. 2016 Jul 13;7(31):50582–50595. doi: 10.18632/oncotarget.10558

Table 5. Multivariate analysis of NSCLC-associated molecular markers based on stage-stratified subpopulations containing early-stage (I–IIIA) NSCLC cases (n = 59) and all non-cancer controls (n = 310). ^#^*.

Variable	^OR	95%CI of ^OR		p	StdEst
DUSP6	5.43	2.46	12.01	0.0000	0.67
GRB2	5.52	2.28	13.36	0.0002	0.70
MDM2	12.36	4.01	38.09	0.0000	0.76
EIF2S3	0.19	0.07	0.57	0.0028	−0.42
POLDIP2	0.16	0.06	0.44	0.0004	−0.68
RNF4	0.21	0.06	0.74	0.0149	−0.49
C statistic	0.883

^OR:

odds ratio; CI: confidence interval; StdEst: standardized coefficients;

The multiple logistic model contains 15 expressed genes, with controlling for age, gender, and smoking status. The other seven molecular markers (CPEB4, EXT2, IRF4, MCM4, MMD, NF1, STAT2, WEE1, and ZNF264) were not significantly associated.

The performance of this model is presented as the sensitivity and specificity depending on the cut-off value chosen:

cut-off value =0.391, Sensitivity = 0.661, Specificity = 0.952

cut-off value =0.224, Sensitivity = 0.763, Specificity = 0.900

cut-off value =0.172, Sensitivity = 0.763, Specificity = 0.855

cut-off value =0.127, Sensitivity = 0.797, Specificity = 0.800

cut-off value =0.100, Sensitivity = 0.847, Specificity = 0.735

Table 5. Multivariate analysis of NSCLC-associated molecular markers based on stage-stratified subpopulations containing early-stage (I–IIIA) NSCLC cases (n = 59) and all non-cancer controls (n = 310). #*.

Table 5. Multivariate analysis of NSCLC-associated molecular markers based on stage-stratified subpopulations containing early-stage (I–IIIA) NSCLC cases (n = 59) and all non-cancer controls (n = 310). ^#^*.