Table 2. Univariable and multivariable Cox regression analyses in each data set.

Variables	Univariable modela			Multivariable model
	HR	95% CI of HR	P value	HR	95% CI of HR	P value
Training set (n=220)
Five-lncRNA risk score	1.151	1.104-1.200	<0.001	1.131	1.086-1.178	<0.001
Age	1.019	1.001-1.038	0.034	1.021	1.002-1.040	0.032
Gender	1.016	0.654-1.580	0.943	0.780	0.491-1.238	0.291
AJCC stage	3.675	2.323-5.816	<0.001	3.390	2.104-5.449	<0.001
Tumor grade	2.820	1.718-4.630	<0.001	1.752	0.999-3.071	0.050
Testing set (n=220)
Five-lncRNA risk score	1.057	1.011-1.104	0.014	1.055	1.014-1.098	<0.001
Age	1.040	1.017-1.063	<0.001	1.041	1.017-1.066	<0.001
Gender	0.904	0.540-1.513	0.701	1.339	0.777-2.309	0.294
AJCC stage	5.121	2.908-9.016	<0.001	4.440	2.419-8.150	<0.001
Tumor grade	2.374	1.326-4.252	0.004	1.542	0.834-2.852	0.167
Entire set (n=440)
Five-lncRNA risk score	1.106	1.073-1.139	<0.001	1.093	1.063-1.125	<0.001
Age	1.029	1.015-1.044	<0.001	1.030	1.015-1.045	<0.001
Gender	0.958	0.685-1.339	0.802	0.964	0.678-1.371	0.179
AJCC stage	4.217	2.956-6.015	<0.001	3.269	2.232-4.787	<0.001
Tumor grade	2.582	1.770-3.767	<0.001	1.675	1.121-2.501	0.012

^a In both univariable and multivariable Cox regression analyses, risk score, age, gender, AJCC stage and Tumor grade were evaluated as continuous variables. P<0.05 was considered statistically significant in all analyses.