A trial screened women by risk instead of age. Late-stage cancers did not rise.

For thirty years the argument over breast cancer screening has been an argument about ages and intervals: start at 40 or 50, screen every year or every two. WISDOM, published in JAMA on 12 December, tested a different premise — that the schedule should be set by a woman's estimated risk rather than by her birthday [s1].

What the trial did

WISDOM is a parallel-group, pragmatic, multicentre randomised clinical trial. It randomised 28,372 women 1:1 to annual mammography or to risk-based screening, with a median follow-up of 5.1 years [s1]. It ran alongside a much larger enrolment — over 80,000 women since the study began in 2016, including an observational arm for participants who preferred to choose their own approach [s2].

Risk was estimated using validated risk models combining age, health history, lifestyle, and breast density, plus genetic information [s2]. Beyond BRCA1 and BRCA2, the study used a polygenic risk score built from many small DNA variants; adding it moved 12–14% of participants into a different risk category than they would otherwise have occupied [s2].

The resulting schedules, and the share of participants in each tier [s2]:

• Lowest risk (26%): no screening until age 50, or until an algorithm indicated risk equivalent to that of a 50-year-old
• Average risk (62%): mammography every two years
• Elevated risk (8%): annual mammography
• Highest risk (2%): twice-yearly imaging, alternating mammography and MRI, regardless of age

What it found

The primary safety question was whether tailoring schedules would let more cancers reach an advanced stage before detection. It did not, by the trial's non-inferiority criterion. Stage ≥IIB cancers numbered 21 in the risk-based group against 31 in the annual group — a rate difference of −18.0 per 100,000 person-years [s1]. Non-inferiority also held for stage ≥IIA cancers [s1].

Biopsy rates did not differ significantly: 1,029 in the risk-based arm against 943, a rate difference of 98.7 per 100,000 person-years [s1].

The risk-based arm received fewer mammograms overall, with the reduction concentrated among low-risk women, and uptake of preventive therapy nearly doubled among the highest-risk group [s1].

A finding from the genetic testing is worth separating out: roughly 30% of women who tested positive for a cancer-risk variant had no family history of the disease [s2] — meaning they would not have qualified for genetic testing under criteria that use family history as the gate.

What the numbers are and are not

The absolute counts are small. Twenty-one advanced cancers versus thirty-one, over five years in a trial of 28,372 women, is a thin margin on which to rest a change in national policy, and non-inferiority was the test — the trial was designed to show the risk-based approach was not worse, not that it was better.

Five years of median follow-up is also short for breast cancer. Screening trials have historically needed a decade or more before mortality differences emerge, and WISDOM's primary outcomes here are cancer stage and biopsy rate, not death.

The authors name three limitations directly [s1]: screening adherence was suboptimal; the participants were predominantly college-educated and non-Hispanic White, which limits generalisability; and the study relied on self-report with variable screening patterns, which can affect the accuracy of outcome ascertainment.

That second limitation carries particular weight for this question. Breast cancer mortality in the United States is markedly higher among Black women, and risk models perform differently across ancestries — polygenic risk scores especially, since they were largely derived in European-ancestry cohorts. A risk-stratification strategy validated mostly in white, college-educated women is not automatically safe to deploy in populations where the models are less accurate, and the trial does not establish that it is.

The preference finding

Among participants in the observational arm who chose their own approach, 89% chose risk-based screening [s2]. That is a statement about what women offered the choice preferred, in a self-selected group that had already enrolled in a screening study — not a population-representative preference estimate. It matters mostly because "women will not accept less screening" has often been asserted as a reason not to test the question.

What the investigators say it means

Laura J. Esserman, MD, MBA, director of the UCSF Breast Care Center and the study's first author, said the findings "should transform clinical guidelines for breast cancer screening and alter clinical practice" [s2]. Allison S. Fiscalini, MPH, who directs the study, described it as "one of the first studies to offer genetic testing to all women, regardless of family history" [s2].

Guideline bodies will decide that on their own timetable, and they will weigh the same limitations listed above.

What to watch

Whether the US Preventive Services Task Force, the American Cancer Society, or comparable bodies take up risk-based schedules, and on what evidence threshold. Longer follow-up from WISDOM, particularly on mortality. Whether the polygenic component performs adequately in more diverse cohorts. And the operational question underneath all of it — whether health systems can actually deliver individualised risk assessment, including genetic testing, at population scale.

This article describes trial results. It is not medical advice, and screening decisions belong with a clinician who knows the individual case.

Sources

1. Risk-Based vs Annual Breast Cancer Screening: The WISDOM Randomized Clinical Trial — JAMA, 12 December 2025
2. UCSF Study Finds a Better Way to Screen for Breast Cancer — University of California, San Francisco, 12 December 2025