Table 2.

Case-level agreement between automated scoring of central tissue microarrays and clinical estrogen receptor (ER) and progesterone receptor (PR) status: impact of core-to-case collapsing method

Clinical ER or PR status in the medical recorda
ERPR
NegativePositiveTotalNegativePositiveTotal
Central TMA ER or PR statusa
Weighted core-to-case collapsing methodb
 Negative427100527Negative565108673
 Positive151,0951,110Positive74803877
 Total4421,1951,637Total6399111,550
 Agreement93%Agreement88%
 Kappa (95% CI)0.83 (0.80–0.86)Kappa (95% CI)0.76 (0.73–0.79)
NegativePositiveTotalNegativePositiveTotal
Any positive core-to-case collapsing methodc
 Negative41970489Negative52983612
 Positive231,1251,148Positive110828938
 Total4421,1951,637Total6399111,550
 Agreement94%Agreement88%
 Kappa (95% CI)0.86 (0.83–0.88)Kappa (95% CI)0.74 (0.71–0.78)
  • aUsing a 10% threshold to define ER and PR positivity.

  • bWeighted method: Case level positivity is weighted by the tumor cellularity of each contributing core.

  • cAny positive method: Case is positive if any core is positive.