Development of a computed tomography-based radiomics nomogram for prediction of transarterial chemoembolization refractoriness in hepatocellular carcinoma

Table 1 Patient characteristics

Characteristic	Training cohort		Validation cohort		P value
	n	%	n	%
Before treatment	137		81
Age (yr)					0.778
< 60	87	63	49	60
≥ 60	50	37	32	40
Sex					0.159
Male	91	67	45	55
Female	46	33	36	45
HBsAg status					0.950
Positive	110	80	65	80
Negative	27	20	16	20
Child-Pugh class					0.879
A	84	61	51	62
B	53	39	30	38
Largest tumor size (mean ± SD, cm)					0.321
< 5	91	66	47	58
≥ 5	46	34	34	42
Tumor distribution					0.360
Solitary	89	64	45	56
Multiple	48	36	36	44
BCLC stage					0.249
A	20	15	16	19
B	117	85	65	81
AFP (IU/mL)					0.410
< 200	29	21	21	26
≥ 200	108	79	60	74
ECOG performance status score					0.321
0	42	30	31	38
1	85	62	43	53
2	10	8	7	9
AST (U/L)					0.106
< 40	50	36	37	46
≥ 40	87	64	44	54
Irregular tumor margin					0.579
Absent	62	45	33	41
Present	75	55	48	59
Capsule					0.165
Absent	67	48	31	38
Present	70	52	50	62

BCLC: Barcelona Clinic Liver Cancer; AST: Aspartate aminotransferase; AFP: Alpha fetoprotein; ECOG: Eastern Cooperative Oncology Group.

Table 2 Univariate and multivariate-Cox regression analyses of prognostic factors in hepatocellular carcinoma patients treated by transarterial chemoembolization in the training cohort

Variable	Univariate analysis			Multivariate analysis				Score
Before first TACE	Hazard ratio	95%CI	P value	Hazard ratio	95%CI	P value	B
Age (yr), < 60/≥ 60	0.781	0.324-1.012	0.124
Sex, male/female	0.891	0.456-1.213	0.211
HBsAg status, positive/negative	0.671	0.319-0.987	0.121
Child-Pugh class, A/B	1.178	0.614-1.418	0.256
Largest tumor size (cm), < 5/≥ 5	1.619	1.671-2.341	0.005	1.312	0.981-1.992	< 0.001	0.4168	1
Tumor number, solitary/multiple	1.987	1.561-2.354	0.007	1.289	1.481-2.002	< 0.001	0.3178	1
BCLC stage, A/B	1.671	1.319-1.987	<0.001	1.789	1.289-2.112	< 0.001	0.4288	1
AFP (IU/mL), < 200/≥ 200	0.678	0.214-1.018	0.128
ECOG performance status score, 0/1/2	1.102	0.781-1.456	0.199
AST (U/L), < 40/≥ 40	0.543	0.178-0.967	0.089
Irregular tumor margin, absent/present	1.562	1.211-1.897	< 0.001	1.457	1.090-2.089	< 0.001	0.404	1
Capsule, absent/present	1.432	1.121-1.976	0.032	1.321	1.007-1.764	0.082

The regression coefficients (B) in multivariate-Cox regression analysis were multiplied by 2 and rounded to calculate the clinical score. The clinical model used the median value (2) of total score (4) to stratify patients into high and low-risk group. HCC: Hepatocellular carcinoma; TACE, Transarterial chemoembolization; BCLC, Barcelona Clinic Liver Cancer; AST: Aspartate aminotransferase; AFP: Alpha fetoprotein; ECOG: Eastern Cooperative Oncology Group; CI: Confidence interval.

Table 3 Univariate analysis of prognostic factors in hepatocellular carcinoma patients treated by transarterial chemoembolization in the training cohort

Variable	n	Overall survival (mo)		P value
		Median	95%CI
Before treatment
Age (yr)				0.213
< 60	87	19.8	16.8-23.4
≥ 60	50	17.6	14.3-21.1
Sex				0.178
Male	91	15.4	12.5-19.6
Female	46	19.6	14.3-23.1
HBsAg status				0.121
Positive	110	14.9	12.3-19.2
Negative	27	18.1	14.1-22.8
Child-Pugh stage
Child-Pugh class				0.301
A	84	23.5	19.1-32.4
B	53	19.8	13.7-23.8
Largest tumor size (cm)				0.032
< 5	91	19.6	19.7-24.2
≥ 5	46	14.1	14.1-18.8
Tumor number				0.029
Solitary	89	19.1	17.4-22.7
Multiple	48	14.7	13.8-18.1
BCLC stage				0.033
A	20	26.7	22.3-31.2
B	117	16.1	13.7-24.2
AFP (IU/mL)				0.041
< 200	29	20.7	17.2-29.4
≥ 200	108	16.8	9.5-18.2
ECOG performance status score				0.195
0	42	21.6	13.2-26.7
1	85	19.3	10.3-24.4
2	10	17.1	9.1-22.3
AST (U/L)				0.261
< 40	50	18.3	15.2-23.7
≥ 40	87	16.9	12.3-22.9
Irregular tumor margin				0.018
Absent	62	20.5	14.7-27.2
Present	75	12.6	8.7-16.4
Capsule				0.087
Absent	67	16.7	12.1-22.4
Present	70	19.3	13.5-25.1
Radiomics score				0.003
< 0.1	102	23.3	19.8-28.6
≥ 0.1	35	10.3	6.5-14.3

HCC: Hepatocellular carcinoma; TACE: Transarterial chemoembolization; BCLC: Barcelona Clinic Liver Cancer; AST: Aspartate aminotransferase; AFP: Alpha fetoprotein; ECOG: Eastern Cooperative Oncology Group; CI: Confidence interval.

Table 4 Multivariate stepwise backward Cox regression analysis of prognostic factors in hepatocellular carcinoma patients treated by transarterial chemoembolization in the training cohort

Variable	Hazard ratio ratio	95%CI	B	Score	P value (Cox regression)
Before treatment
BCLC stage
A	1		0	0
B	2.3	1.2-3.1	0.39	1	0.032
Irregular tumor margin
Absent	1		0	0
Present	1.9	0.7-3.3	0.42	1	0.028
Largest tumor size (cm)
< 5	1		0	0
≥ 5	1.7	0.6-2.9	0.47	1	0.017
Tumor number
Solitary	1		0	0
Multiple	2.1	1.1-3.1	0.33	1	0.021
Radiomics score
< 0.1	1		0	0
≥ 0.1	3.9	3.1-8.8	1.72	3	< 0.001

The regression coefficients (B) in multivariate-Cox regression analysis were multiplied by 2 and rounded to calculate the radiomics-clinical score. The radiomics-clinical model used the median value (3.5) of total score (7) to stratify patients into high and low-risk group. HCC: Hepatocellular carcinoma; TACE: Transarterial chemoembolization; BCLC: Barcelona Clinic Liver Cancer; B: B regression coefficient; CI: Confidence interval.

Table 5 Predictive performance of each model/score

Prediction model	Training cohort C-index (95%CI)	Validation cohortC-index (95%CI)	P value
			3 vs 1	3 vs 2	3 vs 4	vs 5
1 Clinical model	0.643 (0.613-0.712)	0.629 (0.601-0.678)	0.025a/0.023b	0.003a/0.002b	0.007a/0.006b	0.009a/0.004b
2 Radiomics score	0.723 (0.634-0.778)	0.734 (0.641-0.793)
3 CT-based radiomics nomogram	0.844 (0.762-0.901)	0.831 (0.742-0.881)
4 ART score	0.714 (0.632-0.771)	0.690 (0.601-0.761)
5 ABCR score	0.732 (0.646-0.801)	0.701 (0.632-0.789)

^aSignificant difference of C-index between computed tomography (CT)-based radiomics nomogram vs other models/scores in the training cohort.

^bSignificant difference of C-index between CT-based radiomics nomogram vs other models/scores in the validation cohort. CI: Confidence interval; CT: Computed tomography; ART: Assessment for Retreatment with Transarterial Chemoembolization; ABCR: α-Fetoprotein, Barcelona Clinic Liver Cancer, Child-Pugh, and Response.