ӳ����ý

Tuberculosis (TB) can be cured and yet it kills more people worldwide than any other infectious disease. As many as a third of patients with active TB in Africa and Asia aren’t diagnosed and treated, furthering the spread of the disease and underscoring the need for better and simpler diagnostic tests.  

Now, an international research team led by scientists at the ӳ����ý of MIT and Harvard, the Wyss Institute at Harvard University and Brigham and Women’s Hospital has developed a blood-based test that can accurately distinguish patients at increased risk for active TB from those likely to have other lung problems, in geographically diverse regions, including Africa, Asia, and South America.

The new test, described in a in Science Translational Medicine, approaches the performance criteria established by the World Health Organization (WHO) for TB triage tests. The study is also a critical step toward a simple, low-cost blood test that can be used anywhere, including low-income areas, for triaging patients suspected of having active TB.  

“There is an overwhelming need for a field-deployable test for TB — one that does not require extensive laboratory or other infrastructure, can be used by local care providers across the globe, and works on blood samples to stratify at-risk populations or to make a definitive diagnosis,” said senior author Michael Gillette, a senior scientist at the ӳ����ý and a critical care physician at Massachusetts General Hospital. “We aren’t there yet by any means, but our study moves us fundamentally closer to that goal.

from on .

Rapid diagnosis remains a significant challenge, especially in low-resource settings and areas where TB infection rates are high, because current TB tests either rely on expensive lab-based equipment or on decades-old methods that are error-prone and slow to yield results. Moreover, current leading methods require sputum samples, which can be difficult for young children and immunocompromised patients to provide. Some blood tests have tried to rapidly identify patients who most likely have TB and should get tested further, but those tests are not accurate enough.

To address these challenges, Gillette and his colleagues, including Steven Carr, director of the ӳ����ý’s Proteomics Platform, and co-senior author David Walt of Brigham and Women’s Hospital and the Wyss Institute at Harvard University, set out to develop a simple blood test that could distinguish TB from other, more common conditions with similar symptoms, like pneumonia. This test could potentially lower the barrier for TB testing by identifying patients who most likely have the disease and can then receive additional, more definitive diagnostic tests.

“Blood contains most of the proteins that are in the body,” said Carr. “So, to a great extent, this was a fishing expedition where we were trying to find protein signatures that signaled the presence of TB.”

The team began their search with a panel of likely suspects: 47 different immune system signaling proteins and other markers of infection and inflammation. The researchers measured the levels of these proteins in blood samples drawn from patients in the Philippines and Tanzania — 199 people with active TB and 188 with other TB-like diseases. These measurements of patient samples began as part of another global health project, funded by the Bill and Melinda Gates Foundation, that first launched over a decade ago. 

The authors used an antibody-based method to measure the proteins, because antibodies are a key part of many current clinical tests that allow the tests to be configured for use in care settings without the need for advanced equipment. 

Gillette and his colleagues, including first author Rushdy Ahmad of the ӳ����ý (now at True North Bio), mined the data from their initial 47-protein panel, using machine-learning algorithms to look, in part, for the proteins most strongly associated with active TB. 

The researchers narrowed their list of proteins to four markers of inflammation: interleukin-6 (IL-6), IL-8, IL-18, and vascular endothelial growth factor. The team also added a fifth protein marker, an antibody against a TB protein called Ag85B, which improved the panel’s accuracy in follow-up tests.

To further boost the panel’s performance, the team transitioned to a new antibody-based technology called Simoa, which was pioneered by Walt and his colleagues. The highly sensitive technique harnesses single molecule arrays to detect much lower levels of protein than conventional methods. 

TB or not TB

Using their technology, the researchers analyzed a blinded, independent set of more than 300 patient samples from Vietnam, South Africa, and Peru from the FIND repository (Foundation for Innovative New Diagnostics). The five-protein panel distinguished active TB from other similar conditions with 86 percent sensitivity and 69 percent specificity, which are just shy of the WHO’s recommended minimal performance criteria for new TB triage diagnostics tests of 90 percent sensitivity and 70 percent specificity.

Before the new test can be used clinically to triage patients, it needs to be studied in a much larger group of patients. Also, ongoing enhancements of the Simoa technology will make the test suitable for use in low-resource settings and lower the cost to a target of $2 per test. Importantly, if Gillette and his colleagues can further improve the test’s sensitivity and specificity, it could potentially be used to diagnose patients with TB rather than simply triage them.

“We’re on the cusp of the WHO recommendations and our test has held up in more than 700 patients across multiple continents including people who are HIV-positive or negative, different health contexts, and various strains of TB,” said Gillette. “But we haven’t evaluated the test prospectively, and there’s still more work to do to improve the test. Ideally, we want a test that’s powerful enough to be a true diagnostic.”

Funding for this study was provided in part by a Grand Challenge in Global Health grant from the Bill and Melinda Gates Foundation for the study titled “Population Health Metrics Research Consortium Project (contract no. 041222241).