Development and validation of a prognostic index for survival in non-small cell lung cancer: Results from a Tunisian cohort study

• Published in: Cancer epidemiology 2018-04, Vol.53, p.111-118
• Main Authors: Soussi, Ghassen, Ben Alaya, Nissaf, Chaouch, Nawel, Racil, Hajer
• Format: Article
• Language: English
• Subjects: Adenocarcinoma - epidemiology; Adenocarcinoma - mortality; Adenocarcinoma - pathology; Aged; Cancer; Carcinoma, Non-Small-Cell Lung - epidemiology; Carcinoma, Non-Small-Cell Lung - mortality; Carcinoma, Non-Small-Cell Lung - pathology; Carcinoma, Squamous Cell - epidemiology; Carcinoma, Squamous Cell - mortality; Carcinoma, Squamous Cell - pathology; Cell survival; Classification; Cohort analysis; Cox proportional hazards; Epidemiology; Female; Humans; Hypothesis testing; Longitudinal Studies; Lung cancer; Lung Neoplasms - epidemiology; Lung Neoplasms - mortality; Lung Neoplasms - pathology; Male; Medical prognosis; Middle Aged; Modelling; Models, Statistical; Non-small cell lung cancer; Non-small cell lung carcinoma; Patients; Prognosis; Risk Factors; ROC Curve; Smoking; Software; Statistical analysis; Survival; Survival Rate; Tumors; Tunisia - epidemiology; Variables
• ISSN: 1877-7821; 1877-783X
• DOI: 10.1016/j.canep.2018.01.018

•Multiple parameters should be included in the prognostication process in NSCLC.•Additive models seem more relevant than multiplicative ones for index derivation.•A simple prognostic index could predict survival more accurately in NSCLC patients. Despite the continuous efforts made with the TNM system, the issue of heterogeneity of prognosis within the stages of non-small cell lung cancer (NSCLC) could not be resolved. Our aim was to identify prognostic factors and develop an index to predict NSCLC survival with greater accuracy. We conducted a survival study over 5 years on patients with NSCLC. Kaplan–Meier analysis followed by Cox regression modelling were used. Prognostic indices were derived, using either an additive or a multiplicative pattern, and were compared by their receiver operating characteristics (ROC) curves. We then proceeded to a risk stratification and validation of the index on the derivation cohort. Two hundred and sixty-two NSCLC patients were included. Two models were constructed, using the following nine variables as prognostic factors: age, performance status, haemoglobin level, leucocyte count, calcium, lactate dehydrogenase, alkaline phosphatase levels, histological type and TNM stage. Four prognostic indices were derived, and the best one was picked and validated on a population of five risk groups. The higher the risk group, the shorter the survival. This novel and simple prognostic tool could predict survival more accurately in patients with NSCLC.