Overview of the classification and management of cancers of unknown primary site

INTRODUCTION — Cancer of unknown primary site (CUP) is a relatively common clinical entity, accounting for approximately 2 percent of all invasive cancers [1]. Within this category, tumors from many primary sites with varying biology are represented. This heterogeneity has made the design and interpretation of clinical studies difficult.

A general overview of the classification and management of CUP is presented here. Further specific details on the evaluation and management of specific subgroups of CUP are discussed separately.

●(See "Adenocarcinoma of unknown primary site".)

●(See "Squamous cell carcinoma of unknown primary site".)

●(See "Neuroendocrine neoplasms of unknown primary site".)

●(See "Poorly differentiated cancer from an unknown primary site".)

INITIAL DIAGNOSTIC EVALUATION — Patients with CUPs typically present with symptoms referable to metastases. The initial diagnosis of advanced cancer is confirmed by biopsy of a metastatic lesion. The diagnosis of CUP is made only after specific diagnostic tests are performed and do not identify an anatomic primary site. (See 'Subsequent evaluation of CUP' below.)

The tests which are successful in identifying a primary site in most patients with advanced cancer include:

●History, physical examination (including pelvic exam in females).

●Complete blood counts, serum chemistries, urinalysis.

●Serum prostate-specific antigen (PSA; in males).

●Computed tomography (CT) scans of chest, abdomen, and pelvis (these procedures identify a primary site in 10 to 35 percent of patients [2,3]).

●Mammography (females).

●Focused evaluation of specific signs/symptoms.

SUBSEQUENT EVALUATION OF CUP — After the diagnosis of CUP is established, patients are classified into one of the following four categories based on the light microscopic examination of the initial biopsy. These classifications are used to guide further diagnostic evaluation. Patients in these histology-based groups also vary with respect to their clinical characteristics, treatment, and prognosis.

●Adenocarcinoma. (See 'Adenocarcinoma' below.)

●Squamous cell carcinoma. (See 'Squamous cell carcinoma' below.)

●Neuroendocrine carcinoma, which may be either well differentiated or poorly differentiated. (See 'Neuroendocrine tumors' below.)

●Poorly differentiated tumors, most of which are recognized as carcinomas on histologic examination. Less frequently, the lineage (carcinoma, lymphoma, sarcoma, melanoma, germ cell tumor) is unclear after light microscopic examination. (See 'Poorly differentiated cancers' below.)

Most CUP patients receive further evaluation; this may include further clinical diagnostic procedures, but it almost always includes specialized pathologic testing to refine the initial histologic diagnosis. Still, in many patients with CUP, the anatomic primary site is often difficult to identify despite receiving optimal clinical and pathologic diagnostic evaluation [4].

Adenocarcinoma — Adenocarcinomas of unknown primary site constitute approximately 70 percent of CUPs. The clinical presentation of patients with adenocarcinoma is determined by the sites of tumor involvement, which frequently include the liver, lungs, lymph nodes, and/or bones. (See "Adenocarcinoma of unknown primary site".)

In autopsy series of patients with adenocarcinoma of unknown primary site, the most frequently identified malignancies were lung, pancreas, hepatobiliary tree, and kidney, together accounting for approximately two-thirds of cases [5]. However, tumors arising from a wide variety of other primary sites were also identified. Adenocarcinomas of the breast and prostate were infrequently identified in autopsy series, despite being the most common cancers in females and males, respectively [6].

In specific patient subgroups, several additional clinical studies are useful in identifying the anatomic primary site and thus in guiding therapy:

●In females with clinical presentations suggestive of metastatic breast cancer (eg, axillary adenopathy), breast magnetic resonance imaging (MRI) should be considered even if mammograms are normal. Immunochemical staining of the biopsy specimen may detect the expression of breast cancer-specific markers, such as estrogen and progesterone receptors, mammaglobin, and gross cystic duct fluid protein. (See "Axillary node metastases with occult primary breast cancer", section on 'Initial diagnostic workup'.)

●Positron emission tomography (PET) is useful in specific clinical settings [7-9], but in most patients, it does not add substantially to the information obtained by computed tomography (CT) scans [10].

●Additional signs or symptoms, such as a positive test for occult blood in stool, should be evaluated with appropriate radiographic or endoscopic studies. Colonoscopy should also be performed in patients with intra-abdominal metastases (liver, peritoneum) and pathologic findings (histology and/or immunohistochemistry) typical of colon cancer, even in the absence of symptoms referable to the gastrointestinal tract. However, routine endoscopic evaluation of the upper and/or lower gastrointestinal tract has a low yield in other asymptomatic patients.

●In males with clinical presentations suggestive of metastatic prostate cancer (eg, blastic bone metastases), tissue staining for prostate-specific antigen (PSA) is sometimes diagnostic even when serum PSA is not elevated.

The most important improvements to the evaluation of CUP are diagnostic tests performed on biopsy tissue. These tests can usually predict the tissue of origin, even when an anatomic primary site is not evident. If an anatomic primary site is identified, the patient no longer has CUP, and should receive site-specific treatment. These diagnostic methods include:

●Immunohistochemistry – Improved immunohistochemistry facilitates the identification of specific cancer types, including breast, colorectal, lung, renal, and thyroid carcinomas (table 1).

●Molecular cancer classifier assays – Molecular cancer classifier assays (MCCAs) use site-specific gene expression profiles to identify the site of origin of metastatic lesions in patients with CUP. Most studies evaluating the use of MCCA-directed site-specific therapy have been conducted in patients with adenocarcinomas of unknown primary site. (See "Adenocarcinoma of unknown primary site", section on 'Molecular cancer classifier assays'.)

Traditional tumor markers (carcinoembryonic antigen [CEA], cancer antigen [CA] 19-9, CA 15-3, CA 125) generally are not useful as either diagnostic or prognostic tests. These markers are often elevated in the serum of patients with adenocarcinoma of unknown primary site, and serial measurement may be useful in following the response to therapy for individual tumors. (See "Clinical manifestations, diagnosis, and staging of exocrine pancreatic cancer" and "Overview of the approach to metastatic breast cancer" and "Screening for ovarian cancer".)

Neuroendocrine tumors — Neuroendocrine tumors of unknown primary site comprise approximately 1 percent of CUPs and can occur in a variety of sites (table 2). (See "Neuroendocrine neoplasms of unknown primary site".)

The 2000 World Health Organization (WHO) classification system divides these tumors into low-grade and high-grade malignancies:

●Low-grade neuroendocrine tumors – Metastatic carcinoid or islet cell tumors sometimes present without an obvious primary site, usually with liver metastases. In some of these patients, primary sites are subsequently identified in the intestine or pancreas.

●High-grade neuroendocrine carcinoma – Most patients with high-grade neuroendocrine carcinoma have an aggressive malignancy, usually with metastases in multiple sites [11]. Lymph nodes in the retroperitoneum and mediastinum are frequently involved.

•Some patients in this group have metastatic small cell anaplastic carcinoma, which often represents a primary bronchogenic malignancy. CT of the chest and fiberoptic bronchoscopy may identify the primary site. Small cell carcinomas have also been described arising in a wide range of extrapulmonary sites, including the salivary glands, esophagus, bladder, ovary, prostate, and cervix, among others. (See "Extrapulmonary small cell cancer".)

•Other high-grade neuroendocrine malignancies appear histologically as poorly differentiated carcinomas and are identifiable as neuroendocrine carcinomas only when immunohistochemical [IHC] stains or molecular cancer classifier assays (MCCAs) are performed. (See "Poorly differentiated cancer from an unknown primary site", section on 'Molecular cancer classifier assays'.)

Squamous cell carcinoma — Squamous cell carcinoma is uncommon in the absence of an obvious primary tumor, and accounts for approximately 5 percent of CUPs. In many patients, squamous CUP is associated with human papillomavirus (HPV). (See "Squamous cell carcinoma of unknown primary site".)

The clinical presentation and subsequent diagnostic evaluation of patients with squamous cell carcinoma of unknown primary site depend on the predominant area of the metastasis:

●Upper and midcervical lymphadenopathy – Upper and midcervical lymphadenopathy is the most common clinical presentation of squamous CUP and is most frequently due to cancer arising in the head and neck.

●Lower cervical or supraclavicular lymphadenopathy – For patients with lower cervical or supraclavicular lymph nodes should be evaluated for an occult lung or head and neck primary tumor.

●Inguinal, pelvic, or retroperitoneal lymphadenopathy – Most patients with squamous cell carcinoma involving the inguinal, pelvic, or retroperitoneal lymph nodes should be evaluated for an anogenital primary.

●Adenopathy at other sites – Metastatic SCC at other sites is rare. Possible primaries include lung, skin, uterine, cervix, anal canal, or esophagus.

An MCCA may also be of value in identifying the tissue of origin in patients with squamous cell carcinoma of unknown primary site, if the evaluation described above does not identify an anatomic primary site. (See "Adenocarcinoma of unknown primary site", section on 'Molecular cancer classifier assays' and "Poorly differentiated cancer from an unknown primary site", section on 'Molecular cancer classifier assays'.)

Poorly differentiated cancers — Approximately 20 to 25 percent of all CUPs have poorly differentiated histology. Approximately 80 percent of these can be recognized as poorly differentiated carcinoma by light microscopic examination. The remainder are termed "poorly differentiated neoplasms" after initial examination, implying that the pathologist is confident of the diagnosis of malignancy but cannot distinguish between a carcinoma, sarcoma, melanoma, and hematologic malignancy based on light microscopy. (See "Poorly differentiated cancer from an unknown primary site".)

Additional pathologic and molecular evaluation is essential in all poorly differentiated CUPs. The group of cancers initially termed "poorly differentiated neoplasms" contains a number of highly treatable entities, including lymphoma, germ cell tumor, and neuroendocrine carcinoma. The identity of tumors called "poorly differentiated carcinomas" by light microscopy can also be clarified. This group contains poorly differentiated neuroendocrine carcinomas and, occasionally, non-carcinomas. Additional pathologic studies are almost always successful in identifying the lineage of these tumors, and we routinely include immunohistochemistry and MCCA [12]. Electron microscopy and chromosome analysis may occasionally aid in establishing a diagnosis. (See "Poorly differentiated cancer from an unknown primary site", section on 'Clinical evaluation'.)

Diagnosing an occult germ cell tumor is of particular importance in this group because of the responsiveness of germ cell tumors to chemotherapy. In addition to specialized pathologic studies, elevated serum concentrations of human chorionic gonadotropin (hCG) or alpha-fetoprotein (AFP) may suggest the diagnosis of extragonadal germ cell tumor. (See "Extragonadal germ cell tumors involving the mediastinum and retroperitoneum".)

TREATMENT — If an anatomic primary site is identified in a patient with CUP, the patient should no longer be considered to have CUP, and treatment should follow standard guidelines for the identified tumor type.

Several subgroups of CUP patients can be identified using the clinical and pathologic evaluation described above. Patients in the following subsets, which now comprise approximately 40 percent of all CUP patients, should receive specific therapy (table 3):

●Females with adenocarcinoma and isolated axillary lymphadenopathy, who may benefit from treatment as for stage II breast cancer. (See "Axillary node metastases with occult primary breast cancer".)

●Females with peritoneal carcinomatosis, who may respond to treatment patterned after that used for stage III ovarian cancer. (See "Adenocarcinoma of unknown primary site", section on 'Females with peritoneal carcinomatosis'.)

●Patients with clinical features and pathologic studies consistent with a colorectal primary tumor may respond well to treatment for metastatic colon cancer. This profile includes metastases involving the liver and/or peritoneum, typical histology, and immunohistochemical [IHC] staining that is cytokeratin (CK) 20 positive/CK7 negative or caudal type homeobox 2 (CDX2) positive. (See "Adenocarcinoma of unknown primary site", section on 'Colon cancer profile'.)

●Patients with clinical features and pathologic studies consistent with non-small cell lung cancer (NSCLC) may respond to treatment for metastatic lung cancer [13]. Pathologic features for this profile can include (see "Adenocarcinoma of unknown primary site", section on 'Lung adenocarcinoma profile'):

•Mediastinal and/or hilar adenopathy, often accompanied by metastases at other sites

•Adenocarcinoma or squamous histology

•Typical immunohistochemistry (IHC) staining pattern (TTF-1 positive, CK7 positive, CK20 negative)

•Molecular cancer classifier assay (MCCA) prediction of NSCLC

Management includes initial testing for programmed cell death ligand 1 (PD-L1) and critical driver mutations such as epidermal growth factor receptor (EGFR), the anaplastic lymphoma kinase (ALK) fusion oncogene, and other targetable mutations to guide selection of treatment. Further details on these molecular alterations and the treatment approach to lung cancer are discussed in detail separately. (See "Overview of the initial treatment and prognosis of lung cancer" and "Overview of the initial treatment of advanced non-small cell lung cancer".)

●Patients with specific pathologic findings may respond to treatment for metastatic renal cell carcinoma (RCC) [13,14]. Pathologic features for this profile can include: (See "Adenocarcinoma of unknown primary site", section on 'Renal cell carcinoma profile'.)

•Clear cell or papillary histology

•IHC staining for RCC marker, CD10, and PAX8

•MCCA diagnosis of RCC

The management of advanced and metastatic RCC is discussed separately. (See "Overview of the treatment of renal cell carcinoma" and "Systemic therapy of advanced clear cell renal carcinoma" and "The treatment of advanced non-clear cell renal carcinoma".)

●Patients with clinical features and pathologic studies consistent with thyroid carcinoma may respond to treatment for metastatic follicular/papillary thyroid cancer. Pathologic features for this profile can include (see "Adenocarcinoma of unknown primary site", section on 'Thyroid carcinoma profile'):

•Metastases in the cervical or mediastinal lymph nodes, lungs, or bones

•Elevated serum thyroglobulin levels

•IHC staining for thyroglobulin

The management of advanced follicular and papillary thyroid cancer is discussed separately. (See "Follicular thyroid cancer (including oncocytic carcinoma of the thyroid)", section on 'Treatment' and "Differentiated thyroid cancer: Overview of management".)

●Males with blastic bone metastases or an elevated serum prostate-specific antigen (PSA), in whom treatment for metastatic prostate cancer may be useful. (See "Overview of systemic treatment for recurrent or metastatic castration-sensitive prostate cancer".)

●Squamous cell carcinoma limited to the cervical lymph nodes, which may be a manifestation of locally advanced head and neck cancer with an occult primary and should be treated as a locally advanced head and neck cancer. (See "Squamous cell carcinoma of unknown primary site".)

●Young males with poorly differentiated mediastinal or retroperitoneal tumors, which may respond to treatment as extragonadal germ cell tumors. (See "Extragonadal germ cell tumors involving the mediastinum and retroperitoneum".)

●Well-differentiated neuroendocrine carcinoma, for which treatment is similar to that used for advanced carcinoid tumors. (See "Neuroendocrine neoplasms of unknown primary site".)

●Poorly differentiated neuroendocrine carcinoma, for which treatment is similar to that used for small cell lung cancer. (See "Neuroendocrine neoplasms of unknown primary site" and "Extrapulmonary small cell cancer".)

Approximately 60 percent of patients with CUP do not fit into any of the above subgroups. For these patients, empiric chemotherapy has been considered the standard treatment. Such treatment, usually with combinations containing a platinum agent plus other cytotoxic agents (taxanes, gemcitabine, irinotecan), has produced median survivals of 7 to 10 months and two-year survival rates of 20 to 25 percent [15]. This treatment represents an improvement when compared with historical results (median survival four to six months) [16-18]; however, the majority of patients derive only modest benefit. In addition, the improvement of therapy for many advanced solid tumors makes it increasingly unlikely that optimal therapeutic results can be achieved using a single "broad-spectrum" chemotherapy regimen. (See "Adenocarcinoma of unknown primary site", section on 'Empiric chemotherapy'.)

With improvements in diagnosis, particularly IHC and gene expression profiling, the tissue of origin can be accurately predicted in the majority of patients even though the anatomic primary site cannot be identified [19]. For patients who do not fit into a specific treatable subset, data are now compelling to support site-specific treatment based on prediction of the site of origin by an MCCA [19]. For patients who do not fit into a specific treatable subset, data are now compelling to support site-specific treatment based on prediction of the site of origin by an MCCA [20-22]. The benefits of this approach are most evident in patients predicted to have treatment-responsive tumor types [23,24]. (See "Adenocarcinoma of unknown primary site", section on 'Molecular cancer classifier assays'.)

Comprehensive molecular profiling (CMP) is an increasingly important adjunct to optimal cancer treatment. Most common cancer types have one or more identified molecular alterations for which specific targeted therapy has proven efficacious. In addition, therapies targeting several molecular abnormalities (eg, high levels of microsatellite instability [MSI-H], tumor mutational burden [TMB], TRK fusions) are tumor type-agnostic. Targetable molecular alterations are identified in a sizable percentage of CUP patients [25]; ongoing studies are addressing the therapeutic value of these targets. (See "Adenocarcinoma of unknown primary site", section on 'Comprehensive molecular profiling' and "Tissue-agnostic cancer therapy: DNA mismatch repair deficiency, tumor mutational burden, and response to immune checkpoint blockade in solid tumors" and "TRK fusion-positive cancers and TRK inhibitor therapy".)

The treatment of patients with CUP are diagrammed in the algorithm (algorithm 1). Although not yet included in standard treatment guidelines (eg, National Comprehensive Cancer Network [NCCN] guidelines), we are convinced that site-specific therapy based on predictions by immunohistochemistry, MCCA, and CMP should be a part of the standard management of CUP patients who do not fit into a specific subset. Empiric chemotherapy should be considered for patients in whom specialized diagnostic tests are unsuccessful in predicting a primary site.

SUMMARY AND RECOMMENDATIONS

●Epidemiology – Cancer of unknown primary (CUP) accounts for 1 to 2 percent of all cancer diagnoses. In many patients with CUP, the anatomic primary site is often difficult to identify despite optimal clinical and pathologic diagnostic evaluation. (See 'Introduction' above.)

●Initial diagnostic evaluation – The initial diagnosis of advanced cancer is confirmed by biopsy of a metastatic lesion. The diagnosis of CUP is made only after specific diagnostic tests are performed and do not identify an anatomic primary site. (See 'Initial diagnostic evaluation' above.)

●Classification of CUP – CUPs can be classified into four categories (see 'Subsequent evaluation of CUP' above):

•Adenocarcinoma (see "Adenocarcinoma of unknown primary site")

•Squamous cell carcinoma (see "Squamous cell carcinoma of unknown primary site")

•Neuroendocrine carcinoma, which may be either well differentiated or poorly differentiated (see "Neuroendocrine neoplasms of unknown primary site")

•Poorly differentiated cancer (see "Poorly differentiated cancer from an unknown primary site")

●Treatment – Appropriate classification can identify the approximately 40 percent of CUP patients for whom a specific treatment may be particularly useful (table 3). For the remainder of patients, empiric chemotherapy is the standard treatment. (See 'Treatment' above.)

●Predicting the tissue of origin/site-specific therapy – However, accurate prediction of the tissue of origin also allows site-specific therapy in most patients. This can be achieved using techniques such as immunohistochemical [IHC] staining, molecular cancer classifier assays (MCCAs) and/or identification of targetable molecular alterations by comprehensive molecular profiling (CMP). (See "Adenocarcinoma of unknown primary site", section on 'Approach to patients not included in specific subgroups' and "Poorly differentiated cancer from an unknown primary site", section on 'Molecular cancer classifier assays'.)