Less Can Be More in Breast Cancer Treatment, Precision Oncology Editors Argue

A commentary published in PLOS Medicine argues that one of the greatest opportunities in modern breast cancer care lies not in giving more patients more treatment, but in identifying those who can safely receive less. Writing on behalf of the journal’s staff editors, Katarzyna Starowicz of the Public Library of Science frames overtreatment as an unintended consequence of oncology’s own success: as effective systemic therapies have multiplied, so have the acute and long-term harms that can erode patients’ quality of life and function, sometimes with limited clinical benefit in return. In the era of precision oncology, the editorial contends, the decisive question is shifting from who can be treated to who can be spared.

The case rests on a striking backdrop of progress. Over the past decade, early diagnosis and effective therapies have driven annual breast cancer mortality reductions of roughly 2 to 2.5 percent in several high-income countries, although the editorial notes this improvement has not been achieved worldwide. Advances in radiotherapy, chemotherapy, HER2-targeting agents, endocrine therapy, immunotherapy and antibody-drug conjugates have transformed outcomes for many patients. Yet the same advances have created a paradox: while many benefit from systemic treatment, toxic effects are common and can persist long after therapy ends. Historically, the editorial explains, approaches have prioritised minimising any possible recurrence, a stance justified by the absence of reliable biomarkers to tailor treatment strategies to individual patients.

The repercussions of overtreatment extend well beyond transient adverse events. Surgery may cause chronic pain, restricted shoulder movement and changes to physical appearance and body image. Radiation is painful and can lead to skin damage and tissue fibrosis. Chemotherapies and some biologic agents are associated with infertility, fatigue, cardiotoxicity, cognitive impairment and secondary malignancies, with cardiotoxicity seen in HER2-targeted treatments and anthracyclines. Endocrine therapy such as tamoxifen, typically prescribed for five years or longer, can produce musculoskeletal symptoms, depression, sexual dysfunction, osteoporosis and vasomotor symptoms, and carries an increased risk of endometrial cancer and blood clots. Against this catalogue of harm, the editorial highlights the growing implementation of de-escalation, also known as de-intensification, in which treatment duration and intensity are reduced as much as possible to minimise adverse effects without compromising outcomes.

The concept has a precedent in another cancer. Watchful waiting, which prioritises managing symptoms and palliative care, and active surveillance, which involves structured monitoring to detect disease progression early and enable timely treatment where appropriate, are increasingly implemented in prostate cancer management precisely to avoid overtreatment. The editorial suggests breast cancer care is now undergoing a comparable reckoning, enabled by molecular tools that can characterise tumours, predict responses to treatment and estimate risks with unprecedented granularity. The power of precision medicine, in this framing, lies in withholding therapy where the benefit-to-harm ratio would be small due to limited efficacy and harmful side effects.

The clearest evidence for de-escalating breast cancer treatment comes from genomic risk stratification. Trials including TAILORx, RxPONDER, MINDACT and, more recently, OPTIMA are changing the management of different breast cancer subtypes. According to the editorial, these studies show that patients with intermediate or low genomic risk, even when their clinical risk of recurrence is high, derive little or no benefit from chemotherapy and can safely avoid it while maintaining good outcomes. This represents a fundamental shift in how molecular diagnostics are understood: rather than serving only to identify candidates for novel targeted therapies, stratifying patients by genomic risk can reduce intervention. First-generation genomic assays, including the 21-gene Oncotype DX recurrence score and the 70-gene MammaPrint signature, are among the tools already used for treatment de-escalation.

Emerging technologies may extend this logic further. Circulating tumour DNA, or ctDNA, can be used as a biomarker to identify molecular relapse months before clinical recurrence would be detected by imaging and biopsy. The editorial is careful, however, to note that there is not yet enough evidence to show that earlier detection of relapse improves patient outcomes or reduces treatment intensity. This distinction between a measurable biological signal and a demonstrated clinical benefit recurs throughout the commentary and anchors its most cautious claims.

Artificial intelligence is the second technological force the editorial examines. AI-based approaches are being used to analyse mammograms in breast cancer screening, exemplified by the Transpara system in the MASAI trial, in which AI-supported screening was compared with standard screening using double reading by radiologists. AI-supported screening achieved similar cancer detection rates while reducing radiologist workload by 44.3 percent. Elsewhere, AI tools have been implemented in genomic-based recurrence and treatment benefit prediction, such as Oncotype DX, in recurrence risk prediction through PreciseBreast, and in digital pathology analysis and treatment response through ENLIGHT-DeepPT. Despite impressive performance, the editorial stresses that most published studies of these tools evaluate diagnostic accuracy rather than patient outcomes or benefits, a gap that limits what can currently be claimed for them.

The implications of identifying candidates for treatment de-escalation are global. In many low- and middle-income countries, where breast cancer is the leading cause of cancer death among women, oncology services, diagnostic infrastructure, qualified personnel and access to specialised and often expensive medicines can be limited. Avoiding unnecessary treatment that provides little benefit, by matching patients to appropriate personalised regimens, would carry both clinical and economic value and could allow resources to be redirected toward more urgent interventions or other aspects of cancer care. At the same time, the editorial acknowledges that in many such settings, limited access to screening and treatment and delays in diagnosis remain major barriers. Where mammography screening is not feasible, improving symptom awareness, reducing barriers to presentation and establishing rapid diagnostic pathways may have greater impact on survival. AI-assisted tools may help address some infrastructure and workforce constraints, but the editorial insists they require local validation and should complement, not divert resources from, essential services. The Lancet Breast Cancer Commission and the WHO Global Breast Cancer Initiative have both advocated context-specific, phased approaches centred on early detection, timely diagnosis and comprehensive treatments adapted to local resources, with measurement of inequities and outcomes.

Reducing global breast cancer mortality and patient burden, the editorial argues, requires simultaneous progress on early detection, timely diagnosis and the selection of the most suitable treatment regimens using modern tools, while avoiding overtreatment. Small gains in recurrence reduction must be weighed against long-term toxicity, survivorship concerns, financial burden, resource constraints and patient preferences. Doctors and healthcare workers, the authors write, should communicate not only the benefits of treatment but also its possible adverse effects, including long-term ones. Patients should be educated that avoiding unnecessary treatment does not mean withholding therapy; rather, it reflects a commitment to providing the best possible care.

Looking forward, the editorial sets out research priorities centred on randomised non-inferiority trials with long-term follow-up to evaluate treatment reduction or omission, and on biomarker-stratified trials to establish whether genomic assays, including ctDNA tests, improve clinical benefit rather than merely predicting risk. The journal states that it welcomes studies evaluating treatment strategies, including de-escalation and optimisation, and precision medicine approaches such as genomic risk stratification tools, particularly where they provide robust evidence on patient-centred outcomes, treatment burden and quality of life. The closing message is that breast cancer changes lives not only through the diagnosis itself but through the impact of treatment, and that advances in precision oncology should allow patients to make more informed decisions, balancing the potential benefits and harms of undergoing therapy, de-escalation or active surveillance to maximise their chances of remission and survival while preserving or improving their quality of life.

Subject of Research: Treatment de-escalation in breast cancer using genomic risk stratification and artificial intelligence

Article Title: Breast cancer treatment: When less is more

Article References: Starowicz, K., & on behalf of the PLOS Medicine Staff Editors (2026). Breast cancer treatment: When less is more. PLOS Medicine, 23(10), e1005284. https://doi.org/10.1371/journal.pmed.1005284

Image Credits: AI Generated

DOI: 10.1371/journal.pmed.1005284

Keywords: breast cancer, precision oncology, treatment de-escalation, genomic risk stratification, Oncotype DX, MammaPrint, circulating tumour DNA, artificial intelligence, mammography screening, overtreatment, endocrine therapy, global health equity