NSABP B-06

What the trial asked

By the late 1970s, the argument about radical surgery was already cracking. NSABP B-04, completed under Bernard Fisher at the National Surgical Adjuvant Breast and Bowel Project in Pittsburgh, Pennsylvania, had shown that removing less axillary tissue produced survival no worse than Halsted's radical operation. The logical next question was sharper: could the breast itself be left largely intact?

B-06 enrolled 1,843 women between 1976 and 1984. All had primary tumours measuring no more than four centimetres and had operable lymph nodes. They were randomly assigned to one of three groups — total mastectomy; lumpectomy alone; or lumpectomy followed by breast radiotherapy. Axillary dissection was performed in every group, so the surgical variable was clear: the breast, and only the breast, was what differed. The design was Fisher's, and the logic underneath it was his too — the hypothesis, drawn from experimental data and from the failure of ever-larger surgery to improve survival, that breast cancer was a systemic disease from early in its course, not a local one spreading outward in orderly rings.

What the results showed

The primary finding was unambiguous. At twenty years of follow-up — results published in the New England Journal of Medicine in 2002 — overall survival, disease-free survival, and distant-disease-free survival were statistically equivalent across all three groups. Mastectomy offered no survival advantage over lumpectomy with radiotherapy. The NSABP reported a cumulative incidence of ipsilateral breast tumour recurrence of thirty-nine percent in the lumpectomy-alone group at twenty years, falling to fourteen percent when radiotherapy was added. That gap settled the internal argument: radiotherapy was not optional. Lumpectomy without it was an inferior local treatment, even if it did not shorten life.

The survival equivalence held whether or not lymph nodes were positive — an important finding because node-positive disease had long been the justification for aggressive local intervention. Fisher's systemic hypothesis survived the test. So did the operation: conserving surgery moved from a contested alternative to a standard of care backed by level-one evidence.

Significance and context

B-06 did not exist in isolation. Umberto Veronesi and colleagues at the Istituto Nazionale dei Tumori in Milan, Italy, were running Milan I on a closely parallel track, comparing radical mastectomy with quadrantectomy plus radiotherapy in similarly sized tumours. Both trials reached the same conclusion by different surgical definitions of "conserving" — quadrantectomy removed considerably more tissue than lumpectomy — and the convergence gave the result a robustness neither trial could provide alone. The Early Breast Cancer Trialists' Collaborative Group at Oxford subsequently pooled data from these and other trials, and the meta-analyses confirmed that local recurrence rates and long-term mortality tracked consistently with the individual trial findings.

The data-integrity controversy of 1994 — when it emerged that a Montreal investigator had falsified records on a subset of B-06 patients — caused Fisher to be temporarily removed from the NSABP leadership. Reanalysis of the data without the falsified records left the main conclusions intact. The episode is part of the trial's history but not of its science: independent audits found no change in the direction or significance of any primary outcome.

B-06 also had a regulatory consequence. The United States National Cancer Institute issued a clinical announcement in 1990, before the twenty-year data were complete, stating that lumpectomy with radiotherapy was appropriate treatment for most women with Stage I and Stage II breast cancer — one of the few times an ongoing trial produced guidance mid-course. What had been a surgeon's choice became, in principle, a patient's.