Case Definition of Chronic Pulmonary Aspergillosis in Resource-Constrained Settings

Chronic pulmonary aspergillosis (CPA) is a recognized complication of pulmonary tuberculosis (TB). In 2015, the World Health Organization reported 2.2 million new cases of nonbacteriologically confirmed pulmonary TB; some of these patients probably had undiagnosed CPA. In October 2016, the Global Action Fund for Fungal Infections convened an international expert panel to develop a case definition of CPA for resource-constrained settings. This panel defined CPA as illness for >3 months and all of the following: 1) weight loss, persistent cough, and/or hemoptysis; 2) chest images showing progressive cavitary infiltrates and/or a fungal ball and/or pericavitary fibrosis or infiltrates or pleural thickening; and 3) a positive Aspergillus IgG assay result or other evidence of Aspergillus infection. The proposed definition will facilitate advancements in research, practice, and policy in lower- and middle-income countries as well as in resource-constrained settings.

Although coccidioidomycosis, histoplasmosis, and paracoccidioidomycosis are regionally confined, aspergillosis is global. Each year, an estimated 373,000 new CPA cases complicate treated pulmonary TB within 12 months of completion of anti-TB therapy; the 5-year period prevalence is 1,174,000 (range 397,000-2,088,000) cases (9). This wide range results from several factors, notably the extrapolation of CPA diagnosis from a limited UK dataset of only 544 patients with pulmonary cavities (15), substantial variability in the published frequency of cavitation after treatment of pulmonary TB, absence of an estimate of CPA prevalence among patients without cavities, and lack of knowledge of the effects of concurrent HIV infection. The incidence and prevalence of CPA are not known but are probably underestimated, in part because CPA occurs in patients with active pulmonary TB, as a sequela of prior pulmonary TB (16), or as a complication of other pulmonary disorders with symptoms similar to those of pulmonary TB, and is incorrectly diagnosed and treated as pulmonary TB (17).
Diagnostic guidelines for CPA have recently been published in English and Japanese, emphasizing the central role of advanced imaging and serologic testing for Aspergillus (18)(19)(20). Unfortunately, these diagnostics are infrequently available in many resource-constrained settings. In October 2016, the Global Action Fund for Fungal Infections convened an international panel to develop an operational definition of CPA for research and clinical care in resourceconstrained settings. The panel's goals were to adapt the existing European Society for Clinical Microbiology and Infectious Diseases and European Respiratory Society (19) and Infectious Diseases Society of America guideline case definitions of CPA (18) to promote research so that critical data will be available to inform policy and practice, including surveillance, and to enable individualized clinical care for optimal patient management.

Literature and Existing Guidelines
We built on the work of 2 recent CPA expert panels (18)(19)(20). These panels undertook comprehensive searching, appraisal, and synthesis of the relevant literature, including diagnosis and case definitions. We included these papers in a package of materials relevant to diagnosis in different clinical contexts (i.e., underlying disease in patients in whom CPA is developing, CPA and clinically diagnosed pulmonary TB, radiologic assessment and characteristics of CPA, and comparisons of laboratory diagnosis with different immunoassays).

Workshop Participants
The Global Action Fund for Fungal Infections (https://www.gaffi.org) invited 36 experts from all regions of the world, according to expertise. The experts had already implemented CPA diagnostic capacity or were in the process of doing so. Participants also included experts from lower-and middle-income countries that had active clinical and public health programs focused on respiratory diseases including TB. Clinical expertise included internal medicine, pulmonary disease, infectious disease, critical care, thoracic radiology, medical microbiology, and medical mycology, as well as various health system organizational roles and levels (e.g., secondary care consultants, national reference laboratories, national research centers, and international health organizations). Of the 36 invited experts, 33 (the authors) attended a workshop in Liverpool, UK.

Operational Definitions Indicator Selection
Morning presentations and discussions built on prior reading material and provided all participants evidence of CPA burden, risk factors, clinical presentations, diagnostic tools, treatment options, recurrences, and prior case definitions. Three facilitated afternoon breakout groups started with CPA diagnostic indicator types (i.e., clinical presentation, radiology, medical microbiology-mycology-immunology) and discussed options for diagnosis, focusing on secondary care levels and above for lower-and middle-income countries. A recorder took notes, and the pros and cons of different indicators were shared in a plenary session. Subsequently, 3 cross-indicator groups worked to bring the indicators together, constructing key criteria for CPA with different clinical or radiologic presentations.

Development of Case Definitions
Plenary discussion participants compared and contrasted the different approaches and moved toward operational definitions. On the basis of consolidated notes from breakout groups and plenary sessions, descriptions of possible, probable, and confirmed CPA were synthesized to garner consensus on the most critical elements of the diagnosis. Simple graphic representations were developed, shared with breakout group leaders for feedback, and subsequently revised through iteration. Further iteration on the key elements of the algorithms to be used in the field and minimal definitional requirements were conducted via email and online file-sharing services, simplifying the definition to a single, composite definition and algorithm.

Results
Modern diagnostic criteria for CPA date from 2003 (21) and have been used in some prospective clinical trials (Table 1) and refined for specific purposes. The consensus group considered diagnostic criteria in 3 sections: clinical features, radiologic criteria, and microbiological criteria. †This definition includes CNPA cases with 1-3 mo symptoms. Note: precipitins sensitivity even lower than culture. Pleural thickening seems to be a mandatory criterion.

Underlying Diseases
Most CPA patients have prior or concurrent underlying pulmonary disease. Possible risk factors for development of CPA are cavities in the lung caused by pulmonary TB, sarcoidosis, previous Pneumocystis pneumonia, bullae or lung cysts, lung abscess, pulmonary infarction, pulmonary fibrosis, healed abscess cavities, cavitary bronchogenic carcinoma, and infection by NTM (21,(26)(27)(28)(29)(30). In many countries, pulmonary TB is the most common disorder that precedes CPA (9). The most common differential diagnosis is pulmonary mycobacterial infection.

Duration
The duration of disease required to define CPA used by previous studies has been 1-6 months, usually 3 months.
( Table 1). The expert panel consensus was to endorse a 3-month duration as a criterion for diagnosis of CPA. In patients with preexisting pulmonary disease, who often have chronic signs and symptoms, a change in pattern or severity of clinical presentation is considered the trigger point for the 3-month duration. In patients with few or unchanged signs and symptoms, documentation of 3-month duration may also be confirmed with radiographic findings of progression of cavitation, pericavitary infiltrates or fibrosis, development of a fungal ball (which takes weeks to form), or microbiological data.

Signs and Symptoms
Most patients with CPA experience clinical signs and symptoms, although some are asymptomatic and show progression radiologically only. The most distinctive and alarming sign is hemoptysis/hemosputum. Hemoptysis develops in ≈12%-43% of CPA patients (23,25,31,32) and varies from blood streaking in sputum to massive and fatal hemoptysis. Hemoptysis may occur in patients with TB, but it is only streaking of sputum with blood and usually not severe. Another characteristic symptom is mild but persistent chest pain, discomfort, or tightness, experienced by up to 37% of patients. Weight loss and fatigue are also common, although not universal. Cough (usually productive) and dyspnea are common but not sufficiently distinctive to distinguish CPA from other pulmonary disorders, including pulmonary TB. Fever or pyrexia is uncommon in CPA patients and, if present, may indicate a concurrent or alternative diagnosis or subacute invasive aspergillosis. Night or day sweats are occasionally reported but are not discriminatory. The consensus of the panel was that the diagnosis of CPA required the presence of >1 symptoms persisting for 3 months and radiologic progression, a scenario present in most cases. Two other clinical scenarios that also qualify for the diagnosis of CPA are radiologic evidence of a simple aspergilloma (with or without symptoms) and characteristic radiologic appearance without symptoms but showing definite radiologic progression.

Radiologic Criteria
The focus of the discussion revolved around use of chest radiographs alone to diagnose CPA; requiring computed tomography (CT), which is often unavailable, might delay diagnosis. Multiple studies have demonstrated that CT is more sensitive than chest radiography for demonstrating several features, namely, >1 fungal balls, pulmonary nodules, multiple cavities, and disease in the apices and retrocardiac space (33). Considering the limited availability of CT in many healthcare settings, the radiologic criteria adopted are based solely on chest radiograph findings. Where available, CT is recommended in the context of clinical and microbiological suspicion of CPA with a nondiagnostic chest radiograph.

Fungal Ball or Aspergilloma, Intracavitary Material, or Fluid Level on Chest Radiograph
The differential diagnosis for a fungal ball in the lung is limited: echinoccocal cyst, necrotizing bronchogenic carcinoma, or acute or subacute invasive fungal infection. Similar-appearing conditions are lung abscess, Rasmussen aneurysm in a tuberculous cavity, cavitating hematoma, or lung infarct (34). If single and in a localized area of lung, with few or no symptoms, a simple aspergilloma is the most precise diagnosis ( Figure 1) and can be resected or observed without antifungal treatment. Although presence of a fungal ball is highly suggestive of an aspergilloma as a manifestation of CPA, microbiological confirmation is required for a definitive diagnosis.
Occasionally, a fluid level is visible in cavities. Few of these cavities have been sampled with aspiration and microbiological analysis, but when done, cultures are usually positive for A. fumigatus or, less commonly, Staphylococcus aureus, Pseudomonas aeruginosa, and other pathogens. Therefore, a fluid level in a cavity is entirely consistent with CPA but may also represent co-infection (35).

Cavitation
The cardinal feature of CPA is the presence of >1 cavities. The cavities may be small or large; may have thick or, less commonly, thin walls; and usually abut the pleura (33,36). The cavities expand and may coalesce during progression of infection. In patients with extensive bullous emphysema, inflammation around a bulla may resemble cavitation. The differential diagnosis of chronic cavitary lesions includes mycobacterial infection, endemic fungal infection, and malignancy (37). The cavities seen with CPA are often located in the upper lung zone and may mimic TB; in patients with NTM infection and endemic fungal diseases, the cavities may be located in any lung zone. A distinguishing feature of CPA is slow progression of findings over months or years, whereas active pulmonary TB infection progresses faster (34,38) (Figure 2). Patients with NTM infection may also have cavitary lesions; however, lesions are more common among older white men with underlying lung disease (39). Patients with residual coccidioidal or other infectious cavities are usually asymptomatic unless these conditions are complicated by aspergilloma or superinfected (40). Cavitary bronchogenic carcinoma is usually associated with adenopathy and often pleural effusion.

Pleural Thickening
In the pre-CT era, pleural thickening was regarded as a sign of aspergilloma; indeed, pleural thickening is common in patients with CPA (34,38). For many patients, pleural thickening seen on CT images usually consists of 2 components: fibrosis of the pleura overlying a cavity or area of consolidation and indrawing of extrapleural fat (typically seen in chronic inflammatory processes of the lung). This subtle distinction cannot be made readily from a chest radiograph. Pleural thickening should be regarded as a common feature of CPA and useful for diagnosis. Furthermore, pleural thickening is a specific feature of CPA, rarely seen in patients with TB or chronic coccidioidal or other fungal cavities (40,41). Tuberculous empyema is usually exhibited as a large pleural effusion (42) and is commonly accompanied by interlobular septal thickening and micronodules (43   involvement; however, these manifestations are distinct from the focal pleural thickening associated with CPA.

Pericavitary Infiltration
The inflammatory changes seen adjacent to cavities in patients with CPA are often marked, reflecting inflammation but not hyphal invasion. These areas may merge with localized areas of fibrosis, pleural thickening, or both but are usually obvious on plain chest radiographs (36); they indicate active CPA and are a clear indication for therapy (44). Such pericavitary infiltrates, unless very extensive, are uncommon in reactivation of TB and NTM disease. Pericavitary consolidation may be seen in patients with chronic fibrocavitary coccidioidomycosis (40).

CT Image Features
One radiologic aspect of CPA not characterized on a chest radiograph is the interior of a cavity. Aspergillus grows inside the cavity along the wall, resulting in an irregular appearance of the inner border seen on CT images.
In addition, the cavity may contain linear opacities representing mats of fungal growth that have detached from the cavity wall (38). These growths often merge to form sponge-like densities, which can be described as a fungal ball containing air (45). These structures may detach from the cavity wall and may be mobile. For purposes of the definitions outlined in this article, all these characteristic features are deemed to be equivalent to a fungal ball (Figure 3). With antifungal therapy, they resolve more readily than an aspergilloma. One feature for which CT is much more sensitive than chest radiography is the Aspergillus nodule, particularly when the nodule is small (46). These nodules may have a maximum diameter of 5-50 mm, may be single or multiple, and may be solid or have central cavitation. Larger nodules (>3 cm) are more accurately described as masses, which may also be attributable to Aspergillus infection. The differential diagnosis of the Aspergillus nodule is broad and includes carcinoma (primary or secondary), coccidioidomycosis, cryptococcosis, NTM infection, and others.
Establishing a definitive diagnosis of a nodule usually requires resection or biopsy, especially because many patients do not have elevated IgG against Aspergillus or positive sputum cultures (47,48). Because these diagnostic procedures are not available in many medical centers in lower-and middle-income countries, we elected to consider these separately and not include them in our operational definition.

Culture-Based and Non-Culture-Based Evidence of Aspergillosis
The most reliable diagnostic test for CPA is a positive Aspergillus antibody test result, indicative of an immune response to Aspergillus. The second most reliable diagnostic marker for CPA is detection of Aspergillus in the airways by use of culture, antigen, and/or nucleic acid amplification (PCR).

Antibodies against Aspergillus
Elevated circulating levels of Aspergillus antibodies are present in >95% but <100% of patients (49)(50)(51) (21). Serologic testing for Aspergillus relies mainly on the detection of IgG and of precipitating antibodies (known as precipitins), which may be IgG or IgM. Precipitin detection requires immunodiffusion and electrophoresis migration methods, which lack standardization and are too laborious and time-consuming for resource-constrained settings. Consequently, we focused on commercially available enzyme immunoassay kits that detect IgG. Up to 2015, there were few direct comparisons of A. fumigatus serology (55), but more recently, several have been published (Table  2). Crucial to these comparisons is defining the cutoff values for each assay. The control groups of patients used for such comparisons have included healthy persons or those with pulmonary disease without aspergillosis. Few data on antibody titer associations with age, HIV status, or ethnicity of controls are available. New lateral flow devices for Aspergillus antibody detection are in the final stages of development and, if their performance is good, could greatly facilitate diagnosis. The consensus of the expert panel was that any Aspergillus antibody test performance had to be at least 90% sensitive and 85% specific.
An elevated level of Aspergillus IgG is consistent with several conditions, including Aspergillus rhinosinusitis, allergic bronchopulmonary aspergillosis, Aspergillus bronchitis (notably in cystic fibrosis and bronchiectasis), subacute invasive aspergillosis, recovery from invasive aspergillosis, and community-acquired Aspergillus pneumonia. Because an elevated level of IgG is highly sensitive but not specific for CPA, the diagnosis of CPA requires the presence of compatible symptoms and radiologic abnormalities.

Respiratory Tract Microscopy and Culture
Microscopy of sputum may show hyphae morphologically consistent with Aspergillus spp. If present, this finding is most consistent with CPA or Aspergillus tracheobronchitis (63). Despite the substantial amount of Aspergillus spp. in the cavities of patients with CPA, culture positivity from sputum samples is lower than expected (41%-81%) and is probably biased toward culture-positive cases (21,22,(64)(65)(66)(67). One reason for this lower sensitivity is the inoculation of culture plates with very small volumes of sputum (65,68) as is done for bacterial culture. Negative cultures might reflect an inability of the fungus to adapt to in vitro conditions, despite the apparent ease with which environmental contamination occurs in the laboratory.
The merits of positive culture results are substantial, notably for identifying the Aspergillus species causing infection and enabling susceptibility testing to be performed. False-positive cultures do occur as a result of laboratory contamination. Although most CPA cases are caused by A. fumigatus complex (probably sensu stricto), A. niger complex and A. flavus complex and rare cases caused by unusual pathogenic species are reported. In countries such as India, where A. flavus infection is much more common, it is not clear what proportion of cases are attributable to non-A. fumigatus species. Most triazole-resistant A. fumigatus isolated from patients with CPA has arisen while the patient was receiving therapy and probably result from large fungal loads or low drug exposure (low dose, drug interactions, poor bioavailability) (69). Isolates may be resistant or have intermediate susceptibility to 1-4 triazoles.

Aspergillus Antigen and β-1,3-D-glucan
Galactomannan is a detectable carbohydrate antigen produced by the growth of Aspergillus spp. Galactomannan detection is useful for the diagnosis of invasive aspergillosis because it is often detectable in serum and bronchoalveolar lavage (BAL) fluid. Because tissue invasion does not occur in patients with CPA, galactomannan is not usually detectable in serum and is usually detectable in BAL fluid only (59,70). This finding is of limited utility in lower-and middle-income countries because fiberoptic bronchoscopy is infrequently done. Galactomannan is detectable in sputum or tracheal secretions/aspirates, but the cutoff for positivity is not established (71).
A new simple lateral flow assay for different protein antigens specific to A. fumigatus has been commercialized, but no data are available on sputum detection or its utility for patients with CPA. This assay is unlikely to be useful in serum because of antibody masking but may be useful  in BAL fluid (72,73). β-1,3-D-glucan is released by Aspergillus spp. (and many other fungi) during infection but is less specific and probably not more sensitive than serum galactomannan in patients with CPA.

Molecular Detection of Aspergillus
For patients with CPA, PCR detection of Aspergillus spp. in sputum is significantly more sensitive than culture (63). However, the quality and possibly quantity of the sputum sample influence PCR performance (71). Furthermore, studies have shown that the sensitivity of PCR for detection of Aspergillus is ≈80%. Depending on the cutoff value used, low specificity may also be an issue. Molecular assays for Aspergillus spp. are not routinely available in most medical centers worldwide, especially in lower-and middle-income countries.

Consensus Definition and Proposed Algorithm
The definition of CPA recommended for use in resourceconstrained settings in shown in Table 3. In a patient who has had symptoms consistent with chronic Aspergillus infection for >3 months, a chest radiograph should be obtained (Figure 4). If a chest radiograph is not possible, then pulmonary TB should be excluded by examination of respiratory specimens. The chest radiograph appearance determines the next course of action. If the chest radiograph is not remarkable, the signs and symptoms may be caused by bronchiectasis or systemic disease. If the chest radiograph shows consolidation without cavitation, then other diagnoses such as lung malignancy, airway obstruction, or endemic fungal infections should be considered. If the chest radiograph shows cavitation, then work-up for possible pulmonary TB should be initiated according to national practice and standards. A positive result for pulmonary TB should lead to appropriate treatment. If the pulmonary TB results are negative, and especially if the radiographic findings include pleural thickening, fungal ball, or pericavitary fibrosis or infiltrates, serum should be tested for Aspergillus IgG (Figure 4). If the Aspergillus IgG test result is negative (or unavailable), then sputum microscopy for hyphae or fungal culture should be performed; yield is higher when multiple specimens are tested. If the sputum microscopy for hyphae or fungal culture is negative, then other diagnoses, such as atypical mycobacterial infection or endemic fungal infection, should be considered. If sputum microscopy for hyphae or culture is positive, or if a serum Aspergillus antibody test result is positive, then CPA is confirmed and treatment with itraconazole or voriconazole is advised (19).

Conclusions
We present recommendations for diagnosing CPA, adapted for clinical use and public health surveillance in resource-constrained settings. The adapted definition relies on a combination of signs and symptoms, chest radiograph features, and serologic evidence of Aspergillus infection; it is applicable to almost all patient groups in which CPA occurs. Prospective evaluation of the proposed definition and associated algorithms will be valuable in the contexts of different geographic and HIV and pulmonary TB burdens (74). We believe that the proposed operational definition of CPA in lower-and middle-income countries will facilitate advancements in research and practice (clinical, laboratory, radiologic, and public health) and policy (health services and public health) in these countries and globally in resource-constrained settings.