Metastatic Cancer With Unknown Primary Site Workup

Updated: Apr 04, 2022
  • Author: Winston W Tan, MD, FACP; Chief Editor: Wafik S El-Deiry, MD, PhD  more...
  • Print

Approach Considerations

A minimal basic workup for cancer of unknown primary origin comprises basic blood and biochemical analyses along with computed tomography (CT) scans of the abdomen and pelvis. [11] Further investigations are dictated by any positive findings on initial evaluation. Depending on the clinical situation, these may include studies such as chest CT, breast magnetic resonance imaging, or upper or lower endoscopy. In suspected head and neck malignancies, panendoscopy of the upper aerodigestive pathways should be performed, with blind biopsies of the lymphoid tissue in these areas. Diagnostic tonsillectomy may be warranted.

Intensive testing also adds to the morbidity for the weak and frail patient. A large negative cost-to-benefit ratio exists for an extensive unguided clinical evaluation, with one study quoting a 9.5% increase in 1-year survival at a cost of $2-8 million. [12, 13, 14] When these investigations fail to reveal a potential primary lesion, a cancer of unknown primary origin is assumed. The goal of initial evaluation should be to detect the small subset of patients who warrant potentially curative management.


Laboratory Studies

Lab studies for metastatic cancer with an unknown primary site should include the following:

  • Complete blood cell count (iron deficiency anemia may point toward an occult gastrointestinal malignancy leading to chronic blood loss)
  • Urinalysis (microscopic hematuria may be a sign of occult genitourinary malignancy)
  • Liver and renal function tests
  • Stool examination for occult blood
  • Measurement of prostate-specific antigen (PSA) in men

Imaging Studies

Imaging studies should include the following:

  • Chest radiograph
  • Computed tomography (CT) of abdomen and pelvis
  • Mammography in women

Positron emission tomography with 18F-fluoro-2-deoxy-D-glucose (18F-FDG-PET) is increasingly being used in the evaluation of metastatic malignancies. [15] This may be especially the case in suspected head and neck malignancies. However, this modality lacks specificity and may be useful only to identify promising sites for biopsy. Its high cost and false-positive rate of 20% limit its utility in cases of cancer of unknown primary origin. The combination of PE with CT may reduce the false-positive rate. [16]

Depending on the clinical situation, further imaging studies may include CT of the chest or magnetic resonance imaging of the breasts.

The high level of inaccuracy of unguided radiographic studies raises the issue of cost effectiveness for intensive diagnostic workup of this disorder.

See the image below.

CT scan of neck with contrast. The arrows indicate CT scan of neck with contrast. The arrows indicate metastatic lymphadenopathy. Image courtesy of Head and Neck Cancer-Multidisciplinary Approach, Davidson, BJ.

Other Tests

The role of tumor markers, such as alpha-fetoprotein (AFP), beta–human chorionic gonadotropin (beta-HCG), cancer antigen 125 (CA125), CA 27.29, CA 19.9, and carcinoembryonic antigen (CEA), to establish a specific primary site or to identify patients who respond to chemotherapy remains unclear and should probably be limited to cases in which a particular primary site is favored. Most tumor markers are nonspecific and cannot be used to establish definitive diagnoses. [17]

In a large prospective trial, molecular tumor profiling predicted a tissue of origin in the majority of patients with cancer of unknown primary (CUP). Patients who received assay-directed site-specific therapy had a median survival time of 12.5 months, which compares favorably with survival among patients treated with empiric CUP regimens. CUP patients who were predicted to have more responsive tumor types had longer survival than those with less responsive tumor types (14 vs 7 months). [18]

For those tumors in which there is difficulty in making a diagnosis, several commercially available tests for genomic abnormalities and potential help with the diagnosis have been made available. These are often expensive, but at times are very helpful in identifying the primary site of the tumor. These tests include Pathwork, Cancer type Id, and micro-R.

Genomic testing is becoming less expensive, and multiple tests have been claimed to be very effective in finding the primary site of the tumor. Additionally, test results are accompanied by listings of the therapeutic drugs available for the cancer and the therapies that are in clinical trials. All this information must be taken into consideration by the clinician, and the case should be managed on the basis of a thorough synthesis of the data provided and the clinical relevance of the results.

At this time, the National Comprehensive Cancer Network (NCCN) does not recommend gene signature profiling for tissue of origin for standard management of CUP. The NCCN advises that although these tests may offer diagnostic benefit, they do not necessarily provide clinical benefit. [19]

A 2013 technology assessment by the Agency for Healthcare Research and Quality (AHRQ) on commercially available genetic tests reported that the accuracy of the tests is 85-88%, but their accuracy in CUP cases is not easily determined, because in most cases the actual tissue of origin is not identified. The assessment concluded that the available evidence was insufficient to determine whether these tests affect clinician decisions or alter patient survival, and recommended that future studies addressing the clinical value of these tests be carried out by groups who have no evident conflict of interest. [20]

In a study by Hainsworth et al, a 92-gene reverse transcriptase polymerase chain reaction (RT-PCR) cancer classification assay predicted the tissue of origin in 247 of 252 assays (98%). Of the 289 patients enrolled in the study, 12.5% had insufficient tissue for the assay and 10% of patients were not treatment candidates by the time the results became available. [21]

A randomized clinical trial by Hayashi et al compared site-specific therapy directed by gene expression profiling with empirical chemotherapy (carboplatin and paclitaxel) in 101 previously untreated patients with CUP. One-year survival was not significantly different with site-specific treatment than with empirical chemotherapy (44.0% versus 54.9%, respectively ( P = 0.264). However, prediction of the original site seemed to have prognostic value; survival was slightly better for tumor types predicted to be more responsive. [22]



Often the diagnostic procedure of choice is a biopsy of the metastatic site and a careful evaluation of the pathology, which leads the clinician to a cost-effective diagnostic evaluation of the disease.

Depending on the clinical situation, upper or lower endoscopy may be warranted. In suspected head and neck malignancies, panendoscopy of the upper aerodigestive pathways should be performed, with blind biopsies of the lymphoid tissue in these areas, as indicated. Diagnostic tonsillectomy may be warranted.


Histologic Findings

The pathologist has an indispensable role in the evaluation of cancer of unknown primary origin. The help of a pathologist familiar with cancer of unknown primary origin is essential. Tumors provided for pathological review should come from tissue that has whenever possible been excised if such tissue is available and accessible. Needle biopsy specimens may provide insufficient tissue for diagnosis or provide tissue that has been too damaged or distorted by the biopsy procedure for accurate diagnosis. [23]

Various immunoperoxidase stains are available for providing a differential diagnosis for cancer of unknown primary origin. An experienced and knowledgeable pathologist will be familiar with appropriate DNA microarray techniques and proteonomic studies to establish a definitive diagnosis. The pathologist typically puts the tissue specimen through 1-4 steps, depending on the need. These studies include light microscopy, immunohistochemical stains, [24] electron microscopy, and chromosomal analysis including cytogenetics. [25] Deatils are as follows:

  • Light microscopy: After initial light microscopic evaluation, most tumors are classifiable as epithelial cancers, lymphomas, sarcomas, melanomas, or germ cell tumors. When cytologic distinguishing features are limited, the tumor may be classified as undifferentiated or poorly differentiated carcinoma.

  • Immunohistochemistry: These tests help define tumor lineage by using peroxidase-labeled antibody against specific tumor antigens. These include stains for keratin, leukocyte common antigen and S-100 (expressed in melanomas), thyroid transcription factor  thyroid transcription factor–1 (TTF-1; for lung and thyroid cancer), prostate-specific antigen (PSA), human chorionic gonadotropin (hCG; for germ cell tumors), alpha-fetoprotein (AFP; for germ-cell tumors and hepatomas); and estrogen receptors, progesterone receptors, and human epidermal growth factor receptor 2 (HER2) for breast cancer.

  • Electron microscopy: This study has limited utility in identification of the primary site of cancer of unknown primary origin but may rarely be used in poorly differentiated tumors.

  • Chromosomal studies: In cases of cancer of unknown primary origin with suspected occult nasopharyngeal carcinoma, DNA amplification of Epstein-Barr virus (EBV) in tissue may clinch the diagnosis. [26] The presence of iso-chromosome 12p, i(12p), a specific chromosomal marker characteristic of germ cell tumors, can help diagnose extragonadal germ cell tumors in patients with cancer of unknown primary origin. [27, 28]

The majority of cancers of unknown primary origin are adenocarcinomas or undifferentiated tumors (up to 58% in some studies). Less commonly, squamous cell carcinoma, melanoma, sarcoma, and neuroendocrine tumors can also present as metastasis with an unknown primary site of origin. [29] Most studies exclude sarcomas and melanomas from their analysis.

In the approximately 30% of cancers of unknown primary origin in which a full workup establishes a clear pathological diagnosis, the most common epithelial malignancies are lung (15%), pancreas (13%), colon/rectum (6%), kidney (5%), and breast (4%). Sarcomas, melanomas, and lymphomas each contribute 6-8%. The remaining primary tumors are those of stomach (4%), ovary (3%), liver (3%), esophagus (3%), prostate (2%), and a variety of other malignancies (22%).



Patients with cancer of unknown primary origin are presumed to all have stage IV disease at the time of initial presentation.

See Cancer of Unknown Primary Staging for more information.


Immunohistochemistry-Based Classification

Guidelines from the European Society for Medical Oncology (ESMO) recommend selected marker assays as part of the basic immunohistochemical workup of cancers of unknown primary. [11] See the table below.

Table. Immunohistochemical markers for cancers of unknown primary (Open Table in a new window)

Primary Marker Primary Tumor Type Additional Markers
CK7- /CK 20 + Colorectal cancer and merkel cell carcinoma CEA and CDX-2  (for GI malignancy)
CK 7 +/CK 20 - Lung,breast, thyroid, endometrial, cervical, pancreas, and cholangiocarcinoma

TTF-1 (lung, thyroid)

ER, PR (breast)

GCDFP-15 (gynecologic)

CK- 19 (pancreas)

Ck+7/ CK 20+ Urothelial, ovarian, pancreas, cholangiocarcinoma

Urothelin (genitourinary) 

WT-1 (ovarian, mesothelial)

CK = cytokeratin; CEA = carcinoembryonic antigen; TTF1 = thyroid transcription factor 1; ER = estrogen receptor; PR = progesterone receptor; GCDFP-15  = gross cystic disease fluid protein–15; WT-1 = Wilms tumor gene 1; PSA = prostate specific antigen

It is important to note that immunostains are not diagnostic but only supportive of the diagnosis. In some cases, these markers can yield misleading information. In this situation, molecular testing might at times clarify the situation.

New IHC markers can provide a more accurate diagnosis; for example, CDH17 may be a more sensitive marker for gastric cancer than CK20 and CDX2. [30] Nonetheless, due to tumor heterogeneity, both false positive and false negative IHC staining patterns can be found; for example, TTF1 or CDX2 is absent in a minority of lung and colon cancers. [31]