SAN ANTONIO – A margin width beyond ‘no tumor on ink’ may reduce local recurrence in certain subsets of patients undergoing breast-conserving treatment for early-stage breast cancer, according to findings from a meta-analysis.
“Further prospective studies are required to validate the appropriate margin width,” study author Frank A. Vicini, MD, of 21st Century Oncology/Michigan Healthcare Professionals in Farmington Hills, Michigan, said at the San Antonio Breast Cancer Symposium.
Dr. Vicini pointed out that the medical community has been addressing the question of “what is the appropriate margin for breast conserving therapy” for about 3 or 4 decades. A meta-analysis of 33 studies (Ann Surg Oncol. 2014 Mar;21:717-30.) showed that wider margins are unlikely to have a substantial benefit over “no tumor on ink.”
The results of the meta-analysis led to guidelines published by the Society of Surgical Oncology–American Society for Radiation Oncology that supported those conclusions.
“So the question now is that with additional patients and additional modeling, is this conclusion still correct?” Dr. Vicini said.
The 2014 meta-analysis had significant limitations, and to address them, Dr. Vicini and his colleagues conducted an updated analysis of all available data that included 31 of the initial studies. However, stricter criteria were used for inclusion into their review. The authors reviewed all studies published between 1995 and 2016 that had a minimum follow-up of 50 months, and that included explicit pathologic definitions of margin status and local recurrence in relation to margin status.
The final meta-analysis included 55,302 patients (74% T1 tumors; 72% node-negative disease) from a total of 38 studies, and the median follow up was 7.2 years.
The analysis used three generalized linear mixed models for the outcome of local recurrence, and random effects for study and fixed effects for various patient and study characteristics.
In model 1, all patients with at or equal margin width were compared with those who had wider margins. Model 2 examined the impact of margin width “range” as opposed to a set margin width of 0 mm to 2 mm, 2 mm to 5 mm, or greater than 5 mm. Model 3 divided margin distance by cut points of –1 mm, 2 mm, and 5 mm.
The crude rate of local recurrence was 10.3% for patients with positive margins versus 3.8% for those with negative margins that were defined as no tumor on ink or wider (P less than .001). Local recurrence rates declined as the margin distance increased: 7.2% for patients with margins 0-2 mm, 3.6% for margins of 2-5 mm, and 3.2% for margins wider than 5 mm (P less than .001 for each). The use of endocrine therapy and increasing median study year were associated with decreased rates of local recurrence in univariate models but not in multivariable analyses.
For local recurrence in model 1, a benefit for wider margins was observed, with the greatest benefit seen at 1 mm. The odds ratio (OR) for local recurrence for negative vs. close/positive margins was 0.46 (95% confidence interval, 0.4-0.53) for margins larger than 0 mm, 0.43 (95% CI, 0.36-0.51) for those greater than 1 mm, 0.49 (95% CI, 0.42-0.55) for those larger than 2 mm, and 0.53 (95% CI, 0.43-0.66) for greater than 5 mm. Upon multivariate analysis, the only significant predictive variable was margin status.
For model 2, margin width turned out to be the significant variable, and only wider margins were associated with a decrease in local recurrence.
Finally, in model 3, both margin status and margin width were significantly associated with local recurrence. When modeling as negative, close, or positive margins, the lowest rates were at 2 mm (negative, 3.6%; close, 5.5%; positive, 9.5%) and 5 mm (negative, 2.9%; close, 4.1%; positive, 12.8%).
“So the initial question is the same – should we achieve a 1- to 2-mm margin for our patients as compared to a no tumor on ink, taking into account the potential for local control benefits versus morbidity and time, and cost?” said Dr. Vicini.
“The real question in my mind, is which patients with no tumor on ink require additional surgery,” he concluded. “In my opinion, it may not only be related to the margin status itself, but to the volume of disease near the margin but these meta-analyses were not designed to look at that point.”
SOURCE: Vicini et al. SABCS Abstract GS5-01