Establishment of a pattern recognition metabolomics model for the diagnosis of hepatocellular carcinoma

Table 1 General characteristics of patients and healthy controls

Characteristics		C (n = 30)	Y (n = 29)	N (n = 31)	P value
Sex (Male/Female)		25/5	21/8	25/6	0.565
Age (yr)		52.93 ± 11.01	56.63 ± 9.15	51.23 ± 11.79	0.148
Pathogens	HBV	25	24	/	0.720
	HCV	1	2
	HBV + HCV	1	0
	None	3	3
AFP (ng/mL)	> 200	11	0	/	0.000
	50-199	4	1
	< 50	15	28
ALT (U/L)		162.32 ± 201.06	91.02 ± 156.39	20.34 ± 8.43	0.000
AST (U/L)		146.35 ± 112.70	114.49 ± 191.67	21.59 ± 4.51	0.012
TBIL (μmol/L)		39.21 ± 68.38	40.87 ± 42.41	9.66 ± 2.66	0.015
DBIL (μmol/L)		17.91 ± 34.43	17.90 ± 23.03	4.49 ± 1.38	0.044
TP (g/L)		62.42 ± 10.95	74.14 ± 8.05	72.31 ± 3.96	0.000
ALB (g/L)		33.51 ± 6.30	37.65 ± 7.64	45.36 ± 2.62	0.000
Child-Pugh score (A/B)		18/12	15/14	/	0.353

AFP: α-fetoprotein; ALB: Albumin; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; DBIL: Direct bilirubin; TBIL: Total bilirubin; TP: Total protein.

Table 2 Significantly altered metabolites

Retention time	m/z	Metabolites	Adduction	Adduct mass	Delta ppm	Coefficient of variation (%)	Comparison
						3.52	C vs N	Y vs N	C vs Y
9.57	166.0862	Phenylalanine	M + H	166.0863	1.00	8.36	D	U	NS
3.49	118.0864	Valine	M + H	118.0863	1.00	6.76	D	NS	NS
6.63	132.1019	Leucine	M + H	132.1019	0.00	3.34	NS	U	NS
3.58	116.0708	Proline	M + H	116.0706	1.00	14.23	D	D	D
5.42	182.0811	Tyrosine	M + H	182.0812	0.00	6.12	NS	U	NS
6.05	132.1019	Isoleucine	M + H	132.1019	0.00	3.37	NS	U	D
4.89	150.0583	Methionine	M + H	150.0583	0.00	9.28	NS	U	NS
3.16	156.0766	Histidine	M + H	156.0768	1.00	10.83	D	U	D
3.44	148.0602	Glutamic acid	M + H	148.0604	2.00	6.38	U	D	U
3.32	106.0502	Serine	M + H	106.0499	3.00	2.59	D	U	D
3.38	147.0762	Glutamine	M + H	147.0764	1.00	11.41	NS	D	D
3.44	90.0554	Alanine	M + H	90.0550	5.00	12.28	D	D	NS
5.43	165.0546	Hydroxycinnamic acid	M + H	165.0546	0.00	16.17	D	NS	D
5.43	123.0442	Benzoic acid	M + H	123.0441	1.00	10.11	D	U	NS
9.57	149.0596	Cinnamic acid	M + H	149.0597	1.00	12.29	D	U	NS
24.40	190.0497	Kynurenic acid	M + H	190.0499	1.00	6.51	D	D	U
26.39	169.0495	Vanillic acid	M + H	169.0495	0.00	3.41	D	D	U
13.83	239.0912	Trimethoxycinnamic acid	M + H	239.0914	1.00	5.08	D	U	NS
18.85	279.2318	Linolenic acid	M + H	279.2319	0.00	11.26	D	NS	D
3.10	130.0862	Pipecolinic acid	M + H	130.0863	0.00	10.58	D	U	NS
29.42	494.3235	LysoPC 16:1	M + H	494.3241	1.00	4.32	NS	D	D
22.87	542.3234	LysoPC 20:5	M + H	542.3241	1.00	3.09	NS	D	NS
17.33	548.3705	LysoPC 20:2	M + H	548.3711	1.00	2.31	D	D	NS
21.65	550.3857	LysoPC 20:1	M + H	550.3867	2.00	5.58	D	D	NS
23.13	468.3078	LysoPC 14:0	M + H	468.3085	1.00	6.27	D	D	D
19.25	478.2926	LysoPE 20:1	M + H	478.2928	0.00	8.72	D	NS	NS
17.58	181.0857	Propylparaben	M + H	181.0859	1.00	6.39	D	NS	NS
5.42	136.0756	Acetylarylamine	M + H	136.0757	0.00	2.59	D	U	D
18.20	127.0390	Trihydroxybenzene	M + H	127.0390	0.00	13.83	D	U	D
22.22	191.1428	Damascenone	M + H	191.1430	1.00	10.02	U	D	NS
10.70	181.0718	Myoinositol	M + H	181.0707	6.00	8.55	D	NS	NS
4.88	137.0457	Hydroxypurine	M + H	137.0458	1.00	9.74	U	D	U
3.48	114.0664	Creatinine	M + H	114.0662	2.00	7.83	D	NS	NS
3.82	72.0815	Pyrrolidine	M + H	72.0808	10.00	2.09	U	U	NS
11.71	195.0875	Methyl lucopyranoside	M + H	195.0863	6.00	12.43	D	NS	NS

LysoPC: Lysophosphatidyl choline; LysoPE: Lysophosphatidyl ethanolamine; U: Upregulated; D: Decreased; NS: No statistical difference.

Table 3 The efficiency of the diagnostic model

Model		Accuracy (%)	Sensitivity (%)	Specificity (%)	Positive predictive value	Negative predictive value	ROC-AUC (95%CI)	P value
Training set	LDA	95.00	100.00	90.00	0.91	1.00	0.90 (0.81-0.99)	< 0.05
	AFP	65.00	30.00	100.00	1.00	0.59	0.69 (0.52-0.86)
Validation set	LDA	78.95	100.00	60.00	0.69	1.00	0.84 (0.67-1.00)	< 0.05
	AFP	68.42	40.00	100.00	1.00	0.60	0.68 (0.41-0.94)

LDA: Linear discriminant analysis; ROC: Receiver operating characteristic curve; AUC: Area under curve.