Chemotherapy of Carcinoma of Unknown Primary Site
John D. Hainsworth
F. Anthony Greco
Cancer of unknown primary site (CUP) is a common clinical syndrome, accounting for 3% to 5% of all cancer diagnoses. The widespread nature of disease at the time of diagnosis in most patients precludes any consideration of curative surgical resection and has engendered widespread pessimism regarding treatment. Early attempts to devise effective empiric chemotherapeutic regimens were also largely unsuccessful, yielding low response rates and negligible effect on survival. The clinical heterogeneity of these patients also has made the design and clinical evaluation of uniform treatments for this entire group problematic.
As effective combination chemotherapy has developed for certain cancers, it is not surprising that some patients with carcinoma of unknown primary site have also benefited. However, the potential advantage of delivering site-specific therapy versus empiric therapy is also increasing; for example, site-specific therapy for metastatic colorectal cancer now achieves median survival in excess of 24 months and differs markedly from the empiric regimens used for the treatment of carcinoma of unknown primary site (which yield median survivals of 8 to 11 months). Improved diagnostic methods, including more accurate immunohistochemical (IHC) stains and molecular tumor profiling, are developing and will likely allow site-specific therapy to be delivered to a larger percentage of CUP patients.
At present, optimum treatment for patients with CUP requires specific clinical and pathologic evaluation, for the purposes of (1) identifying the primary site if possible and (2) identifying favorable subsets of patients who require specific treatment. Empiric chemotherapy remains the standard treatment for the remainder of patients, but this group is likely to decrease in size as the roles of the new diagnostic methods are defined.
CLINICAL PRESENTATION
Owing to the heterogeneity of this syndrome and the underlying diseases represented, clinical presentations are extremely variable. Typically, the patient develops symptoms at a site of metastatic cancer; routine history, physical examination, chest radiograph, laboratory studies, and selected radiologic evaluation based on symptoms fail to identify the primary site. At this point, a biopsy of the most accessible metastatic site is usually performed, and metastatic carcinoma is documented. The patient should initially be considered to have a CUP if results of the relatively limited evaluation mentioned in the preceding text do not suggest a primary site.
After this limited initial evaluation, most patients require additional pathologic evaluation, as well as additional directed clinical evaluation based on the specific presentation. This chapter is divided into three sections. The first section reviews the pathologic evaluation and includes information on the emerging role of molecular tumor profiling. In the second section, the clinical evaluation of CUP patients is summarized. Finally, the treatment of patients with CUP is discussed, with emphasis on the specific treatable subsets.
PATHOLOGIC EVALUATION
The initial histologic diagnosis provides a useful basis for further evaluation. Almost all initial light microscopic diagnoses fall into one of five categories: poorly differentiated (anaplastic) neoplasm, poorly differentiated carcinoma (PDC), adenocarcinoma, squamous cell carcinoma, and neuroendocrine carcinoma. Occasional sarcomas or melanomas are also diagnosed without an obvious primary site, and management of these patients should follow established guidelines.
Histologic Subtypes
Poorly Differentiated Neoplasm
The diagnosis of poorly differentiated neoplasm using light microscopy implies the inability of the pathologist to distinguish between the major categories of malignant disease, including carcinoma, lymphoma, and various mesenchymal tumors. Approximately 5% of all CUP patients present with this histologic diagnosis, but few remain without undefined lineage after specialized pathologic study.1,2,3 In this group of patients, establishing a precise diagnosis is extremely important because various highly treatable cancers are poorly differentiated. In adults, the most frequent poorly differentiated neoplasm for which specific, highly effective therapy is available is non-Hodgkin lymphoma. When additional pathologic evaluation is performed, either using immunoperoxidase staining or electron microscopy, 34% to 66% of poorly differentiated neoplasms are found to be intermediate- or high-grade non-Hodgkin lymphoma.1,2 In children or young adults, various highly treatable malignancies can be identified, including various sarcomas, neuroblastoma, Ewing tumor, and germ cell tumors.
IHC staining, electron microscopy, and genetic analysis are helpful in the differential diagnosis. Initial communication with the pathologist is important because one common cause of a nonspecific light microscopic diagnosis is a small or poorly preserved biopsy specimen. Fine-needle aspiration biopsy
frequently provides an inadequate amount of tissue for optimal evaluation of poorly differentiated tumors because histology is difficult to evaluate and the ability to perform special studies is limited. A larger biopsy specimen is recommended and is often adequate to make a more specific diagnosis. When the tumor type is unclear after examination of an adequate biopsy specimen, additional pathologic studies are required.
frequently provides an inadequate amount of tissue for optimal evaluation of poorly differentiated tumors because histology is difficult to evaluate and the ability to perform special studies is limited. A larger biopsy specimen is recommended and is often adequate to make a more specific diagnosis. When the tumor type is unclear after examination of an adequate biopsy specimen, additional pathologic studies are required.
Poorly Differentiated Carcinoma
Approximately 20% of patients with carcinoma of unknown primary site have PDC, and an additional 10% have poorly differentiated adenocarcinoma (PDA). It is now recognized that some of these patients have neoplasms that are highly sensitive to chemotherapy and a small percentage is curable with appropriate therapy. Careful initial evaluation is therefore critical in this patient group to ensure that patients with highly responsive neoplasms are identified and treated.
Histologic features that can differentiate chemotherapyresponsive tumors from nonresponsive tumors have not been identified.4 Therefore, all PDCs should have additional pathologic study using IHC staining (see “Immunohistochemical Staining”).
Electron microscopy should be reserved for tumors in which IHC is not contributory because this technique is less widely available and requires specific tissue preparations. Lymphoma can be reliably differentiated from carcinoma in most instances. In addition, sarcoma, melanoma, mesothelioma, and neuroendocrine tumors can occasionally be defined by subcellular features.
Identification of tumor-specific cytogenetic abnormalities may be useful in patients with PDC. Specific chromosomal abnormalities include t(11;22) in patients with peripheral neuroepitheliomas or Ewing tumor5,6 and an isochrome of the short arm of chromosome 12 (i12p) in most patients with germ cell tumors.7 The clinical relevance of molecular and cytogenetic studies has been demonstrated in young patients with PDC or adenocarcinoma involving primarily midline structures or with elevated serum levels of human chorionic gonadotropin (HCG) or α-fetoprotein (AFP).8 In a group of 40 such patients, specific diagnoses were suggested by genetic studies in 17 (42%): germ cell tumor, 12; melanoma, 2; lymphoma, 1; peripheral neuroepithelioma, 1; and desmoplastic small cell tumor, 1. In the 12 patients with germ cell tumors, cisplatin-based therapy produced a 75% overall response rate, with 45% complete responses. Similar therapy yielded responses in only 17% of patients without specific abnormalities, with no complete responses.
Adenocarcinoma
Well-differentiated and moderately well-differentiated adenocarcinomas are the most common tumors identified using light microscopy and account for approximately 60% of CUP diagnoses (about 50,000 patients annually in the United States). As with most types of adenocarcinomas of known primary site, the incidence of adenocarcinoma of unknown primary site increases with advancing age. Most patients have multiple metastases. Common metastatic sites include the liver, lungs, bones, and lymph nodes.
The diagnosis of adenocarcinoma is usually made without difficulty on the basis of light microscopic features and is based on the formation of glandular structures by neoplastic cells. However, adenocarcinomas from many sites share these histologic features, and therefore, the site of tumor origin is usually impossible to pinpoint. Certain histologic features can suggest specific primary sites, but even these are usually not specific enough to make a definitive diagnosis. Examples include papillary features, typically seen in ovarian or thyroid cancers, and signet ring cells, typically associated with gastric adenocarcinoma.
The diagnosis of PDA should be interpreted with caution because patients with this diagnosis may be distinct from patients with well-differentiated adenocarcinoma in both tumor biology and responsiveness to systemic therapy. Criteria for the diagnosis of PDA may differ slightly among pathologists because there is clearly a spectrum of differentiation, ranging from very well-differentiated adenocarcinoma to completely anaplastic carcinoma that fails to show any differentiation. Minimal glandular formation or positive mucin staining in an otherwise PDC often results in the diagnosis of PDA. Additional pathologic study in these patients is therefore appropriate, as described for patients with PDC.
The identification of relatively cell-specific antigens by IHC staining has improved the ability to predict the site of origin in patients with adenocarcinoma of unknown primary site.3,9 Panels of IHC stains are most useful and should be directed by clinical features (e.g., sites of metastases and gender). Molecular tumor profiling assays also appear relatively accurate and often provide additional diagnostic information (see sections on “Immunohistochemical Staining” and “Molecular Tumor Profiling and CUP Classification”).
Squamous Carcinoma
Squamous carcinoma of unknown primary site represents approximately 5% of patients with CUP. The diagnosis of squamous carcinoma is usually made definitively by examination of histology. Additional pathologic evaluation is usually not necessary. Effective treatment is available for the majority of these patients, and appropriate clinical evaluation is important.
Neuroendocrine Carcinoma
Neuroendocrine tumors account for approximately 3% of all CUP. Improved pathologic methods for diagnosing neuroendocrine tumors have resulted in the recognition of an increased incidence and wider spectrum of these neoplasms. Neuroendocrine carcinomas with widely differing histologic and clinical features are represented, and accurate categorization is important in planning treatment.
Two subgroups of neuroendocrine carcinoma can be routinely recognized by histologic features. Well-differentiated (low-grade) neuroendocrine tumors share histologic features with carcinoids and islet cell tumors and frequently secrete bioactive substances. A second histologic group (variously
described as small-cell carcinoma, atypical carcinoid, or poorly differentiated neuroendocrine carcinoma) has typical neuroendocrine features and an aggressive biology.
described as small-cell carcinoma, atypical carcinoid, or poorly differentiated neuroendocrine carcinoma) has typical neuroendocrine features and an aggressive biology.
A third group of neuroendocrine carcinomas appears histologically as a poorly differentiated neoplasm or PDC. Accurate identification of these tumors requires IHC staining, and occasionally electron microscopy.
Immunohistochemical Staining
A large number of relatively sensitive and specific IHC stains are now widely available and frequently aid in the classification of poorly differentiated tumors. Immunoperoxidase reagents consist of either monoclonal or polyclonal antibodies directed at various specific cell components or products, including enzymes, normal tissue components, hormones, oncofetal antigens, and other tumor markers. Most stains can be performed using formalin-fixed, paraffinized tissue. Many new antibodies are being developed against various rather cell-specific proteins, making this area of diagnostic pathology a dynamic and evolving field.
Table 37-1 summarizes the IHC stains that are important in the diagnosis of specific tumor types. In the evaluation of poorly differentiated neoplasms, several important issues can usually be resolved by IHC staining. First, and most important, the leukocyte common antigen (LCA) stain can be used to identify non-Hodgkin lymphomas with a high level of accuracy.10 Poorly differentiated neoplasms diagnosed as non-Hodgkin lymphoma on the basis of IHC staining can be treated effectively with combination chemotherapy.2 Second, neuroendocrine tumors can be identified by staining for chromogranin and/or synaptophysin.11 Third, unsuspected melanomas and sarcomas can be diagnosed with reasonable accuracy.12,13 Finally, staining for germ cell tumors (HCG, AFP, OCT4, and PLAP) is suggestive in an appropriate clinical situation.14
Table 37-1 IHC Tumor Staining Patterns in the Differential Diagnosis of CUP | ||||||||||||||||||||||||||||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
The ability of IHC staining to identify the origin of various adenocarcinomas has improved, but in most cases the staining results must be interpreted in the context of clinical and histologic features. An exception is the PSA stain, which is very specific for prostate carcinoma.15 Stains suggestive of other primary sites are summarized in Table 37-1; specificity is improved using panels of stains.3,16,17,18
Several problems are associated with the IHC stains. Technical expertise is required to perform these tests accurately and reproducibly, and proper interpretation requires an experienced pathologist. None of the stains is entirely specific. For example, some carcinomas stain with vimentin, some sarcomas
stain with cytokeratins, and a wide variety of carcinomas do not always stain in the expected patterns.3,16 The typical staining patterns (Table 37-1) often overlap with the staining patterns of other adenocarcinomas, forcing the pathologist to consider two or three possible primary sites. However, the clinical setting should be used to direct the selection of IHC stains and may narrow the spectrum of possibilities if staining patterns are not completely specific. For example, a CK20+/CK7– staining pattern provides strong evidence for the colon as a primary site in a patient with mucin-positive adenocarcinoma and metastases limited to the liver. Conversely, IHC findings may direct additional diagnostic procedures; in the above example, a colonoscopy should be performed and may result in the identification of a primary site.
stain with cytokeratins, and a wide variety of carcinomas do not always stain in the expected patterns.3,16 The typical staining patterns (Table 37-1) often overlap with the staining patterns of other adenocarcinomas, forcing the pathologist to consider two or three possible primary sites. However, the clinical setting should be used to direct the selection of IHC stains and may narrow the spectrum of possibilities if staining patterns are not completely specific. For example, a CK20+/CK7– staining pattern provides strong evidence for the colon as a primary site in a patient with mucin-positive adenocarcinoma and metastases limited to the liver. Conversely, IHC findings may direct additional diagnostic procedures; in the above example, a colonoscopy should be performed and may result in the identification of a primary site.
In many cases, a single primary site cannot be identified with certainty even after histologic examination, IHC staining, and correlation with clinical features. Additional pathologic evaluation with either electron microscopy or a search for specific chromosomal abnormalities is useful in a few situations. In addition, molecular tumor profiling is a new technique that promises to be of importance in identifying the tissue of origin in patients with CUP.
Molecular Tumor Profiling and CUP Classification
Gene expression or molecular profiling of human neoplasms was first made possible by the development of DNA microarray analysis techniques.19,20 A pivotal study in cancer classification and diagnosis was reported by Golub et al.,21 who demonstrated for the first time that patterns of gene expression alone could discriminate acute myeloid leukemia from acute lymphoblastic leukemia. Other investigators demonstrated that numerous cancer types could be classified accurately by measuring the differential expression of specific gene sets.22,23,24,25 The basis of molecular profiling in recognizing specific cancer types is the identification of the genes responsible for the synthesis of proteins required for specific normal cellular functions (e.g., milk production in breast luminal duct cells, albumin production in hepatocytes, etc.). Cancer cells retain some normal cell-type specific functional characteristics, so their origin can usually be predicted from their gene-expression profile.25 Molecular profiling assays designed to determine the tissue of origin, therefore, measure gene expression dynamics in relation to cell lineage, rather than tumor-specific markers.
Patients with CUP have a clinically undefined primary tumor site, and are ideal candidates for classification by molecular profiling.26 Molecular profiling may identify the specific type of cancer present, and when used in concert with the clinical and pathologic features may be useful in predicting the site of tumor origin. Primary site identification in CUP would probably improve the therapeutic outcome by allowing site-specific therapy to be administered, rather than a single empiric regimen to all patients. In addition to defining the precise tumor type, molecular tumor profiling may aid in unraveling various gene-specific cancer-activated or overexpressed cellular pathways and in identifying new targets for therapy.27,28
Molecular profiling assays have been validated in patients with metastatic tumors of known primary site. When applied to biopsy specimens from a metastatic site, various molecular assays have correctly predicted the primary site in 76% to 89% of patients.23,29,30,31,32,33 Correct identification of the primary tumor type in CUP is difficult to validate because the primary tumor site is unknown and rarely becomes apparent during the subsequent clinical course of these patients. In several retrospective studies using archived biopsy specimens, molecular profiling assays predicted a tissue of origin in most cases.29,34,35,36,37,38 Although these predictions were generally consistent with clinical features, IHC staining patterns, and response to empiric therapy, no confirmation of accuracy was possible.
More direct evidence regarding the accuracy of molecular tumor profiling is now available from a study of CUP patients who had a primary site identified later during their clinical course (latent primary).39 Twenty such patients who had primary sites identified 2 to 54 months (median 10 months) after the initial diagnosis of CUP were identified. Since publication of the initial report, four additional patients have been identified. The initial diagnostic biopsies were evaluated by a molecular profiling assay using RT-PCR methodology (Cancer Type ID; BioTheranostics, Inc.) capable of identifying 32 tumor types. In 18 of 24 biopsies (75%), the primary tumor was accurately predicted (matched the latent primary site identified), providing direct validation of the accuracy of a molecular profiling assay in identifying the tissue of origin in CUP patients.
Developing evidence therefore indicates that molecular tumor profiling can accurately predict the tissue of origin in a majority of patients with CUP and is an important new diagnostic tool. Although its exact role in the diagnostic evaluation awaits definition, molecular tumor profiling is probably more accurate than IHC staining in many patients with CUP. However, there is little published data regarding the impact of diagnoses made by molecular profiling on results of treatment. Until such data exist, these patients should still be considered to have CUP when planning management. In some cases, consideration of treatment based on the predicted primary site is now appropriate (see “Treatment” section).
CLINICAL FEATURES AND EVALUATION
In most patients with CUP, the clinical course is dominated by symptoms related to metastases; the primary site becomes obvious in only 5% to 10% of patients during their life-time. At autopsy, a primary site is identified in about 75% of patients.40,41,42 Primary sites in the pancreas, lung, colorectum, and liver account for approximately 60% of those identified. Primary sites in the breast, ovary, and prostate are uncommon in autopsy series.
Although some clinical differences exist, there is a substantial overlap between the clinical features of patients with adenocarcinoma, PDA, and PDC. Patients with PDC are somewhat younger, and usually exhibit rapid tumor growth. These patients may also have more frequent location of dominant metastatic sites in the mediastinum, retroperitoneum, and peripheral lymph nodes. Because of the similarities, the clinical evaluation of patients with these histologies should follow the same guidelines. Patients with neuroendocrine carcinoma and
squamous carcinoma of unknown primary site are discussed separately.
squamous carcinoma of unknown primary site are discussed separately.
Clinical Evaluation
Routine initial evaluation should include a history and physical examination, standard laboratory screening tests (complete blood count, chemistry profile, and urinalysis), and CT of the chest, abdomen, and pelvis. In addition, specific symptoms should lead to directed radiologic evaluation. A generalized evaluation of asymptomatic areas, including endoscopy of the gastrointestinal (GI) tract, is not recommended.
A few additional diagnostic tests enable the identification of a primary site in a minority of patients. Positron emission tomographic (PET) scanning should routinely be considered, although definitive data in large numbers of patients have not been published.43 Serum PSA should be measured in all males because this marker is quite specific for prostate cancer; all females with clinical presentation compatible with metastatic breast cancer should undergo mammography. Patients with certain histologies (e.g., squamous carcinoma and neuroendocrine carcinoma) should have additional evaluation (see subsequent sections). Serum tumor marker levels, including carcinoembryonic antigen (CEA), CA19-9, CA15-3, and CA-125, are frequently elevated and provide neither diagnostic nor prognostic information in this group of patients.44,45 However, these markers, if elevated, may be useful in assessing response to treatment.
Based on the findings of standard clinical and pathologic evaluation, additional specific evaluation should be triggered in some patients. Table 37-2 summarizes the additional evaluation indicated for several common clinical presentations. These additional tests can sufficiently narrow the diagnostic spectrum so that treatment based on the likely tissue of origin can be selected, even though the primary is not actually identified. Although molecular tumor profiling is not yet considered a standard component of the diagnostic evaluation and is not included in Table 37-2, the authors believe that this test will become standard in the future and can be considered in selected patients.
Table 37-2 Additional Evaluation of Specific Patient Subsets Defined by Initial Clinicopathologic Evaluation | ||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| ||||||||||||||||||||||||||
Neuroendocrine Carcinoma
Although the initial clinical evaluation is the same, patients with neuroendocrine carcinoma require special consideration in determining appropriate treatment. Of major importance is the separation of this group into tumors with low grade histology and indolent clinical course versus those likely to have an aggressive clinical course. This distinction can usually be made by the pathologist: patients with classical carcinoid tumors typically have indolent histology, whereas those with small-cell neuroendocrine carcinoma or PDC/positive neuroendocrine IHC stains have aggressive cancers.
Low-grade neuroendocrine carcinomas, when presenting with an unknown primary site, most frequently involve the liver. Other metastatic sites include lymph nodes (usually abdominal or mediastinal) and bone. As the primary sites for these tumors frequently involve the GI tract, upper and lower GI endoscopy should be performed. Various syndromes caused by secretion of bioactive peptides (carcinoid syndrome, glucagonoma syndrome, VIPomas, and Zollinger-Ellison syndrome) may also be associated with low-grade neuroendocrine carcinomas. Additional clinical evaluation in these patients should include serum or urine screening for these substances.
Aggressive neuroendocrine carcinomas of unknown primary site are usually found in multiple metastatic sites, and rarely secrete bioactive peptides. Patients with a history of cigarette smoking should be suspected of having a lung primary, particularly if the tumor has small-cell histology, and a fiberoptic bronchoscopy should be performed. Patients with a positive tumor cell IHC stain for TTF-1 should also have bronchoscopy. Extrapulmonary small-cell carcinomas arising from various other sites (salivary glands, paranasal sinuses, esophagus, pancreas, colorectum, bladder, prostate, uterus, and cervix) have been described and are occasionally identified during clinical evaluation.
The origin of these poorly differentiated neuroendocrine carcinomas remains unclear, but it is likely that the group is heterogeneous. Some of these patients may have small-cell lung cancer with an occult primary tumor. However, many of these patients have no smoking history, and the absence of overt pulmonary involvement makes this diagnosis unlikely. It is probable that some of these tumors are undifferentiated variants of well-recognized neuroendocrine tumors (e.g., carcinoid tumor), without a recognizable primary site. In the undifferentiated form, the clinical and pathologic characteristics no longer resemble the characteristics of the more differentiated counterpart. Anaplastic carcinoid tumors arising in the GI tract are highly responsive to cisplatin-based chemotherapy, whereas carcinoid tumors with typical histology are usually resistant to cisplatin.46
Squamous Carcinoma
Cervical lymph node presentation
The cervical lymph nodes are the most common metastatic site for squamous carcinoma of unknown primary. Patients developing this syndrome share risk factors with patients who develop squamous cancers of the head and neck area and often have substantial histories of tobacco and alcohol use. In most instances, the upper or middle cervical lymph nodes are involved, and in these patients, a thorough search for a primary site in the head and neck area is an important part of the initial evaluation. Direct visualization of the oropharynx, hypopharynx, larynx, upper esophagus, and nasopharynx is recommended, with biopsy of any suspicious areas. CT of the neck is useful in defining the extent of disease and occasionally identifies a primary site. PET scanning identifies an occult primary in approximately 25% of patients and should be included as a standard diagnostic procedure.47
Detection of the Epstein-Barr viral genome in the tumor tissue is highly suggestive of a nasopharyngeal primary site48,49; this test should be considered in young patients with PDC, particularly nonsmokers. When the lower cervical or supraclavicular lymph nodes are involved, primary lung cancer should be suspected, and fiberoptic bronchoscopy should be included in the diagnostic evaluation. In one large series of patients who developed cervical adenopathy as the initial sign of metastatic squamous carcinoma, this diagnostic workup revealed a primary site in 231 of 267 patients (87%).50
Ipsilateral tonsillectomy has been recommended as a diagnostic modality in patients with squamous carcinoma involving a single subdigastric, midjugulocarotid, or submandibular lymph node, and bilateral tonsillectomy in patients presenting with bilateral subdigastric adenopathy.51 In one series of 87 patients who had tonsillectomy as part of their workup for cervical node presentations, 26% had a primary site identified in the tonsil.52 The advantages of identifying the primary site in this group of patients include a more specific treatment plan, reduced size of radiation therapy ports, more accurate prognosis, and perhaps easier follow-up.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
