Which DCIS lesions will become invasive and when will that happen? These are the most important questions in the DCIS field today. Currently, there is intense molecular biologic study regarding the progression of genetic changes in normal breast epithelium to DCIS and then to invasive breast cancer. Most of the genetic and epigenetic changes present in invasive breast cancer are already present in DCIS. To date, no genes uniquely associated with invasive cancer have been identified. As DCIS progresses to invasive breast cancer, quantitative changes in the expression of genes related to angiogenesis, adhesion, cell motility, and the composition of the extracellular-matrix may occur [2, 16].

It has been recognized for some time that there is a substantial concordance between the grade of a DCIS and its associated invasive carcinoma, such that low grade lesions, regardless of the classification scheme used, are largely associated with lower grade invasive carcinomas while high grade DCIS are associated with higher grade invasive carcinomas. Additionally the frequency of specific biomarkers in DCIS varies with the grade of the lesion: estrogen and progesterone receptors are usually expressed in DCIS, but less so in high grade DCIS, while HER-2 and elevated proliferative markers such as Ki-67 are features of high grade DCIS. More recently surrogate molecular phenotypes defined by immunohistochemistry have identified DCIS phenotypes corresponding to luminal A, luminal B, Her-2 and triple negative/basal phenotypes in invasive breast cancer. Luminal A and B DCIS phenotypes are more frequent in the low = intermediate grade lesions, where HER-2, triple negative and basal phenotypes are more common among high grade DCIS [47, 48].

Currently gene expression profiling for invasive carcinomas can segregate carcinomas at significantly different risks of distant relapse (10 years) and indicate the likelihood of a benefit from chemotherapy. Hannemann et al. were able to segregate DCIS with and without invasive carcinomas by a group of 35 genes, and stratify DCIS by another 43 gene classifer into well differentiated vs. poorly differentiated (high grade) subtypes [49]. Additionally these authors confirmed the association of ER expression in low-intermediate DCIS and HER-2 in high grade DCIS.

Solin et al. presented a new DCIS gene signature RT-PCR assay which can separate out three risk groups based on their gene expression profiles [50]. The genes were selected from the existing Oncotype DX assay and the ECOG 5194 registration trial archived tissue was used for validation. The ECOG trial examined the frequencies of local recurrences in DCIS patients treated by excision alone with minimum 3 mm margins. Low to intermediate grade DCIS had to be 25 mm or less in greatest extent, while high grade DCIS could not exceed 10 mm.

All patients entered into the trial had to have specimens examined by a serial sequential tissue protocol which permits reproducible determination of size (extent), margin widths and excludes microinvasive foci. Cases which did not meet these criteria were excluded. Specimens examined by less rigorous means cannot be reliably evaluated by the assay, e.g., the DCIS may exceed the size limit or exhibit margins which are suboptimal and the RS reported may not be valid for the case submitted. By way of contrast two recent nomograms designed to predict recurrence following breast conservation therapy for DCIS were based on archival materials which were not processed in the serial sequential method and for which data on size and margins was frequently missing. Kerlikowski et al. evaluated specimens which were generated from many institutions without a common protocol, and without accurately determined size or margin status [51]. Rudloff et al. developed a nomogram which was obtained from a single institution but without a single protocol and not examined by the serial subgross technique [52].

The ECOG study dichotomized DCIS into low intermediate vs. high grade and only required a 3 mm margin. It will remain to be seen whether reanalysis of the pathology data will permit greater separation of subsets of DCIS comparable to what can be achieved in the USC/VNPI.

Because most patients with DCIS had been treated with mastectomy, knowledge of the natural history of this disease is relatively recent. The studies of Page et al. [53, 54] and Rosen et al. [55] provide information regarding the outcome of low grade DCIS without treatment. In these studies, patients with noncomedo DCIS, the majority, incidental foci in diagnostic excision biopsies for palpable disease, were initially misdiagnosed as benign lesions and therefore went untreated, aside from initial biopsy. Subsequently, approximately 25–35 % of these patients developed invasive breast cancer, generally within 10 years [53, 54]. Had the lesions been high-grade comedo DCIS, the invasive breast cancer rate likely would have been higher than 35 % and the time to invasive recurrence shorter. With few exceptions, in both of these studies, the invasive breast carcinoma was of the ductal type and located at the site of the original DCIS. These findings suggest that not all DCIS lesions progress to invasive breast cancer or become clinically significant [56]

Page and associates updated their series in 2002 and 2005 [53, 54, 57]. Of 28 women with low-grade DCIS misdiagnosed as benign lesions and treated only by biopsy between 1950 and 1968, 11 patients have recurred locally with invasive breast cancer (39 %). Eight patients developed recurrence within the first 12 years. The remaining three were diagnosed over 23–42 years. Five patients developed metastatic breast cancer (18 %) and died from the disease within 7 years of developing invasive breast cancer. These recurrence and mortality rates, at first glance, seem alarmingly high. However, they are only slightly worse than what can be expected with long-term follow-up of patients with lobular carcinoma in situ, a disease that most clinicians are willing to treat with careful clinical follow-up. In addition, these patients were treated with biopsy only. No attempt was made to excise these lesions with a clear surgical margin. The natural history of low-grade DCIS can extend over 40 years and is markedly different from that of high-grade DCIS.

The incidence of microinvasion is difficult to quantitate because until 1997 there was no formal and universally accepted definition of exactly what constitutes microinvasion. The fifth edition of The Manual for Cancer Staging carried the first official definition of what is now classified as pT1mic and read as follows: “Microinvasion is the extension of cancer cells beyond the basement membrane into adjacent tissues with no focus more than 0.1 cm in greatest dimension. When there are multiple foci of microinvasion the size of only the largest focus is used to classify the microinvasion (do not use the sum of all individual foci). The presence of multiple foci of microinvasion should be noted, as it is with multiple larger invasive carcinomas.”

The reported incidence of occult invasion (invasive disease at mastectomy in patients with a biopsy diagnosis of DCIS) varies greatly, ranging from as little as 2 % to as much as 21 % [58]. This problem was addressed in the investigations of Lagios et al. [32, 38].

Lagios et al. performed a meticulous serial subgross dissection correlated with specimen radiography. Occult invasion was found in 13 of 111 mastectomy specimens from patients who had initially undergone excisional biopsy of DCIS. All occult invasive cancers were associated with DCIS greater than 45 mm in extent; the incidence of occult invasion approached 50 % for DCIS greater than 55 mm. In the study of Gump et al. [59], foci of occult invasion were found in 11 % of patients with palpable DCIS but in no patients with clinically occult DCIS. These results suggest a correlation between the size of the DCIS lesion and the incidence of occult invasion. Clearly, as the size of the DCIS lesion increases, microinvasion and occult invasion become more likely.

If even the smallest amount of invasion is found, the lesion should not be classified as DCIS. It is a T1mic (if the largest invasive component is 1 mm or less) with an extensive intraductal component (EIC). If the invasive component is 1.1 mm to 5 mm, it is a T1a lesion with EIC. If there is only a single focus of invasion, these patients do quite well. When there are many tiny foci of invasion, these patients have a poorer prognosis than expected [10]. Unfortunately, the TNM staging system does not have a T category that fully reflects the malignant potential of lesions with multiple foci of invasion since they are all classified by the largest single focus of invasion. De Mascarel et al. have noted that microinvasive foci consisting of single cells have no impact on outcome while those comprising cohesive groups of cells are associated with a demonstrable increase in distant recurrence and death [60].

Multicentricity is generally defined as DCIS in a quadrant other than the quadrant in which the original DCIS (index quadrant) was diagnosed. There must be normal breast tissue separating the two foci. However, definitions of multicentricity vary among investigators. Hence, the reported incidence of multicentricity also varies. Rates from 0 to 78 % [7, 55, 61, 62], averaging about 30 %, have been reported. Twenty-five years ago, the 30 % average rate of multicentricity was used by surgeons as the rationale for mastectomy in patients with DCIS.

Holland et al. [63] evaluated 82 mastectomy specimens by taking a whole-organ section every 5 mm. Each section was radiographed. Paraffin blocks were made from every radiographically suspicious spot. In addition, an average of 25 blocks were taken from the quadrant containing the index cancer; random samples were taken from all other quadrants, the central subareolar area, and the nipple. The microscopic extension of each lesion was verified on the radiographs. This technique permitted a three-dimensional reconstruction of each lesion. This study demonstrated that most DCIS lesions were larger than expected (50 % were greater than 50 mm), involved more than one quadrant by continuous extension (23 %), but most importantly, were unicentric (98.8 %). Only one of 82 mastectomy specimens (1.2 %) had “true” multicentric distribution with a separate lesion in a different quadrant. This study suggests that complete excision of a DCIS lesion is possible due to unicentricity but may be extremely difficult due to larger than expected size. In a recent update, Holland reported whole-organ studies in 119 patients, 118 of whom had unicentric disease [64]. This information, when combined with the fact that most local recurrences are at or near the original DCIS, suggests that the problem of multicentricity per se is not important in the DCIS treatment decision-making process.

Multifocality is defined as separate foci of DCIS within the same ductal system. The studies of Holland et al. [63, 64] and Noguchi et al. [65] suggest that a great deal of multifocality may be artifactual, resulting from looking at a three-dimensional arborizing entity in two dimensions on a glass slide. It would be analogous to saying that the branches of a tree were not connected if the branches were cut at one plane, placed separately on a slide, and viewed in cross-section [53]. Multifocality may be due to small gaps of DCIS or skip areas within ducts as described by Faverly et al. [43]. Multifocality is more easity recognized when a serial sequential tissue processing technique as opposed to random sampling is employed.

The importance of quality mammography cannot be overemphasized. Currently, most patients with DCIS (more than 90 %) present with a nonpalpable lesion detected by mammography. The most common mammographic finding is microcalcification, frequently clustered and generally without an associated soft-tissue abnormality. More than 80 % of DCIS patients exhibit microcalcifications on preoperative mammography. The patterns of these microcalcifications may be focal, diffuse, or ductal, with variable size and shape. Patients with comedo DCIS tend to have “casting calcifications.” These are linear, branching, and bizarre and are almost pathognomonic for comedo DCIS [66] (Fig. 15.3). It is important to note that many DCIS with prominent comedonecrosis fail to exhibit mammographic microcalcifications, and among others, microcalcification is seen only intermittently.

Thirty-two percent of noncomedo lesions in this series did not have mammographic calcifications, making them more difficult to find and the patients more difficult to follow, if treated conservatively. When noncomedo lesions are calcified, they tend to have fine granular powdery calcifications or crushed stone-like calcifications (Fig. 15.4).

A major problem confronting surgeons relates to the fact that calcifications do not always map out the entire DCIS lesion, particularly those of the noncomedo type. Even though all the calcifications are removed, in some cases, noncalcified DCIS may be left behind. Conversely, in some patients, the majority of the calcifications are benign and map out a lesion bigger than the true DCIS lesion. In other words, the DCIS lesion may be smaller, larger, or the same size as the calcifications that lead to its identification. Calcifications more accurately approximate the size of high-grade comedo lesions than low-grade noncomedo lesions [63].

Before mammography was common or of good quality, most DCIS was usually clinically apparent, diagnosed by palpation or inspection; it was gross disease. Gump et al. divided DCIS by method of diagnosis into gross and microscopic disease [59]. Similarly, Schwartz et al. divided DCIS into two groups: clinical and subclinical [33]. Both researchers thought patients presenting with a palpable mass, a nipple discharge, or Paget’s disease of the nipple required more aggressive treatment. Schwartz believed that palpable DCIS should be treated as though it were an invasive lesion. He suggested that the pathologist simply has not found the area of invasion. Although it makes perfect sense to believe that the change from nonpalpable to palpable disease is a poor prognostic sign, our group has not been able to demonstrate this for DCIS. In our series, when equivalent patients (by size and nuclear grade) with palpable and nonpalpable DCIS were compared, they did not differ in the rate of local recurrence or mortality.

If a patient’s mammogram shows an abnormality, most likely, it will be microcalcifications, but it could be a nonpalpable mass or an architectural distortion. At this point, additional radiologic workup needs to be performed. This generally includes compression mammography and magnification views. Ultrasonography should be performed on all calcifications that require biopsy to rule out the presence of a mass that can be biopsied with ultrasound guidance. Magnetic resonance imaging (MRI) has become increasingly popular and if often used to map out the size and shape of biopsy proven DCIS lesions or invasive breast cancers and to rule out other foci of multifocal, multicentric or contralateral cancer.

For most patients with DCIS, there will be no single correct treatment. There will generally be a choice. The choices, although seemingly simple, are not. As the choices increase and become more complicated, frustration increases for both the patient and her physician [71, 72].

When a patient with DCIS, previously treated by any modality, develops a local invasive recurrence, followed by distant disease and death due to breast cancer, this stepwise progression makes sense. The patient has been upstaged by her local invasive recurrence. The invasive recurrence becomes the source of metastatic disease and death is now a possibility.

In contrast, when a previously treated patient with DCIS develops distant disease and there has been no invasive local recurrence, a completely different sequence of events must be postulated. This sequence implies that invasive disease was present within the original lesion but was never discovered and was already metastatic at the time of the original diagnosis. The best way to avoid missing an invasive cancer is with complete sequential tissue processing at the time the original lesion is treated. Nevertheless, even the most extensive evaluation may miss a small focus of invasion.

If, during histopathologic evaluation, even the tiniest invasive component is found, this patient can no longer be classified as having DCIS. She has invasive breast cancer and she needs to be treated as such. She will need sentinel node biopsy, radiation therapy if treated conservatively, and appropriate medical oncologic consultation and aftercare.

All of the prospective randomized trials have shown a significant reduction in local recurrence for patients treated with radiation therapy compared with excision alone but no trial has reported a survival benefit, regardless of treatment [73, 76, 77, 85–88, 106, 108–111].

Only one trial has compared mastectomy with breast conservation for patients with DCIS and the data were only incidentally accrued. The National Surgical Adjuvant Breast Project (NSABP) performed protocol B-06, a prospective randomized trial for patients with invasive breast cancer [61, 112]. There were three treatment arms: total mastectomy, excision of the tumor plus radiation therapy, and excision alone. Axillary nodes were removed regardless of the treatment assignment.

During central slide review, a subgroup of 78 patients was confirmed to have pure DCIS without any evidence of invasion [61]. After 83 months of follow-up, the percent of patients with local recurrences were as follows: zero for mastectomy, 7 % for excision plus radiation therapy, and 43 % for excision alone [113]. In spite of these large differences in the rate of local recurrence for each different treatment, there was no difference among the three treatment groups in breast cancer-specific survival.

Contrary to the lack of trials comparing mastectomy with breast conservation, a number of prospective randomized trials comparing excision plus radiation therapy with excision alone for patients with DCIS have been performed: the NSABP (protocol B-17) [85], the European Organization for Research and Treatment of Cancer (EORTC), protocol 10,853 [88], the United Kingdom, Australia, New Zealand DCIS Trial (UK Trial) [76], and the Swedish Trial [78].

The results of NSABP B-17 were updated in 1995 [109], 1998; [87], 1999 [86], 2001 [73] and 2011 [106]. In this study, more than 800 patients with DCIS excised with clear surgical margins were randomized into two groups: excision alone versus excision plus radiation therapy. The main endpoint of the study was local recurrence, invasive or noninvasive (DCIS). The definition of a clear margin was non-transection of the DCIS.

After 15 years of follow-up, there was a statistically significant, 50 % decrease in local recurrence of both DCIS and invasive breast cancer in patients treated with radiation therapy. The overall local recurrence rate for patients treated by excision alone was 35 % at 15 years. For patients treated with excision plus breast irradiation, it was 19.8 %, a relative benefit of 43 % [106]. There was no difference in distant disease-free or overall survival in either arm. These data led the NSABP to confirm their 1993 position and to continue to recommend postoperative radiation therapy for all patients with DCIS who chose to save their breasts. This recommendation was clearly based primarily on the decreased local recurrence rate for those treated with radiation therapy and secondarily on the potential survival advantage it might confer.

The early results of B-17, in favor of radiation therapy for patients with DCIS, led the NSABP to perform protocol B-24 [86]. In this trial, more than 1,800 patients with DCIS were treated with excision and radiation therapy, and then randomized to receive either tamoxifen or placebo. After 15 years of follow-up, 16.6 % of patients treated with placebo had recurred locally, whereas, only 13.2 % of those treated with tamoxifen had recurred [106]. The difference, while small, was statistically significant for invasive local recurrence but not for noninvasive (DCIS) recurrence. A recent analysis of the subset of B24 patients known to be estrogen receptor-positive has shown no significant difference in ipsilateral in situ or invasive recurrence [114]. Again, there was no difference in distant disease-free or overall survival in either arm of the B-24 Trial.

The EORTC results were published in 2000 [88, 108]. This study was essentially identical to B-17 in design and margin definition. More than 1,000 patients were included. The data were updated in 2006 [110]. After 10 years of follow-up, 15 % of patients treated with excision plus radiation therapy had recurred locally compared with 26 % of patients treated with excision alone, results similar to those obtained by the NSABP at the same point in their trial. As in the B-17 Trial, there was no difference in distant disease-free or overall survival in either arm of the EORTC Trial. In the initial report, there was a statistically significant increase in contralateral breast cancer in patients who were randomized to receive radiation therapy. This was not maintained when the data were updated.

The United Kingdom, Australia, New Zealand DCIS Trial (UK Trial) was published in 2003 [76] and updated in 2011 [111]. This trial, which involved more than 1,694 patients, performed a two by two study in which patients could be randomized into two separate trials within a trial. The patients and their doctors chose whether to be randomized in one or both studies. After excision with clear margins (same non-transection definition as the NSABP), patients were randomized to receive radiotherapy (yes or no) and/or to tamoxifen versus placebo. This yielded four subgroups: excision alone, excision plus radiation therapy, excision plus tamoxifen, and excision plus radiation therapy plus tamoxifen. After a median follow-up of 12.7 years, those who received radiation therapy obtained a statistically significant decrease in ipsilateral breast tumor recurrence similar in magnitude to the ones shown by the NSABP and EORTC. Tamoxifen significantly reduced the incidence of ipsilateral DCIS recurrences but not invasive recurrences. It reduced the incidence of new contralateral breast cancers in a magnitude similar to NSABP B-24. As with the NSABP and the EORTC, there was no difference in survival, regardless of treatment, in any arm of the UK DCIS trial.

The Swedish DCIS Trial randomized 1,067 patients into two groups: excision alone versus excision plus radiation therapy. 1,046 have been followed for a mean of 8 years. Microscopically clear margins were not mandatory. 22 % of patients had microscopically unknown or involved margins. The cumulative incidence of local recurrence at 10-years was 21.6 % for excision only and 10.3 % for excision plus radiation therapy. There were 15 distant metastases and breast cancer related deaths in the excision only arm and 18 in the excision plus radiation therapy (p = ns) [78, 115].

All of these trials support the same conclusions. They all show that radiation therapy decreases local recurrence by a relative 50 % and they all show no survival benefit, regardless of treatment. Two trials show a decrease in local recurrence and contralateral breast cancer attributable to tamoxifen.

In 2007, Viani et al. published a meta-analysis of the four prospective randomized DCIS trials comparing excision alone with excision plus radiation therapy [77]. Three thousand six hundred and sixty-five patients were available for analysis. Pooled data revealed a 60 % reduction of both invasive and DCIS recurrences (p < 0.00001) with the addition of radiation therapy. There was, however, no decrease in distant metastases in those who received radiation therapy nor was there any survival benefit. Patients with high-grade lesions and involved margins received the most benefit from radiation therapy.

The randomized trials were designed to answer a single broad question: does radiation therapy decrease local recurrence? They have accomplished that goal. All have clearly shown that, overall, radiation therapy decreases local recurrence, but they cannot identify in which subgroups the benefit is so small, that the patients can be safely treated with excision alone.