Methods for Obtaining Specimens

Tissue may be obtained in a number of ways, each with its appropriate place and uses, depending on the clinical circumstances. Cytological examination of exfoliated, scraped, or brushed cells can be a rapid, efficient, and low-risk technique for establishing an accurate diagnosis. This approach, along with the related technique of fine-needle aspiration is discussed in greater detail later; for obvious reasons, these approaches do not always reveal the primary tumor site or the extent of disease. Cutting-needle biopsies, core- needle biopsies, or drill biopsies obtain tissue cores for histologic examinations or special studies that permit evaluation of architectural structure. Incisional biopsy (along with fine-needle aspiration) is often the method of choice for lesions that are inoperable, too large for ready excision, or when excision could lead to functional or cosmetic impairment. Excisional biopsy is often favored because it provides generous amounts of tissue for diagnosis and may itself afford sufficient surgical therapy for some tumors, for example, small- to medium-sized breast cancers.

There are many potential pitfalls in biopsy interpretation. These include inadequate tissue sampling and artifacts induced by the procedure itself, such as thermal damage caused by an electrocautery or laser. Benign findings do not exclude the possibility that a tumor or any other significant pathologic condition is present but was not included in the tissue submitted for examination. Thus, the clinician must be prepared to perform a second, often more extensive procedure if the first does not yield sufficient diagnostic information.

Gross Handling of Specimens

The pathologist must regard, and therefore properly triage biopsies, particularly excisional biopsies, as the definitive surgical specimen. To do this well, the pathologist must be informed about the clinical history, differential diagnosis, relevant laboratory results, gross tissue examination, and frozen section findings, if any, since they may individually or together dictate whether special studies are required. Specimens should be marked with clips, sutures, or ink to provide anatomic orientation, and these should be described in the pathology submission sheet. Often, tissue arrives in the pathology laboratory in formalin or other fixatives. At that stage, it is already too late to perform many special studies (eg, microbiologic cultures, certain types of immunohistochemistry, flow cytometry and optimal electron microscopy) that may prove to be critical for diagnosis. This fact emphasizes the importance of consulting the pathologist in advance in order to avoid the need for repeat biopsy. Frequently the goal of biopsy is to determine whether the lesion is benign or malignant, with the expectation of performing additional surgery if the lesion proves malignant. In this case, supplementary tests may properly be deferred to subsequent, more definitive surgery, at which time larger amounts of tissue become available.

The gross specimen should be described with regard to its appearance and characteristics, taking care to measure in three dimensions the size of the specimen and, if visible, the lesion itself along with the distances between the lesion edges and the excision (resection) margins. Excision margins should be identified and marked with ink prior to any dissection, thus permitting accurate measurement of these distances microscopically. Depending on the type of specimen and the clinical circumstances, margins can be evaluated by analysis of frozen sections. All lymph nodes associated with a cancer specimen need to be dissected out, described along with their location, and processed for histology.

A still more careful examination is required for certain biopsies, for example, those of the breast, where no lesion may be visible to the naked eye. In addition, breast specimens with calcification often require specimen radiography. Ideally, therefore, a radiograph of the intact specimen should be obtained, following which the margins should be inked, the specimen “breadloafed” and radiographs taken of each slice. Sections should then be coded, processed individually for histology, and correlated with the corresponding radiographs.^18,19

Preparation of Microscopic Sections

Microscopic examination requires that tissues be cut with a microtome into thin sections that can be stained with dyes such as H & E, toluidine blue, and other special stains for specific tissue components, such as mucus, glycogen, cytoplasmic granules, collagen, bacteria, and fungi. Two types of sectioning methods are most commonly used: frozen sections and paraffin-embedded or permanent sections. Frozen sections can be prepared rapidly (within minutes) during the course of surgery while the patient is still under anesthesia and therefore are of the greatest practical value in situations where an immediate answer is required for an important clinical question.

Frozen sections have many important applications. First, they are useful for determining whether a lesion is a neoplasm and, if so, whether it is benign or malignant. Second, they can provide information as to the extent of regional tumor metastases which may govern decisions concerning further surgery; for example, mediastinal lymph node involvement in primary carcinoma of the lung, peripancreatic lymph node involvement by carcinoma of the pancreas. Third, they allow the pathologist to determine whether the resection margins are adequate following definitive cancer surgery, such as resection of skin, gut, or pulmonary lesions. If resection margins are inadequate, additional tissue can be removed immediately, without the need for a subsequent operation. Some tumors, such as those arising in the soft tissues or breast, are best evaluated in permanent sections, however. Finally, perhaps the most common current use of frozen sections is to determine the appropriate additional workup necessary for a particular tissue specimen while it is still fresh; for example, if the metastatic tumor found in a lymph node is recognized as a poorly differentiated carcinoma, special fixation and electron microscopy may be required for proper diagnosis. On the other hand, if the tumor is a lymphoma, an entirely different set of studies may be required, such as those for cell surface antigen markers and gene rearrangement.

In contrast to frozen sections, permanent sections are prepared from tissues that have been fixed, dehydrated and embedded in paraffin wax as a supporting medium prior to sectioning. Although requiring more time for preparation (generally 12 to 24 hours), permanent sections offer a number of important advantages over frozen tissue sections. Sections are generally thinner (typically 5 μm) and, by avoiding of freezing artifacts, are of better overall quality and therefore permit greater certainty of interpretation. Certain tissues, such as those containing fat or bone, cut poorly as frozen sections but may be satisfactorily studied in permanent sections. As a general rule, if insufficient tissue is available for both frozen and permanent sections, only permanent sections should be prepared. Although the opinions just expressed certainly represent a majority view, some excellent pathology departments routinely diagnose tumors on the basis of frozen sections and prepare permanent sections primarily for archival purposes.

Microscopic Interpretation of Tissue Sections

In cases of suspected cancer, the first task of the surgical pathologist is to decide whether a neoplasm is present. As noted earlier, the word “tumor” is Latin for “swelling,” and various types of “swelling” can masquerade as neoplasms. These include inflammatory lesions, repair, hypertrophy, hyperplasia (eg, keloids), choristomas (ectopic rests), and hamartomas (masses of mature cells that are appropriate to a given site but are arranged in a disorganized fashion as the result of aberrant differentiation). This initial distinction is often made easily; for example, hyperplastic polyps of the colon, nasal polyps, and skin tags are not likely to be confused with true neoplasms. Sometimes, however, the task is less straightforward. Not infrequently, tumors generate an extensive inflammatory response, and it is not unusual, for example, in endoscopic biopsies of gastric carcinomas, to find only after a prolonged search rare individual cancer cells “buried” in an extensive inflammatory cell infiltrate. Healing ulcerations of the gastrointestinal or cervical mucosa may sometimes closely resemble the carcinomas or premalignant lesions (eg, squamous intra-epithelial lesion, low- or high-grade) that arise in those tissues. Finally, atypical hyperplasia can be very difficult to distinguish from in situ carcinoma and, even when no evidence of tumor is found, may represent an important diagnostic finding. For example, patients whose breast biopsies show atypical hyperplasia have a four- to five-fold increased risk for developing breast cancer at a later time.²⁰

Having decided that a neoplasm is present on the basis of criteria such as cellular abnormalities or invasion (see below), the pathologist's next task is to classify it. A number of classification schemes are possible, but the most important of these is based on the tumor's histogenetic or cytogenetic origin. Histogenetic/cytogenetic classification is often supplemented by other useful descriptors such as those provided by the tumor's gross or microscopic appearance (eg, polypoid, papillomatous), the degree of cellular differentiation (eg, well- or poorly differentiated (Figure 35-4), and, perhaps most importantly, by the expected biologic behavior (benign versus malignant). Broadly speaking, tumors of epithelial cell origin are termed adenomas or papillomas when benign and carcinomas when malignant. Carcinomas account for ~80% of all malignant tumors. Their classification is often further qualified on the basis of the type of epithelium present, for example, glandular (adenocarcinoma, Figure 35-5), squamous (squamous cell carcinoma, Figure 35-6), urothelial cell (urothelial cell carcinoma). Addition of the suffix “-oma” to the cell of origin also describes benign tumors of mesenchymal origin (eg, lipomas, fibromas). Malignant tumors of mesenchymal origin are designated sarcomas (eg, liposarcomas, fibrosarcomas). Most tumors comprise a single type of neoplastic cell. However, a few tumors contain neoplastic cells of more than a single type, for example, Wilms' tumors. Even rarer are tumors containing neoplastic cells that derive from more than a single germ layer, for example, teratomas (dermoid cysts). Certain tumors have long been identified with trivial names that do not follow any well-ordered classification scheme. Examples include seminomas (for carcinomas of testicular epithelial cell origin), hypernephromas (for renal cell carcinomas), and melanomas (for melanocarcinomas) (Figure 35-7). Other tumors, because of prolonged usage, continue to bear eponyms (eg, Hodgkin disease, Kaposi sarcoma).

Figure 35-4

Undifferentiated carcinoma of lung. Tumor is composed of irregularly arranged pleomorphic tumor cells with large nuclei and prominent nucleoli. A central necrotic zone, typical of this type of tumor, is present. There is little stroma. Hematoxylin and (more...)

Figure 35-5

Adenocarcinoma of breast. Tumor is arranged in the form of fairly well-differentiated glands separated by a fibrous connective tissue stroma. Moderate numbers of inflammatory cells, mostly lymphocytes, are present in stroma. Hematoxylin and eosin; magnification: (more...)

Figure 35-6

Squamous cell carcinoma of the skin of the face. Tumor consists of island of well-differentiated squamous epithelium separated by fibrous connective tissue stroma. Note that epithelium forms “keratin pearls.” Hematoxylin and eosin; magnification: (more...)

Figure 35-7

Malignant melanoma arising in skin. Tumor is composed of large, irregular cells with large nuclei, prominent nucleoli, and abundant, clear cytoplasm, peppered with dots and larger accumulations of melanin pigment. Hematoxylin and eosin; magnification: (more...)

The pathologist must carry classification further still. Even within a single organ and within a single type of epithelium, several different types of tumors may arise, each with its own special characteristics, prognosis, and response to therapy. In the breast, for example, the two most common types of malignant tumor are infiltrating ductal carcinoma (sometimes designated as carcinoma “not otherwise specified” or NOS) which accounts for ~80% of infiltrating breast cancers, and infiltrating lobular carcinomas, which account for an additional ~9% of breast cancers. These two tumors, together accounting for nearly 90% of breast cancers, have similar prognoses that are less favorable than those of the other, less common types of breast carcinoma (ie, tubular, mucinous or colloid, medullary, papillary, and adenoid cystic carcinomas).²¹

One of the most important distinctions the surgical pathologist can make is that between tumors that are benign or malignant. In general, benign tumors share certain properties. The neoplastic cells that make up the tumor are usually well differentiated, closely resembling the corresponding cells of normal tissue. Benign tumors tend to expand uniformly in all directions unless impeded from doing so by surrounding structures, for example, compression by the bony skull often causes meningiomas to take on a flattened appearance. As expansile masses, benign tumors cause compression atrophy of surrounding normal tissues that results in the formation of a thin rim of fibrous connective tissue; this enveloping connective tissue rim may serve as a “capsule” that renders benign tumors discrete, readily palpable, and easily movable. However, not all benign tumors have capsules, for example, leiomyomas of the uterus, hemangiomas, and adenomatous polyps of the large intestine.

Malignant tumors, or cancers, are characterized primarily by the abnormality of their neoplastic cells. These cellular abnormalities are of two general types, those involving intercellular relationships and those affecting individual neoplastic cells. With regard to the former, malignant tumors commonly exhibit increased cell number and altered orientation of both neoplastic cells and stroma that may be best described as “helter-skelter” or disorganized. For example, carcinomas of the skin may comprise squamous cells that differentiate and mature fairly normally; however, the cells are commonly organized into nests, in which the least differentiated cells are situated peripherally and the most differentiated cells are positioned centrally where they form keratin pearls. Further, these tumor cell nests are surrounded by disorganized stroma. Disturbed intercellular arrangements such as these are of great help to the pathologist reading a tissue section; much of tumor diagnosis depends on the pathologist's ability to recognize altered microscopic tissue patterns.

Abnormalities of individual neoplastic cells may also be helpful in diagnosis, particularly increased numbers of mitoses and cytological features relating to the state of tumor cell differentiation. Cytological features of malignancy include altered polarity, tumor cell enlargement, increased ratio of nuclear to cytoplasmic area (may approach 1:1 instead of the normal 1:4 or 1:6, although exceptions exist), pleomorphism (variation in size and shape) of tumor cells and their nuclei, clumping of nuclear chromatin and distribution of chromatin along the nuclear membrane, enlarged nucleoli, atypical or bizarre mitoses (eg, tripolar), and tumor giant cells with one or more nuclei. Some malignant tumors, however, are well differentiated, so well differentiated, in fact, that their malignant cells cannot be distinguished from those of benign tumors or even from normal cells by any available diagnostic method. In such instances, the recognition of abnormal cellular relationships becomes especially important for correct diagnosis.

Anaplasia (Greek, “to form backwards”) is a term that describes the degree of tumor cell differentiation or, more correctly, the lack thereof. Although well entrenched, the term is an unfortunate one. It implies that tumors arise from mature, differentiated cells by a process of de-differentiation (ie, differentiation in reverse). Few pathologists hold that view today. Mammalian cells, once differentiated, generally lack the capacity to reverse that process. Also, there is strong and growing evidence for the alternative explanation, namely, that tumors arise from populations of undifferentiated “stem” or “reserve” cells that are present in many, perhaps in all, organs capable of cell renewal.^22–24 Stem cells comprise a minority cell population that lacks differentiation markers, making them difficult to identify. However, positive recognition of stem cells has been achieved in several organs; for example, bone marrow, epidermis, liver, and gastrointestinal tract mucosa.²⁵ Stem cells have a high capacity for cell proliferation but, unless stimulated, may divide infrequently. Stem cells alone have the capacity to regenerate normal tissues and, by extension, tumor cell populations. Oncologists, of course, are well aware that stem cells are the critically important target of cancer therapy. Destruction of differentiated tumor cells, without simultaneous killing of tumor stem cells, will not lead to permanent tumor eradication.

Malignant tumors invariably lack a capsule. Instead, they extend crab-like projections into the surrounding host tissues without respect for normal anatomic boundaries. This behavior is referred to as invasion. Malignant tumor cells often invade lymphatics and veins and are transported by lymph or blood flow to distant sites, opening the possibility of metastasis. Invasion is not a property confined to malignant tumor cells; many proliferations in fetal life, placental trophoblasts, and inflammatory cells also have the capacity to invade tissues. However, cancers need not be invasive at the time of removal, either because they have been “caught” before they had time to invade or because they have not yet progressed to the point where they have acquired the capacity to invade. Epithelial tumors with all other properties of malignancy that have not extended through the underlying basement membrane at the time of diagnosis are described as in situ carcinomas and can almost certainly be cured by complete excision.²⁶

Ancillary Staining and Analytical Methods

Immunocytochemistry is the most commonly employed ancillary technique in surgical pathology (see the following section) but special stains also aid in tumor differential diagnosis and classification. Examples include the Van Gieson's stain or the Masson trichrome method for distinguishing collagen and muscle, the Weigert's stain for elastic tissue, silver stains for reticulin fibers, and special stains for mucins, amyloid, lipids, myelin and glycogen—all substances whose identification may aid in the diagnosis of one or another type of tumor. In other instances, enzyme histochemistry may be essential for defining cell lineage, as in certain types of leukemia; for example, chloroacetate esterase or endogenous peroxidase staining for cells of myelomonocytic lineage, alpha naphthyl butyrate esterase (so-called “nonspecific” esterase) staining for monocytes and macrophages.

Other techniques that may occasionally aid the surgical pathologist in tumor diagnosis are specimen radiography (for localizing and analyzing crystalline material, such as calcium, in breast biopsies) and morphometry.

Excision Margins

An important concern is the adequacy of tumor excision. Depending on the tissue, this decision can be made on either frozen or permanent sections. If the tumor forms a discrete mass and the margins of the specimen are clearly recognizable, determination of excision margins is usually straightforward. Examples of tumors whose excision is likely to give clearly defined margins include those arising in the gastrointestinal tract, lung, and skin. On the other hand, the margins of tumors arising in soft tissues (eg, many sarcomas and breast carcinomas) and diffusely infiltrating tumors (eg, infiltrating lobular carcinoma of the breast, signet ring tumors of the gastrointestinal tract, and nerve-invading tumors, such as adenoid cystic carcinomas of the salivary glands, gliomas, and glioblastomas) may be much more difficult to define. With at least certain histologic patterns of breast carcinoma, factors such as the extent of intraductal growth must be considered when evaluating resection margins.^27,28 In patients treated with excision and radiotherapy for invasive breast cancer, the evaluation of excision margins in the context of an extensive intraductal component provides more prognostic information than the evaluation of margins alone.²⁹

Tumor Grading, Staging, and Prognosis

Pathologists are often called upon to grade tumors or to participate in their staging in order to estimate tumor prognosis. Tumor staging (eg, the well-known TNM system) has proved to be of great value in estimating prognosis. Staging attempts to measure the extent of spread of a cancer within a patient on the basis of such parameters as tumor size, lymph node involvement, and the presence of other metastases. It is obvious that objective determinations made by the pathologist on resected tumor specimens have a critical impact on accurate tumor staging. With rare exceptions, such as papillary carcinoma of the thyroid, the single most important risk factor in determining tumor prognosis is the presence of metastases to regional lymph nodes. Therefore, the pathologist must search diligently to find, examine, and prepare histologic sections from all lymph nodes included in resected tissue.

Tumor grading has traditionally referred to a pathologist's judgment as to a tumor's degree of differentiation and growth rate, often on a scale of I to III where III represents the least differentiated, fastest dividing tumors (ie, those tumors presumed to have the worst prognosis). Formal grading systems have improved in recent years with stricter standardization of criteria (Figure 35-8).³⁰ Tumor grading does, however, have shortcomings. First, a different scale is required for each type of tumor, and scoring is subjective and not always reproducible. Second, tumors are typically heterogeneous so that areas differing significantly in differentiation and mitotic activity exist side by side, with the attendant risk of sampling error. Because prognosis is invariably linked to the most malignant portions of a tumor, it follows that, for accurate diagnosis and grading, sufficient tissue and microscopic sections must be sampled so that the most malignant areas are found. Moreover, a regular (although not invariant) feature of malignant tumors is progression ,^31,32 the property by which tumors become more and more malignant over time. Tumor progression is thought to result from genomic instability that leads to specific mutations of oncogenes or loss of tumor suppressor genes, from other genetic alterations, such as gene amplification, as well as from epigenetic changes that result in altered patterns of gene expression.³³ Overt carcinogens, environmental promoters, and local factors, such as hypoxia and nutrient deprivation, may all contribute to these changes and therefore to tumor progression. Finally, the correlation between histologic appearance and biologic behavior is seldom perfect.

Figure 35-8

Actuarial rates of distant failure for 1,081 patients with invasive breast cancer related to histologic grade of tumor. Grading was performed using Elston's modification of the Bloom and Richardson grading system, which takes into consideration architectural (more...)

Pathologists are continuously on the lookout for more useful tumor-specific features that may be important independent predictors of tumor prognosis. For example, carcinomas of the prostate are now usefully graded on the basis of tissue architecture and neoplastic cell pattern.³⁴ A number of different criteria including nuclear differentiation, degree of gland or tubule formation, and mitotic activity have been usefully combined to grade breast carcinomas.^30,35,36 Carcinoma of the breast illustrates other tumor-specific factors that affect prognosis and will now be discussed in greater detail, as an important example.

Surgical Pathologist's Report

The findings should be presented comprehensively in terms that are understandable to both the pathologist and the clinicians caring for the patient. The report should provide enough information so that the clinician caring for the patient can follow the thought processes of the pathologist, much as if he were viewing the case with the pathologist at a double-headed microscope. The report should contain all the information to which the pathologist has access (ie, tumor size, grade, and nodal status) that is necessary to stage a patient with cancer. This information varies with tumor origin, type, and staging system employed. The report should include the results of all specialized tests performed, their interpretation, and the synthesis and coordination of all clinically useful information available to the pathologist that may be of aid in diagnosis and management. A potentially helpful method to standardize reporting of malignancies has been the introduction of synoptic reporting. Both the Association of Directors of Anatomic Pathology (ADASP) and The College of American Pathologists Cancer Committee (CAP) have produced synoptic reporting guidelines that are available on their websites. Finally, reports should be issued in a timely manner. Failure to report results promptly may delay patient care (thus uselessly adding to the cost of medical care), lead to error and confusion, and at the very least prolong anxiety in patients who are often already distraught.

The Role of the Cytopathologist

Cytology is used for both screening and diagnosing of lesions that may represent cancer or its precursors. Specific benefits include cost-effectiveness, rapid turnaround time, and a diagnosis with minimal patient risk. Since cytologic specimens usually consist of a very small amount of cells or tissue fragments, optimal technique for both sample collection and slide preparation is crucial. Moreover, as in all areas of pathology, cytologic diagnosis should never be made “in a vacuum”; pertinent clinical data and communication between the cytopathologist and the clinician are essential, and will facilitate rapid, accurate and definitive cytologic diagnoses.

Cytology of Body Cavity and Cerebrospinal Fluids

Fluids are removed from body cavities not only for the purpose of therapy (eg, to relieve pressure on vital organs) but also as a form of diagnosis. The presence of tumor cells in body cavity fluids (Figure 35-9) implies advanced stage with the exception of mesotheliomas. It is usually difficult, if not impossible, to determine the site of primary tumor, although certain morphologic clues may occasionally allow the cytopathologist to suggest a site of origin. If consistent with the clinical findings and results from other studies, a positive cytologic diagnosis can lead directly to treatment. Current cytology specimens should always be compared to previous cytology or histology specimens, if available.

Figure 35-9

Ovarian papillary serous adenocarcinoma in ascitic fluid. Cells are present in three-dimensional clusters with nuclear molding and irregular hyperchromatic nuclei. Papanicolaou; magnification: X250. (Four-color version of figure on CD-ROM)

Reactive mesothelial cells share certain characteristics with carcinoma cells and should be considered in the differential diagnosis. Panels of immunocytochemical stains may prove useful (see below). Sometimes (eg, mesotheliomas) tumor growth may be confined to the surfaces lining body cavities without significant exfoliation of malignant cells. Under these circumstances, a biopsy of the cavity wall is required to establish the diagnosis. An alternative approach for future investigation may be to analyze tumor cell-free effusions for secreted products of malignant cells or other tumor cell markers. One successful although preliminary application of this approach has been the finding of high levels of the angiogenic factor VPF/VEGF in effusions by immunoassay.⁶

Aspiration Cytology

The earliest work on aspiration cytology was reported from the Memorial Hospital in New York in the 1930s.⁹⁰ Subsequently, the impetus for this technique shifted to Europe and was not “rediscovered” in the United States until the 1970s.⁹¹ Cytopathologists, surgeons and other clinicians can successfully perform aspirations of a palpable lesion.⁹²

In the case of nonpalpable lesions, aspiration is performed by a radiologist under computed tomography, ultrasound, or fluoroscopic guidance. Such deep aspiration procedures are expensive, time-consuming, and invasive. For these reasons it is desirable that a cytotechnologist or cytopathologist attends the procedure to ensure specimen adequacy and optimal slide preparation. A further advantage of this attendance is the ability of experienced personnel to triage material effectively for special studies, such as immunocytochemistry (Figure 35-10), electron microscopy, flow cytometry and tissue culture.

Figure 35-10

Metastatic neuroendocrine tumor as seen in a fine-needle aspiration of liver. Note eccentric nuclei with “salt and pepper” chromatin and abundant granular cytoplasm. Immunocytochemical stains confirmed the diagnosis. Papanicolaou; magnification: (more...)

Definitive aspiration cytology diagnoses, rendered by an experienced cytopathologist, can provide the basis for definitive therapy. However, such diagnoses need to be viewed in the context of all other laboratory studies and clinical findings. Specific problems and pitfalls that attend aspirations of various sites will now be briefly discussed.

Thyroid

Fine needle aspiration biopsy (FNA) is the first test in most patients with a solitary thyroid nodule because it is safe and inexpensive and leads to a better selection of patients for surgery than any other test.⁹³ FNA permits the accurate diagnosis of papillary, medullary, and anaplastic carcinomas but is less useful in the diagnosis of follicular nodules. There are cytologic features that help to distinguish among the various types of follicular lesions,^94,95 but a definitive diagnosis may not be possible by FNA, particularly the distinction between follicular adenomas and carcinoma. Thus, in some instances, FNA will serve only to distinguish patients needing immediate surgery for thyroid disease from those who may be safely followed with or without hormonal suppression.^93,96 However, even this limited information can eliminate much unnecessary surgery.

Breast

The diagnostic specificity of breast aspiration is very high in the hands of an experienced cytopathologist, and a positive diagnosis may safely lead to mastectomy or other definitive treatment.^78,97 Of course, atypical and suspicious cases will require further workup. Aspiration cytology of the breast may also be performed on nonpalpable lesions under the guidance of conventional or stereotaxic mammography or ultrasound.^98–100 Inherent problems for breast cytology include the inability to distinguish infiltrating from in situ carcinoma and the difficulty in rendering a specific benign diagnosis, when compared with core biopsies.^101,102 However, needles used in aspiration cytology cost much less than the core needles and are more readily tolerated by patients. Although most labs continue to show an increase in FNA of breast, especially for palpable lesions, the increase is much less than that in core biopsies of the breast.¹⁰³

Role of the Immunohistochemist

Immunohistochemistry has become an important adjunct in the evaluation of human neoplasms. A detailed discussion of the technical aspects of immunohistochemistry is beyond the scope of this chapter, and the interested reader is referred to several review articles and monographs.^104–109 The commercial availability of a broad range of reagents (including prediluted reagents in kit form) has made it possible for high-quality immunohistochemistry to be performed in most pathology laboratories. The most commonly employed immunohistochemical techniques are those in which enzymes, such as horseradish peroxidase or alkaline phosphatase, are used in conjunction with specific antibodies to provide color reactions at sites of antigen-antibody interactions. Variations of the avidin-biotin complex (ABC) technique are currently the most widely utilized in current practice. The ABC procedure generally requires three sequential steps: an unlabeled primary antibody, a biotin-labeled anti-immunoglobulin secondary antibody, and, finally, preformed avidin-biotin-peroxidase complexes. One variation of the ABC method employs streptavidin, which has greater sensitivity than avidin and exhibits less nonspecific binding.¹⁰⁹ It should be noted that the sensitivity of any immunohistochemical procedure is, in large part, related to the reagents and detailed procedures employed. As a consequence, it is difficult to compare the results of immunohistochemical studies from different institutions that employ different reagents and methods.

Virtually any type of pathologic specimen may be suitable for immunohistochemical staining, including fresh frozen tissue, fixed tissue, and cytologic preparations. Unfortunately, however, not all antigens are equally well preserved after these various treatments, and the approach taken for immunohistochemical staining must depend on the antigen(s) of interest. For example, although a large number of cytoplasmic antigens are detectable in fixed, paraffin-embedded tissue, other antigens, such as many cell surface-associated antigens, are destroyed or masked by common fixatives and may be demonstrable only in fresh frozen tissue or in cytologic preparations. Antigen retrieval methods, such as pretreatment with proteolytic enzymes or heating (using a microwave oven, steamer, pressure cooker, or autoclave), may permit the identification of otherwise undemonstrable antigens in fixed, paraffin-embedded tissue sections.^110,111 Finally, not all fixatives are equivalent with regard to antigen preservation. Although crosslinking fixatives, such as formaldehyde, are often suitable, they are suboptimal for detecting certain antigens of diagnostic importance, such as those located on intermediate filaments, which are best demonstrated in fresh-frozen or alcohol-fixed tissue.^108,112,113

Applications

Immunohistochemistry has widespread applicability in the evaluation of human tumors. Some of the more common applications are listed in Table 35-1 and are discussed below.

Table 35-1

Common Applications of Immunohistochemistry in the Evaluation of Human Tumors.

Categorization of “Undifferentiated” Malignant Tumors

Not infrequently, a pathologist examining routine H & E-stained paraffin sections recognizes the presence of a malignant tumor but is unable to characterize the tumor further. This is understandable in that “undifferentiated” tumors often lack characteristics that would permit more accurate classification. Yet further classification is often important in making clinical decisions related to appropriate therapy and prognosis. Immunohistochemistry may be helpful in such situations (Figure 35-11). Before performing immunohistochemistry, however, the pathologist must first develop a differential diagnosis, and this will depend on the tumor's histologic appearance, anatomic location, and the clinical setting. Only then is he or she in a position to select antibodies that will permit a more definitive diagnosis.

Figure 35-11

Immunoperoxidase staining using a monoclonal antibody specific for keratin in a poorly differentiated squamous cell carcinoma of skin. This “spindle cell” form of the tumor mimics tumors of connective tissue origin, and its true nature (more...)

One common problem in tumor diagnosis, that of undifferentiated tumors composed of large cells with an epithelioid appearance, will serve as an example. The differential diagnosis in such cases typically includes undifferentiated carcinoma, lymphoma, and melanoma. Distinction among these tumor types can often be made using a panel of antibodies as illustrated in Table 35-2. Unfortunately, this table presents an ideal result that is not always achieved in practice. Some carcinomas show staining for vimentin or S100 protein, some melanomas show immunoreactivity for keratin, and some neoplasms other than melanomas express HMB-45.^113–118 Such results emphasize the need to use a panel of antibodies, rather than a single antibody, when evaluating tumors.

Table 35-2

Idealized Immunohistochemical Evaluation of the “Undifferentiated” Malignant Tumor ^a.

Determination of Site of Origin of Metastatic Tumors

On routine microscopic examination, tumors may be classifiable with regard to general type (eg, carcinoma), but not with regard to site of origin. It would be highly desirable to have available antibodies specific for tumors arising in different sites. At present, however, very few organ- or tissue-specific antigens have been identified, thus limiting the ability of immunohistochemistry to resolve such problems in every instance. A number of useful antigens are listed in Table 35-3. It should be noted that antigens specific for some of the more common tumors, such as carcinomas of the lung, colon, endometrium and pancreas, are not currently available. Furthermore, some of the antigens listed in Table 35-3 have now been demonstrated in neoplasms other than those for which they were initially thought to be “specific.” For example, the melanoma-associated antigen detected by one widely used antibody (HMB-45) has been found in some nonmelanocytic tumors.¹¹⁸ A more recent approach for the subclassification of metastatic carcinomas exploits differences in cytokeratin profiles in tumors from different primary sites.¹¹⁹

Table 35-3

Antigens with Highly Restricted Specificity.

Subclassification of Tumors in Various Organ Systems

In some organs and tissue compartments, it may be difficult to subclassify certain tumors solely on the basis of histologic grounds because of overlapping features. Some of these distinctions are (at least at present) only of academic interest (eg, determining whether a high-grade spindle cell sarcoma shows neural, myogenous, or fibrohistiocytic differentiation), but others have therapeutic and prognostic significance. For example, in some cases, it may be difficult or impossible to distinguish with certainty an anaplastic seminoma from an embryonal carcinoma of the testis by routine microscopic examination, a distinction with both therapeutic and prognostic implications. However, immunostaining for the intermediate filament keratin is often useful in making this distinction because seminomas are typically keratin negative, whereas embryonal carcinomas are usually keratin positive.^120,121 Similar situations are encountered in other organ systems and tissue compartments.

Distinction between Carcinomas and Malignant Mesotheliomas

A common problem encountered by the surgical pathologist is the distinction between metastatic adenocarcinoma and malignant mesothelioma involving the pleura or peritoneum.^122–126 Immunohistochemical staining using a panel of antibodies may be useful in assisting in this distinction (Table 35-4).

Table 35-4

Immunohistochemical Distinction between Metastatic Adenocarcinoma and Malignant Mesothelioma Involving the Pleura or Peritoneum.

Categorization of Leukemias and Lymphomas

One of the most common uses for immunohistochemistry is the correct diagnosis and classification of leukemias and lymphomas. A detailed discussion of this subject is beyond the scope of this chapter, and the interested reader is referred to recent articles and reviews.^127–130 In brief, immunohistochemistry, in conjunction with morphology and histochemistry, is a useful adjunct in making the distinction between acute leukemias of lymphoid and nonlymphoid types and in distinguishing hairy-cell leukemia from other types of leukemic infiltrates in the bone marrow and at other sites. In addition, this technique is useful for subclassifying non-Hodgkin lymphomas and Hodgkin disease and in distinguishing them from each other in problematic cases.

Detection of Antigens of Potential Prognostic or Therapeutic Significance

A variety of antigens of possible prognostic and therapeutic importance can be detected using immunohistochemistry, including estrogen and progesterone receptors in breast cancers, protein products of oncogenes (such as HER-2/neu in breast cancers), antigens associated with tumor cell proliferation such as Ki-67 and the P-glycoprotein product of the multiple drug resistance (MDR) gene.^131–139 Ki-67 is of particular interest.¹⁴⁰ It is a nuclear antigen present in all proliferating cells; that is, present in the G₁, S, G₂ and M phases of the cell cycle but absent in G₀ cells. Therefore, by staining for this antigen, it is possible to measure the tumor growth fraction directly and in a simpler manner, which is more readily applicable to clinical specimens than are radioactive labeling methods using ³H-thymidine. Ki-67 staining also yields results that are more reproducible than those obtained by counting mitotic figures. Recently, a number of antibodies have become available to identify the Ki-67 antigen in formalin-fixed, paraffin-embedded tissue sections.¹⁴¹

Role of the Electron Microscopist

Although making use of radically different technology, electron microscopy seeks the same type of information as that gleaned from immunohistochemistry, that is, detection of differentiated organelles (or markers, in the case of immunohistochemistry) that permit more accurate tumor identification and classification. Electron microscopy (EM) is not useful in determining whether individual cells are malignant or benign. It is a powerful tool for recognizing subcellular structures that are not detectable by light microscopy but which, when present, allow confident identification of cells as, for example, of epithelial or melanocyte origin. Although advances in immunohistochemistry have somewhat reduced the need for EM in tumor diagnosis, they have by no means eliminated this need altogether, and EM remains, at present, a powerful but generally underutilized approach to tumor diagnosis.^142–145 Moreover, validation of new immunohistochemical reagents is often best accomplished by ultrastructural study of replicate tissue samples.

Applications

The great strength of EM lies in its exquisite resolution, which permits the recognition of intracellular structures, organelles, or products that are undetectable by light microscopy.

EM is often helpful in the diagnosis of “undifferentiated” malignant tumors and in determining the origin of metastatic tumors of unknown primary site (Figures 35-12 and 35-13). The recognition of cytoplasmic premelanosomes within tumor cells permits the distinction of amelanotic malignant melanomas from undifferentiated carcinomas and lymphomas with which they can be confused. Other ultrastructural features whose recognition may permit definitive diagnosis are the cytoplasmic granules characteristic of carcinoid tumors; the norepinephrine- and epinephrine-containing granules found in pheochromocytomas; “terminal webs” characteristic of primary gastrointestinal carcinomas of absorptive epithelial cell origin; lamellar (surfactant) bodies found only in type II pneumocytes and therefore diagnostic of alveolar cell carcinomas of the lung tonofilaments and desmosomes found in mesothelial cells and squamous cells; and cytoplasmic glycogen aggregates and calligraphic nucleoli typical of germinomas.^147,148 The presence of intercellular junctions permits the distinction of carcinomas from lymphomas, even if the carcinoma's primary site cannot be determined. EM is also helpful in subclassifying tumors, an exercise that may have important therapeutic implications. The use of ultrastructural cytochemistry for endogenous peroxidase may allow the important distinction of acute myeloblastic leukemia from acute lymphoblastic leukemia. Histiocytosis X may be diagnosed by identification of Birbeck bodies characteristic of Langerhans cells. EM may also permit accurate diagnosis of lysosomal storage diseases and of bacterial, fungal, and viral infections.

Figure 35-12

A, Electron micrograph from lung mass shows typical surfactant-containing lamellar bodies (surfactant bodies) (arrowhead) that fill the cytoplasm of a tumor cell, allowing the specific diagnosis of primary alveolar cell carcinoma of the lung to be made. (more...)

Figure 35-13

High-magnification electron micrograph of pleural tumor shows three typical features of mesothelial cells: numerous elongated, thick surface microvilli that do not display evidence of terminal web differentiation; and desmosomes that connect individual (more...)

Finally, EM is important in identifying the histogenesis of newly recognized neoplasms. A recent example (Figure 35-14) is the recognition that certain spindle cell tumors of the gastrointestinal tract, previously thought to be of smooth muscle origin, in fact arise from autonomic neurons (gastrointestinal tract autonomic nerve tumors or GAN tumors).¹⁴⁷

Figure 35-14

This high-magnification micrograph of a gastric tumor shows tumor cells diagnostic for the newly delineated entity, the gastrointestinal autonomic nerve tumor (or GAN tumor). Elongated tumor cells are neurites, which contain numerous mitochondria and (more...)

Role of the Clinical Pathologist

The role of the clinical pathologist or “laboratorian” is obvious and familiar to the oncologist and requires only brief mention here. Clinical pathologists are interested in measurements of body fluids that lead to the early detection and monitoring of cancer. There has been a great deal of interest in the field of tumor antigens and tumor markers, and some of these are discussed elsewhere in this volume. At least in theory, tumor-specific antigens circulating in the plasma could be of utility in tumor diagnosis and prognosis, assessment of tumor burden, prediction of recurrence, and guidance for treatment. Properties of an ideal tumor marker include great sensitivity, specificity, and accuracy in reflecting total tumor burden. A tumor marker should also be prognostic of outcome and predictive of tumor recurrence. Unfortunately, none of the tumor markers discovered to date fulfills all of these criteria. In fact, none is uniquely produced by tumor cells. Normal cells of one sort or another make all of the tumor markers thus far recognized, and plasma or serum levels in tumor patients differ only quantitatively, not qualitatively, from those of normal controls or patients with other diseases.

The role of the clinical pathologist, however, has recently expanded into other areas of oncology. Particularly important is the molecular biologic diagnosis of tumors, for example, T-cell lymphomas, by the detection of gene rearrangements.^149,150 Demonstration of such rearrangements may be especially important as a supplement to the work of the surgical pathologist and immunohistochemist in distinguishing “clonal” and, therefore, presumably malignant lesions from benign but highly reactive lymph nodes. There is increasing interest in the clinical laboratory in the assessment of solid tumor clonality in understanding tumorigenesis and as an aid to diagnosis and estimation of prognosis. In recent years, the cytogenetics laboratory has enjoyed a renaissance of activity in recognizing consistent chromosome abnormalities in a growing list of leukemias, lymphomas, and solid tumors. At present, specific chromosomal abnormalities are of greatest clinical importance in only a few tumors, primarily lymphomas and leukemias;^151,152 for example, in acute and chronic myeloid leukemia and in acute lymphocytic leukemia. However, it is now clear that nonrandom chromosomal changes are to be found in a variety of solid tumors, and it is likely that cytogenetic information will become increasingly useful in defining tumor progression and prognosis.^{23,45,153–155} Along a similar vein, there has been increasing interest in recent years in the use of flow cytometric analysis of tumor cell suspensions for the determination of tumor ploidy and the fraction of replicating cells (see above). Early results indicate that such information is of use in defining the prognosis of renal, breast, and certain other types of tumors.^{23,48,49,51,156}

Role of the Molecular Pathologist

The molecular pathologist practices at the interface of anatomic and clinical pathology and utilizes the principles, theories and techniques of molecular biology to make or confirm diagnoses of cancer, provide more specific classification of cancer and/or provide prognostic information. Somatic mutations, that is, mutations that occur during the lifetime of an individual in tissue outside of the germ line, are central to the development of cancer.¹⁵⁷ During neoplastic progression, a series of mutations develops in critical genes. These mutations may be gene deletions, duplications, point mutations and/or chromosomal translocations in tumor cell DNA. The mutations affect regulation of the cell cycle, cellular differentiation, apoptosis and cell-cell and cell-matrix interactions and are associated with altered gene expression. Different neoplasms have different combinations of genetic alterations and this can be exploited in the laboratory for diagnostic or prognostic information. Some genetic alterations are required for the development of specific tumor types and therefore, are discrete, objective indicators of the specific neoplasms. The best examples of these are the chromosomal translocations seen in some leukemias, lymphomas and soft tissue tumors. (Table 35-5). Other genetic alterations, such as p53 mis-sense mutations that inactivate the protein's normal role in cell cycle control, seem to be common to many malignancies.¹⁵⁸ Still other genetic changes are seen in subsets of neoplasms such as the amplification of the HER2/NEU gene that is present in 10% to 30% of breast carcinoma and provides both prognostic information as well as targeted therapeutic options with transtuzumab (Herceptin).¹⁵⁹ HPV genotyping has proved to be a useful adjunct tool in the diagnosis of precancerous cervical dysplasias.¹⁶⁰

Table 35-5

Recurrent Chromosomal Translocations in Human Neoplasia.

Many cancer-related molecular pathology tests focus on genetic alterations that are markers of specific tumor types. The new WHO classification of leukemias characterizes subtypes of leukemias by specific chromosomal translocations.¹⁶¹ For example, the Philadelphia chromosome (t(9;22)(q34;q11)) juxtaposes the 5′ portion of the BCR gene to the 3′ portion of the ABL gene creating a new chimeric BCR-ABL gene that is the sine qua non for chronic myelogenous leukemia. The chromosomal breakpoints occur within the introns or noncoding regions of the genes so that the coding regions or exons remain intact and can be transcribed into messenger (mRNA) and translated into a unique 210 Kd fusion protein. It is this fusion protein that is oncogenic for hematopoietic cells and present in all cases of chronic myelogenous leukemia. The molecular pathologist can look for this specific marker of disease in a variety of ways including complete cytogenetic analysis, fluorescence in situ hybridization (FISH) on metaphase spreads or intact interphase nuclei, Southern blot hybridization on tumor DNA, reverse-transcriptase polymerase chain reaction (RT-PCR) on tumor mRNA or Western blot on tumor protein. The most commonly used methods are karyotype, FISH or RT-PCR. All of these technologies are labor-intensive and costly which has limited their application. However, higher throughput technologies, such as real-time PCR, are becoming a reality making more routine application realistic.¹⁶² RT-PCR assays also have the advantage of being sensitive as well as specific. As few as 1 abnormal cell in 100,000 can be detected, making this an effective technique for following minimal residual disease posttherapy.

As the molecular basis of neoplastic transformation and progression becomes elucidated, more applications will become incorporated into the diagnostic laboratory. Recently, several studies have shown the ability of broad gene expression analysis by microarray technology to stratify neoplasms by disease outcome.^163–165 The pathologist will play a central role in determining the best way to incorporate these new biologic markers and technologies into the diagnostic laboratory.

Role of the Autopsy Pathologist

Autopsies do not receive the respect that was at one time accorded them, and the autopsy rate has declined precipitously throughout the United States from ~90% of hospital deaths in some teaching hospitals in the late 1960s to the current average rate of only 15%.^166–170 One widely voiced but erroneous reason for this change in attitude and practice is the belief that autopsies no longer yield much in the way of useful information because they have been preempted by new technologies, such as magnetic resonance imaging and other pathologic and biochemical tests performed on the patient during life. In a series of cases from 1986 to 1995, clinically undiagnosed or misdiagnosed malignancies were found in 44% of autopsies—a rate similar to that reported in studies from 1923 and 1972. Among the undiagnosed cancers, 57% were felt to be directly related to the patient's death.¹⁷¹ Another concern is the clinicians' fear of malpractice suits resulting from new findings revealed at autopsy that had not been diagnosed during life. Still another is the fact that the Joint Commission for the Accreditation of Health Care Organizations has greatly reduced its emphasis on the hospital autopsy rate for accreditation purposes. Psychological factors may also play a negative role. Because of the significant side effects that accompany the longer survivals achieved with modern cancer therapy, the family of the deceased may feel that the patient had “suffered enough.” Also, the autopsy serves as a symbol of failure, reminding the clinician that he was unable to cure the patient. Finally, there is a negative economic incentive for performing autopsies. Neither the pathologist, the hospital, nor the oncologist is reimbursed the cost in time, effort, and materials involved in performing an autopsy or in persuading a reluctant family to permit an autopsy.

Despite these objections, the autopsy continues to have an important role in patient care. In fact, advances in technology have done little to change the incidence of unexpected, clinically significant findings at autopsy. In our experience, it is most unusual for an unexpected autopsy finding to lead to litigation, and if an egregious error in patient management has occurred, is it not the responsibility of the medical profession to discover this?

The autopsy has an important role in evaluating the care of the cancer patient who has succumbed to his illness. Although the autopsy will obviously not offer direct benefit to the dead patient, it may be essential for supporting or refuting clinical impressions, determining the extent of residual disease and the adequacy of therapy, evaluating new therapies, identifying the ultimate and proximate causes of death, and revealing unexpected findings that affected patient care. As a means of advancing knowledge, clinicians should regard an autopsy as the final contribution that they and the deceased patient can make to an understanding of disease. The usefulness of the autopsy is greatly enhanced when the clinician takes the time to address the pathologist with specific questions that he would like answered at postmortem examination.