Unbiased data mining identifies cell cycle transcripts that predict non-indolent Gleason score 7 prostate cancer

BMC Urology

Table 3 Univariate logistic regression and ROC-AUC analyses for predicting BCR in patients with Gleason score 3 + 4 = 7 tumours

Variable	Odds Ratio (OR)	95% CI OR	P value OR	ROC-AUC	SE AUC	P value AUC	Youden cut-off	Sensitivity	Specificity	PPV	NPV
(up in tumour)
BUB1	1.93	1.029–3.604	4.03E-02	0.865	0.045	4.75E-16	1.54	1.000	0.78	0.22	1.00
TPX2	1.40	1.055–1.852	1.98E-02	0.872	0.060	5.07E-10	4.09	1.000	0.74	0.19	1.00
NCAPG	2.72	1.071–6.915	3.55E-02	0.855	0.067	1.23E-07	1.33	0.800	0.83	0.22	0.99
UBE2C	1.20	1.051–1.369	7.00E-03	0.889	0.050	7.21E-15	7.79	1.000	0.80	0.23	1.00
MELK	1.72	0.939–3.152	7.89E-02	0.819	0.082	1.03E-04	2.17	0.800	0.86	0.25	0.99
CCNA2	1.83	1.113–3.000	1.71E-02	0.810	0.106	3.39E-03	2.63	0.800	0.80	0.19	0.99
CCNB1	1.63	1.109–2.383	1.27E-02	0.824	0.083	1.00E-04	4.69	1.000	0.57	0.12	1.00
CDK1	1.28	1.070–1.532	6.80E-03	0.863	0.090	5.50E-05	7.32	0.800	0.90	0.33	0.99
E2F2	5.98	0.824–43.408	7.70E-02	0.720	0.119	6.39E-02	0.37	0.800	0.70	0.14	0.98
DLGAP5	1.95	0.954–3.984	6.73E-02	0.855	0.066	6.61E-08	0.86	1.000	0.64	0.14	1.00
TMEM206	0.98	0.766–1.243	8.40E-01	0.545	0.154	7.72E-01	9.11	0.800	0.52	0.09	0.98
CDKN3	1.50	1.071–2.108	1.85E-02	0.708	0.187	2.65E-01	6.47	0.600	0.95	0.43	0.98
SHMT2	1.03	0.983–1.075	2.21E-01	0.740	0.093	9.94E-03	60.61	0.800	0.71	0.14	0.98
(down in tumour)
SRD5A2	0.58	0.367–0.921	2.09E-02	0.884	0.079	1.03E-06	3.75	0.800	0.88	0.29	0.99
CSRP1	1.00	0.992–1.001	9.71E-02	0.752	0.132	5.66E-02	417.97	0.800	0.80	0.19	0.99
PGM5	0.97	0.930–1.007	1.04E-01	0.745	0.136	7.28E-02	18.57	0.600	0.90	0.27	0.97
NFIB	0.98	0.903–1.054	5.26E-01	0.578	0.129	5.45E-01	24.07	0.600	0.72	0.12	0.97

Data are from TCGA-PRAD (TCGA, 2015). n = 88. CI-confidence interval, SE-standard error. Youden cut-off is the optimal point to separate biochemical recurrence (BCR) from non-BCR. PPV-positive predictive value, NPV-negative predictive value

ISSN: 1471-2490