Cancer of Unknown Primary Site

What every physician needs to know:

Cancer of Unknown Primary (CUP) accounts for 3-5% of all malignant tumors. The CUP syndrome is defined by presence of regional and/or metastatic disease in a patient without a discernible primary site after a thorough and focused search for a primary tumor.

Appropriate management depends on the specific clinical presentation and an expert pathologic review including immunohistochemistry. Although molecular profiling is not recommended for standard management, its role in specific CUP syndromes continues to evolve.

The majority of patients with CUP present with adenocarcinomas (60%).

Continue Reading

Other histologies include squamous carcinomas (5%), poorly differentiated carcinomas (5%), and undifferentiated malignant neoplasms (30%). Although usually distinguishable on histology and immunophenotyping, rarely melanomas, sarcomas, and lymphomas may present as unknown primary tumors and are generally treated in the context of their specific diagnosis.

Are you sure your patient has cancer of unknown primary site? What should you expect to find?

The confirmatory diagnosis of CUP rests on the exclusion of a primary tumor after a systematic appraisal of clinical presentation and an adequate but appropriate workup including imaging and expert pathologic review.

Which individuals are most at risk for developing cancer of unknown primary site:

Each presentation of CUP must be viewed in terms of the most likely source of the primary tumor. Thus patients at increased risk for a given tumor are also at increased risk for CUP.

For example, CUP presenting with axillary adenopathy or abdominal carcinomatosis in women with known or family history of hereditary breast and ovarian cancer syndrome (BRCA-associated) indicates a provisional breast or ovarian primary, respectively. Similarly, prior comorbidities and surgeries can provide clues to occult primaries, such as increased risk of cholangiocarcinoma in patients with primary sclerosing cholangitis. Hence, the importance of an exhaustive family, occupational, and social history cannot be overemphasized in the evaluation of unknown primary tumors.

What laboratory and imaging studies should you order to characterize this patient's tumor (i.e., stage, grade, CT/MRI vs PET/CT, cellular and molecular markers, immunophenotyping, etc.) How should you interpret the results and use them to establish prognosis and plan initial therapy?

All patients should have a thorough clinical examination and appropriate laboratory and imaging studies to delineate the extent of metastatic disease, and to identify a possible primary tumor. In most patients, CT scans of the chest, abdomen and pelvis or a PET-CT scan will suffice. More specific imaging studies and (invasive) procedures will depend on the specific clinical presentation, immunophenotypic profile, and sometimes molecular analyses specifically directed at establishing a putative primary source.

Tumor markers, while helpful in following response to therapy (if elevated) are usually not very helpful in establishing a primary tumor of origin. Adenocarcinoma tumor markers are commonly considered to be disease specific (e.g. CA125 for ovarian cancer and CA19.9 for cancer of the pancreas); however in CUP patients there is broad overlap and often multiple tumor markers are positive. Even high levels of a given marker should not be considered diagnostic of a specific primary.

Both the prognosis and treatment of CUP patients are determined by the clinical presentation (local/regional versus widespread), histology, and the identification of a putative primary tumor.

In our clinic, all patients have imaging studies and pathology studies as noted. If histopathology and immunohistochemistry cannot determine a specific primary or provide a short differential diagnosis, molecular profiling assays can be considered in selected patients, especially where the assay results may help select a specific therapy.

Specific imaging studies and procedures based on presentation.

Patients presenting with squamous carcinoma in neck nodes require endoscopic evaluation of the head and neck and may also require head and neck CT scanning. These patients are good candidates for a PET-CT scan.

Women with axillary adenopathy (adenocarcinoma) or with immunohistochemical and/or molecular studies suggesting breast cancer should be evaluated with mammography, breast ultrasound and, if negative, breast MRI imaging.

Patients presenting with symptoms (e.g. altered bowel movements or dysphagia) or signs (microcytic anemia) indicating a gastrointestinal source should have upper GI endoscopy and colonoscopy. Immunohistochemistry and molecular studies may point to a diagnosis that makes these tests unnecessary (for example, ovarian markers and molecular identification in women with carcinomatosis). In the absence of specific symptoms and signs, the yield of endoscopies is low.

Young men (under 50) with midline adenopathy should have germ cell tumor markers performed in addition to a testicular ultrasound.

Women presenting with disease in the pelvis or inguinal nodes should be evaluated by a gynecologist.

Patients with immunohistochemical or molecular studies suggesting a particular unexpected primary tumor should have the appropriate specialized imaging, particularly if the diagnosis would suggest a specific therapy (eg, thyroid imaging in a CUP patient with a Tissue of Origin test suggesting thyroid cancer).

Immunohistochemical studies in cancer of unknown primary site patients

High quality immunohistochemistry (IHC) is critical to understanding and managing CUP patients. The presence of adequate tissue is a critical component of an effective pathological review. A panel of focused IHC markers is better at delineating a primary than any one single marker. Additionally, the sensitivity and specificity of many IHC markers is limited and should not be interpreted without the clinical context. It can be conceived of in 3 levels:

– Establishing tumor lineage

Both histopathology and IHC studies can help define the tumor lineage. This is the step where patients with melanomas, sarcomas and lymphomas may be identified and separated from carcinomas. In a small number of cases, especially very poorly differentiated tumors, this classification is difficult and the pathologist will be unable to report a specific lineage. Common IHC markers used in this step include:

  • Carcinoma: Pankeratin, Cam 5.2, Epithelial membrane antigen (EMA)

  • Lymphoma: Leukocyte Common Antigen or CD45

  • Melanoma: S100, HMB-45, Vimentin

  • Sarcoma: Vimentin, Desmin

(Basic) – Characterization of Adenocarcinomas/Carcinomas using cytokeratins (CK20, CK7)

The cytokeratins CK-7 and CK-20 are used in the evaluation of CUP samples, and allow the division of tumors into groups based on their expression pattern. Each of the 4 possible combinations of these two markers is helpful in narrowing the differential diagnosis:

  • CK20+/CK7-: Colorectal, Appendiceal, Merkel cell, Ovary (mucinous)

  • CK20-/CK7+: Lung, Breast, Thyroid, Uterus, Cervix, Pancreatobiliary, Salivary, Ovary (serous), Mesothelioma

  • CK20+/CK7+: Urothelial, Ovary (mucinous), Pancreatobiliary

  • CK20-/CK7-: Hepatocellular, Renal, Squamous cell (including head/neck), Small cell lung, Gastroesophageal

– (Specific) Site specificdiagnosticimmunhistochemical studies

Some specific IHC studies or combinations suggest a specific primary tumor, or small differential diagnosis. These tests can be quite helpful in assigning a putative primary tumor. Examples include:

  • Lung: TTF1, Napsin-A

  • Thyroid: TTF1, Thyroglobulin

  • Breast: ER. PR, GCFDP, Mammoglobin, HER2, GATA-3

  • Prostate: PSA, PAP, Prostein

  • Ovarian: WT1,ER/PR, CA-125, Mesothelin

  • Hepatoma: HepPar 1, Arginase-1, Canalicular pCEA, CD10, CD13

  • GI Tumors: CDX2, CEA

  • Urolitheal: URO 111, Thrombomodulin

  • Neuroendocrine: Chromgranin, Synaptophysin, CD56

  • Mesothelioma: Calretinin, Mesothelin, MOC-31 (negative)

When immunohistochemistry is diagnostic of a specific primary, further molecular studies may be indicated and are referred to as the third tier in the CUP algorithm. For example, in patients identified with having breast cancer, Her-2Neu testing is indicated. In patients with a colon cancer profile, KRASmutation evaluation is indicated. In patients with a lung profile EGFR mutation and ALK gene studies may be helpful in selecting therapy.

Molecular studies to determine the tissue of origin in CUP Patients

All cancers are known to have a unique gene expression profile. Exploiting site-specific differences in these profiles, molecular tests to identify a putative primary tumor in patients presenting with CUP have been developed.

These ToO (tissue-of-origin) tests are based on expression profiles of mRNA or microRNA and although the tests differ in terms of the number/types of genes studied, the tumors identified from the results of studies are similar. When used to identify the site of origin in patients with known primaries, they are correct in about 90% of cases. When used in patients with CUP, they can identify a putative primary in about 80% of the cases successfully processed. However, since the defining parameter of CUP is the absence of primary, direct validation of these ToO tests is impossible.

Estimates of ToO test accuracy have thus far depended on indirect validations using concordance with clinicopathological evaluation as the standard metric. Only one prospective outcomes based study has been performed and shown that assay-directed site specific therapy compares favorably with prior studies using empiric therapy.

In clinical studies, because of difficulty processing small samples, the tests return an answer in about 70% of cases. Recent studies suggest that the results of molecular studies correlate well with tumors in which a primary is eventually identified (latent primaries) and with IHC results.

In our clinic all patients have appropriate IHC tests performed. If the IHC is considered diagnostic (as outlined above), they are treated accordingly. If IHC is not considered diagnostic, the tumor may, in specific settings be sent for molecular analysis to determine a tissue of origin.

What therapies should you initiate immediately i.e., emergently?

Emergency measures for patients with CUP are no different than those indicated for other patients with cancer and similar presentations. The suspicion of an unknown primary should not change the emergency management of the patient’s condition.

What should the initial definitive therapy for the cancer be?

In general, management of patients with CUP should be patterned on the treatment of the most likely primary source. Patients presenting with local/regional disease should be managed with appropriate multimodality therapy using chemotherapy and/or radiation and/or surgery to maximize local control. Patients with widespread disease should receive systemic chemotherapy with palliative intent.

A number of clinical subsets with “favorable” prognosis have been defined. They are all situations in which a likely primary source has been identified on the basis of the clinical presentation, histologic review or immunohistochemical characterization. Recent data suggests that patients so identified by molecular studies may also benefit from specific therapies. Examples of favorable subsets and their management are as follows:

Patients with squamous carcinoma in neck nodes (above the supraclavicular region)

These patients should be evaluated and treated as if they have primary head and neck cancers. Evaluation should include endoscopy, CT of the head and neck and PET scanning. They should be managed with combined modality treatments and have a similar prognosis to those patients presenting with a known head and neck primary of similar stage.

Women with axillary adenocarcinoma

These patients should be thoroughly investigated for breast cancer and are generally treated as if they have breast cancer. In some cases, if immunohistochemical studies suggest that the tumor is clearly not breast cancer, other regimens may be considered. It is important in treating these patients to consider both chemotherapy and local control of the ipsilateral breast, usually done with radiation therapy.

Young men with midline adenopathy

A subset of patients with germ cell tumors present with midline adenopathy (para aortic, mediastinal) without an obvious testicular tumor. They may have elevated tumor markers (hCG, AFP) compatible with the diagnosis. Men up to the age of approximately 50 years with this presentation should undergo testicular ultrasound. With the advent of sophisticated IHC, such a presentation is unlikely. If tumor markers are elevated, they should receive germ cell regimens. If tumor markers are negative, they are often treated with platinum-containing regimens.

Women with abdominal carcinomatosis, Mullerian type (serous papillary)

There is a well-defined group of women who present with all of the hallmarks of advanced ovarian cancer, but have normal ovaries or have previously had their ovaries removed. It has been suggested that these patients have formed tumors in remnants of Mullerian tissue. In the literature they are referred to as Mullerian tumors, or Primary Peritoneal Serous Carcinomas.

The diagnosis is usually based on immunohistochemistry (WT1, PAX-8) and a high serum CA125 (not mandatory). These patients must be differentiated from women who have carcinomatosis from other primary sources. Women with primary peritoneal carcinoma should be managed as if they had ovarian cancer both in terms of chemotherapy and surgery. They follow the natural history of ovarian cancer.

Patients with neuroendocrine histology

Patients with unknown primary neuroendocrine tumors need to be managed as other neuroendocrine patients are managed. If they have high-grade tumors (high KI 67, or high mitotic index) they often receive therapy containing platinum and etoposide or platinum and irinotecan. If they have low grade tumors they can be managed in a more expectant fashion and when they require treatment receive either chemotherapy or molecularly targeted agents that have recently been shown to have activity in patients with known neuroendocrine primary tumors, such as everolimus or sunitinib.

Patients with a colon cancer profile unknown primary tumor

The immunohistochemical signature of colon cancer (CK20 +, CK7 -, CDX2 +) can be found in some patients with unknown primary tumors. They respond well to colon cancer regimens (FOLFOX, FOLFIRI, bevacizumab and anti-EGFR antibodies) and should be managed as if they have metastatic colon cancer. Their survival appears to be similar to that of patients with advanced colon cancer. This also appears to be true for patients whose molecular studies identify colon as the primary profile.

Patients with solitary site of disease

Patients who present with local disease can have long survival after local therapy (surgery or radiation therapy) with or without chemotherapy. Our approach to these patients is to give them chemotherapy. If they respond with no evidence of disease dissemination, we consolidate with radiation therapy and/or surgery. We select chemotherapy based on pathology and immunohistochemistry, though most patients with this presentation remain good candidates for taxane and platinum agents in the absence of diagnostic immunohistochemistry. CUP presenting with isolated neck and inguinal lymphadenopathy are representative of this type of disease and should be managed with multimodality therapy similar to head and neck squamous cell carcinomas and cervical/anal squamous cell carcinomas.

Patients with disseminated disease not in a favorable subset

Patients with disseminated disease should have immunohistochemical analyses to attempt to identify a primary tumor. If these studies fail to suggest a specific primary, they should have molecular analyses performed to identify a putative primary tumor. If the studies suggest a specific primary they should be treated for that cancer. If the studies are not diagnostic and the clinical presentation is not helpful in assigning a primary, they should receive one of the commonly used regimens for unknown primary tumors. Commonly used regimens include:

  • Gemcitabine 1250mg/m² on days 1 and 8 and Cisplatin 100mg/m² on day 1 every 3 weeks.

  • An alternative regimen often used is Gemcitabine 1000/m² on days 1 and 8 combined with carboplatin AUC 5 on day 1, repeated every 3 weeks.

  • Paclitaxel 175mg/m² and Carboplatin AUC 5-6 every 3 weeks.

  • FOLFOX – Oxaliplatin 85mg/m² and Leucovorin calcium 200mg/m² day 1, followed by 5FU 400mg/m² bolus and 2400mg/m² over 46 hrs every 2 weeks.

A number of other regimens have been described as active in this group of patients and may be considered (see NCCN guidelines for occult primary), these include:

  • Docetaxel 75mg/m² day 1 and Gemcitabine 1000mg/m² days 1 and 8, repeated every 21 days.

  • Oxaliplatin 85mg/m² with Gemcitabine 1000mg/m² every 2 weeks.

Although Paclitaxel, Carboplatin, and Etoposide is a widely used regimen, randomized studies have shown that it is not superior to the two-drug combinations, and has more toxicity.

The selection of a specific chemotherapy regimen ultimately depends on clinicopathological parameters and the patient’s performance status. Appropriate efforts to identify a putative primary or at the least a narrow differential must be made. Tailoring therapy to suspected primaries may improve patient outcomes; however, the success of this approach over empiric therapy requires testing in prospective trials. There is no data for second-line therapy in CUP and further treatment depends on suspected primary, patient’s performance status, re-evaluation of pathology, and response to first-line therapy.

What should you tell the patient and the family about prognosis?

The literature would suggest that the prognosis of patients with CUP is quite poor (median survival of about 11 months). However, when interpreting this data one must remember that these studies and analyses exclude patients with favorable presentations and patients with regional disease. They also pre-date the advent of molecular studies which may suggest a specific primary tumor. The Culine prognostic model is validated in CUP and indicates that elevated lactate dehydrogenase, liver metastases, and poor performance status are associated with poor prognosis.

In patients with favorable subset presentation or in whom an IHC or molecular signature for a primary source is identified, the prognosis appears to be similar to that of patients with known primary of similar stage. In patients with advanced disease the overall prognosis remains poor.

"What if " scenarios.

A few presentations deserve special comment:

Patients presenting with single skin lesions may be difficult to classify. The tumors may be metastatic from distant sites, or primary eccrine (sweat gland) tumors. Eccrine tumors may be positive for Estrogen Receptor, and other breast cancer markers making pathology-based distinctions challenging. Women with this presentation should have careful breast evaluation, including breast MRI. Clinical presentation holds the key to making an appropriate diagnosis.

Single lesions felt to be eccrine should be managed with surgery, and then with close follow-up. If they recur locally they can be treated with repeat surgery or radiation therapy. Similarly, there is a particular skin related primary tumor of the finger referred to as a Digital Papillary Adenocarcinoma. It presents as a small nodule, often on the third finger. These patients should be treated with amputation of the finger, which is usually curative.

A subset of patients presenting with regional nodes containing squamous carcinoma have metastases from occult skin cancers. These patients are often fair skinned and have had heavy sun exposure. They usually have had multiple skin cancers in the past, although they may not have an obvious primary skin cancer at the time of presentation. The natural history of this presentation is for the tumor to spread contiguously and regionally (usually up the extremity) before disseminating.

Occasionally, patients present with mixed pathology findings such as adenosquamous histology, poorly differentiated carcinomas with neuroendocrine features, or pathology compatible with both carcinoma and sarcoma. The latter are often reported as “sarcomatoid carcinoma.” They are generally treated with chemotherapy regimens that overlap the two entities. Our choice in these patients is usually Gemcitabine and Docetaxel.

Follow-up surveillance and therapy/management of recurrences.

There is no data guiding surveillance in CUP patients. Our empiric approach is as follows:

  • 1. Patients who achieve a complete remission are followed with imaging studies and tumor markers (if elevated markers at presentation) every 3 months for one year, then every 4 months for one year and then every 6 months for 3 years.

  • 2. Patients on active therapy are followed with imaging studies every 10-12 weeks (sooner if clinical deterioration). Decision to continue, change or hold chemotherapy is based on performance status, toxicity and tolerance, and radiographic response.


The pathophysiological basis of unknown primary tumors is still unclear. A number of diverse hypothesis have been proposed to explain the occurrence of this clinical entity. One such hypothesis regards CUP as an early metastatic presentation of primaries with dominant metastatic phenotype. In such cases it is plausible that the metastatic lesions appeared before the primary tumor was large enough to be seen on imaging studies. In a small percentage of cases, a primary tumor becomes apparent during or after therapy; these cases are referred to as “latent primaries.”

Another related controversy has to do with whether the prognosis of CUP patients is related to the prognosis of the primary tumor, or to a genetic profile common to CUP. As IHC and molecular studies have improved our ability to identify a primary tumor it appears that even in the CUP setting most tumors retain the prognosis and responsiveness of the putative primary. Investigators have searched for molecular abnormalities unique to the CUP presentation, but to date these studies have failed to identify a common CUP signature.

What’s the evidence?

“Occult Primary Cancer: NCCN Clinical Practice Guidelines 2014”. (The National Cancer Center Network has compiled a set of evidence and consensus based guidelines for the evaluation and management of patients presenting with unknown primary tumors. The guidelines are frequently revised and should be considered standard of care.)

Pavilidis, N, Pentheroudakis, G. “Cancer of unknown primary site”. Lancet. vol. 379. 2012. pp. 1428-35. (This review provides a broad and up-to-date overview of unknown primary tumors.)

Varadhachary, GR. “Carcinoma of Unknown Primary: Focused Evaluation”. J Natl Compr Canc Netw. vol. 9. 2011. pp. 1406-1412. (This review provides in-depth coverage of evidence-based focused evaluation of carcinoma of unknown primary.)

Oien, KA. “Pathological evaluation of unknown primary cancer”. Semin Oncol. vol. 36. 2009 Feb. pp. 8-37. (This review details the comprehensive pathologic work-up required for cancer of unknown primary.)

Varadhachary, GR. “New strategies for carcinoma of unknown primary: the role of tissue-of-origin molecular profiling”. Clin Cancer Res. vol. 19. 2013 Aug 1. pp. 4027-33. (This review describes the evolution and the current utility of molecular profiling in unknown primary tumors.)