Estimation of malaria haplotype and genotype frequencies: a statistical approach to overcome the challenge associated with multiclonal infections

Aimee R. Taylor, Jennifer A. Flegg, Samuel L. Nsobya, Adoke Yeka, Moses R. Kamya, Philip J. Rosenthal, Grant Dorsey, Carol Hopkins Sibley, Philippe J. Guérin, Chris HolmesMalaria 2014 · DOI

Abstract

The model allows fast, accurate, reliable estimation of the frequency of genetic motifs associated with resistance to anti-malarials using prevalence data collected from malaria patients. The model does not require per-patient MOI measurements and can easily analyse data from five markers. The model will be a valuable tool for monitoring markers of anti-malarial drug resistance, including markers of resistance to artemisinin derivatives and partner drugs.

Introduction: The Challenge of Multiclonal Infections

Monitoring the evolution of drug resistance in malaria parasites is critical for public health. However, in high-transmission regions, a single patient often carries multiple distinct parasite clones simultaneously—a phenomenon known as a multiclonal infection or high Multiplicity of Infection (MOI).

Standard surveillance typically measures prevalence: the proportion of patients carrying a specific mutation. Unfortunately, prevalence is a misleading proxy for the actual frequency of a resistant genotype in the parasite population. Because a single patient acts as a "larger net" in high-transmission areas, prevalence consistently overestimates frequency. This discrepancy makes it difficult to compare resistance levels across regions with different transmission intensities.

A Bayesian Solution

Taylor et al. developed a Bayesian statistical framework to bridge the gap between observed prevalence and unobserved frequency. The model treats the number of clones per patient and their specific genotypes as "hidden" variables, using a probability distribution (such as Poisson or Geometric) to represent the population-level MOI.

Key features of this approach include:

  • MCMC Sampling: Utilizing a Metropolis-Hastings Markov Chain Monte Carlo algorithm to estimate posterior distributions of genotype frequencies.
  • Data Imputation: The model can "fill in the blanks" for partially genotyped samples by using population-level data to predict missing markers.
  • Efficiency: Unlike methods that discard complex multiclonal samples, this model incorporates all available data, turning previously discarded "noise" into valuable information.

Validation and Field Application in Uganda

The researchers validated the model using both simulated datasets and field data from 1,738 patients across five sites in Uganda: Kanungu, Mubende, Jinja, Tororo, and Apac.

Map of the five study sites in Uganda used for field validation.

The field study focused on markers in the Pfdhfr and Pfdhps genes, which confer resistance to Sulfadoxine-Pyrimethamine (SP). The model successfully identified four biologically relevant genotypes, including the "quintuple mutant"—the most resistant form—which was found at high frequencies ranging from 46% in Mubende to 70% in Tororo.

Handling Data Complexity

One of the primary strengths of the Bayesian model is its ability to handle the "messy" reality of field data, where samples often have missing markers or show evidence of both sensitive and resistant strains within a single patient.

Color-coded representation of patient sample data from five Ugandan sites, illustrating the mix of sensitive, resistant, and missing markers.

As shown in the visualization of the Ugandan dataset, the model must navigate significant amounts of missing data and mixed infections. By leveraging the entire population's genetic profile, the model provides more accurate and precise estimates than traditional "direct counting" methods, which are prone to bias when MOI is high.

Conclusion and Future Directions

The Bayesian framework presented by Taylor et al. offers a robust, computationally efficient tool for malaria surveillance. It demonstrates that knowing the exact MOI for every patient is unnecessary; a population-level average is sufficient for accurate frequency estimation. This model is poised to become a vital asset in monitoring emerging resistance to modern treatments, such as artemisinin derivatives, ensuring that public health responses are based on the most accurate data possible.

Summary updated

Abstract

The model allows fast, accurate, reliable estimation of the frequency of genetic motifs associated with resistance to anti-malarials using prevalence data collected from malaria patients. The model does not require per-patient MOI measurements and can easily analyse data from five markers. The model will be a valuable tool for monitoring markers of anti-malarial drug resistance, including markers of resistance to artemisinin derivatives and partner drugs.
BibTeX
@inproceedings{taylor2014estimation,
  author = {Aimee R. Taylor and Jennifer A. Flegg and Samuel L. Nsobya and Adoke Yeka and Moses R. Kamya and Philip J. Rosenthal and Grant Dorsey and Carol Hopkins Sibley and Philippe J. Guérin and Chris Holmes},
  title = {{Estimation of malaria haplotype and genotype frequencies: a statistical approach to overcome the challenge associated with multiclonal infections}},
  booktitle = {Malaria},
  year = {2014},
  doi = {10.1186/1475-2875-13-102},
}
Figure