With bacteriological diagnosis of TB (via sputum samples) in children, particularly in those aged below five years, being riddled with problems, scientists are looking at alternative methods that do not rely on sputum samples. A proof-of-concept study published recently in The New England Journal of Medicine shows promise. The sample used was blood.

Suzanne T. Anderson from the Brighton and Sussex Medical School, University of Sussex and others could confirm TB disease by looking for specific signatures in 51 genes. And the presence of these signatures in 42 genes could help distinguishing TB disease from latent TB. The signatures in question were RNA transcripts.

To test the robustness of the signatures in correctly diagnosing TB disease and differentiating latent TB from TB disease, the scientists had a discovery cohort (where TB was suspected) comprising of 655 South African and 701 Malawian children. A validation cohort (children with one or more typical clinical symptoms and were household contacts of people with smear positive pulmonary TB disease) comprised 1,599 Kenyan children.

In the case of the culture-positive group, the sensitivity was 83 per cent and specificity was 84 per cent for distinguishing TB from other diseases similar to TB. But in the case of the culture-negative group, the sensitivity of the new technique varied widely — 35 to 63 per cent in the different groups (“that are based on the probability of a child having TB”). However, the specificity was the same — 84 per cent.

“One issue is that the ‘true’ proportion of TB in the culture negative group is unknown, as there is no gold standard (culture) [to compare with],” Dr. Michael Levin, Professor of International Child Health, Imperial College London, and a co-author of the paper said in an email to this Correspondent. “We describe a method in the manuscript for estimating the proportion of ‘true’ TB in the culture negative group. And if we correct for the ‘true’ TB rate in the group, then the sensitivity is above 80 per cent.”

What is worth taking note of is that unlike the dismal sensitivity of smear microscopy, the RNA-transcript signatures had higher sensitivity in those aged below five years.

Children under five years of age have great difficulty in producing a sputum sample for examination and even if sputum is collected, there are very few TB bacilli in the sputum (paucibacillary). According to him, the majority of the children, as well as the TB patients were under five years of age. In the discovery cohort, from the 112 TB patients, 57 were under five years of age, 29 between five and 10, and 23 over 10.

Besides smear microscopy, all currently available advanced techniques (like GeneXpert) rely on sputum samples for confirming TB disease.

By relying on blood samples, the new technique, if it passes the litmus test, would be able to side-step the inherent drawbacks of sputum samples and correctly diagnose more children with TB disease. “Yes, we found our RNA-based risk score had a much greater sensitivity than Xpert in both culture-confirmed and culture-negative cases,” he noted. The scientists are planning to undertake further validation in other populations.

A word of caution

Dr. Madhukar Pai, Associate Professor at McGill University, Montreal, sounds a word of caution. “This new study adds to the already existing literature on the potential value of transcriptome signatures in identifying tuberculosis. While the data are promising, the big challenge will be to translate this basic science work into a commercial product that can be used in high burden countries,” Prof. Pai said in an email to this Correspondent. “The technology required is expensive and complex, and thus much more work is needed to convert this into a clinical test. Needless to say, because initial results are invariably over-optimistic, further validation is needed in real-world settings.”

Published in The Hindu on May 15, 2014