Abstract P4-05-01: Results from the first UK NEQAS ICC & ISH external quality assessment for Ki-67 demonstration in breast cancer

Abstract

Abstract AIMS: We previously described use of digital image analysis (DIA) to improve objectivity and reproducibility in external quality assessment (EQA) for immunocytochemical (ICC) staining in pathology [1]. Here we extend the work and report on: (1) use in EQA of the proliferation marker, Ki-67 in breast cancer, (2) use of a commercial prototype cell-line microarray (CLMA). MATERIALS AND METHODS: 34 laboratories stained slides provided by UK NEQAS using their routine ICC methods and returned them for central review and analysis. Slides were provided in duplicate; 29 laboratories returned both and 5 one slide (N = 63). Centrally performed DIA used the Cognition Master Professional Suite (CogM, VMscope GmbH) to quantify Ki67 labelling index (LI) [2]. Qualitative assessment aimed to separate submissions that were not useful from those which could be reported. This followed UK NEQAS standard methods. Materials assessed comprised two different breast cancer samples and a Ki-67 CLMA (Karpas Prototype, Array Science). RESULTS: Descriptive statistics on Ki-67 scores for each sample are presented in Table 1. Analysis of individual sample types for difference in Ki-67 expression associated with primary antibody showed the following: mean Ki-67 LI on breast cancer A had a range of 13.6% (K2, Leica Biosystems) to 17.6% (30-9, Ventana Medical Systems), but no significant difference was found for any primary antibody clone. For breast cancer sample B, the range was 2.0% (MIB-1, Dako Agilent) to 3.2% (30-9), and the difference was significant at P<0.05. In contrast, the CLMA showed no significant difference in expression for either MIB-1 or 30-9 (K2 could not be reported).The Ki-67-positive Karpas cell line in the CLMA had a small nuclear size compared to that of ‘standard’ breast cancer, which led to undercalling by CogM and potentially other DIA software; it also demonstrated a restricted dynamic range of high staining intensity, which reduced its ability to demonstrate differential staining in response to methodological variations.When assessed qualitatively 11.1% of submissions were not reportable (fail), 12.7% were sub-optimal but were scoreable (borderline), 23.8% had minor defects (adequate) and 52.4% were optimally stained (good). For both breast cancer samples, but not for the CLMA these qualitative categories were associated with increasing Ki-67 LI scores. For breast cancer A the mean Ki-67 LI score for the non-reportable group was significantly different from all other categories, which were not significantly different from each other; for breast cancer B, significant differences were seen between the non-reportable group and the adequate and good groups only (all at P<0.05). CONCLUSIONS: Centrally performed DIA of Ki-67 LI on breast cancer tissue samples produces additional information that allows the identification of statistically significant differences between primary antibodies (MIB-1 and 30-9). It also validates qualitative assessment by demonstrating significant differences between qualitative categories and is likely to be useful to identify lab-specific differences in staining performance in future rounds of EQA.The prototype CLMA produced using the Karpas cell-line did not demonstrate statistically significant differences between primary antibody clones or qualitative categories and is not a useful tool for use in EQA.REFERENCES: 1. Warrick N, et al. UK IBCS 2020. Breast Cancer Res Treat 180, 527-596 (2020)2. http://vmscope.com/index BC_ABC_BCLMA_0%CLMA_5%CLMA_10%CLMA_20%CLMA_30%Number of values124125164165110165165Minimum0.30.00.00.01.93.30.025% Percentile11.31.00.01.74.07.613.5Median15.22.20.02.35.09.416.375% Percentile20.03.50.23.26.111.519.8Maximum32.212.112.09.111.721.335.995% CI of medianActual confidence level96.2%95.1%96.5%95.7%95.5%95.7%95.7%Lower confidence limit14.01.80.02.04.99.115.5Upper confidence limit16.62.60.02.65.59.817.1Mean15.72.50.42.65.510.017.1Std. Deviation6.52.11.71.42.23.35.7Std. Error of Mean0.60.20.10.10.20.30.4Lower 95% CI of mean14.52.20.22.45.09.516.2Upper 95% CI of mean16.82.90.72.85.910.518.0 Citation Format: Andrew Dodson, Dawn Wilkinson, Mitch Dowsett, Suzanne Parry. Results from the first UK NEQAS ICC & ISH external quality assessment for Ki-67 demonstration in breast cancer [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P4-05-01.