Bioinformatics and Biostatistics Hub, Institut Pasteur, Université Paris Cité, 75015 Paris, France.
UK MRC Clinical Trials Unit, University College London, London WC1V 6LJ, UK.
Viruses. 2023 May 25;15(6):1244. doi: 10.3390/v15061244.
A deeper understanding of HIV-1 transmission and drug resistance mechanisms can lead to improvements in current treatment policies. However, the rates at which HIV-1 drug resistance mutations (DRMs) are acquired and which transmitted DRMs persist are multi-factorial and vary considerably between different mutations. We develop a method for the estimation of drug resistance acquisition and transmission patterns. The method uses maximum likelihood ancestral character reconstruction informed by treatment roll-out dates and allows for the analysis of very large datasets. We apply our method to transmission trees reconstructed on the data obtained from the UK HIV Drug Resistance Database to make predictions for known DRMs. Our results show important differences between DRMs, in particular between polymorphic and non-polymorphic DRMs and between the B and C subtypes. Our estimates of reversion times, based on a very large number of sequences, are compatible but more accurate than those already available in the literature, with narrower confidence intervals. We consistently find that large resistance clusters are associated with polymorphic DRMs and DRMs with long loss times, which require special surveillance. As in other high-income countries (e.g., Switzerland), the prevalence of sequences with DRMs is decreasing, but among these, the fraction of transmitted resistance is clearly increasing compared to the fraction of acquired resistance mutations. All this indicates that efforts to monitor these mutations and the emergence of resistance clusters in the population must be maintained in the long term.
深入了解 HIV-1 传播和耐药机制可以改进当前的治疗策略。然而，HIV-1 耐药突变（DRMs）的获得率以及哪些传播性耐药突变持续存在是多因素的，在不同的突变之间差异很大。我们开发了一种用于估计耐药获得和传播模式的方法。该方法使用治疗推出日期通知的最大似然祖先特征重建，并允许对非常大的数据集进行分析。我们将我们的方法应用于从英国 HIV 耐药性数据库获得的数据重建的传播树，以对已知的耐药突变进行预测。我们的结果表明，耐药突变之间存在重要差异，特别是多态性和非多态性耐药突变之间，以及 B 和 C 亚型之间。我们基于大量序列的回复时间估计值与文献中已经存在的估计值一致，但更准确，置信区间更窄。我们一致发现，大的耐药性簇与多态性耐药突变和回复时间长的耐药突变有关，需要特别监测。与其他高收入国家（例如瑞士）一样，携带耐药突变的序列的流行率正在下降，但在这些序列中，与获得性耐药突变相比，传播性耐药的比例明显增加。所有这些都表明，必须长期努力监测这些突变以及人群中耐药性簇的出现。
