Table 5.

Endometrial cancer-specific mortality risk prediction after adding dichotomous ER or PR expression at select cut points to standard predictors

	Dichotomous Hormone Receptor Variable Added to Standard Predictors
	ER Cut Point (n=338)			PR Cut Point (n=334)
	30%	40%	50%	10%	80%
C-Indexa	84.6%	84.5%	84.1%	76.5%	79.7%
Reclassification Metricb
Number (%) moved to higher risk category	36 (11%)	43 (13%)	52 (15%)	12 (4%)	167 (50%)
Number (%) moved to lower risk category	14 (4%)	13 (4%)	13 (4%)	9 (3%)	6 (2%)
Total number (%) reclassified	50 (15%)	56 (17%)	65 (19%)	21 (6%)	173 (52%)
Reclassification Calibration Statistic (p-value)	11.04 (0.0009)	8.36 (0.004)	7.58 (0.006)	0.12 (1)	0.28 (0.6)
Net Reclassification Improvement	5.6%	4.0%	1.2%	0.6%	−23.0%
Integrated Discrimination Improvement	−1.5%	−2.0%	−0.9%	3.0%	1.6%

ER and PR each measured on a semi-quantitative percent positive cells scale of 0–100%, then dichotomized at the respective cut point into Marker+ and Marker− groups. Marker+ was defined as expression above the cut point for a cut point of 0% and as expression at or above the cut point for all other cut points. Marker− was defined as expression of 0% at a cut point of 0% and as expression below the cut point for all other cut points.

^aEach c-index value in the table is for a Cox model estimating 10-year risk of endometrial cancer-specific death based on standard clinical predictors (tumor stage, tumor grade, and histologic type) plus the respective dichotomous marker status variable. C-index for a model of standard predictors only was 74.5%.

^bReclassification metrics compare a Cox model estimating 10-year risk of endometrial cancer-specific death based on standard predictors (tumor stage, tumor grade, and histologic type) to a Cox model based on standard predictors plus dichotomous marker status defined by the respective cut point. Mortality risk categories were 0–5%, 5–10%, and >10%.

ER=Estrogen Receptor, PR=Progesterone Receptor