Selected serum cytokines and nitric oxide as potential multi-marker biosignature panels for Parkinson disease of varying durations: a case-control study

Table 3 Performance of tested analytes as best single and multiple biomarker panels under ROC curve analysis

Test Group	Marker (s)	AUC	SE %	SP %	Optimum Cut-off
< 1 year PD	IFNγ	0.817	91.7	71.4	0.198
	IL-10	0.731	100.0	58.9	0.179
	TNFα	0.794	91.7	62.5	0.140
	IFNγ-IL-10	0.859	100.0	75.0	0.173
	IFNγ- TNFα	0.862	83.3	78.6	0.266
	IFNγ-IL-10 & TNFα	0.868	83.3	80.4	0.232
1–3 years PD	IFNγ	0.692	93.3	53.6	0.290
	IFNγ-IL-10	0.665	96.7	53.6	0.309
	IFNγ-NOx	0.797	100.0	51.8	0.192
	IFNγ, IL-10 & NOx	0.847	96.7	66.1	0.192
3–12 years PD	IFNγ	0.910	96.7	82.1	0.355
	IFNγ-IL-10	0.920	96.7	89.3	0.429
	IFNγ-NOx	0.924	93.3	87.5	0.416
	IFNγ, IL-10 & NOx	0.924	93.3	87.5	0.408

ROC curve combines sensitivity (SE) and specificity (SP) of a given marker for a diagnostic test from 0.5 (no discriminating power) to 1.0 (complete separation). The combos with high area under curve (AUC), SE and SP were considered as best performing marker ensembles generated from the combinatorial analysis, while those with small SE and SP were considered to be of negligible importance

ISSN: 1471-2377