PODXL might be a new prognostic biomarker in various cancers: a meta-analysis and sequential verification with TCGA datasets

Table 2 Subgroup analysis of pooled HR for OS

Categories	No. of studies	No. of patients	Pooled HR (95%CI)		Heterogeneity
Categories	No. of studies	No. of patients	Fix/Random	P-value	I² (%)	P-value
OS	13	2272	2.33 (1.76, 3.09)	0	63.4	0.001
Cancer type	8
Colorectal cancer	3	834	1.79 (1.35, 2.37)	0	0	0.499
Pancreatic cancer	2	172	2.98 (1.95, 4.55)	0	0	0.391
Gastric cancer	2	119	2.76 (0.48, 15.84)	0.256	59.9	0.114
Urothelial bladder cancer	2	443	2.14 (1.48, 3.10)	0	0	0.880
Other cancers	4	704	2.60 (1.45, 4.66)	0.001	83.3	0
Analysis
K-M curve	4	355	1.85 (1.17, 2.95)	0.009	0	0.89
Multivariate	9	2017	2.59 (1.77, 3.80)	0	74.7	0
Antibody type
Monoclonal antibody	2	157	2.25 (1.36, 3.73)	0.002	0	0.975
Polyclonal antibody	6	1672	2.55 (1.45, 4.50)	0.001	0	81.6
M + P	2	443	2.14 (1.48, 3.10)	0	0	0.880
Ethnicity
European	6	1361	1.84 (1.47, 2.30)	0	0	0.834
Asian	6	730	3.49 (2.02, 6.02)	0	64.5	0.015
North American	1	181	1.67 (1.23, 2.29)		–	–
Sample size
< 150	8	628	2.46 (1.81, 3.33)	0	0	0.536
≥150	5	1644	2.36 (1.53, 3.65)	0	81.3	0

ISSN: 1471-2407