Development and validation of a molecular diagnostic for tuberculosis
MetadataShow full item record
Background: Tuberculosis is a curable disease that is still responsible for more than 1.4 million deaths annually. In the face of the HIV epidemic the modest gains that have been made in trying to control and eradicate the disease are under serious threat. Lack of affordable, accessible, and geographically available diagnostics is one of the main hindrances in controlling the disease in resource limited settings where the disease burden is highest. Molecular methods due to their good accuracy, short turnaround times (1 to 24 hours), relatively low cost and potential for automation promise to fill current diagnostic gaps. Objectives: The main objective of this study was to develop and validate a low cost user-friendly molecular based test (known as NWU-TB) for tuberculosis diagnosis that provides a result on the same day. Methods: After developing the initial NWU-TB prototype, a preliminary independent evaluation was conducted at the South African National Institute of Communicable Diseases (NICD) using 109 patient sputum samples. Concurrently, a blinded evaluation study was conducted at the North-West University by the development team using 548 patient sputum samples collected at Orkney-Westvaal hospital. Informed by results and recommendations from the two studies, improvements on the initial prototype were made. A new evaluation study was then conceived and executed in order to evaluate the performance of the new prototype using 176 new clinical sputum samples collected at Orkney-Westvaal hospital. This study was combined with a small pilot study that aimed to optimise and evaluate the potential use of propidium monoazide (PMA) to discriminate between live and dead mycobacteria in clinical samples. MGIT culture was used as the reference test (gold standard) in all 3 of the performance evaluation studies conducted. Except in the NICD study, MGIT culture, sputum smear microscopy (SSM) and Xpert MTB/Rif were performed by registered laboratory scientists stationed at Orkney-Westvaal hospital. Finally, the cost per test of the developed diagnostic was estimated from a laboratory perspective using the “ingredients approach” and a preliminary cost-effectiveness analysis based on a simple decision model was also conducted. Results: In the NICD study, NWU-TB performance was compared with SSM against MGIT culture as the gold standard. NWU-TB showed a sensitivity of 67% and specificity of 88%, while SSM showed a sensitivity of 40% and specificity of 100%. For the initial NWU-TB prototype evaluation study conducted at the North-West University, comparisons were made to SSM and Xpert MTB/Rif using MGIT culture as gold standard. NWU-TB showed an overall sensitivity and specificity of 80.8% (95% CI: 75%–85.7%) and 75.6% (95% CI: 64.9%–84.4%) respectively, in comparison to 85.3% (95% CI: 79.9%– 89.6%) and 73.2% (95% CI: 62.2%–82.4%) respectively for Xpert MTB/Rif; and 62.1% (95% CI: 55.3%–68.4%) and 56.1% (95% CI: 44.7%–67%) respectively for SSM. Sensitivity of NWU-TB was lower (p=0.026) than that of Xpert MTB/Rif but better (p=0.001) than that of SSM. However, specificities for all the three assays where significantly lower than expected. Clinical evaluation of the improved NWU-TB prototype showed that both sensitivity and specificity increased. NWU-TB showed a better sensitivity of 94.6% [95% CI: 89.1%; 97.8%) compared to Xpert MTB/Rif 84% (p=0.0025) and SSM 64.3% (p<0.001). Xpert MTB/Rif showed a better specificity of 100% [95% CI: 88.1%-100%), which was significantly higher than NWU-TB 86.2% (p<0.001) and SSM 62.1% (p<0.001). Overall, there was a strong correlation between Xpert MTB/Rif and NWU-TB results (Mc Nemar’s chisquare =11.64, p value = 0.0006). Successful optimization of the PMA protocol was demonstrated using sputum samples. However, when samples were stratified and pooled based on whether individuals were receiving treatment or not, the results showed that there were no statistically significant differences between the two groups. Finally, the total cost per test for NWU-TB was estimated at $5.36, with material cost per test contributing the highest (75%) and equipment cost contributing the least (1.31%). Also, in the preliminary cost-effectiveness analysis, incremental cost effectiveness ratios (ICERs) for community based TB screening within South African mines using NWU-TB and Xpert MTB/Rif were US$50.3 per case detected and US$181.52 per case detected respectively relative to the current base case scenario of passive case finding. Conclusion: A relatively low cost ($5.36 per test), same day (˂ 2 hours TAT) molecular test prototype for tuberculosis was developed and evaluated. Additional product development and longitudinal cohort studies are required in order to produce a final prototype that can be commercialised. Sensitivity of the NWU-TB was comparatively better than both Xpert MTB/Rif and SSM against MGIT culture as gold standard. However, the specificities of all the assays were significantly lower than expected. Since all the samples used in these studies (except for the NICD study) were collected at a single site, any disease dynamics unique to the site may have been replicated throughout the studies. Use of PMA to discriminate between live and dead mycobacteria was optimised and its use demonstrated some potential. This is supported by cases in which culture-negative samples that had tested positive with the NWU-TB test without PMA treatment would test negative after PMA treatment showing that DNA from dead mycobacteria was possibly responsible for the false positive result in the first instance. The pilot study was ill equipped to generate sufficient data that permits conclusive clinical decisions on the utility of PMA, but the developed protocol can be adopted to a larger longitudinal study where such firm conclusions can be drawn.
- ETD@PUK