Multivariate Analysis of Prognostic Factors in Gastric Cancer Patients Using Additive Hazards Regression Models


Background and
Objectives: Gastric cancer is the second leading cause of cancer death worldwide and isthe most common type of cancer in Iran. The objective of this research was to apply additive hazards models tothe study of survival of patients with gastric cancer and to compare with results obtained using the Cox model.
Methods: We retrospectively studied 213 patients with gastric cancer who were registered in one referral cancerregistry center in Tehran, Iran. Age at diagnosis, sex, presence of metastasis, tumor size, histology type, lymphnode metastasis, and pathologic stages were entered into analysis using the Cox model and additive hazardmodels. To visualize a covariate effect over time, the estimated cumulative regression function by the Aalen’smodel was examined.
Results: The five-year survival rate and the median life expectancy in the studied patientswere 14.6% and 29.6 months, respectively. Multivariate Cox and Additive hazards models analysis identifiedage at diagnosis, tumor size and pathologic stage as independent prognostic factors for the survival of patientswith gastric cancer. Moreover, pathologic stage had a late or delayed effect according to the Aalen’s plot. Otherclinicopathological characteristics were not statistically significant.
Conclusion: Since Cox and Aalen models givedifferent aspects of the association between risk factors and the study outcome, it seems desirable to use thentogether to give a more comprehensive understanding of data. Our results also suggest that early detection ofpatients at younger age and in primary stages is important to increase survival of patients with gastric cancer.