Wagner Sarah, Kurz Mario, Klimkait Thomas
Molecular Virology, Department of Biomedicine - Petersplatz, University of Basel, Basel, Switzerland.
Antivir Ther. 2015;20(6):661-5. doi: 10.3851/IMP2947. Epub 2015 Feb 24.
Different genotypic HIV resistance algorithms are based on different rules. They may therefore result in different drug-resistance interpretations for the same patient sample. In particular, for early periods of new retroviral inhibitors or classes, sequence interpretation is expected to vary. One would, however, assume that those differences between systems wane with growing experience and that different algorithms yield similar results for well-established drugs.
To assess the concordance of the Agence Nationale de Recherche sur le SIDA (ANRS), Rega and Stanford-HIVdb algorithms and their evolution over time, we analysed 284 routine samples with the current versions of each algorithm in 2004 and 2013. For 446 recent clinical sequences the differences for actual drugs were analysed. Scoring as 'susceptible' by one algorithm and 'resistant' by a second one defined a discordance.
The longitudinal analysis showed similar overall discordances for both time points as well as an evolution over time. The actual analysis demonstrated a higher overall discordance rate, mainly for certain drugs. Most deviations reflected differences between the ANRS and the other two algorithms.
This study demonstrates discordances between three most commonly used interpretation tools even for long-available drugs. It thereby reveals a need for further adjustment and improvement of current interpretation tools and may point at a possibly crucial role of subtype-specific information.
不同的基因分型HIV耐药性算法基于不同的规则。因此，对于同一份患者样本，它们可能会得出不同的耐药性解读结果。特别是在新型逆转录病毒抑制剂或类别出现的早期阶段，序列解读结果预计会有所不同。然而，人们会认为随着经验的积累，不同系统之间的差异会逐渐减小，并且对于已广泛使用的药物，不同算法会得出相似的结果。
为了评估法国国家艾滋病研究机构（ANRS）、雷加（Rega）和斯坦福-HIV数据库（Stanford-HIVdb）算法的一致性及其随时间的演变，我们在2004年和2013年使用每种算法的当前版本分析了284份常规样本。对于446条近期临床序列，分析了实际药物的差异。一种算法判定为“敏感”而另一种算法判定为“耐药”即定义为不一致。
纵向分析显示两个时间点的总体不一致情况相似，且随时间有所演变。实际分析表明总体不一致率更高，主要针对某些药物。大多数偏差反映了ANRS与其他两种算法之间的差异。
本研究表明，即使对于已广泛使用的药物，三种最常用的解读工具之间也存在不一致情况。因此揭示了进一步调整和改进当前解读工具的必要性，并且可能表明亚型特异性信息可能具有关键作用。
