《学术论文集锦[开错]》紫微客

2、002

　　2.Literature Review

　　Cancer is one of the leading causes of death in humans, and chemotherapy is considered to be an important pillar of defeating cancer[15]. As we know, tumors are heterogeneous, so anticancer therapy should focus on individuation, and currently the most commonly used clinical method to guide individualized drug use is to conduct tumor chemosensitive assay in vitro before clinical practice. In order to understand this field more deeply, it is necessary to review some of the mainstream statistical studies related to tumor chemosensitive assay in vitro.

　　First of all, Frank Bretz et al. pointed out that dose finding studies have guiding significance for the use of chemotherapy drugs, because too low dose will lead to ineffectiveness, while too high dose will lead to safety problems[5]. As an important statistical method in Phase II and III clinical trials, dose finding analysis covers multiple research directions. For example, since nonlinear dose-response model is the main tool for evaluating drug efficacy[16], scholars fitted data into sigmoidal curve (such as Four-parameter Logistic model, Sigmoid Emax model, etc.) to calculate Emax, EC50, IC50 and other sensitivity indexes. According to these indicators, the sensitivity and tolerance of cells to drugs can be measured[2]. However, in 2016, Hafner et al. pointed out that the above indexes are highly sensitive to cell division and thus affect the accuracy of judgment, so they proposed to replace the above indexes with GR values, such as GR50 and the maximal measured effect of the drug (GRmax)[4], because these indexes are immune to confounding factors. That is, it is not affected by natural biological variation and differences in culture conditions. Similarly, Nicholas et al. showed in 2017 that GR method can improve the reliability and repeatability of dose-response analysis[3].