Abstract GS4-03: Validation of Profile for the Omission of Local Adjuvant Radiotherapy (POLAR) in a meta-analysis of three randomized controlled trials of breast conserving surgery +/- radiotherapy
Bibliographic record
Abstract
Abstract Background: There are currently no commercially available tests to identify early stage breast cancer patients treated with breast conserving surgery (BCS) and systemic therapy at low risk of locoregional recurrence (LRR) for whom postoperative radiotherapy (RT) may be safely omitted. Profile for the Omission of Local Adjuvant Radiotherapy (POLAR) is a 16-gene molecular signature developed to identify invasive breast cancer patients who may be candidates for RT omission after BCS. In this work, we seek to validate POLAR in a meta-analysis of three RCTs of BCS +/- RT: SweBCG91RT, Scottish Conservation Trial (SCT) and Princess Margaret Hospital (PMH). Methods: A patient-level meta-analysis was performed in 623 node-negative breast cancer patients with ER+/HER2-negative tumors who were enrolled in the three RCTs and for whom primary tumor material was available for analysis. Contributions from each cohort were as follows: SweBCG91RT N=354 (57%), SCT N=137 (22%), and PMH N=132 (21%). Numbers of LRR events in each cohort were as follows: SweBCG91RT N=72 (20%), SCT N=28 (20%), and PMH N=16 (12%). There was a mix of systemic therapy used (no systemic therapy for SweBCG91RT, chemotherapy or adjuvant endocrine therapy, but not both, in SCT, and tamoxifen but no chemotherapy for PMH). Median follow-up time for the patients who did not have LRR was 13.3 years for SweBCG91RT, 21.1 years for SCT, and 8.6 years for PMH. A multivariable Cox proportional hazards model on time to LRR, including the continuous standardized POLAR score, RT, and interaction, stratified by cohort, was used to test the interaction between the continuous POLAR score and RT. Additional Cox models tested the association between treatment arms separately for patients with a low and high POLAR score using a pre-specified cut point. Cumulative incidences were computed, with distant metastasis and death without recurrence considered as competing events. Results: The test for interaction between RT treatment and POLAR was statistically significant (p = 0.022). Patients with a high POLAR score (N=429 [69%]) had a large benefit from RT (10-year cumulative incidence of LRR: 20% [15%-26%] for those not treated with RT vs 7% [4%-11%] for those treated with RT; hazard ratio for RT vs no RT: 0.37 [0.23-0.60], p < 0.001), whereas there was no evidence of benefit from RT for patients with a low POLAR score (N=194 [31%], 10-year cumulative incidence of LRR: 5% [2%-11%] for those not treated with RT vs 7% [3%-14%] for those treated with RT; hazard ratio for RT vs no RT: 0.92 [0.42-2.02], p = 0.832). Conclusions: To our knowledge, POLAR is the first genomic classifier that is not only prognostic for LRR but also predictive, showing a significant interaction between RT and the classifier. Patients with a high POLAR score should be recommended radiotherapy while patients with a low score may be candidates for omission of radiotherapy after breast conserving surgery. Citation Format: Per Karlsson, Anthony Fyles, S. Laura Chang, Bradley Arrick, Frederick Baehner, Per Malmström, Mårten Fernö, Erik Holmberg, Martin Sjöström, Fei-Fei Liu, David A. Cameron, Linda J. Williams, John MS Bartlett, Joanna Dunlop, Jacqueline Caldwell, Joseph F. Loane, Elizabeth Mallon, Tammy Piper, Wilma J. Jack, Ian Kunkler, Felix Y. Feng, Corey W. Speers, Lori Pierce, John Bennett, Karen J. Taylor. Validation of Profile for the Omission of Local Adjuvant Radiotherapy (POLAR) in a meta-analysis of three randomized controlled trials of breast conserving surgery +/- radiotherapy [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 GS4-03.
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 | 0.038 | 0.057 |
| Meta-epidemiology (narrow) | 0.003 | 0.001 |
| Meta-epidemiology (broad) | 0.011 | 0.049 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.003 | 0.003 |
| Insufficient payload (model declined to judge) | 0.004 | 0.000 |
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".