Predicting hepatocellular carcinoma: A new non-invasive model based on shear wave elastography

Table 1 The initial patient features of the experimental cohort were assessed, n (%)

Variable	All patients (n = 720)	Benign (n = 200)	Malignant (n = 520)	P value1
Clinical data
Age (years)	55.2 ± 12.0	54.6 ± 12.5	55.5 ± 11.8	0.353
Sex (%)				0.423
Men	572 (79.4)	155 (77.5)	417 (80.2)
Women	148 (20.6)	45 (22.5)	103 (19.8)
Body mass index (kg/m2)	24.0 ± 3.3	23.9 ± 3.2	24.0 ± 3.4	0.678
Nationality (%)				0.361
Han nationality	698 (96.9)	192 (96)	506 (97.3)
Not Han nationality	22 (3.1)	8 (4)	14 (2.7)
History of tumor (%)				0.210
Yes	62 (8.6)	13 (6.5)	49 (9.4)
No	658 (91.4)	187 (93.5)	471 (90.6)
Laboratory data
ALT (IU/L)	44.6 ± 105.4	37.8 ± 60.3	47.2 ± 118.2	0.284
AST (IU/L)	41.5 ± 100.5	34.8 ± 62.1	44 ± 111.8	0.269
GGT (IU/L)	83.9 ± 126.4	75.1 ± 139.3	87.4 ± 121.0	0.243
Prothrombin time (s)	12.1 ± 3.1	11.9 ± 1.2	12.2 ± 3.6	0.356
INR	1.0 ± 0.4	1.0 ± 0.1	1.0 ± 0.5	0.372
PLT (109/L)	182.1 ± 73.2	207.4 ± 73.7	172.4 ± 70.7	< 0.001
Albumin (g/L)	41.3 ± 4.1	41.3 ± 4.9	41.4 ± 3.8	0.780
Total bilirubin (µmol/L)	67.6 ± 5.3	67.5 ± 5.5	67.7 ± 5.3	0.638
Ultrasonic data				0.008
Single	562 (78.1)	143 (71.5)	419 (80.6)
Multiple	158 (21.9)	57 (28.5)	101 (19.4)
Diameter (cm)	5.3 ± 3.0	5.2 ± 3.5	5.3 ± 2.8	0.629
Shape (%)				0.102
Roundness	363 (50.4)	91 (45.5)	272 (52.3)
Ellipse	357 (49.6)	109 (54.5)	248 (47.7)
Echo (%)				0.064
Hypoecho	236 (32.8)	76 (38.0)	160 (30.8)
Hyperecho	484 (67.2)	124 (62.0)	360 (69.2)
Edge (%)				0.010
Smooth	398 (55.3)	126 (74.4)	272 (52.3)
Coarse	322 (44.7)	74 (25.6)	248 (47.7)
Cirrhosis (%)				< 0.001
Yes	239 (33.2)	35 (17.5)	204 (39.2)
No	481 (66.8)	165 (82.5)	316 (60.8)
Blood (%)				< 0.001
Yes	462 (64.2)	97 (48.5)	365 (70.2)
No	258 (35.8)	103 (51.5)	155 (29.8)
Emax of the tumors (m/s)	2.4 ± 1.0	2.1 ± 0.6	2.5 ± 1.1	< 0.001
Emean of the tumors (m/s)	1.8 ± 0.5	1.6 ± 0.4	1.9 ± 0.5	< 0.001
Emax of the periphery of tumors (m/s)	1.6 ± 1.0	1.3 ± 0.4	1.7 ± 1.2	< 0.001
Emean of the periphery of tumors (m/s)	1.5 ± 0.7	1.3 ± 0.4	1.5 ± 0.7	< 0.001

¹P value for comparisons between benign cohort and malignant cohort.

The data enclosed between parenthesis represents the intervals of confidence at 95%. ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; GGT: G-glutamyl transferase; PLT: Platelet count; Emax: Maximal elasticity; Emean: Mean elasticity.

Table 2 Both univariable and multivariable logistic regression approaches to assess the probability of liver cancer in the experimental cohort (n = 720)

Variable	Univariable		Multivariable
	Odds ratio	P value	Odds ratio	P value
Clinical data
Age (years)	1.00 (0.99, 1.02)	0.352
Sex (%)	0.85 (0.57, 1.26)	0.424
Body mass index (kg/m2)	1.01 (0.96, 1.06)	0.678
Nationality (%)	0.66 (0.27, 1.61)	0.364
History of tumor (%)	1.50 (0.79, 2.82)	0.213
Laboratory data
ALT (IU/L)	1.00 (0.99, 1.00)	0.306
AST (IU/L)	1.00 (0.99, 1.00)	0.299
GGT (IU/L)	1.00 (0.99, 1.00)	0.249
Prothrombin time (s)	1.00 (0.93, 1.24)	0.328
INR	2.94 (0.50, 17.38)	0.235
PLT (109/L)	1.00 (0.99, 1.00)	< 0.001	1.00 (0.99, 1.00)	< 0.001
Albumin (g/L)	1.00 (0.97, 1.05)	0.780
Total bilirubin (µmol/L)	1.01 (0.98, 1.04)	0.638
Ultrasonic data
Number (%)	0.61 (0.42, 0.89)	0.009	0.61 (0.40, 0.95)	0.027
Diameter (cm)	1.01 (0.96, 1.07)	0.628
Shape (%)	0.76 (0.55, 1.06)	0.102
Echo (%)	1.38 (0.98, 1.94)	0.065
Edge (%)	1.56 (1.11, 2.17)	0.010
Cirrhosis (%)	3.04 (2.03, 4.57)	< 0.001
Blood (%)	2.50 (1.79, 3.50)	< 0.001	2.51 (1.72, 3.67)	< 0.001
Emax of the tumors (m/s)	2.86 (2.10, 3.91)	< 0.001	2.08 (1.50, 2.89)	< 0.001
Emean of the tumors (m/s)	3.95 (2.60, 6.01)	< 0.001
Emax of the periphery of tumors (m/s)	5.72 (3.51, 9.34)	< 0.001
Emean of the periphery of tumors (m/s)	5.84 (3.54, 9.63)	< 0.001	2.70 (1.54, 4.73)	0.001

The data enclosed between parenthesis represents the intervals of confidence at 95%. ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; GGT: G-glutamyl transferase; PLT: Platelet count; Emax: Maximal elasticity; Emean: Mean elasticity.

Table 3 Diagnostic performance of multivariable model, maximal elasticity of the tumors, mean elasticity of the periphery of tumors, platelet count, cirrhosis, blood and number for the diagnosis of malignant tumor in the experimental cohort (n = 720)

Parameter	AUC	Cutoff value1	Sensitivity (%)	Specificity (%)	Positive predictive value (%)	Negative predictive value (%)	Diagnostic accuracy (%)
Multivariable model	0.77 (0.73, 0.81)	0.653	84	61	85	60	78
Emean of the periphery of tumors	0.69 (0.64, 0.73)	1.375	59	71	84	40	62
Emax of the tumors	0.67 (0.63, 0.72)	2.205	61	65	82	39	62
Edge echo	0.61 (0.57, 0.65)	0.500	39	83	85	34	51
Blood	0.62 (0.56, 0.66)	0.500	70	52	79	40	065
Number	0.46 (0.41, 0.50)	0.500	19	72	64	25	34
PLT	0.34 (0.30, 0.39)	39.500	100	1	72	5	72

¹Cutoff values that maximized the sum of sensitivity and specificity were chosen.

PLT: Platelet count; Emax: Maximal elasticity; Emean: Mean elasticity; AUC: Area under the curve.