Prognostic Value of Interferon Gamma Release Assays in Predicting Active Tuberculosis Among Individuals With, or at Risk of, Latent Tuberculosis Infection

Overview

The objective of this study is to assess the efficacy of the two current TB (tuberculosis) blood tests (Interferon Gamma Release Assays (IGRA)) compared with the standard skin test (Mantoux Tuberculin Skin Test (TST)), for predicting active tuberculosis among those at increased risk of TB. Those at increased risk are defined as either newly arrived immigrants or people who have been in contact with TB cases. The study will also provide information on the cost effectiveness of different testing strategies, such as the two step testing approach recommended by NICE. The study is to be funded by the NIHR Health Technology Assessment programme. 10,000 participants will be recruited from 12 hospitals and a network of GP surgeries in London. All participants will have the skin test and blood taken for both assays. Disease status of participants will then be followed up for an average of 24 months using the national register of clinical reports, a phone call and the national microbiological database. The risk of developing active disease is highest in the first two years after exposure. During followup there will be no additional diagnostic procedures unless symptoms occur, i.e. in line with current NICE policy. A sub group of patients, selected as a random 25% of participants, will have a repeat IGRA test shortly after the first test to investigate whether the skin test affects the result of the blood test.

DESIGN: Prospective cohort study of individuals (\> or = 16 yrs) who are: (a) close contacts of TB cases or (b) new entrants from high incidence countries (\>40/100000). SETTING: London TB clinics. NE London TB Network Primary Care Practices. All study sites will be coordinated from the HPA Centre for Infections. HEALTH TECHNOLOGIES (LTBI MEASURES):Participants will be tested by Mantoux TST and two IGRA tests(Quantiferon-Gold In Tube ELISA) and ELISpot assay (same as Tspot.TB). A 25ml blood specimen will be collected with the residuum, after IGRA testing, full blood count and repetition of indeterminate assays, being stored for future research. All tests will be conducted using standardised protocols.(6) ACTION AFTER TESTING: This will follow existing NICE guidance. (1;2) A) If negative by TST and IGRA, follow up only. B) If positive by either TST or IGRA tests, active TB will be excluded. Those without active TB will be followed up. C) If positive by IGRA \& TST test and \> 35 yrs, follow up only and for those 16-34 yrs, chemoprophylaxis will be offered with balanced advice about potential benefits/risks. PRIMARY OUTCOME: Development of active TB. Prognostic values of tests quantified as incidence rate ratios (RR) among contacts and new entrants. SECONDARY OUTCOME: Side effects from chemoprophylaxis. FOLLOW UP: average of 24 months from the date of IGRA/TST testing. a) phone call to GP and or patients at 24 months. b) national enhanced TB surveillance. c) national database of culture proven TB. d) Clinic records. ADDITIONAL DATA: collected on all potential source cases and contacts using a questionnaire and medical records review (see below). HIV status will be determined at the end of the follow-up period through anonymised record linkage with the national HIV surveillance system, which is reliable.(4) DNA finger printing data, from the national strain typing database, will be utilised to ascertain transmission between index cases and subsequent diagnoses among contacts. ANALYSIS PLAN: The predictive performance of each test (TST, ELISpot and ELISA) will be summarised as the RR of test positives in those developing active TB compared to not (analogous to positive likelihood ratio: TPR/FPR). GEE Poisson regression will compare disease RRs between tests accounting for length of follow-up and exploiting within patient comparisons of tests. Absolute risk of disease in different groups will be described in those with no evidence of LTBI and those with LTBI who have / have not received chemoprophylaxis. Other analyses: HIV infected and risk factors for LTBI. SAMPLE SIZE: Assuming a LTBI rate of 30%, 5% progressing to active TB in 2 years if untreated and 20% loss to follow-up, simulations indicate a cohort of 5,000 would have 80% power of detecting clinically important differences in predictive performance (P\<0.05) that would arise from differences in sensitivity and specificity of 10% between tests for detecting LTBI. 50% of TB contacts and new entrants are aged \>35yrs, so a cohort of 10,000 identifies 5,000 for the primary analysis of progression without treatment. Testing and follow-up of 10,000 would allow appropriately powered secondary analyses a) comparing contacts and immigrants, b) estimating progression on treatment using regression models to adjust for test dependent treatment decisions. ECONOMIC ANALYSIS: The cost-effectiveness of alternative screening strategies for patients with suspected LTBI will be assessed. A decision model will be developed to estimate the costs (£) and health effects (QALYs) of the following strategies: a) no screening, b) TST alone, c) different IGRA tests (ELISA or ELISpot), and d) TST followed by IGRA if positive. These strategies will be compared for: contacts and new entrants stratified by age and baseline risk. The model will use data from the cohort study and the published literature, and will follow the NICE reference case (5). Probabilistic sensitivity analysis will be used to assess uncertainty. Benefits due to prevention of transmission will be estimated.