Original Article
Selection for adjuvant chemotherapy in completely resected stage I non-small cell lung cancer: external validation of a Chinese prognostic risk model
Abstract
Background: The ability to sub-stratify survival within stage I is an important consideration as it is assumed that survival is heterogeneous within this sub-group. Liang et al. recently published a nomogram to predict post-operative survival in patients undergoing lung cancer surgery. The aim of our study is external validation of their published nomogram in a British cohort focusing on stages IA and IB to determine applicability in selection of adjuvant chemotherapy within stage I.
Methods: Patient variables were extracted and the score individually calculated. Receiver operative characteristics curve (ROC) was calculated and compared with the original derivation cohort and the discriminatory ability was further quantified using survival plots by splitting our (external) validation cohort into three tertiles and Kaplan Meier plots were constructed and individual curves tested using Cox regression analysis on Stata 13 and R 3.1.2 respectively.
Results: A total of 1,238 patients were included for analysis. For all patients from stage IA to IIB the mean (SD) score was 9.95 (4.2). The ROC score comparing patients who died versus those that remained alive was 0.62 (95% CI: 0.58 to 0.67). When divided into prognostic score tertiles, survival discrimination remained evident for the entire cohort, as well as those for stage IA and IB alone. The P value comparing survival between the middle and highest score with baseline (low score) was P=0.031 and P=0.034 respectively.
Conclusions: Our results of external validation suggested lower survival discrimination than reported by the original group; however discrimination between survival remained evident for stage I.
Methods: Patient variables were extracted and the score individually calculated. Receiver operative characteristics curve (ROC) was calculated and compared with the original derivation cohort and the discriminatory ability was further quantified using survival plots by splitting our (external) validation cohort into three tertiles and Kaplan Meier plots were constructed and individual curves tested using Cox regression analysis on Stata 13 and R 3.1.2 respectively.
Results: A total of 1,238 patients were included for analysis. For all patients from stage IA to IIB the mean (SD) score was 9.95 (4.2). The ROC score comparing patients who died versus those that remained alive was 0.62 (95% CI: 0.58 to 0.67). When divided into prognostic score tertiles, survival discrimination remained evident for the entire cohort, as well as those for stage IA and IB alone. The P value comparing survival between the middle and highest score with baseline (low score) was P=0.031 and P=0.034 respectively.
Conclusions: Our results of external validation suggested lower survival discrimination than reported by the original group; however discrimination between survival remained evident for stage I.
