Gerlovina Inna, Berube Sophie, Briggs Jessica, Murie Kathryn, Murphy Maxwell, Wesolowski Amy, Greenhouse Bryan
EPPIcenter research program, Division of HIV, ID and Global Medicine, Department of Medicine, University of California, San Francisco, CA, USA.
Dept of Biostatistics, University of Florida, Gainesville, FL, USA.
bioRxiv. 2025 May 1:2025.04.28.651146. doi: 10.1101/2025.04.28.651146.
Reliable assessment of antimalarial drug efficacy is crucial for effective response to emerging drug resistance, and therapeutic efficacy studies (TES) are the primary means of estimating efficacy. Accuracy of such estimates rests on correctly classifying recurrent infections developed during follow-up as recrudescences or new infections. Genotyping is used to guide classification, but polyclonal infections and allele chance matching still make classification challenging, especially in high transmission settings. Established methods for analyzing genotyping data are biased or difficult to use; few take full advantage of data from modern genotyping methods such as multiplexed amplicon sequencing. We propose an Adaptable Statistical framework for Therapeutic Efficacy and Recrudescence (Aster) that delivers accurate and consistent results by explicitly incorporating complexity of infection (COI), population allele frequencies, and imperfect detection of alleles in minority strains. Using an identity by descent approach, Aster accounts for alleles matching by chance and background relatedness that can otherwise lead to misclassification. The extensible framework can also utilize external information, such as parasite density and performance characteristics of a genotyping panel. Aster employs efficient combinatorial algorithms to process unphased polyclonal data, making it fast and fully scalable. Using simulations, we show that Aster dramatically outperforms match-counting algorithms currently recommended by WHO in a wide variety of settings and demonstrates consistently balanced performance measures that improve with more informative genotyping panels. Aster provides accurate study-level estimates of treatment failure for TES with any type of genotyping data, facilitating reliable evaluation of drug efficacy and effective management of malaria.
对抗疟药物疗效进行可靠评估对于有效应对新出现的耐药性至关重要，而治疗效果研究（TES）是评估疗效的主要手段。此类评估的准确性取决于将随访期间出现的复发性感染正确分类为复发或新感染。基因分型用于指导分类，但多克隆感染和等位基因偶然匹配仍使分类具有挑战性，尤其是在高传播环境中。现有的分析基因分型数据的方法存在偏差或难以使用；很少有方法能充分利用来自现代基因分型方法（如多重扩增子测序）的数据。我们提出了一种用于治疗效果和复发的适应性统计框架（Aster），该框架通过明确纳入感染复杂性（COI）、群体等位基因频率以及少数菌株中等位基因的不完全检测，提供准确且一致的结果。通过采用同源身份方法，Aster考虑了偶然匹配的等位基因和背景相关性，否则这些因素可能导致错误分类。这个可扩展框架还可以利用外部信息，如寄生虫密度和基因分型面板的性能特征。Aster采用高效的组合算法来处理未分型的多克隆数据，使其快速且完全可扩展。通过模拟，我们表明Aster在各种环境中均显著优于世界卫生组织目前推荐的匹配计数算法，并展示出始终平衡的性能指标，随着基因分型面板信息量的增加而有所改善。Aster可为任何类型基因分型数据的TES提供准确的研究水平治疗失败估计，有助于可靠评估药物疗效和有效管理疟疾。
