Joint modelling of longitudinal CEA tumour marker progression and survival data on breast cancer.

This work proposes the use of Biostatistics methods to study breast cancer in patients of Braga's Hospital Senology Unit, located in Portugal. The primary motivation is to contribute to the understanding of the progression of breast cancer, within the Portuguese population, using a more complex statistical model assumptions than the traditional analysis that take into account a possible existence of a serial correlation structure within a same subject observations. We aim to infer which risk factors aect the survival of Braga's Hospital patients, diagnosed with breast tumour. Whilst analysing risk factors that aect a tumour markers used on the surveillance of disease progression the Carcinoembryonic antigen (CEA). As survival and longitudinal processes may be associated, it is important to model these two processes together. Hence, a joint modelling of these two processes to infer on the association of these was conducted. A data set of 540 patients, along with 50 variables, was collected from medical records of the Hospital. A joint model approach was used to analyse these data. Two dierent joint models were applied to the same data set, with dierent parameterizations which give dierent interpretations to model parameters. These were used by convenience as the ones implemented in R software. Results from the two models were compared. Results from joint models, showed that the longitudinal CEA values were signicantly associated with the survival probability of these patients. A comparison between parameter estimates obtained in this analysis and previous independent survival[4] and longitudinal analysis[5][6], lead us to conclude that independent analysis brings up bias parameter estimates. Hence, an assumption of association between the two processes in a joint model of breast cancer data is necessary. Results indicate that the longitudinal progression of CEA is signicantly associated with the probability of survival of these patients. Hence, an assumption of association between the two processes in a joint model of breast cancer data is necessary. [ABSTRACT FROM AUTHOR]