DCIS represents a broad biologic spectrum of disease, and a wide range of treatment approaches have been proposed. The lack of clear and universally accepted treatment selection criteria has resulted in a diverse array of confusing clinical recommendations. The need for RT as a component of BCT in women with DCIS is controversial. Clinical trials have shown that local excision and RT in patients with negative margins can produce excellent rates of local control [
11‐
16]. Our meta-analysis of four trials that evaluated adjuvant radiotherapy in 3665 patients with CDIS submitted to BCT showed that adjuvant RT leads to a significant reduction (60%) in the risk of a local (invasive and DCIS) in-breast recurrence. However, RT as a component of BCT may represent overly aggressive therapy, since the majority of cases of DCIS do not recur or progress to invasive cancer when treated with excision alone [
3,
17‐
19]. The results of NSABP B-17 have been used to justify radiation for all women; however, pathologic factors affecting local control were largely unrecognized when this and other prospective studies were designed and initiated. Furthermore, deficiencies in pathologic evaluation likely resulted in underdiagnosis of invasive disease and margin involvement. As an example, because neither NSABP B-17 nor EORTC 10853 provided mammographic correlation with either the preoperative imaging and/or with specimen radiography, the completeness of excision is uncertain. In NSABP B-17, microscopic tumor size was not determined, sampling of the surgical specimen could not reliably exclude invasive disease or involved margins, and, as noted previously, management of the resection margins was suboptimal. Reflecting these inadequacies, local recurrence rates were 16 percent at 12 years in the irradiated group [
12]. Similar long-term results were noted in a collaborative study of women undergoing BCT with RT for DCIS in ten institutions in the United States and Europe (15-year actuarial breast recurrence rate 19 percent [
25]). Such data suggest significant limits on the ability of RT to control residual DCIS in the breast, and underscores the importance of complete pathologic specimen evaluation. These deficiencies limit the ability to extrapolate the trial results to patients who undergo optimal pathologic evaluation. Prospective but nonrandomized studies with a larger focus on pathology, including mammographic correlation, have achieved at least comparable results without RT in identifiable subsets of patients with DCIS [
15,
26‐
29]. By weighing factors of prognostic importance (ie, grade, size, age, and margin width), a sizable subset of patients with DCIS (32 percent of those treated by BCT in one study [
27]) can be identified who have a 99 percent twelve-year local recurrence-free survival with excision alone. However, the routine use of these factors, which are included in the USC/Van Nuys Prognostic Index (USC/VNPI) requires a higher standard of pathologic practice than is reflected in the published randomized trials. The original USC/VNPI assigned scores of 1, 2, or 3 for histological type, width of the surgical margin, and lesion size [
26]. A prospectively collected but nonrandomized study suggested that local recurrence rates for lesions with USC/VNPI scores 3 to 4 were acceptably low with excision alone (local recurrence-free survival rates exceeded 99 percent at eight years of follow-up), while those with intermediate scores (5 to 7) required the addition of RT to achieve optimal local control. In contrast, RT provided a significant degree of benefit for the subset of patients whose DCIS was characterized by high grade, large size, and narrow margins (USC/VNPI 8 to 9). The recurrence rate was 100 percent at three years without RT, compared to 60 percent at eight years with RT. Although the relative benefit was large, neither outcome was deemed clinically acceptable, and it was recommended that such patients might be better served by mastectomy. Thus, the identification of subgroups of patients with DCIS for whom RT offers little absolute benefit in local recurrence-free survival is of clinical relevance [
30]. RT is time consuming, and accompanied by significant side effects in a small percentage of patients (eg, cardiac, pulmonary, second malignancies). Radiation fibrosis can alter the texture of the breast and skin, making mammographic follow-up more difficult [
31]. Ongoing trials by both the European and the United States cooperative groups are addressing the issue of benefit from RT. One of them, performed by Radiation Therapy Oncology Group study (RTOG 9804) aims to randomly assign 1790 patients either to radiation therapy or observation, with the option of tamoxifen in either group. The patients must have lesions that are 2.5 cm or less in diameter, low- or intermediate-grade nuclei, and inked margins that are at least 3 mm in diameter. The primary outcomes will be the difference in local recurrence and distant disease-free survival rates. The recently closed European Cooperative Oncology Group (ECOG E-5194) trial was similar. It accrued approximately 1000 patients with DCIS lesions that were 2.5 cm or less and low- or intermediate-grade nuclei or 1 cm or less and high-grade nuclei (both groups had to have negative margins that were at least 3 mm). Patients were treated by lumpectomy alone, and the primary outcome will be local recurrence rates at 5 and 10 years. The results of these studies should provide information on the efficacy of lumpectomy alone with good-risk DCIS and may allow the development of criteria to identify subgroups of patients who do not require adjuvant radiation therapy. Moreover, our data evidenced that contralateral tumor rates are slightly increased by radiotherapy (Figure
3). For the four trials combined there were 49 contralateral tumors in the surgery alone compared to 66 in the patients who also received radiotherapy (P = 0.03). Studies investigating the induction of breast cancer after radiotherapy for Hodgkin's disease have shown intervals of at least 8 years between the primary treatment and presentation of subsequent breast cancer. A large increase was seen in the first report of the EORTC trial [
13], though this may represent a superior finding, given the low standard doses used and short follow up. It has been suggested that the method of delivery of RT in the EORTC study may explain this finding [
32] but has been somewhat reduced with longer follow-up [
33,
34]. The first analysis of the NSABP B-17 did not report a difference in the occurrence of contralateral breast cancer, although in their 8-year update a higher, but non-significant rate of contralateral breast cancer in the treatment group was found when only those contralateral breast cancers occurring as first events (13 in the no further treatment group
vs 19 in the radiotherapy group) were included. However, when all contralateral breast cancers were included in the analysis, this difference disappeared (18
vs 20)[
11]. Studies investigating breast-conserving therapy for invasive breast cancer have not reported an increased rate of contralateral breast cancer. Although our results may have occurred by chance these studies will monitor this finding, if the increase is due to radiotherapy, it will reduce the overall value of this form of therapy for DCIS patients. Finally, none of these trials was sufficiently powered to detect differences in overall survival. Similar rate of distant metastases and death were observed in the two groups. The risk of eventually dying from breast cancer when DCIS is treated with breast-conserving treatment is still no clear. In our study 63 patients died from breast cancer. Although this does not seem a high rate, the follow-up time was relatively short. A few of these deaths would probably have occurred even if the patients had initially had a mastectomy. However, given the number of invasive local recurrences, a significant number will probably die from metastatic disease. Longer follow-up is needed to investigate whether the beneficial effect of radiotherapy on local control will improve survival rates.