FDA clears Artera's AI breast cancer prognosis tool after 1,271-patient validation

The US Food and Drug Administration cleared ArteraAI Breast, a software-only device from Artera, Inc. of Los Altos, California, on 4 May 2026 under 510(k) number K254115 ¹. The FDA received the submission on 19 December 2025 and found the device substantially equivalent to its predicate, ArteraAI Prostate ¹³. Artera announced the clearance on 6 May 2026 and called the product "the first and only FDA-cleared digital pathology-based risk stratification tool for breast cancer" ⁴.

The device analyses scanned whole slide images of haematoxylin and eosin stained breast resection specimens together with three physician-provided clinical variables: age, tumour size and nodal status ². It outputs 5- and 10-year risks of distant metastasis and an ArteraAI risk score for adults with HR+/HER2-, N0 or N1, early-stage invasive breast cancer after surgical resection who are candidates for standard of care adjuvant therapy ². The decision summary states the algorithm is locked, not a continuous learning model ³.

What the validation showed

The model was trained on digitised biopsy and surgical slides from 8,549 patients in four trials: WSG ADAPT, WSG PlanB, NSABP B34 and ABCSG 6 ³. The cut-off between low and high risk was set using 946 held-out patients plus 2,188 patients from NSABP B14 and 1,198 from NSABP B39, and the development data contained 997 distant metastasis events ³.

Clinical performance was assessed in a retrospective study of 1,271 patients with HR+/HER2- early-stage invasive breast cancer diagnosed between 2006 and 2019 at three US sites ³. The tool placed 827 patients (65.1%) in the low-risk group and 444 (34.9%) in the high-risk group ³. Estimated 5-year risk of distant metastasis was 0.9% (95% CI 0.4% to 1.8%) for low-risk patients and 8.7% (95% CI 6.4% to 11.8%) for high-risk patients, against 3.6% overall ³. At 10 years the figures were 2.8% (95% CI 1.6% to 4.7%) for low risk and 16.6% (95% CI 12.8% to 21.5%) for high risk, against 7.6% overall ³. The FDA described each difference from the overall risk as statistically and clinically significant ³.

Raw score reproducibility across three US laboratories gave coefficients of variation from 0.4% to 8.4% on the Philips Ultra Fast Scanner and 0.4% to 4.7% on the Leica Aperio GT450 DX ³.

What it means for the market

For pathology laboratories and cancer centres, the clearance adds a prognostic option that runs on routine H&E slides from the surgical specimen rather than a separate assay; use is limited to slides from FDA-cleared interoperable scanners validated for the device ²³. The FDA also authorised a Predetermined Change Control Plan that lets Artera add further cleared scanners and file formats without a new 510(k), provided each passes the verification process used for the Philips scanner ¹²³. MedTech Dive reported on 12 May 2026 that genomic assays such as Oncotype DX are used to weigh adjuvant chemotherapy in this population and that some clinical facilities lack access to them ⁵.

The cleared indication is prognostic only. The labelling states that the device "is intended to assist physicians with prognostic risk-based decisions along with other clinicopathological factors" and must be used with a complete standard of care evaluation ²³. Artera's press release says data presented at the 2025 San Antonio Breast Cancer Symposium showed the model's potential to inform chemotherapy benefit in certain patient populations, but prediction of treatment benefit is not part of the cleared indication ⁴. Neither the FDA documents nor the press release gives a price, reimbursement code or launch date.

The validation was retrospective and weighted to one institution: Site 3 supplied 1,031 of the 1,271 patients (81%) ³. The cohort was 84% White and 11% African American, and in the high-risk group African American patients had a lower observed 10-year risk than White patients (6.4% versus 20.1%) ³. In that cohort, 65% of patients received endocrine therapy only and 23% received any chemotherapy, so the results describe outcomes under treatment rather than untreated risk ³. The device warnings state that result correctness depends on slide image quality and proper scanner use, and a pathologist must confirm cancer is present in the image and approve the report before release ³.

Under the authorised change control plan, labelling will be updated each time an additional scanner passes validation. The decision summary names two scanners, the Philips Ultra Fast Scanner (K172174) and the Leica Aperio GT450 DX (K232202) ³.