Pulmonary Infiltrates, Eosinophilia, and Cardiomyopathy Following Corticosteroid Withdrawal in Patients With Asthma Receiving Zafirlukast FREE

ZAFIRLUKAST (Accolate) is a novel, potent, orally administered, competitive cysteinyl leukotriene type 1 (CysLT1) receptor antagonist approved in September 1996 for the treatment of mild to moderate asthma. In the clinical trials leading to approval by the US Food and Drug Administration, 6243 patients encompassing 2479 person-years of exposure received zafirlukast; the drug was well tolerated, causing adverse clinical events (headaches, infection, and nausea) and laboratory abnormalities (elevated liver function test results) at rates similar to those with placebo.¹ However, since its approval, zafirlukast has been widely used by a diverse population of asthma patients; an estimated 40000 patient-years of use accrued in the first 6 months of the drug's availability by prescription.² As use of the drug increases, adverse events that occur at low frequency or in populations not examined in clinical trials may become manifest. We describe 8 patients (1 of whom we describe in detail) with steroid-dependent asthma who, in the setting of asthma treatment with zafirlukast, were able to discontinue or taper oral corticosteroid treatment but subsequently developed a clinical syndrome consisting of pulmonary infiltrates, peripheral blood eosinophilia, and cardiomyopathy.

A 45-year-old woman with no history of cigarette smoking and a noncontributory medical history developed recurrent episodes of sinusitis and asthma at age 40 years. Despite therapy with inhaled β-agonists, inhaled steroids, and theophylline, control of her symptoms from the onset of her disease required the frequent use of systemic corticosteroids. Addition of zafirlukast to her treatment regimen in early November 1996 resulted in dramatic alleviation of her asthma. The patient had decreasing asthma symptoms over the ensuing 2 months and was able to discontinue oral corticosteroid treatment by early January 1997. At that time she reported normal peak flow rates and fewer respiratory symptoms than she had had since the onset of her asthma. Approximately 2 weeks after discontinuing oral steroid therapy, she noted a mild, raised erythematous exanthem on her forearms and a large urticaric wheal on her right flank that became increasingly erythematous and hyperpigmented. Over the next few days she developed constitutional symptoms, with nausea, anorexia, fever, and loose stools. When a complete blood cell count revealed leukocytosis of 18.8×10⁹/L with 0.36 eosinophils, zafirlukast therapy was stopped and treatment with high-dose oral corticosteroids was begun.

Over the ensuing 3 days the patient developed dyspnea, a temperature of 38.7°C, a heart rate of 120 beats per minute, a respiratory rate of 24/min, and an oxygen saturation of 90% while breathing air. Results of her chest examination were notable for diffuse wheezes and a prolonged expiratory phase; her skin examination revealed confluent erythema on the chest and back, erythematous papules on the chest, a vesiculopustular lesion on the elbow, and a livedolike blanchable area of erythema on the right thigh. She had a white blood cell count of 25.7×10⁹/L, with an eosinophil fraction of 0.37, normal blood urea nitrogen and creatinine values with otherwise normal blood chemistry values, normal urinalysis results, and a chest x-ray film that revealed patchy diffuse alveolar infiltrates. Results of an antineutrophil cytoplasmic antibody test were positive with a perinuclear pattern, and the serum IgE level was 5.04 mg/L. Results of an assay for strongyloides antibody were negative.

A skin biopsy revealed a superficial perivascular lymphocytic infiltrate with eosinophils as well as epidermal and dermal necrosis with eosinophils and thrombi in superficial blood vessels. An echocardiogram revealed global biventricular hypokinesis with a left ventricular ejection fraction of 0.35 to 0.40. Unilateral foot drop suggestive of mononeuritis multiplex developed, as did angiographic evidence of bilateral lower-extremity arterial occlusion. A thoracoscopic lung biopsy showed organizing pneumonitis with prominent fibrinous exudate and granuloma formation, multifocal necrosis, and necrotizing vasculitis with organizing thrombi inside the pulmonary vessels.

Treatment with cyclophosphamide in addition to systemic corticosteroids was begun. Within 2 weeks, the patient's condition had improved in terms of symptoms, radiography, echocardiography, and pulmonary function testing. After 6 weeks of treatment with oral prednisone (60 mg/d) and cyclophosphamide (100 mg/d), a reduction in corticosteroid dose to 40 mg/d was accompanied by increased blood eosinophilia and deterioration of pulmonary function. The prednisone dose was restored to 60 mg/d, and the peripheral blood eosinophilia resolved.

Subsequent to the presentation of this case, we learned of 3 additional cases following a presentation at a national meeting, and through postmarketing surveillance personnel at Zeneca Pharmaceuticals, Wilmington, Del, we learned of 4 additional patients who, after starting treatment with zafirlukast, were able to reduce their required doses of oral corticosteroids but subsequently developed pulmonary infiltrates, eosinophilia, cardiomyopathy, and a variety of findings consistent with vasculitis (Table 1). The cases were spread over the 48 contiguous states and culture results were all negative, making an infectious process unlikely. Importantly, most of the findings improved significantly (although not necessarily completely) on reinitiation of oral corticosteroid treatment or on the addition of cyclophosphamide.

A prominent feature of all these cases was a markedly elevated peripheral blood eosinophil count. As 6 of the 8 patients in this series had biopsy evidence of eosinophilic tissue infiltration (Table 1) and given the known pathobiologic potential of activated eosinophils to cause the diverse clinical findings that were manifest in these patients,³ it seems likely that this syndrome represents a disorder of eosinophil recruitment and dysfunction. In such syndromes, eosinophils may mediate both airway obstruction (recognized as asthma), via synthesis of bronchoconstrictor mediators, such as the leukotrienes,⁴ and organ dysfunction, via the release of toxic cationic proteins.⁵ In the cases described herein, treatment with zafirlukast probably attenuated the airway obstruction usually associated with these syndromes and thereby resulted in an isolated improvement in the asthmatic component of the disease. We speculate that with the control of airway obstruction, the organ dysfunction arising from eosinophilic infiltration, previously masked by corticosteroid treatment given under the presumption of asthma as the primary abnormality, came to dominate the clinical picture. In this regard, while all 8 patients had findings sufficient for the clinical diagnosis of the Churg-Strauss syndrome (also known as allergic granulomatous angiitis, a rare vasculitis of unknown etiology characterized by moderate to severe asthma, peripheral blood eosinophilia [>0.10], mononeuropathy or polyneuropathy, migratory or transient pulmonary infiltrates, paranasal sinus abnormality, and the presence of extravascular eosinophils⁶), their presentation was atypical of this entity in that all patients had acute dilated cardiomyopathy. We speculate that this and other aspects of the presentation that are atypical for the Churg-Strauss syndrome may reflect the altered biological actions of the infiltrating eosinophils in the presence of leukotriene receptor antagonism.

What is responsible for eosinophil recruitment and activation in these patients? One possibility is that each patient would have developed the syndrome even without zafirlukast treatment. Although we cannot dismiss this possibility, with the detection of this unusual cluster of cases of eosinophilia, pulmonary infiltrates, and cardiomyopathy due to the increased vigilance known to accompany the introduction of a new drug, the incidence of such a syndrome would be extremely high. For example, if the patients in this series are considered to have the Churg-Strauss syndrome, then given the number of patients exposed over this period (about 40000 patient-years), these 8 individuals represent at least 4 times the estimated incidence of the syndrome.⁷

Another possible explanation for the findings in these patients is an allergic reaction to zafirlukast prompting a vasculitic response. The lack of such episodes in the more than 6000 asthma patients exposed to the drug during clinical trials does not rule out this possibility, given the overall low incidence of the syndrome. Furthermore, since patients receiving zafirlukast treatment are asthmatic and therefore may have an allergic diathesis, some of these cases may in fact be drug reactions. However, granulomatous angiitis associated with particular drugs is a rare occurrence: only 8 other cases (due to chlorothiazide,⁸ allopurinol,⁹ glibenclamide,¹⁰ diphenylhydantoin,^11- 12 carbamazepine,¹³ and quinine¹⁴) have been reported in the literature. Furthermore, 2 of the patients in our series had incipient but unappreciated manifestations consistent with Churg-Strauss syndrome before using zafirlukast (patients 2 and 4). Thus, drug-induced vasculitis is a possible but not probable explanation. Another inciting factor could have been treatment with inhaled steroids, medications that were used by all the patients in our series. However, there have been no reported clusters of cases in the literature of similar syndromes temporally associated with or coincident with the introduction of any particular inhaled steroid moiety.

An alternative hypothesis that merits serious consideration is that the patients described herein all suffered from a primary eosinophilic infiltrative disorder in which airway obstruction was such a prominent feature that their disease had been clinically recognized as moderate to severe asthma. For these patients, use of corticosteroids was the only viable therapeutic option until zafirlukast became available. Thus, we speculate that introduction of zafirlukast facilitated a reduction of the corticosteroid dose required to control the airway obstructive component of the syndrome, allowing unmasking of an underlying eosinophilic infiltrative process. Such a mechanism has been proposed previously for formes frustes cases of Churg-Strauss syndrome, and one could easily classify these cases under such a rubric; indeed, such a label could explain the exquisite corticosteroid responsiveness of this cohort of patients.¹⁵ Evidence for this mechanism derives from the fact that all of the people who developed the syndrome had been dependent on steroids, while no non–steroid-dependent patients developed similar symptoms. Indeed, we speculate that if this unmasking is the clinical mechanism involved, this syndrome may occur rarely in patients treated with any leukotriene receptor antagonist. Whether this syndrome will develop when the synthesis of leukotrienes is inhibited rather than when their actions are antagonized is not known.

Zafirlukast is a safe and effective medication for patients with mild to moderate asthma.^16- 17 However, our cases have important implications for physicians who prescribe zafirlukast to patients with corticosteroid-dependent asthma. While the exact pathogenesis of the constellation of findings in these patients remains unclear, vigilance for new symptoms and close monitoring for rising eosinophil counts, pulmonary infiltrates, and cardiomyopathy are recommended for all patients starting therapy with leukotriene antagonists who have previously required oral corticosteroid therapy to control their asthma. While all of the patients in this series had severe physiologic dysfunction on presentation, most of their clinical manifestations resolved in response to treatment with corticosteroids.¹⁸ Indeed, in patients who present with similar clinical findings, a rapid response to corticosteroid treatment may preclude an open lung biopsy, with its associated morbidity. Whether the treatment of choice for these patients is corticosteroids alone or corticosteroids plus cyclophosphamide (the treatment for classic Churg-Strauss syndrome) warrants further study.