Chronic pulmonary aspergillosis (CPA) in association with tuberculosis and AIDS Dr Iain D Page NIHR Academic Clinical Lecturer in Infectious Diseases UK National Aspergillosis Centre, University Hospital South Manchester Manchester Academic Health Sciences Centre (MAHSC) The University of Manchester 1

Chronic Pulmonary Aspergillosis Now well described in case series from UK, France, Japan and Korea. Guidance published by ECCMID and IDSA Diagnostic Criteria SYMPTOMS - cough or breathlessness or fatigue or haemoptysis for > 3 months RADIOLOGY - aspergilloma OR progressive cavitation OR paracavitary infiltrates MICROBIOLOGY -Aspergillus culture in ‘respiratory fluids’ OR anti-Aspergillus antibodies Exclusion of other active illness to explain symptoms Denning et al Clinical Infectious Diseases 2003:37:s265-80, Jhun et al, Medical Mycology;51:811-7, Ohba et al, Respiratory Medicine 2012;106:724-9

Chronic Cavitary Pulmonary Aspergillosis, with aspergilloma Case 004 (also published) in case histories section. Patient JP July 2001 Denning DW et al, Clin Infect Dis 2003; 37:S265 4

Chronic Fibrosing Pulmonary Aspergillosis Case 004 (also published) in case histories section. Patient JP April 2002 Denning DW et al, Clin Infect Dis 2003; 37:S265 5

CPA – without aspergilloma Case 004 (also published) in case histories section. Only 36% of CPA cases have aspergilloma – Jhun 2013 Denning DW et al, Clin Infect Dis 2003; 37:S265 6

CPA – outcome Haemoptysis – 58% (Denning 2003) 5 year mortality - 85% (Ohba 2012)

Chronic pulmonary aspergillosis – Treatment outcomes Oral itraconazole 6 mo 12 mo 35% 41% Stable Improved Standard care 6 mo 12 mo 23% 7% 29% 64% 71% 53% 21% 24% Deterioration 30% relapse off therapy in 6 months Natural history with no therapy over 12 months Agarwal R et al, Mycoses 2013; 56:559.

CPA – primary underlying conditions in UK population Tuberculosis (15.9%) Atypical mycobacteria (14.3%) ABPA (11.9%) COPD / Emphysema (9.5%) Prior pnuemothorax (9.5%) Lung Cancer (9.5%) Sarcoidosis (4.8%) Smith and Denning. The European Respiratory Journal. 2011;386:865-72

Frequency of TB underlying CPA - globally UK - 15.9% (Smith and Denning 2011) Japan - 50% (Ohba et al 2011) China – 71% (Chen et al 2012 – 256 surgical patients ) Korea – 81% (Jhun et al 2011) Senegal – 100% (Ba et al 2000 – 24 surgical patients) Chen et al. Interactive cardiovascular and thoracic surgery. 2012;15:77-80 Ba et al. Dakar Medical. 2000;45:144-6.

Sole published measurement of aspergillosis prevalence post TB Performed in UK between 1968 to 1970 No CT scans!! Medical Research Council study 544 patients previously treated for TB, with persistent cavities 34% developed precipitating antibodies to Aspergillus fumigatus 63% of patients with Aspergillus antibodies develop aspergilloma within 2 years 42% of those with aspergilloma developed haemoptysis British Tuberculosis Association. 1970. Aspergolloma and residual tuberculous cavities - The results of a resurvey. Tubercle 51:227. British Tuberculosis Association. Tubercle;51:227-45.

Frequency of Aspergillus antibodies after TB UK 1970 – 34% Japan 1989 – 20% India 2001 – 27% Brazil 1988 – 21% Iwata et al. Kekkaku. 1989;64:7.13. Shahid et al. Indian Journal of Medical Microbiology. 2001;19:201-5 Kurhade et al. Indian Journal of Medical Microbiology. 2002;20:141-4. Ferreira-da-Cruz et al. Memórias do Instituto Oswaldo Cruz.1988;83:357-60

Estimated Global Prevalence of CPA secondary to TB 36 million cured of TB worldwide 1995- 2008 22% South African patients have cavities post TB 1.1 million cases CPA predicted worldwide 100,000 new cases annually in Africa Prevalence of 43 per 100,000 in DRC Denning, Pleuvry and Cole . Bulletins of the World Health Organization 2011;89:864-72.

Possible reasons for error 1 - Environment

Possible reasons for error 2 – HIV co-infection

Aspergillosis in AIDS Described in pre-HAART era case series Mostly seen in neutropaenic pts Often secondary to zidovudine or ganciclovir 21% documented cases in non-neutropaenics though Almost always with CD4 <50 Neutrophil function often poor in advanced AIDS even if counts normal Survival better if patients received antifungals AND antiretrovirals Adrizzo-Harris et al. Chest 1997;111:612-8 Greenberg et al. Chest 2002;122:886-92 Denning et al. NEJM 1991;324:654-62

Bombay autopsy series – HIV +ve pts 1717 Bombay autopsy series – HIV +ve pts Lanjewar & Duggal, HIV Med 2001;2:266

Japanese AIDS autopsy series Katano et al BMC Infectious Diseases 2014, 14:229

Italian autopsy series Antinori et al, Am J Clin Pathol 2009;132:221 19

Aspergillosis in ‘smear-negative TB’ ‘Smear negative TB’ common in AIDS WHO diagnostic criteria for ‘smear negative TB’ 3 negative sputum smears Abnormal CXR Failure to respond to antibiotics 34% mortality at 2 years Patients NOT routinely tested for any fungal lung diseases 7% of patients with smear negative TB in Kampala grow Aspergillus on BAL culture Banda et al. Int J of Tuberc Lung Dis. 2000;10:1968-74 Yoo et al. Respirology. 2011:16:836-41

Research Questions What is the prevalence of CPA secondary to pulmonary TB? Does HIV co-infection alter the frequency of CPA Can CPA be diagnosed with chest X-ray and symptoms alone? Can CPA be diagnosed with chest X-ray + symptoms + serology? Can patients at high risk of CPA be identified at the time of active pulmonary TB treatment?

CCPA case definition Patients must have ALL of the following:- Cough or haemoptysis for 3 months or longer. Raised Aspergillus-specific IgG or positive precipitins. Radiological features of CPA – at least one from; Progressive cavitation on serial CXR. Paracavitary fibrosis on CT chest. Aspergilloma on CT scan. Plus exclusion of active pulmonary TB

Measuring anti-Aspergillus antibodies 1 – Precipitins / CIE 2 – ELISA Time consuming. Requires highly skilled laboratory staff. May have poor sensitivity. Results are subjective. Can be done in house with antigens from fungal culture. Very difficult to reproduce at other labs. Simple automated systems. Produces objective numerical results. Antigens produced by the manufacturer at a single site and distributed to labs. Intra-lab consistency can be assured. Diagnostic cut-offs for CPA not well defined. Many brands available – comparative efficacy poorly described.

Aspergillus-specific IgG test Exposure to Aspergillus spores probably ubiquitous Most adults have some anti-Aspergillus antibodies Existing published analyses of optimal cut-offs shown below Test Suggested cut-off Type of study Country Author ImmunoCAP 35 mg/L 97.5th centile Netherlands van Toorenebergen 2012 70.1 mg/L Belgium van Hoeyveld 2006 66.4 mg/L South Africa Watkins 2012 35.9 mg/L ROC (20 mixed cases vs 42 controls) Siemens Immulite 19.3 mg/L “Siemens software analysis of healthy controls” BioRAD No published data Serion / Virion Genesis

Study design 242 samples of stored sera from patients with proven CPA at UK National Aaspergillosis Centre tested. Patients not on antifungals at the time of testing. 100 samples from Ugandan healthy controls tested. ROC analysis used to define optimal diagnostic cut- offs. Sensitivity and specificity calculated for those cut-offs. British Tuberculosis Association. 1970tuberculous cavities - The results of a resurvey. Tubercle 51:227. British Tuberculosis Association. Tubercle. 1970;51:227-45.

ROC curves: CPA cases vs. healthy controls

levels between COPD and healthy controls for this assay RESULTS TEST CoV ROC AUC 95% CI New Proposed cut off Sensitivity Specificity in healthy controls ImmunoCAP - 0.996 0.992 - 1 20 mg/L 96% 98% Siemens Immulite 3.4% 0.991 0.982 – 1 10 mg/L Serion 23.2% 0.973 0.96 – 0.987 35 AU/ml 91% Dynamiker 12.1% 0.918 0.890 – 0.946 70 U/ml 75% Genesis 11.1% 0.902 0.871 – 0.933 16 U/ml 78% 97% CIE 59% 100% *a new high-specificity COPD cut off was not calculated for Genesis as there was no difference in mean levels between COPD and healthy controls for this assay

Are the cut-offs found in relation to Ugandan controls relevant to Europe? Environmental exposure to Aspergillus might be different in the two environments Genetic differences might affect results The unexpectedly low age of Ugandan controls recruited might bias results Differences between the current and proposed cut-offs would have significant impacts on test sensitivity and specificity

How to identify optimal European cut-off? Challenges Solution At least 100 cases and 100 controls needed for analysis UK NAC alone has > 100 CPA cases, but does not have access to healthy control sera No funding available Collaboration!! All authors of existing publications on the topic were contacted All agreed to share their data from testing of healthy controls ROC analyses performed comparing UK NAC CPA cases to controls from other European centers

Samples available for testing 241 sera from CPA cases at UK NAC tested by ImmunoCAP, Siemens Immulite and Serion. 118 separate sera from CPA cases at UK NAC tested by Bio-Rad Controls Test Control Group Country Number of controls Median age of controls ImmunoCAP Immulite Serion Blood donors from existing study Uganda 100 19 years Laboratory technicians Belgium 114 - Blood donors Dutch 152 Bio-RAD Pregnant women France 222

ROC AUC results and cut-off comparison Test ROC AUC Ugandan controls cut-off Existing manufacturer’s recommended cut-off European controls cut-off Sensitivity (%) Specificity (%) ImmunoCAP 0.956 20 mg/L 40 mg/L* 50 mg/L 84 96 Siemens Immulite 0.948 10 mg/L 21.4 mg/L 25 mg/L 93 99 Serion / Virion 0.944 35 U/ml Variable with batch 50 U/mL 91 Bio-Rad 0.955 - 10 AU/ml 1.5 AU/ml 98 Performance of existing manufacturer’s cut offs Test Sensitivity (%) Specificity (%) ImmunoCAP 87 90 Immulite 93 99 Bio-Rad 85 100

Potential issues with original study Significantly lower than the manufacturer’s upper reference range of 19.3 mg/L Based on analysis of results in 152 healthy Dutch controls ImmunoCAP cut of was 20mg/L Significantly lower than 40mg/L currently used in UK labs Large number of adolescents in Ugandan control cohort Controls were blood donors Blood donation drive underway in “colleges” at the time Median age = 19 years Age range = 17 – 39 years

Research question Methods Are the lower levels of Aspergillus-specific IgG levels found in the previous Ugandan healthy control cohort due to The young age of those controls? Genuine difference due to environmental or genetic factors? Methods We repeated ROC analyses and optimal cut-off calculations comparing Immulite Aspergillus-specific IgG levels in CPA cases to the following control groups 100 older Ugandan healthy controls 88 older Ugandan control with treated pulmonary tuberculosis, but asymptomatic with no radiological evidence of CPA 152 Dutch healthy controls

Research Questions What is the prevalence of CPA secondary to pulmonary TB? Does HIV co-infection alter the frequency of CPA Can CPA be diagnosed with chest X-ray and symptoms alone? Can CPA be diagnosed with chest X-ray + symptoms + serology? Can patients at high risk of CPA be identified at the time of active pulmonary TB treatment?

Gulu, Northern Uganda

Unpublished cross-sectional survey of CPA prevalence in Uganda CCPA diagnosed in patients with ALL of the following:- Cough or haemoptysis > 1 month Positive Aspergillus – specific IgG Siemens Immulite Cut off 20 mg/L Radiological features (any of three) Progressive cavitation on serial Chest X-ray Paracavitary fibrosis on CT thorax Aspergilloma on CT thorax All reporting done by 2 consultant radiologists blinded to clinical and serological results Deciding report by senior radiologist at UK National Aspergillosis Centre 400 unselected patients with previous pulmonary TB Community based survey Completed TB treatment in last 7 years 50% HIV positive All had CXR and Aspergillus IgG re-surveyed 18 months later Repeat CXR CT scan in those with positive IgG test or suspicion of aspergilloma on CXR

Could anything be done about it? Treatment with Itraconazole prevents CPA progression (ref below) RCT in India without drug level monitoring Stability or improvement in 76% treated pts vs 35% untreated (p=0.02) Voriconazole and Posaconazole also effective Expensive now, but will come off patent Anti-retroviral drugs distributed in Africa despite patents IV Amphotericin for sub-acute invasive disease in AIDS Already used in Africa for cryptococcal meningitis Amphotericin B can be made cheaply Surgery Case series in Senegal shows it can be done in Africa Agarawal et al. Mycoses. 2013;56:599-70.

Conclusion CPA complicating TB is likely to be a major under-recognised Global public health issue Treatment is plausible, even in resource poor setting Further studies now need to be done to confirm prevalence in areas of high TB prevalence Diagnostic tests suitable for use in low-resource settings needed Access to anti-fungal treatments needs to be improved