Multivariate analysis for prognostic predictors in non-small cell lung cancer

Abstract

To determine the prognostic predictors for non-small cell lung cancer ( NSCLC) with multivariate analysis. From 1981 to 1994, 158 patients with NSCLC had been treated by surgery alone or combined with radiotherapy and/ or chemotherapy . Their clinical characteristics were collected. All histological slides were reviewed to evaluate the histological type, tumor differentiation and vascular and lymphatic vessel invasion. Oncoprotein overexpressions were measured by immunohistochemistry assay ( ABC) , including Pan-ras, c-myc, c-erb B2, EGFR and p53. PCNA was also measured by ABC. The endpoints for prognosis consisted of survival, local control and distant metastasis. They were calculated by Kaplan-Meier method with Log-rank test determining the difference between subgroups. Cox model was used for multivariate analysis. The prognostic index ( PI) was calculated basing on the results of multivariate analysis, then the high risk group was determined by PI. The overall actual 5-year survival rate was 44%. The unfavorable predictors were late T and N of TNM stage ( P = 0. 025 and < 0. 001) , lymphatic vessel invasion ( P= 0. 002) and multi-oncoprotein overexpression ( P= 0. 018) . The overall median of PI was 1. 71( 0. 23-3. 75) . The patients were divided into three groups: good prognosis group ( 42 cases, PI ≤ 1. 3) , moderate prognosis group ( 63 cases, 1. 3 < PI ≤ 2. 2) and bad prognosis group ( 53 cases, PI > 2. 2) . There was significant difference in survival rate among the three groups ( P < 0. 001) , 5-year survival rate was 77%, 45% and 21%, respectively. The unfavorable predictors are late T, late N, lymphatic vessel invasion and multi-oncoprotein overexpression. PI could be used to determine the high risk group in patients with NSCLC.