A comparison of liquid and solid culture for determining relapse and durable cure in phase III TB trials for new regimens

BACKGROUND: Tuberculosis kills more people than any other infectious disease, and new regimens are essential. The primary endpoint for confirmatory phase III trials for new regimens is a composite outcome that includes bacteriological treatment failure and relapse. Culture methodology is critical to the primary trial outcome. Patients in clinical trials can have positive cultures after treatment ends that may not necessarily indicate relapse, which was ascribed previously to laboratory cross-contamination or breakdown of old lesions. Löwenstein-Jensen (LJ) medium was the previous standard in clinical trials, but almost all current and future trials will use the Mycobacteria Growth Indicator Tube (MGIT) system due to its simplicity and consistency of use, which will affect phase III trial results. LJ was used for the definition of the primary endpoint in the REMoxTB trial, but every culture was also inoculated in parallel into the MGIT system. The data from this trial, therefore, provide a unique opportunity to investigate and compare the incidence of false ‘isolated positives’ in liquid and solid media and their potential impact on the primary efficacy results. METHODS: All post-treatment positive cultures were reviewed in the REMoxTB clinical trial. Logistic regression models were used to model the incidence of isolated positive cultures on MGIT and LJ. RESULTS: A total of 12,209 sputum samples were available from 1652 patients; cultures were more often positive on MGIT than LJ. In 1322 patients with a favourable trial outcome, 126 (9.5%) had cultures that were positive in MGIT compared to 34 (2.6%) patients with positive cultures on LJ. Among patients with a favourable outcome, the incidence of isolated positives on MGIT differed by study laboratory (p < 0.0001) with 21.9% of these coming from one laboratory investigating only 4.9% of patients. No other baseline factors predicted isolated positives on MGIT after adjusting for laboratory. There was evidence of clustering of isolated positive cultures in some patients even after adjusting for laboratory, p < 0.0001. The incidence of isolated positives on MGIT did not differ by treatment arm (p = 0.845, unadjusted). Compared to negative MGIT cultures, positive MGIT cultures were more likely to be associated with higher grade TB symptoms reported within 7 days either side of sputum collection in patients with an unfavourable primary outcome (p < 0.0001) but not in patients with a favourable outcome (p = 0.481). CONCLUSIONS: Laboratory cross-contamination was a likely cause of isolated positive MGIT cultures which were clustered in some laboratories. Certain patients had repeated positive MGIT cultures that did not meet the definition of a relapse. This pattern was too common to be explained by cross-contamination only, suggesting that host factors were also responsible. We conclude that MGIT can replace LJ in phase III TB trials, but there are implications for the definition of the primary outcome and patient management in trials in such settings. Most importantly, the methodologies differ in the incidence of isolated positives and in their capacity for capturing non-tuberculosis mycobacteria. It emphasises the importance of effective medical monitoring after treatment ends and consideration of clinical signs and symptoms for determining treatment failure and relapse. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (doi:10.1186/s12916-017-0955-9) contains supplementary material, which is available to authorized users.