For a large group of breast cancer patients, one of the hardest questions in oncology isn’t whether to treat — it’s how much. Add chemotherapy to hormone therapy, and you add real toxicity: neuropathy, infection risk, sometimes infertility. Skip it, and for the wrong patient, the cancer can come back. The tools doctors use to sort patients into that decision are decades old, and they have real practical limits.
In May 2026, an AI breast cancer test called ArteraAI Breast became the first FDA-cleared digital pathology tool built to help with exactly that call. It reads a slide that’s already being made during standard care and returns a personalized risk estimate in a day or two — a genuine break from the two- to three-week wait that’s defined this corner of oncology for twenty years. It’s also, by the company’s own data, a test whose most attention-grabbing claim rests on thinner evidence than its risk-stratification claim. Both things are true, and worth understanding together.
Table of Contents
The Problem: A Common, Genuinely Gray-Zone Decision
About 1 in 8 women in the U.S. (roughly 13%) will develop invasive breast cancer in their lifetime, and the World Health Organization estimated 2.3 million new diagnoses and about 670,000 deaths worldwide in 2022 (Medical News Today). For the specific subset this test addresses — early-stage, hormone receptor-positive (HR+), HER2-negative invasive breast cancer — the treatment decision is often genuinely ambiguous. Many patients, particularly postmenopausal women with node-negative tumors, fall into what one oncologist called a “gray area,” where chemotherapy’s benefit isn’t obvious either way.
The existing tools for narrowing that decision are genomic assays, and each has friction built in:
- Oncotype DX, the field’s incumbent since the early 2000s, requires RNA extraction from tissue and can take two to three weeks to process. It’s often covered by insurance after surgery, but not always beforehand.
- MammaPrint requires a fresh tissue sample rather than a standard preserved slide. One feasibility study found that 28% of samples lacked sufficient tissue for analysis — a figure that rose to 88% in tumors under 1 centimeter (PMC).
Both tests are validated by large prospective trials (TAILORx for Oncotype DX, MINDACT for MammaPrint) and remain the clinical standard. ArteraAI Breast competes on a different axis so far: speed and workflow, not yet proven comparative accuracy.
Meet ArteraAI Breast: From Prostate Cancer to a New FDA First
The company behind the test, Artera, was founded in 2021 by CEO Andre Esteva and radiation oncologist Felix Feng, MD, with a stated mission to apply computer vision and digital pathology to cancer treatment decisions. Artera launched commercially in 2023 with $90 million in funding from investors including Johnson & Johnson’s venture arm and Salesforce founder Marc Benioff (MedTech Dive).
Its first product, ArteraAI Prostate, launched as a lab-developed test before the FDA granted it Breakthrough Device Designation in July 2025 and then De Novo authorization in August 2025 — creating a new FDA product code for AI-powered digital pathology risk-stratification tools in the process. ArteraAI Breast followed on May 6, 2026, when the FDA cleared it via the 510(k) pathway (decision summary K254115) for use in early-stage, HR+/HER2-negative invasive breast cancer — described by Artera and independent trade outlets as the first and only FDA-cleared digital-pathology-based risk-stratification tool for breast cancer (Artera; ITN).
Milestone timeline:
| Date | Milestone |
|---|---|
| 2021 | Artera founded by Andre Esteva and Felix Feng |
| 2023 | Commercial launch with $90M in funding |
| Jan. 2024 | CMS sets Medicare payment rate for ArteraAI Prostate |
| July 2025 | FDA Breakthrough Device Designation for ArteraAI Prostate |
| Aug. 2025 | FDA De Novo authorization for ArteraAI Prostate |
| Nov. 2025 | NSABP B-20 breast cancer data presented at SABCS |
| May 6, 2026 | FDA 510(k) clearance for ArteraAI Breast |
How the Test Works
ArteraAI Breast uses the surgical resection slide a pathology lab already prepares during standard care — no additional biopsy or tissue sample required. That slide is digitized, and the resulting image is combined with clinical variables the ordering physician provides (such as age and tumor characteristics). Artera’s software analyzes both inputs and returns a personalized risk score, sorting the patient into an “MMAI Low” or “MMAI High” risk group for distant metastasis, typically within one to two days of the lab receiving the specimen (Artera).
The AI Technology Behind the Score
Artera calls its underlying technology multimodal artificial intelligence (MMAI). Co-founders Esteva and Feng have described it as two linked components: one model that learns from the digital pathology image, and a second that learns from structured clinical data, fused together to generate both a prognosis and a treatment-benefit prediction (UroToday).
Beyond that general description, the specific deep learning architecture is not publicly disclosed. That opacity isn’t just an academic concern — it was raised directly by an independent oncologist evaluating the tool. Donna McNamara, MD, of Hackensack Meridian Health’s John Theurer Cancer Center, told Medical News Today she would need “transparency into the AI model itself, moving beyond the ‘black box’ to understand the key features driving its risk assessments,” adding that she “cannot base a patient’s treatment plan on a tool without thoroughly understanding its accuracy, reliability, and practical utility” (MNT).
The Evidence: What the Data Actually Shows
Artera says the MMAI model was trained on data from more than 8,500 breast cancer patients drawn from clinical trials, according to Calvin Chao, MD, the company’s vice president of medical science. Validation drew on multiple Phase 3 randomized trial cohorts across Germany, Austria, and North America.
| Finding | Result | Evidence type |
|---|---|---|
| Risk stratification, validation cohort | 68% of ~12,000 patients classified low-risk, ~95% estimated 10-year distant-metastasis-free survival in that group | Company-reported, part of FDA clearance record |
| NSABP B-20 trial — overall cohort | MMAI not predictive of chemotherapy benefit in the full trial population | Peer-conference presentation (SABCS 2025) |
| NSABP B-20 — age 50+, MMAI intermediate/high risk subgroup | 52% relative reduction in 10-year distant metastasis with chemotherapy added | Exploratory subgroup analysis, SABCS 2025 |
That last line matters. The subgroup finding is a real result from a randomized trial dataset, but it’s an exploratory subgroup finding, not the trial’s primary, pre-specified outcome — a meaningfully weaker form of evidence. It was also presented as a conference abstract at the San Antonio Breast Cancer Symposium, not (as of this writing) published in a peer-reviewed journal specific to the breast indication (Oncology Nexus).
Independent Scrutiny: An Oncologist’s Conditions for Trust
Reaction from independent clinicians has been mixed. Richard Reitherman, MD, PhD, medical director of breast imaging at MemorialCare Breast Center, called the potential to skip “the costs and time delays associated with currently available methodology” a real advantage.
McNamara was more cautious. Beyond the black-box concern, she told Medical News Today she would want “prospective, head-to-head clinical trials that directly compare its performance against current gold standards like Oncotype DX,” with long-term survival follow-up, before changing how she treats patients. No such independent, head-to-head trial has been published as of this research.
Regulatory Status: What FDA Clearance Covers
The FDA’s 510(k) clearance, granted May 6, 2026 under decision summary K254115, authorizes ArteraAI Breast specifically for early-stage, HR+, HER2-negative invasive breast cancer, describing it as “a software-only device that analyzes scanned histopathology whole-slide images from breast resection specimens, along with physician-provided clinical variables” (nchstats.com). Artera also received CE marking under EU IVDR for the Breast Cancer Assay around the same period, alongside its existing Prostate Biopsy Assay marking.
One open question: while Medicare established a specific payment rate for ArteraAI Prostate in January 2024, it is not confirmed whether an equivalent Medicare payment rate has yet been set for ArteraAI Breast.
Real-World Availability
As of this research, ArteraAI Breast is processed through a single CLIA-certified laboratory Artera operates in Jacksonville, Florida. Ordering is restricted to clinicians within their scope of practice. Artera has also opened an Early Access Program offering the test with zero out-of-pocket costs to enrolled patients. Adoption figures specific to the breast test are not yet available, given its recent clearance; by comparison, ArteraAI Prostate had reached roughly 20% of U.S. prostate cancer clinicians by mid-2025, according to a World Economic Forum profile of the company.
Benefits, Backed by Evidence
- Faster turnaround: results in one to two days, versus two to three weeks for Oncotype DX.
- No added tissue procedure: uses the surgical resection slide already produced during standard care, avoiding MammaPrint’s documented tissue-insufficiency failures.
- Meaningful risk stratification: identified 68% of a large validation cohort as low-risk with an estimated 95% 10-year metastasis-free survival — a company-reported figure from the FDA clearance record.
- Potential to reduce overtreatment: independent oncologist Donna McNamara called the potential to spare low-risk patients from chemotherapy toxicity “a major advantage” — while still calling for more validation before adopting it.
Limitations, Risks, and Open Questions
- No independent head-to-head trial against Oncotype DX or MammaPrint has been published.
- The chemotherapy-benefit finding is an exploratory subgroup result, not validated in the full NSABP B-20 trial population, and not yet in a peer-reviewed journal specific to breast cancer.
- Model architecture remains largely undisclosed, a concern one independent oncologist raised explicitly before she said she’d consider using it.
- Single-laboratory bottleneck: all testing currently routes through one Florida lab.
- Training data demographics for the breast-specific model are not publicly disclosed — notable given that Artera has published such figures for its prostate test.
- List price and Medicare coverage status specifically for the breast test remain unconfirmed.
The Competitive Landscape
| Test | Approach | Turnaround | FDA Status | Key Evidence/Caveat |
|---|---|---|---|---|
| ArteraAI Breast | AI digital pathology + clinical data | 1–2 days | 510(k) cleared, May 2026 | Strong risk stratification; chemo-benefit claim is a trial subgroup finding |
| Oncotype DX | 21-gene RNA expression assay | 2–3 weeks | Established standard | Prospectively validated (TAILORx) |
| MammaPrint | 70-gene expression signature | ~1–2 weeks | Established standard | Prospectively validated (MINDACT); documented tissue-failure rate |
| Paige Predict (Tempus AI) | AI digital pathology, multi-biomarker | Not publicly disclosed | Not confirmed FDA-cleared for this indication | Built on a foundation model trained on 2.3M+ whole-slide images |
What’s Next
Digital pathology AI is consolidating fast: Tempus AI acquired Paige for $81.25 million in August 2025, and Roche agreed in May 2026 to acquire PathAI for up to $1.05 billion — signaling that major diagnostics incumbents are moving directly into the space ArteraAI Breast now occupies (Discoveries in Health Policy). For ArteraAI Breast specifically, the clearest paths to broader clinical trust are a peer-reviewed publication of the NSABP B-20 findings, a prospective trial comparing it directly against Oncotype DX, and clarity on Medicare reimbursement.
Conclusion: A Genuine First, Still Being Tested in the Clinic
ArteraAI Breast is a legitimate regulatory first — the first FDA-cleared AI tool to stratify breast cancer risk directly from a routine pathology slide, arriving in days rather than weeks. Its workflow advantages are real and its risk-stratification data draws on genuine randomized trial cohorts. But its most clinically consequential claim — that it can identify who benefits from chemotherapy — currently rests on an exploratory subgroup analysis, not a validated primary finding, and the independent oncologist most publicly quoted on the technology has said plainly she needs more before she’ll use it. That’s not a reason to dismiss the test. It’s a reason to watch, closely, what comes next.
FAQ
What is ArteraAI Breast and what does it do? It’s an AI digital pathology test that analyzes a patient’s existing breast cancer surgical resection slide, along with clinical data, to estimate the risk of distant metastasis and sort patients into low- or high-risk groups.
Is ArteraAI Breast FDA approved? It received FDA 510(k) clearance on May 6, 2026, for use in patients with early-stage, hormone receptor-positive, HER2-negative invasive breast cancer.
How is ArteraAI Breast different from Oncotype DX or MammaPrint? It uses a standard pathology slide already produced during surgery rather than requiring RNA extraction (Oncotype DX) or fresh tissue (MammaPrint), and returns results in one to two days instead of two to three weeks. Unlike those tests, it does not yet have large prospective trial validation specific to its own use.
Does ArteraAI Breast actually predict who needs chemotherapy? Partially, and with caveats. In an analysis of the NSABP B-20 trial, the model was not predictive of chemotherapy benefit in the overall study population — only in an exploratory subgroup of patients aged 50 and older with intermediate or high AI-assigned risk.
What do independent doctors say about the evidence? Reaction is mixed. One breast imaging specialist highlighted the workflow benefits. A breast medical oncologist said she wants head-to-head trials against Oncotype DX and more transparency into the model before adopting it clinically.
Is ArteraAI Breast covered by insurance? Not confirmed as of this writing. Medicare has an established payment rate for Artera’s earlier prostate cancer test, but it’s not confirmed whether an equivalent rate has been set for the breast test. Artera offers a financial assistance program and an early access program with no out-of-pocket cost for enrolled patients.
Where is ArteraAI Breast available right now? It’s processed through a single CLIA-certified laboratory Artera operates in Jacksonville, Florida, and can be ordered only by clinicians within their scope of practice.
This article discusses breast cancer diagnosis and treatment decisions. It is intended for general information and is not a substitute for advice from a qualified oncologist about any individual’s care.