Silverstein MJ, ... Colburn WJ
Abramson Cancer Center of the University of Pennsylvania
Last Modified: November 1, 2001
Reviewers: John Han-Chih Chang, MD
Source: New England Journal of Medicine 1999; Volume 340: pages 1455 - 1461
Breast conservation therapy usually consists of a lumpectomy or wide local excision of the tumor followed by radiation therapy (RT) or observation. A series of 268 patients from the Hospital of the University of Pennsylvania demonstrated that the 15-year cancer specific survival of patients treated for DCIS with lumpectomy and definitive breast irradiation was 96%. The National Surgical Adjuvant Breast and Bowel Project (NSABP) B - 17 trial is the only published randomized trial comparing lumpectomy with versus without RT. Over 800 patients were evaluated, and the study showed that at 8-years the DCIS recurrence rate was decreased from 13.4% to 8.2% along with a reduction from 13.4% to 3.9% in recurrence as invasive disease. One major criticism of this trial was that subset analyses failed to demonstrate when it would be appropriate not to give RT. Some critics cite that the NSABP definition of negative margins (no tumor at the margin) had no margin width definition. The authors cite that over 40% of the patients did not have information on margin width.
Dr. Silverstein in an earlier manuscript described the formulation of his Van Nuys Prognostic Index (VNPI). It took into account the size, margin width and pathological classification of the DCIS as a predictor of local recurrence. The formulation of his VNPI was based on retrospective data accumulated and analyzed from the patients treated at the Breast Center in Van Nuys, California and the Children's Hospital in San Francisco (the database utilized in this paper, also). The VNPI assigned a score of 1 - 3 for each of the categories listed above, with 3 being the least promising prognosis. For example: a tumor size 40 mm or greater was scored a 3, while tumor size no greater than 15 mm was scored a 1 (obviously, 2 was tumor size 16 - 40 mm). Since there were three factors evaluated, the lowest score was 3 with a high of 9. Based on the VNPI, their retrospective review demonstrated that a score of 3 - 4 did not benefit significantly from RT. Those patients who had scores above 4 did significantly benefit from RT, but the authors claim that the local recurrence rate is so high for the patients with scores of 8 - 9 that they should proceed to mastectomy.
The topic of this article seems to define now that margin width has now become the only significant prognostic factor in the decision of whether to give RT or not.
Pathological evaluation included size of the primary tumor, margin width and nuclear grade. Comedonecrosis was also screened for in each sample. These comprise all three components of the VNPI as previously mentioned.
Table 1 demonstrates the relative risk of local recurrence at 8 years as a function of margin width. I have reproduced it here:
Definitive Breast RT
Number of Patients
Number of LR
LR rate at 8 years
Number of Patients
Number of LR
LR rate at 8 years
1 to < 10
The data here is graphically depicted in the articles in figures 1 - 3.
Table 2 demonstrated the associated characteristics in each of the different margin width arms. The authors evaluated in the median tumor size, mean nuclear grade, the percentage of nuclear grade 3 patients and the incidence of comedonecrosis. Most of the pathological characteristics were similar in the two groups (RT versus excision alone). However, for margins of 1 mm or greater, the median tumor size was greater in the definitive breast RT group.
Table 3 stratifies the data according to the presence or absence of comedonecrosis, nuclear grade and tumor size with the three margin width categories. With stratification, there appears to be a trend for improvement of local control even in patients with margin widths of 1 to < 10 mm (relative risk of 1.75 - 1.87 for excision alone over RT). The benefit is clear in all strata in the close to positive margin category.
One minor criticism is the fact that the authors utilized mean rather than median follow-up, which is the standard convention. It can not be ascertained from the data given, but perhaps the discrepancy between the two follow-up periods would be even greater than the already large rift (20 months difference in the mean follow-ups). The mean follow-up of the excision alone group could be increased significantly by a few long-term patients. Both mean and median should have been reported. Again, it is evident that in the close to positive margin group (high risk), definitive RT is beneficial. The demonstrated benefit is seen earlier since failures tend to occur earlier. Another point of contention was the large difference in the mean follow-up length. This could be the reason that there seemed to be no difference in the low and intermediate risk group in regards to local recurrence. The data is biased, since the definitive RT patients have been out longer and, thus, have a longer at-risk period. With equivalent follow-up, perhaps it would be demonstrated that the low and intermediate risk patients would also benefit from RT.
Table 1 shows the skewed nature of the pathological characteristics, as most of the wide excision margin patients are in the excision only group. In addition, more of the close or positive margin patients are in the definitive RT group. As stated in the results above, patients that were treated with definitive RT had a greater median tumor size.
Thus, we have two major confounding factors -- length of follow-up and patient selection (RT patients had larger tumors and closer margins). These factors obviously work against the patients assigned to receive RT.
The objective of this article was to define a subset of patients with DCIS that could be treated with excision alone. The authors have defined that those with close or positive margins do need RT, but the data on those with 1 mm or more of tumor-free margin remains unclear. To start denying definitive breast RT to DCIS patients off protocol would be dangerous in light of a positive randomized data that supports RT (NSABP B - 17 ) and the significant flaws noted above in this retrospective review. The authors were honest in stating that this a good starting point, but must be confirmed by a randomized trial that stratifies for margin width in DCIS. Currently, the Eastern Cooperative Oncology Group is enrolling patients onto a prospective registration protocol for excision only in DCIS patients who have low grade disease and wide margins on lumpectomy.
The bottom line is that we have STILL not solved the issue of who can safely avoid getting definitive breast RT. The future of DCIS may include the use of systemic therapy (Tamoxifen). In the NSABP B - 24 randomized trial of Tamoxifen versus placebo, there was a significant reduction in the local invasive recurrence and any breast cancer rate with Tamoxifen. This may need to be integrated into any future trials regarding the necessity of definitive breast RT.