Abstract P3-05-01: Development for clinical utility of a validated predictive test of clinical response to aromatase inhibitors
Bibliographic record
Abstract
Abstract Background: Aromatase inhibitors (AIs) have an established role in the treatment of estrogen receptor alpha positive post-menopausal breast cancer. Response rates are only 50-70% even in patients with ER-rich cancers in the neoadjuvant setting and are lower in advanced disease. Recently we developed and validated a microarray-derived 4-gene test to predict response to AIs in the neoadjuvant setting. Whole-genome expression analysis is impractical for clinical utility. There is a need to translate and validate any test utilising clinically accessible and reproducible material and technologies such as polymerase chain reaction (PCR) and immunohistochemistry (IHC). Methods: The original microarray experiment used pre- and on-treatment (at 14 days and 3-months) biopsies from 89 post-menopausal women with ER-rich breast cancer receiving 3 months of neoadjuvant letrozole. Dynamic response was based on periodic 3D ultrasound measurements performed during treatment. The derived 4-gene model was independently validated in a cohort of 44 post-menopausal women with ER-rich breast cancer treated with neoadjuvant anastrozole [table 1]. RNA was extracted from the original biopsies for RT-qPCR analysis using validated primers for the 4 genes with SYBR-green technology normalised to the geometric mean of 3 housekeeping genes. Matched formalin-fixed paraffin embedded (FFPE) tissue sections were used for IHC with optimised antibodies against 3 of the 4 proteins (where validated antibodies were available) using Envision technology. Results: PCR: Microarray and PCR expression levels for each of the four genes were well correlated (Pearson r=0.87-0.65, p<0.0001). Application of the 4-gene model, using PCR expression levels, to a cohort of patients from the initial training set (n=26) resulted in prediction of response with 96% accuracy [table 1]. IHC: stained FFPE tissue sections for proteins corresponding to the 3 most informative genes in the model were independently scored using a histoscore approach (0, 1+, 2+, 3+) in a sample of patients where tissue was available (n=28). Increasing histoscores were associated with increasing gene expression for all 3 proteins (r=0.72, 0.73, 0.60). Each sample was assessed and histoscore determined, then a positive and negative cut-off histoscore was determined for each protein to maximise response prediction. This was then applied to an IHC decision tree, based on the original 4-gene microarray model, and was able to categorise patients as responsive or non-responsive with 89% accuracy. Conclusion: Table 1NAccuracySensitivitySpecificityPPVNPVMicroarray (Training)890.960.890.980.980.96Microarray (Validation)440.910.800.970.920.90PCR260.960.951.001.000.86IHC280.890.861.001.000.89Sensitivity and specificity of model in training (microarray, PCR, IHC) and validation datasets (microarray). PPV/NPV = postivie/negative predictive value. •A 4 gene model has been developed and validated to predict response to neoadjuvant aromatase inhibitors. •This model has been shown to work with a high degree of accuracy using both PCR and IHC technologies. Further independent validation is currently underway. •This new test has the potential to predict accurately the benefit of endocrine therapy and has huge potential clinical value. Citation Format: Arran K Turnbull, Laura Arthur, Victoria Webber, Jeremy Thomas, Charlotte Heerlyn, Phoebe Thornton, Anita Dunbier, Mitch Dowsett, Lorna Renshaw, Andrew H Sims, J Michael Dixon. Development for clinical utility of a validated predictive test of clinical response to aromatase inhibitors [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P3-05-01.
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 | 0.003 | 0.004 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.001 |
| 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".