[A systematic review of international simulation models on the natural history of breast cancer: current understanding and challenges for Chinese-population-specific model development].

Objective: To systematically review the worldwide simulation model studies on the natural history of breast cancer and to summarize related parameters. Methods: A structured literature search was conducted in PubMed and the Cochrane Library to identify articles during 1980-2015. Articles were screened independently by two researchers. Health states in the natural history and relevant parameters were extracted. Results: A total of 36 studies were included for analysis, within the earliest one was published in 1990. Most studies were from Europe and America countries, and 2 studies from China. Markov model was mostly applied to evaluating breast cancer screening programs (n=32). Reported health status included "healthy" (n=36), ductal carcinoma in situ (DCIS, n=17), invasive breast cancer (IBC, n=36), and death (n=27). There were two definite classifications for IBC, tumor size (n=9) and TNM staging (n=9, 3 studies reported transition rates). The median (range) of annual transition rates from DCIS to stage-Ⅰ IBC, Ⅰ to Ⅱ, Ⅱ to Ⅲ, Ⅲ to Ⅳ were 0.279 (0.259-0.299), 0.150 (0.069-0.430), 0.100 (0.060-0.128) and 0.210 (0.010-0.625), respectively. A total of 15 studies reported the mean duration from predinical to clinical stage for IBC was 1.95-4.70 years, which gradually increased with age, and 7 studies reported that for DCIS. Conclusions: Despite closer attention was paid to breast cancer natural history models, in recent years atypical hyperplasia has been neglected. Data on the mean duration of DCIS requires reasonable conversion. Various classifications for IBC exist whereas transition rates are limited. Current findings would be valuable references but challenging for the Chinese-population specific natural history model, development.