Once thought to be impossible by many in the TB community, hundreds of laboratory and academic professionals attending the 44th Union World Conference on Lung Health in Paris listened as they were introduced to TBDx™, the first fully-automated, sputum microscopy technology. Dr. Nazir Ismail, Director of the Center for Tuberculosis and primary investigator of the study, presented evidence of this reality as he described the results from the independent evaluation of the technology conducted this spring at the National TB Reference Lab in Johannesburg.
In presenting the background for the study, Dr. Ismail acknowledged that “for better or for worse, microscopy is here to stay. The challenge is whether technology can improve its use.” He went on to comment that GeneXpert is an excellent diagnostic tool, but the deployment of the technology has been limited, in part because of the costs and the absence of required infrastructure. Can this computer-vision technology act as a screening, or triage, technology? Can it perform more consistently than routine microscopy? Can it deliver high performance and identify the most-likely positive cases that can be confirmed by a molecular test? If so, such a triage technology could tightly focus the use of molecular tests on these specific cases, incurring diagnostic expenses where they should be borne and providing a balance of the best performing AND most affordable combinations of these technologies!
In using the evaluation results to answer these questions, a study team of renown industry professionals, concluded:
- As a stand-alone test TBDx™ provides a level of sensitivity and specificity that would be equivalent to microscopists with 80 years of experience, or more. Nazir noted that having microscopists with this level of experience is both extraordinary and almost impossible to replace.
- TBDx™ performance shows it can be a triage tool for GeneXpert, and can achieve high sensitivity:
- Sensitivity of 78%; Specificity of 98.8%, with 75% fewer GXP tests.
- Sensitivity of 73%; Specificity of 99.3%, with 90% fewer GXP tests.
- The TBDx™ performance occurred among a low prevalence cohort (10%). It is very possible the TBDx™ performance will be greater in a higher prevalence setting.
- One limitation of the study was the high performing standard of the microscopists participating in the study, which is well above the accepted norm. For example, comparing TBDx™ to a standard microscopist performance (60%), TBDx™ would show sensitivity improvements of 25%-30%, with equivalent specificity.
These conclusions led the study team to recommend:
Demonstrations studies, similar to evaluations on molecular tests in recent years, should be conducted broadly on TBDx™ to test and confirm these findings. The evaluators can see several potential applications:
- If a laboratory can test ALL TB suspects with GeneXpert, use TBDx™ automated microscopy to monitor treatment.
- If a laboratory has a limited budget for GeneXpert, and drug resistance is not a concern, use TBDx™ automated microscopy on ALL TB suspects and use GeneXpert to confirm scanty cases. If drug resistance is a concern use GeneXpert on all positive cases identified by TBDx™ automated microscopy.
- If there is no budget for GeneXpert, use TBDx™ automated microscopy throughout, or use it in central locations.
In our upcoming posts we will provide a link to the slides presented by Dr. Ismail, and we will further detail the performance results. We will describe the enthusiastic response to the technology demonstrations. We will outline what we learned, with whom we spoke, and identify a specific quality control challenge in laboratories today that TBDx™ could address immediately. There is much to report, and a series of articles will give us a chance to develop the details that readers want. One post cannot treat all of it adequately.
So, please stay tuned to our blog! There is so much more to convey! For now, what we want our readers to know is that TBDx™ has bridged the gap between the theoretical to the possible. Automated sputum microscopy for the detection of TB is a reality, the technology performs at a high level, and there is scientific evidence that proves it!
The independent study team will present further details of the results in a peer-reviewed scientific journal, which will be forthcoming shortly.