THE FEMALE BREAST

The class Mammalia is distinguished from other animals by highly evolved modified skin appendages, known as mammary glands or breasts, that provide a complete source of nourishment and an important degree of immunological protection for offspring. In humans, paired mammary glands rest on the pectoralis muscle on the upper chest wall. The breasts are composed of specialized epithelium and stroma that may give rise to both benign and malignant lesions (Fig. 23-1).

Diseases of the breast are best understood in the context of its normal anatomy. The human breast contains six to ten major ductal systems. The keratinizing squamous epithelium of the overlying skin dips into the orifices at the nipple and then abruptly changes to a double-layered cuboidal epithelium lining the ducts. Successive branching of the large ducts eventually leads to the terminal duct lobular unit. In adult women the terminal duct branches into a grapelike cluster of small acini to form a lobule (Figs. 23-1 and 23-2B). Each ductal system typically occupies more than one quadrant of the breast, and the systems extensively overlap one another. In some women, ducts extend into the subcutaneous tissue of the chest wall and into the axilla.

1

Two cell types line the ducts and lobules. Contractile myoepithelial cells containing myofilaments lie in a meshlike pattern on the basement membrane. These cells assist in milk ejection during lactation and provide structural support to the lobules. Luminal epithelial cells overlay the myoepithelial cells. Only the lobular luminal cells are capable of producing milk. A committed stem cell in the terminal duct is postulated to give rise to both luminal and myoepithelial cells.1

Figure 23-1 Anatomic origins of common breast lesions.

There are also two types of breast stroma. The interlobular stroma consists of dense fibrous connective tissue admixed with adipose tissue. The intralobular stroma envelopes the acini of the lobules and consists of breast-specific hormonally responsive fibroblast-like cells admixed with scattered lymphocytes. There is important cross-talk between breast epithelium and stroma that promotes the normal structure and function of the breast.2

In the prepubertal breast in males and females, the large duct system ends in terminal ducts with minimal lobule formation. Changes in the breast are most dynamic and profound during the reproductive years (Fig. 23-2). Just as the endometrium grows and ebbs with each menstrual cycle, so does the breast.3 In the first half of the menstrual cycle the lobules are relatively quiescent. After ovulation, under the influence of estrogen and rising progesterone levels, cell proliferation increases, as does the number of acini per lobule. The intralobular stroma also becomes markedly edematous. Upon menstruation, the fall in estrogen and progesterone levels induces the regression of the lobules and the disappearance of the stromal edema.

page 1067

Figure 23-2 Life cycle changes. A, Mammograms in young women are typically radiodense or white in appearance, making mass-forming lesions or calcifications (which are also radiodense) difficult to detect. B, The density of a young woman's breast stems from the predominance of fibrous interlobular stroma and the paucity of adipose tissue. Before pregnancy the lobules are small and are

invested by loose cellular intralobular stroma. Larger ducts connect lobules. C, During pregnancy, branching of terminal ducts produces more numerous, larger lobules. Luminal cells within lobules undergo lactational change, a precursor to milk

formation. D, With increasing age the lobules decrease in size and number, and the interlobular stroma is replaced by adipose tissue. E,Mammograms become more radiolucent with age as a result of the increase in adipose tissue, which facilitates the detection of radiodense mass-forming lesions and calcifications. (A, E, Courtesy of Dr. Darrell Smith, Brigham and Women's

Hospital, Boston, MA.)

Only with the onset of pregnancy does the breast become completely mature and functional. Lobules increase progressively in number and size. As a consequence, by the end of the pregnancy the breast is composed almost entirely of lobules separated by relatively scant stroma (Fig. 23-2C).

Immediately after delivery of the baby the luminal cells of the lobules produce colostrum (high in protein), which changes to milk (higher in fat and calories) over the next 10 days as progesterone levels drop. Not surprisingly, given these profound morphologic changes, the terminally differentiated breast has a specific pattern of gene expression.4

Breast milk not only provides complete nourishment from birth until several years of age, but it also provides protection against infection, allergies, and some autoimmune diseases. Maternal antibodies (chiefly secretory IgA), vitamins, enzymes, and numerous other mediators (e.g., cytokines, antioxidants, fibronectin, and lysozyme) augment the infant's own developing immune defenses. However, certain drugs, radioactive compounds given during diagnostic procedures, and viruses can also be passed to the

infant through breast milk.

Upon the cessation of lactation, the breast epithelium and stroma undergo extensive remodeling.5 Epithelial cells undergo apoptosis, lobules regress and atrophy, and the total breast size is diminished. However, full regression does not occur, and as a result pregnancy causes a permanent increase in the size and number of lobules.

After the third decade, long before menopause, lobules and their specialized stroma start to involute. Lobular atrophy may be almost complete in elderly females (Fig. 23-2D). The interlobular stroma also changes, since the radiodense fibrous stroma of the young female (see Fig. 23-2A) is progressively replaced by radiolucent adipose tissue (Fig. 23-2E).

Disorders of Development

Milkline Remnants

Supernumerary nipples or breasts result from the persistence of epidermal thickenings along the milk line, which extends from the axilla to the perineum. The disorders that affect the normally situated breast rarely arise in these heterotopic, hormone-responsive foci, which most commonly come to attention as a result of painful premenstrual enlargements.

Accessory Axillary Breast Tissue

In some women the normal ductal system extends into the subcutaneous tissue of the chest wall or the axillary fossa (the "axillary tail of Spence"). This epithelium can undergo lactational changes (resulting in a palpable mass) or give rise to carcinomas outside the breast proper. Therefore, prophylactic mastectomies markedly reduce, but do not completely eliminate, the risk of breast cancer.

Congenital Nipple Inversion

The failure of the nipple to evert during development is common and may be unilateral. Congenitally inverted nipples usually correct spontaneously during pregnancy, or can sometimes be everted by simple traction. Acquired nipple retraction is of more concern, since it may indicate the presence of an invasive cancer or an inflammatory disorder (e.g., recurrent subareolar abscess or duct ectasia).

Clinical Presentations of Breast Disease

The most common symptoms reported by women are pain, a palpable mass, "lumpiness" (without a discrete mass), or nipple discharge (Fig. 23-3). Asymptomatic women with abnormal findings on mammographic screening also require further evaluation.

Pain (mastalgia or mastodynia) is a common symptom that may be cyclic with menses or noncyclic. Diffuse cyclic pain has no pathologic correlate, and most effective treatments target hormone levels. Noncyclic pain is usually localized to one area of the breast. Causes include ruptured cysts, physical injury, and infections, but most often no

specific lesion is identified. Although roughly 95% of painful masses are benign, it must be remembered that about 10% of breast cancers are painful.

page 1067

page 1068

Figure 23-3 Common clinical symptoms of breast disease.

Discrete palpable masses are also common and must be distinguished from the normal nodularity (or "lumpiness") of the breast. The most common palpable lesions are invasive carcinomas, fibroadenomas, and cysts. A breast mass generally becomes palpable when it is at least 2 cm in size. Palpable masses are most common in premenopausal women (Fig. 23-4), but the likelihood of a palpable mass being malignant increases with age. For example, only 10% of breast masses in women under age 40 are malignant as compared with 60% of masses in women over age 50. Approximately 50% of carcinomas arise in the upper outer quadrant, 10% in each of the remaining quadrants, and about 20% in the central or subareolar region.

Mammographic screening was introduced in the 1980s as a means to detect small, nonpalpable, asymptomatic breast carcinomas (discussed later). The sensitivity and specificity of mammography increase with age, as a result of replacement of the fibrous, radiodense tissue of youth with the fatty, radiolucent tissue of the elderly (see Fig. 23-2). At age 40, the probability that a mammographic lesion is cancer is only 10%, but this rises to greater than 25% in women over 50 (see Fig. 23-4). The principal mammographic signs of breast carcinoma are densities and calcifications:

Densities. Mammographic densities are produced most commonly by invasive carcinomas, fibroadenomas, or cysts (see Fig. 23-4). Most neoplasms are radiologically denser than the intermingled normal breast tissue. The value of mammography lies in its ability to identify small, nonpalpable cancers. For example, the average size of an invasive carcinoma detected by mammography (1.1 cm) is less than half that of carcinomas detected by palpation (2.4 cm).

Calcifications. Calcifications form on secretions, necrotic debris, or hyalinized

stroma. Benign calcifications are often associated with clusters of apocrine cysts, hyalinized fibroadenomas, and sclerosing adenosis. Calcifications associated with malignancy are usually small, irregular, numerous, and clustered. Ductal carcinoma in situ (DCIS) is most commonly detected as mammographic calcifications, which are often deposited in a linear, branching pattern as the carcinoma fills the ductal system. Mammographic screening has increased the number of breast cancers diagnosed as DCIS (see Fig. 23-13). Invasive carcinomas presenting as calcifications without an accompanying radiodensity are uncommon, generally small in size, and rarely associated with lymph node metastases.

Figure 23-4 Frequency of pathologically diagnosed benign and malignant breast lesions by clinical presentation and age. Based on 914 women undergoing diagnostic breast surgery at Brigham and Women's Hospital (Boston, MA) from January to June 2001.

page 1068

page 1069

In about 10% of cases, carcinomas are missed by mammography. The principal causes of these failures are the presence of surrounding radiodense tissue (especially in younger women) that obscures the tumor, the absence of calcifications, small size, a diffuse infiltrative pattern with little or no desmoplastic response, or a location close to the chest wall or in the periphery of the breast. The inability to image a palpable mass does not indicate that it is benign, and all palpable masses require further investigation.

Other imaging modalities are useful adjuncts. Ultrasonography distinguishes between solid and cystic lesions and can define more precisely the borders of solid lesions. Most palpable masses that are not imaged by mammography are detectable by ultrasound. Magnetic resonance imaging (MRI) detects cancers by the rapid uptake of contrast agents due to increased tumor vascularity and blood flow. It is useful in screening for cancer in women with dense breasts or at very high risk for cancer, in determining the extent of chest wall invasion by locally advanced cancers, and in the evaluation of breast implant rupture. A high rate of false-positive results limits its usefulness in screening women outside of these groups.

Inflammatory Disorders

Inflammatory diseases of the breast are uncommon, accounting for less than 1% of women with breast symptoms. Women usually present with an erythematous swollen painful breast. "Inflammatory breast cancer" mimics inflammation by obstructing dermal vasculature with tumor emboli, resulting in an enlarged erythematous breast, and should always be suspected in a nonlactating woman with the clinical appearance of mastitis.

ACUTE MASTITIS

Almost all cases of acute mastitis occur during the first month of breastfeeding. During this time the breast is vulnerable to bacterial infection because of the development of cracks and fissures in the nipples. From this portal of entry, Staphylococcus aureus or, less commonly, streptococci invade the breast tissue. The breast is erythematous and painful, and fever is often present. At the outset only one duct system or sector of the breast is involved. If not treated the infection may spread to the entire breast.

Morphology. Staphylococcal infections usually produce a localized area of acute inflammation that may progress to the formation of single or multiple abscesses. Streptococcal infections tend to cause (as elsewhere) a diffuse spreading infection that eventually involves the entire breast. The involved breast tissue is infiltrated by neutrophils and may be necrotic.

Most cases of lactational mastitis are easily treated with appropriate antibiotics and continued expression of milk from the breast. Rarely, surgical drainage is required.

PERIDUCTAL MASTITIS

This condition is known by a variety of names, including recurrent subareolar abscess, squamous metaplasia of lactiferous ducts, and Zuska disease. Women, and sometimes men, present with a painful erythematous subareolar mass that clinically appears to be an infectious process. More than 90% of the afflicted are smokers. This condition is not associated with lactation, a specific reproductive history, or age. In recurrent cases, a fistula tract often tunnels under the smooth muscle of the nipple and opens onto the skin at the edge of the areola. Many women with this condition have an inverted nipple, most likely as a secondary effect of the underlying inflammation. The strong association with cigarette smoking is intriguing. It has been suggested that the vitamin A deficiency associated with smoking or toxic substances in tobacco smoke alter the differentiation of the ductal epithelium.7

Morphology. The key histologic feature is keratinizing squamous metaplasia of the nipple ducts (Fig. 23-5). Keratin shed from these cells plugs the ductal system, causing dilation and eventually rupture of the duct. An intense chronic and granulomatous inflammatory response develops once keratin spills into the surrounding periductal tissue. Sometimes a secondary bacterial infection supervenes and causes acute inflammation.

In most cases en bloc surgical removal of the involved duct and contiguous fistula tract is curative.7 Simple incision drains the abscess cavity, but the offending keratinizing epithelium remains and recurrences are common. When bacterial infection is present, antibiotics also have a therapeutic role.

Figure 23-5 Recurrent subareolar abscess. When squamous metaplasia extends deep into a nipple duct, keratin becomes trapped and accumulates. If the duct ruptures, the ensuing intense inflammatory response to keratin results in an erythematous painful

mass. A fistula tract may burrow beneath the smooth muscle of the nipple to open at the edge of the areola.

page 1069

page 1070

Figure 23-6 Mammary duct ectasia. Chronic inflammation and fibrosis surround an ectatic duct filled with inspissated debris. The fibrotic response can produce a firm irregular mass that mimics invasive carcinoma on palpation or mammogram.

MAMMARY DUCT ECTASIA

This disorder tends to occur in the fifth or sixth decade of life, usually in multiparous

women. Unlike periductal mastitis, it is not associated with cigarette smoking. Patients present with a poorly defined palpable periareolar mass that is often associated with thick, white nipple secretions and sometimes with skin retraction. Pain and erythema are uncommon.

Morphology. This lesion is characterized chiefly by dilation of ducts, inspissation of breast secretions, and a marked periductal and interstitial chronic granulomatous inflammatory reaction (Fig. 23-6). The dilated ducts are filled by granular debris that contains numerous lipid-laden macrophages. The periductal and interductal tissue contains dense infiltrates of lymphocytes and macrophages, and variable numbers of plasma cells. On occasion, granulomatous inflammation forms around cholesterol deposits. Fibrosis may eventually produce skin and nipple retraction. Squamous metaplasia of nipple ducts is absent.

The principal significance of this disorder is that it produces an irregular palpable mass that mimics the mammographic appearance of carcinoma.

FAT NECROSIS

Fat necrosis can present as a painless palpable mass, skin thickening or retraction, a mammographic density, or mammographic calcifications. The majority of affected women have a history of breast trauma or prior surgery.

Morphology. Acute lesions may be hemorrhagic and contain central areas of liquefactive fat necrosis. In subacute lesions the areas of fat necrosis take on the appearance of ill-defined, firm, gray-white nodules containing small chalky-white foci or dark hemorrhagic debris. The central region of necrotic fat cells is initially associated with an intense neutrophilic infiltrate mixed with macrophages. Over the next few days proliferating fibroblasts associated with new vessels and chronic inflammatory cells surround the injured area. Subsequently, giant cells, calcifications, and hemosiderin make their appearance, and eventually the focus is replaced by scar tissue or is encircled and walled off by fibrous tissue.

As with other inflammatory breast disorders, the major clinical significance of the condition is its possible confusion with breast cancer.

LYMPHOCYTIC MASTOPATHY (SCLEROSING LYMPHOCYTIC LOBULITIS)

This condition presents with single or multiple hard palpable masses. The masses may be bilateral and may be detected as mammographic densities. The lesions are so hard that it can be difficult to obtain tissue with a needle biopsy. Microscopically, they show collagenized stroma surrounding atrophic ducts and lobules. The epithelial basement membrane is often thickened. A prominent lymphocytic infiltrate surrounds the epithelium and small blood vessels. This condition is most common in women with type

1 (insulin-dependent) diabetes or autoimmune thyroid disease. Based on this association, it is hypothesized to have an autoimmune basis. Its only clinical significance is that it must be distinguished from breast cancer.

GRANULOMATOUS MASTITIS

Granulomatous inflammation is present in less than 1% of all breast biopsy specimens. The causes include systemic granulomatous diseases (e.g., Wegener granulomatosis or sarcoidosis) that occasionally involve the breast, and granulomatous infections caused by mycobacteria or fungi. Infections of this type are most common in immunocompromised patients or adjacent to foreign objects such as breast prostheses or nipple piercings. Granulomatous lobular mastitis is an uncommon breast-limited disease that only occurs in parous women. The granulomatous inflammation is confined to the lobules, suggesting that it is caused by a hypersensitivity reaction to antigens expressed by lobular epithelium during lactation.

Benign Epithelial Lesions

A wide variety of benign alterations in ducts and lobules are observed in the breast. Most come to clinical attention when detected by mammography or as incidental findings in surgical specimens. These lesions have been divided into three groups, according to the subsequent risk of developing breast cancer: (1) nonproliferative breast changes, (2) proliferative breast disease, and (3) atypical hyperplasia.

page 1070

page 1071

NONPROLIFERATIVE BREAST CHANGES (FIBROCYSTIC CHANGES)

This group includes a number of very common morphologic alterations that are often grouped under the term fibrocystic changes. To the clinician the term might mean "lumpy bumpy" breasts on palpation; to the radiologist, a dense breast with cysts; and to the pathologist, benign histologic findings. These lesions are termed nonproliferative to distinguish them from "proliferative" changes, which are associated with an increased risk of breast cancer.

Figure 23-7 Apocrine cysts. A, Clustered, rounded calcifications are seen in a radiograph. B, Gross appearance of typical cysts filled with dark, turbid fluid contents. C, Cysts are lined by apocrine cells with round nuclei and abundant granular cytoplasm. Note the

luminal calcifications, which form on secretory debris.

Morphology. There are three principal morphologic changes: (1) cystic change, often with apocrine metaplasia; (2) fibrosis; and (3) adenosis.

Cysts. Small cysts form by the dilation and unfolding of lobules, and in turn may coalesce to form larger cysts. Unopened cysts contain turbid, semi-translucent fluid that produces a brown or blue color (blue-dome cysts) (Fig. 23-7B). Cysts are lined either by a flattened atrophic epithelium or by metaplastic apocrine cells. The latter cells, which have an abundant granular, eosinophilic cytoplasm and round nuclei, closely resemble the normal apocrine epithelium of sweat glands (Fig. 23-7C). Calcifications are common and may be detected by mammography (see Fig. 23-7A). "Milk of calcium" is a term mammographers use to describe calcifications that line the bottom of a rounded cyst. Cysts are alarming when they are solitary and firm to palpation. The diagnosis is confirmed by the disappearance of the cyst after fine-needle aspiration of its contents.

Fibrosis. Cysts frequently rupture, releasing secretory material into the adjacent stroma. The resulting chronic inflammation and fibrosis contribute to the palpable firmness of the breast.

Adenosis. Adenosis is defined as an increase in the number of acini per lobule. A normal physiologic adenosis occurs during pregnancy. In nonpregnant women, adenosis can occur as a focal change. The acini are often enlarged (blunt-duct adenosis), but are not distorted as is seen in sclerosing adenosis, described later. Calcifications are occasionally present within the lumens. The acini are lined by columnar cells, which may appear benign or show atypical features ("flat epithelial atypia"). These lesions may be the earliest recognizable precursor of epithelial neoplasia.8-10

Lactational adenomas present as palpable masses in pregnant or lactating women. They are formed by normal-appearing breast tissue with physiologic adenosis and lactational changes. These lesions are probably not true neoplasms but an exaggerated focal response to hormonal influences.

PROLIFERATIVE BREAST DISEASE WITHOUT ATYPIA

These changes are commonly detected as mammographic densities, calcifications, or as incidental findings in specimens from biopsies performed for other reasons. Although

each can be found in isolation, typically more than one lesion is present, frequently in association with nonproliferative breast changes.

These lesions are characterized by proliferation of ductal epithelium and/or stroma without cytologic or architectural features suggestive of carcinoma in situ.

Morphology

Epithelial Hyperplasia. Normal breast ducts and lobules are lined by a double layer of myoepithelial cells and luminal cells (Fig. 23-8A). Epithelial hyperplasia is defined by the presence of more than two cell layers. The additional cells consist of both luminal and myoepithelial cell types that fill and distend ducts and lobules. Irregular lumens can often be discerned at the periphery of the cellular masses (Fig. 23-8B). Epithelial hyperplasia is usually an incidental finding.

page 1071

page 1072

Sclerosing Adenosis. The number of acini per terminal duct is increased to at least double the number found in uninvolved lobules. The normal lobular arrangement is maintained. The acini are compressed and distorted in the central portions of the lesion but characteristically dilated at the periphery. Myoepithelial cells are usually prominent. On occasion, stromal fibrosis may completely compress the lumens to create the appearance of solid cords or double strands of cells lying within dense stroma, a histologic pattern that at times closely mimics the appearance of invasive carcinoma (Fig. 23-9). Sclerosing adenosis can come to attention as a palpable mass, a radiologic density, or calcifications.

Complex Sclerosing Lesion. Complex sclerosing lesions have components of sclerosing adenosis, papillomas, and epithelial hyperplasia. One member of this group, the radial sclerosing lesion ("radial scar"), is the only commonly occurring benign lesion that forms irregular masses and can closely mimic invasive carcinoma mammographically, grossly, and histologically (Fig. 23-10). There is a central nidus of entrapped glands in a hyalinized stroma with long radiating projections into stroma. The term radial scar is a misnomer, as these lesions are not associated with prior trauma or surgery.

Papillomas. Papillomas are composed of multiple branching fibrovascular cores, each having a connective tissue axis lined by luminal and myoepithelial cells (Fig. 23-11). Growth occurs within a dilated duct. Epithelial hyperplasia and apocrine metaplasia are frequently present. Large duct papillomas are usually solitary and situated in the lactiferous sinuses of the nipple. Small duct

papillomas are commonly multiple and located deeper within the ductal system.

Figure 23-8 A, A normal duct or acinus with a single basally located myoepithelial cell layer (cells with dark, compact nuclei and scant cytoplasm) and a single luminal cell layer (cells with larger open nuclei, small nucleoli, and more abundant

cytoplasm). B,Epithelial hyperplasia. The lumen is filled by a heterogeneous, mixed population of luminal and myoepithelial cell types. Irregular slitlike fenestrations are prominent at the periphery.

More than 80% of large duct papillomas produce a nipple discharge. Large papillomas may undergo infarction, possibly because of torsion on the stalk, resulting in a bloody discharge. Nonbloody discharge probably results from intermittent blockage and release of normal breast secretions or irritation of the duct by the papilloma. The remaining large duct papillomas and most small duct papillomas come to clinical attention as small palpable masses, or as densities or calcifications seen on mammograms.

Figure 23-9 Sclerosing adenosis. The involved terminal duct lobular unit is enlarged, and the acini are compressed and distorted by dense stroma. Calcifications are present within some of the lumens. Unlike carcinomas, the acini are arranged in a swirling pattern,

and the outer border is well circumscribed.

page 1072

page 1073

Figure 23-10 Radial sclerosing lesion. A, The radiograph shows an irregular central mass with long radiodense projections. B, Grossly the mass appears solid and has irregular borders, but it is not as firm as an invasive carcinoma. C, The mass

consists of a central nidus of small tubules entrapped in a densely fibrotic stroma and numerous projections containing epithelium with varying degrees of cyst formation and hyperplasia.

PROLIFERATIVE BREAST DISEASE WITH ATYPIA

Proliferative disease with atypia includes atypical ductal hyperplasia and atypical lobular hyperplasia. Atypical ductal hyperplasia is present in 5% to 17% of specimens from biopsies performed for calcifications and is found less frequently in specimens from biopsies for mammographic densities or palpable masses. Occasionally, atypical ductal hyperplasia is associated with radiologic calcifications; more commonly it is adjacent to another calcifying lesion. Atypical lobular hyperplasia is an incidental finding and is found in fewer than 5% of specimens from biopsies performed for any reason.

Morphology. Atypical hyperplasia is a cellular proliferation resembling carcinoma in situ but lacking sufficient qualitative or quantitative features for diagnosis as carcinoma. Unlike other benign changes, atypical hyperplasias harbor some of the same acquired genetic losses and gains that are present in carcinoma in situ.

Atypical ductal hyperplasia is recognized by its histologic resemblance to ductal carcinoma in situ (DCIS). It consists of a relatively monomorphic proliferation of regularly spaced cells, sometimes with cribriform spaces. It is distinguished from DCIS by being limited in extent and only partially filling ducts (Fig. 23-12A).

Atypical lobular hyperplasia is defined as a proliferation of cells identical to those of lobular carcinoma in situ (LCIS, described later), but the cells do not fill or distend more than 50% of the acini within a lobule (Fig. 23-12B). Atypical lobular hyperplasia can also involve contiguous ducts through pagetoid spread, in which atypical lobular cells lie between the ductal basement membrane and overlying normal ductal epithelial cells.

CLINICAL SIGNIFICANCE OF BENIGN EPITHELIAL CHANGES

Multiple epidemiologic studies have classified benign histologic changes in the breast and determined their association with the later development of invasive cancer11-

13 (Table 23-1). Nonproliferative changes do not increase the risk of cancer. Proliferative disease is associated with a mild increase in risk, while proliferative disease with atypia confers a moderate increase in risk. Both breasts are at increased risk, although a few more subsequent carcinomas occur in the same breast.14 Risk reduction can be achieved by bilateral prophylactic mastectomy or treatment with estrogen antagonists, such as tamoxifen.15 However, more than 80% of women with atypical hyperplasia will not develop breast cancer, and many choose careful clinical and radiologic surveillance over intervention.

Carcinoma of the Breast

Figure 23-11 Intraductal papilloma. A central fibrovascular core extends from the wall of a duct. The papillae arborize within the lumen and are lined by myoepithelial and luminal cells.

page 1073

page 1074

Figure 23-12 A, Atypical ductal hyperplasia. A duct is filled with a mixed population of cells consisting of oriented columnar cells at the periphery and more rounded cells within the central portion. Although some of the spaces are round and regular, the peripheral

spaces are irregular and slitlike. These features are highly atypical, but fall short of a diagnosis of DCIS. B, Atypical lobular hyperplasia. A population of monomorphic small, round, loosely cohesive cells partially fill a lobule. Some intracellular lumens can

be seen. Although the cells are morphologically identical to the cells of LCIS, the extent of involvement is not sufficient for this diagnosis.

Table 23-1. Epithelial Breast Lesions and the Risk of Developing Invasive Carcinoma

Pathologic Lesion Relative Risk (Absolute Lifetime Risk)*

NONPROLIFERATIVE BREAST CHANGES (Fibrocystic changes) 1.0 (3%)

Duct ectasia

Cysts

Apocrine change

Mild hyperplasia

Adenosis

Fibroadenoma w/o complex features

PROLIFERATIVE DISEASE WITHOUT ATYPIA 1.5 to 2.0 (5% to 7%)

Moderate or florid hyperplasia

Sclerosing adenosis

Papilloma

Complex sclerosing lesion (radial scar)

Fibroadenoma with complex features

PROLIFERATIVE DISEASE WITH ATYPIA 4.0 to 5.0 (13% to 17%)

Atypical ductal hyperplasia (ADH)

Atypical lobular hyperplasia (ALH)

CARCINOMA IN SITU 8.0 to 10.0 (25% to 30%)

Lobular carcinoma in situ (LCIS)

Ductal carcinoma in situ (DCIS) *Relative risk is the risk compared to women without any risk factors. Absolute lifetime risk is the percentage of patients expected to develop invasive carcinoma if untreated.

Carcinoma of the breast is the most common non-skin malignancy in women. A woman who lives to age 90 has a one in eight chance of developing breast cancer. In 2007 an estimated 178,480 women were diagnosed with invasive breast cancer, 62,030 with carcinoma in situ, and over 40,000 women died of the disease (Surveillance Epidemiology and End Results [SEER] data at http://seer.cancer.gov/). As the demographic bulge of the "baby boomers" continues to grow older, the number of women with breast cancer is expected to increase by about a third over the next 20 years. It is both ironic and tragic that a neoplasm arising in an exposed organ, readily accessible to self-examination and clinical diagnosis, continues to exact such a heavy toll. Only lung cancer causes more cancer deaths in women living in the United States.

It has long been appreciated that breast cancer is a heterogeneous disease with a wide array of histologic appearances. Recent gene profiling studies have confirmed that there are many types of cancers but also show that most carcinomas cluster into several major groups with important biologic and clinical differences. The majority of carcinomas are estrogen receptor (ER) positive and are characterized by a gene signature dominated by the dozens of genes under the control of estrogen. Among the ER-negative tumors, many fall into a distinctive "basal-like" group that is discussed later.

ER-positive and ER-negative carcinomas show striking differences with regard to patient characteristics, pathologic features, treatment response, and outcome. In the past, most studies grouped all breast cancers together, but it is now widely recognized that the diagnosis of breast cancer encompasses multiple molecular subclasses of disease, as discussed later.

INCIDENCE AND EPIDEMIOLOGY page 1074

page 1075

Figure 23-13 Breast cancer incidence and mortality rates for women over 50 years of age. Rates are per 100,000 women and are age-adjusted to the 2000 US standard million population. (SEER Cancer Statistics Review; http://seer.cancer.gov/)

Figure 23-14 Change in stage of breast cancer at presentation from 1983 to 1996. (SEER Cancer Statistics Review, http://seer.cancer.gov/)

After remaining constant for many years (except for a transient rise in 1974 attributed to increased awareness surrounding the recurrence of breast cancer in Betty Ford and Happy Rockefeller), the incidence of breast cancer began to increase in older women (Fig. 23-13). What seemed to be an alarming trend was, in part, due to the introduction of mammographic screening in the early 1980s. Rates of screening gradually increased but have recently reached a plateau of 60% to 80% of eligible women. The main benefit of screening is the detection of small, predominantly ER-positive invasive carcinomas and in situ carcinomas. DCIS is almost exclusively detected by mammography, providing an explanation for the sharp increase in the diagnosis of DCIS since 1980

(see Fig. 23-13). Small node-negative carcinomas (stage I), which are best detected by mammography, increased in frequency as the number of large, advanced-stage breast carcinomas (stages II to IV) diminished modestly (Fig. 23-14). Over the same time period the incidence of breast carcinoma in younger women, for whom screening is not recommended, did not change.

From 2001 to 2004, the incidence of ER-positive invasive cancer decreased. The reasons for this trend are probably multifactorial. The plateau in the number of women screened should be associated with a decrease in incidence back to prescreening levels. In addition, in 2002 many women stopped using postmenopausal hormone replacement therapy after the results of the Women's Health Initiative trial showed that this treatment had limited benefits.16 It is possible that this treatment stimulated the growth or development of ER-positive cancers. During the same time period the incidence of breast cancer for African American women remained stable and the number of ER-negative cancers increased, suggesting that these cancers are not affected by hormonal treatment. Finally, there may have been changes in modifiable risk factors (e.g., the frequency and length of breastfeeding) or the use of chemopreventive agents that can lower risk. Whatever the reason or reasons, the decrease in breast cancers is a promising trend that hopefully will continue.

page 1075

page 1076

During the 1980s the number of women dying of breast cancer remained constant, while the incidence of breast cancer was increasing. Since 1994 the breast cancer mortality rate for all women has slowly declined from 30% to 20% (Fig. 23-13). The decrease is attributed to the detection of clinically significant cancers at a curable stage due to screening, as well as better and more effective treatment modalities. The number of women dying from their breast cancer has decreased from 30% to 20%. However, the decline in the death rate has been less impressive for African American women, women in other ethnic groups, and women with ER-negative cancers. The mortality is higher in these groups even though the incidence of cancer is lower than in white women.

Risk Factors

The most important risk factor is gender; only 1% of breast cancer cases occur in men. Common risk factors for women identified by epidemiologic studies have been combined into the Breast Cancer Risk Assessment Tool (BCRAT), which now includes information from the Contraceptive and Reproductive Experiences study,17,18 which provides more accurate information for African American women. The model can be used to calculate the absolute risk of an individual woman developing invasive cancer within the next 5 years or over a lifetime. The BCRAT incorporates the following risk factors.

Age

The incidence rises throughout a woman's lifetime, peaking at the age of 75-80 years and then declining slightly thereafter. The average age at diagnosis is 61 for white

women, 56 for Hispanic women, and 46 for African American women. Only 20% of non-Hispanic white women are diagnosed under the age of 50, compared with 35% of African American women and 31% of Hispanic women. Breast cancer is very rare in all groups before the age of 25.

Although carcinoma is uncommon in young women, almost half of these are either ER negative or human epidermal growth factor receptor 2 (HER2/neu) positive, whereas these cancers make up less than a third of cancers in women over the age of 40.

Age at Menarche

Women who reach menarche when younger than 11 years of age have a 20% increased risk compared with women who are more than 14 years of age at menarche. Late menopause also increases risk.

Age at First Live Birth

Women who experience a first full-term pregnancy at ages younger than 20 years have half the risk of nulliparous women or women over the age of 35 at their first birth. It is hypothesized that pregnancy results in terminal differentiation of milk-producing luminal cells, removing them from the potential pool of cancer precursors.4 This protective effect might be overshadowed in older women by stimulation of proliferation early in pregnancy of cells that have already undergone preneoplastic changes. It is also possible that the changes in stroma that allow the growth and expansion of lobules during pregnancy facilitate the transition from in situ to invasive carcinoma. These pregnancy-related changes may help explain the transient increase in cancer risk that follows a pregnancy, an effect that is most pronounced in older women.5 Age at first live birth is not a strong risk factor for African American women.

First-Degree Relatives with Breast Cancer

The risk of breast cancer increases with the number of affected first-degree relatives (mother, sister, or daughter), especially if the cancer occurred at a young age. However, most women do not have a family history. Only 13% of women with breast cancer have one affected first-degree relative, and only 1% have two or more. In turn, over 87% of women with a family history will not develop breast cancer. Most family risk is probably due to the interaction of low-risk susceptibility genes and nongenetic factors. The BCRAT is not designed to calculate the risk for women with a mutation in a high-risk breast cancer gene, such as BRCA1 orBRCA2 (see the section "Hereditary Breast Cancer" below).

Atypical Hyperplasia

A history of prior breast biopsies, especially if revealing atypical hyperplasia, increases the risk of invasive carcinoma. There is a smaller increase in risk associated with proliferative breast changes without atypia (see Table 23-1).

Race/Ethnicity

Non-Hispanic white women have the highest rates of breast cancer. The risk of developing an invasive carcinoma within the next 20 years at age 50 is 1 in 15 for this

group, 1 in 20 for African Americans, 1 in 26 for Asian/Pacific Islanders, and 1 in 27 for Hispanics.19 However, women of African or Hispanic ancestry present at a more advanced stage and have an increased mortality rate. Social factors such as decreased access to health care and lower use of mammography may well contribute to these disparities, but biologic differences also play an important role.20 African American and Hispanic women tend to develop cancers at a younger age, prior to menopause, that are more likely to be poorly differentiated and ER negative. Mutations in p53 are more common in African American women but less common in Hispanic women, as compared with non-Hispanic white women. It is suspected that variation in breast cancer risk genes across ethnic groups is responsible, at least in part, for these differences. One known example is the incidence of BRCA1 andBRCA2 mutations, which occur at different frequencies in different ethnic groups.21

Additional risk factors (listed below) are recognized, but have not been incorporated into the BCRAT model because of their rarity or uncertainties about quantifying the magnitude of risk.

Estrogen Exposure

Postmenopausal hormone replacement therapy increases the risk of breast cancer 1.2- to 1.7-fold, and adding progesterone increases the risk further. Most excess cancers are ER-positive carcinomas, including invasive lobular carcinomas, that tend to be of small size when detected. As a result, any effect on the death rate is expected to be small. After publication of the Women's Health Initiative trial in 2002, the number of postmenopausal women receiving hormone replacement therapy dropped from approximately 17% to 7%, a change that was followed by a substantial drop in ER-positive invasive breast cancers in 2003 and 2004 (see Fig. 23-13).16

Oral contraceptives have not been shown convincingly to affect breast cancer risk but do decrease the risk of endometrial and ovarian carcinomas. Reducing endogenous estrogens by oophorectomy decreases the risk of developing breast cancer by up to 75%. Drugs that block estrogenic effects (e.g., tamoxifen) or block the formation of estrogen (e.g., aromatase inhibitors) also decrease the risk of ER-positive breast cancer.

Breast Density

High breast radiodensity is a strong risk factor for developing cancer. High density is correlated with young age and hormone exposure, and clusters in families. High breast density may be related to less complete involution of lobules at the end of each menstrual cycle, which in turn may increase the number of cells that are potentially susceptible to neoplastic transformation.

Dense breasts also make detection of cancer more difficult by mammography. Other modalities, such as MRI, may be helpful in such women.

Radiation Exposure page 1076

page 1077

Radiation to the chest, whether due to cancer therapy, atomic bomb exposure, or nuclear accidents, results in a higher rate of breast cancer. The risk is greatest with exposure at young ages and with high radiation doses. For example, women in their teens and early 20s who received radiation to the chest for Hodgkin lymphoma have a 20% to 30% risk of developing breast cancer over 10 to 30 years. Recognition of this iatrogenic complication has led to a much more judicious use of radiation therapy in adolescents and young women undergoing cancer treatment. The risks of radiation exposure are substantially lower in women over the age of 25. Current mammographic screening uses low doses of radiation and is unlikely to have an effect on the risk of breast cancer.

Carcinoma of the Contralateral Breast or Endometrium

Approximately 1% of women with breast cancer develop a second contralateral breast carcinoma per year. The risk is higher for women with germline mutations in high-risk breast cancer genes such as BRCA1 andBRCA2, who frequently develop multiple cancers. Breast and endometrial carcinomas have several risk factors in common, the most important of which is exposure to prolonged estrogenic stimulation.

Geographic Influence

Breast cancer incidence rates in the United States and Europe are four to seven times higher than those in other countries. Unfortunately, the rates are rising worldwide, and by 2020 it is estimated that 70% of cases will be in developing countries.

The risk of breast cancer increases in immigrants to the United States with each generation. The factors responsible for this increase are of considerable interest because they are likely to include modifiable risk factors. Reproductive history (number and timing of pregnancies), breastfeeding, diet, obesity, physical activity, and environmental factors all probably play a role.

Diet

Large studies have failed to find strong correlations between breast cancer risk and dietary intake of any specific type of food. Coffee addicts will be pleased to know that caffeine consumption may decrease the risk of breast cancer. On the other hand, moderate or heavy alcohol consumption increases risk. Higher estrogen levels and lower folate levels may underlie this association.

Obesity

There is decreased risk in obese women younger than 40 years as a result of the association with anovulatory cycles and lower progesterone levels late in the cycle. In contrast, the risk is increased for postmenopausal obese women, which is attributed to the synthesis of estrogens in fat depots.

Exercise

There is a probable small protective effect for women who are physically active. The decrease in risk is greatest for premenopausal women, women who are not obese, and women who have had full-term pregnancies.

Breastfeeding

The longer women breastfeed, the greater the reduction in risk. Lactation suppresses ovulation and may trigger terminal differentiation of luminal cells.4 The lower incidence of breast cancer in developing countries largely can be explained by the more frequent and longer nursing of infants.22

Environmental Toxins

There is concern that environmental contaminants, such as organochlorine pesticides, have estrogenic effects on humans. Possible links to breast cancer risk are being investigated intensively, but definitive associations have yet to be made.

Tobacco

Cigarette smoking has not been clearly associated with breast cancer but is associated with the development of periductal mastitis (subareolar abscess; discussed earlier). Breast cancer was the leading cause of cancer deaths in women until the early 1990s, when lung cancer deaths surged ahead. Currently, twice as many women die from lung cancer-surely a good reason to avoid tobacco use.

ETIOLOGY AND PATHOGENESIS

The major risk factors for the development of breast cancer are hormonal and genetic. Breast carcinomas can therefore be divided into sporadic cases, probably related to hormonal exposure, and hereditary cases, associated with germline mutations. Hereditary carcinoma has received intense scrutiny in the hopes that the specific genetic mutations can be identified and that these alterations will illuminate the causes of nonfamilial breast cancers as well. Recent studies have borne out these hopes. We begin our discussion with hereditary breast cancer and follow with sporadic breast cancer.

Hereditary Breast Cancer

The inheritance of a susceptibility gene or genes is the primary cause of approximately 12% of breast cancers.23,24 The probability of a hereditary etiology increases with multiple affected first-degree relatives, when individuals are affected before menopause and/or have multiple cancers, or there are family members with other specific cancers (discussed below).

In some families the increased risk is the result of a single mutation in a highly penetrant breast cancer gene (Table 23-2). Mutations in BRCA1 and BRCA2 account for the majority of cancers attributable to single mutations and about 3% of all breast cancers. Penetrance (the percentage of carriers who develop breast cancer) varies from 30% to 90% depending on the specific mutation present. Mutations in BRCA1 also markedly increase the risk of developing ovarian carcinoma, which occurs in as many as 20% to 40% of carriers. BRCA2 confers a smaller risk for ovarian carcinoma (10% to 20%) but is associated more frequently with male breast cancer. BRCA1 and BRCA2 carriers are also at higher risk for other epithelial cancers, such as prostatic and pancreatic carcinomas.

BRCA1 and BRCA2 are both large genes over 80 kilobases in size. Hundreds of different mutations distributed throughout the coding regions have been reported for each. The frequency of mutations that increase breast cancer risk is only 0.1% to 0.2% in the general population, and inconsequential polymorphisms are common. As a result, genetic testing is difficult and generally restricted to individuals with a strong family history or those belonging to certain ethnic groups. For example, 2% to 3% of people of Ashkenazi Jewish descent carry one of three specific mutations, two in BRCA1 and one in BRCA2. Identification of carriers is important, since increased surveillance, prophylactic mastectomy, and oophorectomy can reduce cancer-related morbidity and mortality.

page 1077

page 1078

Table 23-2. Most Common "Single Gene" Mutations Associated with Hereditary Susceptibility to Breast Cancer

GENE (location) Syndrome (Incidence)*

% of "Single Gene" Hereditary Cancers†

Breast Cancer Risk by Age 70‡

Changes in Sporadic Breast Cancer

Other Associated Cancers Functions Comments

BRCA1 (17q21) Familial breast and ovarian cancer (1 in 860)

52% (∼2% of all breast cancers)

40% to 90%

Mutations rare; inactivated in 50% of some subtypes (e.g. medullary and metaplastic) by methylation

Ovarian, male breast cancer (but lower than BRCA2), prostate, pancreas, fallopian tube

Tumor suppressor, transcriptional regulation, repair of double-stranded DNA breaks

Breast carcinomas are commonly poorly differentiated and triple negative (basal-like), and have P53mutations.

BRCA2 (13q12-13) Familial breast and ovarian cancer (1 in 740)

32% (∼1% of all breast cancers)

30% to 90%

Mutations and loss of expression rare

Ovarian, male breast cancer, prostate, pancreas, stomach, melanoma, gallbladder, bile duct, pharynx

Tumor suppressor, transcriptional regulation, repair of double-stranded DNA breaks

Biallelic germline mutations cause a rare form of Fanconi anemia (Chapter 7)

p53 (17p13.1) Li-Fraumeni (1 in 5,000)

3% (<1% of all breast cancers)

>90% Mutations in 20%, LOH in 30% to 42%; most frequent in triple negative cancers

Sarcoma, leukemia, brain tumors, adrenocortical carcinoma, others

Tumor suppressor with critical roles in cell cycle control, DNA replication, DNA repair, and apoptosis

p53 is the most commonly mutated gene in sporadic breast cancers

CHEK2 (22q12.1) Li-Fraumeni variant (1 in 100)

5% (∼1% of all breast cancers)

10% to 20%

Mutations rare (<5%); loss of protein expression in at least one third by unknown mechanism(s)

Prostate, thyroid, kidney, colon

Cell cycle checkpoint kinase, recognition and repair of DNA damage, activates BRCA1 and p53 by phosphorylation

May increase risk for breast cancer after radiation exposure

*Frequency of heterozygotes in the U.S. population; the incidence of gene mutations is higher in some ethnic populations (e.g., BRCA1 and BRCA2 mutations occur at high frequencies in Askenazi Jews). †Defined as familial breast cancers showing a pattern of inheritance consistent with a major effect of a single gene.

‡Risk varies with specific mutations and is likely modified by other genes.

BRCA1-associated breast cancers are commonly poorly differentiated, have "medullary

features" (a syncytial growth pattern with pushing margins and a lymphocytic response), and do not express hormone receptors or overexpress HER2/neu (the so-called "triple negative" phenotype). Their gene profiling signature is very similar to basal-like breast cancers, a distinct molecular subtype that is discussed later. BRCA1 cancers are also frequently associated with loss of the inactive X chromosome and reduplication of the active X, resulting in the absence of the Barr body.25 BRCA2-associated breast carcinomas also tend to be relatively poorly differentiated, but are more often ER positive than BRCA1 cancers.

Other known susceptibility genes are much less commonly implicated; together, this group accounts for fewer than 10% of hereditary breast carcinomas (see Table 23-2). Li-Fraumeni syndrome (due to germline mutations in p53) and Li-Fraumeni variant syndrome (due to germline mutations in CHEK2) together account for about 8% of breast cancers caused by single genes. Three other tumor suppressor genes, PTEN(Cowden syndrome), LKBI/STK11 (Peutz-Jeghers syndrome), and ATM (ataxia telangiectasia), are mutated in less than 1% of all breast cancers and are described elsewhere.

The known high-risk breast cancer genes account for only about one quarter of familial breast cancers. The search for a high-risk "BRCA3" gene has been unsuccessful, and other highly penetrant genes may not exist. As a result, it is likely that the remaining familial cancers are caused by multiple genes with weak effects. As with other multigenic diseases, genome-wide association studies (GWAS) have commenced and have identified a number of candidate genes associated with risk, including the fibroblast growth factor receptor-2 (FGFR2).24 Such studies will need to take into account genetic variation across different ethnic groups, which (as we have seen) correlates with both overall breast cancer risk and susceptibility to particular molecular subtypes.

The major susceptibility genes for breast cancer are tumor suppressors that have normal roles in DNA repair, cell cycle control, and the regulation of apoptosis in many tissues (Chapter 7). Except for p53, mutations in genes implicated in hereditary breast cancer are uncommon in sporadic breast cancers. However, decreased expression of BRCA1 and CHEK2 is common in sporadic cancers, particularly those that are "triple-negative" or poorly differentiated, and basal-like cancers, which comprise a large subset of the triple-negative group, have a gene expression profile that bears a striking resemblance to hereditary cancers arising inBRCA1 carriers. Based on these observations, it is suspected that the pathways that these genes participate in are frequently disrupted in sporadic cancers through currently unknown mechanisms.

page 1078

page 1079

Sporadic Breast Cancer

The major risk factors for sporadic breast cancer are related to hormone exposure: gender, age at menarche and menopause, reproductive history, breastfeeding, and exogenous estrogens. The majority of sporadic cancers occur in

postmenopausal women and are ER positive.

Hormonal exposure increases the number of potential target cells by stimulating breast growth during puberty, menstrual cycles, and pregnancy. Exposure also drives cycles of proliferation that place cells at risk for DNA damage. Once premalignant or malignant cells are present, hormones can stimulate their growth, as well as the growth of normal epithelial and stromal cells that may aid and abet tumor development.

Estrogen may also play a more direct role in carcinogenesis. Metabolites of estrogen can cause mutations or generate DNA-damaging free radicals in cell and animal model systems.26 It also has been proposed that variants of genes involved in estrogen synthesis and metabolism could increase the risk of breast cancer. Such variants would be analogous to cytochrome P-450 alleles that alter the metabolism of tamoxifen in some women.27

Overview of Carcinogenesis and Tumor Progression

The diverse histologic appearances of carcinomas and putative precursor lesions are the outward manifestations of the complex genetic and epigenetic changes that drive carcinogenesis. One model of carcinogenesis postulates that a normal cell must acquire several new capabilities to become malignant (see Chapter 7).28,29 Each may be achieved by a change in the activity of one of many different genes that regulate common cellular activities.

Populations of cells that harbor some, but not all, of the genetic and epigenetic changes that are required for carcinogenesis give rise to morphologically recognizable breast lesions (discussed earlier) that are associated with an increased risk of progression to cancer. The earliest such alterations are proliferative changes, which may stem from the loss of growth-inhibiting signals, aberrant increases in pro-growth signals, or decreased apoptosis. For example, most early lesions (such as atypical ductal hyperplasia and atypical lobular hyperplasia) show increased expression of hormone receptors and abnormal regulation of proliferation.10,30 LOH is rarely detected in typical proliferative change but becomes more frequent in atypical hyperplasias and is almost universally present in carcinoma in situ. Profound DNA instability in the form of aneuploidy, which manifests morphologically by nuclear enlargement, irregularity, and hyperchromasia, is observed only in high-grade DCIS and some invasive carcinomas. At some point during tumor progression the malignant clone also becomes immortalized and acquires the ability to drive neo-angiogenesis. The morphologic and biologic features of carcinomas are usually established at the in situ stage, since in the majority of cases the in situ lesion closely resembles the accompanying invasive carcinoma.

The cell of origin of breast cancers is of interest, since this has important implications for etiology and treatment. The "cancer stem cell hypothesis" proposes that malignant changes occur in a stem cell population that has unique properties distinguishing them from more differentiated cells.31,32 Although the majority of tumor cells would consist of non-stem cell progeny, only the malignant stem cells would contribute to tumor progression or recurrence. Effective treatment would need to target only this population,

which to date has been difficult to define.

The most likely cell type of origin for the majority of carcinomas is the ER-expressing luminal cell, since the majority of cancers are ER-positive and precursor lesions, such as atypical hyperplasias, are most similar to this type of cell (Fig. 23-15). ER-negative carcinomas may arise from ER-negative myoepithelial cells.33,34 This would explain why many proteins found in myoepithelial cells are shared by the "triple-negative" or basal-like cancers. An alternative possibility is an origin from an ER-positive precursor that loses ER expression.10,35 The precursor lesion for ER-negative tumors is unknown (see Fig. 23-15).

The final step of carcinogenesis, the transition of carcinoma in situ to invasive carcinoma, is the most important and unfortunately the least understood. Genetic markers specific for invasive carcinomas have been difficult to identify. It is important to remember that the structure and function of the normal breast depend on a complex interplay between luminal cells, myoepithelial cells, and stromal cells. The same molecular events that allow for the normal formation of new ductal branch points and lobules during puberty and pregnancy-abrogation of the basement membrane, increased proliferation, escape from growth inhibition, angiogenesis, and invasion of stroma-may be recapitulated during carcinogenesis.2 Remodeling of the breast, which involves inflammatory and "wound healing-like" tissue reactions, could explain the transient increase in breast cancers during and shortly after pregnancy, since such changes could facilitate the transition of carcinoma in situ to invasive cancer.5,36,37

As can be surmised from this discussion, there are many paths that can lead to the development of breast cancer. Breast cancer is not one disease, but many, each with its own clinical characteristics and optimal prevention and treatment strategies. This recognition has led to the introduction of molecular classification systems, which are discussed below.

CLASSIFICATION OF BREAST CARCINOMA

Greater than 95% of breast malignancies are adenocarcinomas, which are divided into in situ carcinomas and invasive carcinomas. Carcinoma in situ refers to a neoplastic proliferation that is limited to ducts and lobules by the basement membrane. Invasive carcinoma (synonymous with "infiltrating" carcinoma) has penetrated through the basement membrane into stroma. Here, the cells have the potential to invade into the vasculature and thereby reach regional lymph nodes and distant sites.

page 1079

page 1080

Figure 23-15 Proposed precursor-carcinoma sequences in breast cancer. Morphologic changes are displayed from left to right according to the risk for subsequent invasive carcinoma.

Despite evidence that all breast carcinomas arise from cells in the terminal duct lobular unit38, the use of the terms lobular and ductal to describe both in situ and invasive carcinomas persists. Carcinoma in situ was originally classified as ductal or lobular based on the resemblance of the involved spaces to normal ducts or lobules. However, it is now recognized that varied patterns of growth in situ are not related to the site or cell of origin, but rather reflect differences in tumor cell biology, such as whether the tumor cells express the cell adhesion protein E-cadherin or not. By current convention, "lobular" refers to carcinomas of a specific type, and "ductal" is used more generally for adenocarcinomas that have no other designation.

Carcinoma in Situ

Ductal Carcinoma in Situ (DCIS; Intraductal Carcinoma)

With the advent of mammographic screening, the diagnosis of DCIS rapidly increased from fewer than 5% of all carcinomas to 15% to 30% of carcinomas in well-screened populations (see Fig. 23-13).39 Among cancers detected mammographically, almost half are DCIS. Most are detected as a result of calcifications; less commonly, periductal fibrosis surrounding DCIS forms a mammographic density or a vaguely palpable mass. Rarely, DCIS (often of micropapillary type) produces a nipple discharge or is detected as an incidental finding upon biopsy for another lesion.

DCIS consists of a malignant clonal population of cells limited to ducts and lobules by the basement membrane. The myoepithelial cells are preserved, although they may be diminished in number. DCIS can spread throughout ducts and lobules and produce extensive lesions involving an entire sector of a breast. When DCIS involves lobules, the acini are usually distorted and unfolded and take on the appearance of small ducts.

Morphology. Historically, DCIS has been divided into five architectural subtypes: comedocarcinoma, solid, cribriform, papillary,

and micropapillary. Some cases of DCIS have a single growth pattern, but the majority show a mixture of patterns.

Comedocarcinoma is characterized by the presence of solid sheets of pleomorphic cells with "high-grade" hyperchromatic nuclei and areas of central necrosis (see Fig. 23-16C). The necrotic cell membranes commonly calcify and are detected on mammography as clusters or linear and branching microcalcifications (Fig. 23-16A). Periductal concentric fibrosis and chronic inflammation are common, and extensive lesions are sometimes palpable as an area of vague nodularity (Fig. 23-16B).

Noncomedo DCIS consists of a monomorphic population of cells with nuclear grades ranging from low to high. Several morphologic variants can be seen. In cribriform DCIS, intraepithelial spaces are evenly distributed and regular in shape (cookie cutter-like) (Fig. 23-17A). Solid DCIS completely fills the involved spaces (Fig. 23-17B). Papillary DCIS grows into spaces along fibrovascular cores that typically lack the normal myoepithelial cell layer (Fig. 23-18A). Micropapillary DCIS is recognized by bulbous protrusions without a fibrovascular core, often arranged in complex intraductal patterns (Fig. 23-18B). Calcifications may be associated with central necrosis but more commonly form on intraluminal secretions.

page 1080

page 1081

Paget disease of the nipple is a rare manifestation of breast cancer (1% to 4% of cases) and presents as a unilateral erythematous eruption with a scale crust. Pruritus is common, and the lesion may be mistaken for eczema. Malignant cells (Paget cells) extend from DCIS within the ductal system, via the lactiferous sinuses, into nipple skin without crossing the basement membrane (Fig. 23-19). The tumor cells disrupt the normal epithelial barrier, allowing extracellular fluid to seep out onto the nipple surface. The Paget cells are readily detected by nipple biopsy or cytologic preparations of the exudate.

A palpable mass is present in 50% to 60% of women with Paget disease, and almost all of these women have an underlying invasive carcinoma. In contrast, the majority of women without a palpable mass have only DCIS. The carcinomas are usually poorly differentiated, ER negative, and overexpress HER2/neu.

Prognosis of Paget disease depends on the features of the underlying carcinoma and is not affected by the presence or absence of DCIS involving the skin when matched for other prognostic factors.

DCIS with microinvasion is diagnosed when there is an area of

invasion through the basement membrane into stroma measuring no more than 0.1 cm. Microinvasion is most commonly seen in association with comedocarcinoma. If only one or a few foci of microinvasion are present, the prognosis is very similar to DCIS.

Figure 23-16 Ductal carcinoma in situ (DCIS) comedo type. A, The specimen radiogram reveals linear and branching calcifications within the ductal system. B, Ducts filled with punctate areas of necrosis ("comedone" like) and surrounded by periductal fibrosis are

seen. C, DCIS with large central zones of necrosis and calcifications fills several adjacent ducts.

Figure 23-17 Noncomedo DCIS. A, Cribriform DCIS composed of cells forming round, regular ("cookie cutter") spaces. The lumens are filled with calcifying secretory material. B, This solid DCIS has almost completely filled and distorted this lobule with only a few remaining normal luminal cells visible. This type of DCIS is not usually associated with calcifications and may be clinically occult.

The natural history of DCIS has been difficult to determine because, until recently, all women were treated with mastectomy, and the current practice of surgical excision, usually followed by radiation, is largely curative. If untreated, women with small, low-grade DCIS develop invasive cancer at a rate of about 1% per year.40 The majority of these cancers are in the same quadrant and have a similar grade and expression pattern of ER and HER2/neu as the DCIS. It is assumed that women with high-grade or extensive DCIS progress to invasive carcinoma at higher rates. Specific biologic features that predict recurrence or progression to invasion are being sought so as to target treatment to these patients.35

Mastectomy for DCIS is curative for over 95% of patients. Rare recurrences and/or death are usually due to residual DCIS in ducts in subcutaneous adipose tissue that was not removed during surgery, or occult foci of invasion that were not detected at the time of diagnosis.

page 1081

page 1082

Figure 23-18 Noncomedo DCIS. A, Papillary DCIS. Delicate fibrovascular cores extend into a duct and are lined by a monomorphic population of tall columnar cells. Myoepithelial cells are absent. B, Micropapillary DCIS. The papillae are connected to the duct wall

by a narrow base and often have bulbous or complex outgrowths. The papillae are solid and lack fibrovascular cores.

Figure 23-19 Paget disease of the nipple. DCIS arising within the ductal system of the breast can extend up the lactiferous ducts and into the skin of the nipple without crossing the basement membrane. The malignant cells disrupt the normally tight squamous

epithelial cell barrier, allowing extracellular fluid to seep out and form an oozing scaly crust.

Breast conservation is appropriate for most women with DCIS but results in a slightly higher risk of recurrence. The major risk factors for recurrence are (1) grade, (2) size, and (3) margins. However, if wide margins (i.e., at least 1 cm) can be achieved, the rate of recurrence is quite low. Complete excision of DCIS presents a challenge, since its extent can only be reliably predicted by pathologic evaluation. Postoperative radiation therapy and tamoxifen also reduce the risk of recurrence. The benefit of tamoxifen may be restricted to women with ER-positive DCIS.41 If DCIS is treated adequately, the risk of recurrence in the same breast is only slightly higher than the risk in the contralateral breast for subsequent carcinoma.42 Whatever the treatment, deaths from breast cancer are very uncommon, occurring in fewer than 2% of women with DCIS.

Lobular Carcinoma in Situ (LCIS)

LCIS is always an incidental biopsy finding, since it is not associated with calcifications or stromal reactions that produce mammographic densities. As a result, its incidence (1% to 6% of all carcinomas) has not been affected by the introduction of mammographic screening. When both breasts are biopsied, LCIS is bilateral in 20% to 40% of cases, compared with 10% to 20% of cases of DCIS. LCIS is more common in young women, with 80% to 90% of cases occurring before menopause.

The cells of LCIS and invasive lobular carcinoma are identical in appearance and share genetic abnormalities, such as those that lead to loss of expression of E-cadherin, a transmembrane cell adhesion protein that contributes to the cohesion of normal breast epithelial cells.

page 1082

page 1083

Morphology. Atypical lobular hyperplasia, LCIS, and invasive lobular carcinoma all consist of dyscohesive cells with oval or round nuclei and small nucleoli (Fig. 23-20A). The cells lack the cell adhesion protein E-cadherin, resulting in the cells appearing rounded without attachment to adjacent cells (Fig. 23-20B). Mucin-positive signet-ring cells are commonly present. LCIS rarely distorts the underlying architecture, and the involved acini remain recognizable as lobules. LCIS almost always expresses ER and PR. Overexpression of HER2/neu is not observed.

Figure 23-20 Lobular carcinoma in situ. A, A monomorphic population of small, rounded, loosely cohesive cells fills and expands the acini of a lobule. The underlying lobular architecture can still be recognized. The cells extend into the adjacent lobule by pagetoid

spread. B, An immunoperoxidase study shows E-cadherin-positive normal luminal cells that have been undermined by E-cadherin-negative LCIS cells spreading along the basement membrane.

Women with LCIS develop invasive carcinomas at a frequency similar to that of women with untreated DCIS. In patients observed for more than 20 years, invasive carcinoma develops in 25% to 35%, or at about 1% per year. Although both breasts are at increased risk, the risk is slightly higher in the ipsilateral breast.42-44 Invasive carcinomas developing in women after a diagnosis of LCIS are threefold more likely to be of the lobular type, but the majority do not show specific lobular morphology. Treatment choices include bilateral prophylactic mastectomy, tamoxifen, or, more typically, close clinical follow-up and mammographic screening.

Rare cases of carcinoma in situ that lack E-cadherin have high-grade nuclei and/or central necrosis. The cells may be ER negative, and some overexpress HER2/neu. The natural history of this type of CIS is not known and may well be different from typical LCIS.44

Invasive (Infiltrating) Carcinoma

In the absence of mammographic screening, invasive carcinoma almost always presents as a palpable mass. Palpable tumors are associated with axillary lymph node metastases in over 50% of patients. Larger carcinomas may be fixed to the chest wall or cause dimpling of the skin. When the tumor involves the central portion of the breast, retraction of the nipple may develop. Lymphatics may become so involved as to block the local area of skin drainage and cause lymphedema and thickening of the skin. In such cases, tethering of the skin to the breast by Cooper ligaments mimics the appearance of an orange peel, an appearance referred to as peau d'orange.

In older women undergoing mammography, invasive carcinomas most commonly present as a radiodense mass (Fig. 23-21A). Mammographically detected cancers are, on average, half the size of palpable cancers. Fewer than 20% will have nodal metastases. Invasive carcinomas presenting as mammographic calcifications without an associated density are very small in size, and metastases are unusual.

The term inflammatory carcinoma is reserved for tumors that present with a swollen, erythematous breast. This gross appearance is caused by extensive invasion and obstruction of dermal lymphatics by tumor cells. The underlying carcinoma is usually diffusely infiltrative and typically does not form a discrete palpable mass. This can result in confusion with true inflammatory conditions and a delay in diagnosis. Many patients have metastases at diagnosis or recur rapidly, and the overall prognosis is poor.45

Rarely, breast cancer presents as an axillary nodal metastasis or distant metastasis before cancer is detected in the breast. In most cases, the primary carcinoma is either small or obscured by dense breast tissue. The number of primary carcinomas that remain occult in such cases has been minimized with imaging using mammography, ultrasound, and MRI.

The most common histologic types of breast adenocarcinoma are listed in Table 23-3. These special types are important to recognize because of their specific clinical associations.

Invasive Carcinoma, No Special Type (NST; Invasive Ductal Carcinoma)

Invasive carcinomas of no special type include the majority of carcinomas (70% to 80%).

Morphology. On gross examination, most tumors are firm to hard and have an irregular border (Fig. 23-21B). When cut or scraped, they typically produce a characteristic grating sound (similar to cutting a water chestnut) due to small, central pinpoint foci or streaks of chalky-white elastotic stroma and occasional small foci of calcification. Less frequently, carcinomas have a well-circumscribed border and a softer consistency.

page 1083

page 1084

There is a wide range of histologic appearances. Well-differentiated carcinomas show prominent tubule formation, small round nuclei, and rare mitotic figures (Fig. 23-22A). Moderately differentiated carcinomas may have tubules, but solid clusters or single infiltrating cells are also present. These tumors have a greater degree of nuclear pleomorphism and contain mitotic figures (Fig. 23-22B). Poorly differentiated carcinomas often invade as ragged nests or solid sheets of cells with enlarged irregular nuclei. A high proliferation rate and areas of tumor necrosis are common (Fig. 23-22C).

Figure 23-21 Invasive ductal carcinoma. A, The radiograph shows an invasive cancer with a characteristic irregular border. B, Grossly, the irregular firm white mass contains chalky areas of elastotic stroma that extend out into the surrounding yellow adipose

tissue. (B, Courtesy of Dr. Anna Laury, Brigham and Women's Hospital, Boston, MA.)

Table 23-3. Distribution of Histologic Types of Breast Cancer

Total Cancers Percentage

CARCINOMA IN SITU* 15-30

Ductal carcinoma in situ 80

Lobular carcinoma in situ 20

INVASIVE CARCINOMA 70-85

No-special-type carcinoma ("ductal") 79

Lobular carcinoma 10

Tubular/cribriform carcinoma 6

Mucinous (colloid) carcinoma 2

Medullary carcinoma 2

Papillary carcinoma 1

Metaplastic carcinoma <1

*The proportion of in situ carcinomas detected depends on the percentage of women undergoing mammographic screening and ranges from less than 5% in unscreened populations to almost 50% in populations that are well screened. Current observed numbers are between these two extremes. The data on invasive carcinomas are modified from Dixon JM et al.: Long-term survivors after breast cancer. Br J Surg 72:445, 1985.

page 1084

page 1085

Recently developed techniques that examine the DNA, RNA, and proteins of

carcinomas globally have provided a framework for new molecular classifications of this group of breast cancers (Fig. 23-23). Gene expression profiling, which can measure the relative quantities of mRNA for essentially every gene, has identified five major patterns of gene expression in the NST group: luminal A, luminal B, normal, basal-like, and HER2 positive.46 These molecular classes correlate with prognosis and response to therapy, and thus have taken on clinical importance.

"Luminal A" (40% to 55% of NST cancers): This is the largest group and consists of cancers that are ER positive and HER2/neu negative. The gene signature is dominated by the dozens of genes under the control of ER (see Fig. 23-23). ER-positive carcinomas also show increased transcription of genes thought to be characteristic of normal luminal cells. The majority are well- or moderately differentiated, and most occur in postmenopausal women. These cancers are generally slow growing and respond well to hormonal treatments. Conversely, only a small number will respond to standard chemotherapy. Commercial tests, some already available for use with formalin-fixed tissues, have been developed to identify this and other molecular classes.47 In addition, clinical trials are attempting to identify different types or combinations of chemotherapeutic agents that may be efficacious for ER-positive cancers.

"Luminal B" (15% to 20% of NST cancers): This group of cancers also expresses ER but is generally of higher grade, has a higher proliferative rate, and often overexpresses HER2/neu. They are sometimes referred to as triple-positive cancers. They compose a major group of ER-positive cancers that are more likely to have lymph node metastases and that may respond to chemotherapy.

"Normal breast-like" (6% to 10% of NST cancers): This is a small group of usually well-differentiated ER-positive, HER2/neu-negative cancers characterized by the similarity of their gene expression pattern to normal tissue. It is not yet clear whether or not this is a specific tumor expression pattern.

"Basal-like" (13% to 25% of NST cancers): These cancers are notable for the absence of ER, PR, and HER2/neu and the expression of markers typical of myoepithelial cells (e.g., basal keratins, P-cadherin, p63, or laminin), progenitor cells, or putative stem cells (e.g., cytokeratins 5 and 6) (see Fig. 23-23). "Basal" was chosen as a general term that covers all of these cell types. By strict definition this group is defined by their gene expression profile. Basal-like cancers are a subgroup of ER-PR-HER2/neu "triple-negative" carcinomas.48,49 Members of this group include medullary carcinomas, metaplastic carcinomas (e.g., spindle cell carcinomas or matrix-producing carcinomas), and carcinomas with a central fibrotic focus. Basal-like cancers are of particular interest because of their distinct genetic and epidemiologic features. Many carcinomas arising in women with BRCA1 mutations are of this type. There is also an increased incidence in certain ethnic populations and in young women. These cancers are generally high grade and have a high proliferation rate. They

are associated with an aggressive course, frequent metastasis to viscera and the brain, and a poor prognosis. However, approximately 15% to 20% will have a pathologic complete response to chemotherapy; cure may be possible in this chemosensitive subgroup.

"HER2 positive" (7% to 12% of NST cancers): This group comprises ER-negative carcinomas that overexpress HER2/neu protein. In over 90% of HER2/neu positive cancers, overexpression is due to amplification of the segment of DNA on 17q21 that includes the HER2/neu gene and varying numbers of adjacent genes. This amplicon dominates the gene signature of this group (see Fig. 23-23). HER2/neu assays, which include measurement of gene copy number by fluorescence in situ hybridization, mRNA level by gene arrays, and protein by immunohistochemistry, are all abnormal in the majority of these cancers. In rare cases, HER2/neu protein overexpression may occur as a result of mechanisms other than gene amplification.50 These cancers are usually poorly differentiated, have a high proliferation rate, and are associated with a high frequency of brain metastasis.

Figure 23-22 A, A well-differentiated invasive carcinoma of no special type consists of tubules or a cribriform pattern of cells with small monomorphic nuclei. B, A moderately differentiated carcinoma shows less tubule formation and more solid nests of cells and pleomorphic nuclei. C, This poorly differentiated invasive carcinoma of no special type infiltrates as ragged sheets of pleomorphic

cells with numerous mitotic figures and central areas of tumor necrosis.

Trastuzumab (Herceptin) is a humanized monoclonal antibody specific to HER2/neu. The combination of trastuzumab and chemotherapy is highly effective in treating carcinomas that overexpress HER2/neu. Demonstrating the first gene-targeted therapeutic agent for a solid tumor, these results have created great excitement within the community of physicians and scientists involved with treating cancer patients. Unfortunately, trastuzumab does not cross the blood-brain barrier, leaving patients susceptible to metastatic disease to this site. Newer agents, such as the dual tyrosine

kinase inhibitor lapatinib, that targets both EGFR and HER2/neu, will hopefully overcome these limitations.51 Other genes on the same segment of amplified DNA may influence the sensitivity of HER2 tumors to these agents.

Invasive Lobular Carcinoma

Invasive lobular carcinomas usually present as a palpable mass or a mammographic density with irregular borders. However, in about one fourth of cases the tumor infiltrates the tissue diffusely and causes little desmoplasia. Such tumors are difficult to detect by palpation and may cause only very subtle mammographic changes. Metastases can also be difficult to detect clinically and radiologically because of this type of invasion.

page 1085

page 1086

Figure 23-23 Gene expression portraits of breast carcinomas. Alterations in DNA, messenger RNA (mRNA), and protein expression identify breast cancer subtypes previously recognized by morphology (e.g., lobular carcinomas) and define new subtypes ("luminal

A," "HER2/neu positive," and "basal-like"). (Array data courtesy of Dr. Andrea Richardson, Brigham and Women's Hospital, Boston, MA, as modified from Signoretti S et al.:

Oncogenic role of the ubiquitin ligase subunit skp2 in human breast cancer. J Clin Invest 110:633, 2002.)

page 1086

page 1087

Lobular carcinomas have been reported to have a greater incidence of bilaterality. However, many studies have been biased by the greater likelihood of performing contralateral surgery in women with lobular carcinoma. The actual fraction of women who develop invasive carcinomas in the contralateral breast is only 5% to 10%, which is similar to the incidence for NST carcinomas.

Morphology. The histologic hallmark is the presence of

dyscohesive infiltrating tumor cells, often arranged in single file or in loose clusters or sheets (see Fig. 23-23). Tubule formation is absent. The cytologic appearance is identical to the cells of atypical lobular hyperplasia and LCIS. Signet-ring cells containing an intracytoplasmic mucin droplet are common. Desmoplasia may be minimal or absent.

Invasive lobular carcinoma is graded using the same criteria as those applied to other breast carcinomas.52 Well-differentiated and moderately differentiated invasive lobular carcinomas are usually diploid, ER positive, and associated with LCIS. HER2/neu overexpression is very rare. These cancers have a gene expression profile similar to luminal A cancers (see Fig. 23-23).53 In contrast, poorly differentiated lobular carcinomas are generally aneuploid, often lack hormone receptors, and may overexpress HER2/neu. If matched by grade and stage, lobular carcinomas have the same prognosis as NST carcinomas.

Lobular carcinomas have a different pattern of metastasis than other breast cancers. Metastasis tends to occur to the peritoneum and retroperitoneum, the leptomeninges (carcinoma meningitis), the gastrointestinal tract, and the ovaries and uterus.53 In some cases, metastatic lobular carcinoma may be mistaken for signet ring carcinoma of the GI tract, which it closely resembles. The morphologic resemblance of these two tumors is not coincidental, but rather reflects a common underlying molecular etiology. Both lobular carcinoma and signet ring carcinoma of the gastrointestinal tract are characterized by the loss of E-cadherin, a cell adhesion molecule that functions as a tumor suppressor. In lobular carcinoma, biallelic loss of expression of CDH1, the gene that encodes E-cadherin, stems from a combination of deletions, mutations, and promoter silencing via methylation. Loss of E-cadherin is also seen in atypical lobular hyperplasia and LCIS, indicating that this alteration is a relatively early event in the development of lobular carcinoma. Rare patients with heterozygous germline mutations in CDH1 are at very high risk of developing lobular carcinoma (if female) and gastric signet ring carcinoma (males and females), emphasizing the close molecular relationship between these two tumors and the importance of E-cadherin loss in their pathogenesis.54,55

Medullary Carcinoma

Medullary carcinoma is most common in women in the sixth decade and presents as a well-circumscribed mass. It may closely mimic a benign lesion clinically and radiologically, or present as a rapidly growing mass.

Morphology. These tumors produce little desmoplasia and are distinctly more yielding on palpation and cutting than typical breast carcinomas. The tumor is soft, fleshy (medulla is Latin for "marrow"), and well circumscribed. Histologically, the carcinoma is characterized by (1) solid, syncytium-like sheets of large cells with vesicular, pleomorphic nuclei, and prominent nucleoli, which compose more than 75% of the tumor mass; (2) frequent mitotic

figures; (3) a moderate to marked lymphoplasmacytic infiltrate surrounding and within the tumor; and (4) a pushing (noninfiltrative) border (Fig. 23-24C). All medullary carcinomas are poorly differentiated. DCIS is minimal or absent.

Medullary carcinomas have a slightly better prognosis than do NST carcinomas, despite the almost universal presence of poor prognostic factors, including high nuclear grade, aneuploidy, absence of hormone receptors, and high proliferative rates. HER2/neu overexpression is not observed. Lymph node metastases are infrequent and rarely involve multiple nodes. The syncytial growth pattern and pushing borders may stem from the overexpression of adhesion molecules, such as intercellular cell adhesion molecule and E-cadherin, which could potentially limit metastatic potential.53

Medullary carcinomas have a basal-like gene expression profile.56 Among cancers arising in BRCA1 carriers, 13% are of medullary type, and up to 60% have a subset of medullary features (see Table 23-3). Although, the majority of medullary carcinomas are not associated with germline BRCA1 mutations, hypermethylation of the BRCA1 promoter is observed in 67% of medullary carcinomas, suggesting an association of this morphology with underlying gene expression.

Mucinous (Colloid) Carcinoma

These carcinomas occur in older women (median age 71) and tend to grow slowly over the course of many years.

Morphology. The tumor is soft or rubbery and has the consistency and appearance of pale gray-blue gelatin. The borders are pushing or circumscribed. The tumor cells are arranged in clusters and small islands of cells within large lakes of mucin (Fig. 23-24D).

Mucinous carcinomas are usually diploid, well to moderately differentiated, and ER positive. Lymph node metastases are uncommon. The overall prognosis is slightly better than that of NST carcinomas.

Tubular Carcinoma

Tubular carcinomas are typically detected as small irregular mammographic densities in women in their late 40s. They are uncommon, but constitute up to 10% of tumors that are smaller than 1 cm in size. In a significant minority of cases, tumors are multifocal within one breast or detected bilaterally.

page 1087

page 1088

Morphology. These tumors consist exclusively of well-formed tubules and are sometimes mistaken for benign sclerosing lesions (Fig. 23-25). However, the myoepithelial cell layer is absent, placing the tumor cells in direct contact with the stroma. A cribriform pattern is sometimes seen. Apocrine snouts are typical, and calcifications may be present within the lumens. Tubular carcinomas are

frequently associated with atypical lobular hyperplasia, LCIS, or low-grade DCIS.8

Figure 23-24 Invasive carcinoma variants. A, The specimen radiogram shows a well-circumscribed mass. The majority of such masses are benign, but approximately 6% are carcinomas. B, Grossly, this carcinoma has a pushing border and a fleshy

appearance. C, Medullary carcinoma. Note the pushing border, the sheetlike growth of the pleomorphic tumor cells, and the prominent lymphoplasmacytic infiltrate. D, Mucinous (colloid) carcinoma. The malignant cells lie within pools of extracellular mucin.

The tumor also has a pushing border and is very deceptively soft in texture.

Figure 23-25 Tubular carcinoma. This carcinoma is composed of well-formed angulated tubules lined by a single layer of cells with small uniform nuclei.

More than 95% of all tubular carcinomas are diploid, ER positive, and HER2/neu negative. By definition, all are well differentiated. Axillary metastases occur in fewer than 10% of cases unless multiple foci of invasion are present. This subtype is important to recognize because of its excellent prognosis.

Invasive Papillary Carcinoma

Invasive papillary carcinomas and invasive micropapillary carcinomas are rare, representing 1% or fewer of all invasive cancers. Papillary or micropapillary architecture is more commonly seen in DCIS. Invasive papillary carcinomas are usually ER positive and have a favorable prognosis. In contrast, invasive micropapillary carcinomas are more likely to be ER negative and HER2 positive. Lymph node metastases are very common, and the prognosis is poor.

Metaplastic Carcinoma page 1088

page 1089

"Metaplastic carcinoma" includes a variety of rare types of breast cancer (<1% of all cases), such as matrix-producing carcinomas, squamous cell carcinomas, and carcinomas with a prominent spindle cell component. They are ER-PR-HER2/neu "triple negative," often express myoepithelial proteins, and appear to be related to the basal-like carcinomas. Lymph node metastases are infrequent, but the prognosis is generally poor.

PROGNOSTIC AND PREDICTIVE FACTORS

The outcome for women with breast cancer varies widely. Many women have a normal life expectancy, whereas others have only a 10% chance of being alive in 5 years. Except in women who present with distant metastasis (<10%) or with inflammatory carcinoma (<5%) (in whom the prognosis is poor regardless of other findings), prognosis is determined by the pathologic examination of the primary carcinoma and the axillary lymph nodes. Prognostic information is important in counseling patients about the likely outcome of their disease, choosing appropriate treatment, and the design of clinical trials.

Major prognostic factors that are the strongest predictors of death from breast cancer are incorporated into the American Joint Committee on Cancer (AJCC) staging system,57 which is used to divide patients into five stages (0 to IV) that are correlated with survival (Table 23-4). The major prognostic factors are as follows:

1. Invasive carcinoma versus in situ disease. By definition, in situ carcinoma is confined to the ductal system and cannot metastasize. Breast cancer deaths associated with DCIS are due to the subsequent development of invasive carcinoma or areas of invasion that were not detected at the time of diagnosis. The great majority of women with adequately treated DICS are cured. In contrast, at least half of invasive carcinomas have metastasized locally or distantly at the time of diagnosis.

2. Distant metastases. Once distant metastases are present, cure is unlikely, although long-term remissions and palliation can be achieved, especially in women with hormonally responsive tumors. As mentioned earlier, the tumor type influences the timing and location of metastases.58,59

3. Lymph node metastases. Axillary lymph node status is the most important prognostic factor for invasive carcinoma in the absence of distant metastases. The clinical assessment of lymph node status is unreliable due to both false positives (e.g., palpable reactive nodes) and false negatives (e.g., lymph nodes with small metastatic deposits). Therefore, biopsy is necessary for accurate assessment. With no nodal involvement, the 10-year disease-free survival rate is close to 70% to 80%; the rate falls to 35% to 40% with one to three positive nodes, and to 10% to 15% when more than 10 nodes are positive. Lymphatic vessels in most breast carcinomas drain first to one or two sentinel nodes, which can be identified with radiotracer or colored dyes. If a biopsy restricted to the sentinel nodes is negative for metastasis, it is unlikely that other more distant nodes will be involved and the patient can be spared the morbidity of a complete axillary dissection. For these reasons, sentinel node biopsy has been adopted in many centers as part of the assessment of lymph node status. In some tumors of the medial breast, the sentinel node is an intrathoracic internal mammary node. These nodes are generally not biopsied owing to the morbidity associated with the procedure. Macrometastases (greater than 0.2 cm) are of proven prognostic importance. Through more sensitive approaches, including serial sectioning of lymph nodes,

immunohistochemistry for keratins, and RT-PCR-based detection of tumor-specific mRNA, increased numbers of women with micrometastases (0.2 cm or less) are being identified. The clinical significance of these small metastases is unclear and is being addressed by current clinical trials. Approximately 10% to 20% of women without axillary lymph node metastases have a recurrence outside of the breast and about the same number die from breast cancer. In these patients, metastasis may occur via the internal mammary lymph nodes or hematogenously.

4. Tumor size. The size of an invasive carcinoma is the second most important prognostic factor. The risk of axillary lymph node metastases increases with the size of the primary tumor, but both are independent prognostic factors. Women with node-negative carcinomas <1 cm in size have a 10-year survival rate of over 90%, whereas survival drops to 77% for cancers >2 cm. Unfortunately, breast self-examination does not lower breast cancer mortality,60 suggesting that by the time breast cancers become palpable (typically when at least 2 to 3 cm), tumors capable of metastasizing have already done so. Mammographically detected cancers are smaller and less likely to have metastasized.

5. Locally advanced disease. Carcinomas invading into skin or skeletal muscle are usually large and may be difficult to treat surgically. With increased awareness of breast cancer detection, such cases have fortunately decreased in frequency and are now rare at initial presentation.

6. Inflammatory carcinoma. Breast cancers presenting with breast swelling and skin thickening due to dermal lymphatic involvement have a particularly poor prognosis. The 3-year survival rate is only 3% to 10%. Less than 3% of cancers are in this group, but the incidence is higher in African American women and younger women.61

Minor Prognostic and Predictive Factors page 1089

page 1090

In addition to the six factors used by the AJCC, a number of other factors are predictive of outcome; some of these also direct therapies against particular molecular targets.

Histologic subtype. The 30-year survival rate of women with special types of invasive carcinomas (tubular, mucinous, medullary, lobular, and papillary) is greater than 60%, compared with less than 20% for women with NST cancers. With the exception of medullary carcinoma, most of these carcinomas will be well to moderately differentiated, ER positive, and HER2/neu negative. This favorable prognosis probably does not apply to unusual special-type carcinomas without these characteristics.

Histologic grade. The most commonly used grading system, the Nottingham Histologic Score (also referred to as Scarff-Bloom-Richardson), combines nuclear grade, tubule formation, and mitotic rate to classify invasive carcinomas

into three groups that are highly correlated with survival.52 Survival for patients with well-differentiated grade 1 carcinomas (approximately 20% of the total) gradually declines to 70% at 24 years. In contrast, most deaths for poorly differentiated grade 3 carcinomas (approximately 46% of the total) occur in the first 10 years, and 45% of patients survive long-term. Women with moderately differentiated grade 2 carcinomas (approximately 35% of the total) have better survival initially, but their long-term survival is only slightly better than grade 3 carcinomas.

Estrogen and progesterone receptors. Current assays use immunohistochemistry to detect nuclear hormone receptors, a finding that is correlated with a better outcome and is an important predictor of response to hormonal therapy (see Fig. 23-23). Eighty percent of carcinomas that are ER and PR positive respond to hormonal manipulation, whereas only about 40% of those with either ER or PR alone respond. ER-positive cancers are less likely to respond to chemotherapy. Conversely, cancers that fail to express either ER or PR have a less than 10% likelihood of responding to hormonal therapy but are more likely to respond to chemotherapy.

HER2/neu. HER2/neu overexpression is associated with poorer survival, but its main importance is as a predictor of response to agents that target this transmembrane protein (e.g., trastuzumab or lapatinib). Several different assays are used to determine HER/neu gene amplification and protein overexpression (see Fig. 23-23).

Lymphovascular invasion. Tumor cells are present within vascular spaces (either lymphatics or small capillaries) in about half of all invasive carcinomas. This finding is strongly associated with the presence of lymph node metastases. It is a poor prognostic factor for overall survival in women without lymph node metastases and a risk factor for local recurrence. As already mentioned, extensive plugging of the lymphovascular spaces of the dermis with carcinoma cells (inflammatory carcinoma) bodes a very poor prognosis.

Proliferative rate. Proliferation can be measured by mitotic counts (e.g., as part of histologic grading), by immunohistochemical detection of cellular proteins produced during the cell cycle (e.g., cyclins, Ki-67), by flow cytometry (as the S-phase fraction), or by thymidine labeling index. Carcinomas with high proliferation rates have a poorer prognosis but may respond better to chemotherapy.

DNA content. The amount of DNA per tumor cell can be determined by flow-cytometric analysis or by image analysis of tissue sections. Tumors with a DNA index of 1 have the same total amount of DNA as normal diploid cells, although marked karyotypic changes may be present. Aneuploid tumors are those with abnormal DNA indices and have a slightly worse prognosis.

Response to neoadjuvant therapy. Most patients complete their surgery and subsequently receive systemic treatment (referred to as adjuvant therapy). Neoadjuvant therapy is an alternative approach in which the patient is treated before surgery. Although this approach does not improve survival, the degree that the tumor responds to chemotherapy is a strong prognostic factor. Clinical and radiologic examinations are useful to monitor changes during treatment, but

often underestimate or overestimate the amount of residual carcinoma. Cancers most likely to respond well are poorly differentiated, ER negative, and have areas of necrosis. The subgroup of patients who achieve a pathologic complete response (i.e., no residual cancer in the breast or lymph nodes) have a greater than 95% long-term survival, in contrast to the poor prognosis of this group as a whole.62 Pathologic response can be used as a short-term end point for clinical trials (which thus can yield useful information with fewer patients in shorter periods of time) and is being linked to research studies investigating the molecular basis of tumor sensitivity or resistance to therapy.

Gene expression profiling. Expression profiling has been shown to predict survival and recurrence-free interval, and also identifies patients who are most likely to benefit from particular types of chemotherapy. Methods that require rapidly frozen tissue will be difficult to apply in clinical practice, but alternative approaches that use formalin-fixed paraffin-embedded tissues are beginning to enter clinical practice.47

Table 23-4. AJCC Staging*

Stage T: Primary Cancer Lymph Nodes (LNs) M: Distant Metastasis

5-Year Survival (%)

0 DCIS or LCIS No metastases Absent 92

I Invasive carcinoma ≤2 cm No metastases Absent 87

II Invasive carcinoma >2 cm No metastases Absent 75

Invasive carcinoma <5 cm 1 to 3 positive LNs Absent

III Invasive carcinoma >5 cm 1 to 3 positive LNs Absent 46

Any size invasive carcinoma ≥4 positive LNs Absent

Invasive carcinoma with skin or chest wall involvement or inflammatory carcinoma

0 to >10 positive LNs Absent

IV Any size invasive carcinoma Negative or positive lymph nodes

Present 13

*The groups listed in the table are based on the characteristics of the primary carcinoma and the axillary lymph nodes. For rare women with involved internal mammary lymph nodes or supraclavicular lymph nodes, there are additional staging criteria.

57

DCIS, ductal carcinoma in situ; LCIS, lobular carcinoma in situ. page 1090

page 1091

Although gene expression profiles provide a vast amount of information about carcinomas, they are not well correlated with tumor size or lymph node status-two of the strongest prognostic factors.63 This suggests that while patterns of gene expression likely determine metastatic potential, time and chance also influence whether and when metastasis occurs. It is likely that future means of estimating prognosis will involve some combination of these "old" and "new" factors.

Current therapeutic approaches directed at local and regional control consist of combinations of surgery (mastectomy or breast conservation) and postoperative radiation, whereas attempts at systemic control rely on hormonal treatment, chemotherapy, or both. Axillary node dissection or sentinel node sampling is performed for prognostic purposes, but the axilla can also be treated with radiation alone. Newer therapeutic strategies include inhibitors of membrane-bound growth factor receptors (e.g., HER2/neu), stromal proteases, and angiogenesis.

Such therapies are based on models of breast cancer dissemination that have evolved as our understanding of its biology has changed. Earlier models proposed that breast cancer spreads in a contiguous fashion by direct extension from breast to nodes and could therefore be cured by en bloc surgical resection. However, radical surgery, including mastectomies with removal of pectoralis muscles, internal mammary nodes, and even supraclavicular nodes, failed to decrease mortality. A subsequent model, based on studies demonstrating that breast-conserving surgery and radiation were equivalent to radical mastectomy, postulated that all cancers had spread systemically by the time of diagnosis and that local or regional treatment was unimportant for overall survival. In the current era of increased detection of early-stage carcinomas by mammography, a third model that combines the first two is thought to be a more appropriate guide to therapy.64

STROMAL TUMORS

The two types of stroma in the breast, intralobular and interlobular (see the introductory section on the normal female breast), give rise to distinct types of neoplasms. The breast-specific biphasic tumors fibroadenoma and phyllodes tumor arise from intralobular stroma. This specialized stroma may elaborate growth factors for epithelial cells, resulting in the proliferation of the non-neoplastic epithelial component of these tumors. Interlobular stroma is the source of the same types of tumors found in connective tissue in other sites of the body (e.g., lipomas and angiosarcomas) as well as tumors arising more commonly in the breast (e.g., pseudoangiomatous stromal hyperplasia, myofibroblastomas, and fibrous tumors).

Fibroadenoma

This is the most common benign tumor of the female breast. Most occur in women in their 20s and 30s, and they are frequently multiple and bilateral. Young women usually present with a palpable mass and older women with a mammographic density (Fig. 23-26A) or mammographic calcifications. The epithelium of the fibroadenoma is hormonally responsive, and an increase in size due to lactational changes during pregnancy, which may be complicated by infarction and inflammation, can mimic carcinoma. The stroma often becomes densely hyalinized after menopause and may calcify. Large lobulated ("popcorn") calcifications have a characteristic mammographic appearance, but small calcifications may appear clustered and require biopsy to exclude carcinoma.

Morphology. Fibroadenomas grow as spherical nodules that are usually sharply circumscribed and freely movable. They vary in size from less than 1 cm to large tumors that can replace most of the

breast. The tumors are well-circumscribed, rubbery, grayish white nodules that bulge above the surrounding tissue and often contain slitlike spaces (Fig. 23-26B).

The delicate, cellular, and often myxoid stroma resembles normal intralobular stroma. The epithelium may be surrounded by stroma or compressed and distorted by it (Fig. 23-26C). In older women, the stroma typically becomes densely hyalinized and the epithelium atrophic.

Figure 23-26 Fibroadenoma. A, The radiogram shows a characteristically well-circumscribed mass. B, Grossly, a rubbery, white, well-circumscribed mass is clearly demarcated from the surrounding yellow adipose tissue. The absence of adipose tissue accounts

for the radiodensity of the lesion. C, The proliferation of intralobular stroma surrounds, pushes, and distorts the associated epithelium. The border is sharply delimited from the surrounding tissue.

page 1091

page 1092

Some fibroadenomas are polyclonal hyperplasias of lobular stroma due to some type of stimulus. For example, almost half of women receiving cyclosporin A after renal transplantation develop fibroadenomas. In this setting the tumors are frequently multiple and bilateral. Regression may occur after cessation of cyclosporin treatment. Other fibroadenomas are benign neoplasms associated with clonal cytogenetic aberrations that are confined to the stromal component. No consistent cytogenetic changes have been found.

Fibroadenomas were originally grouped with other "proliferative changes without atypia" in conferring a mild increase in the risk of subsequent cancer. However, in one study the increased risk was limited to fibroadenomas associated with cysts larger than 0.3 cm, sclerosing adenosis, epithelial calcifications, or papillary apocrine change ("complex fibroadenomas") (see Table 23-1).65

Phyllodes Tumor

Phyllodes tumors, like fibroadenomas, arise from intralobular stroma. Although they can occur at any age, most present in the sixth decade, 10 to 20 years later than the peak age for fibroadenomas.66 The majority are detected as palpable masses, but a few are

found by mammography. The term cystosarcoma phyllodes is sometimes used for these lesions. However, the term phyllodes tumor is preferred, since the majority of these tumors behave in a relatively benign fashion, and most are not cystic.

Figure 23-27 Phyllodes tumor. Compared to a fibroadenoma, there is increased stromal cellularity, cytologic atypia, and stromal overgrowth, giving rise to the typical leaflike architecture.

Morphology. The tumors vary in size from a few centimeters to massive lesions involving the entire breast. The larger lesions often have bulbous protrusions (phyllodes is Greek for "leaflike") due to the presence of nodules of proliferating stroma covered by epithelium (Fig. 23-27). In some tumors these protrusions extend into a cystic space. This growth pattern can also occasionally be seen in larger fibroadenomas and is not an indication of malignancy. Phyllodes tumors are distinguished from the more common fibroadenomas on the basis of cellularity, mitotic rate, nuclear pleomorphism, stromal overgrowth, and infiltrative borders. Low-grade lesions resemble fibroadenomas but are more cellular and contain mitotic figures. High-grade lesions may be difficult to distinguish from other soft-tissue sarcomas and may have foci of mesenchymal differentiation (e.g., rhabdomyosarcoma or liposarcoma). The frequency of chromosomal changes increases with grade and the majority of high-grade lesions are reported to have amplification of EGFR.67 Recurrent phyllodes tumors are often of a higher grade than the presenting lesion.

Phyllodes tumors must be excised with wide margins or by mastectomy to avoid local recurrences. Axillary lymph node dissection is not indicated, because the incidence of nodal metastases (as for other stromal malignancies) is exceedingly small. The majority are low-grade tumors that may recur locally but only rarely metastasize. Rare high-grade lesions behave aggressively, with frequent local recurrences and distant hematogenous metastases in about one third of cases. Only the stromal component

metastasizes.

Benign Stromal Lesions

Tumors of the interlobular stroma of the breast are composed of stromal cells without an accompanying epithelial component. Pseudoangiomatous stromal hyperplasia and fibrous tumors present as circumscribed palpable masses or mammographic densities in premenopausal women or older women on hormone replacement therapy and are benign proliferations of interlobular fibroblasts and myofibroblasts. Myofibroblastoma consists of myofibroblasts and is unusual in that it is the only breast tumor that is more common in males. Lipomas and hamartomas are often palpable but can also be detected mammographically as fat-containing lesions. The only importance of these lesions is to distinguish them from malignancies.

Fibromatosis is a clonal proliferation of fibroblasts and myofibroblasts. It presents as an irregular, infiltrating mass that can involve both skin and muscle. Though locally aggressive, this lesion does not metastasize. Most cases are sporadic, but some occur as part of familial adenomatous polyposis, hereditary desmoid syndrome, and Gardner syndrome. You will recall that familial adenomatous polyposis is caused by mutations in the adenomatosis polyposis coli (APC) gene, which negatively regulates the nuclear translocation of β-catenin. The abnormal presence of β-catenin in the nucleus is a useful diagnostic feature.68

Malignant Stromal Tumors page 1092

page 1093

Malignant stromal tumors include angiosarcoma, rhabdomyosarcoma, liposarcoma, leiomyosarcoma, chondrosarcoma, and osteosarcoma. Sarcomas usually present as bulky palpable masses. Lymph node metastases are rare; hematogenous spread to the lung is commonly seen.

Angiosarcomas of the breast can be sporadic or arise as a complication of radiation therapy.69 Most sporadic angiosarcomas occur in young women (mean age 35), are of high grade, and have a poor prognosis. There is an approximate 0.3% risk of sarcoma after radiation therapy for breast cancer, with most cases arising 5 to 10 years after treatment. Two thirds are angiosarcomas, most of which arise in the overlying skin. Angiosarcomas can also arise in the skin of an arm rendered chronically lymphedematous by prior mastectomy and lymph node dissection (Stewart-Treves syndrome). Fortunately, this complication has become much less common with better surgical techniques.

OTHER MALIGNANT TUMORS OF THE BREAST

Malignant tumors may arise from the skin of the breast, sweat glands, sebaceous glands, and hair shafts; these tumors are identical to their counterparts found in skin elsewhere. Lymphomas may arise primarily in the breast, or the breasts may be secondarily involved by systemic lymphomas. Most are of diffuse large B-cell type. Young women with Burkitt lymphoma may present with massive bilateral breast

involvement and are often pregnant or lactating. Metastases to the breast are rare, and most commonly arise from a contralateral breast carcinoma. The most frequent nonmammary metastases are from melanomas and lung cancers.

Printed from STUDENT CONSULT: Robbins and Cotran Pathologic Basis of Disease 8E (on 11 January 2010)

© 2010 Elsevier