Bulletin of the World Health Organization

Global burden of chronic pulmonary aspergillosis as a sequel to pulmonary tuberculosis

David W Denning a, Alex Pleuvry b & Donald C Cole c

a. National Aspergillosis Centre, University Hospital of South Manchester, Southmoor Road, Manchester, M23 9LT, England.
b. Independent pharmaceutical consultant, High Peak, England.
c. Dalla Lana School of Public Health, University of Toronto, Toronto, Canada.

Correspondence to David W Denning (e-mail: ddenning@manchester.ac.uk).

(Submitted: 19 April 2011 – Revised version received: 13 July 2011 – Accepted: 25 July 2011 – Published online: 27 September 2011.)

Bulletin of the World Health Organization 2011;89:864-872. doi: 10.2471/BLT.11.089441

Introduction

With more than 36 million people cured of tuberculosis between 1995 and 20081 and 9 million new cases diagnosed worldwide each year,2 the health of those affected over the long term warrants attention. Treated pulmonary tuberculosis (PTB) can lead to complications, including progressive loss of lung function,3 persistent pulmonary symptoms3 and chronic pulmonary aspergillosis (CPA).46 Of the long-term sequelae of PTB, CPA is perhaps the most subtle, yet the most severe.411 In the 1960s the Research Committee of the British Thoracic and Tuberculosis Association estimated the prevalence of CPA in patients who had a residual cavity of at least 2.5 cm on the chest radiograph following treatment for PTB.9,12 It assessed more than 500 patients from 55 chest clinics twice – once about 12 months after the sputum became negative for acid fast bacilli,12 and again three years later.9 Remarkably, 25% of the patients had detectable Aspergillus precipitins in blood and both precipitins and radiological features of an aspergilloma were detectable in 14% at 12 months and in 22% at 3–4 years. PTB and CPA present with similar symptoms. This, combined with inadequate facilities for testing for immunoglobulin G (IgG) antibodies (precipitins) against A. fumigatus in many places, probably results in the underdiagnosis of CPA both at initial presentation13 and following treatment for PTB. For example, in early case series of people with respiratory illness and negative acid fast bacillus (AFB) sputum smears in sub-Saharan Africa, A. fumigatus was among the pathogens identified.14 CPA is an important differential diagnosis of what appears to be smear-negative tuberculosis.

CPA occurs in various forms: simple aspergilloma, chronic cavitary pulmonary aspergillosis and chronic fibrosing pulmonary aspergillosis, both with and without an aspergilloma.4 Unlike invasive aspergillosis, CPA occurs in immunocompetent patients. Morbidity is considerable and is marked by both systemic and respiratory symptoms and haemoptysis.7,8Weight loss, profound fatigue, severe shortness of breath and life-threatening haemoptysis are common. Progressive pulmonary fibrosis and loss of lung function, also common, could partly account for the unexplained loss of lung function in these patients. Even when treated, CPA has a case fatality rate of 20–33% in the short-term and of 50% over a span of 5 years.5,8

The country-specific PTB statistics and mortality rates published by the World Health Organization (WHO)15 make it possible to estimate the burden of chronic sequelae after treatment for PTB. Our objective was to use these published clinical and population data as inputs to model estimates of the likely burden of CPA related to PTB worldwide.

Methods

We developed a deterministic scenario model using Excel (Microsoft, Bellevue, United States of America). Fig. 1 shows our approach to estimating the adult burden of CPA in the largest countries of every WHO region. We started with WHO estimates of the number of new cases of PTB and of deaths from PTB15 and assumed that the mortality figures quoted by the WHO were for the point 12 months after the diagnosis of PTB.

Fig. 1. Factors used to estimate the annual incidence and five-year period prevalence of chronic pulmonary aspergillosis (CPA)
Fig. 1. Factors used to estimate the annual incidence and five-year period prevalence of chronic pulmonary aspergillosis (CPA)
PTB, pulmonary tuberculosis, TB, tuberculosis.a Attrition applied annually over 5 years.

We searched the literature with the following questions in mind: (i) What is the frequency of pulmonary cavitation after completion of the treatment for PTB? (ii) How common is CPA following PTB? (iii) Are there any radiological risk factors (such as cavitation) for CPA? (iv) What is the range of the 12-month survival for PTB (to estimate the numbers at risk of developing CPA development)? and (v) What is the range of the 12-month survival for CPA (to estimate attrition and convert incidence to period prevalence)? We initially adopted a systematic search strategy but quickly realized that the literature was limited and that scoping reviews for all five questions were more appropriate.16

To identify the primary literature on cavitation after PTB, we searched several electronic bibliographic databases using the search terms “tuberculosis” and all of (separately) “follow up”, “radiology”, “cavitation”, “cavity”, “CT”, “radiograph”, “radiographic”, “outcome”, “fungal ball”, “aspergilloma”, “aspergillosis” and “haemoptysis” (using both British and American English spellings). All literature on aspergilloma was retrieved back to 1936. Our primary source was Medline, but we also searched the Aspergillus web site (including the historical paper archive [pre-1964]). We did not search meeting abstracts, doctoral theses or other grey literature sources, but we hand searched our extensive files of pre-1990 papers on all forms of aspergillosis. Every paper retrieved with information on cavitation following PTB was scrutinized and additional referenced papers were also retrieved. The term “aspergilloma” entered the medical literature in the 1940s, but the phrase “chronic pulmonary aspergillosis” was not formally used until 2003. Thus, searches with the terms “aspergillosis” and “aspergilloma” yielded very different numbers of papers. We initially identified over 400 papers and scanned their abstracts. We read over 100 papers to determine the availability of data on the treatment of pulmonary cavitation. Five sources contained relevant quantitative information on cavitation; the rest contained only qualitative data. To minimize selection bias, we established as inclusion criteria a cohort study or case series design and a minimum of 20 study subjects. We also contacted the original authors of the primary combination studies of PTB conducted by the Medical Research Council of the United Kingdom of Great Britain and Northern Ireland to establish if the original end-of-treatment chest radiographs or readings were available for review. Unfortunately, they were not.

We identified five papers that provided data on the proportion of cases with pulmonary cavities following anti-tuberculous therapy. The studies yielded different figures: they ranged from 21% in the United States17 to 35% in Taiwan, China18 and > 50% in South Africa in some studies,19 while in others the figures were 21–23% in South Africa20 and the United States17 and 30% in Brazil21 (Table 1). We decided on an intermediate figure of 22% and applied it to countries across the globe except for the 22 members of the European Community. For Europe we took into account our own (unpublished) observations from the United Kingdom, which showed a post-treatment cavitation rate of less than 10%, and selected an intermediate figure of 12% that will require prospective validation. We also performed sensitivity analyses using figures of 10% and 30% for countries outside the European Community.

We examined the papers found through our searches for data on the frequency of an association between PTB and either aspergilloma or chronic pulmonary aspergillosis, as well as on the radiological characteristics of these entities. We excluded case reports, qualitative studies and papers describing an association with invasive aspergillosis. Papers reporting the frequency of aspergilloma or fungal balls without serological confirmation were included but those describing cavitation without a fungal ball were included only if CPA was confirmed serologically or through direct histological exam or culture of the lesions. Finally, only two papers from the same Medical Research Council study in the United Kingdom linked radiological findings with the subsequent development of aspergilloma (and included serologic testing for Aspergillus) and, therefore, chronic pulmonary aspergillosis (Fig. 2).9,12 The figures shown in Fig. 2 were checked and confirmed by all authors. We found no papers describing the rate of CPA after PTB without reference to pulmonary cavitation or in patients without cavitation.

Fig. 2. Relative frequency of chronic pulmonary aspergillosis after standard anti-tuberculous treatment
Fig. 2. Relative frequency of chronic pulmonary aspergillosis after standard anti-tuberculous treatment
TB, tuberculosisa Including resection (n = 7) showing an aspergilloma.Source: Anonymous9 and Research Committee of the British Tuberculosis Association.12

On a country by country basis, we multiplied the number of survivors 12 months following initiation of therapy for PTB by the designated percentage of patients with pulmonary cavities. We then estimated the number of patients with cavities who were likely to develop CPA (22% according to the studies from the United Kingdom,9,12 Fig. 2). We found little published data on CPA among patients without visible cavities on a chest radiograph following PTB, so we chose 2% as the best estimate, with upper and lower bounds of 1% and 4% for sensitivity analyses. This allowed us to estimate the number of cases with CPA 1 year following completion of treatment of PTB.

The purpose of the fourth review was to establish the range of survival rates 12 months after diagnosis of PTB. A full systematic review could have been conducted to address this question, but since the rates of survival were used to identify patients at risk for CPA for modelling purposes only, we chose to perform a scoping review to establish the survival range. We assumed that all cases of PTB had been correctly diagnosed in these studies. Five studies were identified and 12-month mortality following PTB ranged from 5% to 26%.

We also conducted a scoping review to estimate12-month survival after the development of CPA and found only three published studies that allowed us to make such an estimate (85%) and convert annual incident cases into five-year period prevalence. Patients undergoing resection surgery for aspergilloma are so few compared with the number who die from CPA that we did not estimate their proportion.

Long-term survival from PTB varies widely and is affected by co-infection with the human immunodeficiency virus (HIV), age, treatment adherence and the presence of multidrug-resistant or extensively drug-resistant PTB (MDR-PTB and XDR-PTB, respectively). Annual mortality following PTB has been shown to vary from < 2% in Denmark22 and 9% in Guinea-Bissau23 to 15% in Uzbekistan.24 (Table 2). Annual mortality from MDR-PTB may be no higher than mortality from PTB responsive to medication (41% over 5 years),26 but it is higher in HIV-positive individuals (26%) before they are treated with antiretroviral therapy.25,27 Thus, to estimate the five-year period prevalence of CPA as a complication of PTB, we applied 10%, 15% and 25% annual attrition rates to deduct deaths annually over the 5-year period.

All papers retrieved that contained quantitative data were reviewed by all authors to ensure that the samples surveyed in each paper were not highly selected and hence biased. The proportions of patients with pulmonary cavities and the annual post-treatment mortality rates were extracted independently by two authors (DWD and DCC) and discrepancies were resolved by discussion. The list of excluded papers is available from the authors on request.

Results

According to WHO, in 2007 an estimated 7.7 million cases of PTB occurred worldwide, and 5.96 million (77.1%) of them survived to at least 12 months after diagnosis. Nearly 50% of these cases occurred in China and India, whose combined population in 2005 was 2.4 billion (Table 3).28 We estimate that annually at least 372 385 patients in the world develop CPA following PTB, but incidence estimates vary substantially by country and WHO region. The number of new cases of CPA associated with PTB was estimated at 11 420 in the European Region, 20 615 in the Eastern Mediterranean Region and 12 610 in the Region of the Americas (Table 4). However, minimum estimates were 98 551 for the African Region, 83 815 for the Western Pacific Region and 145 372 for the South-East Asia Region. Individual country estimates for the 24 most populous countries are provided in Table 3.

Our best estimate of the global five-year period prevalence of CPA following PTB was 1 173 881 patients, with a range from 852 048 at 25% annual attrition to 1 372 457 at 10% annual attrition (Table 4). Sensitivity analyses using rates of cavitation after PTB of 10% and 30% and rates of CPA in people without cavities of 1% and 4% altered the estimates from a low global five-year period prevalence of 546 844 to a high of 1 786 421 when a 15% attrition rate was applied.

The five-year period prevalence of CPA indicated a predicted prevalence rate of 18 per 100 000 population (Table 3). The prevalence rate of CPA varies widely, however. Among the 23 largest countries in the world it ranges from as low as 0.4 per 100 000 in Germany to 42.9 per 100 000 in both Nigeria and the Democratic Republic of the Congo. China and India have intermediate predicted prevalence rates of 16.2 and 23.1 per 100 000, respectively. In the largest developed countries, the predicted prevalence rate is invariably below 1 per 100 000.

Discussion

According to our model, which resembles many models used by WHO to estimate the burden of other diseases, around 1.2 million people in the world have CPA as a sequel to PTB. Most CPA cases occur in WHO’s South-East Asia, Western Pacific and African regions, where PTB has the highest prevalence. In many series PTB is an underlying condition among CPA cases, but this varies widely. Only 17% of referred CPA patients in Manchester, England,29 had underlying PTB, compared with 93% in Seoul, Republic of Korea.6 This variation reflects differences in clinician awareness, in diagnostic approaches in patients with persistent pulmonary shadowing, and in the relative frequency of underlying pulmonary diseases in each locality. The progressive loss of pulmonary function and/or the presence of symptoms after PTB could be caused by CPA, but this possibility has never been studied. In cases in which CPA is diagnosed, symptoms such as fatigue, cough, shortness of breath, weight loss and haemoptysis are best managed with antifungal therapy. Identifying CPA early in patients with residual pulmonary shadows from PTB is only possible by means of microbiological testing (principally for Aspergillus IgG antibodies). If tests are not conducted, patients are often diagnosed as having “smear-negative pulmonary tuberculosis”, “progressive upper lobe fibrosis” or “recurrent pulmonary tuberculosis”, all of which result in inappropriate therapy or none at all. In areas with a high prevalence of tuberculosis, criteria for the diagnosis of CPA and PTB are so similar that distinguishing between the two entities is not possible, without serological testing for Aspergillus precipitins, even if sputum culture is positive for A. fumigatus.14

Accuracy of pulmonary tuberculosis case estimates

We based our estimates of CPA prevalence following PTB on WHO tuberculosis rates.15 The data are robust in some countries but not others. Under-reporting is common, especially in countries such as China. Therefore, we have probably underestimated the burden of CPA. In addition, both incident cases and cure rates are changing relatively rapidly, thanks to the Stop TB Partnership. Case fatality rates dropped from 8% to 4% between 1995 and 2008.1 A recent estimate of mortality in patients with HIV and TB co-infection yielded a rate of 5% for countries in Africa with a low prevalence of HIV infection, but closer to 20% in those with a high prevalence,25 consistent with our estimates. Increased survival is likely to lead to greater numbers of people at risk of sequelae, including CPA. Estimating post-treatment survival was challenging, mostly because accurately estimating prognostic denominators was difficult, as others have found.27

CPA case ascertainment

To estimate CPA burden we have used radiographic findings primarily. Our own data suggests that about 25% of patients with CPA have an aspergilloma.30 The original United Kingdom survey on PTB was conducted with chest radiographs, yet computerized tomography (CT) scanning of the thorax is much more sensitive, especially in the apex of the lungs, which is the site of most PTB and CPA. The cavitation rates of 30%18 and 35%21 after PTB are based on CT, whereas estimates based on chest radiographs are generally 21–23%.17,20 An even higher rate of residual cavitation (> 50%) was demonstrated in a population with MDR-PTB.19 We arbitrarily applied to all of Europe the cavitation rate in the United Kingdom instead of conducting a prospective assessment, a clear study limitation. Plain chest X-rays have reasonable sensitivity (70%) for the detection of pulmonary cavities when anti-tuberculous treatment is initiated and while it is being administered, but it drops by the end of treatment (49%)17 and few centres undertake CT scans at that time. Although we have accounted for this variation in our upper and lower estimates, additional work is necessary to validate these frequencies in different populations, especially in North America and Europe. We recognize that a robust estimate of CPA based on rates of cavitation after PTB needs to be fully validated at the local level, especially in countries where data are old or do not exist, such as the United Kingdom.

All patients who have had a pulmonary insult are probably at some risk of developing CPA. The relative risk of CPA following PTB in patients with smaller cavities or with none has not been estimated. A cavity is thought to be important in pathogenesis because the insult to the lung in that area probably undermines local host defences, allowing Aspergillus conidia to germinate. In other groups of patients who develop CPA, notably those with sarcoidosis and emphysematous bullae, pulmonary cavities predate the development of CPA. In others, including patients with survived lung cancer or who have allergic bronchopulmonary aspergillosis, cavities are not present before the development of CPA. We have estimated that the risk of CPA among PTB patients without discernible cavities is about 2% (range: 1–4%), but this may not be an accurate estimate across all populations. Prospective studies are needed to substantiate these rates.

In addition to pulmonary cavitations on chest X-ray, with or without an aspergilloma, either the presence of hyphae in a pulmonary cavity (preferably with a positive culture) or the presence of Aspergillus IgG antibodies must be definitively demonstrated for the diagnosis of CPA to be made. Serologic tests for A. fumigatus IgG antibodies have a sensitivity of about 90% for CPA.31,32 Rare cases of CPA caused by A  niger, A. nidulans and A. flavus instead of the more common pathogen, A. fumigatus, have been documented8,3335 and may be a source of false negative serology results. Thus, case detection is likely to be incomplete. Detectable Aspergillus desoxyribonucleic acid (DNA) in respiratory samples36 could suggest CPA, prior to A. fumigatus IgG antibodies being requested.

Accounting for CPA-related mortality

CPA can be cured in 1% to 17% of patients who undergo surgery, usually within a year of diagnosis.9,11 Mortality from surgery is extremely low for simple aspergilloma3740 but much higher for complex disease.4,11 Even with antifungal treatment, CPA develops gradually and leads to progressive loss of lung function. The case fatality rate after admission to hospital ranges from 10% to 30%.6,8,10 We have therefore introduced an annual attrition rate of 15% by default, with a range from 10% to 25%, when converting annual incident cases to five-year period prevalence. In our experience, survival is determined primarily by the combined effect of the severity of the underlying pulmonary disease and the extent and pace of lung destruction.

Other risk factors

Many risk factors for CPA probably exist and they include some genetic defects. Deficiency of surfactant A2 and toll-like receptor 4 has been shown to alter innate immune function.41,42 In CPA patients, cytokine production profiles typically show a Th2 cytokine profile41 and gamma interferon production may be absent or poor.43 Other risk factors such as these, whose frequency probably varies in different ethnic groups, could affect both the incidence and progression of CPA.

Future directions

CPA is a sequel of PTB more commonly than is generally appreciated. It can account for progressive lung destruction and the persistence of symptoms after successful anti-tuberculous treatment and can mimic smear-negative PTB. Antifungal therapy is very beneficial in CPA patients, as it reduces both morbidity and mortality.68,30,4346 Little data exist on the development of CPA after PTB. Prospective clinical and epidemiological studies using the best diagnostic tools available are needed to ascertain its frequency in different places and among different ethnic groups. Recognition of CPA and treatment with generic itraconazole have the potential to reduce morbidity and mortality from CPA worldwide at a modest cost.


Acknowledgements

We are indebted to Joanne Gill and the library staff for sourcing the papers and to multiple colleagues who answered questions about the frequency of disease in their countries, including Ashok Shah (Vallabhbhai Patel Chest Institute, University of Delhi), Won-Jung Koh (Sungkyunkwan University School of Medicine, Seoul) and Peter Omerod (Blackburn, England). Helen Carruthers drew Fig. 2.

Funding:

Funding was provided by the University Hospital of South Manchester, Manchester, England.

Competing interests:

David W Denning holds founder shares in F2G Ltd, a University of Manchester spin-out company, and has received grant support from F2G as well as the Fungal Research Trust, the Wellcome Trust, the Moulton Trust, The Medical Research Council, The Chronic Granulomatous Disease Research Trust, the National Institute of Allergy and Infectious Diseases, National Institute of Health Research and the European Union, AstraZeneca and Basilea. He has been as an advisor/consultant over the last 5 years to F2G, Lab21, Basilea, Vicuron (now Pfizer), Pfizer, Schering Plough (now Merck), Nektar, Daiichi, Astellas, Gilead and York Pharma. He has been paid for talks on behalf of Schering, Astellas, Novartis, Merck, Dainippon and Pfizer. Alex Pleuvry is a Director and shareholder in Oncalex, an independent consultancy, with no specific financial interest in respiratory or fungal disorders. Donald C Cole is a tenured professor with consultancies on environmental health to public health units but none on respiratory or fungal disorders or their treatment.

References