Pharmacology Research Laboratories, University of Liverpool, Liverpool, UK.
HIV Med. 2009 Nov;10(10):620-6. doi: 10.1111/j.1468-1293.2009.00734.x. Epub 2009 Jul 6.
Genotypic algorithms used to predict the clinical efficacy of lopinavir/ritonavir (LPV/r) have included a range of mutation lists and efficacy endpoints. Normally, HIV clinical trials are powered to detect a difference between treatment arms of 10-12% for the endpoint of viral load suppression <50 HIV-1 RNA copies/mL. The TITAN trial evaluated LPV/r vs. darunavir/ritonavir (DRV/r) in treatment-experienced patients with viral load >1000 copies/mL. This analysis aimed to re-evaluate resistance algorithms for LPV/r in the TITAN trial.
Baseline genotype data were classified using seven genotypic resistance algorithms: International AIDS Society USA (IAS-USA) LPV mutations (current cut-off=6), Abbott 2007 mutation list (cut-off=3), ANRS mutations (cut-off=4), FDA mutations (cut-off=3), Stanford, REGA and IAS-USA major protease inhibitor (PI) mutations. Efficacy in the TITAN trial (HIV-1 RNA <50 copies/mL at week 48) was correlated with the number of mutations from each list, to show the 'efficacy advantage cut-off level': the number of mutations from each list associated with a difference in efficacy between treatment arms of at least 12%.
Multivariate logistic regression analysis identified lower genotypic cut-off levels than previously reported where there was at least 12% lower efficacy for LPV/r vs. DRV/r. These efficacy advantage cut-off levels were: IAS-USA LPV mutations, cut-off=3; Abbott 2007, cut-off=2; ANRS LPV, cut-off=3; FDA LPV mutations, cut-off=2; major IAS-USA PI mutations, cut-off=1; Stanford algorithm, cut-off=low-level LPV resistance; REGA algorithm, cut-off=intermediate-level LPV resistance. There were linear falls in HIV-1 RNA suppression rates with rising mutation counts in the TITAN, French LPV ATU, BMS-045 and RESIST trials.
The analysis identified more sensitive cut-off levels for LPV genotypic algorithms, below those currently used.
用于预测洛匹那韦/利托那韦（LPV/r）临床疗效的基因型算法包括一系列突变列表和疗效终点。通常，HIV 临床试验的目的是检测治疗组之间病毒载量抑制<50 HIV-1 RNA 拷贝/ml 的终点有 10-12%的差异。TITAN 试验评估了 LPV/r 与达芦那韦/利托那韦（DRV/r）在病毒载量>1000 拷贝/ml 的治疗经验丰富的患者中的疗效。本分析旨在重新评估 TITAN 试验中 LPV/r 的耐药性算法。
使用七种基因型耐药算法对基线基因型数据进行分类：美国国际艾滋病协会（IAS-USA）LPV 突变（现行截止值=6）、雅培 2007 年突变列表（截止值=3）、ANRS 突变（截止值=4）、FDA 突变（截止值=3）、斯坦福、REGA 和 IAS-USA 主要蛋白酶抑制剂（PI）突变。将 TITAN 试验的疗效（第 48 周 HIV-1 RNA<50 拷贝/ml）与每个列表的突变数量相关联，以显示“疗效优势截止水平”：每个列表的突变数量与治疗组之间至少 12%的疗效差异相关。
多变量逻辑回归分析确定了比以前报告的更低的基因型截止值，在 LPV/r 与 DRV/r 相比，至少有 12%的疗效降低。这些疗效优势截止值为：IAS-USA LPV 突变，截止值=3；雅培 2007 年，截止值=2；ANRS LPV，截止值=3；FDA LPV 突变，截止值=2；主要 IAS-USA PI 突变，截止值=1；斯坦福算法，截止值=低水平 LPV 耐药；REGA 算法，截止值=中水平 LPV 耐药。随着 TITAN、法国 LPV ATU、BMS-045 和 RESIST 试验中突变数量的增加，HIV-1 RNA 抑制率呈线性下降。
分析确定了更敏感的 LPV 基因型算法截止值，低于目前使用的截止值。
