Corticosteroids in Acute Respiratory Failure

MICHAEL A. JANTZ and STEVEN A. SAHN

Division of Pulmonary and Critical Care Medicine, Allergy and Clinical Immunology, Medical University of South Carolina, Charleston, South Carolina

	CONTENTS

TOP CONTENTS PHARMACOLOGY AND MECHANISMS OF... CORTICOSTEROIDS IN ACUTE... REFERENCES

Introduction

Pharmacology and Mechanisms of Action

Corticosteroids in Acute Respiratory Failure

Acute Respiratory Distress Syndrome

Status Asthmaticus

COPD

Pneumocystis carinii Pneumonia

Acute Eosinophilic Pneumonia

Alveolar Hemorrhage Syndromes

Systemic Lupus Erythematosus

Wegener's Granulomatosis

Microscopic Polyangiitis

Anti-Glomerular Basement Membrane Disease (Goodpasture's Syndrome)

Bone Marrow Transplantation

Acute Lupus Pneumonitis

Bronchiolitis Obliterans Organizing Pneumonia (Crypto-     genic Organizing Pneumonia)

Radiation Pneumonitis

Miliary Tuberculosis

Pulmonary Toxicity Secondary to Drugs or Toxins

Bleomycin Pneumonitis

Mitomycin

Amiodarone

Cocaine

Acute Complications of Corticosteroid Therapy

Corticosteroids have been advocated for a host of conditions which the pulmonologist/intensivist may encounter as causes of acute respiratory failure. The purpose of this article is to examine the data and rationale for use of corticosteroids in conditions that are most likely to cause acute respiratory failure requiring admission to the intensive care unit (ICU). At the conclusion of each section discussing a specific disease or condition, we provide our recommendation for corticosteroid treatment of that entity, cognizant that randomized, double-blind, placebo-controlled trials may be lacking. We focus primarily on human studies and not in vitro or animal model studies. In the following section on the mechanisms of action of corticosteroids, we provide a general overview and do not cite each individual study as our focus is the clinical treatment of conditions encountered in the ICU. For additional in-depth discussion of the molecular actions of corticosteroids, the reader is referred to the excellent discussions provided by Stellato and coworkers (1) and by Barnes and Adcock (2).

	PHARMACOLOGY AND MECHANISMS OF ACTION

TOP CONTENTS PHARMACOLOGY AND MECHANISMS OF... CORTICOSTEROIDS IN ACUTE... REFERENCES

Corticosteroids are mainly transported in the blood complexed to transcortin (corticosteroid-binding globulin) and albumin, although a small portion is in a free, metabolically active state. The free corticosteroid molecules readily cross the plasma membrane into the cytoplasm. Once in the cytoplasm, corticosteroids bind to their specific receptor, the glucocorticoid receptor (GR). The GR is located in the cytoplasm of nearly all human cells. In the cytoplasm, the GR exists as a heterocomplex that contains the GR in association with two subunits of the heat shock protein hsp90, one subunit of the heat shock protein hsp56, one subunit of the heat shock protein hsp70, and an acidic 23 kD protein (3, 4). After hormone binding, the GR complex undergoes dissociation and the ligand-activated GR translocates to the nucleus. In the nucleus, the activated GR can bind to glucocorticoid response elements on DNA or can interact with transcription factors (5). Depending on the specific glucocorticoid regulatory element to which the GR binds, the transcription rate of a specific protein's messenger RNA (mRNA) can be induced (upregulated) or suppressed (downregulated) (11). In addition to effects mediated through glucocorticoid response elements located in promoter regions of DNA and interactions with other transcription factors, the activated GR can alter mRNA stability (12, 13).

Corticosteroids inhibit the transcription of several cytokines that are relevant to inflammatory conditions including interleukin (IL)-1, IL-3, IL-4, IL-5, IL-6, IL-8, tumor necrosis factor (TNF)-, and granulocyte-macrophage colony stimulating factor (GM-CSF) (14). Corticosteroids interact with the transcription factors activator protein (AP)-1 and nuclear factor (NF)-B (7, 9, 10). The transcription factors AP-1 and/or NF-B are known to be involved in the upregulation of a number of gene products that play a central role in inflammation including IL-1, IL-2, IL-3, IL-6, IL-8, TNF-, GM-CSF, and RANTES (Regulated upon Activation, Normal T-cell Expressed and Secreted) (15). By interfering with these transcription factors, corticosteroids can inhibit expression of these genes. Transcription of the IL-2 gene is predominantly regulated by nuclear factor of activated T cells (NF-AT), which is dependent on a nuclear factor thought to be AP-1, to form a transcriptional complex (16, 17). Thus, corticosteroids may inhibit IL-2 gene transcription indirectly by binding to AP-1 (18). Corticosteroids have also been demonstrated to increase the degradation of mRNA encoding for IL-1, IL-6, and GM-CSF (12, 13).

Corticosteroids have an effect on a variety of inflammatory mediators. Nitric oxide synthase may be induced by various cytokines, resulting in increased nitric oxide production (19, 20). The induction of the inducible form of nitric oxide synthase is inhibited by corticosteroids (21, 22), presumably by effects on NF-B which is an important transcription factor in regulating inducible nitric oxide synthase (23). Corticosteroids inhibit the gene encoding for inducible cyclo-oxygenase (COX-2), which also appears to be dependent on NF-B activation (24). Gene transcription of cytosolic phospholipase A₂, which is induced by cytokines and cleaves arachidonic acid providing precursors for the leukotriene and prostaglandin mediators of the COX and lipoxygenase pathways, is inhibited by corticosteroids (27, 28). Platelet-activating factor activates AP-1 binding in inflammatory cells, and this may be inhibited by steroids. In addition, corticosteroids inhibit the synthesis of endothelin-1, a peptide with potent bronchoconstrictor properties, in lung and airway epithelial cells (29, 30).

Corticosteroids increase transcription of the ₂-receptor gene. Desensitization of ₂-adrenergic receptors is prevented by corticosteroids, and downregulated receptors are restored to normal levels by corticosteroids (31). Corticosteroids also have effects on the IL-1 surface receptor, as they induce a soluble form of the IL-1 receptor, known as the IL-1 RII or decoy receptor, which reduces the functional activity of IL-1 (32, 33). In addition, corticosteroids have effects on adhesion molecules. Corticosteroids inhibit the gene transcription of intercellular adhesion molecule-1 (ICAM-1) and E-selectin (34, 35). ICAM-1 expression is dependent on NF-B activation (36). Lastly, corticosteroids increase the synthesis of secretory leukocyte protease inhibitor (SLPI) by airway epithelial cells. SLPI is thought to be the predominant antiprotease in conducting airways and may be important in reducing airway inflammation (37).

There are a number of cellular effects of corticosteroids. Steroids cause an expansion in the number of circulating neutrophils secondary to diminished adherence to the vascular endothelium (demargination) and stimulation of bone marrow production. It has been reported that corticosteroids stabilize lysosomes, inhibit the release of lysosomal enzymes, and inhibit chemotaxis and other neutrophil functions; however, these studies used extremely high concentrations of steroids and may not be pharmacologically or physiologically meaningful (38). Levels of circulating mononuclear cells, eosinophils, and basophils are decreased after steroid administration. IL-1 and TNF- release by macrophages is inhibited by corticosteroids (39). Inhibition of genes encoding for the chemokines macrophage inflammatory protein-1 (MIP-1) and monocyte chemotactic protein-1 (MCP-1) by steroids is also observed (44, 45). Corticosteroids have a direct inhibitory effect on mediator release from eosinophils and suppress the permissive action of the cytokines IL-3, IL-5, and GM-CSF that prolong eosinophil survival (46). The downregulation of IL-1 and IL-2 by steroids affects T lymphocytes as they act as growth factors for these cells (50). The release of various cytokines by T lymphocytes is also inhibited by corticosteroids (1). Although steroids do not appear to have a direct inhibitory effect on mast cell mediator release in the lung, chronic steroid treatment is associated with a marked reduction in mucosal mast cell number (53). Mucus secretion is stimulated by several inflammatory mediators and cytokines. Corticosteroids have been noted to decrease mucus secretion in the airways (54, 55). Finally, corticosteroids appear to have an inhibitory effect on leakage in the microvasculature. This effect is thought to be due to diminishing the cellular sources of proinflammatory and vasoactive mediators, and possibly by a direct antipermeability effect on the microvasculature (56).

Some investigators have proposed that corticosteroids may not simply inhibit immune responses but may enhance some immune functions and optimize the biologic response to various immunologically related stressors (60, 61). Upregulation of receptors for IL-1, IL-2, IL-4, interferon- (IFN-), GM-CSF, and TNF in various cell types by glucocorticoids has been demonstrated (60). Glucocorticoids also strongly potentiate the IL-1- and IL-6-induced expression of acute phase proteins as part of the acute phase response to tissue injury or infection, which is important in limiting local and systemic inflammation (60, 61). In human B cells, glucocorticoids are synergistic with IL-1 and IL-6 in inducing production of IgM and IgG (60). Further study is needed to understand the seemingly paradoxical effects of corticosteroids in decreasing cytokine production while upregulating cytokine receptors, and how corticosteroids optimize the immune response. Further research is also needed in assessing potential differences between the effects of endogenous corticosteroids at physiologic concentrations and the effects of exogenous, synthetic corticosteroids at pharmacologic doses on the immune response.

Cortisol is the major endogenous glucocorticoid, whereas aldosterone is the major mineralocorticoid. Cortisol is produced by the zona fasciculata and zona reticularis in the adrenal cortex; aldosterone is produced in the zona glomerulosa. The effects on immune and inflammatory modulation by glucocorticoids have previously been discussed. In addition to these effects, effects on glucose and lipid are seen with glucocorticoids. The primary action of the mineralocorticoids is to facilitate sodium reabsorption and excretion of potassium and hydrogen ions in the renal tubular cell. The normal unstressed daily production of cortisol is estimated to be 13 to 20 mg/d; the maximum stress-induced output of cortisol is estimated to be 200 to 300 mg/d (62).

The properties of the corticosteroids commonly prescribed in the ICU are listed in Table 1. Because many of the effects of corticosteroids are mediated through gene transcription and protein synthesis, the biologic half-lives of the various preparations may not directly correlate with plasma levels and, thus, are only estimates. Hydrocortisone is the synthetic equivalent to cortisol. Because of its mineralocorticoid effects in addition to its glucocorticoid effects, it is the preferred agent for physiologic replacement. This same property, however, increases the possibility of fluid and electrolyte imbalances. Prednisone, in contrast to prednisolone, requires hepatic hydroxylation to become active. Methylprednisolone may be better concentrated in the lungs than prednisolone because it has a larger volume of distribution, longer mean residence time, and greater retention in the epithelial lining fluid of the alveoli (63). Because of its negligible mineralocorticoid activity, dexamethasone may be the agent of choice in situations in which fluid and sodium retention is undesirable, such as treatment of cerebral edema.

	CORTICOSTEROIDS IN ACUTE RESPIRATORY FAILURE

TOP CONTENTS PHARMACOLOGY AND MECHANISMS OF... CORTICOSTEROIDS IN ACUTE... REFERENCES

Acute Respiratory Distress Syndrome (ARDS)

A number of trials have demonstrated that patients with ARDS do not benefit from a short course of large doses of corticosteroids administered early in the disease. Weigelt and coworkers performed a prospective, double-blind, randomized study of early methylprednisolone therapy in 81 acutely ill, mechanically ventilated patients felt to be at high risk for ARDS (64). Patients were given methylprednisolone 30 mg/kg every 6 h for 48 h. ARDS developed in 25 of 39 (64%) of the steroid-treated patients and 14 of 42 (33%) of the placebo-treated patients (p = 0.008). Mortality was 46% in the steroid- and 31% for placebo-treated patients (p = 0.18). Infectious complications occurred in 77% of the steroid group versus 43% in the placebo group (p = 0.001). Luce and coworkers conducted a prospective, double-blind, randomized trial evaluating the efficacy of methylprednisolone to prevent ARDS in patients with septic shock (65). Patients received 4 doses of methylprednisolone 30 mg/kg every 6 h or placebo. In the methylprednisolone group, 13 of 38 (34%) developed ARDS compared with 14 of 37 (38%) in the placebo group (p = not significant [NS]). Mortality was 58% in the steroid group versus 54% in the placebo group (p = NS).

Bone and coworkers also performed a prospective, double-blind, randomized study to determine if early treatment with corticosteroids would decrease the incidence of ARDS in patients at risk from sepsis (66). Patients were treated with either methylprednisolone 30 mg/kg every 6 h for 24 h or placebo. A trend toward an increased incidence of ARDS was observed in the steroid-treated group, 50 of 152 (32%) compared with the placebo-treated group, 38 of 152 (25%) (p = 0.10). Mortality at 14 d for patients who developed ARDS was 52% (26 of 50) in the steroid group and 21% (8 of 38) in the placebo group (p = 0.004), while reversal of ARDS was noted in 31% of the steroid-treated patients versus 61% in the placebo-treated patients (p = 0.005). A greater mortality from secondary infections was also noted in the steroid group. Bernard and coworkers conducted a prospective, double-blind, randomized controlled trial of methylprednisolone in 99 patients with early ARDS (67). Patients were administered methylprednisolone 30 mg/kg every 6 h for a total of 4 doses or placebo. Mortality at 45 d was 60% (30 of 50) in the steroid group and 63% (31 of 49) in the placebo group (p = 0.74). Reversal of ARDS was 36% (18 of 50) in the steroid group and 39% (19 of 49) in the placebo group (p = 0.77). Infectious complications were similar between the two groups, 16% versus 10% in the steroid and placebo groups, respectively (p = 0.60).

In contrast to early ARDS, there is evidence that corticosteroids may be beneficial in the fibroproliferative phase, or late phase, of ARDS. Ashbaugh and Maier described 10 patients who did not respond to conventional treatment (68). Open lung biopsies demonstrated cellular proliferation, obliteration of alveoli, and fibrosis. Patients were treated with methylprednisolone 125 mg every 6 h beginning 6 to 22 d after onset of ARDS. Following a clinical response, the dose of steroids was then tapered over 3 to 6 wk. Of the 10 patients, eight recovered. Two patients died from sepsis. Hooper and Kearl initially reported 10 patients followed by an additional 16 patients who were treated with corticosteroids for "established" ARDS (69, 70). Established ARDS was defined as ARDS where the causative process had resolved and the duration of ARDS was greater than 72 h. Clinically, patients were not thought to have active infections and patients with positive blood or wound cultures were excluded from the study. Bronchoalveolar lavage (BAL) or lung biopsy was not performed. ARDS was present 3 to 40 (mean 9.2) d before steroid therapy. The initial dose of methylprednisolone was 125 to 250 mg every 6 h based on the severity of respiratory compromise. The initial dose was maintained for 72 to 96 h before tapering was begun based on clinical response with the dosage reduced approximately 50% every 2 to 3 d. All patients showed improvement in respiratory parameters. Overall survival was 81% (21 of 26). Three patients died of multisystem organ failure, one died of a cardiac arrhythmia, and one died of systemic candidiasis. Biffl and coworkers utilized corticosteroids in six patients with persistent, severe ARDS who were failing conventional therapy (71). BAL was performed before initiating steroid therapy to exclude infection. Patients were treated with methylprednisolone 1 to 2 mg/kg every 6 h starting 12 to 26 (mean 15.8) d after ventilatory support was initiated. The ratio of arterial oxygen pressure to fraction of inspired oxygen (Pa_O2/FI_O2) improved from 84 to 172 (p = 0.01) and the lung injury score (LIS) decreased from 3.6 to 2.9 (p = 0.01) by Day 7. Overall survival was 83% (5 of 6). The mean duration of corticosteroid therapy was 21.3 d (range 13 to 42). One patient developed a Staphylococcus aureus lung abscess and two patients had catheter-related sepsis.

Meduri and colleagues have published a number of studies reporting the efficacy of corticosteroids for the fibroproliferative phase of ARDS. They initially described eight patients followed by an additional 17 patients (72, 73). Patients were given an initial bolus of 200 mg followed by 2 to 3 mg/kg/d in divided doses every 6 h of methylprednisolone. Bronchoscopy with bilateral BAL and protected brush sampling was performed before initiation of steroid treatment to exclude infection. Weekly bronchoscopy with BAL was also done for early detection of nosocomial pneumonia. Patients required mechanical ventilation for 5 to 23 (mean 15) d before initiation of steroid therapy. By Day 7 of treatment, the Pa_O2/FI_O2 ratio increased from 164 to 234 (p = 0.0004) and LIS decreased from 3.0 to 2.1 (p = 0.001). Three patterns of response were noted. Rapid responders showed improvement by Day 7 (n = 15), delayed responders showed improvement by Day 14 (n = 6), and nonresponders exhibited no improvement by Day 14 (n = 4). Overall survival rate was 72% (18 of 25). ICU survival was 87% (13 of 15) in the rapid responders, 83% (5 of 6) in the delayed responders, and 25% (1 of 4) in the nonresponders. The average duration of corticosteroid treatment was 36 d. Pneumonia developed in 38% of responders and 75% of nonresponders.

Meduri and coworkers recently published the results of a randomized, double-blind, placebo-controlled trial of corticosteroids for unresolving ARDS (74). Patients were eligible for the study if they had required mechanical ventilation for 7 d with a LIS of  2.5 and less than a 1-point reduction in the LIS from Day 1 of their ARDS, as well as no evidence of untreated infection. Sixteen patients were randomized to treatment with methylprednisolone while eight patients received placebo. Enrollment was stopped after 24 patients based on interim statistical analysis. The treatment protocol for methylprednisolone was a loading dose of 2 mg/kg, followed by 2 mg/ kg/d from Day 1 to Day 14, 1 mg/kg/d from Day 15 to Day 21, 0.5 mg/kg/d from Day 22 to Day 28, 0.25 mg/kg/d on Days 29 and 30, and 0.125 mg/kg/d on Days 31 and 32. If the patient was extubated before Day 14, treatment was advanced to Day 15 of drug therapy. One-fourth of the total daily methylprednisolone dose was given every 6 h. Bronchoscopy with bilateral BAL was performed prior to initiation of steroid therapy as well as weekly to exclude ventilator-associated pneumonia. Infections were identified in 10 patients prior to study entry, eight in the methylprednisolone arm and two in the control group. These patients were treated with antibiotics for at least 3 d before treatment randomization.

Weekly bronchoscopy with bilateral BAL was included in the protocol because the two previous uncontrolled studies by Meduri and coworkers had shown that ventilator-associated pneumonia may develop in the absence of fever (72, 73). Thus, surveillance bronchoscopy was performed to allow for early detection of occult pneumonia. Significant changes were observed for Pa_O2/FI_O2 ratio (262 versus 148, p < 0.001), LIS (1.7 versus 3.0, p < 0.001), mean pulmonary artery pressure (22.5 versus 30.0 mm Hg, p = 0.01), and multiple-organ dysfunction syndrome (MODS) score (0.7 versus 1.8, p < 0.001) in the corticosteroid-treated group versus the placebo group. In accordance with the study protocol, four patients in the placebo arm were blindly crossed over to the methylprednisolone arm when no improvement was observed 10 d after treatment. Of the four patients who were blindly crossed over from placebo to methylprednisolone, only one survived. ICU survival was 100% (16 of 16) in the steroid group versus 37% (3 of 8) in the placebo group (p = 0.002); overall survival was 87% (14 of 16) versus 37% (3 of 8), respectively (p = 0.03). The two deaths in the methylprednisolone arm occurred after ICU discharge and were related to a cardiac arrhythmia in one patient and recurrent aspiration in the other patient with neurologic dysfunction. An increased incidence of infections was noted in the steroid group with a risk ratio of 1.80 compared with placebo, although the 95% confidence interval was 0.86 to 3.76. Four of 16 surveillance bronchoscopies in patients without fever identified a significant growth of pathogens.

Meduri has written an excellent review of the host defense response in the progression of ARDS and how that response may be affected by corticosteroid therapy, which we will summarize (75). Fibroproliferation is a stereotypical reparative reaction to tissue injury and is characterized by the replacement of damaged epithelial cells by accumulation of mesenchymal cells and their connective tissue products in the airspaces and walls of the intra-acinar microvessels. This process occurs within 7 d of the onset of ARDS with a rapid increase in the second and third week of respiratory failure. Unchecked fibroproliferation results in extensive fibrotic remodeling of the lung parenchyma. A number of cytokines mediate the host defense response to injury. Nonsurvivors of ARDS have been reported to have higher initial plasma and BAL levels of TNF-, IL-1, IL-2, IL-4, IL-6, and IL-8 compared with survivors (76, 77). Persistent elevation of plasma and BAL cytokines TNF-, IL-1, IL-6, and IL-8 has also been reported in nonsurvivors of ARDS (76, 77). Similar observations have been noted in patients with sepsis, and it is postulated that disruption of the alveolocapillary membrane in patients with ARDS allows the release of cytokines into the systemic circulation, which may contribute to the development and/or maintenance of the systemic inflammatory response syndrome (SIRS) and MODS (75, 78, 79). In the absence of inhibitory signals, these mediators of the host defense response sustain ongoing inflammation with tissue injury and stimulate proliferation of mesenchymal cells with deposition of extracellular matrix products and collagen, resulting in fibrosis. Thus, an overaggressive and protracted host defense response, rather than the inciting condition, is likely the major factor influencing outcome in ARDS.

It is hypothesized that activity of glucocorticoids produced by the hypothalamic-pituitary-adrenal (HPA) axis as well as anti-inflammatory cytokines such as IL-4, IL-10, IL-1 receptor antagonist, and soluble TNF receptors, are necessary to regulate termination of the host defense response (75). Corticosteroids inhibit the host defense response at many levels and inhibit the transcription of TNF, IL-1, IL-2, and IL-6. Corticosteroids also suppress the synthesis of phospholipase-A₂, COX-2, and nitric oxide synthase-1 genes, decreasing the production of prostanoids, platelet-activating factor, and nitric oxide, three additional key molecules in the inflammatory pathway. In addition, corticosteroids have an inhibitory effect on fibrogenesis and the expression of adhesion molecules. Cytokines can produce resistance to glucocorticoids by reducing their receptor binding affinity which may offset the modulatory activity of the HPA axis in limiting the host defense response in some patients (80, 81).

Thus, there appears to be some rationale for use of corticosteroids in the fibroproliferative phase in ARDS. In a separate study by Meduri and coworkers, the effects of corticosteroid treatment on plasma and BAL cytokine levels in nine patients with late ARDS were compared with 12 previous nonsurvivors from ARDS who had undergone cytokine concentration measurements (82). Baseline plasma and BAL cytokine levels in the nine corticosteroid-treated patients were similar to the comparison group. The surviving patients treated with corticosteroids were found to have significant reductions in plasma and BAL TNF-, IL-1, IL-6, and IL-8 concentrations. The decreases in the various cytokine levels were seen only after 5 to 14 d of steroid administration. Meduri and coworkers also examined the effects of corticosteroids on plasma and BAL levels of procollagen aminoterminal propeptide type I (PINP) and type III (PIIINP) in patients with nonresolving ARDS (83). PINP and PIIINP are secreted by fibroblasts and reflect collagen synthesis at the site of disease. Previous studies reported that nonsurvivors of ARDS had persistent elevations of plasma and BAL PIIINP levels (84). Meduri and coworkers found elevated plasma levels of PINP and PIIINP in their patients at the time of study enrollment and observed that the concentrations of PINP and PIIINP increased over time in nonsurvivors as opposed to survivors in whom the levels did not change significantly (83). BAL concentrations of PINP and PIIINP were also noted to be higher in nonsurvivors compared with survivors, although the differences were not statistically significant. In that study 11 patients who had not shown an improvement in LIS greater than 1 point were randomized to receive methylprednisolone using the same protocol as in their previous randomized trial (74), and six patients received placebo. In the patients treated with methylprednisolone, significant decreases in plasma and BAL PINP and PIIINP levels were observed, whereas no changes in these concentrations were seen in patients receiving placebo. Decreases in plasma and BAL PINP and PIIINP levels correlated with improvements in LIS and Pa_O2/FI_O2 ratios.

Timing and duration of corticosteroid therapy are also likely important. In experimental acute lung injury, corticosteroid administration was effective in decreasing lung collagen and edema formation as long as treatment was prolonged, whereas steroid withdrawal rapidly negated this positive effect (87). Limiting corticosteroid therapy to the first 6 d after acute lung injury was also shown to enhance accumulation of collagen after discontinuing treatment in an experimental model (88). These observations, together with the appreciation of the role of an exaggerated and protracted host defense response in producing nonresolving ARDS, may explain the difference between initial studies using a short course of corticosteroids early in ARDS versus the results of later studies reporting benefit with prolonged corticosteroid administration in the late, or fibroproliferative phase, of ARDS. In addition, the use of corticosteroids after a certain period of time may not be effective in changing the outcome of ARDS. In an earlier study by Meduri and coworkers, open lung biopsies were obtained in 12 patients before treatment with methylprednisolone for late ARDS (73). Histology from the responders demonstrated myxoid cellular fibrosis, preserved alveolar architecture, and absence of arteriolar subintimal fibroproliferation; whereas nonresponders had dense acellular fibrosis, distorted alveolar architecture, and arteriolar subintimal proliferation. Thus, it appears a therapeutic window exists for treating fibroproliferative ARDS and treatment will be ineffective once end- stage fibrosis has developed.

Given the current data of case series and one, albeit small, double-blind randomized controlled trial, it appears that there may be a role for corticosteroids in the fibroproliferative phase of ARDS, as opposed to early in the course or as preventative therapy for ARDS. Additional double-blind, randomized trials are needed before making a firm recommendation for treating all patients with corticosteroids. Nevertheless, in patients who have not exhibited clinical improvement after 7 to 14 d of conventional therapy, we think that it is reasonable to initiate a trial of corticosteroids after excluding sources of ongoing infection, including bronchoscopy to evaluate for lower respiratory tract infection, recognizing that ARDS and SIRS may clinically simulate infection. In the absence of any trials evaluating dosage or duration of steroid treatment, we would recommend the protocol used by Meduri and colleagues including weekly surveillance bronchoscopy. In their studies, weekly surveillance bronchoscopy with bilateral BAL was performed and believed to be useful in detecting early, occult pneumonia. It is beyond the scope of this article to discuss the role of bronchoscopy in evaluating ventilator-associated pneumonia, albeit controversial. A low threshold in evaluating patients for development of a new infection while undergoing corticosteroid therapy should be maintained. A National Institutes of Health (NIH)-sponsored multicenter randomized, controlled trial using Meduri and coworkers' protocol is planned to further assess the efficacy of corticosteroid therapy for fibroproliferative ARDS.

Status Asthmaticus

Although not all studies have reported a benefit, the majority support the efficacy of corticosteroids in treating acute, severe asthma or status asthmaticus. Fanta and coworkers conducted a randomized, double-blind, placebo-controlled trial in 20 patients who were refractory to 8 h of treatment with -agonists and aminophylline (90). Eleven patients were randomized to a 2 mg/kg bolus followed by a 0.5 mg/kg/h infusion of hydrocortisone, and nine patients received placebo. At 24 h the FEV₁ increased by 118% from baseline (p < 0.025) in the corticosteroid group, whereas FEV₁ increased by 36% from baseline (p = NS) in the placebo group. All of the patients who received corticosteroids had at least a 10% improvement in FEV₁, whereas only four patients who received placebo had similar improvement (p < 0.025). Younger and coworkers evaluated the use of corticosteroids in a pediatric population in a double-blind, randomized, placebo-controlled fashion (91). After no improvement with three bronchodilator treatments, patients were randomized to an initial bolus of 2 mg/kg of methylprednisolone followed by 1 mg/kg every 6 h or placebo. By 24 h the patients in the methylprednisolone arm (n = 15) had a significant improvement in a clinical pulmonary index score compared with patients in the placebo arm (n = 13) (p < 0.02), as well as FEF_25-75 at 36 h (p < 0.05). However, no differences were observed in changes in peak expiratory flow rate (PEFR), FEV₁, or FVC between the two groups. Pierson and coworkers also conducted a double-blind, randomized, placebo-controlled trial in pediatric patients (92). Patients were randomized to receive hydrocortisone, dexamethasone, betamethasone, or placebo. A significantly greater increase in Pa_O2 was noted in the steroid-treated patients (all three steroid groups combined) compared with the patients who received placebo, 58.5 to 78.1 mm Hg and 57.9 to 64.4 mm Hg, respectively (p < 0.005). No differences in changes in FEV₁ were seen. Seven of 15 patients in the placebo group, however, required reassignment to a steroid group. No differences were observed between the three steroid groups. Other studies have been published reporting efficacy for corticosteroids in treating acute, severe asthma (93).

Other researchers have reported no benefit with corticosteroids in treating acute, severe asthma or status asthmaticus (94). Corticosteroids are known not to have any immediate effect on pulmonary mechanics in acute asthma, with a number of studies noting a 6-h or greater delay from administration to a measurable effect on pulmonary function (90, 92, 93, 95). This initial delay most likely reflects the time necessary for the actions of corticosteroids on the ₂-receptor including synthesis of new ₂-receptors with increased receptor density and reversal of ₂-receptor desensitization and downregulation (31, 96). Given these observations on the delayed effect of corticosteroids, some studies may be discounted for having observation times that were too short (97, 98). Morell and coworkers found no difference in pulmonary function in patients treated with 10 mg/kg of methylprednisolone or 2 mg/kg of methylprednisolone every 4 h for 48 h versus placebo (99). However, a significant portion of each of the three groups who were on maintenance steroids prior to study entry continued to receive oral steroids at undisclosed doses. A negative study was reported by Luksza in comparing 1,200 mg/d and 400 mg/d of hydrocortisone to no steroid treatment as assessed by PEFR (100). The study was not randomized or blinded, however, and six of 30 (20%) patients not receiving steroids required mechanical ventilation compared with seven of 60 (12%) of patients treated with hydrocortisone. In addition to too short observation times, other potential factors may explain studies finding no benefit for corticosteroids in acute asthma. The presence of airway remodeling with a component of fixed airway obstruction secondary to previous undertreatment or lack of treatment is one possibility. Some patients in the adult studies may have had a component of chronic obstructive pulmonary disease (COPD) that would tend to negate the effect of steroid therapy. Some patients may simply have had a particularly severe and prolonged episode of status asthmaticus that requires a more extended course of corticosteroids. Finally, it has become appreciated that some patients with asthma exhibit corticosteroid resistance owing to various mechanisms (101, 102).

A number of studies have evaluated various doses of corticosteroids in treating status asthmaticus. Haskell and coworkers randomized eight patients each to 15 mg (low), 40 mg (medium), and 125 mg (high) of methylprednisolone every 6 h for 3 d in a double-blind fashion (103). Compared with initial spirometry, the high-dose group significantly improved (p < 0.01) in the first day, the medium-dose group significantly improved (p < 0.01) in the second day, and the low-dose group did not show significant improvement. Marquette and coworkers conducted a double-blind, randomized trial of 1 mg/kg/d versus 6 mg/kg/d of methylprednisolone in 45 patients (104). No differences in improvements in FEV₁ were observed. Bowler and coworkers performed a double-blind, randomized trial comparing three different doses of hydrocortisone (105). Patients received 50 mg (n = 22), 100 mg (n = 20), or 500 mg (n = 24) of hydrocortisone every 6 h for 48 h followed by oral prednisone. PEFR and FEV₁ improved with no significant differences between the three groups. Raimondi and coworkers randomized 20 patients to treatment with 80 mg/kg/d of hydrocortisone and 20 patients to 6 mg/kg/d of hydrocortisone given in divided doses every 6 h (106). No differences in spirometry were noted between the two groups. Ratto and coworkers conducted a trial in 70 patients comparing 80 mg or 160 mg twice a day of oral methylprednisolone to 125 mg or 250 mg of methylprednisolone given intravenously four times a day (107). Improvements in PEFR and FEV₁ were similar between the oral and intravenous groups. Studies comparing different corticosteroid preparations have not shown any significant differences in efficacy (92, 95).

We think that corticosteroids should be used in all patients admitted to the ICU for status asthmaticus. Given less mineralocorticoid effects with methylprednisolone compared with hydrocortisone, we favor the former. Given the available data, we would recommend a dose of at least 40 mg every 6 h although some clinicians prefer a dose of 60 mg every 6 h. Doses above that range probably do not confer additional benefit. Given the lack of data, we do not recommend a specific tapering protocol; dose reduction should be based on clinical response.

COPD

Although corticosteroids are used by many clinicians in treating acute respiratory failure from COPD, few controlled studies exist to support its use. The most widely referenced trial is by Albert and colleagues who studied 44 patients with COPD and acute respiratory insufficiency (defined as Pa_O2  65 mm Hg on room air or Pa_CO2  50 mm Hg with pH  7.35 or both) in a double-blind, randomized, placebo-controlled trial with 0.5 mg/kg of methylprednisolone every 6 h for 72 h or placebo (108). Patients also received intravenous aminophylline, inhaled isoproterenol, and antibiotics. Bedside spirometry was done before and after bronchodilator inhalation three times daily. At each time interval, the percent change in FEV₁ was greater in the methylprednisolone group (p < 0.0001). Twelve of 22 (55%) patients in the methylprednisolone arm had 40% or more improvement in prebronchodilator FEV₁ compared with three of 21 (14%) in the placebo arm (p < 0.01). For the postbronchodilator FEV₁, nine of 22 (41%) patients receiving methylprednisolone improved by at least 40% compared to three of 21 patients (14%) receiving placebo (p < 0.05). However, the statistical methods and analysis used in the study by Albert and colleagues have been questioned. In an editorial, Glenny challenged the study on the basis that the use of absolute spirometric volumes, rather than percent changes, should be used to assess differences between the methylprednisolone and placebo groups (109). When analyzed with this method, Glenny concluded that there was no effect of corticosteroid treatment. In addition, although not statistically significant, the admitting FEV₁ levels for the steroid group were lower than the placebo group, 602 ± 240 ml versus 675 ± 267 ml (p < 0.1), which may have affected the results because of the effect of regression to the mean.

Rubini and coworkers have recently published a study evaluating the effects of corticosteroids on respiratory mechanics in eight patients requiring mechanical ventilation for acute respiratory failure secondary to COPD (110). Patients were given 0.8 mg/kg of methylprednisolone and respiratory mechanics were measured before and 90 min after steroid administration. Bronchodilators were withheld at least 12 h before beginning the study. Maximal inspiratory resistance decreased from 20.3 to 15.3 cm H₂O/L/s and minimal airway resistance decreased from 16.2 to 11.9 cm H₂O/L/s (p < 0.01). Self-controlled positive end-expiratory pressure (auto-PEEP) also decreased by 16% from the baseline of 9.0 cm H₂O (p < 0.05). Given the time course for effects of corticosteroids observed in status asthmaticus, we are surprised at the results of this study.

A study often cited for showing no efficacy for corticosteroids in treating acute exacerbations of COPD is that of Emerman and coworkers (111). In their study, 96 patients with COPD who presented to the emergency department with acute respiratory distress were randomized to receive either 100 mg of methylprednisolone or placebo in a double-blind fashion. Patients also received hourly inhaled isoetharine and intravenous aminophylline. Spirometry performed after the third and fifth aerosol treatments revealed no difference in improvement in FEV₁ between the methylprednisolone group (37%) and the control group (43%) (p = NS). There was also no difference in the rate of hospitalization between the two groups (33% in the steroid group and 30% in the placebo group). As the treatment effect was only assessed for an average of 4.5 h, the negative results may reflect insufficient time to observe an effect for corticosteroid therapy.

Given the available data, it is problematic to make a firm recommendation either for or against the use of corticosteroids in acute respiratory failure secondary to COPD and the appropriate dose. In our practice, we tend to use corticosteroids for patients who are severely ill and require admission to the ICU, particularly if they require mechanical ventilation. In the absence of any controlled data, we use 40 to 60 mg of methylprednisolone every 6 to 12 h for 72 h.

Pneumocystis carinii Pneumonia

It is beyond the scope of this article to discuss the clinical features of Pneumocystis carinii pneumonia (PCP) and the various treatment modalities. The reader is referred to several recent excellent reviews of the subject (112). This section will focus specifically on the use of corticosteroids for PCP-induced respiratory failure. Although the incidence of PCP has decreased with the use of chemoprophylaxis, PCP remains the most common opportunistic infection associated with human immunodeficiency virus (HIV) infection. In the Multicenter AIDS Cohort Study the incidence of PCP as the acquired immunodeficiency syndrome (AIDS) index diagnosis was 15% among those receiving chemoprophylaxis, with 28% developing PCP at some point during their illness. For those not receiving chemoprophylaxis, 46% developed PCP as their initial AIDS-related illness (115). PCP continues to be the most common diagnosis among HIV-infected patients requiring ICU admission (116).

Prior to the use of corticosteroids, the survival for patients with AIDS-related PCP who required admission to the ICU for mechanical ventilation was below 15% (117, 118). The first suggestion that corticosteroids might be beneficial in PCP was noted in case reports in 1985 and 1987 (119). A number of case series were then published between 1987 and 1990 reporting a beneficial effect of corticosteroids in PCP-induced acute respiratory failure. MacFadden and coworkers described 10 patients with acute respiratory failure (Pa_O2 < 60 mm Hg with FI_O2  0.6) who were treated with methylprednisolone 40 mg intravenously every 6 h for 7 d compared with eight patients who received conventional treatment (122). Nine of the 10 (90%) steroid-treated patients survived whereas only two of the eight (25%) conventionally treated patients survived (p < 0.01). Walmsley and colleagues reported on treatment of 21 episodes of PCP in 20 patients with corticosteroids compared with 12 retrospective controls who were similar to the steroid treatment group except for being less hypoxemic (Pa_O2 47 mm Hg steroid group versus 56 mm Hg controls) (123). Six patients were treated with 20 to 120 mg/d of prednisone or methylprednisolone for 4 to 20 d, and 15 patients were treated with 80 mg/d of either steroid for 5 d. Corticosteroid treatment was associated with a significantly greater improvement in oxygenation and a trend toward lower mortality (10% versus 25%). Five patients worsened after the 5-d regimen of steroids was suddenly discontinued with a repeat course resulting in improvement of three of the five patients.

Two additional nonrandomized trials were then published suggesting decreased mortality and decreased need for mechanical ventilation in patients treated with corticosteroids (Table 2). After these case series, additional randomized trials were performed to evaluate the role of corticosteroid treatment in PCP (Table 3). These trials demonstrated that the use of corticosteroids within 48 to 72 h after initiating anti-Pneumocystis therapy decreased the incidence of early deterioration in oxygenation as well as lowered mortality and the development of respiratory failure. The beneficial effects of corticosteroids as adjunctive treatment for PCP have also been demonstrated in children (Table 4).

In addition to causing infection in the HIV-positive patient, PCP may occur in solid organ transplant recipients, bone marrow transplant recipients, patients undergoing chemotherapy for hematologic and solid organ malignancies, and patients with chronic inflammatory diseases requiring prolonged use of corticosteroids or other immunosuppressives (e.g., methotrexate) (133). The mortality of PCP in these patient populations is higher than the HIV population, ranging from 34 to 58% (133). Pareja and coworkers conducted a retrospective review examining the effects of adjunctive corticosteroids in cases of severe PCP in non-HIV patients (139). Thirty patients were identified who had a Pa_O2 < 5 mm Hg or O₂ saturation < 90% on room air. Sixteen patients received increased steroids ( 60 mg prednisone or equivalent daily) while 14 patients were maintained on a low dose of steroid ( 30 mg prednisone or equivalent daily) or had steroid therapy tapered. The increased high-dose steroid group demonstrated a shorter duration of mechanical ventilation (6.3 versus 18.0 d, p = 0.047), a shorter duration of ICU stay (8.5 versus 15.8 d, p = 0.025), and a shorter duration of supplemental oxygen requirement (10.0 versus 32.2 d, p = 0.05). No difference in mortality was noted (44% versus 36%, p = NS).

Although there has been some concern about precipitating serious infections or worsening opportunistic infections in HIV patients with the use of corticosteroids based upon case reports in the literature, analysis of a number of series is not supportive. In the largest series of 251 patients, there was an excess of localized herpes infections (26% versus 15% in the standard therapy group), but no other opportunistic infections (129). In a retrospective analysis of 23 patients treated with corticosteroids versus 16 patients treated without corticosteroids, Lambertus and coworkers did not find an excess toxicity in the steroid-treated group with one patient developing cryptococcal meningitis 3 wk after completing treatment (140). In a study by Gallant and coworkers (141) of 53 patients who received adjunctive corticosteroids and 121 patients who did not receive steroid therapy, no differences were seen between the two groups in the incidence of cytomegalovirus (CMV), Mycobacterium avium complex, cryptococcal meningitis, toxoplasmosis, Kaposi's sarcoma, herpes simplex, herpes zoster, or non-Hodgkin's lymphoma. Esophageal candidiasis was more common among the patients who received corticosteroids. Comparing 94 patients receiving corticosteroids with 50 patients who did not, Jones and coworkers found that use of corticosteroids did not increase the morbidity from undiagnosed tuberculosis (n = 8) or increase the frequency of reactivation tuberculosis (142). No increase in risk of developing tuberculosis or nontuberculous mycobacterial infection was noted by Martos and coworkers in 72 patients treated with corticosteroids versus 57 patients who did not receive steroid therapy (143). Jensen and colleagues did note a trend toward higher mortality in 21 patients with PCP treated with corticosteroids versus 21 patients not treated with corticosteroids who also had CMV cultured from BAL fluid; however, this difference was not statistically significant (p = 0.08) (144).

Based upon the available literature, we concur with the recommendations for adjunctive corticosteroid therapy as stated by the National Institutes of Health-University of California Expert Panel consensus statement with the regimen of 40 mg of prednisone twice a day on Days 1-5, 40 mg/d on Days 6-10, and 20 mg/d on Days 11-21 (145). Intravenous methylprednisolone can be given at 75% of the aforementioned doses in patients unable to take oral therapy. Most patients requiring ICU care will meet the definition of moderate-severe pulmonary dysfunction for which corticosteroids are recommended (Pa_O2 < 70 mm Hg or alveolar-arterial O₂ gradient > 35 mm Hg). To obtain maximal benefit, corticosteroids should be initiated within 24 to 72 h of the diagnosis of PCP. Based on the limited data, we would also recommend corticosteroid treatment for non-HIV patients with respiratory failure. To limit the possibility of an adverse outcome, we would recommend obtaining appropriate studies to evaluate for coexisting bacterial or opportunistic pathogens.

Acute Eosinophilic Pneumonia

Idiopathic acute eosinophilic pneumonia (AEP) was first described by Allen and Davis (146) as well as Badesch and coworkers in 1989 (147). AEP is a clinical entity distinct from other eosinophilic lung diseases. Based on their observations Allen and Davis proposed the following clinical criteria for the diagnosis of AEP: (1) acute febrile illness less than 5 d duration; (2) hypoxemic respiratory failure; (3) diffuse alveolar or mixed alveolar-interstitial infiltrates on chest radiograph; (4) BAL eosinophils greater than 25%; (5) absence of parasitic, fungal, or other infection; (6) prompt and complete response to corticosteroids; and (7) failure of relapse after discontinuation of corticosteroids.

Since then additional cases of AEP have been reported in the literature (148). The average age at presentation in the two largest series was 28.4 and 37.2 yr (n = 15 and 13, respectively) (152, 155). The youngest patient reported was 13 yr old (149) while the oldest was 86 yr old (153). In general, the mean age at onset is lower for AEP than for chronic eosinophilic pneumonia (CEP). The mean duration of symptoms has been reported to be from 2.8 d and 7.8 d (152, 155). The progression of symptoms may be very rapid with less than 24 h from time of symptom onset to respiratory failure requiring mechanical ventilation. A duration of symptoms ranging from 15 to 22 d prior to presentation has also been reported in two smaller series (150, 156). Cough, fever, and dyspnea are the prominent symptoms with pleuritic chest pain and myalgias also reported.

Most patients with AEP are febrile on presentation with temperatures ranging from 99.0 to 104.0° F. On auscultation the majority of patients have inspiratory crackles; however, a normal exam may be noted in some patients (146, 155). Although not originally described by Allen and coworkers, the presence of wheezes has been noted (151, 153, 155). The vast majority of patients demonstrate hypoxemia with Pa_O2 less than 60 mm Hg on room air with the remainder having an increased alveolar-arterial gradient. Peripheral white blood cell (WBC) counts are usually elevated with a neutrophilic predominance. In contrast to patients with CEP, patients with AEP typically do not have peripheral blood eosinophilia, although some patients have been reported with peripheral blood eosinophilia either initially or during the course of their illness (151, 152, 156). Pulmonary function testing in nonintubated patients has demonstrated the presence of either obstructive or restrictive defects with a decrease in diffusion capacity being seen in all patients (149, 151, 153).

Most patients present with bilateral infiltrates on chest radiograph, although two patients have presented with unilateral findings (154). In their review of 15 patients, Pope-Harman and coworkers found that 27% had interstitial infiltrates, 27% had alveolar infiltrates, and 46% had mixed alveolar and interstitial infiltrates (155). The progression of findings was usually one of increased interstitial markings followed by the development of alveolar infiltrates. Small pleural effusions were also common, being seen bilaterally in 67% of patients at some time during their illness. In an analysis of computed tomographic (CT) scans obtained on five patients, Cheon and coworkers found that patchy areas of ground glass opacity was the most common finding with areas of consolidation being seen in two patients (154).

BAL fluid obtained in AEP is most remarkable for an increase in eosinophil percentages. The average percentage of BAL eosinophils has been reported as 36.9 ± 2.5% and 40.3 ± 27.5% (152, 155). Transbronchial biopsies or open lung biopsies reveal infiltration of the alveoli and interstitium by eosinophils (151, 153, 156). The appearance of acute and organizing diffuse alveolar damage was also noted on open lung biopsies (156).

The diagnosis of AEP should be considered in all patients with acute respiratory failure accompanied by unexplained diffuse pulmonary infiltrates, and BAL should be performed immediately because it is the only method of establishing the diagnosis of AEP. Other conditions that may produce BAL eosinophilia include CEP, idiopathic pulmonary fibrosis, sarcoidosis, Churg-Strauss syndrome, bronchiolitis obliterans organizing pneumonia (BOOP), idiopathic hypereosinophilic syndrome, drug reactions, PCP, parasitic infections, and fungal infections (158, 159). In particular, fungal pneumonia can resemble AEP and BAL fluid should be stained and cultured for fungi when AEP is suspected. Fatal invasive Aspergillus infection in immunocompetent patients has been reported to cause BAL eosinophilia as has infection with Coccidiodes immitis (160, 161). CEP has rarely been reported to cause respiratory failure requiring mechanical ventilation (162, 163).

Corticosteroids are the cornerstone of treatment for AEP. A number of case series and reports have demonstrated the efficacy of corticosteroids (146, 149, 151, 152, 155, 157). In their original report and in their subsequent larger series of 15 patients, Allen and Davis treated patients with methylprednisolone at a dose of 60 mg to 125 mg every 6 h (146, 155). Badesch and colleagues treated their patient with 125 mg of methylprednisolone every 6 h (147). The pediatric patient treated by Buchheit and coworkers was given 2 mg/kg of methylprednisolone every 6 h (149). In Tazelaar and coworkers' series of nine patients, all patients were treated with high-dose corticosteroids, although the dose was not specified (156). Most patients demonstrated rapid improvement within 1 to 2 d with some patients showing improvement in several hours. All patients responded by 6 to 7 d. After resolution of hypoxemia and termination of mechanical ventilation, patients were then treated with 40 to 60 mg of prednisone a day which was then tapered over 4 to 12 wk (146, 147, 149, 155). Corticosteroids have been tapered as quickly as 10 d and 14 d in two patients (146, 157). Although spontaneous improvement without the use of corticosteroids has been reported (150), these patients likely had less severe disease. We recommend the use of corticosteroids for any patient with AEP requiring admission to the ICU and certainly for any patient requiring mechanical ventilation for AEP. Based on the available literature, we suggest the use of intravenous methylprednisolone at an initial dose of 60 mg to 125 mg every 6 h and then switching to oral prednisone at a dose of 40 to 60 mg a day after the patient has stabilized. This may then be tapered over a 2- to 6-wk time period depending on the clinical course.

Alveolar Hemorrhage Syndromes

Alveolar hemorrhage has been associated with a wide variety of disorders, including connective tissue diseases, vasculitides, coagulopathies, cardiac disease, and drug or toxin exposure. The alveolar hemorrhage syndromes are often misdiagnosed initially on presentation as pulmonary edema or pneumonia. To discuss all of these entities is beyond the scope of this article, thus we will focus on those conditions most likely to be encountered and in which benefit from corticosteroid therapy has been reported.

Systemic lupus erythematosus (SLE). Alveolar hemorrhage is a well-recognized complication of SLE with acute, life-threatening hemorrhage described in the literature. The incidence of massive pulmonary hemorrhage in SLE is unknown, but Marino and Pertschuk noted that of their 140 patients followed for 2 yr, three (2%) developed diffuse alveolar hemorrhage (164). Diffuse alveolar hemorrhage may be the presenting manifestation of SLE (165). Mortality greater than 50% has been reported with alveolar hemorrhage at any time during the course of SLE (164, 166). In a recent series by Zamora and coworkers of 15 patients with 19 episodes of diffuse alveolar hemorrhage, 13 of whom required mechanical ventilation, mortality was 42% (169). Schwab and coworkers, however, reported a mortality of 25% (1 of 8) despite four patients requiring mechanical ventilation (170).

No prospective or controlled trials assessing therapy have been performed, although high-dose corticosteroids have been advocated by many investigators (171, 172). In the series by Zamora and coworkers, 18 of the 19 episodes were treated with methylprednisolone at doses ranging from 500 to 2,000 mg/d for 2 to 6 d followed by a taper (169). Cyclophosphamide was used in 10 episodes and plasmapheresis in 12 episodes, in addition to corticosteroids. In Schwab's series of eight patients, methylprednisolone 1,000 mg/d was given for 3 d with cyclophosphamide in five patients (170). Given these data and observations with treatment for other causes of alveolar hemorrhage, we would recommend a course of pulse methylprednisolone at a dose of 1,000 mg/d followed by a taper based on clinical course. It is beyond the scope of this article to discuss the merits of treatment with cyclophosphamide and plasmapheresis.

Wegener's granulomatosis (WG). Although the lung is the most common organ system involved in WG (173), diffuse alveolar hemorrhage is uncommon (174). Between 5 and 15% of patients presenting with diffuse alveolar hemorrhage will have WG as the underlying condition (174), and in patients with WG who develop alveolar hemorrhage, alveolar hemorrhage is often the initial presentation of their disease (175, 176). Patients with WG presenting with alveolar hemorrhage have a more fulminant course and a higher mortality compared with patients with a more typical presentation (175, 177). Acute renal failure is often associated with alveolar hemorrhage in these patients.

Corticosteroids alone are not sufficient therapy for WG (173, 183). The current standard treatment of WG, developed at the NIH, consists of prednisone 1 mg/kg/d and cyclophosphamide 2 mg/kg/d based on ideal body weight (173, 183). After 4 weeks, the prednisone is tapered over 1 to 2 months to 60 mg every other day. In patients with fulminant disease, 2 mg/ kg/d of prednisone and 3 to 5 mg/kg/d of cyclophosphamide have been used. Some have suggested the use of pulse methylprednisolone at a dose of 1,000 mg/d for 3 d for life-threatening alveolar hemorrhage due to WG (184, 185). We concur with these recommendations. Other immunosuppressives have been used but the discussion of these therapies is beyond the scope of this article.

Microscopic polyangiitis (MPA). MPA is a systemic necrotizing vasculitis that clinically and histologically affects small vessels (capillaries, venules, or arterioles) without formation of granulomas, as is seen with WG. Like WG, MPA is an antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis. MPA has the same spectrum of manifestations as WG and is thought to be the most common cause of the pulmonary-renal syndrome (186). A number of patients have been described with diffuse alveolar hemorrhage due to MPA (187- 189). Pulmonary capillaritis is thought to be common with MPA. In their series of 18 patients with MPA, Gaudin and coworkers found that 11 (61%) had capillaritis and 12 (67%) had acute alveolar hemorrhage (190). In the series of 34 patients with MPA described by Savage and coworkers, 10 (29%) had pulmonary hemorrhage (191). Respiratory failure requiring mechanical ventilation has been reported (188, 191).

For treatment of MPA, Savage and coworkers used a regimen of prednisolone 60 mg/d, and cyclophosphamide 3 mg/kg/d as initial therapy (191). Azathioprine at a dose of 3 mg/kg/d was used as additional therapy in 17 patients, and instead of cyclophosphamide in three patients. Plasma exchange was also performed in 18 patients. The success with this regimen was 79%. The prednisolone dose was reduced at weekly intervals to 20 mg/d by 4 wk and then to a maintenance dose of 5 to 10 mg/d for 1 yr or longer. Cyclophosphamide was given for 8 wk and the azathioprine was continued for 1 yr or longer. Relapses were observed in 12 patients. Treatment similar to that for fulminant WG with pulse methylprednisolone and cyclophosphamide have been advocated by others (186, 192). Given the lack of available data and clinical similarities between MPA and WG, such recommendations seem appropriate.

Anti-glomerular basement membrane disease (Goodpasture's syndrome). Pulmonary involvement by anti-glomerular basement membrane (anti-GBM) disease ranges from 60 to 80% (184). Virtually all patients with anti-GBM disease will have evidence of glomerulonephritis with microscopic hematuria or proteinuria although frank renal failure may be present in only 50% (193). Glomerulonephritis alone may be seen in 20 to 40% of patients, whereas anti-GBM disease limited to the lung alone occurs in less than 10% of patients (194). Patients may exhibit hemoptysis ranging from blood-streaked sputum to massive life-threatening hemorrhage. A small number of patients may require mechanical ventilation for severe alveolar hemorrhage.

Pulse methylprednisolone at a dose of 1,000 mg/d for 1 to 3 d has been reported to be effective in controlling alveolar hemorrhage in patients with anti-GBM disease (195, 196). However, corticosteroids as monotherapy have not changed overall mortality owing to their inability to prevent progression to end-stage renal disease (194, 197). Combination therapy with cytotoxic agents and corticosteroids (prednisone 1 to 2 mg/kg/d) coupled with the use of plasmapheresis has been reported to achieve the best results (195, 197, 198). We concur that pulse methylprednisolone should be used for patients with life-threatening alveolar hemorrhage caused by anti-GBM disease.

Bone marrow transplantation (BMT). Diffuse alveolar hemorrhage has been reported in patients undergoing autologous and allogeneic BMT. In a study of 141 consecutive patients undergoing autologous BMT, Robbins and coworkers found that 29 (21%) patients developed alveolar hemorrhage (199). The median time of onset of alveolar hemorrhage was 12 d with a range of 7 to 40 d after transplantation. Two patients developed alveolar hemorrhage before transplantation. Mortality was 79% in their series. In an analysis of 77 patients treated with autologous BMT, Jules-Elysee and coworkers reported that 10 (13%) cases of acute respiratory failure caused by diffuse alveolar hemorrhage occurred with a mortality of 100% (200). Crilley and coworkers found that of 84 patients undergoing allogeneic BMT, five (6%) patients died of alveolar hemorrhage (201). In a review of 111 open lung biopsies of 109 BMT recipients with diffuse infiltrates, Crawford and coworkers noted nine (8%) cases with extensive alveolar hemorrhage (202). In a postmortem study of 47 patients who had received allogeneic BMTs and died of pulmonary complications, Agusti and coworkers noted that 11 (23%) patients had evidence of diffuse alveolar hemorrhage (203). In a postmortem study of 21 pediatric patients who developed hypoxemic respiratory failure, Bojko and coworkers reported that two patients (10%) had pulmonary hemorrhage (204).

Administration of corticosteroids to patients with diffuse alveolar hemorrhage after BMT may improve outcome. Chao and coworkers reported successfully treating four patients with alveolar hemorrhage after autologous BMT, two of whom required mechanical ventilation (205). Patients were treated with methylprednisolone 1,000 mg/d for 3 d, then 500 mg/d for 3 d, then 250 mg/d, followed by 60 mg/d with a taper over 2 mo. In a retrospective analysis, Metcalf and coworkers studied 65 episodes of alveolar hemorrhage in 63 of 603 consecutive patients who had undergone BMT (206). Patients were divided into three groups according to the therapy they received. Twelve patients received supportive care alone, 10 patients received low-dose corticosteroids (30 mg/d or less of methylprednisolone or its equivalent), and 43 patients received high-dose corticosteroids (more than 30 mg/d of methylprednisolone or its equivalent). The need for endotracheal intubation was 100% in the group that did not receive steroids, 80% in the low-dose steroid group, and 45% in the high-dose steroid group. Mortality was 92% in the nonsteroid group, 90% in the low-dose steroid group, and 67% in the high-dose steroid group. In the high-dose steroid group, most patients were treated with 125 to 250 mg of methylprednisolone every 6 h for 4 to 5 d followed by a taper over 2 to 4 wk based on clinical improvement. No increased risk of infection was noted between the steroid- and nonsteroid-treated groups.

In the absence of other data, and if the presence of infection has been sufficiently excluded, we recommend treatment for BMT-associated alveolar hemorrhage with methylprednisolone at a dose of 125 to 250 mg every 6 h for 4 to 5 d followed by a taper over 2 to 4 wk.

Acute Lupus Pneumonitis

The syndrome of acute lupus pneumonitis occurs in 1 to 4% of patients with SLE (171). Patients typically present with abrupt onset of dyspnea, fever, cough, and pleuritic chest pain. Lupus pneumonitis may be the presenting manifestation in some patients. Matthay and coworkers described 12 patients with acute lupus pneumonitis, representing 12% of their SLE patients who were hospitalized during a 6-yr period (207). Mortality was 50% in their series despite treatment with corticosteroids and the addition of azathioprine in some patients. Other fulminant cases of acute lupus pneumonitis have been reported (208).

Little data exists for the optimal treatment of acute lupus pneumonitis. In their series of 12 patients, Matthay and coworkers specified the dose of corticosteroids in only three patients, which was 100 mg/d of prednisone in two patients and 400 mg/d of hydrocortisone in one patient. Azathioprine at a dose of 2 mg/kg/d was used in seven patients (207). Inoue and coworkers successfully treated two patients with corticosteroids alone, one with 80 mg/d of prednisone and one with 200 mg/d of prednisone (213). In a patient requiring mechanical ventilation, Domingo-Pedrol and coworkers reported successful treatment with 1,000 mg/d of methylprednisolone for 5 d followed by 60 mg/d of prednisone (212). Freter and coworkers also used pulse methylprednisolone at a dose of 750 mg/d for 3 d followed by prednisone 60 mg/d to successfully treat a patient with fulminant lupus pneumonitis (211). Cyclophosphamide and plasmapheresis have also been used in conjunction with methylprednisolone 250 mg every 6 h in a patient who was failing corticosteroid therapy alone (210).

Based on the case series and data available for alveolar hemorrhage caused by SLE, we recommend initial treatment with 1 mg/kg/d of prednisone or equivalent. Pulse methylprednisolone of 1,000 mg/d for 3 d followed by 1 mg/kg/d of prednisone may be of benefit in patients with life-threatening lupus pneumonitis. The adjunctive use of cytotoxic agents and plasmapheresis has been advocated by some investigators, although their role is undefined.

Bronchiolitis Obliterans Organizing Pneumonia (BOOP) (Cryptogenic Organizing Pneumonia)

The clinical syndrome of idiopathic bronchiolitis obliterans organizing pneumonia (BOOP), also known as cryptogenic organizing pneumonitis (COP), was initially described in eight patients by Davison and coworkers in 1983 (214) and in 50 patients by Epler and coworkers in 1985 (215). BOOP has also been associated with a variety of conditions including infections, drugs and toxins, irradiation, organ and bone marrow transplantation, collagen vascular diseases, malignancies, and inflammatory bowel disease (216, 217). Patients with BOOP typically have a subacute presentation with dyspnea, cough, fever, malaise, weight loss, and a flulike syndrome (215, 218- 220). On chest radiograph multiple patchy alveolar opacities, diffuse interstitial infiltrates, or a focal pulmonary opacity may be seen. The clinical presentations of idiopathic BOOP and secondary BOOP are similar (220).

The overall prognosis for BOOP is good, however progression to death occurs in 6 to 15% of cases (215, 218, 221); patients with BOOP secondary to connective tissue diseases may have a worse outcome (215, 220). Fulminant and life-threatening BOOP has been reported. Cohen and coworkers described 10 patients with rapidly progressive BOOP and severe respiratory failure requiring mechanical ventilation in nine patients (224). Of these 10 patients, seven died despite aggressive therapy with corticosteroids and the use of cytotoxic therapy in four. The doses of corticosteroids used were not specified. Nizami and coworkers reported five patients with life-threatening BOOP causing hypoxemic respiratory failure (225). Four patients required mechanical ventilation, and two subsequently died. All patients were treated with corticosteroids, although the dose was reported for only one patient (methylprednisolone 80 mg every 12 h). Fulminant respiratory failure in a patient with BOOP unsuccessfully treated with 250 mg of methylprednisolone every 6 h for 3 d followed by 60 mg/d was described by Iannuzzi and coworkers (226). Schwarz, however, successfully treated a patient requiring mechanical ventilation with 2 g/d of methylprednisolone for 5 d followed by an unspecified taper (227). Bellomo and coworkers also reported successful treatment of two patients having severe hypoxemia with 50 to 60 mg/d of prednisolone (228).

Corticosteroids are considered the treatment of choice for BOOP although the ideal dose and duration of treatment have not been clearly defined. Given the lack of data regarding the treatment of life-threatening BOOP, extrapolation from treatment of non-ICU patients is necessary. A number of case series have reported successful treatment with corticosteroids (214, 215, 217, 219, 229). Doses ranged from 20 mg/d to 1 mg/ kg/d of prednisone. Patients were treated for several months. Relapses were noted when therapy was discontinued or steroids were rapidly tapered during the first months of treatment. King and Mortenson have recommended a dose of prednisone of 1.5 mg/kg/d (using ideal body weight) not to exceed 100 mg/d for 4 to 8 wk followed by a dose of 0.5 to 1.0 mg/kg/d for 4 to 6 wk (222). After 3 to 6 mo, if the patient's condition is stable or improving, the prednisone may be gradually tapered. For rapidly progressive, severe cases of BOOP they recommend 250 mg of methylprednisolone intravenously every 6 h for 3 to 5 d followed by oral prednisone. Cordier has recommended a regimen of methylprednisolone 60 mg every 12 h for 3 to 5 d followed by daily prednisone at a dose of 1 mg/ kg (216). After clinical and radiographic improvement the dose is decreased to 0.5 mg/kg/d for 1 to 2 mo followed by a gradual taper. Although spontaneous improvement has been noted in some patients (218, 219, 223), for patients with moderate to severe BOOP we recommend treatment with corticosteroids. Based on the current case series, we would suggest an initial dose of 1 mg/kg/d of prednisone or equivalent, not to exceed 100 mg/d. After 4 to 8 wk, if clinical response is seen, this may be decreased to 0.5 mg/kg/d for 4 to 8 wk followed by a gradual taper. Patients should probably be treated for at least 6 mo, although some patients may require 12 mo of therapy. In patients with life-threatening BOOP, an initial pulse of 1,000 mg/d of methylprednisolone for 3 to 5 d may be appropriate.

Radiation Pneumonitis

Radiation lung injury presents as two distinct clinical syndromes, radiation pneumonitis and radiation fibrosis. This section will focus on the former as it may result in respiratory failure. The estimated incidence of symptomatic radiation pneumonitis is 7 to 8%, although the percentage of patients who develop radiologic changes is substantially higher, averaging 43% across several studies (230, 231). The typical time course for the onset of radiation pneumonitis is 2 to 3 mo after completion of radiation therapy, although some patients develop symptoms as late as 6 mo (232). Presentations as early as 1 to 2 wk after receiving radiation therapy have been reported (233), usually in association with previous chemotherapy or high-dose, short-course radiotherapy. The cardinal symptom of radiation pneumonitis is dyspnea, which can vary from mild to severe. Patients may also have cough and fever, thus simulating an infectious process. In a small number of patients, the radiation pneumonitis may produce acute respiratory failure requiring mechanical ventilation, and in some patients, fatal respiratory insufficiency and acute cor pulmonale (236). The development of ARDS with diffuse bilateral infiltrates following limited thoracic irradiation has also been reported (234, 239).

Corticosteroids are the mainstay for treatment of radiation pneumonitis, although data from large human studies are lacking. However, a number of animal models have shown efficacy. In a mouse model, Philips and coworkers demonstrated that prednisolone (1 mg/kg) given at the time of irradiation reduced mortality, as did administration when animal deaths began to appear in another cohort group (100 to 160 d) (240). Gross found that methylprednisolone given 80 d postirradiation increased survival in mice; and when the steroid was later discontinued, mortality increased to that of the control group (241). These findings were confirmed by Gross and coworkers in a subsequent study (242). In a 30 to 40 kg sheep model of radiation pneumonitis, Loyd and coworkers observed that treatment with 1 g of methylprednisolone every 6 h prevented acute lung dysfunction as measured by increases in lung lymph protein clearance, mean pulmonary artery pressures, lung lymph thromboxane concentrations, and arterial oxygen tensions (243).

Although there are no controlled clinical trials in humans, there is some evidence that corticosteroids are beneficial in treating radiation pneumonitis. Abrupt discontinuation of corticosteroids in patients receiving corticosteroids as part of a chemotherapeutic regimen was reported to precipitate severe radiation pneumonitis in some patients (244). When seven of 10 of these patients resumed prednisone, at doses of 20 mg to 120 mg daily, all showed clinical improvement. Rubin and Casarett summarized eight studies in humans published before 1960 (247). In patients treated at the onset of acute radiation pneumonitis, seven of nine (78%) patients responded as opposed to seven of 13 (54%) who were treated after radiation pneumonitis had been established. Four of these studies demonstrated no benefit with prophylactic use of corticosteroids. Yamada and coworkers reported that 16 of 17 patients with moderate to severe radiation pneumonitis improved with steroid therapy, although doses were not provided (248). Other case reports and case series have reported benefit with corticosteroids (233, 238, 249); however, some cases of severe radiation pneumonitis may be unresponsive to corticosteroids (234, 237).

Based on the available data, we recommend the use of corticosteroids for moderate or severe radiation pneumonitis. In the absence of clinical trials, we would suggest the use of at least 60 mg of prednisone per day or equivalent. Gross has suggested a dose range of 60 to 100 mg daily of prednisone (232). Following a satisfactory response, this dose should be decreased to 20 to 40 mg a day followed by a gradual taper over several weeks to prevent recurrence.

Miliary Tuberculosis

Miliary tuberculosis (TB) may occasionally lead to respiratory failure requiring ICU admission. A number of cases of ARDS have been reported in association with miliary TB (250). Miliary TB with respiratory failure has also been reported in the HIV-infected patient population (254, 255). A number of investigators have advocated the use of corticosteroids in this clinical setting. To our knowledge, only one controlled trial exists examining the utility of corticosteroids for miliary TB. Tongnian and coworkers treated 28 patients with isoniazid (INH), streptomycin (STM), and para-aminosalicylic acid (PAS) alone, whereas 27 patients were treated with prednisone in addition to antituberculous therapy (256). The corticosteroid regimen consisted of prednisone 10 mg four times a day for 1 wk, then 20 mg daily for 2 to 7 wk, followed by a gradual taper for a total course of 3 to 5 mo. Mortality was 7% in the corticosteroid group compared with 18% in the control group.

With the paucity of data, no firm recommendations for the use of corticosteroids in miliary TB can be given. We believe that corticosteroids should be given in conjunction with antituberculous chemotherapy in patients with miliary TB and hypoxemic respiratory failure or ARDS. The possibility of coexisting adrenal insufficiency in patients with disseminated TB should be kept in mind.

Pulmonary Toxicity Secondary to Drugs or Toxins

Bleomycin pneumonitis. Bleomycin is an antitumor antibiotic with activity against a variety of tumors including squamous cell carcinoma of the head and neck, cervix, and esophagus, as well as germ cell tumors, Hodgkin's lymphoma, and non-Hodgkin's lymphoma (257). Interstitial pneumonitis is the most serious adverse effect, leading to respiratory compromise and death in some patients. The incidence of bleomycin interstitial pneumonitis has ranged from 3 to 11% and the overall mortality from interstitial pneumonitis has been estimated to be between 1 and 2% (257). Mortality may be as high as 60% in patients with severe bleomycin toxicity (260).

There appears to be a critical point, at approximately 400 total units of bleomycin, above which the incidence of pulmonary toxicity increases (257). A number of cases with fatal pneumonitis have been reported in patients receiving lower doses, and doses as low as 60 units have caused toxicity (261, 262). The onset of bleomycin interstitial pneumonitis is usually insidious with progression to interstitial fibrosis; however, acute presentations as well as fulminant courses with rapid development of symptoms and acute respiratory failure