Background. Several mathematical models have been developed for predicting individual breast cancer risk. Such models can help clinicians to choose appropriate preventive and therapeutic interventions for each patient. Unfortunately, the validity of these models has not been tested outside the USA.
Methods. The authors describe a case–control study in the Czech Republic with a similar design to that of the US Breast Cancer Detection and Demonstration Project (BCDDP). The main objective of the study was to evaluate the validity of the Gail model in the Czech female population, and to develop a local model using the same statistical approach as the Gail model. Between November 2000 and May 2004, 14,566 questionnaires containing case history data from both healthy women (control group) and women with breast cancer were collected. Case–control age-matched pairs (n = 4598) have subsequently been matched and analyzed.
Results. Our results show that the original Gail model was not able to properly distinguish between controls and breast cancer cases in the Czech female population. Based on paired data, the mean 5-year and life-time breast cancer risk was 1.379 ± 0.668 and 7.990 ± 3.184 in the control group and 1.375 ± 0.692 and 8.028 ± 3.506 in the patients with breast cancer group. The original Gail model was also not able to properly describe age-specific baseline risk of breast cancer development in the Czech population. In response the authors developed two variants of modified/locally adjusted models.
Conclusion. The original Gail model is not an accurate breast cancer risk assessment tool for the Czech female population.