Huang Christine, Zheng Ming, Yang Zheng, Rodrigues A David, Marathe Punit
Metabolism and Pharmacokinetics, Pharmaceutical Candidate Optimization, Bristol-Myers Squibb Co., PO Box 5400, Princeton, New Jersey 08543-5400, USA.
Pharm Res. 2008 Apr;25(4):713-26. doi: 10.1007/s11095-007-9411-4. Epub 2007 Sep 25.
Preclinical and clinical data for 35 proprietary Bristol-Myers Squibb discovery compounds (years 1997 to 2005) were collected and analyzed. In each case, exposure and efficacy in human subjects were projected at the time of nomination (for development) prior to first-in-human dosing.
Projections of area under the plasma concentration-time curve (AUC) in humans involved the use of one or more methods: (1) allometric scaling of animal pharmacokinetic data; (2) clearance projection employing in vitro data (liver microsomes and hepatocytes); (3) chimpanzee as an animal model; (4) the species-invariant time method; and (5) the Css-mean residence time or "Css-MRT" method. Whenever possible, prior clinical experience with lead compounds enabled the selection of the most appropriate method(s). Multiple approaches were also available at the time of the human efficacious dose projections: (1) efficacious exposure from animal efficacy models; (2) in vitro potency; and (3) prior experience with clinical leads.
Over the 8 year period described, AUC in humans was projected within 2-fold (20 out of 35 compounds; 57%), greater than 2-fold to 4-fold (11 out of 35 compounds; 32%), and greater than 4-fold (4 out of 35 compounds; 11%) of the observed value. At the time of writing, clinical efficacy data were available for 10 compounds only. In this instance, the efficacious doses were also projected within 2-fold (7 out of 10 compounds; 70%), greater than 2-fold to 4-fold (2 out of 10 compounds; 20%), and greater than 4-fold (1 out of 10 compounds; 10%) of the actual clinical dose.
Overall, it was possible to project human exposure and efficacious dose within 4-fold of observed clinical values for about 90% of the compounds.
收集并分析了百时美施贵宝公司35种自主研发化合物（1997年至2005年）的临床前和临床数据。在每种情况下，在首次人体给药前提名（用于开发）时预测人体受试者的暴露量和疗效。
预测人体血浆浓度-时间曲线下面积（AUC）涉及使用一种或多种方法：（1）动物药代动力学数据的异速缩放；（2）采用体外数据（肝微粒体和肝细胞）进行清除率预测；（3）以黑猩猩作为动物模型；（4）物种不变时间法；以及（5）Css-平均驻留时间或“Css-MRT”法。只要有可能，先导化合物的既往临床经验有助于选择最合适的方法。在预测人体有效剂量时也有多种方法可用：（1）来自动物疗效模型的有效暴露量；（2）体外效力；以及（3）临床先导化合物的既往经验。
在所描述的8年期间，人体AUC预测值在观察值的2倍以内（35种化合物中的20种；57%）、大于2倍至4倍（35种化合物中的11种；32%）以及大于4倍（35种化合物中的4种；11%）。在撰写本文时，仅10种化合物有临床疗效数据。在这种情况下，有效剂量预测值也在实际临床剂量的2倍以内（10种化合物中的7种；70%）、大于2倍至4倍（10种化合物中的2种；20%）以及大于4倍（10种化合物中的1种；10%）。
总体而言，约90%的化合物能够将人体暴露量和有效剂量预测在观察到的临床值的4倍以内。
