The increasing incidence of DCIS, its biological heterogeneity, and controversy about its treatment have made the management of this condition challenging. We review data on the natural history of this disorder and the results obtained with various management approaches. Specifically, we address the following points: the incidence of DCIS, the classification of the condition, the molecular genetic abnormalities associated with it, its usual clinical presentation, its natural history, the proportion of lesions that are multicentric, the treatment options, whether axillary lymph node dissection should be performed, whether all patients need radiotherapy if they undergo lumpectomy, the survival rates associated with these treatments, the risk factors for recurrence, the best treatment for recurrence, and the possible role of hormonal therapy.

Methods

Top Methods Author & Article Info References

We performed a computerized search of the MEDLINE database for English-language articles on DCIS of the breast published since 1966. The search was done by using the words carcinoma in situ, ductal carcinoma in situ, intraductal breast cancer, and DCIS. References in the identified articles were reviewed. Because randomized trials were given higher value but were relatively scarce, retrospective studies were also reviewed. In addition, published abstracts were reviewed, but no personal communications and case reports were considered. The authors reviewed all sources critically but conducted no formal statistical analyses.

Data Synthesis

Biology and Natural History

Incidence

The exact incidence of DCIS in the general population is unknown and has been a source of controversy. In an autopsy study of 519 women of various ethnic backgrounds, only one case (0.2%) of DCIS was found (in a woman who died at 40 years of age) [2]. Of these 519 women, only 117 were older than 55 years of age. Alpers and Wellings [3] performed autopsies on 185 patients, of whom 9 (4.9%; 11 breasts) had evidence of the tumor on subgross sampling (which involves cutting breast tissue into 2-mm slices, examining the slices with a dissecting microscope, and evaluating any focal lesion [4]). The findings of the autopsy studies suggest that incidental DCIS of uncertain clinical relevance is not common. Variations in incidence among autopsy studies probably reflect differences in sampling techniques and diagnostic criteria. Ductal carcinoma in situ of the breast also occurs in men and represents 3.5% to 7% of all cases of male breast cancer [5, 6].

The incidence of DCIS has increased in recent years, partly because screening mammography is more refined and more widely used [7-10]. In 1992, DCIS was diagnosed in more than 23 300 women (age-adjusted incidence rate, 15.9 per 100 000 women) [11]. The rate of increase in incidence has been higher for DCIS than for any other type of breast cancer [12]. The reported incidence in women 50 years of age or older increased 235% from 1979 to 1986; in contrast, the incidence of invasive cancer increased 50% [13]. However, mammography alone does not seem to explain the increasing incidence of DCIS. Data from the metropolitan Atlanta Surveillance, Epidemiology, and End Results (SEER) study (1979 to 1986) showed that asymptomatic tumors (those detected on screening mammography alone) accounted for only 25% to 40% of the increase in incidence [14]. These reports suggest that increased detection accounts for some but not all of the increase in incidence. The National Cancer Institute SEER data for the entire United States confirmed this trend [11]. Overall, the total number of cases of DCIS in 1992 was 200% greater (23 368 cases) than expected for that year as calculated from the trends between 1973 and 1983.

Pathology

Ductal carcinoma in situ is a heterogeneous entity with several morphologic variants that markedly differ in gross appearance, growth pattern, and cytologic features. Furthermore, it is part of a spectrum of proliferative ductal lesions of the breast that extends from epithelial hyperplasia without atypia to microinvasive carcinoma. Various classification schemes for DCIS exist. Most pathologists and clinicians recognize two major subtypes of DCIS according to the presence or absence of comedo necrosis, but this is a matter of considerable debate. These two subtypes differ not only in pathologic features but also in clinical presentation, appearance on mammography, and malignant potential.

The comedo necrosis type of DCIS is diagnosed when at least one duct in the breast is filled and expanded by large, markedly atypical cells and has abundant central luminal necrosis (Figure 1). This necrotic material is usually partially calcified and thus may be recognized on mammography as linear and branching calcifications. Prominent periductal fibrosis is common and may render the lesion clinically palpable. The resulting distortion of the breast parenchyma may present the pathologist with the difficult and sometimes impossible task of excluding stromal invasion.

View larger version (139K): [in this window] [in a new window]   Figure 1. Comedo necrosis ductal carcinoma in situ. (Hematoxylin and eosin; magnification bar = 200 µm.).

All other forms of DCIS are of the noncomedo necrosis type and include the cribriform, micropapillary, and solid types (Figure 2). Many examples of the noncomedo necrosis type consist of a combination of the various histologic patterns. Although necrosis may be present, it is less prominent than in the comedo necrosis type and is not as prone to calcification. Just as the comedo necrosis type may overlap with microinvasive carcinoma, distinguishing small examples of the noncomedo necrosis type from atypical ductal hyperplasia may be difficult. However, if careful attention is paid to standardized diagnostic criteria, this distinction can be made with reasonable certainty [15-17].

View larger version (144K): [in this window] [in a new window]   Figure 2. Noncomedo necrosis ductal carcinoma in situ. (Hematoxylin and eosin.).

Many pathologists think that the comedo necrosis-noncomedo necrosis classification scheme is too simplistic and does not account for the marked heterogeneity of DCIS. This has led to several new classification schemes for DCIS [18-21]. All of these schemes take into account some or all of the following features: nuclear grade, amount and type of necrosis, growth pattern, tumor size, and architectural differentiation. The optimal classification scheme remains controversial.

Molecular Genetics and Cytogenetic Studies

Little information is available on chromosomal abnormalities in DCIS [22]. With the use of interphase cytogenetic techniques, 18 of 21 (86%) samples of the tumor were aneusomic for chromosome 1 [23]. In another study, aneuploidy, as determined by image analysis, was seen in 77.5% of cases of the tumor, mainly in the high- and intermediate-grade subtypes [24].

Specimens of DCIS have several alterations at the molecular genetic level. Overexpression of c-erbB-2 occurs in 46% to 60% of cases; the frequency is higher with the comedo necrosis type. In contrast, this overexpression is seen in only 20% to 25% of cases of invasive breast cancer [25-30]. At least some cases of invasive breast cancer that express c-erbB-2 seem to be derived from preexisting DCIS [31]. Maguire and colleagues [31] studied 12 specimens containing the noncomedo necrosis type and derivative invasive adenocarcinomas in the same section, all of which were negative for c-erbB-2 expression. However, of 13 comedo necrosis carcinomas in situ with an invasive component, 10 had definite c-erbB-2 expression that was similar in the in situ and invasive components in every case.

Mutations of the p53 gene are present in DCIS and are more common in the comedo necrosis type [32]. Allelic imbalance (loss of intensity of one allele) for the BRCA1 gene has been found in 74% of cases of the tumor [33]. Loss of heterozygosity has been described in the tumor at loci reported to show allelic loss in invasive breast cancers [34]. Identical loss of heterozygosity in the in situ component of tumors with invasive tumor cells has been reported on chromosome 11q13 [35], further supporting the idea that the tumor is preinvasive. Over-expression of cyclin D1, an oncogene on 11q13, is present in nearly 90% of malignant breast lesions (both DCIS and invasive breast cancer) [36].

Although some molecular genetic data support the hypothesis that invasive breast cancer can arise from preexisting in situ lesions, this has not been definitively proven. More work is needed to broaden our understanding of the relation between DCIS and invasive breast cancer.

Clinical Presentation

Ductal carcinoma in situ can present clinically as a palpable mass; Paget disease (which may be diagnosed independently); nipple discharge; or, most commonly, a nonpalpable abnormality on mammography [37]. It now accounts for 20% to 30% of all newly diagnosed cases of breast cancer in medical centers where screening is standard [38]. The most frequent radiologic finding is a collection of malignant-appearing or indeterminate calcifications [39], often with no other associated abnormalities. In our experience with nonpalpable lesions seen on mammography, 22% of indeterminate calcifications alone are associated with a histologic diagnosis of malignancy; 76% of these are minimal cancers, and almost all are DCIS [40]. Of all malignant-appearing calcifications, 92% are associated with a malignant histologic diagnosis. Fifty-eight percent of these are minimal cancers, and most are DCIS [40]. In addition, atypical features may include circumscribed nodules, ill-defined masses, ductal asymmetry, architectural distortion, and negative findings on mammography [41]. In a review of 100 mammographically detected cases of DCIS, 72% presented as calcifications alone and 12% presented as calcifications and an associated soft tissue abnormality [42]. Of note, microcalcifications on mammography usually underestimate the size of the tumor [43]. However, with state-of-the-art mammography (including magnification views), this discrepancy is less than 2 cm in up to 85% of cases [44].

Natural History

Much of what is known about the natural history of DCIS comes from studies of patients in whom the lesion was originally interpreted as benign but was found to be DCIS on subsequent analysis [45]. Page and colleagues [46] found 28 such cases among 11 760 women in whom benign lesions were originally diagnosed between 1959 and 1968. Most of these cases were low-grade noncomedo necrosis lesions with uncertain surgical margins. According to a follow-up report [47], ipsilateral invasive cancer later developed in the quadrant in which the original biopsy was done in 9 of the 28 women (32%). In similar studies, Rosen, Betsill, and colleagues [48, 49] reported the long-term outcome of 15 women in whom benign lesions were originally diagnosed. After a median of 9.7 years, 10 women had clinically apparent carcinoma in the ipsilateral breast; it was invasive in 8 of them. Eusebi and colleagues [50] also reported on 80 women in whom benign lesions were originally diagnosed (mean follow-up, 17.5 years). Invasive cancer subsequently developed in 11 women, and DCIS recurred in 5. Again, the recurrence was ipsilateral in most patients (11 of 16).

These studies showed that DCIS treated with excision alone is associated with a significant risk for subsequent cancer. Approximately 50% of these recurrences have an invasive component, and nearly all occur in the ipsilateral breast in the region of the initial biopsy.

Multicentricity

The earlier rationale for mastectomy as the preferred treatment for DCIS was derived from the concept that DCIS is a multicentric disease (that is, that the tumor cells arise independently in separate quadrants of the breast). Studies indicating multicentricity have reported an incidence as high as 32% [51-56]. However, most of these studies were done in mastectomy specimens from patients who had previously had biopsy for DCIS. Residual cancerous cells were considered as separate foci, and tumors were therefore called "multicentric."

As shown recently, most cases of DCIS are not multicentric. Radiologic and pathologic correlative studies of 82 mastectomy specimens from patients with DCIS, with subgross sectioning and extensive sampling, found only one multicentric lesion [43]. The tumor often recurs at or near the site of the original lesion [57]. This is a crucial factor in determining the feasibility of breast-conserving therapy as a treatment option because resection of the affected area must encompass the entire tumor. Current data do not support the concept of multicentric disease.

Treatment Options

Mastectomy

Mastectomy as primary treatment for DCIS is associated with near-total avoidance of recurrence (Table 1). Farrow [45] found only one case of recurrence among 181 patients with a minimum follow-up of 5 years. Ashikari and colleagues [59] followed 74 patients for up to 11 years and did not find any recurrences after mastectomy. However, with accurate information about the curative potential of breast-conserving therapy, patients less frequently have mastectomy as a front-line treatment [11]. Mastectomy was performed in 71% of cases of DCIS in 1983 but only 43.8% in 1992. In contrast, the use of breast-conserving therapy increased from 25.6% in 1983 to 53.3% in 1992. Currently, there is substantial geographic variation in the use of breast-conserving therapy and mastectomy [11, 66]. Recent and ongoing clinical trials in DCIS have not included mastectomy as standard treatment [67].

Axillary Lymph Node Dissection

The incidence of lymph node metastases in DCIS is negligible (<1%) [58, 65, 68]. Therefore, axillary lymph node dissection is not routinely indicated. Silverstein and colleagues [69] performed 200 axillary lymph node dissections for patients with DCIS without microinvasion; none of these patients had positive nodal disease.

The challenge for pathologists lies in identifying those lesions with microinvasion and thereby identifying the potential for metastasis. The risk for microinvasion appears to correlate with the size of the primary lesion. Lagios and colleagues [18] analyzed specimens for microinvasion by using intraoperative radiography and sequential sectioning at 2- to 3-mm intervals. The adequacy of the surgical excision was confirmed by postoperative mammography. In a series of 115 specimens, the overall rate of microinvasion was 14.7%. The incidence of microinvasion was 1.6% in lesions less than 25 mm and 29% in those larger than 26 mm [18]. Of 87 patients who had lymph node sampling, only 2 had nodal metastases. However, in both of these patients, the tumors were larger than 68 mm and had stromal invasion; thus, they did not represent pure DCIS. For this tumor, there is no definitive indication for axillary lymph node dissection [70]. Nevertheless, up to 50% of patients still undergo this procedure at the time of surgery [71].

Lumpectomy

The use of lumpectomy, or excision alone, as primary treatment for DCIS is associated with a high recurrence rate (mean, 20%; range, 8% to 63%), as reported in many prospective and retrospective studies (Table 2). The wide range of recurrence rates in these studies reflects different inclusion criteria, different rigor of pathologic assessment, and different follow-up times. From 1982 to 1987, Ottesen and colleagues [78] prospectively evaluated 112 women (93 with pure DCIS and 19 with a combination of ductal and lobular carcinoma in situ). The median patient age was 48 years, and the median follow-up period was 53 months. The women had simple excisional biopsy (except for those with lesions >5 mm, for which wider local excision was recommended). Of the 25 cases of recurrent cancer (22%), 5 were invasive (20%). In contrast to the results of a recent series, only 7 of the 112 cases were discovered on mammography. For a median of 44 months, Lagios and colleagues [18] followed 79 patients with mammographically detected DCIS lesions less than 25 mm (median, 6.8 mm) treated with lumpectomy alone from 1975 to 1987. There were eight recurrences (10%), four of which were cases of invasive cancer. The data from the National Surgical Adjuvant Breast and Bowel Project (NSABP)-B17 showed that at a mean of 43 months after lumpectomy alone (in 391 patients), the chance of recurrence was 16% and half of the recurrent cases were cases of invasive breast cancer.

Breast-Conserving Therapy

One of the most important and controversial questions about the appropriateness of lumpectomy for DCIS is whether radiotherapy is needed after excision (breast-conserving therapy). Several studies have addressed the results of radiotherapy after lumpectomy (Table 3).

The NSABP-B6 was a randomized trial designed to compare lumpectomy, breast-conserving therapy, and mastectomy in patients with early invasive breast cancer [76]. A subset of 76 patients was found to have DCIS on subsequent pathologic review [57]. The lesions were larger than 10 mm in 84% of cases, and moderate or marked comedo necrosis was present in 60% of specimens. Twentyseven patients received breast-conserving therapy, 21 had excision alone, and 28 were treated with mastectomy. Two patients (7%) who received breast-conserving therapy and 9 (43%) who had excision alone had recurrence in the ipsilateral breast. Of the 11 recurrences, 6 were cases of invasive breast cancer. The average time to in-breast recurrence was 35 months. The recurrent tumors were morphologically similar to the original ones. On multivariate analysis, the only feature associated with a decreased risk for recurrence was the use of radiotherapy.

Solin and colleagues [83] studied 268 women with 270 DCIS lesions who received breast-conserving therapy. After a median follow-up of 10.3 years, 45 in-breast recurrences (16.7%) were detected. The 15-year actuarial overall survival rate was 87%. It is encouraging that the actuarial rate of freedom from metastases was 96% at 15 years. The median time to local failure was 5.2 years; relapses were detected up to 16.8 years after initial diagnosis. The histologic diagnosis at the time of recurrence included invasive breast cancer in 53% of cases. Median times to recurrence were 6.7 years for invasive breast cancer and 3.6 years for recurrent DCIS.

The NSABP-B17 is the only published randomized trial specifically designed to compare lumpectomy alone with breast-conserving therapy for DCIS [67]. In this trial, 818 women with mammographically or clinically detected tumors were studied. Eligible patients were to have negative margins, but subsequent pathologic analysis found that this was not always true [84]. Seventy-three percent of the lesions were smaller than 1 cm, and 43% could not be measured grossly (that is, they were < 0.1 cm). After a mean follow-up of 43 months, the 5-year event-free survival rate was significantly better in the breast-conserving therapy arm than in the lumpectomy arm (84.4% compared with 73.8%; P = 0.001). The event-free survival rate was estimated by using the presence of ipsilateral or contralateral breast cancer, regional or distant metastases, a second primary tumor other than a breast tumor occurring after surgery, or death without recurrent disease as an event. The use of radiotherapy significantly decreased the rate of in-breast recurrences from 16.4% to 7%. Significant reductions were seen in cases of invasive (10.5% to 2.9%; P < 0.001) and noninvasive recurrent cancer (10.4% to 7.5%; P = 0.055) (Table 4). The difference between groups with respect to events other than ipsilateral inbreast recurrence was not significant. This trial has been criticized for several possible flaws, including lack of universal pathologic review, possible inclusion of patients with atypical ductal hyperplasia, lack of thorough examination of each sample, and short follow-up [85].

Survival

No data from randomized trials compare survival after breast-conserving therapy with that after mastectomy for DCIS. Using a Markov decision analysis, Hillner and colleagues [86] postulated that if there is a difference in overall survival, it is minimal (0.7%). This translates into an absolute improvement of 10 to 27 days at 10 years and 74 to 202 days at 20 years with mastectomy. The long-term survival data for breast-conserving therapy were projected from the preliminary results of NSABP-B17 at 5 years. However, this model assumes 100% efficacy for the treatment of the disease with mastectomy, an assumption that is incorrect (Table 1). In the subset of patients enrolled in NSABP-B06 who were later found to have DCIS, one disease-related death occurred among those who had had mastectomy (n = 28) and no deaths occurred among those who had had breast-conserving therapy (n = 27) [57].

Silverstein and colleagues [60] retrospectively evaluated 98 women with DCIS who had mastectomy and 103 who received breast-conserving therapy. Although mastectomy had a better actuarial 7-year disease-free survival rate, no difference in overall survival was seen. It could be argued that this difference may become significant with extended follow-up. In the SEER study, in which half of the women had mastectomy and half had lumpectomy or breast-conserving therapy, the survival rates were the same or better than those expected for the general population [11]. Thus, although we do not have data from randomized trials showing equivalent survival rates, existing data with similar survival figures for both treatments support the use of breast-conserving therapy unless this therapy is contraindicated or the patient prefers mastectomy.

Risk Factors for Recurrence

The presence of comedo necrosis has been associated with a higher rate of recurrence after initial treatment of DCIS [21, 67, 79, 82]. In Lagios and associates' [18] series of 79 patients treated with lumpectomy, recurrence rates varied according to histologic subtypes. Patients with the micropapillary subtype had no recurrences (0 of 33), whereas patients with evidence of comedo necrosis had a 16% recurrence rate (5 of 31) and patients with the cribriform pattern with necrosis had a 40% recurrence rate (2 of 5).

The presence of tumor at the surgical margin leads to a higher rate of relapse [61, 87]. However, there is no consensus about what represents negative margins, and this criterion has varied markedly among studies. In the original design of the NSABP-B17 protocol, patients with DCIS had to have negative margins for inclusion. However, pathologic review of 573 of 790 specimens found that the margins were either involved or uncertain in 101 cases (17.6%) [84]. This analysis showed that the presence of negative margins in combination with the absence of comedo necrosis correlated with excellent tumor control (Table 5). In a subset analysis of 125 such patients who had lumpectomy alone, the average annual rate of in-breast recurrence was 1.97 per 100 patients. In contrast, in 25 women with questionable or positive margins and comedo necrosis who were treated with lumpectomy alone, the in-breast annual recurrence rate was 10.46 per 100 patients. It is important to note that this was a retrospective analysis of patient subsets, and the presence of comedo necrosis and the presence of negative margins were not stratification factors in the protocol.

Length of follow-up is an important factor in judging these studies. For example, Solin and colleagues [83], in a study of DCIS with longer follow-up (10.3 years), showed that the risks for recurrence at 5 years with comedo necrosis and noncomedo necrosis histology were 11% and 2%, respectively. At 10 years, the risks were 17% for comedo necrosis histology and 15% for noncomedo necrosis histology (P > 0.2). This study suggests that lesions with comedo necrosis histology and high nuclear grade recur earlier; therefore, extended follow-up is needed to document recurrences in cases with a "favorable" histologic subtype.

Van Nuys Prognostic Index

A recently proposed prognostication index, the Van Nuys Prognostic Index, considers size of the lesion; status of the surgical margins; and pathologic features of aggressiveness, such as the presence of comedo necrosis and high-grade tumors [19, 88, 89]. Three hundred thirty-three lesions were evaluated retrospectively for tumor size, margin width, and grade. The treatment had been lumpectomy in 195 patients and breast-conserving therapy in 138. Lesions were scored according to these features, and the resulting score was correlated with recurrence rates (Table 6). The Van Nuys Prognostic Index scoring categories are low (3 to 4), intermediate (5 to 7), or high (8 to 9). According to the Van Nuys Prognostic Index, the local recurrence rate did not differ for lumpectomy and breast-conserving therapy in patients with low scores (3 or 4). The authors concluded that for patients with high scores (8 or 9), the risk for recurrence is so high after breast-conserving therapy that mastectomy is justified.

As noted earlier, other authors have suggested that excision alone can be used in patients with favorable features, such as small lesion, absence of comedo necrosis, and negative margins [88-91]. However, in the subset of patients in NSABP-B17 with negative margins and noncomedo necrosis, the annual recurrence rate was 1.97 per 100 patients with lumpectomy alone and 1.18 per 100 patients with breast-conserving therapy (median follow-up, 43 months) (Table 5).

The widespread attention paid to this report demonstrates the clinical need for an accurate and reliable prognostication system for DCIS [92]. However, the Van Nuys Prognostic Index was developed in a retrospective study and must be validated prospectively before it can serve as the basis for treatment recommendations. The editorial accompanying the Van Nuys Prognostic Index report [92] thoroughly outlined difficulties in the development and use of such a prognostic model.

Treatment after Recurrence

Several studies have reported the outcome of patients with tumor recurrence after primary surgical treatment [61, 72, 79, 81]. Solin and colleagues [93] reported 42 recurrences in patients with DCIS who received breast-conserving therapy: Twenty-three had invasive carcinoma, and 19 had DCIS. Mastectomy was used for local salvage treatment in 40 patients, and excision alone was used in 2. Two patients received adjuvant chemotherapy, tamoxifen treatment was started in 8 patients, and both adjuvant chemotherapy and tamoxifen were given to 1 patient. Thirty-two patients did not receive systemic therapy. The overall survival rate at 5 years was 78%, and the 5-year cause-specific survival rate was 84% [93].

Graham and colleagues [73] reported no cancer-related deaths among 14 patients who had recurrence after primary treatment for DCIS, 7 of whom had invasive breast cancer. All patients (n = 53) had had surgery as primary treatment (37 had lumpectomy alone). The salvage treatment was breast-conserving therapy in 6 patients, mastectomy in 4 patients, lumpectomy in 2 patients, and radiotherapy alone in 2 patients. Price and colleagues [75] followed 60 patients who had 26 recurrences after treatment. Eight patients had local excision alone, 10 received breast-conserving therapy, 7 had mastectomy, and 1 received radiotherapy alone. Distant metastases developed in 2 patients with and 2 patients without local recurrence.

In summary, of the recurrences that occurred after primary treatment for DCIS in the aforementioned studies, nearly half were cases of invasive cancer. Although no consensus exists [93, 94], most authors recommend mastectomy for patients with recurrence if breast-conserving therapy was the initial treatment. On the basis of the available information, we conclude that selected patients may also undergo breast-conserving therapy at the time of relapse if lumpectomy alone was used initially. With regard to the use of systemic therapy in patients with an invasive recurrence, we use standard prognostic factors (nodal and hormonal receptor status and tumor size) to predict the risk for distant micrometastatic disease and make our treatment recommendations accordingly.

Hormonal Therapy

Expression of estrogen and progesterone receptors in DCIS has been studied by several investigators. Of 151 specimens evaluated immunohistochemically by Poller and colleagues [95], 31% were positive for estrogen receptor. In this and other studies, estrogen receptor positivity was associated with the noncomedo necrosis type [96], small-cell type, and lack of c-erbB-2 overexpression [95]. Pallis and colleagues [97] showed that 80% of the comedo necrosis tumors were negative for estrogen receptor. Estrogen receptor staining provides no additional clinical information.

In the clinical setting, two studies are currently evaluating the role of tamoxifen in the management of DCIS. The NSABP-B24 is comparing tamoxifen with placebo after breast-conserving therapy, and an ongoing European study (UKCCCR-DCIS Phase III-90001) is comparing tamoxifen alone with tamoxifen and radiotherapy [98].

Conclusions

The incidence of DCIS is increasing, and, because of screening mammography, the disease is diagnosed in an increasing proportion of asymptomatic patients. The most common clinical presentation is that of calcifications on mammography. Ductal carcinoma in situ is a heterogeneous pathologic condition. With the use of more refined molecular analysis, the relation of DCIS to invasive breast cancer will be better defined. If left untreated, a proportion of these tumors will evolve into cases of invasive breast cancer. Convincing evidence suggests that most cases of DCIS are not multicentric.

The goal in the treatment of patients with DCIS is to control local disease and prevent subsequent development of invasive cancer. Mastectomy is associated with a risk for chest wall recurrence of approximately 1%. Axillary lymph node dissection is not routinely recommended in the management of DCIS. No randomized trials have compared mastectomy with breast-conserving therapy for the treatment of this tumor.

According to the results of NSABP-B17, breast-conserving therapy is an effective option in the management of DCIS. The use of radiotherapy after lumpectomy significantly decreases the rate of recurrence. No prospective information supports the conclusion that the use of radiotherapy after lumpectomy can be omitted in patients with good prognostic features. The presence of comedo necrosis and surgical margin involvement are the most commonly used predictors of the likelihood of recurrence. If relapse occurs after breast-conserving therapy, the chance of an invasive component is approximately 50%. The risk for in-breast recurrence at 5 years after breast-conserving therapy is approximately 8%. In NSABP-B17, the mortality rate due to breast cancer in patients who had breast-conserving therapy was between 0.3% and 0.5%. Most patients with recurrent disease can be treated effectively. There is no defined role for hormonal therapy.

Dr. Petersen: Division of Radiation Oncology, Mayo Clinic and Mayo Foundation, 200 First Street SW, Rochester, MN 55905.

Dr. Donohue: Department of Surgery, Mayo Clinic and Mayo Foundation, 200 First Street SW, Rochester, MN 55905.

Dr. Crotty: Division of Anatomic Pathology, Mayo Clinic and Mayo Foundation, 200 First Street SW, Rochester, MN 55905.

Dr. Gisvold: Department of Diagnostic Radiology, Mayo Clinic and Mayo Foundation, 200 First Street SW, Rochester, MN 55905.