4-year mortality in patients with non-small-cell lung cancer: development and validation of a prognostic index

Abstract

Background: Lung cancer is the commonest cause of death due to cancer in the world. Non-small-cell lung carcinoma (NSCLC) represents about 80% of overall lung cancer cases worldwide. An accurate predictive model of mortality in patients with NSCLC could be useful to clinicians, policy makers, and researchers involved in risk stratification. The objective of this study was to develop and validate a simple prognostic index for 4-year mortality in patients with NSCLC by use of information obtained at the time of lung cancer diagnosis.

Methods: In 2000, 4669 patients with histologically or cytologically proven NSCLC were enrolled prospectively from 137 pneumology departments in French general hospitals. Patients not lost to follow-up (n=4479) were randomly assigned to the development cohort (n=2979) or the validation cohort (n=1500). Every patient's physician completed a standard and anonymous questionnaire. We used a Cox model to identify variables independently associated with mortality and weighted the variables to create a prognostic index.

Findings: Median follow-up for survivors was 49 months (IQR 46-51). There were 2585 deaths (87%) in the development cohort and 1310 deaths (87%) in the validation cohort. Five independent predictors of mortality were identified: age (>70 years, 1 point); sex (male, 1 point); performance status at diagnosis (reduced activity, 3 points; active >50%, 5 points; inactive >50%, 8 points; and total incapacity, 10 points); histological type (large-cell carcinoma, 2 points); and tumour-node-metastasis (TNM) staging system (IIA or IIB, 3 points; IIIA or IIIB, 6 points; and IV, 8 points). The minimum and maximum possible point scores were 0 and 22, respectively. Scores of the prognostic index were strongly associated with 4-year mortality in the development cohort: 0-1 points predicted a 35% (95% CI 28-43) risk, 2-4 points a 59% (52-66) risk, 5-7 points a 77% (72-81) risk, 8-10 points an 88% (85-90) risk, 11-14 points a 97% (96-98) risk, and 15-22 points a 99% (97-100) risk. The corresponding percentages in the validation cohort were 36% (24-47), 60% (50-70), 77% (71-83), 89% (86-93), 96% (95-98), and 99% (98-100), respectively. The prognostic index showed good discrimination, with mean bootstrap c statistics of 0.85 (95% CI 0.84-0.86) in the development cohort and 0.86 (95% CI 0.85-0.87) in the validation cohort.

Interpretation: This prognostic index, incorporating personal, tumour, and functional information would be helpful in guiding patient management, resource use, and the design of clinical trials.