Determine TB prevalence and incidence, select trial sites and choose target populations for TB vaccines Suzanne Verver, epidemiologist 3rd Global Forum on TB vaccines, 26 March 2013, Cape Town Outline • • • Choose target populations Epi studies to determine TB prevalence and incidence Select trial sites: hotspots Target groups • • • • • • • • Infants Easy to access Include vaccine in EPI Slow effect on epidemic Adults Highest incidence The older the more stable More difficult to access and follow-up • High risk groups: HIV infected, urban population • Adolescents 12-18 • Mix pre- and post-infection (South Africa 50%) • Easy to access; • the younger the easier to followup but the lower the incidence • Probably need to be included for licensure WHO TB impact working group mtg 2010 background paper no 9 by Ana Bierenbach WHO TB Impact working group mtg 2010 background paper no 9 by Ana Bierenbach Objectives of epidemiological studies to prepare for vaccine trials • To assess TB prevalence (adolescents) & incidence • To assess proportion MTB infected • To assess comorbidities (Malaria, HIV) & mortality (verbal autopsy) and strategies to reduce these • To assess recruitment and cohort retention strategies • To assess yield of different case finding strategies and diagnostic algorithms • To assess current coverage of BCG & EPI • To build capacity for vaccine trials Assess TB incidence and prevalence: Cohort studies (see poster 43) • Adolescent cohort studies and neonatal cohort studies • South Africa (SATVI): compare active and passive case finding strategies – ACS (n=6363) ; NCS (n = 4786) – Moyo Int J Tuberc Lung Dis 2012; Mahomed PlosOne 2013 • India, Kenya and Uganda: – NCS (n~2500-2900); ACS (n~5000) – publications ongoing • Excellent studies but took > 5 years to get results Alternative designs • Infants: – Shorter cross sectional study among infants 0-3 yr to assess period prevalence (CHC Cambodia and CISM Mozambique): challenges to assign prevalent and incident cases • HIV infected adults: – Retrospective paper based cohort study to estimate TB incidence among HIV infected adults: link TB and HIV records and compare to district data (Aurum, South Africa): challenges to link records – Combined retrospective and prospective estimate of TB incidence among HIV infected persons: South Africa (Aurum) and Tanzania (Ifakara in Bagamoyo): analysis ongoing – Estimates change with IPT and ART use No quick way to get accurate incidence estimates Lessons from neonatal cohort studies for trials • Number of culture positive cases small -> adapted endpoints (Moyo 2012) • Active case finding finds more cases, and they are equally severe as in passive group -> decided to use active follow-up in future trials (Moyo 2012) • High mortality needs mortality reduction strategies (Nabongo submitted) • High concordance TST and quantiferon, but low sensitivity for TB disease > both tests needed (Moyo 2011) • Large proportion of participants was suspects and needed admission in ward; -> more efficient ways? (Moyo 2012) • Importance of testing mother and infant for HIV (in prep) • Relevance of surveillance of TB treatment registers (Moyo 2012) • Enrollment strategies (Moyo 2012) • Relevance of harmonization of CXR reading (Wajja in prep) Lessons from adolescent studies for trials • Proportion LTBI differed by site: in South Africa 50% (Mahomed 2011). • Adolescents with positive TST and/or Quantiferon have higher TB incidence than those with no infection at baseline (Mahomed 2011). Quantiferon converters have 9x higher risk to develop TB (Machingaidze 2012). • Relevance of surveillance site specific: in SA half the cases detected through regular health system but not in Uganda (Mahomed 2013) • Differences between specific schools (Mahomed 2013) • Ruler and calliper comparable for TST (Geldenhuys 2010) • Low proportion smear-positive: early case finding leads to less severe cases (Waako submitted) • TB among adolescents not HIV related; probably due to high transmission (Mahomed 2013, Nduba submitted, Waako submitted) • High proportion non-tuberculous mycobacteria (Asiimwe 2013) • Usefulness of early morning sample (Ssengooba 2013) Learn lessons during preparatory epi studies rather than during trials*? • Cheaper since no investigational product, less ethical requirements, less intensive follow-up for adverse events, less screen failures • Do not want to make errors in trials • Capacity building • In infants no quick and easy way to obtain incidence; and highest in third year • Adaptive designs also need some estimates • Lessons applicable to several groups (eg adolescents/adults) • Epi studies can identify high risk groups (smokers, urban slums) • Epi studies can collect samples for biomarkers Lessons from trials see Tameris 2012 Challenge of epi studies • Define scientific objectives: – add on studies, – biomarkers – link with questions of national TB control programmes • Takes more time • Need to analyse data in detail to predict numbers for more strict exclusion criteria in trial Develop trial sites HIV/malaria vaccine trial sites TB diagnostic/drug trial sites • • • • • • • TB epidemiology: calculate expected number of TB suspects, LTBI and TB cases by type. Clinical: CV ward, tuberculin skin testing, gastric washing, induced sputum, harmonize chest x-ray reading of nonsevere cases Laboratory: TB culture capacity for large number of samples with low proportion positive ; sample handling TB treatment start and follow-up for outcome Collaborate with TB control programme Community knowledge, attitudes and practice • TB epidemiology: calculate expected number of TB suspects, LTBI and TB cases by type. Clinical; depends on target group • Lab: expand capacity • Follow-up of large cohort by routine screening Community knowledge, attitude and practice & work with community in general • TB hotspots – HIV negative adults Find geographical hotspots within countries • • • • Use NTP data by district/subdistrict Calculate TB incidence among HIV uninfected = A*(1-B)/C*(1-D) A = number of TB cases notified B = proportion of TB cases co-infected with HIV C = population D = proportion of the population HIV infected Limitations: differences in age, sex, underdiagnosing & underreporting of TB and HIV, mortality, population estimates Literature review on TB incidence among HIV negative persons (Mitchell et al in preparation): • Very little information: IPT trials, active case finding/screening studies, prevalence surveys Use modelling TB incidence in trial differs from notified incidence Expect a decrease due to: • Strict selection criteria • Exclusion of high risk groups • Exclusion of prevalent cases: wash out effect • Early case finding & preventive therapy Expect an increase due to: • Under notification: assess by inventory studies • Under diagnosis Notified cases may not cover the age category or geographical area where the trial takes place • Study to quantify the level of under-reporting of diagnosed cases of TB to national surveillance systems. • Compare the number of cases recorded in public and private health facilities with the records of cases notified • To estimate TB incidence using capture–recapture methods. • Need 3 fairly independent data sources • Use record linkage • Simpler: – assess initial defaulters – Assess cases lost from hospital referrals – Collaborate private sector Under diagnosis: develop hotspots Improve current case finding of health system by: • Improving diagnostics • establishing contact tracing • Following up all suspects and make sure they do not get lost in system • Needs collaboration with TB control programmes and private sector Acknowledgements • • • • KNCV: Ellen Mitchell Sites: • SATVI/UCT South Africa • KEMRI/CDC, Kisumu Kenya • IDI/Makerere Uganda & TB team Iganga/Mayuge • CISM Mozambique • Aurum South Africa • Ifakara Institute Tanzania Partners/funders: • GSK biologicals, Belgium • Aeras, USA • EDCTP • ITM Belgium • Karolinska Institute, Sweden More info: ververs@kncvtbc.nl; www.kncvtbc.org