Adenocarcinoma 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]. CUP is diagnosed in patients with metastatic cancer, but no anatomic primary site identified on a comprehensive initial evaluation. Cancers from many primary sites with varying biology are represented in patients with CUP. In many patients with CUP, improved diagnostic methods such as molecular cancer classifier assays (MCCAs; also known as gene expression profile assays) and immunohistochemistry (IHC) staining can predict the site of tumor origin. However, patients with CUP remain a clinically distinct group since the anatomic primary site is usually not detected during the clinical course. (See "Overview of the classification and management of cancers of unknown primary site".)

The diagnosis and management of patients with adenocarcinoma of unknown primary site are reviewed here. The diagnosis and management of the other CUPs are discussed separately.

●(See "Overview of the classification and management of cancers of unknown primary site".)

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

●(See "Head and neck squamous cell carcinoma of unknown primary".)

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

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

●(See "Axillary node metastases with occult primary breast cancer".)

CLINICAL PRESENTATION — 

Adenocarcinomas of unknown primary sites comprise approximately 70 percent of cancer of unknown primary sites (CUPs). In autopsy series, although these cancers may arise from a wide variety of primary tumor sites, the most frequently identified primary sites are lung, pancreas, hepatobiliary tree, and kidney, together accounting for approximately two-thirds of cases [2]. Adenocarcinomas of the breast and prostate are identified infrequently at autopsy, despite being the most common cancers in females and males, respectively. No primary site can be identified in 20 to 30 percent of patients. However, very small primaries that would require multiple sections for microscopic identification may be missed at autopsy. Autopsy series include patients who were not evaluated with modern imaging such as computed tomography (CT) and positron emission tomography (PET), and therefore published data may not accurately reflect contemporary patient population with adenocarcinoma of unknown primary site.

The incidence of adenocarcinoma of unknown primary site increases with age. The clinical presentation is determined by the sites of metastatic tumor involvement, which are frequently multiple and often include the liver, lungs, lymph nodes, and bones. Many patients with adenocarcinoma of unknown primary site have widespread metastases and poor performance status at diagnosis. The prognosis for most of these patients is poor, although this is influenced by the type of adenocarcinoma, sites of metastases, extent of tumor burden, performance status, and disease management.

In an analysis of almost 19,000 patients with CUP from the Swedish Cancer Registry from 1987 to 2008, the median survival for those with adenocarcinoma (70 percent of the group) was three months, with a one-year overall survival (OS) of 17 percent [3]. Another analysis from this cancer registry found an improvement over time in OS in patients with adenocarcinoma (median OS of approximately six months between 2001 and 2008 versus four months between 1987 and 1993) [4]. In both registry analyses, all patients with the diagnosis of CUP were included, regardless of performance status or treatment received. Most of these patients were treated with empiric chemotherapy.

However, several subsets of patients have a more favorable outlook, and the initial evaluation should attempt to identify these patients. Additional diagnostic tests including immunohistochemistry (IHC) stains, molecular cancer classifier assays (MCCAs), and comprehensive molecular profiling (CMP) have improved the identification of the site of tumor origin and improved the prognosis of selected patients when treated with site-specific and molecularly-guided therapy (MGT). (See 'Initial evaluation of the tumor specimen' below.)

DIAGNOSTIC EVALUATION — 

Most of the necessary clinical studies should have already been performed in the process of making the diagnosis of cancer of unknown primary site (CUP). These studies include a thorough history and physical examination, complete blood count, urinalysis, basic serum chemistries, and contrast-enhanced CT or gadolinium-enhanced magnetic resonance imaging (MRI) of the chest, abdomen, and pelvis. In males, assessment should incorporate a prostate examination and measurement of serum prostate-specific antigen (PSA). In females, the evaluation should include a pelvic examination and mammography. In addition, patients should have a focused clinical evaluation of any signs or symptoms suggestive of metastatic cancer.

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

●In females with a clinical presentation suggestive of metastatic breast cancer (eg, axillary adenopathy), breast MRI should be performed even if mammograms are normal. Immunohistochemistry (IHC) 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 'Females with axillary lymph node metastases' below.)

●Colonoscopy should be performed in patients with intra-abdominal metastases who have histology typical of lower gastrointestinal cancer and either typical IHC staining (cytokeratin 20 [CK20]-positive, cytokeratin 7 [CK7]-negative, caudal-type homeobox transcription factor 2 [CDX-2]-positive) or a molecular cancer classifier assay (MCCA) diagnosis of a colorectal site of origin. (See 'Colon cancer profile' below.)

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

Positron emission tomography (PET) imaging is an additional standard diagnostic staging procedure that may be useful in certain presentations. In a number of retrospective series, PET identified a primary site in approximately 40 percent of patients [5,6]. However, in a single existing prospective study, PET was not superior compared with CT [7]. Therefore, the use of PET should be restricted to the evaluation of patients with specific clinical presentations (eg, those with squamous carcinoma in cervical lymph nodes or with a single metastasis). Although unrelated to the diagnostic evaluation, PET scans are also useful for monitoring response to treatment in patients with bone-predominant metastases.

Exhaustive imaging and endoscopic testing should not be performed since these studies rarely detect the primary site in the asymptomatic patient, and confusion can result from false-positive results.

Traditional serum tumor markers (carcinoembryonic antigen [CEA], cancer antigen [CA] 19-9, CA 15-3, CA 125) are generally not useful as diagnostic or prognostic tests. However, they are commonly elevated and may be useful in following the response to therapy.

INITIAL EVALUATION OF THE TUMOR SPECIMEN

General approach — Cancer of unknown primary site (CUP) is, by definition, a metastatic tumor for which pretreatment evaluation does not reveal an anatomic primary site. In cases of suspected CUP, a biopsy of the most accessible site should be performed, preferably using a core needle or excisional biopsy to obtain sufficient tissue for all necessary studies. Adenocarcinoma can usually be distinguished from other histologies by light microscopic examination. Immunohistochemistry (IHC) and additional studies should be guided by the tumor histology, as described below. Treatment should then proceed based on results of these assessments (algorithm 1).

Light microscopy — The diagnosis of adenocarcinoma is usually based on the identification of glandular structures that are formed by the neoplastic cells. These features are shared by all adenocarcinomas, and the site of the primary tumor usually cannot be determined by light microscopy. Although certain morphologic features can be associated with a particular tumor type (eg, papillary features with ovarian cancer and signet ring cells with gastric cancer), they generally are not sufficiently specific to provide a definitive diagnosis.

The diagnosis of poorly differentiated adenocarcinoma is usually made when only minimal glandular formation is seen on histologic examination or in tumors that lack glandular differentiation but stain positively for mucin. Adenocarcinoma, poorly differentiated adenocarcinoma, and poorly differentiated carcinoma are histologic diagnoses that represent a spectrum of tumor differentiation rather than well-demarcated entities. Different pathologists may use somewhat different criteria for each of these diagnoses.

The light microscopic diagnosis of poorly differentiated adenocarcinoma should be interpreted with caution since some of these patients have a distinctive tumor biology and responsiveness to systemic chemotherapy. For this reason, evaluation and treatment of patients with poorly differentiated adenocarcinoma of unknown primary site should follow the guidelines outlined for poorly differentiated carcinoma of unknown primary site, including the use of IHC staining, molecular cancer classifier assay (MCCA), and electron microscopy (if necessary) to identify potentially chemotherapy-responsive cancer and rule out other tumor types such as hematologic malignancies, sarcomas, neuroendocrine carcinomas, or germ cell tumors. (See "Poorly differentiated cancer from an unknown primary site", section on 'Clinical evaluation'.)

Immunohistochemistry — In most instances, IHC is successful in defining the tumor lineage of poorly differentiated neoplasms (table 1). However, IHC allows the determination of the tissue of origin in only a minority of adenocarcinomas of unknown primary site [8]. In part, this is due to the atypical staining patterns present in many adenocarcinomas. However, selection of the appropriate IHC stains is also problematic. It is not possible to do every possible IHC stain (table 2) due to limitations of available tissue and expense; rather, the pathologist must select stains based on suggestive histologic or clinical findings [9].

In evaluating a CUP, most pathologists start with a panel of four IHC stains that form the basis of several potentially diagnostic patterns (cytokeratin 7 [CK7], cytokeratin 20 [CK20], thyroid transcription factor-1 [TTF-1], caudal-type homeobox transcription factor 2 [CDX-2]) to narrow the diagnostic possibilities and add stains based on histology, clinical presentation, and results of the initial IHC panel [8]. With this approach, results are strongly suggestive of a single cancer type in only 33 percent of patients with CUP [10]. Even in these patients, clinicians have hesitated to use the IHC predictions to guide site-specific treatment, since the pathology reports are often somewhat equivocal, with phrases like "consistent with" or "favor," rather than a firm diagnosis.

In a few situations, IHC provides strong evidence regarding the primary site (table 1 and table 2) [11]:

●Positive staining for prostate-specific antigen (PSA) is quite specific for prostate cancer and should be included in the evaluation of males with adenocarcinoma of unknown primary site.

●Positive staining for thyroglobulin in concert with TTF-1 is relatively specific for thyroid cancer.

●Positive staining for CDX-2, or the combination of CK20-positive/CK7-negative, is highly suggestive of colorectal cancer [12].

●Positive staining for CK7 and TTF-1, with negative staining for CK20, is highly suggestive of lung adenocarcinoma.

●Positive staining for CK7, gross cystic fluid protein 15, and GATA-binding protein 3 (GATA3) is highly suggestive of breast adenocarcinoma.

●Positive staining for renal cell carcinoma (RCC; also called renal cell carcinoma marker [RCC-Ma]) and paired box gene 8 (PAX8), with negative staining for CK20 is highly suggestive of RCC.

●Positive staining for CK7, Wilms tumor 1 (WT-1), and PAX8 is highly suggestive of ovarian adenocarcinoma.

●Positive staining for octamer-binding transcription factor 4 (OCT-4) and placental alkaline phosphatase is highly suggestive of germ cell carcinoma.

The pattern of staining with the cytokeratins CK20 and CK7 may be helpful in narrowing the diagnostic spectrum (table 3 and table 2). CK20 is a low molecular weight cytokeratin that is normally expressed in the lower gastrointestinal tract, urothelium, and in Merkel cells [12]. CK7 is expressed in cancers of the lung, ovary, endometrium, and breast but not in cancers of the lower gastrointestinal tract. The CK20-positive/CK7-negative combination is the most specific, particularly if the CDX-2 stain is also positive, and allows a strong prediction of colorectal cancer in patients with compatible clinical and histologic features. In a study of 93 autopsy cases of adenocarcinoma of unknown primary site involving the liver, a CK20-positive/CK7-negative pattern correctly predicted a colorectal primary in 17 of 21 cases (81 percent) [12]. Other CK20/CK7 combinations were not specific enough for confident identification of a primary site. Furthermore, these and other IHC stains are sometimes falsely positive or negative, further confounding interpretation.

None of these IHC results have been prospectively evaluated for accuracy in identifying the primary site in patients with CUP. Likewise, no studies have adequately addressed the question of whether treatment based on these IHC "diagnoses" improves the outcome for these patients. However, the good correlation between IHC and MCCA diagnoses when a single primary site is suggested by IHC supports the accuracy of these IHC diagnoses. (See 'Molecular cancer classifier assay' below.)

Molecular evaluation of the tumor biopsy

Molecular cancer classifier assay — If standard pathologic evaluation does not identify the tissue of origin, tumor tissue should be analyzed using an MCCA, also known as a gene expression profile assay. The diagnostic efficacy of MCCAs is based on distinct gene expression profiles present in different normal body tissues. When cancer arises from normal tissues, the distinct gene expression profiles are usually retained, at least in part, by the neoplastic cells, which allows identification of the primary tumor site of origin. Several MCCAs, using either reverse transcription polymerase chain reaction (RT-PCR) or gene microarray technology, are commercially available and provide results with a clinical turnaround time of approximately one to two weeks [13-15]. Based on clinical validation studies, MCCAs can correctly identify the tissue of origin (ie, primary tumor site) in 85 to 95 percent of metastatic cancers from various known primary sites [13-15], and in most CUPs [10,11,16,17].

Comprehensive molecular profiling (CMP) — Comprehensive molecular profiling (CMP) involves assaying a broad group of genes to identify potentially actionable oncogenic molecular alterations (eg, HER2, EGFR, BRAF, ALK, TRK, RET, BRCA, ROS1, FGFR, among others), as well as genomic signatures (eg, tumor mutational burden [TMB], high microsatellite instability [MSI-H], and genome-wide loss of heterozygosity). These molecular alterations can be used to predict activity (and thus select) for targeted agents and/or immunotherapy. CMP is usually performed on tissue biopsies, but blood-based liquid biopsies (circulating tumor deoxyribonucleic acid [DNA]) are also feasible [18].

Of note, CMP differs from an MCCA, which measures differential expressions of normal genes to enable the identification of a tissue of origin. Although data suggest that CMP using a next-generation sequencing panel may also be able to predict tissue of origin, further clinical validation is necessary, and available next-generation sequencing platforms do not contain this function [19].

The number of actionable molecular alterations that CMP can identify continues to increase. In addition, CMP can identify predictors of response to immunotherapy, including MSI-H and TMB. Identical molecular alterations can occur in various tumor types, although the frequency varies widely. When detected by CMP, these molecular alterations can guide effective targeted therapy or immunotherapy regardless of tumor type; this concept is also known as "tumor agnostic" therapy [20,21].

CMP identifies actionable molecular alterations in approximately 20 percent of patients with CUP [22-24]. In addition, genomic signatures that predict response to immune checkpoint inhibitors (MSI-H, high TMB, amplified or overexpressed programmed cell death ligand 1 [PD-L1]) are also identified in CUP [21,25-29]. High TMB is relatively common in CUP, occurring in 8 percent of adenocarcinomas, 11 percent of carcinomas, and 23 percent of squamous carcinomas [25]. MSI-H and PD-L1 amplification are less common (1 to 2 percent) [26]. PD-L1 overexpression is more common among CUP; staining in tumor-infiltrating lymphocytes was seen in 63 percent, and staining of cancer cells was observed in 21 percent [21]. (See "Overview of advanced unresectable and metastatic solid tumors with DNA mismatch repair deficiency or high tumor mutational burden" and "Principles of cancer immunotherapy".)

TREATMENT — 

After completion of clinical, pathologic, and molecular tumor evaluations, groups of patients who fit into favorable subsets requiring specific therapy can be defined. These patients comprise 30 to 35 percent of all patients with adenocarcinoma of unknown primary sites. The remaining 65 to 70 percent of patients with cancer of unknown primary site (CUP) do not fit into any of these specific subgroups, although most have a tissue of origin predicted by molecular cancer classifier assay (MCCA). Our approach to the management of patients with adenocarcinoma of unknown primary site by specific subgroups is presented below. Clinical trials are encouraged, where available. Patients who are unable to tolerate treatment are offered best supportive care.

Specific clinicopathologic subgroups — The diverse group of patients with adenocarcinoma of unknown primary site contains several specific clinicopathologic subgroups, which are defined by clinical and/or pathologic features for which specific therapy is available. All patients who fit into one of these subgroups should receive specific therapy.

Females with peritoneal carcinomatosis — In females, adenocarcinoma (usually with serous histologic features) causing diffuse peritoneal involvement without an obvious primary tumor usually originates in the ovary or in extraovarian tissues with similar histogenesis. Disease-directed therapy can result in a relatively favorable outlook in these patients compared with adenocarcinoma of unknown primary site of nonovarian origin. (See "First-line chemotherapy for advanced (stage III or IV) epithelial ovarian, fallopian tube, and peritoneal cancer".)

●Pathology and pathogenesis – In some cases, these tumors arise from the peritoneal surface or fallopian tubes, which share a common histogenesis with ovarian tissues. Many have morphologic features that are typical for epithelial ovarian carcinoma, such as papillary configuration or psammoma bodies. In such cases, this syndrome has various terms such as primary peritoneal carcinoma, serous carcinoma of the peritoneum, or multifocal extraovarian serous carcinoma. However, some patients may present with a poorly differentiated adenocarcinoma that does not exhibit a papillary configuration (analogous to poorly differentiated epithelial ovarian carcinomas); they should be approached similarly to those with typical serous histology. Both immunohistochemistry (IHC) and MCCAs often corroborate this diagnosis. (See "Epithelial carcinoma of the ovary, fallopian tube, and peritoneum: Histopathology", section on 'Microscopic pathology'.)

Primary peritoneal carcinoma may share a common biology with ovarian carcinoma, a concept that is supported by the following observations:

•Females at high risk for ovarian cancer may also develop primary peritoneal carcinoma. For example, these cancers occur more commonly in females with breast cancer susceptibility gene 1 (BRCA1) mutations and occasionally in those from families at high risk for ovarian cancer who have undergone prophylactic oophorectomy. (See "Risk-reducing salpingo-oophorectomy in patients at high risk of epithelial ovarian and fallopian tube cancer" and "Screening for ovarian cancer".)

•The clinical features of primary peritoneal carcinomas are often typical of advanced ovarian cancer, with tumor involvement limited to the peritoneal surfaces and elevated serum concentrations of cancer antigen (CA) 125. (See "Screening for ovarian cancer".)

●Management – Management of females with peritoneal carcinomatosis of unknown primary site may include a multimodality approach that includes surgical debulking and systemic chemotherapy:

•Surgical cytoreduction should be considered in patients with bulky disease. In patients with epithelial ovarian cancer, debulking may provide the best chance for long-term remission, although the optimal timing is controversial [30]. (See "Cancer of the ovary, fallopian tube, and peritoneum: Surgical cytoreduction".)

•Patients with peritoneal carcinomatosis of unknown primary site often respond well to chemotherapy regimens that are effective in the treatment of advanced epithelial ovarian cancer. Several studies have documented high initial response rates similar to those seen in patients with advanced ovarian carcinoma [31-35]. (See "First-line chemotherapy for advanced (stage III or IV) epithelial ovarian, fallopian tube, and peritoneal cancer".)

Females with axillary lymph node metastases — Breast cancer should be suspected in females (and, rarely, in males) who have an adenocarcinoma of unknown primary site and axillary lymphadenopathy. Multiple series have demonstrated that females presenting with axillary lymphadenopathy have a much better prognosis than those with adenocarcinoma of other unknown primary sites. (See "Axillary node metastases with occult primary breast cancer", section on 'Initial diagnostic workup' and "Axillary node metastases with occult primary breast cancer", section on 'Prognosis'.)

To support a diagnosis of breast cancer, IHC staining (for estrogen receptor, progesterone receptor, human epidermal growth factor receptor 2 [HER2], and other breast cancer-specific markers such as GATA-binding protein 3 [GATA3] and gross cystic duct fluid protein 15) should be obtained on the biopsy material in such patients. Even with normal physical examination and mammography, bilateral breast MRI is successful in identifying a primary site in the majority of such cases. (See "Axillary node metastases with occult primary breast cancer", section on 'Breast MRI'.)

If a focal breast lesion is identified, further diagnostic evaluation should follow standard guidelines for suspected breast cancer. (See "Diagnostic evaluation of suspected breast cancer" and "Clinical features, diagnosis, and staging of newly diagnosed breast cancer".)

Females with adenocarcinoma or poorly differentiated carcinoma in axillary nodes, compatible IHC staining, and no metastatic sites other than the axillary lymph nodes may have potentially curable breast cancer. These patients are treated according to guidelines for stage II breast cancer. (See "Axillary node metastases with occult primary breast cancer", section on 'Locoregional treatment' and "Overview of the treatment of newly diagnosed, invasive, non-metastatic breast cancer".)

If metastatic sites in addition to axillary lymph nodes are present, such patients may have metastatic breast cancer. These patients should receive systemic therapy according to guidelines for the treatment of metastatic breast cancer. (See "Axillary node metastases with occult primary breast cancer", section on 'Metastatic disease' and "Overview of the approach to metastatic breast cancer".)

Males with skeletal metastases and elevated prostate-specific antigen — When bone metastases are the first manifestation of metastatic adenocarcinoma, the most common primary tumor sites are the lung, prostate, and less often, liver, kidney, thyroid, and colon [22].

Metastatic prostate cancer should be suspected in males with adenocarcinoma predominantly involving bone, particularly if the metastases are osteoblastic or sclerotic. Elevated serum levels of prostate-specific antigen (PSA) or tumor staining with PSA provide confirmatory evidence of prostate cancer, and such patients should be treated using guidelines for metastatic prostate cancer. Occasional patients have a significantly elevated serum PSA or tumor staining for PSA but a clinical presentation that is atypical for prostate cancer (eg, metastases to the lung or mediastinal or upper abdominal lymph nodes, without concomitant involvement of bone or pelvic lymph nodes) [36,37]. Such patients, in the absence of data supporting another primary, should also be considered for treatment for metastatic prostate cancer. (See "Bone metastases in advanced prostate cancer: Clinical manifestations and diagnosis".)

Metastatic prostate cancer is amenable to treatment with a range of therapies that differ from those used for metastatic adenocarcinoma of other primary sites; therefore, identification of these patients is important in guiding appropriate therapy. (See "Overview of systemic treatment for recurrent or metastatic castration-sensitive prostate cancer".)

Patients with specific tumor profiles

Colon cancer profile — Systemic treatment for patients with metastatic colorectal cancer has resulted in a substantial improvement in overall survival (OS). (See "Initial systemic therapy for metastatic colorectal cancer".)

Accurate recognition of patients with adenocarcinoma of unknown primary site who are likely to respond to similar treatments is therefore increasingly important. A "colon cancer profile" has been described and includes:

●Predominant metastatic sites in the liver and/or peritoneum

●Adenocarcinoma with histology typical of gastrointestinal origin

●Typical IHC staining pattern including cytokeratin 20 (CK20)-positive/cytokeratin 7 (CK7)-negative and caudal-type homeobox transcription factor 2 (CDX-2)-positive

Patients with this profile respond well to chemotherapy with regimens developed for patients with metastatic colorectal carcinoma (eg, FOLFOX/bevacizumab) [38,39].

Increasing evidence also supports the use of site-specific treatment in CUP patients who have a colorectal tissue of origin identified by MCCA. In two retrospective series, patients with CUP who had a colorectal site of origin predicted by MCCA and received standard regimens for advanced colon cancer had median survival >20 months [40,41]. In both series, approximately 45 percent of patients predicted to have colorectal cancer by MCCA had atypical IHC staining and would not have been identified by standard pathologic evaluation. Although prospective data are needed, the survival documented in these retrospective studies, which is similar to patients with advanced colon cancer, suggests accurate identification by molecular profiling.

Tumors with a colon cancer profile should also be tested for high microsatellite instability (MSI-H) and high tumor mutational burden (TMB) to assess treatment eligibility for immune checkpoint inhibitors. The approach to such testing is discussed separately. (See "Overview of advanced unresectable and metastatic solid tumors with DNA mismatch repair deficiency or high tumor mutational burden".)

Lung adenocarcinoma profile — Lung cancer is one of the most common cancers represented among patients with adenocarcinoma of unknown primary site. In one autopsy series, lung primary sites were found in approximately 20 percent of patients with CUP [2]. In another study of 252 patients with CUP whose biopsies were tested with MCCA, 28 patients (11 percent) were predicted to have a lung primary site [42].

The recognition of patients with CUP from a lung tissue of origin is important because it influences therapy. The treatment of non-small cell lung cancer (NSCLC) has evolved with the addition of immunotherapy, molecular targeted therapies, and maintenance therapy, among other approaches. None of these specific therapies are included in the empiric chemotherapy options for CUP. (See "Overview of the initial treatment and prognosis of lung cancer".)

Patients with CUP and a lung adenocarcinoma profile often exhibit the following features:

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

●Typical IHC staining pattern (thyroid transcription factor-1 [TTF-1]-positive, CK7-positive, CK20-negative, thyroglobulin-negative

●MCCA prediction of lung tissue of origin [13,14,43,44]

In patients with lung adenocarcinoma identified by MCCA, standard first-line NSCLC chemotherapy produced median survival of 16 months [42]. However, no prospective studies have applied contemporary NSCLC treatment in its entirety to patients with CUP. While awaiting further data, patients with CUP and a lung adenocarcinoma profile can be treated using the same approach for advanced NSCLC. (See "Overview of the initial treatment and prognosis of lung cancer".)

Treatment selection can also be guided by early initial testing for programmed cell death ligand 1 (PD-L1) as well as critical driver mutations such as epidermal growth factor receptor (EGFR), the anaplastic lymphoma kinase (ALK) fusion oncogene, and other targetable mutations. (See "Overview of the initial treatment of advanced non-small cell lung cancer", section on 'Driver mutation present'.)

Renal cell carcinoma profile — Renal cell carcinoma (RCC) accounts for approximately 5 percent of CUP in autopsy series and in MCCA series [42].

Treatment options for advanced RCC include immunotherapy, vascular endothelial growth factor receptor (VEGFR) inhibitors, and mechanistic target of rapamycin (mTOR) inhibitors. None of these agents are included in the empiric chemotherapy regimens options for CUP, which have no activity in the treatment of RCC. (See "Systemic therapy for advanced and metastatic clear cell renal cell carcinoma".)

Although patients may have diverse clinical presentations, the diagnosis of advanced RCC can often be suspected based on pathologic findings, which can include:

●Clear cell or papillary histology

●IHC staining for RCC marker, CD10, and paired box gene 8 (PAX8)

●MCCA diagnosis of a kidney site of origin [44]

In a typical IHC evaluation of a CUP biopsy, kidney-specific stains are not included unless the diagnosis is suspected, so the recognition of these patients can be missed.

The efficacy of treatment for patients with CUP predicted to have advanced RCC by IHC and/or MCCA has not been extensively evaluated, but data from observational studies are encouraging [45,46]. As examples:

●In one retrospective series, 10 patients with CUP predicted to have RCC by histology and IHC were treated with targeted RCC-specific therapy [45]. None of the patients had detectable kidney lesions and over half (60 percent) were in the poor risk category. In all patients, the objective response rate (ORR) was 40 percent. Among those with intermediate-risk disease, median OS was 18.5 months.

●Site-specific RCC treatment based on MCCA prediction has also been reported. In another retrospective series, 24 patients with CUP were predicted to have RCC by MCCA; papillary histology was present in 11 patients [46]. Among the 20 patients treated with VEGFR inhibitors or mTOR inhibitors, the ORR was 19 percent and median OS was 16 months.

Although further data are necessary, particularly with immunotherapy, patients with CUP identified as having an RCC profile can be treated using the same approach for advanced or metastatic RCC. (See "Systemic therapy for advanced and metastatic clear cell renal cell carcinoma" and "The treatment of advanced and metastatic non-clear cell renal cell carcinoma".)

Thyroid carcinoma profile — Although rare, follicular/papillary thyroid carcinoma can present as a CUP. The thyroid carcinoma profile includes:

●Metastases in cervical or mediastinal lymph nodes, lungs, or bones

●Tumor IHC staining for thyroglobulin

●Elevated serum thyroglobulin levels

The treatment of patients with CUP identified as having a thyroid carcinoma profile should follow guidelines for metastatic follicular/papillary thyroid cancer. (See "Follicular thyroid cancer (including oncocytic carcinoma of the thyroid)", section on 'Treatment' and "Differentiated thyroid cancer: Overview of management".)

Adenocarcinoma of unknown primary in a single site — In occasional patients, only a single metastatic lesion is identified after a complete staging evaluation. Such single lesions have been described in a variety of sites including lymph nodes, brain, lung, adrenal gland, liver, and bone. The possibility of an unusual primary site (eg, apocrine, eccrine, or sebaceous carcinoma) mimicking a metastatic lesion should be considered but can usually be excluded based on clinical or pathologic features.

In most of these patients, other metastatic sites become evident within a relatively short time. A positron emission tomography (PET) scan may be helpful to rule out additional unrecognized sites of metastatic disease prior to definitive local therapy [47]. (See 'Diagnostic evaluation' above.)

If no evidence of additional disease is found, resection of the solitary lesion should be considered. If resection is not feasible because of the location of the metastatic lesion, definitive local radiation therapy should be administered. Local treatment sometimes results in long disease-free intervals.

The benefit of surgical resection of more than one metastatic lesion in patients with CUP is not well documented. Since this approach is recommended in highly selected patients with several tumor types (eg, kidney carcinoma, colon cancer with liver metastases, NSCLC), it may also be reasonable in occasional patients with CUP.

In some instances (eg, after resection of a solitary brain metastasis), local radiation therapy may also be appropriate to maximize the chance of local control [48,49]. (See "Overview of the treatment of brain metastases".)

The role of adjuvant chemotherapy in this setting is undefined. However, a primary site can be predicted in many of these patients using IHC and MCCA. Adjuvant therapy is reasonable to consider if indicated in the management of the tumor type predicted. When a primary site cannot be predicted, empiric chemotherapy may be useful in patients with poorly differentiated carcinoma. (See 'Empiric chemotherapy' below.)

Patients not in specific clinicopathologic subgroups — Most patients (65 to 70 percent) with adenocarcinoma of unknown primary site do not fit into any of the described clinicopathologic subgroups. Although these patients were traditionally treated with empiric chemotherapy, the approach to treatment is evolving to integrate the use of MCCAs to select site-specific therapy and comprehensive molecular profiling (CMP) to provide options for molecularly-guided therapy (MGT).

Approach to therapy — We include MCCA and CMP in the standard evaluation of all patients with CUP who do not fit into specific clinicopathologic subgroups. Treatment should be guided by the findings of both molecular tests (algorithm 1).

●Tissue of origin predicted – For patients with CUP where MCC predicts a tissue of origin (ie, primary tumor site), we offer initial site-specific therapy based on the MCCA. Site-specific therapy should follow standard guidelines for that tumor type, including chemotherapy, targeted treatment, and immunotherapy, as appropriate. (See 'Site-specific treatment based on MCCA' below.)

•CMP should be used to identify tumor-specific molecular alterations for targeted therapy that are part of standard treatment for that tumor type (eg, EGFR, ALK, ROS1, KRAS, RET, BRAF, among others).

•If CMP identifies potential molecular alterations that are not part of standard treatment for that tumor type, we administer such targeted therapy at some point during the disease course.

●No tissue of origin predicted and actional molecular targets – For those where MCCA fails to predict a tissue of origin (or is not available) and CMP identifies potentially actionable targets, we offer empiric chemotherapy and MGT based on CMP (eg, targeted agents and/or immunotherapy), as this approach improves progression-free survival (PFS). The optimal sequencing of empiric chemotherapy and MGT (ie, concurrent versus sequential use) has not been addressed in clinical trials. (See 'Molecularly-guided therapy based on CMP' below.)

●No tissue of origin predicted and no actionable molecular targets – For those where MCCA fails to predict a tissue of origin and CMP does not identify any actionable targets, we offer empiric chemotherapy alone or a clinical trial. (See 'Empiric chemotherapy' below.)

Most patients with CUP were traditionally treated with empiric chemotherapy. However, several important advances have influenced the management of patients with CUP.

●In most patients with CUP, the tissue of origin can be accurately predicted using MCCA and/or IHC testing. These tests, along with autopsy data in CUP, have confirmed that a wide variety of tumor types are represented and many patients with CUP have tumor types that respond poorly to chemotherapy.

●Many critical molecular abnormalities have been identified in the cells of different cancer types, including CUP. These abnormalities can be identified by CMP, which is widely available.

●There have been rapid advances in the use of targeted therapy and immunotherapy for cancer. CMP is now part of the diagnostic evaluation for nearly all patients with advanced cancer, and specific targeted therapy and/or immunotherapy is used to treat most cancer types.

These improvements in cancer therapy have led to an evolution in the management of CUP. Since the tissue of origin can be accurately predicted in most patients with CUP, studies are using this information to optimize chemotherapy and select relevant targeted therapy and immunotherapy to improve outcomes over empiric chemotherapy. Although these hypotheses seem intuitive, confirmatory data from clinical trials have emerged slowly for many reasons. For example, MCCAs tested in initial trials varied in efficacy and the number of different cancer types recognized. In addition, most of these initial clinical trials included limited numbers of patients. The administration of optimal standard site-specific therapy has been a rapidly moving target. Finally, some randomized trials only evaluated the use of MCCA, while others only evaluated the use of CMP. We believe it is ideal to use both MCCA and CMP to guide site-specific treatment. Studies that only evaluate MCCA cannot identify the optimal targeted treatment, and studies that only evaluate CMP cannot identify the optimal chemotherapy regimens.

Multiple clinical trials have addressed the use of site-specific treatment directed by results of IHC and MCCA [11,40-42,46,50-56], although some questions still remain. Although studies using MCCA are mixed, all studies demonstrate clinical benefits in patient subgroups predicted to have treatment-sensitive (ie, chemotherapy-sensitive) tumor types. Initial trials evaluating MCCA-directed site-specific treatment included few targeted agents and no immunotherapy, so they essentially compared site-specific chemotherapy to empiric chemotherapy. The high percentage of patients with treatment-resistant (ie, chemotherapy-resistant) tumor types has made interpretation of these trials more difficult. One randomized trial (Fudan CUP-001), which showed a PFS benefit for MCCA-directed therapy, included a higher proportion of patients with treatment-sensitive tumor types and contained treatment more consistent with contemporary standard therapy for the predicted tumor types [52]. However, since immunotherapy was not part of standard therapy at the time of the study, enrolled patients whose tumors contained molecular alterations responsive to immunotherapy did not receive such agents. (See 'Site-specific treatment based on MCCA' below.)

The management of tumor types that were previously classified as chemotherapy-resistant has evolved to include targeted agents and/or immunotherapy. Examples of such tumor types include biliary, pancreatic, gastroesophageal, endometrial, kidney, urothelial, and salivary glands. At the same time, treatments for chemotherapy-sensitive tumor types have been further improved by the addition of targeted therapy and/or immunotherapy. In most cases, improvements have been made by identifying subsets of these populations for specific additional treatment. Since all these tumor types can be identified within the CUP population (using MCCA), studies began to evaluate approaches beyond the traditional approach of "empiric chemotherapy for all."

In a subsequent randomized phase II trial (CUPISCO), treatment with empiric chemotherapy and molecularly targeted therapy (as identified by CMP) demonstrated a PFS advantage compared with empiric chemotherapy alone [18]. These results support a change in the approach to patients with CUP who do not fit into any of the prior clinicopathologic subgroups. However, we believe it is ideal to obtain both MCCA and CMP to guide precision-based, site-specific therapy. Since no MCCA was performed in the CUPISCO study, it is possible that some patients (ie, those with predicted chemotherapy-sensitive CUP) received less than optimal chemotherapy. In addition, the delay of targeted therapy for three months may not have been the optimal way to deliver this treatment. (See 'Molecularly-guided therapy based on CMP' below.)

Site-specific treatment based on MCCA — Site-specific therapy includes:

●The use of site-specific first-line and subsequent-line chemotherapy.

●Molecular testing for specific molecular alterations pertinent to the tissue of origin (eg, HER2 for breast cancer) and treatment with targeted therapy in appropriate subsets.

●Treatment with immunotherapy if indicated for the tumor type or predicted effective by CMP.

In these trials, site-specific therapy differed based on treatment availability. Many targeted agents and immunotherapy became part of standard treatment during or after the time in which these trials were conducted.

Several prospective trials and retrospective studies used site-specific therapy in CUP patients with specific sensitive tumor types predicted by MCCA (colon, kidney, poorly differentiated neoplasm) [11,39,41,46]. All these trials showed a substantially longer OS than the median 9 to 11 months reported with empiric chemotherapy. For example, a prospective phase II trial of 194 patients with CUP evaluated MCCA-directed site-specific therapy [42]. Patients mostly received site-specific chemotherapy only; a few also received EGFR or VEGF inhibitors. Median OS for the entire study population was 12.5 months, which was better than historical comparisons to empiric chemotherapy. Median OS was higher in patients with cancer types that were chemotherapy-sensitive versus chemotherapy-resistant (median OS was 13.2 months, versus 7.6 months).

Subsequent randomized trials compared MCCA-directed site-specific therapy to empiric chemotherapy [50-52]. Comparing these trials is difficult due to the different patient populations.

●In a randomized phase III trial (GEFCAPI 04), 243 treatment-naïve patients with CUP received either site-specific treatment or empiric chemotherapy (gemcitabine/cisplatin) [50]. In preliminary results, no difference in OS was seen between the two treatment arms. In the subgroup of 60 patients with cancer types predicted to be unresponsive to empiric gemcitabine plus cisplatin (eg, kidney, colorectal, liver, sarcoma, neuroendocrine, breast, melanoma, salivary gland), OS favored site-specific therapy (one-year OS 39 percent versus 30 percent; two-year OS 24 percent versus 10 percent).

●Another randomized phase II trial of 101 patients with CUP used the same trial design as GEFCAPI 04 but used a different MCCA [51]. Relative to empiric chemotherapy, site-specific chemotherapy demonstrated similar OS (median 9.8 versus 12.5 months) and PFS (median 5.1 versus 4.8 months). In this trial, only 17 of 101 patients were predicted to have chemotherapy-sensitive cancers, and 26 patients were predicted to have lymphoma. Since almost all lymphomas can be identified using standard pathologic and IHC evaluation, questions are raised about either the pathologic evaluation or the MCCA used in this study.

●In an open-label phase III trial (Fudan CUP-001), 182 patients with previously untreated metastatic CUP were randomly assigned to either site-specific therapy (as directed by a 90-gene expression assay) or empiric chemotherapy (with either taxane plus platinum or gemcitabine plus platinum) [52]. Most patients had adenocarcinoma (46 percent), followed by poorly differentiated carcinoma (31 percent), squamous cell carcinoma (22 percent), or poorly differentiated neoplasms (1 percent); neuroendocrine tumors were excluded. Compared to prior randomized trials, a much higher percentage of patients had treatment-sensitive tumor types predicted by MCCA (eg, lung [13 percent], ovary and cervix [12 percent each], and breast [10 percent]). In addition, a larger proportion of patients in the site-specific treatment group received molecularly targeted therapy and/or immunotherapy, although immunotherapy was not available for most patients with molecular alterations known to respond to immunotherapy (eg, MSI-H, high TMB tumors).

At a median follow-up of 33 months, site-specific therapy improved PFS relative to empiric chemotherapy (median PFS 9.6 versus 6.6 months, hazard ratio [HR] 0.68, 95% CI 0.49-0.93). At a median follow-up of 44 months, median OS for site-specific therapy and empiric chemotherapy were 29 and 19 months, respectively (HR 0.74, 95% CI 0.52-1.06). Grade ≥3 toxicity was similar between the treatment arms (56 versus 61 percent).

Molecularly-guided therapy based on CMP — MGT is the selection of therapy (ie, targeted agents and/or immunotherapy) based on the molecular alterations identified by comprehensive molecular profiling (CMP), in the absence of a confirmed tissue of origin.

In retrospective studies, various actionable molecular alterations have been identified using CMP on the biopsies of patients with CUP [21,23-29,38]. MGT is also effective in patients with CUP, based on data from initial case reports and observational studies [43,57-62]. In most of these studies, targeted therapy was administered as single agents after progression on first-line empiric chemotherapy.

These initial data led to the conduct of an open-label, randomized phase II trial (CUPISCO), which compared initial management with empiric chemotherapy and MGT based on CMP findings versus empiric chemotherapy alone [18]. In this study, 636 patients with newly diagnosed CUP (either adenocarcinoma or poorly differentiated carcinoma) were initially treated using empiric platinum-based regimens (paclitaxel plus carboplatin or gemcitabine plus platinum). Among the 573 patients who completed three months of induction chemotherapy, those who did not progress after induction chemotherapy (436 patients) were randomly assigned 3:1 to either MGT (as directed by CMP from either tissue or liquid samples) or continuation of the same empiric chemotherapy regimen. Those who progressed after induction chemotherapy (135 patients) were assigned to MGT. Targeted agents were available for the following targets: ALK, EGFR, HER2, BRAF, PIK3CA, BRCA, PTCH1, high TMB, and deficient mismatch repair (dMMR)/MSI-H. Patients assigned to MGT but with no targetable alterations or genomic signatures were treated with the same chemotherapy regimen plus atezolizumab, an immune checkpoint inhibitor. Of note, MCCAs were not used in this study, so the tissue of origin for these patients was unknown.

●Among the 326 patients randomly assigned to MGT, 88 patients (27 percent) had one or more actionable targets. The remaining 238 patients in this group received continued empiric chemotherapy plus atezolizumab.

●At a median follow-up of 24 months, among the 436 patients who did not progress on induction chemotherapy, MGT improved PFS relative to chemotherapy (median 6.1 versus 4.4 months, HR 0.72, 95% CI 0.56-0.92) with similar or less toxicity.

●Among the subgroup with an actionable molecular profile, MGT also improved PFS relative to chemotherapy (median eight versus five months, HR 0.65, 95% CI 0.42-0.99).

●Among the subgroup without an actionable mutation, MGT (chemotherapy plus atezolizumab) demonstrated a nonstatistically significant trend towards higher PFS than chemotherapy alone (median PFS six versus four months, HR 0.76, 95% CI 0.54-1.06).

●Although data are immature, MGT and chemotherapy demonstrated median OS of 15 and 11 months, respectively.

●Outcomes have not been reported for the 135 patients who progressed after induction chemotherapy and were treated with MGT.

However, we believe it is ideal to obtain both MCCA and CMP to guide precision-based, site-specific therapy. Since no MCCA was performed in the CUPISCO study, it is possible that some patients (ie, those with predicted chemotherapy-sensitive CUP) received less than optimal chemotherapy. In addition, the delay of targeted therapy for three months may not have been the optimal way to deliver treatment.

Empiric chemotherapy — Empiric chemotherapy was previously the standard initial treatment for patients with CUP who are not included in specific clinicopathologic subgroups. Patients with CUP traditionally received empiric chemotherapy that was designed to be effective across a broad spectrum of cancer types. When these chemotherapy regimens were evaluated, chemotherapy was the only option for the treatment of metastatic cancer. These empiric regimens were of modest benefit, mainly since many of the patients with CUP had cancer types that were poorly responsive to chemotherapy.

Although there have been no randomized studies, substantial evidence supports a modest improvement in survival with empiric chemotherapy versus supportive care only. As opposed to the registry data, patients in empiric chemotherapy trials are restricted to those with a good performance status; in this population, empiric chemotherapy produced median survivals of 7 to 10 months in most phase II studies [16,63-66].

Several two-drug combinations have similar activity when used for initial therapy. The most frequently used combinations are paclitaxel plus carboplatin, gemcitabine plus cisplatin (or carboplatin), and gemcitabine plus irinotecan [16,63-66]. These regimens demonstrate similar efficacy in clinical trials and selection of therapy is based on patient and provider preference regarding relative toxicities (eg, neuropathy and neutropenia with paclitaxel plus carboplatin; neurotoxicity, ototoxicity, kidney toxicity, and cytopenias with gemcitabine plus cisplatin; and cytopenias and diarrhea with gemcitabine plus irinotecan). These empiric chemotherapy regimens result in ORRs of 25 to 45 percent, with median OS ranging between 7 and 10 months. In a pooled analysis of five phase II trials that included 396 patients with CUP treated with empiric chemotherapy, the combined ORR was 30 percent. Median PFS was nine months, and two-year OS was 19 percent [16].

A few later-line empiric chemotherapy regimens also have modest activity [67-70]. Chemotherapy combinations with some activity (following first-line taxane plus platinum combinations) included oxaliplatin plus capecitabine (ORR 19 percent, median OS 10 months) [67] and gemcitabine plus irinotecan (ORR 10 percent, median OS 5 months) [69]. Empiric systemic regimens containing other agents also showed some activity, including bevacizumab plus erlotinib (ORR 10 percent, median OS seven months) [68] and nivolumab (ORR 22 percent, median OS 16 months) [70].

PROGNOSTIC FACTORS — 

Retrospective analyses have identified clinical and pathologic features that are associated with a favorable response to treatment using empiric chemotherapy in patients with cancer of unknown primary site (CUP) [71-77]. Many of these features are related to tumor grade or extent of disease and are prognostic factors for many types of advanced cancer. These include the following:

●Tumor location in lymph nodes or soft tissue. Patients with involvement of the liver or bones have a relatively poor prognosis.

●Fewer sites of metastatic disease.

●Female sex.

●Poorly differentiated carcinoma histology.

●Good performance status.

●Normal serum lactate dehydrogenase level.

●Normal serum albumin.

●Normal lymphocyte count.

Of note, these prognostic factors have not been studied in patients receiving site-specific therapy. It is likely that use of molecular cancer classifier assay (MCCA) and comprehensive molecular profiling (CMP) will lead to the identification of other prognostic factors.

In one prognostic factor analysis that included 150 patients with CUP who were seen at a single institution over a 10-year period, the performance status and serum lactate dehydrogenase could be used to separate patients into good- and poor-risk categories. The median survival durations for good- and poor-risk patients were 11.7 and 3.9 months, and one-year survival rates were 45 and 11 percent, respectively [76].

A multivariate analysis of prognostic factors based on a series of 317 consecutive patients found that a normal serum albumin and the absence of liver metastases identified a favorable subset of patients (median survival 371 days, versus 103 days in patients with a low serum albumin and/or liver metastases) [77]. In the same report, the favorable prognosis associated with the combination of normal serum albumin and the absence of liver metastases was validated in a second cohort of 124 patients with CUP.

SUMMARY AND RECOMMENDATIONS

●Definition and epidemiology – Cancer of unknown primary site (CUP) is diagnosed in patients with metastatic cancer but no anatomic primary site identified by a comprehensive initial evaluation. CUPs account for approximately 2 percent of all invasive cancers. Adenocarcinomas of unknown primary site comprise approximately 70 percent of CUPs. (See 'Introduction' above.)

●Initial clinical evaluation – In order to make the diagnosis of CUP, the following evaluation should be obtained, yet fail to identify the anatomic primary site (see 'Diagnostic evaluation' above):

•History and physical examination

•Complete blood count

•Urinalysis

•Basic serum chemistries

•Contrast-enhanced CT or gadolinium-enhanced MRI of the chest, abdomen, and pelvis

•In males, a prostate examination and measurement of serum prostate-specific antigen (PSA)

•In females, a pelvic examination and mammography

●Pathologic evaluation – Once the diagnosis of CUP is established, additional evaluation may be warranted based on the findings of clinical and pathologic assessment. (See 'Initial evaluation of the tumor specimen' above.)

•Histologic evaluation – Histologic characteristics of a biopsy specimen usually allow for classification of the lineage of a CUP (ie, carcinoma versus sarcoma, lymphoma, melanoma). (See 'Light microscopy' above.)

Histologic examination cannot distinguish among various adenocarcinomas, but immunohistochemistry (IHC) staining strongly suggests the primary site in approximately one-third of patients. (See 'Immunohistochemistry' above.)

•Molecular evaluation of the tumor biopsy

-Molecular cancer classifier assays – Molecular cancer classifier assays (MCCAs) accurately predict the tissue of origin (ie, primary tumor site) in 85 to 90 percent of patients with CUP patients and guides the selection of site-specific therapy. (See 'Molecular cancer classifier assay' above.)

-Comprehensive molecular profiling – Comprehensive molecular profiling (CMP) identifies patients with actionable molecular alterations and genomic signatures that predict responsiveness to treatment with targeted agents and/or immunotherapy. (See 'Comprehensive molecular profiling (CMP)' above.)

●Treatment of patients in specific clinicopathologic subgroups – Patients with adenocarcinoma of unknown primary site should be evaluated to determine whether their clinical features and pathology classify them as belonging to any of the following subgroups (table 4) for which individualized therapy is appropriate (algorithm 1) (see 'Specific clinicopathologic subgroups' above):

•Females with peritoneal carcinomatosis – Females with adenocarcinoma of unknown primary and peritoneal carcinomatosis and consistent pathology should be treated like those with ovarian carcinoma. This approach may include both surgical debulking and systemic chemotherapy. (See 'Females with peritoneal carcinomatosis' above and "First-line chemotherapy for advanced (stage III or IV) epithelial ovarian, fallopian tube, and peritoneal cancer" and "Cancer of the ovary, fallopian tube, and peritoneum: Surgical cytoreduction".)

•Females with axillary lymphadenopathy – Females presenting with adenocarcinoma of unknown primary and axillary lymphadenopathy should be treated as if they have primary breast cancer, as long as the pathology and clinical presentation are consistent with that diagnosis. (See 'Females with axillary lymph node metastases' above and "Axillary node metastases with occult primary breast cancer".)

•Males with skeletal metastases and elevated PSA – Males with metastatic adenocarcinoma and skeletal metastases and elevated serum levels of PSA and/or tumor staining with PSA should be treated for advanced prostate cancer. (See 'Males with skeletal metastases and elevated prostate-specific antigen' above and "Overview of systemic treatment for recurrent or metastatic castration-sensitive prostate cancer".)

•Colon cancer profile – Patients who present with a colon cancer profile (ie, predominant metastatic sites in the liver and/or peritoneum, an adenocarcinoma with histology typical of gastrointestinal origin, and typical IHC staining pattern [cytokeratin 20 [CK20]-positive/cytokeratin 7 [CK7]-negative, or caudal-type homeobox transcription factor 2 [CDX-2]-positive]) should be treated as if they have metastatic colorectal cancer. Treatment for colorectal cancer should also be offered for patients with a colorectal tissue of origin predicted by an MCCA. (See 'Colon cancer profile' above and "Initial systemic therapy for metastatic colorectal cancer".)

•Lung adenocarcinoma profile – Patients who present with a lung adenocarcinoma profile (ie, mediastinal and/or hilar adenopathy often accompanied by metastases at other sites; typical IHC staining pattern [thyroid transcription factor-1 [TTF-1]-positive, CK7-positive, CK20-negative]; MCCA prediction of non-small cell lung cancer [NSCLC]) should be treated using the same approach for advanced NSCLC. (See 'Lung adenocarcinoma profile' above and "Overview of the initial treatment of advanced non-small cell lung cancer".)

•Renal cell carcinoma profile – Patients who present with a renal cell carcinoma (RCC) profile (ie, clear cell or papillary histology; typical IHC staining [RCC marker, CD10, paired box gene 8 [PAX8]]; MCCA prediction of RCC) should be treated using the same approach for advanced or metastatic RCC. (See 'Renal cell carcinoma profile' above and "Systemic therapy for advanced and metastatic clear cell renal cell carcinoma" and "The treatment of advanced and metastatic non-clear cell renal cell carcinoma".)

•Thyroid carcinoma profile – Patients who present with a thyroid carcinoma profile (ie, metastases in cervical or mediastinal lymph nodes, bones, or lungs; elevated serum thyroglobulin levels, IHC staining for thyroglobulin) should be treated using the same approach for metastatic follicular/papillary thyroid carcinoma. (See 'Thyroid carcinoma profile' above and "Follicular thyroid cancer (including oncocytic carcinoma of the thyroid)", section on 'Treatment' and "Differentiated thyroid cancer: Overview of management".)

•Disease at a single site – Patients with a single metastatic focus of adenocarcinoma who do not fit any of the patterns above should be carefully evaluated to exclude any other sites of disease involvement. If no other site of disease involvement can be identified, we offer definitive local therapy, consisting of either surgical resection or radiation therapy. Although most patients will develop disseminated disease relatively rapidly, this approach is associated with prolonged survival in some cases. (See 'Adenocarcinoma of unknown primary in a single site' above.)

●Treatment of patients not in specific clinicopathologic subgroups – In patients who do not fit into a specific subset, treatment should be guided by the results of MCCA (and/or IHC stains) and CMP (algorithm 1). (See 'Patients not in specific clinicopathologic subgroups' above.)

•Tissue of origin predicted – For patients with CUP where MCC predicts a tissue of origin (ie, primary tumor site), we use initial site-specific therapy based on the MCCA. Site-specific therapy should follow standard guidelines for that tumor type including chemotherapy, targeted treatment, and immunotherapy, as appropriate. (See 'Site-specific treatment based on MCCA' above.)

-CMP should be used to identify tumor-specific molecular alterations for targeted therapy that are part of standard treatment for that tumor type (eg, EGFR, ALK, ROS1, KRAS, RET, BRAF, among others).

-If CMP identifies potential molecular alterations that are not part of standard treatment for that tumor type, we administer such targeted therapy at some point during the disease course.

•No tissue of origin predicted and actionable molecular targets – For those where MCCA fails to predict a tissue of origin (or is not available) and CMP identifies potential actionable molecular targets, we use empiric chemotherapy and molecularly-guided therapy (MGT) based on CMP (eg, targeted agents and/or immunotherapy), as this approach improved progression-free survival (PFS) in a randomized trial. The optimal sequencing of empiric chemotherapy and MGT (ie, concurrent versus sequential use) has not been addressed in clinical trials. (See 'Molecularly-guided therapy based on CMP' above.)

•No tissue of origin predicted and no actionable molecular targets – For those where MCCA fails to predict a tissue of origin, and CMP does not identify any actionable targets, we offer empiric chemotherapy alone or a clinical trial. The most frequently used regimens for empiric chemotherapy are paclitaxel plus carboplatin, gemcitabine plus cisplatin (or carboplatin), and gemcitabine plus irinotecan. These regimens are similarly effective and selection of therapy is based on patient and provider preference regarding relative toxicities. (See 'Empiric chemotherapy' above.)