To the best of our knowledge, this report was the first one discussing the diagnostic approach and included the largest series of comparisons in clinical manifestations of patients with IPA and PM. Approximately half of our patients had DM and chronic lung disease. Only 11 patients (16%) had hematologic disease and eight of these were under neutropenic status. Influenza infection was independently associated with IPA. Prior exposure to voriconazole provided clinical hint for diagnosing PM. While PM had lower culture rate comparing to IPA, the GM level from serum and BAL were significantly higher among IPA patients. Seventy-two percent of patients with IPA and 95% of patients with PM were proven cases and two third of the diagnosing tissues were obtained from bronchoscopy. The mortality was greater than 60% in both IPA and PM patients. Systemic steroid exposure and APACHE II score on admission were independently correlated to mortality in IPA, and concurrent bacterial sepsis independently predicted the in-hospital mortality among PM patients.
In recent decades, we found that not only immunocompromised patients were susceptible to fungal infections, diabetes mellitus, chronic lung diseases may also experience invasive fungal infections [4]. In one review article, Patterson et al. demonstrated that critically ill patients who are admitted to intensive care units (ICUs) also have higher risk of pulmonary fungal infection [2]. Wauters et al. revealed that IPA is a more frequent complication in critically ill H1N1 patients and suggest that use of systemic steroid in these patients is an independent risk factor for fungal infections [5]. Moreover, in one largest multicenter retrospective study, Schauwvlieghe et al. collected 457 critically ill patients with influenza comparing to 321 patients with community acquired pneumonia and demonstrated that influenza was independently associated with IPA [14]. In current study, DM and chronic lung disease was the most common underlying disease in both IPA and PM patients. The underlying disease was similar in IPA and PM. However, we found that after influenza infection, patients were independently more susceptible to IPA rather than PM (19% vs. 0%, p=0.036). IPA is the most common mold infection, followed by PM [1]. Nevertheless, the incidence of PM seems to rise in recent years [4, 7, 15]. Neofytos et al found that after bronchoscopy initiation, the observed rate of invasive mucormycosis among HSCT recipients increased from 0.6% annually to 3.0% [7]. In our previous report, 25% of bronchoscopy diagnosed invasive fungal tracheobronchitis were caused by Mucorales [4]. In current investigation, among these invasive pulmonary mold infection patients, one third of them had PM. Carol Garcia-Vidal et al. had proposed that voriconazole prophylaxis had significantly associated with zygomycosis in hematologic patients [16]. Moreover, in animal models, the virulence of zygomycytes may increase after exposure to voriconazole [17]. In current report, prior voriconazole exposure was strongly associated with PM (14% vs. 0%). Moreover, two of three patients had concomitant IPA and PM based on the histopathologic report was under voriconazole treatment before definite diagnosis. Although the exposure to voriconazole may not a directly leading to PM, the clinical awareness of PM should be risen when the patients had prior voriconazole exposure.
Typical findings in CT scan, such as halo sign (HS) for IPA, reverse halo sign (RHS) for PM and air-crescent sign, were useful in diagnosing invasive pulmonary mold infection in immunocompromised patients. In one review article, Georgiado et al. found lots of diseases other than invasive pulmonary mold infection presenting with those findings [8]. Nam et al. recently presented that consolidation or mass with halo sign were the most common finding on CT scan in PM. The serial morphologic changes into reverse halo sign, central necrotic cavity or air-crescent sign were noted after treatment and recovery of neutropenia [18]. However, articles discussing about CT finding in non-neutropenic patients were scarce. Jung et al. conducted a retrospective study focusing on comparison of CT scan in IPA and PM. They found the RHS was more common in PM. But the RHS was more frequent developed in neutropenic patients and half of them only appeared within five days from symptoms onset [11]. In our study, which mainly composed by non-neutropenic patients, the most common finding on CT scan was consolidation (29 of 33, 83% in IPA vs. 19 of 20, 95% in PM, in patients who receiving CT scan, respectively). There were a total 8 patients (12%) were under neutropenia status while diagnosing invasive mold infections and five had consolidation in CT scan, one had merely airway invasion, one had cavitation and one had abscess formation. None of our patients had HS or RHS. In the CT scan findings among our invasive pulmonary mold infection patients, only abscess formation was higher in PM patients than IPA (35% vs. 9%, p=0.028), but it was not independently significant in multivariate analysis.
The fungal culture takes lots of time and positive rate in lower respiratory tract samples is 48 to 76% in IPA patients [11, 19] and ranged from 29% to 46% in PM patients [11, 20]. In current report, the positive culture rate was significantly higher in IPA than in PM (68% vs. 41%, respectively, p = 0.0363). Among these patients, most of the positive culture specimen were obtained from bronchoscopy (78% in both IPA and PM). GM is a fungal cell wall component which is released during tissue invasion by Aspergillus hyphae and can be detected in serum and bronchoalveolar lavage (BAL) fluid [21]. Serum levels in non-neutropenic patients may be underestimate because circulating neutrophils are able to clear the antigen [22]. Although the value of serum GM can be influenced by neutrophil counts [22, 23], GM in BAL sample provides reliable diagnosing tools in IPA [21]. Moreover, GM had no role in diagnosing PM. In current study, we demonstrated that GM in both serum and BAL fluid were significantly higher in IPA than in PM (3.2 ± 0.5 vs 0.7 ± 0.6, p = 0.0259; 4.1 ± 0.6 vs. 1.5 ±1.0, p=0.0435). Considering the possible influence of neutrophil counts in serum GM, samples from BAL is a good choice for differentiate IPA from PM. Furthermore, in our research, 72% (34 of 47) of IPA and 95% (21 of 22) of PM were proven cases. Among them, 80% (28 of 34) proven IPA and 67% (14 of 21) proven PM tissue were obtained from bronchoscopy. Moreover, none of these patients developed massive hemoptysis or major complications after bronchoscopy. Although we found that prior use of voriconazole was independently associated with PM, there were three patients had concomitant IPA and PM. Chamilos et al. also had found five cancer patients had concomitant IPA and PM [10]. Considering the concomitant IPA and PM, the precise and timely diagnosis may still mainly depend on histopathology. In patients with high risk of surgical intervention, biopsy from bronchoscopy and specimen from BAL for GM test and fungal culture may be an ideal diagnosing method.
Many studies reveal that underlying conditions, such as neutropenia, active malignancy, liver disease, hematopoietic stem cell or solid organ transplant, are risk factors for mortality in invasive mold infection [7, 24-26]. In current study, only 11 patients (16%) had hematologic disease and eight were under neutropenic status. We found that underlying disease had no association with in-hospital mortality in both IPA and PM. In addition to uncontrolled underlying disease, Taccone et al. showed that mechanical ventilation, renal replacement therapy during ICU stay and higher Sequential Organ Failure Assessment score at diagnosis were independent predictors for death of invasive aspergillosis in critically ill patients [27]. Two third of our patients (68% in IPA and 64% in PM) developed respiratory failure before diagnosis and had admitted to ICU. The usage of systemic steroid, respiratory failure before diagnosis, surgical intervention were correlated to mortality in IPA, but only systemic steroid exposure and high APACHE II score on admission were independent risk factor for in-hospital mortality among IPA patients. Regarding to PM, Lin et al. conducted a retrospective study which composed more than 70% patients had hematologic malignancy and were under neutropenic status. They demonstrated that concurrent bacteremia is the sole independent predictor for mortality among those PM patients [20]. In our PM patients, which mainly composed with underlying diabetes mellitus (DM) and chronic lung disease (59% and 36%, respectively), concurrent bacterial sepsis was still the only independent predictor for in-hospital mortality. These finding suggest that the outcomes of IPA and PM were more closely related to the clinical disease severity rather than underlying conditions.
The present study has some limitations. First, for the retrospective nature of this study, the sample size was limited and some of the data are incomplete, this probably led to diminished generalizability; second, we only included proven and probable cases, patients with possible invasive mold infection may have been overlooked, and the outcomes may have been underestimated.