Invasive aspergillosis in patients with chronic obstructive pulmonary diseases

1Respiratory Epidemiology and Clinical Research Unit of the Montreal Chest Institute; 2The Jewish General Hospital, McGill University, Montreal, Quebec Correspondence: Dr Dick Menzies, Montreal Chest Institute, room K1.24, 3650 Street Urbain, Montreal, Quebec H2X 2P4. Telephone 514-934-1934, fax 514-483-2083, e-mail dick.menzies@mcgill.ca A Muquim, S Dial, D Menzies. Invasive aspergillosis in patients with chronic obstructive pulmonary diseases. Can Respir J 2005;12(4):199-204.

A spergillus species can cause a wide spectrum of pulmonary illnesses ranging from the saprophytic colonization of airways to a necrotizing form with vascular invasion, termed invasive pulmonary aspergillosis (IPA).Because the incidence of IPA has increased over the past 40 years, it is now one of the most common mycoses (1,2).For example, a European study (3) reported a 14-fold increase in its detection at autopsy between 1978 and 1992, and another study (4) reported that Aspergillus was found in 6.9% of all patients admitted to a medical intensive care unit (ICU).This condition has an important economic impact -ICU length of stay is almost doubled with invasive Aspergillus (5), and in the United States, each case has been estimated to cost US$65,000 (1).Because mortality is 50% to 100% even with appropriate therapy (1,2,(4)(5)(6)(7)(8)(9), the identification of risk factors to prevent its occurrence would be highly beneficial.
IPA is a well-recognized complication of immunosuppressive therapy following organ transplantation (1,2), chemotherapy for cancer (particularly for leukemias and lymphomas) (1,2) and corticosteroid therapy (10).Although case reports ©2005 Pulsus Group Inc.All rights reserved ORIGINAL ARTICLE and case series have been published (6,9,(11)(12)(13), chronic obstructive pulmonary disease (COPD) was not included as a high-risk condition for the development of IPA in recent published authoritative reviews (1,2) or guidelines (14).In a recent study (4), COPD was the underlying disease in 49% of proven or probable IPA in patients without malignancy (4).This suggests that IPA may be under-recognized in patients with COPD, and the factors contributing to its development in these patients have not been defined.
At the Montreal Chest Institute in Montreal, Quebec, IPA developed in 12 patients who were hospitalized for the management of acute respiratory failure from COPD.We hypothesized that treatment of these patients with high-dose corticosteroids and/or broad-spectrum antibiotics was contributing to the occurrence of IPA; therefore, we conducted a case control study to investigate this hypothesis.

Definition of cases
Cases were defined as patients with COPD who developed IPA while hospitalized between January 1994, and January 2001 (considered the 'index' hospitalization).IPA was defined following the guidelines of an international consensus committee (15): • confirmed IPA: histopathological evidence from biopsy or autopsy and a positive culture for Aspergillus species from sputum, tracheal aspirate or bronchoalveolar lavage (BAL).
• probable IPA: positive culture for Aspergillus species from sputum, tracheal aspirate or BAL and all of the following: new respiratory symptoms (cough, dyspnea, hemoptysis or pleuritic chest pain) and/or pleural rub; new undiagnosed radiographic infiltrate; and no clinical response to broadspectrum antibiotics and/or did respond to therapy with amphotericin B.

Definitions of control subjects
Control subjects were selected from patients with the same underlying pulmonary disease who were hospitalized within a year of the case's index hospitalization and who did not develop IPA.Records of consecutive patients admitted during a single month, selected randomly from the 12 months before or after the case's index hospitalization, were reviewed to identify control subjects.Months were selected randomly to assess seasonal differences in occurrence, but within a year to reduce potential confounding from changes in therapy over time.Control subjects with cystic fibrosis (CF) were identified from our institutional registry of all CF patients.Two 'colonized' control subjects/case had a positive sputum or BAL culture for Aspergillus species during their hospitalization.Control subjects without colonization had no positive culture (out of at least one sputum culture performed) for Aspergillus during their hospitalization or the preceding year.

Data gathering
Demographic, clinical and therapeutic information were gathered from hospital records.The diagnoses of COPD and bronchiolitis (as the major underlying obstructive respiratory disease) were taken from the primary discharge diagnosis made by the treating respirologist, and were based on chest x-rays, pulmonary function testing and clinical course.Diagnosis of CF was based on genetic testing.For all patients, a series of three chest radiographs, from admission, the approximate time of onset of IPA and two to seven days later were reviewed by two board certified chest specialists (SD and DM), without knowledge of case status or the dates of the radiographs.When there was a discrepancy between the two readings for a particular patient, a consensus reading was organized (by AM).Microbiological data on Aspergillus and therapy with antibiotics and corticosteroids (in milligram doses equivalent to prednisone) were reviewed in detail for period during the index hospitalization, one month and one to 12 months preceding the index hospitalization.For the patients with COPD who developed IPA, therapy was counted until the date when clinical deterioration related to the occurrence of IPA was first noted.Outcomes recorded included ICU admission, invasive or noninvasive mechanical ventilation, death and length of stay in the ICU and acute care service.

Data analysis
The patients with COPD who developed IPA (cases) were compared with all controls and with each set of controls separately; in addition, the two groups of controls were compared.Differences were tested for statistical significance using Wilcoxon rank sum tests for non-normally distributed variables, including all variables relating to corticosteroid and antibiotic therapy.For normally distributed continuous variables, which included the clinical and demographic characteristics of patients, Student's t tests were used, while χ 2 tests were used to test differences for categorical variables (16).Multivariate logistic regression provided adjusted estimates of the association of IPA and Aspergillus colonization with clinical characteristics and therapy (17).Due to the very small number of cases, stepwise backward regression was performed.Initial models included all factors significantly associated in univariate or bivariate analysis.Final models included only those variables that were significantly associated.All analyses were conducted using SAS, Version 8.02 for Windows (SAS Institute, USA).

RESULTS
Between January 1994 and January 2001, 14 possible cases and 56 matched controls were identified.However, on further review, one patient was reclassified as a control with Aspergillus colonization and a second patient had possible invasive Candidiasis, but no Aspergillus was ever identified.This left 12 cases with IPA, four of whom were confirmed at autopsy (three with COPD and one with CF), and eight probable cases that met the international diagnostic criteria outlined.Four of these probable cases died and the other four manifested slow improvement with therapy with intravenous amphotericin B.
Aspergillus flavus was isolated in one case and Aspergillus fumigatus was isolated in the remainder of the cases.Of the 10 cases with underlying COPD and one with bronchiolitis, no patient had documented bronchiectasis.The case with underlying CF was diagnosed to have extensive IPA at autopsy.Based on discharge diagnoses during the study period, 3267 patients with COPD, 195 with CF and 40 with bronchiolitis were hospitalized, resulting in estimated case rates of invasive aspergillosis of 3.1/1000, 5.0/1000 and 25.0/1000 admissions, respectively.There were no significant differences between the cases and all controls, nor between colonized and noncolonized controls in terms of the season of admission, room on the acute care ward or the day hospital where the patients received care.During the study period, there were no major construction or renovation projects in the hospital buildings.Limited environmental sampling was performed to assess mold contamination in the rooms and ventilation systems of the hospitals in 2001.No Aspergillus contamination was found.
As shown in Table 1, there were no significant differences in the clinical characteristics among the three patient groups, including cause and severity of illness at the time of the initial hospitalization.Onset of IPA occurred after five to 37 days in hospital (median 13 days, interquartile range 10 to 18 days).All four patients who developed IPA within 10 days of hospitalization had received oral prednisone (or equivalent) in the preceding month for an average of 18 days and an average total dose of 800 mg (of prednisone or equivalent).When admitted to hospital, eight patients with IPA had normal chest x-rays (considered to have exacerbation) and four had evidence of airspace infiltrates (pneumonia) which improved in the first few days.After a period of relative stability, the patients with IPA manifested sudden clinical deterioration, with the development of new, rapidly progressive radiographic infiltrates, accompanied by acute respiratory failure requiring ICU admission.Eight of the cases had multifocal airspace patterned opacification, two had lobar airspace disease and one had multinodular disease.Only one patient had cavitary infiltrates, which were multiple and small.
The most significant difference between the cases and either group of controls was the average daily corticosteroid dose (expressed as prednisone equivalent in milligrams) during the index hospitalization before the onset of IPA (Table 2).As seen in Figure 1, no cases occurred among patients who received an average of less than 80 mg per day.A total of 15 patients studied had received an average daily dose of more than 80 mg; 12 (80%) of these patients developed IPA after a median of 12 days; three controls received this dose for two to 28 days.The cases also received a significantly higher total dose of corticosteroids in the month before the index hospitalization, as well as more broad-spectrum antibiotics during hospitalization and in the preceding month.Compared with noncolonized controls, Aspergillus colonized controls received significantly higher total corticosteroid doses and more broad-spectrum antibiotics in the month preceding, as well as during the index hospitalization (Table 2).
In multivariate analysis, IPA was not associated with clinical characteristics but was associated with more antibiotics and higher dose corticosteroid therapy during the index hospitalization.Aspergillus colonization was also not associated with clinical characteristics but was associated with antibiotic therapy during the index hospitalization and corticosteroid therapy in the preceding month (Table 3).

DISCUSSION
In the present study, 12 cases of IPA occurred among hospitalized patients with COPD.When compared with 29 similar patients colonized with Aspergillus and 28 others with no Aspergillus, IPA was associated with very high doses of corticosteroid therapy and multiple broad-spectrum antibiotics.Aspergillus colonization was also associated with treatment of higher doses of corticosteroids and more broad-spectrum antibiotics, albeit to a lesser extent than for the patients with IPA.In animal studies, the development of fungal pneumonia has been related to the dose and duration of corticosteroids (17).In a recent study of patients without malignancy who developed proven or probable IPA (4), 94% had received corticosteroids but at unspecified doses.In two uncontrolled case series of IPA complicating prednisone therapy, 16 patients with COPD had received an average of 73 mg/day (8) and 39 patients "without major immune suppression" received 140 mg/day for an unspecified duration (9).Of the 24 COPD patients with IPA reported until 1998 (9), 10 (42%) had taken more than 80 mg per day, eight (33%) had taken prednisone for at least two months and only two had not taken any corticosteroids.In a more recent report, Agusti et al (18) described nosocomial fungal pneumonia in nine patients (six patients with Aspergillus) who had received at least 30 mg/day of corticosteroids for at least 30 days.With very prolonged corticosteroid therapy following bone marrow transplantation, IPA was uncommon in patients who received less than 0.25 mg/kg/day and increasingly common at doses above 0.5 mg/kg/day (2,10).
In the present study, all of the patients with IPA received an average daily dose of corticosteroids of 80 mg or more compared with 4% of all controls, resulting in a crude incidence of 4.3/100 patients who received similar doses.(The incidence by dose could be estimated because the controls were selected randomly from all patients with similar obstructive lung diseases).IPA was not observed in patients who received less than 80 mg/day.This is in contrast to earlier studies (2,10,18), where IPA was associated with a lower dose but longer therapy duration.From this, it seems likely that IPA is rare in patients with COPD, simply because they usually receive much shorter   recommend the use of much lower doses of corticosteroids than received by some of the patients with COPD who developed IPA, and practice in our institution has since changed to use much lower doses.These changes may help prevent IPA.Antibiotic therapy before or during the index hospitalization was also an important risk factor, as has been observed in earlier case series (6,8,9,20).The risk for IPA increased according to the number of antibiotics used.Because many broad spectrum antibiotics were used and because the small number of cases limited the power to detect significant associations, no single agent could be identified that was associated with IPA.
Although colonization with Aspergillus was first described in patients with COPD over 40 years ago (21), the incidence, risk factors (besides cigarette smoking [13]) and clinical significance remain undefined.Our observation that corticosteroid and antibiotic therapy of patients with Aspergillus colonization was between that of patients with IPA and patients without any Aspergillus, suggests that Aspergillus colonization may be a marker of moderate but potentially important immune suppression.In patients with obstructive lung diseases, Aspergillus colonization may be a precursor of IPA, as has been reported in patients receiving cancer or transplant therapy (22).
The occurrence of IPA in hospitalized patients has been linked with Aspergillus species contamination of hospital air handling systems (23), mechanical ventilator filters (12) or nearby construction projects (1).However, no environmental contamination or source has been identified in most series of IPA.In the current study, patients with IPA or colonized controls were unlikely to have acquired Aspergillus as a result of a nosocomial source.This is because there was no seasonal, temporal or spatial clustering of the cases or colonized controls.In the years reviewed, there were one to two cases of IPA each year.
There were a number of limitations in the present study.First, only four cases were histologically confirmed (all at autopsy).No patients had lung biopsies because they were judged to be too unstable at the time.However, the four patients who recovered had a well-documented clinical response to amphotericin B therapy after failure to respond to multiple broad spectrum antibiotics.Some diagnostic uncertainty is inevitable regarding the four patients who died without autopsy, although they did meet the criteria of the international consensus guidelines (3).The difficulties of antemortem diagnosis are well-known (6,9), related in part to the low sensitivity of sputum cultures (2).In an autopsy series of 22 patients with IPA (24), while still alive, only two patients (9%) had confirmed IPA, six patients (27%) met the criteria for probable IPA and the remainder were considered possible or unclassifiable (24).The small number of cases with IPA limited the statistical power, which was another weakness.However, despite the small number, highly significant differences in the corticosteroid and antibiotic therapy between cases and controls were detected in univariate and multivariate analyses, which is consistent with a very strong effect.
The strengths of the study included the use of a case-control design and matching of each case to four controls (25) to maximize the efficiency of the study.The inclusion of controls randomly selected from all other patients with similar obstructive lung diseases allowed the estimation of the overall incidence and by dose of corticosteroid therapy, plus the assessment of clinical and therapeutic risk factors.By including controls with Aspergillus colonization, the factors associated with invasive disease could be distinguished from factors simply associated with colonization.The requirement that controls had been hospitalized and had the same underlying pulmonary disease should have ensured reasonable matching of the severity of underlying, as well as acute, illnesses (26).

CONCLUSIONS
Therapy with high-dose corticosteroids with or without broad-spectrum antibiotics for as little as one to two weeks can result in Aspergillus colonization or IPA in patients with COPD.If a hospitalized patient with COPD develops severe  nosocomial pneumonia, IPA should be excluded, corticosteroid therapy re-evaluated and the institution of systemic antifungal therapy considered.

ACKNOWLEDGEMENTS:
The authors are grateful to the staff of the Montreal Chest Institute (respiratory intensive care and acute care units) for their assistance in patient care and data gathering, and the staff of the medical records department for assistance in data gathering.
Invasive aspergillosis in COPD Can Respir J Vol 12 No 4 May/June 2005 201 Muquim et al Can Respir J Vol 12 No 4 May/June 2005 202

Figure 1 )
Figure 1) Average daily dose of prednisone (or equivalent) before the development of invasive aspergillosis for patients with invasive pulmonary aspergillosis (cases) and during index hospitalization for controls Invasive aspergillosis in COPD Can Respir J Vol 12 No 4 May/June 2005 203

FUNDING:
The study was not directly funded, although Dr Muqim was a trainee in the McGill Respiratory training programme and Dr Menzies was the recipient of a Scientist award from the Canadian Institutes of Health Research.

TABLE 1
Personal and clinical characteristics, and hospital course for the three patient groups *P for comparison between groups, from χ 2 tests for categorical variables or Student's t tests for continuous variables: 1 -Comparison of patients with COPD (cases) who developed invasive pulmonary aspergillosis with controls with Aspergillus, 2 -Comparison of patients with COPD who developed invasive pulmonary aspergillosis with controls without Aspergillus, and 3 -Comparison of controls with Aspergillus to controls without Aspergillus; † Best result in one year preceding index hospitalization.CHF Congestive heart failure; FEV 1 Forced expiratory volume in 1 s; FVC Forced vital capacity; IQR Interquartile range; NS Not significant

TABLE 2
Comparison of corticosteroid and antibiotic treatment given to the three patient groups

TABLE 3
Adjusted odds of therapeutic factors with development of invasive pulmonary aspergillosis (IPA) or Aspergillus colonization (from multivariate logistic regression) Adjusted for therapy and age, sex Adjusted for selected aspects of therapy* and pack-years smoking †