Modified Methods For Ki67 Label Index Assessment In Breast Cancer And Inter Observer Reproducibility

Objective The proliferation marker Ki67 plays an important role in subtypes of breast cancer and is a key index for breast cancer patients in choosing cytotoxic therapy and estimating prognosis.However a formal standard method for Ki67 label index(LI)assessment based on scientific observations has not been described,and the reproducibility of Ki67 LI assessment is poor,which would result in breast cancer patients being overtreated or undertreated.It is imminent to establish a Ki67 LI assessment method which is scientific and well operable.Here,we modified methods for Ki67 LI assessment and investigate inter observer reproducibility.Methods 100 invasive breast cancer tissues were set immunostained for Ki67 in one laboratory,and scanned as digital slides,41 pathologists in 16 hospitals laboratories assessed Ki67 LI using the method as follows.1.Selected of hot spots of Ki67 immunostaining.2.Demarcated three square frames in hot spot.3.Assessed Ki67 LI in the frames with demarcating-square visual assessment in hot spot(DVAHS),demarcating-square manual counting in hot spot(DMCHS),demarcating-square automatic counting in hot spot(DACHS).Measures of dispersion of Ki67 LI were described by M ± Q.The intraclass correlation coefficient(ICC)and Bland-Altman plot were used to evaluate interobserver reproducibility of Ki67 LI assessment using DMCHS and DVAHS respectively.The difference in Ki67 LI assessment with DMCHS and DACHS was analyzed with Wilcoxon signed-rank test.Results 1.In our study,among 100 Ki67 immunohistochemical slices of breast cancer,the percent of slices that had areas of hot spot was 83%.There were 32 cases that hot spot being distributed centrally,and 51 cases hot spot were scattered distributed.The hot spots in 29 cases were distributed in the invasive edge of the tumor,and hot spot in 54 cases were distributed in the internal tumor tissue.The diameter of hot spot were from 50?m to 11740?m 13 cases were<100?m,17 cases were 100-200?m and 53 cases were>200?m.2.Each 100x1OO?m frames could covered 50 tumor cells,200x200?m frames could covered 200 tumor cells and 300x300?m could covered 50 tumor cells.3.Measures of dispersion of Ki67 LI assessed by DMCHS was less than that by DVAHS.4.The ICC(95%CI)of 100 Ki67 label index(Ki67 LI)assessed by 41 pathologists using DMCHS was 0.942(0.926,0.957),and the ICC of DVAHS was 0.802(0.756,0.846).The reproducibility of DMCHS and DVAHS are both acceptable and DMCHS demonstrated a better reproducibility than DVAHS.5.The Ki67 LI assessment by experienced pathologists(n=35)showed that the ICC(95%CI)of DMCHS was 0.953(0.940,0.965);the ICC(95%CI)of DVAHS was 0.805(0.760,0.849).The Ki67 LI assessment by inexperienced pathologists(n=6)showed that the ICC(95%CI)of DMCHS was 0.903(0.873,0.928);the ICC(95%CI)of DVAHS was 0.786(0.730,0.837).Experienced pathologists showed a better reproducibility than inexperienced pathologists.6.DMCHS assessment revealed that the ICC(95%CI)in homogeneous group was 0.965(0.939,0.985);the ICC(95%CI)in dense hot spot group was 0.930(0.894,0.959);the ICC(95%CI)in sparse hot spot group was 0.833(0.776,0.886).DVAHS assessment revealed that the ICC(95%CI)in homogeneous group was 0.875(0.792,0.942);the ICC(95%CI)in dense hot spot group was 0.704(0.601,0.810);the ICC(95%CI)in sparse hot spot group was 0.597(0.505,0.698).The DMCHS and DVAHS assessment both showed that reproducibility was relatively low in sparse hot spot group,compared with homogenous group and dense hot spot group.7.The difference in reproducibility between state hospitals and provincial hospitals is small.8.The Bland Altman plot investigated that the reproducibility of DMCHS was better than DVAHS.9.The Wilcoxon signed-rank test confirmed that Ki67 LI assessment with DMCHS was obviously higher than that with DACHS.Conclusion The method of DMCHS has better inter-observer reproducibility than DVAHS and is a better choice for Ki67 LI assessment.