Surveillance Data Point the Way to Light-Based Diagnostics for Drug-Resistant Tuberculosis in Rural South Africa

In the rural Eastern Cape Province of South Africa, where drug-resistant tuberculosis (DR-TB) remains one of the most stubborn public health challenges, a team of researchers has taken an unusual approach to diagnostic innovation. Rather than starting in the laboratory, they started with the data already flowing through the local health system. In a study published in Frontiers in Tuberculosis, researchers from Walter Sisulu University and the Council for Scientific and Industrial Research (CSIR) analyzed routine surveillance records from 402 patients treated between 2020 and 2024 at two decentralized DR-TB facilities, Mthatha Gateway Clinic and Holy Cross Clinic, and used the findings to build a framework for deciding which genetic targets a future point-of-care diagnostic should detect first.

The scale of the global problem provides the backdrop. According to the World Health Organization, approximately 10.8 million people developed tuberculosis and 1.25 million died from the disease in 2023, and drug-resistant strains continue to undermine progress toward the End TB Strategy. Rifampicin-resistant tuberculosis (RR-TB), multidrug-resistant tuberculosis (MDR-TB), pre-extensively drug-resistant tuberculosis (pre-XDR-TB), and extensively drug-resistant tuberculosis (XDR-TB) all demand complex, prolonged treatment and are associated with higher costs, worse clinical outcomes, and ongoing transmission. Rapid and accurate identification of resistance is therefore central to controlling the epidemic, yet in rural settings access to molecular testing can be constrained by laboratory capacity, shortages of trained personnel, specimen transport logistics, and infrastructure limitations.

The Eastern Cape illustrates these constraints vividly. Its communities are geographically dispersed, and specialized diagnostic services may be far away. Molecular technologies such as Xpert MTB/RIF Ultra and line probe assays have improved TB detection nationally, but the researchers argue that decentralized, point-of-care approaches could complement existing laboratory-based services where centralized testing falls short. Their study, led by Ncomeka Sineke and colleagues including senior author Lindiwe Modest Faye, set out to answer a deceptively simple question: if you could build only a small, affordable diagnostic panel for this region, which resistance genes should it target first?

To answer it, the team conducted a retrospective secondary analysis of routinely collected programme data drawn from the Electronic Drug-Resistant Tuberculosis Register (EDRWeb), facility treatment registers, district reports, and laboratory records. The dataset covered patients with bacteriologically confirmed DR-TB who initiated treatment between January 2020 and December 2024. Resistance categories followed established WHO definitions: RR-TB for resistance to rifampicin with or without other drugs, MDR-TB for resistance to at least rifampicin and isoniazid, pre-XDR-TB for MDR/RR-TB with additional fluoroquinolone resistance, and XDR-TB for MDR/RR-TB that is also resistant to bedaquiline or linezolid.

The headline finding was the overwhelming dominance of first-line drug resistance. RR-TB accounted for 191 of the 402 cases (47.5 percent) and MDR-TB for 178 cases (44.3 percent), meaning these two categories together made up 91.8 percent of the DR-TB burden. Pre-XDR-TB represented just 23 cases (5.7 percent) and XDR-TB only 10 cases (2.5 percent). This distribution mattered enormously for the prioritization exercise, because it suggested that a diagnostic capable of detecting rifampicin and isoniazid resistance would address the vast majority of locally relevant resistance, while fluoroquinolone and second-line resistance detection could follow in later iterations.

The researchers formalized this reasoning in a conceptual tool they call the Diagnostic Priority Index (DPI), which weighs five domains: disease burden, frequency of resistance categories, treatment outcome vulnerability, programme impact, and potential suitability for decentralized implementation. Resistance categories accounting for at least 40 percent of the caseload were deemed high epidemiological priority, those between 5 and 39 percent medium priority, and those below 5 percent candidates for future expansion. The authors are careful to note that these thresholds were pragmatic, study-specific decision boundaries rather than statistically validated cutoffs. Applying the framework, the genes rpoB, katG, and inhA emerged as high-priority targets, because mutations in rpoB drive most rifampicin resistance while katG and the inhA promoter region are the principal determinants of isoniazid resistance. The medium-priority tier comprised gyrA, gyrB, rrs, and eis, which are associated with fluoroquinolone and injectable drug resistance, while embB, pncA, Rv0678, atpE, and pepQ were flagged for future expansion.

The surveillance data also revealed a troubling trend in treatment outcomes. Among patients with MDR-TB, treatment success declined from 80.9 percent in 2021 to 38.5 percent in 2024, while the proportion of outcomes classified as not evaluated rose from zero to 30.8 percent over the same period. Treatment success among RR-TB patients fell from 79.5 percent to 51.6 percent, with unevaluated outcomes climbing to 25.8 percent. The researchers interpret these patterns as signals of programme vulnerability, but they are emphatic that the data do not permit causal attribution to diagnostic limitations. DR-TB outcomes are shaped by many factors, including HIV status, treatment regimens, adherence, loss to follow-up, comorbidities, socioeconomic circumstances, and healthcare access, none of which could be assessed from the aggregated records.

The second half of the framework is where the study earns its molecular-biophotonic label, though the authors stress that this component is entirely conceptual. They mapped the prioritized genes onto candidate optical biosensing technologies drawn from the published literature: surface plasmon resonance (SPR), localized surface plasmon resonance (LSPR), fluorescence-based biosensing, surface-enhanced Raman spectroscopy (SERS), photonic crystal biosensors, and integrated optical waveguide biosensors. These approaches exploit optical phenomena to detect nucleic acids or biomolecules, and published studies suggest some offer advantages in analytical sensitivity, multiplexing capability, miniaturization, and potential portability. In the conceptual compatibility assessment, SPR, LSPR, and fluorescence-based platforms appeared compatible with all three high-priority targets, while SERS and photonic crystal sensors could accommodate high- and medium-priority targets, and waveguide biosensors could theoretically span the full panel.

The limitations of the study are as important as its contributions, and the authors enumerate them with unusual candor. No isolate-level mutation analysis or sequencing was performed, so the study cannot report the actual frequency of specific mutations in the population; the gene priorities reflect the distribution of resistance categories combined with established resistance mechanisms from the literature. No biosensor was fabricated or tested, and no analytical sensitivity, specificity, limit of detection, reproducibility, diagnostic accuracy, or clinical feasibility was evaluated. The DPI thresholds were exploratory rather than externally validated, and the two-facility setting may limit generalizability to regions with different resistance patterns. The authors also acknowledge that the dataset lacked bacterial load, diagnostic turnaround times, detailed phenotypic susceptibility results, and patient-level clinical characteristics.

What the study offers, then, is a hypothesis-generating translational pathway rather than a finished technology. The researchers recommend that future work begin with mutation-level characterization of the local resistance spectrum using targeted or whole-genome sequencing, followed by laboratory development and analytical validation of candidate assays starting with rpoB, katG, and inhA, and then prospective clinical validation against reference methods such as Xpert MTB/RIF Ultra, line probe assays, and phenotypic drug-susceptibility testing. Implementation studies would ultimately need to assess cost, workflow integration, quality assurance, acceptability, and compatibility with national TB programmes before any claims about faster diagnosis or better outcomes could be made. For now, the value lies in the demonstration that routine surveillance data, often collected merely for programme monitoring, can be repurposed to steer early-stage diagnostic research toward the targets that matter most in the communities where the burden is heaviest.

Subject of Research: Epidemiology-guided prioritization of molecular targets for drug-resistant tuberculosis diagnostics in rural Eastern Cape, South Africa

Article Title: Translating drug-resistant tuberculosis surveillance into molecular–biophotonic diagnostic innovation in rural Eastern Cape, South Africa

Article References: Sineke, N., Sabisa, K., Dlatu, N., Ombinda-Lemboumba, S., Apalata, T., & Faye, L. M. (2026). Translating drug-resistant tuberculosis surveillance into molecular–biophotonic diagnostic innovation in rural Eastern Cape, South Africa. Frontiers in Tuberculosis, 4, Article 1937755. https://doi.org/10.3389/ftubr.2026.1937755

Image Credits: AI Generated

DOI: 10.3389/ftubr.2026.1937755

Keywords: drug-resistant tuberculosis, molecular diagnostics, biophotonics, point-of-care testing, South Africa, Eastern Cape, surveillance, rpoB, katG, inhA, biosensors, MDR-TB