AI-enabled analysis of H&E-stained prostate cancer tissue images: Assessing risk for metastasis prior to apalutamide (APA) treatment of patients with non-metastatic castration-resistant prostate cancer (nmCRPC).
Bibliographic record
Abstract
5027 Background: Most prostate cancer patients (pts) treated with androgen deprivation therapy (ADT) for progressive disease will experience progression to CRPC. APA has been approved for the treatment of nmCRPC and metastatic castration sensitive prostate cancer (mCSPC). AI-enabled tools developed to predict oncological outcomes from digitized whole-slide images (WSIs) of H&E-stained tissues are a practical alternative to costly genomic testing tools that require sufficient tumor material. PathomIQ has developed the AI-enabled prognostic test PRAD-DX that predicts risk of metastasis from WSIs of H&E-stained core biopsies or radical prostatectomy specimens. It has been validated as a research tool on more than 2000 tissue samples across multiple institutions. The objective of this study is to evaluate PRAD-DX on predicting risk of metastatic progression using archived primary tumor samples from a randomized, double-blind, phase 3 trial in nmCRPC pts [1]. Methods: WSIs were collected from 471 pts (APA+ADT (n=315); placebo+ADT (n=156)), de-identified and anonymized prior to the AI analysis. Patient outcomes were blinded to the PathomIQ team. 35 pts were excluded due to lack of tumor or poor image quality; PRAD-DX scores were generated for 436 (93%) pts. The PRAD-DX test generated a risk score for each patient between 0 and 1, with higher values representing increased risk of metastasis. A pre-determined cut-off of 0.55 was previously developed to provide the best predictive accuracy and stratification with respect to time-to-metastasis on multiple clinical cohorts and was applied to this dataset. Kaplan-Meier analysis was performed on Metastasis-free-survival (MFS). Results: All pts receiving APA+ADT had improved outcomes compared with pts receiving ADT alone, independent of PRAD-DX risk score category. 53% of pts had high PRAD-DX scores and significantly benefited from treatment with APA+ADT compared to placebo+ADT with regard to MFS (hazard ratio, 0.19; 95% CI, 0.1 – 0.37; P<0.005). 47% were assigned low PRAD-DX risk; also in this cohort, treatment with APA+ADT resulted in a significantly improved MFS (hazard ratio, 0.39; 95% CI, 0.17 – 0.86; P=0.02). Conclusions: These results indicate that PRAD-DX score derived from AI-powered image analysis could be used as a biomarker to identify pts at highest risk of metastatic progression in the nmCRPC setting. Although, there are limitations related to sample size, this may have significant impact in identifying pts for informed clinical trial patient selection based on their individual risk profile for development of novel therapies. 1) Smith MR, Saad F, Chowdhury S, et al. N Engl J Med. 2018;378(15):1408-1418. doi:10.1056/NEJMoa1715546.
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 | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 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".