啮齿类动物致癌性试验是新药研发中非临床安全性评价的重要组成部分。致癌性试验周期较长、数据量大,其统计学方法选用的合理性、科学性将可能影响试验结果的结论。首先介绍了目前致癌性试验常用的生存分析和肿瘤发生率统计方法;其次,汇总了生存分析和肿瘤发生率统计方法的选用情况;最后,通过比较申报方、美国食品药品监督管理局审评方和美国国家毒理学项目中心在统计学方法选用的差异,分析和讨论了这些差异的来源和影响,并总结了致癌性试验统计分析设计思路。希望通过对以上内容的汇总和分析,能够为新药非临床致癌性试验提供参考。

Carcinogenicity studies in rodents are important in nonclinical safety evaluation of new drugs. Carcinogenicity studies have long durations and large amounts of data, the appropriateness and selection scientificity of statistical methods are also likely to affect their study results and conclusions. This review firstly introduces statistical methods of survival analysis and tumor incidence currently used in carcinogenicity studies. Secondly, the selection of statistical methods for survival analysis and tumor incidence is summarized. Finally, by comparing the differences between statistical selecting methods chosen by sponsors, the reviewers of the U.S. Food and Drug Administration(FDA) and the National Toxicology Program(NTP), the sources and implication of these differences are analyzed and discussed. Furthermore, the review summarizes the ideas for designing statistical analysis in carcinogenicity studies. We hope that all the summary and analysis above can provide a reference for designing non-clinical carcinogenicity studies.