Factors Associated with Overall and Attributable Mortality in Invasive Aspergillosis

Transcription

1 MAJOR ARTICLE Factors Associated with Overall and Attributable Mortality in Invasive Aspergillosis Yasmine Nivoix, 1 Michel Velten, 7 Valérie Letscher-Bru, 2 Alireza Moghaddam, 3 Shanti Natarajan-Amé, 3 Cécile Fohrer, 3 Bruno Lioure, 3 Karin Bilger, 3 Philippe Lutun, 4 Luc Marcellin, 5 Anne Launoy, 6 Guy Freys, 6 Jean-Pierre Bergerat, 3 and Raoul Herbrecht 3 1 Pharmacie, 2 Institut de Parasitologie et de Pathologie Tropicale, 3 Service d Hématologie et d Oncologie, 4 Service de Réanimation Médicale, 5 Service de Pathologie Générale, and 6 Service de Réanimation Chirurgicale, Hôpitaux Universitaires de Strasbourg, and 7 Laboratoire d Epidémiologie et de Santé Publique, Faculté de Médecine, Université Louis Pasteur, Strasbourg, France (See the editorial commentary by Kohno on pages ) Background. Invasive aspergillosis is associated with high death rates. Factors associated with increased mortality have not yet been identified in a large population of patients with various underlying conditions. Methods. We retrospectively reviewed 385 cases of suspected or documented aspergillosis that occurred during a 9-year period. We identified 289 episodes that fulfilled the criteria for possible, probable, or proven invasive aspergillosis according to the international definition criteria and that was treated with an anti-aspergillus active antifungal drug. Clinical and microbiological variables were analyzed for their effects on overall and attributable mortality. Significant variables in univariate analysis were introduced into a multivariate Cox model. Results. Twelve-week overall and disease-specific survival rates were 52.2% (95% confidence interval, 46.5% 57.9%) and 59.8% (95% confidence interval, 54.0% 65.4%), respectively. Receipt of allogeneic hematopoietic stem cell or solid-organ transplant, progression of underlying malignancy, prior respiratory disease, receipt of corticosteroid therapy, renal impairment, low monocyte counts, disseminated aspergillosis, diffuse pulmonary lesions, pleural effusion, and proven or probable (as opposed to possible) aspergillosis are predictors of increased overall mortality. Similar factors are also predictors of increased attributable mortality, with the following exceptions: pleural effusion and low monocyte counts have no impact, whereas neutropenia is associated with a higher attributable mortality. Conclusions. Identification of predictors of death helps in the identification of patients who could benefit from more-aggressive therapeutic strategies. Initiation of therapy at the stage of possible infection improves outcome, and this finding calls for the development of efficient preemptive strategies to fill the gap between empirical and directed therapy. Invasive aspergillosis (IA) affects severely immunocompromised patients [1 6]. Despite new treatment options, almost one-half of severely immunocompromised patients with IA do not respond to treatment, and 30% of such patients die within 12 weeks after treatment initiation [7, 8]. Little is known about prognostic factors associated Received 27 April 2008; accepted 6 July 2008; electronically published 22 September Presented in part: Meeting of the American Society of Hematology, Orlando, Florida, December 2006 (abstract 2852). Reprints or correspondence: Dr. Raoul Herbrecht, Dépt. d Hématologie et d Oncologie, Hôpital de Hautepierre, Strasbourg, France Clinical Infectious Diseases 2008; 47: by the Infectious Diseases Society of America. All rights reserved /2008/ $15.00 DOI: / with IA. Factors have been identified in selected subgroups of patients, but no information is available for the general population of patients who develop IA [9 15]. Here, we present a large, single-center study with an analysis of all relevant clinical, radiological, microbiological, and biological parameters. To assess the role of host factors as a potential cause of death, we separately analyzed the prognostic factors for overall and attributable mortality. PATIENTS AND METHODS Patients. We retrospectively identified all cases of IA occurring in adult patients hospitalized in the oncology-hematology department or occurring in patients with cancer who were hospitalized in intensive care units from 1 February 1997 through 30 April The charts were selected for the following reasons: (1) doc CID 2008:47 (1 November) Nivoix et al.

2 Figure 1. Kaplan-Meier probability of overall survival after initiation of treatment according to first-line therapy. AmB d, amphotericin B deoxycholate; L-AmB, lipid formulation of amphotericin B; Other, mostly itraconazole administered alone or in combination with AmB deoxycholate or a combination of an echinocandin plus a lipid formulation of amphotericin B; Vor, voriconazole. umented aspergillosis; (2) unexplained focal infection not responding to broad-spectrum antibiotics; (3) positive findings of direct microscopic examination or positive fungal culture results; (4) 1 serum sample collected for galactomannan detection, except in cases of routine surveillance; or (5) positive histopathological findings. This was a retrospective analysis, and there were no interventions with respect to the diagnostic procedures or the treatment. The study was approved by the local ethics committee (Comité d Ethique des Facultés de Médecine, d Odontologie, de Pharmacie et du Centre Hospitalier Régional de Strasbourg). The committee waived the need for informed consent. Therapy for invasive aspergillosis. First-line therapy evolved over time, from an amphotericin B (AmB) formulation to voriconazole, when the latter agent became available in France in October Overall, 104 patients were enrolled in various prospective clinical trials for primary or salvage therapy. In each of these cases, an ethical committee approved the protocol, and all patients gave informed consent. Definitions. Neutropenia was defined as a neutrophil count of!500 cells/ml. Proven, probable, or possible invasive fungal infections were defined according to the European Organisation for Research and Treatment of Cancer/Mycosis Study Group definition criteria, with the exception of the galactomannan antigen test [16]. On the basis of previous publications [17, 18], we accepted as positive a single value Degree of certainty was assessed using information available to the clinicians within 4 days after initiation of therapy. When a patient developed 2 episodes of IA, only the first episode was analyzed. Imaging review. All chest radiographs and CTs were reviewed by a physician who was experienced in the assessment of IA to analyze baseline images [7, 19, 20]. The extent of Figure 2. Kaplan-Meier probability of overall survival according to the period of therapy. Voriconazole largely replaced amphotericin B products after October 2002 for patients not included in clinical trials (hazard ratio, 1.56; 95% CI, ; P p.01). pulmonary lesions was categorized into localized (i.e., restricted to a single lobe) or diffuse (i.e., found in 11 lobe). Causes of death. Death was attributed to aspergillosis (1) in patients who did not respond to therapy (i.e., who had stable disease or disease progression) at time of death; (2) in patients with a partial response to therapy who died as the result of an acute event involving any of the sites of infection or of an unknown cause; and (3) in patients who died as a result of the toxicity of antifungal therapy. Data collection. The following data were collected: demographic data; underlying condition; date of diagnosis and status of cancer; anticancer and corticosteroid therapy received during the 3 preceding months; medical history; receipt of prior antimicrobial therapy, including agents that can induce falsepositive galactomannan test results [21, 22]; biological parameters, including leukocyte and differential cell counts, creatinine clearance level, C-reactive protein level, and fibrinogen level; radiological and clinical signs; microbiological data, including microscopy findings, culture results, and the results of tests to detect anti-aspergillus antibody and galactomannan antigen; Figure 3. Probability of all-cause mortality and mortality attributable to aspergillosis. Prognosis Factors in IA CID 2008:47 (1 November) 1177

3 Table 1. Univariate analysis of factors of prognostic significance for overall survival and disease-specific survival at 12 weeks among patients with invasive aspergillosis (IA). Variable No. of patients Disease-specific survival a (95% CI) P Overall survival (95% CI) P Main host factor Hematological malignancy ( ) ( ).006 Allogeneic HSCT or SOT ( ) 33.3 ( ) Other ( ) 56.5 ( ) Progression of underlying disease No ( )! ( )!.001 Yes ( ) 39.4 ( ) Concomitant diabetes mellitus No ( ) ( ).015 Yes ( ) 39.6 ( ) Prior noninfectious respiratory disease No ( ) ( ).025 Yes ( ) 40.3 ( ) Intubation at onset of IA No ( )! ( )!.001 Yes ( ) 21.4 ( ) Corticosteroid dosage on day 1 of IA!0.2 mg/kg ( )! ( )! mg/kg ( ) 23.3 ( ) Creatinine clearance level 60 ml/min ( )! ( )! ml/min ( ) 40.0 ( ) 0 29 ml/min ( ) 28.6 ( ) Neutrophil count cells/ml ( ) ( ) cells/ml ( ) 55.2 ( ) Monocyte count 0 99 cells/ml ( ) ( ) cells/ml ( ) 59.3 ( ) Lymphocyte count cells/ml ( ) ( ) cells/ml ( ) 59.8 ( ) Site of IA infection Lungs only ( )! ( )!.001 Other single organ Disseminated ( ) 21.1 ( ) Extent of pulmonary lesions Single pulmonary lobe involvement ( )! ( )!.001 Diffuse lung involvement ( 2 lobes) ( ) 39.4 ( ) No lung lesion ( ) 92.9 ( ) Pleural effusion Yes ( )! ( )!.001 No ( ) 68.9 ( ) First-line antifungal therapy Amphotericin B deoxycholate ( )! ( ).016 Lipid-amphotericin B ( ) 47.1 ( ) Voriconazole ( ) 69.4 ( ) Other antifungal treatment ( ) 49.0 ( ) Positive mycological finding b No ( ) ( ).013 Yes ( ) 46.2 ( ) (continued)

4 Table 1. (Continued.) Variable No. of patients Disease-specific survival a (95% CI) P Overall survival (95% CI) P Anti-Aspergillus antibody positive No ( ) ( ).060 Yes ( ) 60.0 ( ) NA ( ) 35.7 ( ) Galactomannan antigen positive No ( ) ( ).011 Yes ( ) 46.7 ( ) Degree of certainty of IA diagnosis Proven or probable ( ) ( ).009 Possible ( ) 61.7 ( ) NOTE. Only factors with a P value!.20 for at least 1 of the 2 end points are listed in this table. HSCT, hematopoietic stem cell transplantation; NA, not available; SOT, solid-organ transplantation. a Deaths not attributed to aspergillosis were censored. b By microscopy or culture. histopathological test results; anti-aspergillus therapy; and cause and date of death. Parameters for prognostic significance were abstracted from the list above. All parameters that were included in the analysis had to be available within 4 days after the onset of the first anti-aspergillus therapy. Statistical methods. End points of the analyses were 12- week overall survival and disease-specific survival. Overall survival evaluated the risk of death from all causes, and diseasespecific survival evaluated the risk of death from aspergillosis. Overall survival was measured from the date of diagnosis to the date of death, regardless of cause, censored at 12 weeks. Specific survival was measured from the date of diagnosis to the date of death from aspergillosis (patients dying from other causes were censored at the time of death). All assessments were updated on 1 July All survivors had a minimum follow-up period of 12 weeks. Univariate analysis was performed according to the method described by Kaplan and Meier [23]. Curves were compared with use of the log-rank test, and 95% CIs were computed according to the method described by Rothman [24]. All variables with a P value!.2 were introduced into a backward stepwise Cox regression model, allowing for interactions [25]. Hazard ratios were computed, as well as their 95% CIs. Categorical variables with g classes were recoded into g-1 dummy variables. Continuous variables were categorized and treated similarly. The statistical software used was SAS, version 9.1 (SAS Institute). P values!.05 were considered to be statistically significant. Because major progress in therapeutic methods had occurred over the study period as a result of the availability of voriconazole, we stratified the multivariate analysis by first-line therapy to produce a consistent analysis of the other variables, independent of the therapy applied [7]. RESULTS Case selection and patient characteristics. We identified 385 cases of suspected or documented IA over the 9-year study period. We excluded 82 cases in which, after an analysis of prognostic factors, the diagnosis was doubtful. Reasons for exclusion were colonization by Aspergillus species in the absence of any clinical or radiological sign or symptom of lower or upper respiratory tract infection (10 cases); false-positive serum galactomannan test results, mostly occurring in patients who received piperacillin-tazobactam (26 cases); noninvasive aspergillosis (10 cases); bacterial, viral, or non-aspergillus fungal pneumonia (19 cases); noninfectious cause (4 cases); and subsequent episode of IA (13 cases). Because we analyzed the impact of the factors present at onset of therapy, we also excluded 14 cases of IA that had not been treated. Absence of therapy was explained by a confounding cause of pulmonary signs and symptoms or by the death of the patient before results of diagnostic tests were obtained. Thus, 289 cases fulfilled selection criteria. Most patients had an underlying hematological malignancy, received a hematopoietic stem cell transplant (HSCT), or a had a solid tumor. Ten patients had a history of solid-organ transplantation, but all of them had a malignancy. Primary host condition included nonmalignant disease in 21 cases (7.3%), as follows: aplastic anemia in 3; AIDS in 3; rheumatoid arthritis in 3; chronic granulomatous disease in 2; drug-induced agranulocytosis in 2; monoclonal gammopathy of unknown significance combined with acute renal failure or diabetes mellitus in 2; and Prognosis Factors in IA CID 2008:47 (1 November) 1179

5 Table 2. Results of multivariate analysis of 12-week overall mortality for patients with invasive aspergillosis (IA) after stratification by treatment. Variable Reference category Hazard ratio (95% CI) P Main host factor Allogeneic HSCT or SOT Hematologic malignancy 1.7 (1 2.8).045 Other 1.6 ( ) Progression of underlying disease Yes No 4.2 ( )!.001 Prior noninfectious respiratory disease Yes No 1.7 ( ).009 Corticosteroid dosage on day 1 of IA 0.2 mg/kg!0.2 mg/kg 2.4 ( )!.001 Creatinine clearance level ml/min 60 ml/min 1.4 ( ).007!30 ml/min 2.1 ( ) Monocyte count 100 cells/ml 1100 cells/ml 1.6 ( ).015 Site of IA infection Disseminated Lungs only 3.3 ( ).001 Other single organ NA (NA) Extent of pulmonary lesions Diffuse lung involvement ( 2 lobes) Single pulmonary lobe involvement 2.3 ( ).007 No lesion 1.0 ( ) Pleural effusion Present Absent 1.6 ( ).030 Degree of certainty of IA diagnosis Possible Proven or probable 0.5 ( ).001 NOTE. Only statistically significant ( P!.05) predictors of overall survival are listed. HSCT, hematopoietic stem cell transplantation; NA, not available; SOT, solid-organ transplantation. alveolar proteinosis, autoimmune hemolytic anemia, pulmonary Langerhans cell histiocytosis, hypogammaglobulinemia, scleroderma, and celiac disease in a pregnant woman in 1 case each. Patients without neutropenia, impaired granulocyte function, prolonged corticosteroid therapy, or HIV infection had a proven diagnosis of IA. Seventy-seven (26.6%) of 289 cases occurred in patients with concomitant noninfectious lung disease, as follows: primary or secondary localization of the malignancy in 31 cases, chronic obstructive bronchopulmonary disease in 11, emphysema in 8, bronchiolitis obliterans due to graft-versus-host disease (GVHD) in 5, bronchiectasis in 3, pulmonary fibrosis in 3, tuberculosis sequels in 3, lung transplantation in 2, asthma in 2, and pulmonary Langerhans cell histiocytosis, diaphragm paralysis, inhaled bronchial foreign body, aspiration pneumonia, right pneumonectomy, silicosis, alveolar proteinosis, alveolar hemorrhage, and sarcoidosis in 1 case each. In 28 (9.7%) of the cases, patients were intubated and received mechanical ventilation before onset of aspergillosis. Aspergillosis was localized to the lungs in 258 cases, disseminated in 19 cases, and localized to another single organ in 12 cases (sinuses in 6 cases, kidney in 2, nose in 1, ear and temporal bone in 1, thyroid in 1, and toe in 1). In a majority of cases (178 cases; 61.6%), patients were treated primarily with an AmB formulation (deoxycholate AmB in 127 cases and a lipid formulation of AmB in 51 cases). Voriconazole was administered as first-line therapy in 62 cases (21.5%). In the 49 remaining cases, patients received another primary therapy, mainly based on itraconazole or a combination of 2 agents. Survival according to first-line therapy differed ( P p.016) between the various therapeutic strategies used over the 9-year period (figure 1). Similarly, we observed an effect of treatment period: patients who were treated before October 2002 had a 12-week overall survival rate of 47.5%, whereas patients who were treated after October 2002 had a 12-week overall survival rate of 60.4% ( P p.01) (figure 2). Because no modification occurred in diagnostic strategy, it is likely that this difference was related to the use of voriconazole after this date for most patients other than those who were enrolled in clinical trials. Univariate and multivariate analysis. A total of 138 patients died within 12 weeks after initiation of therapy. Of these deaths, 114 (82.6%) were attributed to aspergillosis. Overall survival and disease-specific survival at 12 weeks were 52.2% (95% CI, 46.5% 57.9%) and 59.8% (95% CI, 54.0% 65.4%), respectively (figure 3) CID 2008:47 (1 November) Nivoix et al.

6 Table 3. Results of multivariate analysis of 12-week attributable mortality for patients with invasive aspergillosis (IA) after stratification by treatment. Variable Reference category Hazard ratio (95% CI) P Main host factor Allogeneic HSCT or SOT Hematologic malignancy 1.9 ( ).005 Other 2.4 ( ) Progression of underlying disease Yes No 5.4 ( )!.001 Corticosteroid dosage on day 1 of IA 0.2 mg/kg!0.2 mg/kg 2.8 ( )!.001 Creatinine clearance level ml/min 60 ml/min 1.4 ( ).001!30 ml/min 2.8 ( ) Neutrophil count!500 cells/ml 500 cells/ml 1.8 ( ).007 Site of IA infection Disseminated Lungs only 3.2 ( ).002 Other single organ NA (NA) Extent of pulmonary lesions Diffuse lung involvement ( 2 lobes) Single pulmonary lobe involvement 3.3 ( ).002 No lesion 0.8 ( ) Degree of certainty of IA diagnosis Possible Proven or probable 0.5 ( ).010 NOTE. Only statistically significant ( P!.05) predictors of overall survival are listed. HSCT, hematopoietic stem cell transplantation; NA, not available; SOT, solid-organ transplantation. All variables with a P value of!.20 for at least 1 of the 2 end points in univariate analysis are listed in table 1. The following parameters did not reach a P value of!.20 for any of the end points in the univariate analysis: age, sex, time from diagnosis of cancer to onset of aspergillosis, receipt of anticancer chemotherapy or antibiotics during the preceding month, leukocyte count, duration of neutropenia before onset of aspergillosis, C-reactive protein level, fibrinogen level, maximum daily corticosteroid dose during the preceding 3 months, prior anti-aspergillus prophylaxis, and time from first clinical, radiological, or mycological sign of infection to initiation of therapy. Tables 2 and 3 list the factors with a prognostic significance in multivariate analysis for overall and attributable mortality, respectively. DISCUSSION Factors of prognostic significance in IA aspergillosis have been analyzed only in small series or in very specific populations of patients, such as recipients of HSCTs, patients with AIDS, or critically ill patients [9 15]. In our study, we reviewed all of the cases of IA that occurred in patients hospitalized in an oncology-hematology service or that occurred in patients with cancer who were hospitalized in intensive care units over a 9- year period, allowing us to collect a large number of cases associated with various underlying oncological or hematological conditions. To ensure exhaustivity, we reviewed the charts of all patients with suspected aspergillosis. Of these 385 cases, 82 were confidently excluded, either because another explanation was identified or because the case did not fulfill definition criteria [16, 26]. An additional 14 patients were excluded because they had not received anti-aspergillus therapy and, therefore, did not qualify for analysis of the impact on outcome of factors present at the onset of therapy. Underlying conditions in our patients included hematological malignancies, solid tumors, receipt of autologous and allogeneic HSCTs, solid-organ transplantation, HIV infection, prolonged corticosteroid therapy, and other immune deficiencies. These are fully representative of the conditions associated with the occurrence of IA: they represented 99% of the total number of patients in the 2 largest randomized clinical trials conducted during the previous decade [7, 27]. In addition, the international definition criteria for invasive fungal infections fully apply only to these patient populations, allowing us to classify these cases according to the consensus criteria [16, 26]. Our patient population is, therefore, representative of the patients at risk for IA and is well defined, because we used internationally accepted definition criteria. In 94 (32.5%) of our cases, the patient did not reach the level of probable or proven disease within 4 days after initiation of therapy. However, when the cases were classified at the end of the episode by integrating all available information (including autopsy data, if such data existed), 52 of the cases were Prognosis Factors in IA CID 2008:47 (1 November) 1181

7 upgraded to the level of probable or proven disease, suggesting that we were right to include these cases in the analysis. The remaining 42 cases remained at the level of possible IA according to the European Organisation for Research and Treatment of Cancer/Mycosis Study Group definition criteria, but no other infectious or noninfectious cause could be demonstrated. IA was the most likely diagnosis. Our patient selection appears to have been adequate and, interestingly, included a large proportion of patients who usually are not eligible for clinical trials. We demonstrate that survival is higher when therapy is started before microbiological confirmation of the disease, compared with when therapy is delayed until positive mycological or serological test results are obtained. This suggests that preemptive therapy of IA improves patient outcome. The 12-week overall survival rate (52.3%) was lower by nearly 20% for our patients, compared with the survival rate in the best arm of 2 recent comparative randomized studies, the comparative voriconazole trial (survival rate, 70.8% for patients treated with voriconazole) and the Ambiload study (survival rate, 72% for patients receiving 3 mg/kg/day of liposomal AmB) [7, 27]. This major difference can be explained by the use of AmB deoxycholate as the primary therapy for most patients treated until October 2002 (before voriconazole was available). The 12-week survival rate for AmB deoxycholate treated patients is known to be significantly lower (57.9% in a comparative trial), which is much closer to our results [7]. Another explanation is the absence of selection of our cases, compared with cases included in clinical trials. Two conditions that are usually excluded in clinical trials, severe renal impairment and/or intubation at the onset of therapy, were present in 40 cases (13.8%). Not surprisingly, allogeneic HSCT recipients had a worse outcome than did leukemic patients. Most of these transplant recipients had acute or chronic GVHD, a factor that we did not include in the multivariate analysis because of its interaction with the underlying condition of allogeneic HSCT in our analysis. GVHD has already been associated with a poor outcome in several studies [9, 11, 28]. Progressive underlying malignancy has also been associated with a lower overall survival and lower specific survival according to previous reports [12, 29]. The presence of a concomitant noninfectious respiratory disease is a significant predictor of poor overall survival, but it has no impact on specific survival, which emphasizes the interference of the host condition with outcome. Severe pretransplant pulmonary function impairment has been associated with a higher death rate among HSCT recipients [9]. We demonstrated that any pre-aspergillosis respiratory function impairment, irrespective of its severity, was critical in our patient population. Concomitant respiratory disease may also represent a major additional risk factor for aspergillosis in immunosuppressed patients, because it was present in more than one-quarter of our patients with IA. Renal function has always been considered to be critical in patients with invasive fungal infection because of the potential nephrotoxicity of AmB. We demonstrate that even a moderate decrease in creatinine clearance level is associated with a significant impact on the survival of patients with IA, independent of the primary antifungal therapy. Receipt of prior corticosteroid therapy has been consistently associated with poor outcome of allogeneic HSCT [9, 11, 30, 31]. We show that, in patients with various hematological malignancies or solid tumors, treatment with corticosteroids at a dose as low as 0.2 mg/kg/day at onset of aspergillosis is associated with higher mortality, independent of the underlying host condition. Previous studies have suggested that much higher doses of corticosteroids (at least 2 mg/kg/day) are required to have a positive impact on outcome [9, 11, 30, 31]. These studies involved HSCT recipients who received unusually high dosages of corticosteroids; therefore, the impact of administering lower dosages may be more difficult to assess for such patients, compared with the patients in our study. It is well accepted for many types of infections that neutropenic patients have a worse outcome if they do not recover from neutropenia. This was also highly significant for our patients, but this parameter was not introduced into the multivariate analysis, because this information was not available at the onset of therapy. Neutropenia at baseline has no impact on overall mortality, but it affects the attributable mortality. On the contrary, a low monocyte count is associated with increased overall mortality. This finding has already been reported by Cordonnier et al. [9] among HSCT recipients and has been confirmed by Upton et al. [30]. The extent of the infection, as determined by the presence of diffuse pulmonary involvement or dissemination of the infection to other organs, is associated with a major increase in the probability of death. Although multivisceral dissemination of the disease is a well-known cause of death among patients with IA, the impact of the extent of pulmonary infection has not yet been fully investigated. The presence of a halo sign has been associated with a better outcome in a univariate analysis [19]. One explanation for this may be that patients with a halo sign receive an earlier diagnosis and have earlier onset of therapy, compared with patients who have a nodule without a halo. This hypothesis is supported by the fact that the occurrence of a halo sign is associated with a very early and transient stage of IA [10, 19]. Similarly, Yeghen et al. [29] showed that patients with typical signs and symptoms associated with IA (e.g., a halo sign, a cavitation, or a sequestrum) have a better outcome, compared with patients who have diffuse lung involvement. In this study, we have shown that the extent of pulmonary lesions, defined according to a simple classification as localized (single 1182 CID 2008:47 (1 November) Nivoix et al.

8 pulmonary lobe involvement) or diffuse (involvement of 2 lobes), has an influence on both overall and attributable mortality. Presence of a pleural effusion was associated with a higher overall mortality rate but had no effect on attributable mortality. Upton et al. [30] reviewed cases from January 1990 through December 2004 and showed a significant decrease over time in mortality among patients with a diagnosis of IA after receipt of an HSCT. The 90-day probability of survival was 22% for the period from 1990 through 2001 and was 47% for the period from 2002 through This decrease in the death rate was attributed to changes in transplantation practices, including the use of nonmyeloablative regimens and receipt of peripheral stem cells; improved diagnostic procedures, including the introduction of the galactomannan test in 2003; and the widespread use of voriconazole after October We also show an improvement in outcome over time: the 12-week survival was 60.4% from October 2002 through the end of the study, compared with 47.5% before October Because all patients were screened for galactomannanand had routine CTs throughout the study period, the only significant change was the increased use of voriconazole monotherapy from October 2002 through the end of the study for all patients who were not included in a clinical trial. This study identifies several factors that are predictive of death among patients with IA, independent of the treatment given. They should be considered for stratification in future clinical trials that assesses the efficacy of new antifungal agents or for selecting subgroups of patients who might benefit from more-aggressive therapeutic strategies, such as combination antifungal therapy. In addition, initiation of therapy at the stage of possible infection, rather than after confirmation of the diagnosis, improves the outcome; this calls for a better definition of possible cases. Inclusion of well-defined possible cases into clinical trials should also be considered and could help in the development of more-efficient preemptive strategies, filling the current gap between empirical therapy and treatment of documented infection. Acknowledgments Financial support. Pfizer and Alsace Thérapie Génique et Cancer. Potential conflicts of interest. R.H. has been a consultant for Astellas, Gilead Sciences, Pfizer, Merck Sharp and Dohme, Novartis, and Schering- Plough; has received honoraria from Gilead Sciences, Pfizer, Merck Sharp and Dohme, Schering-Plough, and Zeneus; and has received a grant from Pfizer. A.M. has received a grant from Alsace Thérapie Génique et Cancer. All other authors: no conflicts. References 1. Denning DW, Marinus A, Cohen J, et al. An EORTC multicentre prospective survey of invasive aspergillosis in haematological patients: diagnosis and therapeutic outcome. EORTC Invasive Fungal Infections Cooperative Group. J Infect 1998; 37: Denning DW. Invasive aspergillosis. Clin Infect Dis 1998; 26: Marr KA, Carter RA, Boeckh M, Martin P, Corey L. Invasive aspergillosis in allogeneic stem cell transplant recipients: changes in epidemiology and risk factors. Blood 2002; 100: Marr KA, Carter RA, Crippa F, Wald A, Corey L. Epidemiology and outcome of mould infections in hematopoietic stem cell transplant recipients. Clin Infect Dis 2002; 34: Patterson TF, Kirkpatrick WR, White M, et al. Invasive aspergillosis: disease spectrum, treatment practices, and outcomes. I3 Aspergillus Study Group. Medicine (Baltimore) 2000; 79: Lin SJ, Schranz J, Teutsch SM. Aspergillosis case-fatality rate: systematic review of the literature. Clin Infect Dis 2001; 32: Herbrecht R, Denning DW, Patterson TF, et al. Voriconazole versus amphotericin B for primary therapy of invasive aspergillosis. N Engl JMed2002; 347: Cornely OA, Maertens J, Bresnik M, et al. Liposomal amphotericin B as initial therapy for invasive mold infection: a randomized trial comparing a high-loading dose regimen with standard dosing (AmBiLoad trial). Clin Infect Dis 2007; 44: Upton A, Kirby KA, Carpenter P, Boeckh M, Marr KA. Invasive aspergillosis following hematopoietic cell transplantation: outcomes and prognostic factors associated with mortality. Clin Infect Dis 2007;44: Caillot D, Casasnovas O, Bernard A, et al. Improved management of invasive pulmonary aspergillosis in neutropenic patients using early thoracic computed tomographic scan and surgery. J Clin Oncol 1997; 15: Ribaud P, Chastang C, Latge JP, et al. Survival and prognostic factors of invasive aspergillosis after allogeneic bone marrow transplantation. Clin Infect Dis 1999; 28: Ribrag V, Dreyfus F, Venot A, Leblong V, Lanore JJ, Varet B. Prognostic factors of invasive pulmonary aspergillosis in leukemic patients. Leuk Lymphoma 1993; 10: Subira M, Martino R, Franquet T, et al. Invasive pulmonary aspergillosis in patients with hematologic malignancies: survival and prognostic factors. Haematologica 2002; 87: Tumbarello M, Tacconelli E, Pagano L, et al. Comparative analysis of prognostic indicators of aspergillosis in haematological malignancies and HIV infection. J Infect 1997; 34: Vandewoude KH, Blot SI, Benoit D, Colardyn F, Vogelaers D. Invasive aspergillosis in critically ill patients: attributable mortality and excesses in length of ICU stay and ventilator dependence. J Hosp Infect 2004;56: Ascioglu S, Rex JH, de Pauw B, et al. Defining opportunistic invasive fungal infections in immunocompromised patients with cancer and hematopoietic stem cell transplants: an international consensus. Clin Infect Dis 2002; 34: Herbrecht R, Letscher-Bru V, Oprea C, et al. Aspergillus galactomannan detection in the diagnosis of invasive aspergillosis in cancer patients. J Clin Oncol 2002; 20: Marr KA, Balajee SA, McLaughlin L, Tabouret M, Bentsen C, Walsh TJ. Detection of galactomannan antigenemia by enzyme immunoassay for the diagnosis of invasive aspergillosis: variables that affect performance. J Infect Dis 2004; 190: Greene RE, Schlamm HT, Oestmann JW, et al. Imaging findings in acute invasive pulmonary aspergillosis: clinical significance of the halo sign. Clin Infect Dis 2007; 44: Zaspel U, Denning DW, Lemke AJ, et al. Diagnosis of IPA in HIV: the role of the chest X-ray and radiologist. Eur Radiol 2004; 14: Mattei D, Rapezzi D, Mordini N, et al. False-positive Aspergillus galactomannan enzyme-linked immunosorbent assay results in vivo during amoxicillin-clavulanic acid treatment. J Clin Microbiol 2004;42: Sulahian A, Touratier S, Ribaud P. False positive test for Aspergillus antigenemia related to concomitant administration of piperacillin and tazobactam. N Engl J Med 2003; 349: Kaplan EL, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc 1958; 53: Prognosis Factors in IA CID 2008:47 (1 November) 1183

9 24. Rothman KJ. Estimation of confidence limits for the cumulative probability of survival in life table analysis. J Chronic Dis 1978; 31: Peto R, Pike MC. Conservatism of the approximation sigma (O-E)2- E in the logrank test for survival data or tumor incidence data. Biometrics 1973; 29: Humar A, Michaels M. American Society of Transplantation recommendations for screening, monitoring and reporting of infectious complications in immunosuppression trials in recipients of organ transplantation. Am J Transplant 2006; 6: Cornely OA, Maertens J, Bresnik M, et al. Liposomal amphotericin B as initial therapy for invasive mold infection: a randomized trial comparing a high-loading dose regimen with standard dosing (AmBiLoad trial). Clin Infect Dis 2007; 44: Jantunen E, Ruutu P, Piilonen A, Volin L, Parkkali T, Ruutu T. Treatment and outcome of invasive Aspergillus infections in allogeneic BMT recipients. Bone Marrow Transplant 2000; 26: Yeghen T, Kibbler CC, Prentice HG et al. Management of invasive pulmonary aspergillosis in hematology patients: a review of 87 consecutive cases at a single institution. Clin Infect Dis 2000; 31: Cordonnier C, Ribaud P, Herbrecht R, et al. Prognostic factors for death due to invasive aspergillosis after hematopoietic stem cell transplantation: a 1-year retrospective study of consecutive patients at French transplantation centers. Clin Infect Dis 2006; 42: Fukuda T, Boeckh M, Carter RA, et al. Risks and outcomes of invasive fungal infections in recipients of allogeneic hematopoietic stem cell transplants after nonmyeloablative conditioning. Blood 2003;102: CID 2008:47 (1 November) Nivoix et al.