Characterisation of long-term cancer survivors and application of statistical cure models: a protocol for an observational follow-up study in patients with colorectal cancer

Abstract

Background

Improved colorectal cancer (CRC) survival rates have been reported over the last years, with more than half of these patients surviving more than 5 years after the initial diagnosis. Better understanding these so-called long-term survivors could be very useful to further improve their prognosis as well as to detect other problems that may cause a significant deterioration in their health-related quality of life (HRQoL). Cure models provide novel statistical tools to better estimate the long-term survival rate for cancer and to identify characteristics that are differentially associated with a short or long-term prognosis. The aim of this study will be to investigate the long-term prognosis of CRC patients, characterise long-term CRC survivors and their HRQoL, and demonstrate the utility of statistical cure models to analyse survival and other associated factors in these patients.

Methods

This is a single-centre, ambispective, observational follow-up study in a cohort of n = 1945 patients with CRC diagnosed between 2006 and 2013. A HRQoL sub-study will be performed in the survivors of a subset of n = 485 CRC patients for which baseline HRQoL data from the time of their diagnosis is already available. Information obtained from interviews and the clinical records for each patient in the cohort is already available in a computerised database from previous studies. This data includes sociodemographic characteristics, family history of cancer, comorbidities, perceived symptoms, tumour characteristics at diagnosis, type of treatment, and diagnosis and treatment delay intervals. For the follow-up, information regarding local recurrences, development of metastases, new tumours, and mortality will be updated using hospital records. The HRQoL for long-term survivors will be assessed with the EORTC QLQ-C30 and QLQ-CR29 questionnaires.

An analysis of global and specific survival (competitive risk models) will be performed. Relative survival will be estimated and mixture cure models will be applied. Finally, HRQoL will be analysed through multivariate regression models.

Discussion

We expect the results from this study to help us to more accurately determine the long-term survival of CRC, identify the needs and clinical situation of long-term CRC survivors, and could be used to propose new models of care for the follow-up of CRC patients.