Abstract P1-07-10: Comparison of three commercial ER/PR assays on a single clinical outcome series.
Bibliographic record
Abstract
Abstract Introduction: Adjuvant tamoxifen is the standard therapy for early stage hormone receptor+ breast cancers (BC). Estrogen receptor (ER) and progesterone receptor (PR) analysis is routinely performed by immunohistochemical (IHC) testing on BC specimens to assist in determining hormone receptor status and patient treatment. ER/PR IHC methodologies and guidelines have recently come under review. Many clinical laboratories have opted to use platform specific, ready-to-use (RTU) ER/PR assays provided by three companies: Dako, Leica and Ventana. While each of these companies are using antibodies that were validated on BC clinical outcome series, these platform specific RTU assays have never been directly compared using the same clinical outcome series. We present a systematic comparison of the three platform specific RTU ER/PR assays, using a retrospective BC cohort, to evaluate the concordance and reproducibility of the RTU ER/PR assays, and to assess the ability of the RTU ER assays to predict tamoxifen response. Methods: The Calgary Tamoxifen Cohort is a retrospective database containing demographic, clinical and pathological data for 820 BC patients diagnosed between 1985–2000 at the Tom Baker Cancer Centre (Calgary, Canada). Formalin-fixed paraffin-embedded tissue blocks were available for 511 patients, and replicate 0.6mm cores were taken and built into tissue microarrays (TMAs). The TMAs were stained using the platform specific assays on the DakoLink Plus, Ventana BenchMark Ultra, or Bond-III Leica autostainers. Slides were manually scored by the Allred method. Results: Ventana and Dako had the best concordance for ER (κ=0.90). Substantial agreement was seen for ER staining between Leica and Ventana (κ=0.79), and Dako and Leica (κ=0.66). Agreement was more consistent between the three platforms for PR staining (κ=0.78–0.82). Inter-observer reproducibility was evaluated for all three platforms between three observers: Dako ER (κ=0.80–0.92) and PR (κ=0.69–0.90); Leica ER (κ=0.67–0.83) and PR (κ=0.70–0.89); Ventana ER (κ=0.88–1.00) and PR (κ=0.78–0.94). TMAs were rescored and intra-observer agreement was calculated: Dako ER (κ=1.00) and PR (κ=0.98); Leica ER (κ=0.91) and PR (κ=0.94); Ventana ER (κ=1.00) and PR (κ=0.94). ER Allred scores were dichotomized using current standards and univariate analysis for 5-year disease free survival was performed. All platforms achieved significance with the logrank test and hazard ratio (HR) estimates (p < 0.0001). Cox models were also run to adjust for lymph node status, grade, size and HER2 status. ER status determined by Dako [HR=0.37(0.19–0.74), p = 0.005] and Ventana [HR=0.40(0.18–0.87), p = 0.021] maintained significance, while ER status determined by Leica [HR=0.61(0.31–1.20), p = 0.154] did not. Conclusions: Concordance between RTU assays demonstrated more variation for ER than PR. All assays showed substantial agreement for inter- and intra- observer reproducibility. Although ER RTU assays from all vendors performed as expected in univariate analysis, multivariate models demonstrated differences. Dako and Ventana appeared equivalent in the multivariate analysis, each providing prognostic information, whereas Leica did not achieve independence in this analysis. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P1-07-10.
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 | 0.022 | 0.036 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.002 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".