Prediction and prognostic factors of post-recurrence survival in recurred patients with early-stage NSCLC who underwent complete resection.

Background: This study aimed to evaluate prognostic factors of post-recurrence survival (PRS) and to improve survival in recurred patients with early-stage non-small cell lung cancer (NSCLC).

Methods: The 141 patients with recurrence after complete resection of stage I and II NSCLC between 1995 and 2012 was retrospectively reviewed. Overall PRS and PRS of the patient groups stratified according to the sum of their own risk scores were analyzed.

Results: The patterns of recurrence of 141 patients included local only in 40(28.4%), distant only in 86 (61%) and both in 15 (10.6%) patients. Of 141 patients, 110 patients received post-recurrence therapy. The overall 1- and 3-year PRS rates were 50.7% and 28.4%, respectively. Extensive pulmonary resection (P=0.001), poor histologic differentiation (P=0.009), symptom at initial recurrence (P=0.000), no pulmonary metastasis (P=0.006), no post-recurrence therapy (P=0.001) were significant risk factors in univariate analysis. Multivariate analysis revealed that extent of pulmonary resection [hazard ratio (HR), 2.039; 95% confidence interval (CI), 1.281 to 3.244; P=0.003; risk score 1.0], histologic differentiation HR, 3.125; 95% CI, 1.976 to 4.941; P=0.000; risk score 1.5), symptom (HR, 3.154; 95% CI, 2.000 to 4.972; P=0.000; risk score 1.5) and post-recurrence therapy (HR, 2.330; 95% CI, 1.393 to 3.899; P=0.001; risk score 1.1) were significant prognostic factors. The recurred patients whose risk score sums were 1.1 or less were assigned to Group I; between 1.5 and 2.1, to Group II; and more than 2.5, to Group III. Significant differences in their PRS rates were confirmed (P=0.000).

Conclusions: Extent of pulmonary resection, histologic differentiation, symptom and post-recurrence therapy are a prognostic factor for PRS. Based on the hazard ratios of each factors, the risk scores were yielded. And the recurred patients were stratified according to the sum of their risk scores based on their PRS rates. Therefore, these results may help advancements in making predictions for their prognosis and the improvement of PRS.