AI Guest Author: Review of H&E AI Breast Cancer Prognosis, versus Standard Methods (21-gene)

Update:  See the FDA's 28-page review of the Artera AI breast cancer H&E prognostic test, here.

####

I have seen a scattering of articles that study whether AI interpretations of H&E slides can mirror the prognostic value of molecular tests.   

The linked PDF white paper is entirely machine generated.

Here were the steps:

1. I asked it to seek literature on the topic from PubMed.
2. It got about 8.
3. I asked it to download all the (public access) PDFs in a zip file for me.  (See footnote).
4. I then gave the PDFs back to it, and asked it to write a review article.
1. I gave it about 8 or 10 topics or talking points to include.
5. It produced a 12-page review.
6. I asked it to produce an interesting report cover.

##

Find the white paper in the cloud here.  

(See also footnote below).

##

Here's the executive summary:

From Oncotype DX to “Virtual Recurrence Scores”:

AI Histopathology for  Breast Cancer Risk Stratification

Abstract

Oncotype DX is the best-established multigene assay for guiding adjuvant chemotherapy in hormone receptor-positive, HER2-negative early breast cancer. Its evidence base spans retrospective validation, TAILORx, RxPONDER, and major guidelines, but cost, turnaround time, access, and tissue consumption remain limitations. A rapidly expanding literature now uses routine H&E whole-slide images, often combined with clinicopathologic data, to approximate Oncotype DX or related transcriptomic scores and, increasingly, to predict recurrence and chemotherapy benefit directly.

Recent studies show multicenter generalization, outcome discrimination, and analytical reproducibility. Yet score imitation, prognosis, treatment-benefit prediction, and clinical utility remain distinct claims and they will require separate validation before AI could replace genomic testing.

###

Footnote.

Regarding Chat GPT fetching and downloading PDFs: It did do that that, though after a puzzling half-hour of delay.   For other white papers, Chat GPT just failed to download PDFs (colliding with PubMed Central firewalls, bot detection, etc).   One would have to go further with downloaded AI and true "agentic" modes to reliably get it to find and download PDFs.  On the other hand, freeware reference manager software like Zotero can fetch any number of PMC PDFs nearly instantly.

Regarding US Companies.  A colleague suggested that a future second version could give more direct attention to preciseDx and ArteraAI.  Here are two AI bullets:

PreciseDx is developing PreciseBreast, which combines H&E-derived features with clinicopathologic data to predict six-year recurrence risk.  It has reported performance comparable or superior to Oncotype DX in retrospective cohorts.

ArteraAI has advanced further commercially: its FDA-cleared ArteraAI Breast test uses digitized H&E slides plus clinical variables to stratify distant-metastasis risk, with separate trial evidence suggesting prediction of chemotherapy benefit in selected patients.  (The Oncotype test is not FDA-cleared.)

In the first-version white paper, a third company, the German company Stratif-AI was represented by Boehm et al. 2025 in Nature Communications.