Metabolomics Identifies Biomarker Signatures to Differentiate Pancreatic Cancer from Type 2 Diabetes Mellitus in Early Diagnosis

Table 2

Relation of shared common metabolites to risk of future diseases (diabetes mellitus and pancreatic cancer).

Model	LysoPC (16 : 0)	Catelaidic acid	LysoPC (22 : 6)	Docosanamide	Cerebronic acid	Asparaginyl-histidine
Logistic regression between diabetes mellitus patients and healthy volunteers
Metabolite as a continuous variable
Per SD	1.381 (1.163–1.930)	0.137 (0.041–0.341)	1.264 (0.922–1.776)	0.271 (0.105–0.555)	0.177 (0.056–0.415)	1.845 (1.288–3.069)
	0.0063	0.0002	0.1557	0.0018	0.0006	0.0051
Metabolite as a categorical variable
1st quartile	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)
2nd quartile	0.895 (0.496–1.653)	0.616 (0.217–1.312)	1.000 (0.644–1.552)	0.405 (0.144–0.822)	0.533 (0.187–1.149)	3.886 (1.966–10.953)
3rd quartile	1.759 (0.957–3.515)	0.585 (0.210–1.192)	1.063 (0.642–1.775)	0.669 (0.238–1.416)	0.526 (0.190–1.043)	3.175 (1.556–9.048)
4th quartile	3.119 (1.536–8.877)	0.232 (0.079–0.478)	1.442 (0.680–4.067)	0.306 (0.110–0.596)	0.263 (0.093–0.521)	5.241 (2.328–17.749)
for trend	0.0001	<0.0001	0.8107	0.0005	0.0002	<0.0001
Logistic regression between pancreatic cancer patients and healthy volunteers
Metabolite as a continuous variable
Per SD	1.147 (1.015–1.364)	0.107 (0.028–0.291)	4.026 (2.079–11.805)	0.126 (0.032–0.330)	0.058 (0.009–0.210)	2.052 (1.386–3.556)
	0.0559	0.0002	0.0011	0.0004	0.0003	0.0022
Metabolite as a categorical variable
1st quartile	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)
2nd quartile	0.493 (0.248–0.853)	0.693 (0.246–1.472)	1.000 (0.329–3.036)	0.550 (0.200–1.079)	0.771 (0.274–1.635)	2.371 (1.250–5.114)
3rd quartile	1.326 (0.781–2.390)	0.579 (0.207–1.191)	2.433 (1.232–6.771)	0.638 (0.225–1.364)	0.531 (0.190–1.080)	2.988 (1.652–6.307)
4th quartile	1.671 (0.841–4.628)	0.191 (0.056–0.430)	5.241 (2.328–17.749)	0.199 (0.059–0.450)	0.203 (0.060–0.459)	4.160 (2.012–12.181)
for trend	0.0010	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001
Logistic regression between pancreatic cancer patients and diabetes mellitus patients
Metabolite as a continuous variable
Per SD	0.917 (0.844–0.989)	0.807 (0.436–1.453)	2.296 (1.547–3.921)	0.775 (0.458–1.276)	0.851 (0.468–1.517)	1.116 (0.860–1.464)
	0.0287	0.4781	0.0004	0.3234	0.5857	0.4117
Metabolite as a categorical variable
1st quartile	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)	1.0 (referent)
2nd quartile	0.550 (0.250–1.089)	1.126 (0.722–1.775)	1.000 (0.329–3.036)	1.357 (0.838–2.289)	1.446 (0.919–2.355)	0.610 (0.209–1.428)
3rd quartile	0.754 (0.418–1.280)	0.991 (0.627–1.562)	2.289 (1.170–6.334)	0.953 (0.606–1.490)	1.010 (0.629–1.614)	0.941 (0.325–2.202)
4th quartile	0.536 (0.296–0.896)	0.822 (0.286–1.894)	3.634 (1.824–10.275)	0.652 (0.234–1.326)	0.771 (0.274–1.635)	0.794 (0.276–1.833)
for trend	0.0824	0.8753	<0.0001	0.2421	0.2513	0.3034

Values are odds ratios (95% confidence intervals) for diabetes mellitus and pancreatic cancer, from logistic regressions.