Intraoperative radiotherapy for breast cancer: powerful evidence to change practice

We believe that the recent News and Views article (Sasieni, P. D. & Sawyer, E. J. Intraoperative radiotherapy for early breast cancer — insufficient evidence to change practice. Nat. Rev. Clin. Oncol. 17, 723–724 (2020))¹ about the TARGIT-A trial contains several factual and logical errors. This article overlooks both the long-term positive findings² and the all-important patient perspective.

Risk-adapted single-dose targeted intraoperative radiotherapy during lumpectomy (TARGIT-IORT) is a method of partial breast irradiation (PBI) for early breast cancer. Most patients (80%) receiving TARGIT-IORT² during their lumpectomy complete their local treatment entirely during this single session, under the same anaesthetic. Supplemental whole breast external beam radiotherapy (WBRT) is only recommended for a minority of patients (20%) if unexpected prespecified tumour-related factors such as invasive lobular cancer and positive margins are found postoperatively. However, most patients with conventional ‘high risk’ features were treated without supplemental WBRT, including four-fifths of those with grade 3 or ER-negative disease, and two-thirds of node-positive cases. By contrast, traditional WBRT or other PBI approaches require up to 30 additional hospital visits — TARGIT-IORT involves far fewer clinic appointments³. Other benefits include fewer toxicities, less pain, better cosmetic results and better quality of life².

The TARGIT-A randomized trial compared risk-adapted TARGIT-IORT with WBRT. The long-term results² revealed no significant differences in local and distant control, breast preservation or breast cancer mortality. Local control was also comparable to that achieved with TARGIT-IORT alone². A significant reduction in non-breast cancer mortality (from cardiovascular causes and other cancers) was also observed with TARGIT-IORT, from 9.85% to 4.41% at 12 years². For patients, who sit on the more uncomfortable side of the consultation desk, these are most welcome results, particularly in the COVID-19 era.

The authors complain¹ that TARGIT-IORT was not compared with ‘no radiotherapy’; however, we emphasize that the TARGIT-A cohort had a much higher proportion of high-risk patients than trials investigating this approach (Supplementary information). In fact, more than three-quarters of patients (1,737 of 2,298) in TARGIT-A² would not have fulfilled the low-risk criteria for inclusion in a trial of ‘no radiotherapy’ such as PRIME-II (inclusion criteria: age >65 years, tumour diameter ≤3 cm, grade 1 or 2, node-negative and ER positive). Despite this higher-risk cohort, local recurrence with TARGIT-IORT was 2–3 times lower than with ‘no radiotherapy’ in those trials (Supplementary information). Crucially, for a more inclusive population such as this, which is more representative of clinical practice, a ‘no radiotherapy’ arm would be unethical. We agree that “discriminating … those who can safely avoid radiotherapy altogether remains a fundamental challenge”¹, therefore, patients should not be recommended ‘no radiotherapy necessary’ without first discussing options such as TARGIT-IORT. We emphasize that with TARGIT-IORT completed during lumpectomy, 80% of patients do not need postoperative radiotherapy².

The proportion of high-risk patients in the TARGIT-A cohort (PBI versus WBRT) is remarkably similar to that of the Fast-Forward cohort (shorter-course WBRT versus 3-week daily WBRT) (Supplementary information), which the authors recommend¹. The 5-year local recurrence with 3-week WBRT in Fast-Forward and TARGIT-IORT was virtually identical at 2.1%. If the authors¹ seriously question whether TARGIT-IORT is better than ‘no radiotherapy’¹, should the same question not also apply to the Fast-Forward WBRT regimen? In any event, ‘no radiotherapy’ is not considered the standard of care for such patients and therefore is not the correct comparator.

The effectiveness of PBI approaches such as TARGIT-IORT has been repeatedly demonstrated (Supplementary information), yet the authors do not mention this important concept. Instead, they promote¹ the intensive ‘Fast-Forward’ whole-breast-radiotherapy approach, which we argue represents overtreatment for the majority of patients and comes with well-known hazards: the most important adverse effects of an increased irradiated volume and the associated scattered irradiation are the substantially increased risks of cardiovascular^4,5 and cancer-related mortality^4,6, which are avoided by PBI techniques⁷ such as TARGIT-IORT^2,8. Conversely, as expected with WBRT techniques, there is no mortality benefit with Fast-Forward. Fast-Forward also entails inevitable post-operative delay plus 7–15 hospital visits (for consultation and planning followed by daily WBRT with or without boost).

The authors criticize the TARGIT-A non-inferiority margin of 2.5%² and surprisingly claim¹ that no radiotherapy (as used in PRIME-II, Supplemental information) is non-inferior to WBRT. We argue that the data disprove this claim — the actual difference in 5-year local recurrence in PRIME-II was 2.9%, with an upper confidence interval of 4.8% — both well above the 2.5% margin². The 2.5% non-inferiority margin used in TARGIT-A² is one of the most stringent (in both absolute and relative terms) among trials involving PBI (Supplementary information). Nonetheless, the actual difference in 5-year local recurrence between the two treatment arms of TARGIT-A was just 1.16%.

The Kaplan-Meier model, which we used² to analyse local control, includes all relevant events^9,10 in addition to time of occurrence and length of follow-up monitoring, for every patient. This is not the case for a chi-square test, which was employed by the authors¹ to test for superiority, even though TARGIT-A was a non-inferiority trial — a very different concept: “Non-inferiority trials … test new treatments that have obvious non-oncological advantages … The non-inferiority statistical test … is not meant to check for superiority, but to assess if the difference is within an acceptable margin and the experimental treatment is not meaningfully worse than the control”². The protocol-specified non-inferior 5-year local recurrence associated with TARGIT-IORT was clearly confirmed in TARGIT-A.

Many countries across the world have enthusiastically embraced TARGIT-IORT, with > 45,000 patients treated so far. TARGIT-IORT is now recommended in many international guidelines. Patient choice, informed by clearly presented evidence, is now recognized as being much more important than clinician preferences, a point powerfully underscored by the UK Supreme Court (Montgomery v Lanarkshire Health Board, 2015), the Royal College of Surgeons of England, and the UK General Medical Council.

There is a reply to this letter by Sasieni, P. D. & Sawyer, E. J. Nat. Rev. Clin. Oncol. https://doi.org/10.1038/s41571-021-00472-6 (2020).