Table 2.

Bias-adjusted hazard ratios and optimal cutpoints for dichotomized metrics.

Metric	Dichotomization Threshold (#)	Hazard Ratio (*)	LL 95%CI	UL 95%CI	p (&)
High vs. low number of CD${163}^{+}$ TAMs	70 (12–102)	2.97	1.73	5.11	<0.001
High vs. low proportion of CD${163}^{+}$ TAMs	0.038 (0.004–0.149)	1.94	1.20	3.14	0.026
High vs. low 10th PCTL of cancer-to-CD${163}^{+}$ NNDs	13 (7–55)	0.60	0.38	0.94	0.027
High vs. low 25th PCTL of cancer-to-CD${163}^{+}$ NNDs	17 (9–42)	0.46	0.26	0.82	0.026
High vs. low median cancer-to-CD${163}^{+}$ NND	19 (13–74)	0.48	0.27	0.85	0.026
High vs. low average number of adjacent CD${163}^{+}$ TAMs	0.034 (0.022–0.087)	2.71	1.70	4.32	<0.001
High vs. low average number of communicating CD${163}^{+}$ TAMs	46 (21–57)	2.96	1.80	4.87	<0.001

(#) bootstrap-based 90% confidence interval in parentheses. (&) adjusted for multiple testing using Holm’s method. (*) bootstrap-based optimistic bias-adjusted hazard ratio.