Immune-associated molecular occurrence and prognosis predictor of hepatocellular carcinoma: an integrated analysis of GEO datasets

Figures & data

Figure 1. Identification of a differentially expressed immune-associated gene signature for HCC. (a) Volcano plot shows differentially expressed genes based on an absolute log₂ fold change (FC) >1.5 and adjusted P value < 0.05; black circles indicate four immune-associated genes. (b) GO analysis of the top 250 differentially expressed genes. (c) Circos plot illustrating the relationship between differentially expressed immune-associated genes and GO terms for HCC. (d) Univariate analysis of the relationship between differentially expressed immune-associated genes and HCC overall survival. (e) Univariate analysis of the relationship between differentially expressed immune-associated genes and HCC disease-free survival

Table 1. Sensitivity and specificity of the immune-associated molecular occurrence predictor in the training cohort, validation cohort, and low-AFP cohort

Clinical diagnosis	Training cohort			Validating cohort			Low-AFP cohort
	HCC tissues	adjacent normal tissues	Percent	HCC tissues	adjacent normal tissues	Percent	HCC tissues	adjacent normal tissues	Percent
HCC tissues	220	5	97.78%	230	11	95.44%	112	6	94.92%
adjacent tissues	11	210	95.02%	8	185	95.85%	9	211	95.91%

Figure 2. Development and validation of an IMOP for HCC. (a) Development of the IMOP via hierarchical clustering analysis of the training cohort. (b) ROC validation of IMOP in the training cohort, validation cohort, and low-AFP cohort

Figure 3. Development and validation of an IMPP for HCC. (a) Development of the IMPP via hierarchical clustering. (b) Overall survival was significantly lower in the immunodeficient group than in the normal immunity group. (c) Disease-free survival was also significantly lower in the immunodeficient group than in the normal immunity group. (d) Nomogram for predicting 5- and 10-year overall survival for HCC patients based on IMPP and clinicopathological characteristics (MN, Multinodular; BCLC, Barcelona Clinic liver cancer; CLIP, Cancer of the Liver Italian Program). (e) Calibration plot illustrating agreement between observed and nomogram-predicted 5- and 10-year outcomes. The 45° dashed line represents perfect prediction; the actual results from our nomogram are indicated by blue and red lines

Table 2. Univariate and multivariate Cox regression analyses of disease-free survival and overall survival between HCC patient groups stratified by the immune-associated molecular prognosis predictor

Characteristics	Univaries analysis		Multivaries analysis
	Hazard ratio (95% CI)	P Value	Hazard ratio (95% CI)	P Value
Disease-free survival
Group (immunodeficient group or normal immunity group)	1.537(1.074–2.200)	0.019	-	0.145
Gender (male or female)	1.480(1.071–2.047)	0.018	1.387(1.001–1.921)	0.049
MTS (≤5 cm or >5 cm)	0.832(0.692–1.001)	0.051	-	0.880
Cirrhosis (no or yes)	-	0.053	-	0.122
TNM stage (I-II stage or III stage)	0.668(0.547–0.816)	0.000	-	0.746
BCLC stage (0-A stage or B-C stage)	0.609(0.500–0.741)	0.000	0.627(0.514–0.764)	0.000
CLIP stage (0–1 stage or 2–5 stage)	0.699(0.573–0.854)	0.000	-	0.754
Overall survival
Group (immunodeficient group or normal immunity group)	2.029(1.316–3.129)	0.001	1.832(1.173–2.859)	0.008
MTS (≤5 cm or >5 cm)	0.707(0.570–0.877)	0.002	-	0.588
MN (no or yes)	-	0.056	1.729(1.241–2.410)	0.001
Cirrhosis (no or yes)	0.464(0.230–0.935)	0.032	-	0.052
TNM stage (I-II stage or III stage)	0.535(0.428–0.671)	0.000	-	0.119
BCLC stage (0-A stage or B-C stage)	0.533(0.426–0.666)	0.000	0.540(0.380–0.767)	0.001
CLIP stage (0–1 stage or 2–5 stage)	0.561(0.449–0.701)	0.000	0.628(0.444–0.889)	0.009

MTS, Main tumor size; MN, Multinodular; BCLC, Barcelona Clinic liver cancer; CLIP, Cancer of the Liver Italian Program.

Table 3. Correlation analysis between immune-associated molecular prognosis predictor and the clinicopathological characteristics of HCC patients

Characteristics		immunodeficient group	normal immunity group	Χ^2	P Value
Gender	Male	90(85.6)	101(105.4)	3.073	0.080
	Femal	9(13.4)	21(16.6)
Age(years)	≤60	87(81.1)	94(99.9)	4.324	0.038^▲
	>60	12(17.9)	28(22.1)
HBV virtal status	AVR-CC	27(24.8)	29(31.2)	0.463	0.496
	CC	67(69.2)	89(86.8)
ALT	≤50 U/L	53(58.2)	77(71.8)	2.071	0.150
	>50 U/L	46(40.8)	45(50.2)
AFP	≤300 ng/ml	40(53.0)	78(65.0)	12.707	<0.001^▲
	>300 ng/ml	58(45.0)	42(55.0)
Main tumor size	≤5 cm	61(63.0)	79(77.0)	0.317	0.573
	>5 cm	38(36.0)	42(44.0)
Multinodular	no	73(78.8)	103(97.2)	3.851	0.050^▲
	yes	26(20.2)	19(24.8)
Cirrhosis	no	5(8.9)	13(11.1)	3.348	0.067
	yes	94(90.1)	109(102.9)
TNM stage	I-II stage	68(76.1)	102(93.9)	6.930	0.008^▲
	III stage	30(21.9)	19(27.1)
BCLC stage	0-A stage	68(75.2)	100(92.8)	5.327	0.021^▲
	B-C stage	30(22.8)	21(28.2)
CLIP stage	0–1 stage	69(76.5)	102(94.5)	6.104	0.013^▲
	2–5 stage	29(21.5)	19(26.5)

NA, not available; HBV, Hepatitis B Virus; AVR-CC, active viral replication chronic carrier; CC, chronic carrier; ALT, alanine aminotransferase; AFP, Alpha Fetal Protein; BCLC, Barcelona Clinic liver cancer; CLIP,Cancer of the Liver Italian Program.

▲ There are statistically significant between the two groups.

Figure 4. The proportion of CD8⁺ T cells significantly differed between the HCC IMOP–stratified immunodeficient and normal immunity groups. (a) Stacked bar chart showing the relative distribution immune cells (22 cell types) in each sample. (b) Box plot illustrates relative percent of tumor-infiltrating immune cells between the HCC IMOP–stratified immunodeficient and normal immunity groups. (c) Violin plot shows differential expression of CD8⁺ T-cell markers between the HCC IMOP–stratified immunodeficient and normal immunity groups

Data availability statement

Data analyzed during the present study are available at NCBI GEO: GSE14520 and GSE36376.