The risk to develop a recurrence of a gastric cancer—is it independent of time?

Abstract

Recurrence is the main reason for early death of cancer patients. Therefore, survival curves are regarded as crucial tools for clinicians to evaluate therapies and to estimate the patients' prognosis. Current models for the development of recurrence are based on the assumption of a residual cancer cell burden after therapy. Accordingly, in assumption of an exponential cell growth of the cancer cells an S-shaped decline of the survival curve is expected with earlier onset in case of advanced cancer and a later manifestation in case of little residual tumour. However, many survival curves do not reflect any S-shaped configuration and thus may question the current concept. To test, whether the incidence for developing a recurrence may be considered as remaining constant over time, we analysed survival data of 446 patients with gastric cancer, operated from 1975-2001 in the Surgical Department of the RWTH Aachen. All survival curves, even after sub-grouping according to UICC stage, show a monotonous decline without any apparent S-shape. The impact of TNM and UICC stage to predict the survival in patients is estimated by Cox regression, for estimation the risk for death a logistic regression is performed. Whereas the presence of metastasis lowers the prognosis significantly with a hazard ratio of 1.57 and an odds ratio of 7.56, respectively, a significant relevance for the UICC stage, the tumour size or the lymph node status cannot be proven. Furthermore, the two assumptions: (1) that 20% of patients who are still alive after 5 years have been cured, and (2) that the remainder develop a recurrence in constantly 7.3% per month, are able to configure the survival curve almost precisely (correlation coefficient between calculated and observed survival rate r > 0.99). The absence of any S-shaped survival curve configuration is not in accordance with the focus on residual tumour clones with its exponential growth as the decisive process for recurrence development. In contrast, the monotonous decline of surviving patients is best reflected by a constant incidence of recurrence. The (time) constancy of the recurrence incidence encourages the view of recurrent cancer as a chronic problem of a carcinogenic environment. Furthermore, it supports new anticancer therapies which rather targets cell regulation and immunology instead of acting cytotoxic.