Transmission of rifampicin-resistant tuberculosis in Ho Chi Minh City, Viet Nam: a prospective genomic epidemiology study

This prospective genomic epidemiology study in Ho Chi Minh City reveals that the majority of rifampicin-resistant tuberculosis cases are driven by the ongoing transmission of already-resistant strains rather than *de novo* resistance acquisition, highlighting the need for strategies that combine early diagnosis to interrupt transmission with efforts to identify and control the drivers of newly acquired resistance.

The Gist

The Big Picture: A City Under Siege by a "Super-Bug"

Imagine Ho Chi Minh City as a massive, bustling metropolis. In this city, there is a dangerous enemy called Tuberculosis (TB). But this isn't just any TB; it's a "super-bug" version that has learned to ignore its most powerful weapon: a medicine called Rifampicin.

The big question the researchers wanted to answer was: How is this super-bug spreading?

There are only two ways this happens:

1. The "Bad Neighbor" Theory (Transmission): You catch the super-bug directly from someone else who already has it.
2. The "Bad Recipe" Theory (Acquired Resistance): You catch a normal, weak TB bug, but because your treatment went wrong (maybe you missed doses or the drugs weren't strong enough), the bug mutated inside your body and became a super-bug.

The researchers wanted to know: Is the city mostly being overrun by bad neighbors passing the bug around, or is the "bad recipe" (treatment failure) creating new super-bugs constantly?

---

The Investigation: Genetic Fingerprinting

To solve this mystery, the researchers didn't just ask patients questions. They acted like genetic detectives.

They collected samples from nearly 1,500 patients over four years. They took the bacteria from these patients and looked at their DNA fingerprints (Whole Genome Sequencing).

Think of it like this:

• If two people have TB and their bacterial DNA fingerprints are almost identical, they likely got it from the same source (Transmission).
• If the fingerprints are totally different, or if the "super-power" (resistance) appeared suddenly in one person's bacteria that didn't have it before, it likely happened inside that person's body (Acquired).

They also looked at the patients' medical histories. Had they been treated for TB before? If yes, did the super-bug appear after that treatment?

---

The Findings: What They Discovered

1. The "Bad Neighbor" is the Main Culprit (72% - 87%)

The study found that most of the super-bug cases in Ho Chi Minh City are due to transmission.

• The Analogy: Imagine a game of "telephone" where a bad rumor spreads. Most people in the city aren't inventing the rumor; they are just hearing it from someone else.
• The super-bug strains are so successful that they have been spreading around the city for decades. Some of these "families" of bacteria started spreading as far back as the 1980s and are still active today. They have become like a long-running, city-wide family tree of super-bugs.

2. The "Bad Recipe" is Still a Problem (13% - 28%)

Even though the city has good treatment programs, about 1 in 4 to 1 in 7 cases are still happening because the treatment failed.

• The Analogy: Imagine a chef (the doctor) giving a recipe (medicine) to a cook (the patient). Sometimes, the cook doesn't follow the recipe perfectly, or the ingredients are slightly off, and the dish turns out burnt (resistant).
• The study found that people with diabetes were more likely to develop this "bad recipe" resistance. It suggests that even with good drugs, some people's bodies or other health issues make it harder to kill the bug before it evolves.

3. The Spread is Everywhere, Not Just in Houses

A common assumption is that TB spreads mostly within families living in the same house.

• The Analogy: You might think the fire is spreading from room to room in a single house.
• The Reality: This study found the fire is spreading across the whole city. The "super-bug families" are scattered all over Ho Chi Minh City. People who are related by bacteria often live in completely different neighborhoods and commute to work in different areas. It's not a "household" problem; it's a "city-wide" problem.

---

The Takeaway: What Should We Do?

The researchers concluded that to stop this super-bug, we need a two-pronged attack:

1. Stop the Spread (The Main Priority): Since most cases come from transmission, we need to find sick people sooner and treat them faster. If we catch the "bad neighbors" before they pass the bug to others, we can break the chain. It's like putting out the fire before it jumps to the next building.
2. Fix the "Bad Recipes": We also need to figure out why some people are still developing resistance despite treatment. Is it because of hidden health issues like diabetes? Is the treatment not working for certain types of bacteria? We need to fix the "recipe" so the bug doesn't evolve inside the patient.

In a Nutshell

The city is fighting a war against a super-bug that is mostly being passed from person to person across the whole city, rather than just inside homes. While the city's treatment programs are good, they aren't perfect, and a significant number of new super-bugs are still being created by treatment failures. To win, the city needs to be faster at catching the spread and smarter about why treatment sometimes fails.

Technical Summary

1. Problem Statement

Rifampicin-resistant tuberculosis (RR-TB) remains a critical public health challenge in Viet Nam, with approximately 10,000 incident cases annually. A central epidemiological question is whether this burden is driven primarily by the transmission of pre-existing resistant strains or by the de novo acquisition of resistance mutations during inadequate treatment of susceptible TB.

• Context: While Viet Nam has a mature National TB Programme (NTP) with DOTS strategies and fixed-dose combinations designed to prevent acquired resistance, historical data suggests most RR-TB patients have a history of prior TB treatment.
• Gap: Previous studies often relied on simple SNP-based clustering thresholds, failed to account for incomplete sampling, and lacked integration of clinical treatment history with phylogenetic inference. Consequently, the relative contributions of transmission vs. acquired resistance in Viet Nam were unknown.

2. Methodology

The study employed a prospective, city-wide genomic epidemiology design with advanced phylogenetic and statistical modeling.

• Study Population & Sampling:

  • Setting: Ho Chi Minh City (HCMC), Viet Nam.
  • Period: March 2020 – April 2024.
  • Cohort: 2,033 adults newly diagnosed with pulmonary RR-TB via Xpert MTB/RIF.
  • Sequencing: 1,491 isolates (64% of estimated true RR-TB cases) were successfully cultured and whole-genome sequenced (WGS).
  • Data Integration: Clinical data (demographics, treatment history, comorbidities) and geocoded residential addresses were collected.

• Phylogenetic & Evolutionary Analysis:

  • Dataset: Combined 1,491 new isolates with 2,922 previously published HCMC genomes and 3,050 contemporaneous sequences (total $N=5,718$).
  • Classification Strategy: Resistance origin was classified using a dual approach:
    1. Clinical History: Patients with no prior TB were automatically classified as "transmitted."
    2. Phylogenetic Placement: For patients with prior TB, resistance mutations ($rpoB$) were mapped to the phylogeny.
       • Approach N: All descendants of an internal resistance node are classified as transmitted (lower bound for acquired).
       • Approach N-1: One descendant is classified as acquired (if they had prior TB), and the rest as transmitted (upper bound for acquired).
  • Dating: Bayesian phylogenetic dating (BactDating) was used to estimate the timing of resistance emergence events for major lineages (1, 2, and 4).

• Statistical Corrections & Uncertainty:

  • SIMEX (Simulation-Extrapolation): To correct for the 36% of cases not sampled, the authors simulated progressively sparser sampling fractions (8%–92%) and extrapolated estimates to 100% coverage using Generalized Additive Models (GAMs).
  • Bootstrap Resampling: 100 iterations were performed to account for phylogenetic and sampling uncertainty.
  • Network Analysis: Transmission networks were analyzed for geographic dispersion (using commute times via OSRM) and demographic structuring (using treeSeg and Mantel tests).

3. Key Contributions

• Methodological Rigor: The study moves beyond simple SNP clustering by integrating clinical history, homoplasy-aware phylogenetic inference, and explicit correction for incomplete sampling coverage (SIMEX).
• Granular Resolution: It provides the first population-level estimates distinguishing transmitted vs. acquired RR-TB in a high-burden Vietnamese setting.
• Evolutionary Context: It reconstructs the temporal history of resistance, dating the emergence of resistant clones back to the 1980s.
• Spatial Dynamics: It utilizes commute-time-based analysis to demonstrate that transmission is city-wide and diffuse rather than localized to specific neighborhoods or households.

4. Key Results

• Transmission vs. Acquisition:
  • After correcting for sampling coverage, 72% to 87% of RR-TB cases in HCMC are attributed to the transmission of already-resistant strains.
  • 13% to 28% of cases arise from de novo acquired resistance, a substantial minority despite long-standing programmatic controls.
• Risk Factors for Acquired Resistance:
  • Diabetes: Significantly associated with higher odds of acquired resistance (OR 1.58).
  • Lineage: Infection with Lineage 2 was associated with lower odds of acquired resistance compared to transmission.
  • Prior TB: Among those with prior TB, ~36% of RR-TB cases were estimated to be acquired de novo.
• Temporal Emergence:
  • Resistance emergence events occurred repeatedly from the 1980s to the present.
  • The earliest emergence in Lineage 2 was dated to 1980 (95% CI: 1970–1989), coinciding with the introduction of rifampicin in Viet Nam (1986).
  • Early resistance events seeded long-lived, city-wide transmission networks that persist today.
• Spatial and Demographic Structure:
  • Geographic Dispersion: Transmission networks are geographically dispersed across the city. There is no correlation between genetic relatedness and geographic distance (Mantel $r = 0.001$).
  • Household Clustering: Only 2.6% of cases shared a household, and in less than half of these, the strains belonged to the same transmission network.
  • Demographics: Transmission is not strongly assortative by age, sex, HIV status, or birthplace.

5. Significance and Implications

• Control Strategy Shift: The findings confirm that RR-TB in HCMC is predominantly driven by ongoing transmission. Therefore, control efforts must prioritize interrupting transmission through active case finding, rapid diagnosis, and immediate treatment initiation, rather than focusing solely on preventing acquired resistance in treated patients.
• Persistent Vulnerability: The 13–28% rate of acquired resistance indicates that current treatment pathways are failing a significant minority of patients. Potential drivers include:
  • Undiagnosed isoniazid mono-resistance leading to functional rifampicin monotherapy.
  • Host factors like diabetes and fast-acetylator genotypes.
• Policy Recommendations:
  • Broad Interventions: Community-wide screening is justified given the diffuse, city-wide nature of transmission.
  • Targeted Investigation: Further research is needed to identify why acquired resistance persists (e.g., drug susceptibility testing for isoniazid, management of comorbidities like diabetes).
• Generalizability: The analytic framework (integrating WGS, clinical history, and SIMEX correction) provides a robust model for estimating resistance origins in other high-burden settings where sampling is incomplete.