BACKGROUND: The prognostic significance of cyclin A2 overexpression in non-small cell lung cancer (NSCLC) is controversial. METHODS: To understand the effect of cyclin A2 on recurrence in NSCLC, we retrospectively analyzed the expression of Bcl-2, cyclin A2, E-cadherin, Ki-67, and p53 using immunohistochemistry in 635 NSCLCs. RESULTS: Overexpression of cyclin A2 was found in 466 (73%) of 635 NSCLCs, and recurrence occurred in 291 (46%) of 635 NSCLCs with a median follow-up of 5.4 years. The relationship between recurrence and cyclin A2 overexpression was not homogenous by pathologic stage (Breslow-Day test for homogeneity, P = 0.007). Overexpression of cyclin A2 was associated with poor recurrence-free survival (RFS) in 374 stage I NSCLCs (P = 0.02), and RFS was worse in patient with negative expression of Bcl-2 than those with positive expression of Bcl-2. Cox proportional hazard analysis showed that stage I NSCLC patients with overexpression of cyclin A2 and negative expression of Bcl-2 had poorer RFS (hazard ratio = 3.86, 95% confidence interval = 1.07-15.77; P = 0.03) than those with normal expression of cyclin A2 and Bcl-2, after adjusting for age, adjuvant radiotherapy, and histology. Neural network and generalized linear model including cyclin A2 and Bcl-2 showed best performance in the prediction of recurrence; error rates for neural network and generalized linear model were 15% and 12%, respectively. CONCLUSIONS: Negative effect of cyclin A2 on RFS in stage I NSCLC was aggravated by negative expression of Bcl-2.
BACKGROUND: The prognostic significance of cyclin A2 overexpression in non-small cell lung cancer (NSCLC) is controversial. METHODS: To understand the effect of cyclin A2 on recurrence in NSCLC, we retrospectively analyzed the expression of Bcl-2, cyclin A2, E-cadherin, Ki-67, and p53 using immunohistochemistry in 635 NSCLCs. RESULTS: Overexpression of cyclin A2 was found in 466 (73%) of 635 NSCLCs, and recurrence occurred in 291 (46%) of 635 NSCLCs with a median follow-up of 5.4 years. The relationship between recurrence and cyclin A2 overexpression was not homogenous by pathologic stage (Breslow-Day test for homogeneity, P = 0.007). Overexpression of cyclin A2 was associated with poor recurrence-free survival (RFS) in 374 stage I NSCLCs (P = 0.02), and RFS was worse in patient with negative expression of Bcl-2 than those with positive expression of Bcl-2. Cox proportional hazard analysis showed that stage I NSCLCpatients with overexpression of cyclin A2 and negative expression of Bcl-2 had poorer RFS (hazard ratio = 3.86, 95% confidence interval = 1.07-15.77; P = 0.03) than those with normal expression of cyclin A2 and Bcl-2, after adjusting for age, adjuvant radiotherapy, and histology. Neural network and generalized linear model including cyclin A2 and Bcl-2 showed best performance in the prediction of recurrence; error rates for neural network and generalized linear model were 15% and 12%, respectively. CONCLUSIONS: Negative effect of cyclin A2 on RFS in stage I NSCLC was aggravated by negative expression of Bcl-2.