Could a blood microRNA test help diagnose tuberculosis?
A few small RNA molecules were raised in the blood of people with active tuberculosis and could separate them from healthy people moderately well, but the finding is preliminary.
Source
Altered serum microRNAs as biomarkers for the early diagnosis of pulmonary tuberculosis infection
Study at a glance
- Design
- Case-control — Two-stage diagnostic case-control study: pooled-serum microRNA array screen, then individual-sample qRT-PCR validation with ROC analysis
- N
- Participant counts are inconsistent in the text: Methods list 155 participants (30 active TB, 65 healthy controls, plus 60 paediatric sera with other infections), while Results say 65 participants including 30 TB patients and 65 controls; the array stage used pools of 20 TB and 20 control sera
- Population
- HIV-negative adults with untreated smear-positive pulmonary TB and healthy check-up attendees in Jiangsu Province, China; comparison pools from children with pertussis, varicella-zoster or enterovirus infection
- Outcome
- Serum microRNA levels and their ability to distinguish TB from controls (area under the ROC curve)
Structured fields used in claim comparison tables when every cited study has a complete layer.
What they did
The researchers pooled serum from 20 patients with confirmed, untreated pulmonary TB and 20 healthy controls and screened 667 microRNAs on an array to find ones that differed. They picked the ten most strongly raised candidates and measured them in individual samples by quantitative PCR, then used ROC curves to judge how well each one, and a combination, distinguished TB patients from controls. Serum pools from children with three other infections were used to check whether the signals were specific to TB.
What they found
The array flagged 97 microRNAs that differed at least two-fold (90 higher and 7 lower in TB). Seven of the ten candidates were confirmed as raised by PCR, but only three separated the groups reasonably: miR-361-5p (AUC 0.848), miR-889 (AUC 0.765) and miR-576-3p (AUC 0.711). Combining the three barely improved accuracy (AUC 0.863), and these three also differed between TB and the other-infection pools.
The limits
What it doesn't show
This is an early discovery study, not proof of a usable test: cases were clear-cut smear-positive patients compared with healthy people, which tends to overstate accuracy compared with real clinical use, and no separate validation cohort was tested. Latent TB and patients with other lung diseases were not studied, and the other-infection comparison used pooled serum from children rather than individual adults. The reported participant numbers are internally inconsistent, and the authors themselves describe the diagnostic ability as only moderate and call for larger studies.
Key terms
- MicroRNA (miRNA)
- A short non-coding RNA that regulates gene expression; some circulate stably in blood and can act as disease biomarkers.
- ROC curve and AUC
- A plot of sensitivity against 1 minus specificity across all cut-offs; the area under it, ranging from chance level to perfect, summarises how well a marker separates groups.
- qRT-PCR
- Quantitative reverse-transcription PCR, which measures how much of a specific RNA is present in a sample.
- Spectrum bias
- Inflated accuracy that occurs when a test is evaluated in clear-cut cases versus healthy controls rather than in the full range of patients seen in practice.
- Endogenous control
- A reference molecule assumed to be stable across samples (here miR-16) used to normalise measurements.
Flashcards
0 of 10 answers reviewed
Research intelligence for this paper
See its role on concept claims, tensions it is part of, placement history, and related discoveries.
Quiz yourself
What design did this microRNA study use?
Common questions
Why test blood microRNAs at all when sputum culture exists?
Culture takes weeks and smear microscopy misses many cases, so a fast blood test could speed diagnosis if it proved accurate.
Is an AUC of 0.848 good enough to diagnose TB?
Not by itself; it indicates moderate discrimination, and in a case-control design with healthy controls real-world accuracy is likely to be lower.
Why did some array hits fail in the PCR check?
Screening on pooled samples can produce false leads, so individual-sample validation is needed; three of ten candidates, including miR-144*, did not differ significantly.
More on Diagnostic accuracy