Study identifies patients at higher risk from resistant tuberculosis

An analysis of tuberculosis programmes and clinical trials has identified a group of patient characteristics associated with poorer outcomes in multidrug- or rifampicin-resistant tuberculosis. The findings, reported by Nature Communications, were based on records from 7,750 people and were used to create a score for estimating treatment risk.

The study found that people living with HIV, older patients, those with a low body mass index and smokers were more likely to experience an unfavorable clinical outcome. The same association was seen among patients whose samples were positive for acid-fast bacilli, as well as those with disease outside the lungs or cavities in the lung tissue.

Together, these characteristics gave the researchers a way to separate patients into risk groups. People placed in the low-risk category had a lower probability of an unfavorable result as treatment progressed than those classified as high risk. The difference between the groups was statistically significant, with p < 0.001.

A difficult form of tuberculosis

Tuberculosis remains the deadliest infectious disease, according to the article. Its drug-resistant forms pose an additional challenge because the medicines used to treat them can be both toxic and difficult to take over long periods.

Multidrug-resistant tuberculosis is resistant to important anti-tuberculosis medicines, while rifampicin-resistant tuberculosis is resistant to rifampicin, one of the drugs used in standard treatment. The analysis focused on the patient-level factors that could help distinguish people more likely to respond well from those who may need closer attention.

The researchers assessed how well the combined risk information separated more and less favorable outcomes. Their model produced a receiver operating characteristic area under the curve of 0.728, with a 95% confidence interval of 0.726 to 0.731. This measure describes the model’s ability to discriminate between different outcome groups; the reported interval indicates the estimated performance in the analysed population.

The value of the approach lies in bringing several clinical features together rather than considering them in isolation. HIV status, nutritional state, smoking history, the laboratory result from a smear, the location of infection and the appearance of lung disease each contributed to the overall patient profile. Age was also included as part of the assessment.

Evidence from a six-month regimen

The researchers then tested the risk categories against an external group treated with a highly effective six-month BPaL regimen. In that validation analysis, treatment succeeded for 97% of patients in the low-risk group. Among patients in the medium- and high-risk categories, the success rate was 88%.

The difference does not mean that people in the higher-risk groups cannot be cured. Instead, it shows that the score was able to identify groups with different observed probabilities of success, even when they were assessed using a treatment regimen that performed strongly overall.

External validation is important because a prediction method can appear useful in the population from which it was developed but perform differently in another set of patients. Applying the score to people treated with the six-month regimen provided a separate test of whether the risk pattern remained meaningful beyond the original analysis.

Toward more tailored care

The findings support using patient risk phenotypes to guide tuberculosis interventions. A risk profile could help clinicians recognise patients who are more likely to have an unfavorable course and direct greater attention towards their treatment, while also identifying those with a lower predicted risk.

The article does not present the score as a replacement for treatment. Rather, its conclusion is that separating patients by risk may make treatment planning more effective and more personalised. That could be especially relevant in drug-resistant tuberculosis, where therapy is already lengthy and toxic and where the consequences of an unfavorable outcome are serious.

The analysis also offers a framework for comparing outcomes across programmes and clinical trials. By recording the same patient characteristics and assigning people to comparable risk categories, health teams may be better able to interpret treatment results and determine where additional support is needed.

According to Nature Communications, this kind of risk stratification is central to improving cure rates. The study’s results indicate that patient-level information can be used to distinguish treatment prospects and support more targeted tuberculosis care.