Hydrocortisone and Tumor Necrosis Factor in Severe Community-Acquired Pneumonia

Hydrocortisone and Tumor Necrosis Factor in Severe Community-Acquired Pneumonia* A Randomized Controlled Study lbul Marik, M.B.B.Ch., M.Med.;t Peter Kraus, M.B.B.Ch.; Juan Sribante, R.N.; Ivan Havlik , Ph .D.; Jeffery Lipman, M.B.B.Ch.; and David W johnson, M.B .B.Ch . Background: Community-acquired pneumonia is a major cause of death in third world countries. Antimicrobial therapy may have little impact on the natural history of patients with severe pneumonia. We hypothesized that the intrapulmonary production of tumor necrosis factor-a (TNF-a) may be responsible for the progressive lung injury and shock commonly seen in patients with severe pneumonia after commencing antibiotic therapy. Aim: To investigate the effects of a single bolus of hydrocortisone on the clinical course and serum TNF-a levels of patients with severe community-acquired pneumonia. Design: Randomized placebo-controlled study. Setting: Multidisciplinary ICU of a tertiary care teaching hospital. lbtients and methods: Patients with three or more British Thoracic Society criteria of severe pneumonia were studied. Patients were randomized to receive either a single dose of hydrocortisone (10 mg/kg) or placebo 30 min prior to commencing antibiotic therapy. Patients were treated with cefotaxime and other antibiotics as clinically indicated. Blood for TNF-a was taken at the time of hospital admission

and repeated 2, 6, and 12 h after starting antibiotic therapy. Results: Thirty patients were studied: 16 received placebo and 14 received hydrocortisone. The patients who received placebo tended to be sicker than the patients who received hydrocortisone. The baseline TNF-a value was 989±374 pg/ml in the placebo group and 827±394 pg/ml in the hydrocortisone group. In both groups of patients, the TN Fa levels did not change significantly with time. There was no correlation between the TN F-a levels and the APACHE II score, lung injury score, or outcome. The only variable that predicted outcome was the APACHE II score. Conclusion: Bactericidal antibiotics do not increase serum TNF-a levels in patients with severe pneumonia. Hydrocortisone given prior to antibiotic treatment had no effect on the serum TN F-a levels or the clinical course of patients with severe community-acquired pneumonia. (Chest 1993; 104:389-92)

pneumonia ranks among the C ommunity-acquired five major causes of death in both Western and

to die." The reason why an immunocompetent patient with a community-acquired pneumonia treated with appropriate antibiotics should die is unknown. Pneumolysin released from the cell wall of the pneumococcus after treatment with an antibiotic may play a role in the death of the host. 7 We postulated that bacterial lysis and antigen release following the use of bactericidal antibiotics might result in a "second wave" of tumor necrosis factor-a (TNF-a) production, with progressive lung injury and multiple organ dysfunction syndrome (MODS). Recently, a number of studies have shown that "low dose" corticosteroids may be beneficial in pediatric patients with bacterial meningitis.'l-1° Hydrocortisone, prednisone, and dexamethasone were used in these studies, in a dosage that ranged from the equivalent of 2.5 to 15 mglkgld of hydrocortisone. Low doses of corticosteroids have been shown to inhibit the production ofTNF-a. 11 · 12 We hypothesized that low-dose hydrocortisone given prior to antibiotic therapy would prevent the second wave ofTNF-a release in patients with severe pneumonia.

third world countries. I-4 Despite the availability of effective antimicrobial agents, the mortality of community-acquired pneumonia currently ranges from 6 to 24 percent. H The prognosis of patients with severe pneumonia admitted to the ICU is bleak. Hook and colleagues5 and Feldman and colleagues6 reported a mortality of 76 percent and 53 percent, respectively, for patients with community-acquired pneumonia admitted to the ICU. Austrian and Gold, 1 having studied the effect of penicillin on the natural history of pneumonia, concluded that "the data suggests that antimicrobial therapy has little or no impact upon the outcome of infection among those destined, at the onset of illness, *From the Intensive Care Unit, Baragwanath Hospital and Department of Clinical and Experimental Pharmacology (Dr. Havlik), University of the Witwatersrand, Johannesburg, South Africa. tCurrently Assistant Professor, Division of Critical Care, Wayne State University, Detroit. Manuscript received October 20; revision accepted December 14. Reprint requests: Dr. Marik, Detroit Receiving Hospital, 4201 St. Antoine, Rm 55-10, Detroit 48201

MODS=multiple organ dysfunction syndrome; TNF-o=tumor necrosis factor-a

CHEST I 104 I 2 I AUGUST. 1993

389

METHODS This study was conducted in the ICU of a tertiary care teaching hospital. The study was approved by the Institutional Review Board for research involving human subjects. Informed consent was obtained frnm all patients. Patients with community-acquired pneumonia admitted to the medical admissions ward with three or more of the li1llowing criteria (British Thoracic Society criteria of severe pneumonia) were identified and admitted to the ICU prior to the commencement of antibiotic therapy:• (1) respiratory rate >30/min; (2) diastolic BP <60 mm Hg; (3) confusion; (4) PaO, <55 mm Hg (on room air); (5) WBC count <4 or >30X 10"/L; (6) semm urea >7 mmoVL; (7) platelet count < 140 X 10"/L; and (8) radiographic evidence of multilobar involvement. The li>llowing patients were excluded from the study: patients allergic to ~-lactam antibiotics, patients with a malignancy or patients receiving immunosuppressive therapy, patients with active tuberculosis, human immunodeficiency vims-positive patients, and patients younger than 18 years or older than 70 years of age. Patients were randomized by a random number generator to receive either hydrocortisone or placebo. The patients received 10 mg/kg of hydrocortisone or placebo (normal saline solution) intravenously 30 min prior to starting antibiotic therapy. All patients had a Gram stain, culture of the sputum, and blood cultures performed prior to commencing antibiotic therapy. In addition, a full blood count, urea and electrolytes, liver function tests, and human immunodeficiency vims, Legionella, Mycoplasma, and Chlamydia pneumonia serologic tests were performed at the time of hospital admission and repeated as clinically indicated. All patients were treated with cefotaxime 1 g intravenously every 6 h, the first dose given 30 min after either hydrocortisone or placebo. Additional antibiotics were added according to sputum Gram stain and blood culture results. If the sputum Gram stain showed Gramnegative rods, amikacin, 15 mg/kgld, was added . Cloxacillin, 1 every 6 h, was added if Gram-positive cocci in chains were seen. Erythromycin, 1 g intravenously every 6 h (started 6 h after the first cefotaxime dose), was added in patients in whom the Gram stain showed no predominant organism . Blood for TNF-a was sampled in pyrogen-free EDTA tuhes 13 at the time of admission to the ICU and repeated 2 h , 6 h and , 12 h after <.-ommencing antibiotic therapy. The specimens were centrifuged immediately and the serum was stored at -70°C until assayed . The TNF-a was assayed using a TN F-a ""')-assay system (with Amerlex-M, Amersham, UK). This system uses a highly specific '~'1-TNF-a (human, recombinant) tracer, together with a specific antisemm . Separation of bound antibody from free fraction is achieved by magnetic separation. All the samples were assayed in duplicate and the concentrations of the TN F-a were established from a calibration curve using standards provided with the kit. Results were expressed in picograms per milliliter of semm . The patient received appropriate supportive treatment, including ventilatory support and inotropic agents as indicated. The patients were followed up until discharge from the ICU or until death . Each patient"s illness severity at the time of ICU admission was assessed using the APACHE II and Lung Injury Score."·'' The patient's demographic, clinical, and laboratory data were stored in a database .

Table !-Hospital Admission Characteristics of the Hydrocortisone and Placebo Groups

Age, yr Duration of symptoms, days APACHE II score Admission criteria Systolic BP, mm Hg Pulse rate, heats/min Temperature, •c Respiratory rate, breaths/min PaO, room air, mm Hg P(A-a)O,, * mm Hg Lung injury score WBC count, x 10"/L Blood urea, mmoVL

390

Hydrocortisone (n=14)

40.6± 14.7 3.3± 1.8 14.2±6.4 4. 1 ± 1.3 105±25 115±14.7 37.9± 1.0 40± 12.1 45±9.9 190.5± 109 1.68±0.88 11.9±8.9 11.5±7.9

31.7±12.8 3.2±2.3 11.2± 1.96 3.6±0.6 106± 12.4 113± 16.6 38.4±0.7 36±9.1 44.8±5.7 152.1±62.8 1.66±0.71 18.9±7.8t 7.8±4.2

*P(A-a)O, =alveolar arterial 0, difference. tp=0.03. as the mean ± S D, with statistical significance declared for probability values <0.05. RESULTS

Thirty patients were entered into the study: 16 received placebo and 14 received hydrocortisone. The admission characteristics of the two groups are listed in Table 1. The patients who received placebo tended to be sicker than the hydrocortisone group; however, only the WBC count was significantly different between the two groups. Serum TN F-a levels were elevated in all the patients studied. The serial TNF levels are listed in Table 2. There was no difference in the TN F-a levels between the placebo and hydrocortisone group at any of the time intervals. In addition, there was no significant change within each group with time . Furthermore, when analyzing the change in TNF-a in individual patients, there was no difference between the two groups. Blood cultures were positive in four patients, with pneumococci present in three and Klebsiella pneumoniae present in one patient. Gram-positive diplococci were seen on Gram stain in 11 additional patients, Gram-positive cocci in chains in two, and Gram-negative rods in three. There was no correlation between the offending organism, the antibiotic regimen, and the TNF-a levels. Table 2-Serial Tumor Necrosis Factor-a (Picogram per Milliliter) in the Hydrocortisone and Placebo Groups

Statistics Summary statistics of the placebo and hydrocortisone groups are presented. These variables and the TNF-a at each time interval were <.-ompared using unpaired t tests. To examine if significant changes in TNF-a occurred within each group with time, the baseline, 2-h, 6-h and 12-h values were examined using one-way analysis of variance. The t-orrelation between the baseline TNF levels and the APACHE II score and lung injury score were analyzed by linear regression. Unless otherwise stated, all data are expressed

Placebo (n=16)

Admission 2h 6h 12 h

Placebo (n= 16)

Hydrocortisone* (n= 14)

989±374 1,016±347 891 ±447 853±267

827±394 829±361 850±333 804±312

*No difference between the placebo and hydrocortisone groups at any time interval and no difference in each group with time. Hydrocortisone and TNF in Severe Community-Acquired Pneumonia (Marik at a/)

Six patient~ were ventilated: two in the hydrocortisone group and four in the placebo group. Four patients died , all of whom were mechanically ventilated (p = 0.()()1). Of the four patients who died, three were in the placebo group and one was in the hydrocortisone group (NS). The mean length of stay in the ICU was 4.6±5.9 days in the placebo group and 4.3±3.8 days in the hydrocortisone group. No obvious steroidrelated complications were recorded in the hydrocortisone group. The TNF-alevel was 908±386 pglml in the survivors compared with 936 ± 422 pglml in the nonsurvivors (NS). Only the APACHE II score predicted outcome: 11.3±3.5 in the survivors and 19.0±6.7 in the nonsurvivors (p=0.03). There was a poor correlation between the baseline TNF levels and the APACHE II score (r= -0.04) and the lung injury score (r= -0.09). DISCUSSION

In this study, we have documented that patients with severe community-acquired pneumonia have high baseline levels of serum TN F-a that are unaffected by bactericidal antibiotics and remain stable when followed fqr up to 12 h. Furthermore, the serum TNF-a levels were unaffected by treatment with low-dose hydrocortisone. In addition, we were unable to demonstrate a relationship between the serum TNF-a level and illness severity as measured by the APACHE II score, lung injury score, or patient outcome. It has been suggested that the variation in the severity of a patient's response to infection may be related to the amount ofTNF-a produced.l6·17 Tumor necrosis factor-a and other cytokines have been measured in a number of infective disorders. In most of these studies, an association between the level ofTNFa in the serum and mortality has been found . 1s-22 Such an association has not been demonstrated with the other cytokines. 1s- 22 Gram-positive cell wall as well as endotoxin stimulate TN F-a production by macrophages. 12 ·16 ·20 Alveolar macrophages are very efficient TNF-a producers. 23 •24 The intrapulmonary TNF-a production may contribute to the acute lung injury and MODS seen in patients with severe pneumonia. There is significant compartmentalized production of TNF-a. Systemic administration of endotoxin results in increased circulating levels of TNF-a. In contrast, intratracheal administration of endotoxin significantly increases pulmonary alveolar lavage TNF-alevels without altering circulating TNF-a levels.25 Suter and colleagues26 found high bronchoalveolar levels of TNF-a with normal levels in the serum of patients with severe adult respiratory distress syndrome. The levels of TNF-a in the cerebrospinal fluid of patients with meningococcemia are much higher than those meas-

ured in the circulation. 27 It is possible that the compartmentalized production ofTNF-a may explain the poor correlation between the serum TN F-a levels and illness severity and outcome in this study. In experimental models, injection of endotoxin is associated with a rapid and brief pulse of circulating TNF-a. In animals and humans injected with endotoxin, serum TNF-a peaks between 1 to 2 h and returns to the base value after 6 h , with a half-life of between 14 to 28 min.28-32 Waage33 studied the TNF-a response in rats that received endotoxin either as a bolus injection or as a continuous infusion. In this experiment, the TNF-a level peaked between 60 and 90 min returning to baseline within 2 h irrespective of the mode of endotoxin challenge . When these animals were rechallenged with endotoxin 24 h later, they showed a diminished response , achieving peak TNF-a levels that were on average only 15 percent of the initial response. In the clinical setting, it is unlikely that the release of bacterial antigens into the blood stream occurs as a bolus or continuous infusion as has been used in experimental models. Rather small amounts of bacterial antigens may enter the blood stream at irregular intervals stimulating macrophages in a way very different to that of the experimental models. This may explain the stable TNF-a levels found in this study. In a study of patients in septic shock, Damas et al 19 found the TNF-a levels to be stable with time, with most patients not showing a peak level. Kwiatkoski et al 22 measured the TNF-a levels in patients with cerebral malaria over a 24-h period. The serum levels taken 24 h after admission were not significantly different from the admission levels. In this study, we have demonstrated that low-dose hydrocortisone does not affect serum TNF-a levels in patients with severe pneumonia. This finding is supported by experimental data. Glucocorticoids in nanomolar concentrations strongly inhibit TNF-a gene transcription and TNF-a production if given prior to endotoxin challenge .ll.l2.33 However, once translation of the mRNA is in progress, glucocorticoids are unable to arrest the production and secretion of TNF-a . 11 •12 This finding may· explain the failure of corticosteroids to favorably affect outcome in patients with malaria, adult respiratory distress syndrome, and septic shock who were treated with high doses of methylprednisolone .34 · 37 Furthermore, systemic corticosteroids may be associated with an increased incidence of potentially hazardous side effects.34· 37 CoNCLU S ION

The use of bactericidal antibiotics in patients with severe pneumonia was not associated with an increase in serum TNF-a levels. Furthermore, hydrocortisone CHEST I 104 I 2 I AUGUST. 1993

391

given at the time of commencing antibiotic treatment had no effect on the serum TN F-a levels or the clinical course of patients with severe community-acquired pneumonia. ACKNOWLEDGMENT: The authors acknowledge the technical assistance of Avril Mer, Dip, Medical Tech .

REFERENCES 1 Austrian R, Gold J. Pneumococcal bacteremia with special reference to bacteremic pneumococcal pneumonia. Ann Intern Med 1964; 60:759-70 2 Burman LA, Norby D , Trollfors B. Invasive pneumococcal infections: incidence, predisposing factors and prognosis. Rev Infect Dis 1985; 7:133-42 3 Macfarlane J. Community-acquired pneumonia. Br J Dis Chest 1987; 81:116-27 4 British Thoracic Society Research Committee. Communityacquired pneumonia in adults in British Hospitals in 1982-1983: a survey of aetiology, mortality, prognostic factors and outcome. Q J Med 1987; 62:195-220 5 Hook EW, Horton CA, Schaberg DR. Failure of intensive care unit support to in8uence mortality from pneumococcal bacteremia. JAMA 1983; 249:1055-57 6 Feldman C, Kallenbach JM, Levy H, Reinach SG, Hurwitz MD, Thorburn JR, et al . Community acquired pneumonia of diverse etiology: prognostic features in patients admitted to an intensive care unit and a s' everity of illness' score. Int Care Med 1989; 15:302-07 7 Klugman K. Pneumococcal resistance to antibiotics. Clin Microbioi Rev 1990; 3:171-96 8 Lebel MH, Freij BJ, Syrogiannopoulos GA. Dexamethasone therapy for bacterial meningitis: results of two double-blind, placebo controlled trials. N Engl J Med 1988; 319:964-71 9 Lebel MH, Hoyt MJ, Waagner DC, Rollins NK , Finitzo T, McCracken GH Jr. Magnetic resonance imaging and dexamethasone therapy for bacterial meningitis. AJDC 1989; 143:301-06 10 Bademosi 0, Osuntokun BO. Prednisolone in treatment of pneumococcal meningitis. Trop Geogr Med 1979; 31 :53-6 11 Beutler B, Cerami A. Cachectin and tumor necrosis factor as two sides of the same biological coin. Nature 1986; 320:584-88 12 Beutler B, Krochin N, Milsark I, Luedke C, Cerami A. Control of cachectin (tumor necrosis factor) synthesis: mechanism of endotoxin resistance. Science 1986; 232:977-90 13 Leroux-Roels G, Offner F, Philippe J, Vermeulin A. In8uence of blood collecting-systems on concentrations of tumor necrosis factor in serum and plasma. Clin Chern 1988; 34:2373-74 14 Knaus WA , Draper EA , Wagner DP, Zimmerman JE. APACHE II : a severity of disease classification system . Crit Care Med 1985; 13:818-29 15 Murray JF, Matthay MA, Luce JM , Flick MR. An expanded definition of the adult respiratory distress syndrome . Am Rev Respir Dis 1988; 138:720-23 16 Tracey KJ, Lowry SF, Cerami A. Cachectin!fNF in septic shock and septic adult respiratory distress syndrome [editorial] . Am Rev Respir Dis 1988; 138:1377-79 17 Tracey KJ, Beutler B, Lowry SF, Merryweather J, Wolpe S, Milsar 1\V, et al. Shock and tissue injury induced by recombinant human cachectin. Science 1986; 234:470-74 18 Debets JMH, Kampmeijer R, van der Linden MPMH, Buurman WA, van der Linden CJ. Plasma tumor necrosis factor and mortality in critically ill septic patients. Crit Care Med 1989; 17:489-94 19 Damas P, Reuter A, Gysen P, Demonty MD , Lamy M, Franchimont P. Tumor necrosis factor and interleukin-1 serum levels

392

during severe sepsis in humans. Crit Care Med 1989; 17:975-78 20 Marks JD, Marks CB, Luce JM , Montgomery AB, Turner J, Metz CA, et al. Plasma tumor necrosis factor in patients with septic shock: mortality rate, incidence of adult respiratory distress syndrome, and effects of methylprednisolone administration. Am Rev Respir Dis 1990; 141:94-7 21 Waage A, Halstensen A, Espevik T. Association between tumor necrosis factor in the serum and fatal outcome in patients with meningococcal disease . Lancet 1987; 1:355-57 22 Kwiatowski D, Hill AVS , Sambou I, Twumasi P, Castacane J, Manogue KR, et al . TNF mncentration in fatal cerebral, nonfatal cerebral, and uncomplicated plasmodium falciparum malaria. Lancet 1990; 336:1201-04 23 Spatafora M, Merendino A, Gjomarkaj M, Chippara G, Melis M, Bellia V, et al. Tumor necrosis factor release by pulmonary alveolar macrophages. Respiration 1988; 54(suppl1):73-7 24 Matthews N. Tumor necrosis factor from the rabbit: V. Synthesis in vitro by mononuclear phagocytes from various tissues of normal and BCG-injected rabbits. Br J Cancer 1981; 44:111321 25 Nelson S, Bagby GJ, Bainton BG, Wilson LA, Thompson JJ, Summer WR. Compartmentalization of intra-alveolar and systemic liposaccharide-induced tumor necrosis factor and pulmonary in8ammatory response . J Infect Dis 1989; 159:189-94 26 Suter PM, Suter S, Girardin E , Roux-Lombard P, Grau GE, Dayer JM. High bronchoalveolar levels of tumor necrosis factor and its inhibitors, interleukin-1 , interferon, and elastase in patients with adult respiratory distress syndrome after trauma, shock or sepsis.. Am Rev Respir Dis 1992; 145:1016-22 27 Leist TP, Frei K, Kam-Hansen S, Zinkernagel RM, Fontana A. Tumor necrosis factor in cerebrospinal 8uid during bacterial but not viral meningitis. J Exp Med 1988; 167:1743-48 28 Michie HR, Manogue KR, Spriggs DR, Revhaug A, O'Dwyer S, Dinarello CA, et al. Detection of circulating tumor necrosis factor after endotoxin administration. N Engl J Med 1988; 318:1481-86 29 Hesse DG, Tracy KJ, Fong Y. Cytokine appearance in human endotoxemia and primate bacteremia. Surg Gyneml Obstet 1988; 166:147-52 30 Beutler BA, Milsark IW, Cerami A. Cachectin/tumor necrosis factor: production, distribution and metabolic fate in vivo. J Immunol1985; 135:3972-77 31 Zuckerman SH, Bendele AM. Regulation of serum tumor necrosis factor in glucocorticoid-sensitive and resistant rodent endotoxin shock model. Infect lmmun 1989; 57:3009-13 32 Fromm RE, Sulfredini AF, Kovacs JA, Schlesinger T, Parrillo JE . Circulating tumor necrosis factor in normal volunteers receiving endotoxin. Crit Care Med l988; 16:397 33 Waage A. Production and clearance of tumor necrosis factor in rats exposed to endotoxin and dexamethasone. Clin lmmunol Immunopathol1987; 45:348-55 34 Sprung CL, Caralis PV, Macial RRT, Pierce M, Gelbard MA , Long WM, et al. The effects of high-dose mrtimsteroids in patients with septic shock: a prospective, controlled study. N Engl J Med 1984; 311:1137-43 35 Bone RC, Fisher CJ, Clemmer TP, Slotman GJ, Metz CA. Early methylprednisolone treatment for septic syndrome and adult respiratory distress syndrome. Chest 1987; 92:1032-36 36 Bernard GR , Luce JM, Sprung CL, Rinaldo JE , Tate RM, Sibbald WJ, et al. High-dose corticosteroids in the patients with the adult respiratory distress syndrome. N Engl J Med 1987; 317:1565-70 37 Warrell DA, Looareesuwan S, Warrel MJ, Kasemsarn P, Intaraprasert R, Bunnad D, et al. Dexamethasone proves deleterious in cerebral malaria. N Engl J Med 1982; 306:313-19

Hydrocortisone and TNF in Severe Community-Acquired Pneumonia (Marik eta/)