IL-4 36.0 � 235.8 44.9 � 161.7 39.0 � 241.4IL-5 2.7 � 2.7 2.7 � 8.5 2.7 � 3.2IL-6 111.4 � 185.8 126.9 � 219.8 144.1 � 251.1IL-8 16.9 � 32.2 19.8 � 40.1 19.1 � 44.2IL-10 51.9 � 82.0 52.6 � 90.7 46.3 � 102.6IL-12 2.7 � 15.6 5.6 � 14.0 5.3 � 12.0IFN 6.4 � 11.9 7.6 � 16.8 6.0 � 15.8GM-CSF 7.3 � 9.8 5.4 � 12.6 6.1 � 9.1TNF-� 4.6 � 4.8 5.0 � 6.6 4.7 � 7.3

*Values are mean � SD.†Pairwise comparison: <3 days vs >7 days.‡

Pairwise comparison: 4 to 6 days vs >7 days.GM-CSF, granulocyte-macrophage colony-stimulating factor; IFN, interferon; IL,

The most significant results of this study, therefore, arehe unique differences noted in systemic cytokine expres-ion profiles, which corroborate the results of select humannd animal studies. As demonstrated here, alterations inytokine expression afterTHI exist and have been noted foreveral cytokines, including IL-1, IL-6, IL-8, IL-10, IL-12,NF-�, and transforming growth factor-�.7,13,20-25 IL-1 isne of the most studied cytokines and classically describeds a proinflammatory cytokine. In multiple rodent models,n IL-1 receptor antagonist has been shown to reduce neu-odegeneration and contusion volume and prevent cog-itive deficits when administered in either an intracere-roventricular or subcutaneous route.26,27 Similar to theurrent study, IL-6 is found in elevated levels after THI ands a significant mediator of fever, tachycardia, and the

Severe Head Injuries Stratified by In-Hospital MortalityNo severe head injury (n � 687)

eSurvivor*(n � 623)

Mortality*(n � 64) p Value

2.7 � 2.5 2.7 � 3.6 0.552.8 � 10.2 2.7 � 4.7 0.50

31.2 � 190.4 112.7 � 393.3 0.0012.7 � 1.6 2.7 � 1.7 0.71

162.0 � 335.8 247.4 � 827.6 0.00333.9 � 53.2 55.9 � 126.9 0.00353.1 � 110.3 66.9 � 112.2 0.212.7 � 5.4 2.7 � 5.2 0.703.0 � 7.5 2.7 � 5.9 0.843.6 � 6.3 4.9 � 11.2 0.205.2 � 7.2 7.5 � 10.1 0.01

interleukin; TNF, tumor necrosis factor.

Severe Head Injuries Stratified by Total Ventilator DaysNo severe head injury (n � 687)

ue<3 Days*(n � 74)

4 to 6 Days*(n � 127)

> Days*(n � 251) p Value

2.7 � 3.9 2.7 � 5.8 2.7 � 15.4 0.367.1 � 19.1 4.5 � 11.3 4.3 � 15.4 0.22

12.1 � 104.4 43.3 � 188.6 37.5 � 191.2 0.742.7 � 2.3 2.7 � 3.9 2.7 � 3.9 0.80

135.7 � 357.4 180.5 � 283.0 215.6 � 517.6 0.1924.7 � 39.7 27.3 � 42.4‡ 36.3 � 62.2 0.0454.4 � 119.8 44.2 � 92.3‡ 67.4 � 114.6 0.063.2 � 7.6 2.7 � 7.7 3.7 � 7.7 0.555.4 � 9.7 4.7 � 11.5 5.4 � 13.4 0.626.5 � 8.0 5.1 � 9.8 5.9 � 8.2 0.865.3 � 5.1* 6.8 � 6.4 7.2 � 10.1 0.02

thout

Valu

0.980.830.0010.330.070.110.690.850.850.840.38

thout

p Val

0.630.120.200.200.700.160.820.760.490.950.67

interleukin; TNF, tumor necrosis factor.

ssro

t

tr

IIIIIIIIGT

G feronp

484 LaPar et al Traumatic Head Injury and Cytokine Expression J Am Coll Surg

acute-phase response.13 In one noted study, although con-iderably lower than cerebrospinal fluid levels after THI,erum levels of IL-6 peaked 2 days after non-head injury–elated trauma and a considerable acute-phase response wasbserved.13 IL-6 has demonstrated neuroprotective effects,

as it has been shown to stimulate nerve growth factor andaid in survival and differentiation of neurons.28-30 In addi-ion, TNF-� is a potent proinflammatory cytokine and

functions much like IL-1, activating a multitude of addi-tional cytokine and chemokine cascades.7 Consistent withour results, both animal and human studies reveal thatTNF-� is significantly upregulated after THI and re-leased by numerous resident cells in the CNS.21,31,32 Al-ernatively, IL-2 and interferon-� have been found to beeduced after traumatic injury,20 supporting our obser-

vations that higher IL-2 levels were associated with re-duced mechanical ventilation duration and the absenceof VAP.

This study has important clinical relevance. The re-ported results not only provide novel data related to theoverall inflammatory response after THI, but also provideimportant correlations between various cytokine profilesand patient morbidity and mortality. As these data andothers have demonstrated, severe traumatic head injuryresults in a varied neuroimmunologic response comparedwith traumatic injuries without SHI. In addition, to ourknowledge, this study represents the largest series to exam-ine the relationship between admission serum cytokine lev-els and outcomes among THI patients. The identificationof certain cytokine expression profiles, such as those forIL-2, IL-4, IL-6, and TNF-� among both SHI and NHIpatients, can provide clinicians with an objective measure

Table 6. Cytokine Profiles for Trauma Patients with and wiAssociated Pneumonia

Cytokine

Severe head injury (n � 335)VAP*

(n � 99)No VAP*

(n � 236) p

IL-1 2.7 � 1.8 2.7 � 5.9L-2 3.1 � 11.9 4.4 � 11.6L-4 90.5 � 311.8 29.6 � 141.5L-5 2.7 � 2.6 2.7 � 4.0L-6 149.0 � 263.5 111.8 � 222.9L-8 33.6 � 55.5 20.4 � 43.2L-10 60.5 � 121.7 40.0 � 85.4L-12 3.9 � 10.6 2.7 � 8.7FN 2.7 � 6.5 5.4 � 14.2M-CSF 5.3 � 9.1 5.0 � 8.2NF-� 4.8 � 6.2 4.5 � 6.2

*Values are mean � SD.M-CSF, granulocyte-macrophage colony-stimulating factor; IFN, interneumonia.

of subacute systemic inflammation for which to provide

earlier directed therapy. Our results remain hypothesis-generating and provide a legitimate clinical context fromwhich future, prospective studies can be derived to helpdirect targeted therapeutic interventions.

Certain aspects of this study deserve additional discus-sion. Various clinical scoring systems (eg, ISS score andGCS score) or objective clinical parameters (eg, intracra-nial pressure measurements) are frequently used to identifypatients with varying degrees of head injury. Accordingly,in the present series, the SHI group, defined by ISS score,identified patients with SHI with a median GCS score of 6,and distinguished this population from those trauma pa-tients suffering either mild to moderate head injury, theNHI cohort, which had a median GCS score of 9. Unfor-tunately, although of great interest, intracranial pressuremeasurements to correlate within these populations wereavailable in a minority of patients even with SHI and werenot collected routinely. With respect to the observed cyto-kine profiles, these results reflect differences in cytokinelevels drawn during a period of 72 hours from the time ofadmission. Although these results might not provide acompletely accurate representation of the initial cytokineresponse after traumatic injuries, nearly half of all cytokinelevels included in this analysis were collected on the firstday of admission. In addition, little variation existed inexpressed cytokine levels during a 72-hour period (data notshown).The analysis of cytokine levels expressed within thecerebrospinal fluid were not collected, but serve as an areaof future investigation. In addition, these data to do notexclude the effects of prehospital interventions before cy-tokine collection, which might have impacted reportedprofiles. Finally, the effect of penetrating head injuries was

t Severe Head Injuries Stratified by Incidence of Ventilator-

No severe head injury (n � 687)

eVAP*

(n � 178)No VAP*

(n � 509) p Value

2.7 � 3.5 2.7 � 2.4 0.733.3 � 15.2 2.7 � 8.9 0.33

39.2 � 202.9 34.5 � 226.6 0.782.7 � 2.6 2.7 � 1.5 0.99

199.4 � 429.5 166.4 � 340.1 0.2239.1 � 59.0 33.8 � 53.2 0.1460.3 � 117.1 52.9 � 108.7 0.152.7 � 5.0 2.7 � 5.4 0.613.8 � 8.6 2.8 � 7.1 0.343.4 � 6.0 4.0 � 6.6 0.626.4 � 9.8 5.0 � 6.9 0.01

; IL, interleukin; TNF, tumor necrosis factor; VAP, ventilator-associated

thou

Valu

0.060.300.160.350.110.010.020.280.060.850.57

not specifically addressed; however, the relative contribu-

485Vol. 214, No. 4, April 2012 LaPar et al Traumatic Head Injury and Cytokine Expression

tion of severe penetrating head injuries to the SHI groupwas very small, as many of these patients were exclu-ded from TIMD because of death within 24 hours ofpresentation.

This study has several identifiable limitations. The influ-ence of selection bias should be considered as a potentialfactor in any retrospective study design. The issue of con-founding in the reported results must also be consideredbecause of the inclusion of a relatively heterogeneoustrauma patient population within 2 separate institutions.Reported results represent short-term outcomes, and cyto-kine profiles do not represent variations that can occurbeyond 72 hours (ie, 5 to 7 days). Longer follow-up mighthave helped to define additional small differences betweenstudy groups. Additional mortality details, such as cause ofdeath or deaths following multiorgan failure/dysfunctionor brain death, as well as the ability to correlate cytokineresponse and outcomes to more detailed blood-producttransfusion profiles can provide an additional context fromwhich to scrutinize the reported results.

CONCLUSIONSBased on these results, the null hypothesis is rejected. Thepresence of severe traumatic head injury significantly alterssystemic cytokine expression and exerts an immunomodu-latory effect. Systemic cytokine expression profiles amongtraumatically injured patients vary depending on nature ofinjury. These results suggest that the early recognition ofcharacteristic cytokine profiles might allow for targeted in-tervention to reduce patient morbidity and mortality.

Author ContributionsStudy conception and design: LaPar, Rosenberger, Walters,

Hedrick, Young, SawyerAcquisition of data: LaPar, Swenson, Dossett, May, SawyerAnalysis and interpretation of data: LaPar, Young, SawyerDrafting of manuscript: LaPar, Rosenberger, Walters,

SawyerCritical revision: LaPar, Rosenberger, Walters, Hedrick,

Sawyer

Acknowledgment: We would like to thank the patients andfamilies who contributed to this study, as well as the followinghealth care practitioners: Sera L Downing, Michelle Pom-phrey, Stewart M Lowson, MD, Judy L Jenkins, and Donna LHollis.

REFERENCES

1. Centers for Disease Control and Prevention. Deaths: final

data for 2007. National Vital Statistics Report 2010;58(19).

Available at: http://www.cdc.gov/NCHS/data/nvsr/nvsr58/nvsr58_19.pdf. Accessed November 11, 2011.

2. Ghajar J. Traumatic brain injury. Lancet 2000;356:923–929.3. Morganti-Kossmann MC, Rancan M, Otto VI, et al. Role of

cerebral inflammation after traumatic brain injury: a revisitedconcept. Shock 2001;16:165–177.

4. Morganti-Kossmann MC, Satgunaseelan L, Bye N, KossmannT. Modulation of immune response by head injury. Injury 2007;38:1392–1400.

5. Kadhim HJ, Duchateau J, Sebire G. Cytokines and brain injury:invited review. J Intensive Care Med 2008;23:236–249.

6. Ziebell JM, Morganti-Kossmann MC. Involvement of pro- andanti-inflammatory cytokines and chemokines in the pathophys-iology of traumatic brain injury. Neurotherapeutics 2010;7:22–30.

7. Lenzlinger PM, Morganti-Kossmann MC, Laurer HL, McIn-tosh TK. The duality of the inflammatory response to traumaticbrain injury. Mol Neurobiol 2001;24:169–181.

8. Bone RC. The pathogenesis of sepsis. Ann Intern Med 1991;115:457–469.

9. Bone RC, Grodzin CJ, Balk RA. Sepsis: a new hypothesis forpathogenesis of the disease process. Chest 1997;112:235–243.

10. Hranjec T, Swenson BR, Dossett LA, et al. Diagnosis-dependentrelationships between cytokine levels and survival in patientsadmitted for surgical critical care. J Am Coll Surg 2010;210:833–844, 845�836.

11. Baker SP, O’Neill B, Haddon W Jr, Long WB. The injury sever-ity score: a method for describing patients with multiple injuriesand evaluating emergency care. J Trauma 1974;14:187–196.

12. Delgado-Rodriguez M, Gomez-Ortega A, Llorca J, et al. Noso-comial infection, indices of intrinsic infection risk, and in-hospital mortality in general surgery. J Hosp Infect 1999;41:203–211.

13. Kossmann T, Hans VH, Imhof HG, et al. Intrathecal and seruminterleukin-6 and the acute-phase response in patients with se-vere traumatic brain injuries. Shock 1995;4:311–317.

14. Helling TS, Evans LL, Fowler DL, et al. Infectious complica-tions in patients with severe head injury. J Trauma 1988;28:1575–1577.

15. Miller CH, Quattrocchi KB, Frank EH, et al. Humoral andcellular immunity following severe head injury: review and cur-rent investigations. Neurol Res 1991;13:117–124.

16. Quattrocchi KB, Frank EH, Miller CH, et al. Impairment ofhelper T-cell function and lymphokine-activated killer cytotox-icity following severe head injury. J Neurosurg 1991;75:766–773.

17. Quattrocchi KB, Frank EH, Miller CH, et al. Severe head in-jury: effect upon cellular immune function. Neurol Res 1991;13:13–20.

18. Helmy A, De Simoni MG, Guilfoyle MR, et al. Cytokines andinnate inflammation in the pathogenesis of human traumaticbrain injury. Prog Neurobiol 2011;95:352–372.

19. Correale J, Villa A. The neuroprotective role of inflammation innervous system injuries. J Neurol 2004;251:1304–1316.

20. Boddie DE, Currie DG, Eremin O, Heys SD. Immune suppres-sion and isolated severe head injury: a significant clinical prob-lem. Br J Neurosurg 2003;17:405–417.

21. Csuka E, Morganti-Kossmann MC, Lenzlinger PM, et al. IL-10levels in cerebrospinal fluid and serum of patients with severetraumatic brain injury: relationship to IL-6, TNF-alpha, TGF-beta1 and blood-brain barrier function. J Neuroimmunol 1999;

101:211–221.

2

2

2

2

2

2

2

3

3

3

486 LaPar et al Discussion J Am Coll Surg

22. Goodman JC, Robertson CS, Grossman RG, Narayan RK. El-evation of tumor necrosis factor in head injury. J Neuroimmu-nol 1990;30:213–217.

3. Morganti-Kossman MC, Lenzlinger PM, Hans V, et al. Produc-tion of cytokines following brain injury: beneficial and deleteri-ous for the damaged tissue. Mol Psychiatry 1997;2:133–136.

4. Ott L, McClain CJ, Gillespie M, Young B. Cytokines and met-abolic dysfunction after severe head injury. J Neurotrauma1994;11:447–472.

5. Whalen MJ, Carlos TM, Kochanek PM, et al. Interleukin-8 isincreased in cerebrospinal fluid of children with severe headinjury. Crit Care Med 2000;28:929–934.

6. Jones NC, Prior MJ, Burden-Teh E, et al. Antagonism of theinterleukin-1 receptor following traumatic brain injury in themouse reduces the number of nitric oxide synthase-2-positivecells and improves anatomical and functional outcomes. EurJ Neurosci 2005;22:72–78.

7. Sanderson KL, Raghupathi R, Saatman KE, et al. Interleukin-1receptor antagonist attenuates regional neuronal cell death andcognitive dysfunction after experimental brain injury. J CerebBlood Flow Metab 1999;19:1118–1125.

8. Hama T, Miyamoto M, Tsukui H, et al. Interleukin-6 as a neu-rotrophic factor for promoting the survival of cultured basalforebrain cholinergic neurons from postnatal rats. Neurosci Lett1989;104:340–344.

9. Kossmann T, Hans V, Imhof HG, et al. Interleukin-6 released inhuman cerebrospinal fluid following traumatic brain injury maytrigger nerve growth factor production in astrocytes. Brain Res1996;713:143–152.

0. Satoh T, Nakamura S, Taga T, et al. Induction of neuronaldifferentiation in PC12 cells by B-cell stimulatory factor2/interleukin 6. Mol Cell Biol 1988;8:3546–3549.

1. Fan L, Young PR, Barone FC, et al. Experimental brain injuryinduces differential expression of tumor necrosis factor-alphamRNA in the CNS. Brain Res Mol Brain Res 1996;36:287–291.

2. Taupin V, Toulmond S, Serrano A, et al. Increase in IL-6, IL-1and TNF levels in rat brain following traumatic lesion. Influenceof pre- and post-traumatic treatment with Ro5 4864, aperipheral-type (p site) benzodiazepine ligand. J Neuroimmu-nol 1993;42:177–185.

Discussion

DR MARTIN A CROCE (Memphis, TN): The authors collectedblood on a number of trauma patients and analyzed it for various pro-and anti-inflammatory mediators. They then analyzed these datarelative to the presence or absence of severe brain injury to see ifthere’s any relation to subsequent infection. The mystifying patternof these pro- and anti-inflammatory mediators reminds me of theScarecrow in the Wizard of Oz. Which direction is the right one?

I have a number of questions for the authors, and I’ll start with aneasy one.

There’s a rapid sequence of cytokine peaks and nadirs after injury.Do you think that collecting baseline data up to 3 days after admis-sion leads to valid data?

Inclusion of the severe head injury group required a head Abbre-

viated Injury Score (AIS) of 2 or more, which does not necessarily

identify a severely injured brain, yet the Injury Severity Score (ISS)for the group was 34, suggesting significant multisystem injury. Howcan you separate the contribution of associated injuries from braininjury to the perturbations in these cytokine levels?

Why not study patients with isolated brain injuries to eliminatethis variable? Furthermore, why use AIS as a marker of severe braininjury instead of something like a 24-hour score on the GlasgowComa Scale (GCS)? You would likely identify patients using thismethod with diffuse axonal injury, eliminate the effects of sedation,and keep with the more conventional methods of stratification, suchas a GCS of 8 or less indicates severe brain injury and coma, 9 to 12indicates moderate brain injury, and 13 or higher equals a mild braininjury.

There was no method of severity of shock or stratification oftransfusion requirements. How can we interpret these cytokine changeswithout these data, especially when infection is one of the outcomes?Virtually every other study involving patients with severe brain injurieshas demonstrated an increased infection rate in these patients, yet there’sno such difference in this particular study. What’s different? Could thisbe related to the definition of severe brain injury as mentioned earlier?

Finally, those with severe brain injury had a higher mortality,which is not particularly surprising. What were the causes of death? Ifdeath was due to progression of brain injury, then the cytokine datareally mean very little. If death was due to progression of multipleorgan failure, then the cytokine data would be very intriguing butmay have little to do with severity of brain injury.

I would like to thank Dr Sawyer and his colleagues for this pro-vocative paper. His group has a serious interest in surgical infectionsand, believe it or not, I really do enjoy their work.

DR GLEN FRANKLIN (Louisville, KY): I too share Dr Croce’sconcerns with regard to the way head injury was identified, using theAIS for the brain as opposed to a standard score such as GCS. AIS forthe head could also include other injuries about the head and neckthat may or may not correlate with a true brain injury. I’m curious asto why the authors chose this instead of a standard use of a GCS.

In the Methods section, they say that in patients who arrivedsedated, or chemically perturbed, that the GCS was then estimatedbased on what was assumed to be their GCS had they not hadsedation. I do find this concerning. Perhaps things are different inCharlottesville than they are in Louisville, but nearly all of my pa-tients are sedated by my emergency department before I get an op-portunity to see them, due to their flailing about on the bed. And I’mnot sure that I would be able to accurately estimate the GCS were Igiven the opportunity.

The authors noted a difference in penetrating injury between thehead-injured group and the nonhead-injured group of 14% vs 6%,respectively. And I wonder if these data should be excluded whentalking about cytokine analysis, particularly when stratifying by in-jury severity scores because penetrating trauma sometimes is a muchdifferent animal. But how many of these patients had gunshotwounds to the head? Those, for sure, I think, should be eliminatedfrom the group.

There was no difference seen in ICU infection, and this waspointed out by Dr Croce. I’m also concerned about whether the

higher incidence of pulmonary disease, transfusions, older patients,