An Interobserver Reproducibility Analysis of Ki67 Visual Assessment in Breast Cancer

Background Ki67 labeling index (LI) is used as a predictive marker and is associated with prognosis in breast cancer. However, standardised methodologies for measurement are lacking which has limited its application in clinical practice. In this study, we evaluated the interobserver concordance of visual assessment of Ki67 LI in breast cancer. Methods Ki67-immunostained slides of 160 cases of primary invasive breast cancer were visual assessed by five breast pathologists with two different methods to choose the scoring fields: (1) hot-spot score, (2) average score. Proportions of positive invasive tumor cells at 10 % intervals were scored. The intra-class correlation coefficient (ICC) was used to assess the interobserver reproducibility. Results (1) A perfect concordance of Ki67 LI was demonstrated according to both score methods (P<0.0001). Average score method (ICC, 0.904) demonstrated a better correlation than hotspot score method (ICC, 0.894). (2) By respective means according to two score methods, all cases were classified into three groups (<= 10%, 11%-30% and >30% Ki-67 LI). The concordance was relatively low in intermediate Ki67 LI group compared with low and high Ki67 LI groups. (3) All cases were classified into three groups by paired-difference (d) between means of hot-spot score and average score (d<5, 5 <= d<10, d >= 10). The consistency was observed to decrease with increasing paired-difference according to both methods. Conclusions Visual assessment of Ki67 LI at 10% intervals is a candidate for a standard method in breast cancer clinical practice. Average score and hot-spot score of visual assessment both demonstrated a perfect concordance, and an overall average assessment across the whole section including hot spots may be a better method. Interobserver concordance of intermediate Ki67 LI in which most cutoffs are located for making clinical decisions was relatively low.