Table 3.

Firth’s penalized logistic regression for stratification of F-IR (n = 220) by Decipher compared to VL/LR (n = 427) predicting AP

Model	Variable	Odds ratio (95% CI)	p value
NCCN	NCCN F-IR vs. VL/LR	1.71 (1.04−2.79)	0.034*
NCCN stratified by Decipher	NCCN F-IR + GC low vs. NCCN VL/LR	1.48 (0.85−2.53)	0.160
	NCCN F-IR + GC int vs. NCCN VL/LR	1.26 (0.33−3.59)	0.700
	NCCN F-IR + GC high vs. NCCN VL/LR	6.83 (2.45−18.31)	<0.001*

F-IR NCCN favorable-intermediate risk, VL/LR NCCN very low/low risk, AP adverse pathology, NCCN National Comprehensive Cancer Network, GC genomic classifier–Decipher

*p value < 0.05