Reply to ‘Intraoperative radiotherapy for breast cancer: powerful evidence to change practice’

The TARGIT-A investigators claim in their Correspondence (Vaidya, J. S. et al. Intraoperative radiotherapy for breast cancer: powerful evidence to change practice. Nat. Rev. Clin. Oncol. https://doi.org/10.1038/s41571-021-00471-7 (2020))¹ that our 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))² contains several factual and logical errors but we are unclear as to what these are. We stand by our assertion that the treatment that they recommend has not been shown to have any effect against local recurrence and suggest that, were radiotherapy held to the same regulatory standards as chemotherapy, this particular usage would not be licensed.

It seems that we failed to convey our central argument in that Vaidya et al.¹ complain that we overlooked the long-term positive findings and the patient perspective. We accept that targeted intraoperative radiotherapy during lumpectomy (TARGIT-IORT) is more convenient than conventional whole-breast external beam radiotherapy (WBRT). However, a new treatment needs to have been shown (at least indirectly) to be superior to no treatment³, and TARGIT-IORT fails in this regard. We know of no convincing argument that TARGIT-IORT might reduce non-breast cancer mortality relative to no radiotherapy, therefore, the question as to whether TARGIT-IORT results in fewer non-breast cancer deaths than WBRT is only relevant if TARGIT-IORT is shown to reduce breast cancer recurrence.

In their analysis of indirect evidence of superiority to no radiotherapy, the authors make an error — they focus on the wrong end of the confidence interval (CI). In Prime-II⁴ the 95% CI for the additional risk of ipsilateral recurrence at 5 years comparing no radiotherapy with WBRT was 1.1–4.8%. TARGIT-A has not shown the difference in local recurrence at 5-years comparing TARGIT-IORT to WBRT to be <1.1% (and may therefore be no better than no radiotherapy). Vaidya et al. make the point that patients enrolled in PRIME-II had a generally lower risk of recurrence, but 5-year local recurrence in women assigned WBRT was in fact higher in PRIME-II (1.3%) than in TARGIT-A (1.0%).

Rather than using the 95% CI from a single trial, we would suggest using meta-analysis⁵: we calculate the 95% CI for the risk ratio as 0.26–0.38 (Supplementary information). Thus, to show that TARGIT-IORT is superior to no radiotherapy, one would have to set a non-inferiority margin on the relative risk of 1/0.38 = 2.6, and require that the upper limit of the 95% CI comparing TARGIT-IORT with WBRT is at most 2.6. Based on the number of local recurrences by 5 years in the per-protocol analysis, the 95% CI for the risk ratio in TARGIT-A is 1.14–4.99 (ref.⁶) (Supplementary information). By contrast, in FAST-Forward⁷, the upper limit of the 95% CI for the hazard ratio comparing 26 Gy in 5 fractions (over 1 week) to 40 Gy in 15 fractions (over 3 weeks) was 1.16: well below the margin required to infer superiority to no radiotherapy.

TARGIT-IORT is inferior to WBRT in terms of local recurrence: in all four analyses, the 95% CI for the excess local recurrence at 5 years with TARGIT-IORT does not include 0 (ref.⁶); and at a median follow-up duration of 8.6 years, the excess is 3.2% (95% CI 1.7–4.7%) (Supplementary information). Compared with WBRT, TARGIT-IORT (with WBRT added for the 20% of patients at greatest risk of recurrence) leads to at least an additional 1.7% having local recurrence. The authors emphasize that there was no significant difference in recurrence-free survival in women randomized pre-surgery; however, that was a post-hoc sub-group analysis. Combining with data from the post-surgery sub-group⁸ yields a hazard ratio for recurrence-free survival of 1.20 (95% CI 1.01–1.43) (Supplementary information).

We agree that “patient choice, informed by clearly presented evidence”¹ is important. When presenting the benefits of TARGIT-IORT to patients it is essential to also make clear that it is somewhat inferior to standard treatment and might be no better than no treatment.