Chronic pulmonary aspergillosis is common among patients with presumed tuberculosis relapse in Ghana

Abstract Chronic pulmonary aspergillosis (CPA) may mimic pulmonary tuberculosis (PTB). The two diseases are clinically indistinguishable and may result in CPA misdiagnosed as PTB or vice versa. Although PTB is largely recognised as a differential diagnosis of CPA and often ruled out prior to CPA diagnosis, the reverse is uncommon. The aim of this study was to determine the proportion of CPA cases among patients being assessed for PTB. A cross-sectional survey was conducted among consecutive patients referred for GeneXpert Mycobacterium tuberculosis test for the diagnosis of PTB at the Korle-Bu Teaching Hospital, Accra, Ghana. Patients’ demographics, clinical and socioeconomic details were obtained using a structured questionnaire. Blood was collected for Aspergillus and HIV serology, and sputum samples obtained for Aspergillus culture. Chest radiograph was obtained, and computed tomography scan was also done for patients with positive Aspergillus serology or cavitation. CPA was defined using an algorithm developed by the Global Action for Fungal Infections (GAFFI) international expert panel. A total of 154 patients were included in the analysis, of whom 134 (87%) did not have a prior PTB diagnosis. There were 41 (26.6%) GeneXpert positive cases. CPA prevalence was 9.7% overall, but 50% in patients with a prior history of PTB and 3.7% in those without previous PTB. Although CPA is rarely considered as a differential diagnosis of PTB in Ghana, our findings show that CPA may affect half of patients being assessed for PTB relapse. Efforts to diagnose CPA should be prioritised in this patient group.


Introduction
Pulmonary fungal infections have increased in clinical significance in recent times, and although many of them mimic pulmonary tuberculosis ( PTB ) , chronic pulmonary aspergillosis ( CPA ) is one of the most common. 1 CPA is a slow, progressive, and destructive lung disease associated with both respiratory and systemic symptoms. Globally, approximately 3 million people suffer from CPA, with 1.2 million occurring as a sequel of PTB. 2 In Ghana, CPA among PTB patients is estimated at 2600 cases annually. 3 PTB is a common differential diagnosis of CPA, and could occur before, after, or infrequently, together with CPA. 4 There are many similarities between PTB and CPA in terms of risk factors, clinical presentation, and radiological features, making the two diseases clinically indistinguishable. 5 This may result in misdiagnosis of CPA as PTB, or vice versa. As PTB is more common and largely recognised globally, the index of suspicion for PTB is likely higher compared to CPA, particularly in settings with a high PTB burden. Being mostly diagnosed as a post-PTB complication, CPA may be misdiagnosed as relapsed PTB infection and managed as such. 5 CPA may also be occasionally misdiagnosed as primary TB infection. 5 Some studies have reported CPA misdiagnosed as acid-fast bacilli ( AFB ) smear-negative or GeneXpert Mycobacterium tuberculosis ( MTB ) -negative PTB and resulting in worsening symptoms and anti-TB treatment failure. 6 , 7 Previous and present guidelines for CPA diagnosis have recommended a necessary exclusion of PTB. 4 , 8 However, with emerging concerns of primary CPA and CPA co-existing with PTB, it may be equally important to rule out CPA when making a diagnosis of PTB to avoid inappropriate exposure of patients to anti-TB medications.
Unfortunately, differentiating CPA from PTB in many high TB burden countries, which are mostly resource-constrained, is a major challenge. This is probably because of inadequate awareness, unavailable diagnostic laboratory support, and omission of CPA as a differential diagnosis in existing local guidelines. To improve the status quo , the Global Action for Fungal Infections ( GAFFI ) convened an international expert panel in 2016 to develop a CPA guideline specific for resourceconstrained settings. 8 However, the lack of Aspergillus serology testing capacity, which is key to CPA diagnosis, poses a significant limitation to the general use of these guidelines. 9 -11 Until recently, the common standard and commercially available methods were precipitins and enzyme immunoassay. The drawbacks of these techniques include cost, long turnaround time, poor inter-laboratory reproducibility, and variable cut-off values. 9 , 12 -14 Additionally, these tests require sophisticated equipment and adequate laboratory expertise. LDBio Diagnostics introduced a new rapid diagnostic test ( RDT ) in the form of a lateral flow assay ( LFA ) for the detection of Aspergillus -specific IgG and IgM antibodies based on immunochromatography technology that meets the World Health Organization ( WHO ) ASSURED ( 'Affordable, Sensitive, Specific, User-friendly, Rapid and robust, Equipment-free, and Deliverable to end users' ) criteria and may be more suitable for use in resource-constrained settings. Several evaluation studies or clinical use of the LFA have reported a general high analytical performance and strong clinical relevance. 15 -19 In this study, we screened for CPA among patients presumed to have PTB using the LDBio Aspergillus -specific IgG and IgM LFA and the CPA guideline for resource-constrained settings. 8 We also evaluated the significance of CPA as a differential diagnosis of PTB and assessed the clinical relevance of the LD-Bio Aspergillus -specific IgG and IgM LFA in CPA diagnosis.

Study design and site
The study was a cross-sectional survey conducted at the Chest Clinic, Chest Diseases Unit, Department of Medicine and Therapeutics, Korle-Bu Teaching Hospital, Accra. The clinic acts as the national TB referral centre and hosts a specialised TB laboratory that receives samples from different parts of Ghana.

Study population
Patients seen at the Chest Clinic of the Korle-Bu Teaching Hospital for suspected PTB or those referred from other parts of the country to the TB laboratory for GeneXpert MTB ( Xpert ® MTB/RIF, Cepheid, CA, USA ) testing were recruited, irrespective of their symptoms. Also, blood donors determined to have no symptoms and signs of a respiratory condition or history of PTB, or any other chronic respiratory condition via an interview, were also recruited as control group participants from the National Blood Service, Ghana. This was mainly done to assess the specificity of the LFA test.
Ethical approval was obtained from the Institutional Review Board of the Korle-Bu Teaching Hospital ( STC/IRB/00058/2020 ) and the National Blood Services Ghana ( NBSGRD/201410/02 ) and University Research and Ethics Committee of the University of Manchester ( Ref: 2020-9368-16168 ) . Written informed consent was obtained from all participants.

Investigations
Patients' demographics, clinical and socioeconomic details were collected via interviews using a questionnaire. Serum samples were obtained for Aspergillus serology with LD-Bio Aspergillus IgG and IgM LFA ( LDBio Diagnostics, Lyon, France ) and HIV antibody testing with HIV ½ RDT ( Healgen Scientific LLC, Houston, TX, USA ) and confirmed with OraQuick HIV ½ RDT ( OraSure Technologies, Bethlehem, PA, USA ) . Sputum was obtained for high-volume culture to enhance Aspergillus detection, a modified version of Vergidis et al ., 20 by inoculating an aliquot ( 1-2 ml ) of undiluted specimen on Sabouraud dextrose agar and incubated at 37°C for up to 8 days. Chest radiograph was done for all patients unless patient had obtained one within the previous month. Xpert MTB/RIF results were retrieved from laboratory records. Chest computed tomography ( CT ) scan was done for patients with positive Aspergillus serology or cavitation on chest radiograph including MTB-positive cases. Among the MTB-positive cases with cavitation, CT scan was done to unravel any concealed imaging features of CPA to identify possible PTB-CPA coinfection. In the control group, only serum samples were obtained for Aspergillus serology.

Case definition
TB was diagnosed if MTB was detected in a patient's sputum by Xpert MTB/RIF assay. Patients with suspected or confirmed PTB were classified as new PTB and relapsed PTB as follows: i. new PTB; patients with no prior history of PTB ii. relapsed PTB; patients who had been treated successfully for PTB in the past A case of CPA was defined following the guidelines for CPA diagnosis in resource-constrained settings, developed by the GAFFI international expert panel ( 2018 ) . 8 The panel defined a case of CPA as follows: i. weight loss, persistent cough, and/or haemoptysis for > 3 months ii. chest images showing progressive cavitary infiltrates and/or a fungal ball and/or pericavitary fibrosis or infiltrates or pleural thickening iii. positive Aspergillus IgG assay or other evidence of Aspergillus infection.
Patients who met criteria ( i ) and ( ii ) above, but not ( iii ) , or met criteria ( i ) and ( iii ) , but not ( ii ) were categorised as probable CPA, a modified version of a classification described by Setianingrum et al . 21

Data analysis
Data were analysed with SPSS version 25 ( IBM, Armonk, NY, USA ) at 5% significance level, using either Chi-square or Fisher's exact tests. Summary statistics were presented using frequencies and percentages for categorical variables, and median values for non-normally distributed continuous variables. Fisher's exact tests were employed to compare proportions between groups. Logistic regression was carried out to assess the effect of individual symptoms and socioeconomic details on the likelihood of acquiring CPA.

Laboratory results
Laboratory findings of the 154 patients are shown in Table 2 . MTB was detected in 41 ( 26.6% ) patients, of whom 35 ( 85.4% ) were classified as new PTB and 6 ( 14.6% ) had relapsed PTB. Aspergillus serology was positive in 9.9% ( n = 16 ) of participants, but there was no imaging data for two ( 12.5% ) of these and hence they were excluded from the analysis. There were 44 ( 28.6% ) HIV-positive patients, of whom 2 ( 4.5% ) had positive Aspergillus serology. Aspergillus serology 4 Ocansey et al.   Table 2 ) .

CPA classification
Of the 154 patients, 15 ( 9.7%, 95% CI 7.8%-11.6% ) met the criteria for CPA, including three probable CPA cases ( Supplementary Table S1 ) . Ten ( 66.7% ) CPA cases had pre-vious PTB, representing 50% ( 10/20 ) of the patients being assessed for relapsed PTB. The predisposing conditions in the remaining CPA patients were chronic obstructive pulmonary disease ( COPD ) ( n = 3 ) and asthma ( n = 2 ) . Four cases classified as CPA, all with previous history of PTB infection, had trace or very low levels of MTB detected on Gen-eXpert, and were placed on TB treatment. The diagnosis of CPA was made in these patients because the imaging features were more suggestive of CPA including cavitation with irregular intraluminal lining of cavity, pleural thickening adjacent cavity, paracavitary fibrosis, and one with a fungal ball. Additionally, Aspergillus antibody test was positive in all four patients and Aspergillus spp. was isolated in three patients. Three subsequently had a negative sputum Xpert MTB and/or AFB smear 1 month later but the fourth patient had a positive AFB smear which showed scanty organisms. The proportion of CPA based on HIV status were 10.9% ( 12/110 ) and 6.8% ( 3/44 ) for HIV-negative and HIV-positive, respectively ( P = .560 ) . The most common symptoms among CPA cases were fatigue ( 73.3%, n = 11 ) , weight loss ( 73.3%, n = 11 ) , and haemoptysis ( 46.7%, n = 7 ) . Haemoptysis was proportionately more common in CPA than other symptoms. No association was found between CPA diagnosis and socioeconomic details as potential risk factors.

Discussion
This study is the first epidemiological study on CPA from Ghana and provides data for differential PTB diagnosis. It is common practice in high TB prevalence areas that patients previously treated for PTB who present with new symptoms are generally diagnosed with PTB relapse even when sputum AFB smear or Xpert MTB are negative. Our study showed that half of these patients had CPA. Therefore, a prior history of PTB treatment in a patient presenting with suspected relapse should raise suspicion for CPA.
The current study revealed a CPA prevalence of 9.7% among patients presenting with presumed PTB. Varying rates have also reported in Uganda 22 and Nigeria, 6 Iran, 23 Pakistan, 24 Indonesia, 15 , 21 Uganda, 16 and Brazil 25 largely due to difference in study designs, population, investigations, and sampling methods. Also, the present study identified 50% prevalence of CPA in patients with prior PTB. This is similar to recent reports from Vietnam ( 54.3% ) 26 and India ( 57% ) . 27 CPA and PTB often present with clinically indistinguishable symptoms; fever is more common in PTB, and haemoptysis in CPA, but are not sufficiently distinctive features to be used as a definite diagnostic tool. A recent review established a common association between CPA and TB in Africa. 28 Considering the high burden of TB in many African countries such as Ghana, CPA is likely to be misdiagnosed as PTB. 5 In recent times, the new GAFFI case definition for CPA for resource-constrained healthcare settings, utilising the new Aspergillus IgG and IgM LFA, is improving epidemiological studies and providing more clinical experience of CPA in resource-constrained settings. 7 , 15 -19 In the present study, three of four patients with concomitant CPA and PTB had a negative AFB smear within a month. The average interval from completion of TB treatment to recruitment for these patients was 4 years. Possibly, their Xpert MTB could have been false positives due to detection of residual MTB DNA or non-viable non-intact bacilli. 29 -31 The role of Xpert MTB in detecting reinfection or relapse accurately is unclear, often associated with high occurrence of false positives especially with low mycobacterial burden, and short time post-treatment, as noted in our study. 30 , 31 The likelihood of false positivity is reported to decrease with the longer time since successful treatment of PTB but the total duration is not known. 30 Nevertheless, one study indicates that this can be up to 4 years after successful completion of appropriate treat-ment. 31 The fourth patient had a positive AFB smear result, and probably had PTB-CPA co-infection.
The profound drug-drug interaction between rifampicin and oral antifungal azoles requires that clinicians to select between these diagnoses, and not attempt treating both PTB and CPA together. Some very ill patients with CPA require intravenous antifungal therapy while those with a single aspergilloma can be cured with surgical resection. 4 However, the majority require at least 12 months of oral antifungal ( itraconazole or voriconazole ) treatment for chronic cavitary pulmonary aspergillosis to reduce symptoms, prevent progression or relapse and improve overall quality of life. 4 A recent randomised controlled trial in India demonstrated that 12 months of oral itraconazole was superior to a 6-month regimen in reducing relapses of CPA at 2 years. 32 The most common symptoms for CPA cases recorded in the current study were fatigue, weight loss, and haemoptysis. Although patients with active PTB also had these symptoms, haemoptysis was more common in CPA, and should raise suspicion for diagnosis. 5 COPD was more common in patients with CPA than PTB, although overall numbers were small. We observed no significant statistical difference in the rate of CPA in HIV-positive patients and CPA in HIV-negative patients ( 6.8% vs 10.9% ) , contrary to other studies. 6 , 16 , 22 , 33 Though Aspergillus serology was positive in less than 10% of our screened patients, it contributed to greater than 90% of CPA diagnosis. One CPA patient who was HIV-positive had a negative Aspergillus IgG and IgM LFA result. This maybe because of a reduced capacity to elicit production of antibodies; the correlation between immunodeficiency and negative Aspergillus serology has been previously described. 6 , 9 , 33 Unfortunately, the participant in the control group with a positive Aspergillus antibody test was unable to be reached for imaging and further laboratory investigations to evaluate the clinical relevance. It could have been a possible case of asymptomatic pulmonary aspergillosis or a false positive test. In addition, we did not observe an association with various sociodemographic practices common in Ghana that could potentially lead to fungal exposure.
The most common organism implicated in CPA is A. fumigatus ; it is especially reported in Europe and United States, where it accounts for over 90% of all cases. 34 -36 However, in Africa and Asia, A. flavus and A . niger are frequently isolated, as demonstrated in our study and elsewhere. 6 , 15 , 16 , 34 The frequent imaging findings of CPA, as stipulated in many guidelines, reported in several epidemiological studies, and considered to be more linked to CPA are cavitation, pericavitary fibrosis, and pleural thickening. 4 , 5 , 8 , 15 , 21 , 22 These three features were present in 67% ̶ 80% of CPA cases in our study on chest radiograph and/or CT scan. Three patients without cavitation on chest radiograph were diagnosed with probable CPA based on parenchymal fibrosis and/or bronchiectasis coupled with positive Aspergillus serology and/or culture. It is possible that these patients had cavitation that could have been revealed by CT scan as observed in two other patients with CPA. CT scan is an important complementary investigation to chest radiography when available especially when Aspergillus serology is positive. In the present study, additional cavitation and fungal balls detected by CT scan in four patients were missed on chest radiographs. Similar observations were made by Page et al . 22 and Nguyen et al . 26 In conclusion, the early differentiation of active PTB, post-TB lung disease, and PTB plus CPA co-infection, in settings with high TB burden, may require a broader screening strategy at the investigation stage. The present study contributes to the efforts of identifying an efficient framework for routine or systematic screening for CPA in PTB. Access to readily available diagnostics, in addition to algorithms that easily identify patients with CPA, will improve patient care and outcomes.
Our study is not without limitations. Our inability to do CT scans for all eligible patients due to loss to follow-up or death was a major challenge. Secondly, prior to PTB retreatment, a repeat Xpert MTB and/or culture was not done for CPA patients from whom MTB was detected to rule out false positive results. Additionally, patients being assessed for PTB relapse were few, and the CPA rates in this group may not be representative.