Recent Insights Into the Unique Biology of Ductal Carcinoma in Situ and Lobular Carcinoma in Situ

In many respects, DCIS should be regarded as carcinoma of the ductal system, for it possesses all of the molecular and biological abnormalities as frankly invasive carcinoma. DCIS is clonal and frequently expresses abnormal p53 and HER2/neu. It has the same loss of heterozygosity patterns as its invasive counterpart. Using comparative genomic hybridization studies, DCIS exhibits no gains or losses of chromosomal regions compared with its invasive counterpart. Rather, DCIS is held in check by a surrounding layer of myoepithelial cells that exert paracrine suppressive effects on invasion. Because of this, the histopathologic patterns of DCIS usually strongly correlate with its invasive counterpart, when present in individual cases. For example, papillary DCIS ( Fig. 9.1 ) tends to invade as papillary adenocarcinoma. Cribriform DCIS ( Fig. 9.2 ) and solid DCIS ( Fig. 9.3 ) of intermediate nuclear grades tend to invade as a moderately differentiated adenocarcinoma. Comedo DCIS of high nuclear grade ( Fig. 9.4 ) tends to invade as a poorly differentiated adenocarcinoma. The precursor lesion of DCIS is thought to be atypical ductal hyperplasia (ADH) for the DCIS of low or intermediate nuclear grades. Simple ductal hyperplasia, based on loss of heterozygosity and genomic hybridization studies, is no longer thought to be a precursor lesion of either ADH or DCIS. A recently described lesion, flat epithelial atypia or columnar cell atypia ( Figs. 9.5 and 9.6 ), based on its abnormal nuclear features and its presence juxtaposed to DCIS, especially high-grade DCIS, is thought to be a possible precursor lesion. However, this has not yet been proved in prospective studies. The evidence is incontrovertible that DCIS can and often does progress to frank invasive adenocarcinoma, and clearly, the same clone is involved.

The micropapillary type of noncomedo ductal carcinoma in situ is characterized by intraductal papillations projecting into a central lumen.

The cribriform type of noncomedo ductal carcinoma in situ, in contrast, consists of intraductal proliferations showing bridging and “Roman arch” formation.

The solid type of ductal carcinoma in situ exhibits a solid sheet of intraluminal cells.

The comedo type of ductal carcinoma in situ exhibits a solid pattern of intraluminal proliferation with high nuclear grade and central necrosis.

The lesion of flat columnar epithelial atypia is characterized by dilated ducts lined by a single luminal cell layer.

At higher magnification, the single luminal cell layer is columnar in shape and manifests marked nuclear atypia.

With LCIS, the story is both similar and different. Emerging evidence has suggested that LCIS is a heterogeneous disease. Some types of LCIS are associated with a 4- to 10-fold increased risk of invasive breast carcinoma ( Fig. 9.7 ). The increased risk can be associated with any type of infiltrating breast carcinoma, including both ductal as well as lobular carcinoma. Other forms of LCIS may be more innocuous. Still others, for example, those that express more nuclear pleomorphism such as pleomorphic LCIS ( Fig. 9.8 ), may actually progress to invasive lobular carcinoma ( Fig. 9.9 ). This latter type of LCIS resembles DCIS in its biology. Still other types of LCIS can mimic other features of DCIS. Some LCIS spreads laterally through the ductal system analogous to Paget disease ( Fig. 9.10 ). This spread of LCIS is aptly termed pagetoid spread . Preliminary molecular studies of LCIS by loss of heterozygosity and genomic hybridization confirm the molecular heterogeneity of LCIS. Some LCIS has few, if any, obvious chromosomal abnormalities. Other types of LCIS exhibit evidence of genomic instability. Still other types of LCIS exhibit the same clonal abnormalities as its invasive lobular counterpart. This molecular heterogeneity suggests that some forms of LCIS may be innocuous, others may confer increased risk for the development of breast cancer, and still others can directly progress to invasive lobular breast cancer. This latter type of LCIS is therefore analogous to DCIS. Clearly, the appropriate therapy would depend on the type of LCIS present. So-called innocuous LCIS could be treated by “watchful waiting,” genomically unstable LCIS with an increased risk of breast cancer could be treated with tamoxifen, and the LCIS that directly progresses to invasive carcinoma could be treated with surgical extirpation. Prospective randomized studies and not just historical controls are needed to resolve the LCIS question.

Lobular carcinoma in situ of the classic type fills and distends acini.

Lobular carcinoma in situ of the pleomorphic type exhibits individual cells that are larger and more individually defined.

Lobular carcinoma in situ, especially of the pleomorphic type (bottom) , can directly progress to invasion (central and top ).

Lobular carcinoma in situ can also exhibit pagetoid spread into adjacent ducts.

One consistent molecular difference between LCIS and DCIS is the loss of epithelial cadherin (E-cadherin) expression by mutation, promoter methylation, or cis/trans promoter silencing in LCIS but the maintenance of expression of E-cadherin in DCIS. This observation can be extended to invasive lobular versus invasive ductal carcinoma as well.

Pathology of Ductal Carcinoma in Situ

DCIS comprises a heterogeneous group of noninvasive neoplastic proliferations with diverse morphologies and risks of subsequent recurrence and invasive transformation ( Figs. 9.11 to 9.17 ). Although DCIS probably arises predominantly in the terminal ductal-lobular unit, it often extends out to involve extralobular ducts. Compared with LCIS, DCIS is generally more variable histologically and cytologically, with larger and more pleomorphic nuclei and a tendency to form microacini, cribriform spaces, or papillary structures. In some cases, the periphery of these lesions may include patterns overlapping with atypical hyperplasia. Pathologists do not agree on whether small lesions should be considered atypical hyperplasia or in situ carcinoma. In general, lesions that involve only a few membrane-bound spaces and that measure less than 2 to 3 mm in greatest dimension should be regarded as hyperplastic lesions (with or without atypia) and not in situ carcinoma. There is a greater degree of concordance in larger lesions, however. Pathologists tend to agree about the diagnosis of difficult, smaller, borderline lesions if they have agreed on criteria ( Box 9.1 ). Occasionally, it may be difficult to distinguish DCIS and LCIS histologically, with some forms of DCIS characterized by small uniform cells with a solid growth pattern simulating LCIS. In rare instances, in situ neoplastic proliferations are indeterminate. In such cases, they are presumed to have the prognostic implications of both diagnoses (e.g., local evolution to invasion for DCIS and increased general risk in each breast for LCIS). It has been proposed that E-cadherin stains are useful in such overlap cases, but it should be pointed out that no long-term follow-up study has examined the implications of E-cadherin staining or absence thereof for regional breast cancer risk. It has been our approach to diagnose such cases as in situ carcinoma, mixed pattern, and to indicate that a regional risk for local recurrence should be assumed.

Low-power photograph demonstrating the full extent of the evidence supporting a diagnosis of atypical ductal hyperplasia (ADH). Note that there are only three or four spaces in which a central population of uniform cells may be seen. In the others, only narrow bars cross from one side to the other. Thus, there are pattern and cell population features of ductal carcinoma in situ (DCIS). However, in the three largest spaces involved, there are cells adjacent to the basement membrane that appear different; thus a diagnosis of ADH rather than DCIS is made. Magnification: ×75.

Photomicrograph exhibiting a detail of the polarization of luminal cells near the basement membrane that are quite different from the evenly placed and “suspicious” cells present in the central proliferation. This is atypical ductal hyperplasia. Magnification: ×200.

This central cribriform pattern of similar cells with outer cells normally polarized (above basement membrane) is probably the most common pattern of atypical ductal hyperplasia. (×150.)

(From Anderson TJ, Page DL. Risk assessment in breast cancer. In: Anthony PP, MacSween RNM, Lowe DG, eds. Recent Advances in Histopathology . Vol 17. Edinburgh: Churchill Livingstone; 1997.)

Collagenous spherulosis, a pattern sometimes confused with atypical ductal hyperplasia or ductal carcinoma in situ. Note that the spaces are defined by a secreted material that may be seen faintly. The spaces are surrounded by a sparse population of cells that everywhere is tapered or thinned in its extent. Such a pattern is not recognized as atypical. Magnification: ×150.

High-power view of comedo ductal carcinoma in situ demonstrating necrosis in the upper left-hand corner. Note also that the stroma is altered about this area, which occurs frequently in this type of carcinoma in situ. Magnification: ×700.

Occasionally cellular necrosis in comedo ductal carcinoma in situ is so extensive that very few atypical cells remain. Indeed, the necrosis may appear to extend to the basement membrane. Magnification: ×125.

Characteristic of more advanced and comedo carcinoma–type examples of ductal carcinoma in situ is the spread of highly atypical cells into lobular units. Here this phenomenon of so-called cancerization of lobules is demonstrated. Magnification: ×200.

Terminology is an important consideration here, and as noted earlier, there are distinct clinical implications concerning the terms lobular and ductal. Although the inherited terms have a historical legacy that may carry other significance, they have the impelling merit of familiarity. It is for us to develop more specific criteria for subtypes and to accept that it is the criteria linked to clinical end point analysis that guides clinical practice. One significant problem is the wonderfully earthy word comedo . It refers to the lowly comedones (e.g., acne) of common experience from our teen years, an unfailing image of the gross appearance of these lesions. Bloodgood coined the term comedo-adenoma because when treated with mastectomy in the halstedian era, it was associated with long-term survival. For Bloodgood, the alternative to carcinoma was adenoma, a lesion capable of cure when adequately excised. The term comedo remains descriptive and is now somewhat confusingly used to indicate both a type of DCIS with a coagulation-type necrosis and frequent nuclear debris, as well as merely the evidence of necrosis alone (comedo necrosis). Today most students of DCIS use comedo as a modifier for DCIS, signifying high-grade lesions that exhibit necrosis.

A major transition in our thinking regarding DCIS was the idea that perhaps not all DCIS cases were the same and that the different histologic appearances of DCIS might in fact have important clinical implications. Translating this concept into practical terms, it was suggested that if comedo-type DCIS did have a more menacing clinical import, then one should err on the side of including any questionable case within this category. Thus in 1989, the critically important concept of further stratification was introduced as a part of the inception of the modern era of understanding of DCIS. Because minor amounts of necrosis may be seen in the common hyperplasias without features of atypia, specific guidelines are necessary to make appropriate stratifications. Thus, inclusion of an intermediate-grade category has been adopted by the majority of DCIS classifications put forth since the early 1990s ( Table 9.1 ) to recognize examples with minimal necrosis and a moderate degree of nuclear pleomorphism.

Conventionally, DCIS has been classified on the basis of architectural features, such as comedo, cribriform, papillary, solid, and micropapillary. Although comedo-type DCIS includes advanced nuclear abnormalities within the neoplastic proliferation as part of the definition, the diagnosis of other patterns of DCIS were based on architecture alone. These patterns were accepted to be overlapping when DCIS was considered one entity and not held to have separate clinical or biological implications. The first indication that distinguishing among comedo, noncomedo, and micropapillary subtypes, as well as separation by grade, was of clinical utility certainly inaugurated the modern era of DCIS in 1989.

The increasing use of breast conserving therapy (BCT) in the treatment of mammographically detected DCIS has permitted studies on factors that predict local recurrences and invasive events in the remaining breast after excisional biopsy. Before a large number of small mammographically detected DCIS cases were found with screening, mastectomy was the only acceptable treatment for DCIS and remains a standard treatment for multifocal and multicentric disease.

Three prognostic factors have been shown to be important in local control of DCIS after attempts at BCT : (1) the extent (size) of disease in the breast (and its corollary, the residuum after an attempt at excision), (2) the status of margins (also reflecting residual disease in the breast), and (3) the grade of the DCIS (and possibly pure subtype, particularly micropapillary). The most significant of these factors appears to be margin status, followed by histologic grade. High nuclear grade and necrosis together define forms of DCIS at much higher risk of local recurrence and invasive transformation. The grade of a DCIS is largely independent of the conventional pattern classification. For example, lesions of high nuclear grade can exhibit any architectural pattern (although lack of precise patterns is most common). However, as recognized in the classification system of Holland and associates (see Box 9.1 ), ordered intercellular relationships are most common in lesions of low nuclear grade. There is a growing consensus that classifications based on nuclear grade and necrosis can identify the majority of patients with DCIS who are at risk for short-term local recurrence and invasive transformation after excision with or without irradiation. Most of these short-term recurrences are associated with DCIS exhibiting high (3/3 or grade III) nuclear grade morphology and significant coagulative necrosis. Such lesions would be conventionally classified as comedo DCIS. Studies using conventional classification schemes have shown that most short-term failures are associated with comedo-type DCIS. It should be recalled that the term comedo-type DCIS is not synonymous with high nuclear grade when it is used to indicate necrosis only. Some lesions exhibiting comedo-type necrosis and a solid growth pattern are composed of intermediate-grade and, in rare cases, borderline low-grade nuclei ( Figs. 9.18 to 9.27 ).

This low-power view of a common form of ductal carcinoma in situ shows solid cellular masses distending basement membrane–bound spaces. Within these cellular masses are sharply defined, rounded secondary lumens. There are central areas of necrosis as well as evident distention and distortion of the involved spaces. Magnification: ×75.

This high-power view of Fig. 9.11 demonstrates that the nuclei are of low grade, being similar one to another and without demonstrated irregularity. The presence of necrosis and low-grade nuclei is indicative of a condition intermediate between well-developed comedo ductal carcinoma in situ (DCIS) and the usual noncomedo DCIS. Magnification: ×200.

This example of ductal carcinoma in situ is characterized by sinuous, interconnecting strands of hyperchromatic cells. Note the few necrotic cells centrally. Magnification: ×100.

This example of intermediate-grade ductal carcinoma in situ (DCIS) presents highly atypical nuclei but not the most advanced, bizarre, and varied cytologic patterns often seen in comedo DCIS. Although one might debate whether these represent intermediate-grade nuclei, the limited luminal necrosis (here at upper left and elsewhere in this case) indicate an intermediate-grade designation. Magnification: ×350.

In this example of intermediate-grade ductal carcinoma in situ, the presence of necrosis helps define the category. Low- to intermediate-grade nuclei are present (×200).

(From Anderson TJ, Page DL. Risk assessment in breast cancer. In: Anthony PP, MacSween RNM, Lowe DG, eds. Recent Advances in Histopathology . Vol 17. Edinburgh: Churchill Livingstone; 1997.)

Rigid arches of a cribriform pattern variant of ductal carcinoma in situ. Note: calcified material in central spaces is not indicative of cellular necrosis. Magnification: ×225.

An example of a solid pattern variant of ductal carcinoma in situ. There are no evident intercellular spaces, and the slightly irregular placement of cells and sharply defined intercellular contours are not consistent with lobular carcinoma in situ. Magnification: ×450.

Micropapillary ductal carcinoma in situ with necrosis. Although some cells have lighter cytoplasm, the nuclear pattern is similar throughout. Magnification: ×400.

This solid variant of atypical ductal hyperplasia (ADH) is diagnostically very similar to solid ductal carcinoma in situ. The more vesicular nuclei in the second population of cells render a diagnosis of ADH. Magnification: ×150.

The microglandular or “endocrine” pattern of solid ductal carcinoma in situ. Magnification: ×150.

Information regarding the potential for recurrence of low-grade (noncomedo-type) DCIS after biopsy or BCT resides in a small number of published studies. It is clear that in the short term (5–10 years), few local recurrences or invasive transformations occur. However, a recent update of the only study of low-grade DCIS present at biopsy (without planned excision ) and with extended follow-up noted a substantially delayed recurrence rate of invasive lesions (e.g., approximately 37% and 50% at 25 years and more than 40 years of follow-up, respectively). Although the sample is small, it is significant that recurrences were generally in the same quadrant and, in some women, in the site of the prior biopsy, thus representing a biology identical to that of higher-grade DCIS and a risk that does not diminish after menopause. This biology of DCIS should be contrasted with that of risk marker lesions (e.g., ADH and atypical lobular hyperplasia [ALH]), which do not predict the side of involvement and in which risk diminishes postmenopausally (at least for ALH and LCIS).

There are several published classifications of DCIS, many of which use nuclear grade and necrosis as the major distinguishing features in a general classification applying to most cases of specific subtypes. The separations achieved by these classifications are different and in part may affect the interpretation of outcome results ( Table 9.2 ). DCIS characterized by nuclear morphology (high grade III; e.g., advanced atypia) and necrosis is uniformly classified as high grade. The European Organization for Research and Treatment of Cancer (EORTC) classification, although it does not use conventional nuclear grade or necrosis as major discriminates, would also regard this as a high-grade or, in their terminology, a poorly differentiated DCIS. Fisher and colleagues summarized the pathology analysis from the National Surgical Adjuvant Breast Project (NSABP) studies on DCIS and noted that DCIS with grade III nuclei and DCIS that exhibited larger areas of necrosis (greater than one third of ducts involved) had a higher local recurrence rate. The authors reported these results separately, not analyzing the risk associated with the two features in concert. Despite the differences in classification, it would appear that high-grade DCIS can be recognized uniformly, with all investigators showing that the high-grade subtype, so defined, has the highest risk of local recurrence and invasive transformation.

The recognition that necrosis and high nuclear grade usually cluster together may foster agreement between observers by using limited necrosis as a way of defining an intermediate-grade category. The separate classifications are less consistent with regard to the remainder of the heterogeneous noncomedo-type group (see Table 9.1 ). DCIS with grade III nuclei but without necrosis, an uncommon situation, is classified as high grade by Silverstein and coworkers but “noncomedo” (a lower grade) by Solin and associates. Lagios and colleagues and Silverstein and coworkers use nuclear grade to separate the remaining DCIS groups. However, Lagios and colleagues classify low-grade DCIS as grade I nuclei without necrosis and intermediate-grade DCIS as grade II with or without necrosis. Silverstein and coworkers separate DCIS with nuclear grades I and II on the basis of necrosis. Group I (low grade) may exhibit grade I or II nuclei but no necrosis, whereas DCIS with grade I or II nuclei but with any necrosis is classified as intermediate (group II). Solin and associates regard all DCIS without grade III nuclei and necrosis as noncomedo-type DCIS. Despite these differences in classification, all investigators have shown a substantially diminished local recurrence rate for DCIS that is not characterized by grade III nuclei and necrosis. Moreover, in those studies in which DCIS is divided into three groups, as opposed to the dichotomous comedo/noncomedo structure, there is a recognizable intermediate group (intermediate grade, group II, intermediately differentiated) that exhibits a morphology and risk intermediate between low- and high-grade DCIS.

Classification of Ductal Carcinoma in Situ

Although nuclear grade and necrosis would appear to define most of the risk associated with DCIS, certain architectural patterns appear to bear clinical significance independent of the nuclear grade. For example, DCIS with almost pure micropapillary architectural features is strongly associated with extensive disease, that is, within seemingly separate foci of different quadrants. This growth pattern makes adequate excision extremely difficult. In some cases, mammographic and histopathologic evidence of disease is present in all four quadrants of the breast. As a result, most clinical studies that define DCIS with micropapillary features and low nuclear grade were based on excisions without theoretically adequate margins of resection.

Conventional classification of DCIS covers perhaps 85% of what is recognized as noninvasive ductal carcinoma. A number of less common subtypes remains to be fully defined morphologically and with regard to risk. Proliferations with apocrine features, bridging the spectrum from minimal atypia to frank DCIS, were the subject of a proposed classification by O’Malley and associates. Because of the difficulty of applying traditional rules regarding cellular atypia and architecture to apocrine lesions, this schema proposed that definitive diagnoses of low-grade apocrine DCIS be limited to cases measuring at least 8 mm in size. In addition, a borderline category was proposed for lesions measuring 4 to 8 mm, with the suggestion that these lesions had the relative risk implications of at least ADH. Furthermore, apocrine DCIS, which is characteristically estrogen receptor negative, progesterone receptor negative, and androgen receptor positive and represents a heterogeneous group of lesions ranging from low-grade lesion (which should be differentiated from atypical hyperplastic apocrine lesions) to obvious malignant, high-grade tumors (which are difficult to recognize as apocrine), was more recently classified into three histologic grades based on nuclear grade and necrosis by Leal and colleagues, similar to schema for classical DCIS. Confirmation of the utility of any of these approaches to classifying such apocrine lesions awaits long-term follow-up analysis of cases treated with conservative surgery.

Another contender for special-type status is the so-called endocrine type of DCIS, which presents a particularly low-grade pattern of disease. Similarly, a possible special type characterized by hypersecretory features is discussed subsequently. In all of these less common special types of DCIS, a major limitation is the lack of precise confines of histologic definition that specifically and reproducibly describes the entire spectrum of changes with a linkage to clinical implications corroborated by long-term follow-up studies. With this background of remaining uncertainty, it is generally recommended that all the traditional rules for characterization and classification be applied to these less common entities, so as not to misdiagnose or mistreat such in situ lesions.

The classification of DCIS has been subjected to different approaches, each with advantages and disadvantages. The purpose of any classification scheme for DCIS is to predict the likelihood of recurrences and the likelihood of progression to invasion, and no classification scheme is ideal from these perspectives. For these reasons, newer classification schemes are continually evolving. The classification scheme proposed by Page and Lagios is summarized (see Tables 9.1 and 9.2 ). Several other DCIS classifications are presented (see Box 9.1 ), providing a basis from which to understand the slightly varied approaches. The major differences between Page’s classification and most of the other schemes are that Page’s uses the intersection of two variables—necrosis and nuclear grade—to foster agreement. It is common to debate between adjacent nuclear grades, viz, 1 or 2, and 2 or 3. The extensiveness of the necrosis is to be used to aid in the resolution of these issues. In a test set, agreement was fostered by this approach. The second feature of Page’s scheme is to separate some special types of DCIS because they present patterns not readily allowing grading. In the special case of pure (not intermixed with solid or cribriform) micropapillary DCIS, Page believes that the usual extensiveness of disease is independent of the nuclear grade. It should be noted that separating special types from the majority of cases is precisely what we do with invasive disease.