P5-11-11: Automated Quantification Methods Improve the Accuracy of PR as an Independent Prognostic Factor in Tamoxifen Treated Breast Cancer Patients.
Bibliographic record
Abstract
Abstract Background Adjuvant endocrine therapy reduces the risk of recurrence and death in hormone receptor positive breast cancer patients. However, 40–50% of estrogen receptor (ER) positive tumors are resistant to endocrine therapy. We have previously shown that quantitative measurement of ER protein expression has limited value as a prognostic marker in tamoxifen treated patients. The presence of progesterone receptor (PR) expression has shown promise as prognostic and/or predictive marker for endocrine therapy, including adjuvant tamoxifen, but reports are contradictory. Automation of scoring methods will improve the accuracy of PR scoring and its value as a prognostic factor. Methods: This retrospective study evaluated breast cancer patients treated with adjuvant tamoxifen (n=458) from 1985–2000 at the Tom Baker Cancer Centre, Calgary, Canada. Tissue microarrays were assembled from formalin fixed paraffin embedded tumor tissue. Clinico-pathologic data was obtained from chart review. Five-year disease-free survival (DFS) was the primary outcome. DAB-based PR staining was used to generate Allred PR scores, as well as to generate scores for PR percent area expression and PR integrated optical density (PR IOD) using the DAKO ACIS III scanner and image analysis software. PR tumor nuclear pixel intensity density scores were obtained using fluorescence-based PR staining, scanned and quantified with a HistoRx PM2000 scanner andAQUA image analysis software. Continuous variables were dichotomized using Xtile. Results: In our tamoxifen cohort, 5-year DFS was associated with tumor grade [HR 4.9(3.5−6.9), p<0.001], tumor size [HR 2.7(1.9−3.8), p<0.001], lymph node status [HR 5.7(4.0−8.0), p<0.001] and ER status [HR 2.4(1.3−4.7), p=0.008]. Low PR was associated with significantly worse DFS regardless of the method used to quantify its expression. Methods of analysis requiring less subjective input had stronger associations between PR expression and outcome. The subjective identification of tumor and subjective biomarker quantification used in Allred scoring [HR 2.2(1.4−3.3), p<0.001] was less effective than ACIS scoring [PR percent area: HR 2.8(1.8−4.3), p<0.001; PR IOD: HR 2.9(1.9−4.6), p<0.001], which only requires subjective identification of the tumor. All methods were inferior to fluorescence-based AQUA scoring [HR 4.2(2.7−6.5), p<0.001], which automatically detects the tumor nuclear area using pan-cytokeratin and DAPI staining and measures the PR expression within this compartment. Multivariate analysis, which included age, tumor size, tumor grade, lymph node status, and ER status, confirmed that AQUA [HR 3.3(2.0−5.6), p<0.001] was superior to ACIS PR percent area [HR 2.3(1.4−3.9), p=0.002] and ACIS PR IOD [HR 2.7(1.6−4.8), p<0.001], all of which were superior to Allred scoring [HR 1.2(0.7−2.1), p=0.511]. Conclusions: We conclude that: 1) PR is an independent prognostic marker in our cohort, 2) using less subjective, automated quantitative scoring methods improves the value of PR as a prognostic biomarker, 3) AQUA appears superior to other automated methods and that 4) the incorporation of digital image analysis into practice would improve the prognostic value of PR expression in the clinical setting. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P5-11-11.
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 | 0.002 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.003 | 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".