Development of a diagnostic nomogram for alpha-fetoprotein-negative hepatocellular carcinoma based on serological biomarkers

Table 1 Baseline characteristics of the study population, n (%)

	HCC (n = 180)	Healthy (n = 100)	F/χ2/Z value	P value
Demographics
Age, yr, mean (SD), gender	52.46 (9.18)	49.79 (8.26)	2.318	0.129
Male	165 (91.7)	85 (85)	2.987	0.084
Female	15 (8.3)	15 (15)
LFTs
AST (U/L), median (IQR)	47 (33, 76.75)	20 (16, 25)	-11.534	< 0.001
ALT (U/L), median (IQR)	38 (26, 59.75)	21 (18, 25.75)	-9.764	< 0.001
GGT (U/L), median (IQR)	86 (45, 147)	26 (20, 36)	-9.744	< 0.001
ALP (U/L), median (IQR)	100.5 (67.25, 149)	62 (53, 78)	-7.354	< 0.001
HCC biomarkers
AFP (U/L) median (IQR)	25 (7.49, 572.6)	4 (3, 5)	-11.269	< 0.001
GP73 (ng/mL) median (IQR)	147.5 (111, 217.5)	54 (45, 73.75)	-11.775	< 0.001
DCP (ng/mL) median (IQR)	10 (7, 513)	6 (2.25, 8)	-8.172	< 0.001
CK18-M65 (U/L) median (IQR)	802 (469.5, 1769)	119 (94, 194.5)	-12.536	< 0.001
CK18-M30 (U/L) median (IQR)	407.5 (219, 890.75)	99 (80, 144)	-10.572	< 0.001

HCC: Hepatocellular carcinoma; AST: Aspartate transaminase; ALT: Alanine transaminase; GGT: Gamma-glutamyl transferase; ALP: Alkaline phosphatase; AFP: Alpha-fetoprotein; LFTs: Liver function tests; DCP: Des-γ-carboxyprothrombin.

Table 2 Comparison of positive rates of biomarkers

	HCC (n = 180)	AFP (-) (n = 86)	Healthy (n = 100)	P value
GP73				< 0.001
Positive rate	80 (144/180)	75.6 (65/86)	7 (7/100)
Negative rate	20 (36/180)	24.4 (21/86)	93 (93/100)
DCP				< 0.001
Positive rate	70 (126/180)	72.1 (62/86)	29 (29/100)
Negative rate	30 (54/180)	21.4 (24/86)	71 (71/100)
CK18-M65				< 0.001
Positive rate	87.8 (158/180)	87.2 (75/86)	6 (6/100)
Negative rate	12.2 (22/180)	12.8 (11/86)	94 (94/100)
CK18-M30				< 0.001
Positive rate	79.4 (143/180)	73.3 (63/86)	12 (12/100)
Negative rate	20.6 (37/180)	26.7 (23/86)	88 (88/100)

HCC: Hepatocellular carcinoma; AFP: Alpha-fetoprotein.

Table 3 Univariate and multivariate logistic regression analysis for alpha-fetoprotein-negative hepatocellular carcinoma

Variable	Univariate analysis		Multivariate analysis
	OR (95%CI)	P value	OR (95%CI)	P value
GP73 > 103 ng/mL	41.122 (16.5, 102.3)	< 0.001	14.13 (4.1, 48.8)	< 0.001
DCP > 8 ng/mL	6.325 (3.34, 12)	< 0.001	3.5 (1.15, 10.9)	0.028
CK18-M65 > 286 U/L	106.818 (37.76, 302.21)	< 0.001	44.141 (13.76, 141.56)	< 0.001
CK18-M30 > 183U/L	20.08 (9.31, 43.35)	< 0.001	-	-

Table 4 Performance of models for alpha-fetoprotein-negative hepatocellular carcinoma

Model performance measures				CK18-M65
	Model 1	Model 2	Model 3
	CK18-M65, GP73, DCP	CK18-M65, GP73	CK18-M65, DCP
AIC	41	44	65	70.5
Mean absolute error	0.016	0.018	0.03	0.022
Mean squared error	0.00048	0.00039	0.00125	0.00049
0.9 Quantile of absolute error	0.031	0.025	0.047	0.023
C statistic	0.951 (0.916-0.986)	0.946 (0.913-0.979)	0.931 (0.89-0.97)	0.906 (0.864-0.948)
P value	< 0.001	< 0.001	< 0.001	< 0.001
IDI (95%CI), %	1 (Reference)	0.0036 (-0.0239-0.031)	0.186 (0.1223-0.2496)	0.0721 (0.0363-0.1079)
P value		0.79	< 0.001	< 0.001
NRI (95%CI), %
Categorical	1 (Reference)	0.04 (-0.0101-0.0901)	0.287 (0.1484 - 0.4246)	-0.0019 (-0.0698-0.0661)
P value		0.12	< 0.001	0.96
Continuous	1 (Reference)	-0.2326 (-0.5166-0.0515)	0.866 (0.6135 - 1.1186)	0.8781 (0.6227-1.1336)
P value		0.11	< 0.001	< 0.001

AIC: Akaike information criterion; IDI: Integrated discrimination improvement; NRI: Net reclassification improvement.