Abstract P6-04-02: Ki67 Assessment Protocol: Companion Diagnostic Biomarker for LUMINA Prospective Cohort Study
Bibliographic record
Abstract
Abstract Introduction: Luminal A breast cancer is associated with low proliferation, indolent disease biology and limited benefit from chemotherapy. The LUMINA prospective study recently demonstrated a very low 5 year local recurrence rate (2.3%) in women ≥55 years with grade I-II, T1N0 luminal A breast cancer (defined as ER ≥ 1%, PR>20%, HER2 negative and Ki67 index ≤ 13.25%) treated with breast conservation surgery and endocrine therapy without radiation, supporting the safe omission of radiation in this molecularly defined low risk group. Here, we report the protocol for multicentre Ki67 scoring, the embedded integral companion diagnostic employed in LUMINA. Methodology: Ki67 immunohistochemistry was performed on full-face sections at one of the 3 labs and scored by pathologists using an adaptation of the International Ki67 Working Group (IKWG) method. Prior to the start of the study, quality assurance and quality control programs were set up to standardize staining and scoring protocols. All pathologists completed the IKWG training and calibration exercise using a tissue microarray-based series of 18 breast cancers. Inter-laboratory variability was assessed annually during the study period on a set of 9 breast cancer cases with a range of Ki67 scores that purposely over-represented the 13.25% threshold. Stained slides were scanned and images annotated to demarcate invasive carcinoma. Next, 5 random, non-overlapping, 1 mm virtual cores were generated via software and 100 nuclei assessed per core using a keyboard-based counting aid. Ki67 index was derived as the percentage of all counted tumor nuclei that are positively stained. For cases with high Ki67 heterogeneity, additional virtual cores were generated and scored and a 95% confidence interval (CI) of Ki67 index was estimated. The goal was to confidently assign a case as luminal A (≤13.25%) or B (> 13.5%). If the 95% CI crossed 13.25% a recount was performed by an additional pathologist. Results: Quality Assurance Programs: Mean Ki67 index across all cases, labs and years was 13% with high concordance across specimens and score ranges. Observed intra-class correlation coefficients (ICC) were ≥ 0.9, showing near perfect agreement in quantitative Ki67 evaluation. About the 13.25% cutpoint, the observed Kappa statistics were ≥ 0.7 indicating excellent agreement for assignment of luminal A vs. B status. A sub-study was conducted to compare the method of randomly selected virtual fields with the IKWG ‘global weighted score’ method for visual assessment of full-face sections. For this purpose, the 9 quality control cases were reassessed by the same pathologist using the updated IKWG method. Results showed an ICC of 0.96 (0.95% CI: 0.91-0.98) indicating that the Ki67 score generated by the methodology employed in LUMINA trial is highly concordant with the IKWG scoring methodology validated for use on full face sections. Ki67 index summary statistics across LUMINA: Of the 724 eligible cases, 69% (n=500) were assigned as luminal A (median Ki67=7.5%; IQR 5.2-9.8%) and 31% (n=224) as luminal B (median Ki67=19%; IQR 17-23%). Median pathologist scoring time was 4 minutes/case; 45% of cases required scoring of > 5 virtual cores. Per protocol, 39% cases where the initial CI crossed 13.25% were rescored by additional pathologist for final luminal A consensus assignment. Conclusions: Ki67 is a practical biomarker for identifying molecularly defined low-risk luminal A cancers. Our structured quality assurance approach for the trial led to excellent reproducibility and concordance among decentralized labs, supporting applicability of a distributed, inexpensive methodology beyond clinical trial settings and in resource restricted environments. Citation Format: Torsten Nielson, Samuel Leung, Nazia Riaz, Zuzana Kos, Anita Bane, Timothy J. Whelan. Ki67 Assessment Protocol: Companion Diagnostic Biomarker for LUMINA Prospective Cohort Study [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P6-04-02.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.018 | 0.026 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.003 | 0.002 |
| Insufficient payload (model declined to judge) | 0.024 | 0.019 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".