Journal of Clinical Oncology November 17, 2026

Partial-Breast Irradiation Versus Whole-Breast Irradiation for Low-Risk Invasive Breast Carcinoma: 5-Year Results of the Multicenter, Randomized, Phase III IRMA Trial

Bruno Meduri, Antonella Baldissera, Cinzia Iotti, et al.

Bottom Line

The phase III IRMA trial demonstrated that external-beam accelerated partial-breast irradiation yields exceptionally low 5-year local recurrence rates comparable to whole-breast irradiation in low-risk breast cancer patients, making it a viable clinical alternative despite not strictly meeting statistical noninferiority.

Key Findings

1. At 5 years, the cumulative incidence of ipsilateral breast tumor recurrence (IBTR) was 1.0% (95% CI, 0.6 to 1.6) in the accelerated partial-breast irradiation (APBI) group and 0.8% (95% CI, 0.5 to 1.4) in the whole-breast irradiation (WBI) group.
2. The hazard ratio for APBI versus WBI for local recurrence was 1.18 (90% CI, 0.63 to 2.20).
3. Formal statistical noninferiority was not met because the upper confidence limit (2.20) crossed the prespecified margin of 1.5, though the absolute difference between the arms was less than 0.2%.
4. Both treatment arms resulted in exceptionally low (≤1%) 5-year local recurrence rates in this low-risk population.

Study Design

Design
Randomized Phase III Trial
Open-Label
Sample
3,309
Patients
Duration
97.5 mo
Median
Setting
Multicenter, Europe
Population Patients age ≥49 years with invasive breast cancer, resection margins ≥2 mm, tumor diameter ≤3 cm, and pN0-1
Intervention External-beam accelerated partial-breast irradiation (APBI) (38.5 Gy in 10 fractions twice daily)
Comparator Whole-breast irradiation (WBI)
Outcome Ipsilateral breast tumor recurrence (IBTR)

Study Limitations

• The trial failed to formally demonstrate statistical noninferiority because the upper confidence limit of the hazard ratio crossed the prespecified margin.
• Neither patients nor investigators were masked to treatment allocation (open-label design), which could introduce bias, though objective endpoints like tumor recurrence are less susceptible to this.
• The very low event rate in both arms limited the statistical power to rule out a relative increase in recurrence, even with a large sample size.
• Median follow-up of 97.5 months, while substantial, may still miss very late recurrences, especially in a hormone-sensitive breast cancer population.

Clinical Significance

Although the trial technically missed its noninferiority endpoint, the absolute 5-year recurrence rates were exceedingly low (≤1%) with an absolute difference of <0.2% between the two treatments. Consequently, external-beam APBI (38.5 Gy in 10 fractions twice daily) represents a highly reasonable, time-saving, and effective de-escalation alternative to standard whole-breast irradiation for carefully selected women aged ≥49 years with low-risk early-stage invasive breast carcinoma.

Historical Context

The standard of care for early-stage breast cancer following breast-conserving surgery has traditionally been whole-breast irradiation. Over the past two decades, there has been a strong push toward treatment de-escalation, specifically accelerated partial-breast irradiation (APBI), which reduces both the treatment volume and duration, easing the burden on patients. Trials such as NSABP B-39 and RAPID established APBI as largely comparable to whole-breast irradiation in selected low-risk patients. The IRMA trial was launched to further validate the noninferiority of external-beam APBI against whole-breast irradiation, with preliminary toxicity and cosmetic results published in 2023. This 2026 5-year outcomes report provides mature efficacy data showing excellent disease control, reinforcing APBI's position as a standard option in modern breast oncology.

Guided Discussion

High-yield insights from every perspective

Med Student
Medical Student

What is the radiobiological and anatomic rationale for using accelerated partial-breast irradiation (APBI) instead of whole-breast irradiation (WBI) following breast-conserving surgery in low-risk patients?

Key Response

Pathologic studies demonstrate that the vast majority of local breast cancer recurrences occur in the tumor bed (the index quadrant). APBI leverages this anatomic fact by targeting only the lumpectomy cavity with a margin, sparing healthy breast tissue, heart, and lungs, while allowing for larger doses per fraction (hypofractionation) to safely shorten the treatment course.

Resident
Resident

When evaluating a patient with early-stage breast cancer in the radiation oncology clinic, what specific clinicopathologic features make her an ideal low-risk candidate for partial-breast irradiation according to current consensus standards?

Key Response

Residents must know patient selection criteria for safe APBI application. Ideal candidates are typically older (e.g., age >50), with small tumors (T1, <2cm), node-negative disease, estrogen receptor-positive status, clear surgical margins, and no lymphovascular space invasion (LVSI) or extensive intraductal component (EIC).

Fellow
Fellow

The IRMA trial failed to strictly meet its statistical noninferiority margin, yet authors concluded APBI is a viable alternative. How does an unexpectedly low baseline event rate compromise noninferiority trial designs, and how should a fellow interpret this discrepancy clinically?

Key Response

Noninferiority trials are statistically powered based on an expected event rate. With modern advances in surgery and systemic therapies, local recurrence rates often fall to 1-2 percent. At these low rates, trials lose statistical power, and confidence intervals can cross the noninferiority margin even if the absolute difference in recurrence is a clinically negligible fraction of a percent. Fellows must understand that failing mathematical noninferiority does not invariably equate to clinical inferiority.

Attending
Attending

Given that APBI in the IRMA trial did not technically meet noninferiority for local control but offers substantial advantages in treatment duration and convenience, how should attendings frame this trade-off during shared decision-making?

Key Response

Attendings must balance statistical thresholds with clinical pragmatism. If the absolute risk of recurrence difference is less than 1 percent, the physician should explain this tiny absolute risk increase against the massive quality-of-life benefit of completing radiation in 1 week instead of 3 to 5 weeks, factoring in the patient's lifestyle, travel burden, and personal anxiety regarding recurrence.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD
PhD

From a biostatistical and trial design perspective, how could the IRMA investigators have optimized their noninferiority margin or interim analysis plan to account for the contemporary, systemic-therapy-driven decline in breast cancer recurrence rates?

Key Response

When baseline event rates are extremely low, relative noninferiority margins (like a hazard ratio upper bound) become overly sensitive to tiny absolute differences. Alternative trial designs could utilize an absolute risk difference margin (e.g., strictly ruling out a >1.5 percent absolute increase in recurrence) or employ adaptive, event-driven designs that allow for sample size re-estimation if the pooled event rate is lower than expected.

Journal Editor
Journal Editor

As a peer reviewer assessing the IRMA trial, why might you argue that reporting 5-year local recurrence data in a predominantly estrogen receptor-positive, low-risk breast cancer cohort is insufficient for definitively establishing APBI's safety profile?

Key Response

Low-risk, hormone-receptor-positive breast cancers are notorious for late recurrences, often relapsing 5 to 15 years post-treatment. A 5-year endpoint in this specific tumor biology might significantly underestimate the true cumulative recurrence differences between APBI and WBI, representing a major limitation that an editor should require authors to explicitly address.

Guideline Committee
Guideline Committee

How should guideline committees, such as ASTRO or ESTRO, integrate the IRMA trial's results into the next iteration of APBI suitability criteria given the discordance between the strict statistical failure of noninferiority and the excellent absolute clinical outcomes?

Key Response

Current ASTRO guidelines categorize patients into suitable, cautionary, and unsuitable groups for APBI. Committees must decide if real-world absolute recurrence rates of <2 percent at 5 years justify maintaining or expanding the suitable category for external beam APBI, acknowledging that clinically, a fraction of a percent difference in local control does not outweigh the patient-centered benefits of shorter treatment, even if strict statistical noninferiority criteria were not met.

Clinical Landscape

Noteworthy Related Trials

2017

IMPORT LOW Trial

n = 2,018 · Lancet

Tested

Partial-breast radiotherapy

Population

Women aged 50 or older with early breast cancer

Comparator

Whole-breast radiotherapy

Endpoint

Local relapse

Key result: Partial-breast irradiation was found to be non-inferior to whole-breast irradiation for local relapse at five years, with reduced late toxicity.
2019

NSABP B-39/RTOG 0413 Trial

n = 4,216 · Lancet

Tested

Accelerated partial-breast irradiation

Population

Women with early-stage breast cancer post-lumpectomy

Comparator

Whole-breast irradiation

Endpoint

In-breast tumor recurrence

Key result: PBI did not meet the strict statistical criteria for equivalence to WBI for local recurrence, though the absolute difference was small at ten years.
2019

RAPID Trial

n = 2,135 · NEJM

Tested

Accelerated partial-breast irradiation

Population

Women aged 40 or older with node-negative breast cancer

Comparator

Whole-breast irradiation

Endpoint

Ipsilateral breast tumor recurrence

Key result: APBI was non-inferior to WBI for tumor recurrence but was associated with an increase in late radiation toxicity and adverse cosmetic outcomes.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis
↑