Table 1. Demographics and clinicopathological characteristics of GBM patients in the TCGA training cohort and CGGA validation cohort based on the hypoxia signature.

Variables	TCGA cohort (Training set)			CGGA cohort (Validation set)
	Total (n=151)	Low risk (n=76)	High risk (n=75)	Total (n=350)	Low risk (n=175)	High risk (n=175)
Age (years)	59.6±13.7	58.8±13.4	60.4±14.0	48.1±13.3	47.2±13.1	48.9±13.5
Sex
Female	53	21	32	139	63	76
Male	98	55	43	211	112	99
KPS
< 80	32	15	17	NA
>= 80	81	41	40	NA
NA	38	20	18	NA
Pharmacotherapy
TMZ	64	37	27	61 (No)	24	37
TMZ+BEV	26	10	16	269 (Yes)	139	130
Others (No TMZ)	19	10	9	-	-	-
No or NA	42	19	23	20 (NA)	12	8
Radiotherapy
No	22	11	11	48	19	29
Yes	122	63	59	283	146	137
NA	7	2	5	19	10	9
Surgery
Biopsy only	16	10	6	NA
Tumor resection	135	66	69	NA
IDH status
Wildtype	147	68	75	270	113	157
Mutant	8	8	0	80	62	18
MGMT promoter status
Methylated	66	30	36	NA
Unmethylated	85	46	39	NA
TERT status
Wildtype	146	73	73	NA
Mutant	5	3	2	NA
BRAF status
Wildtype	146	74	72	NA
Mutant	5	2	3	NA
ATRX status
Wildtype	140	68	72	NA
Mutant	11	8	3	NA
EGFR status
Wildtype	97	42	55	NA
Mutant	54	34	20	NA
1p/19q status
Non-codeletion	NA			323	152	171
Codeletion	NA			17	15	2
NA	NA			10	8	2

GBM, glioblastoma; NA, not available; KPS, Karnofsky performance score; TMZ, temozolomide; BEV, bevacizumab; PCV, procarbazine lomustine vinCRISTine.

“Others (No TMZ)” in pharmacotherapy included PCV, PCV+BEV, and other drugs, including avastin, carmustine, and irinotecan.