TL;DR
This study uses whole genome sequencing (WGS) data from over 8,500 Mycobacterium tuberculosis isolates to reconstruct the historical acquisition of drug resistance across the globe. They found a consistent order of resistance development, with isoniazid resistance appearing first, and identified that wealthier nations tend to have older, established resistance patterns, suggesting successful programmatic interventions can slow amplification, but that fit resistant strains persist.
What Problem Does This Paper Address?
Tuberculosis (TB) remains a major global health threat, and the rise of drug-resistant strains (MDR-TB and XDR-TB) significantly hinders efforts to control the epidemic. Understanding how, when, and where drug resistance emerges and spreads is crucial for developing effective public health strategies. While previous studies have looked at resistance, this paper aims to provide a comprehensive, global, and temporal analysis by leveraging a large dataset of whole genome sequences and phenotypic drug susceptibility data to untangle the complex interplay of resistance acquisition and transmission.
Methods
The researchers employed a retrospective analysis of whole genome sequencing (WGS) data from clinical M. tuberculosis isolates. A key technique was Bayesian molecular dating using BEAST, a method that infers evolutionary relationships and estimates the timing of past events, in this case, the acquisition of drug resistance. They constructed phylogenetic trees and used these to estimate the age of the "most recent common susceptible ancestor" (MRSCA) for resistance to various drugs. This allowed them to date when resistance likely first emerged in different geographical locations and for different drug classes. They also analyzed the geographical distribution of M. tuberculosis lineages and correlated resistance acquisition timing with economic indicators like GDP per capita.
Datasets
The study utilized a curated dataset of 10,299 M. tuberculosis clinical isolates with both WGS and culture-based drug susceptibility test (phenotypic) data. These isolates were sourced from public databases (Patric and ReSeqTB) and literature curation, spanning 52 countries across Africa, Asia, the Americas, and Europe. Of these, 8,550 isolates met the quality and origin criteria for the analysis. The study also referenced data from commercial diagnostic tests (Hain Line-probe and GeneXpert) for comparison.
Key Insights and Ideas
- A Consistent Order of Resistance: The study provides strong evidence for a consistent global order of drug resistance acquisition: isoniazid resistance typically arises first, followed by rifampicin and streptomycin, then ethambutol, and subsequently other drugs like pyrazinamide, fluoroquinolones, and second-line injectables. This suggests a predictable evolutionary pathway for resistance.
- Wealthier Nations, Older Resistance: A striking finding is the correlation between a country's gross domestic product (GDP) and the age of rifamycin resistance. Wealthier nations tend to have older resistance patterns (MRSCA >20 years ago), implying that robust public health programs and effective interventions can slow down the amplification of resistance, even if it has already emerged.
- Transmission is a Major Driver: The number of independent resistance acquisition events was consistently lower than the number of resistant isolates. This strongly suggests that ongoing transmission of already resistant strains plays a significant role in the spread of drug resistance, rather than solely new mutations arising in individual patients.
- Second-Line Drug Resistance is Recent: Resistance to fluoroquinolones and second-line injectable drugs, often seen in extensively drug-resistant (XDR) TB, appears to have emerged much more recently, with a median ancestral age of only 4.7 years before sample collection. This highlights the rapid evolution and spread of resistance to these critical drugs.
- Diagnostic Gaps Exist: The sensitivity of commercial molecular diagnostics for detecting second-line drug resistance varies significantly by country, indicating a need for geographically tailored diagnostic approaches.
Key Results
- 8,550 isolates from 52 countries were analyzed.
- 6,099 isolates were included in the molecular dating analysis.
- 4,860 MRSCA dates were estimated for ten drugs across 179 country-lineage-drug groups.
- Isoniazid resistance was dated significantly earlier than rifampicin resistance (median MRSCA 11.4 years vs. 7.6 years ago, respectively).
- Recent resistance amplification (within 5 years) to pyrazinamide and ethambutol was observed in several countries, with rates ranging from 2% to 33%.
- A significant correlation (r²=0.47, p=0.014) was found between GDP per capita and older rifamycin MRSCA dates.
- The study estimated that 14-52% of the resistance burden could be attributed to transmission.
Limitations and Future Work
The authors acknowledge several limitations:
- Oversampling of drug-resistant isolates: The study sample was enriched for drug-resistant isolates, which might skew some prevalence estimates compared to general population data.
- Inability to distinguish importation vs. new acquisition: The analysis could not definitively differentiate between resistance imported through migration and new resistance emerging locally. This means transmission burden estimates are likely underestimates.
- Reliance on phenotypic data: The assumption of accuracy in culture-based phenotypic drug susceptibility tests, despite known variability, was necessary.
- Scarcity of immigration data: This limited the ability to draw conclusions about specific countries and assess cross-border transmission.
Future work could involve reassessing these patterns with WGS data that is now being systematically generated through routine TB surveillance and resistance diagnosis. Further refinement of diagnostic prediction models using variants outside of commercially targeted regions is also suggested. The study also points to the need for continued investment in TB care, surveillance, and the development of geographically relevant diagnostics.
Summary updated