Handbook of Cancer Chemotherapy (Lippincott Williams & Wilkins Handbook Series), 8th Ed.

24. Metastatic Cancer of Unknown Origin

James M. Leonardo

The primary site remains unknown in about 5% to 10% of patients with newly diagnosed cancer (excluding nonmelanoma skin cancer), despite a detailed pre treatment evaluation. Even after extensive evaluation and postmortem examination, a primary tumor is not found in up to 30% of these patients. However, the frequency of cancers with truly occult primary sites is decreasing, in part because of advances in technology to detect the primary site(s). The problem of metastatic cancer of unknown origin raises difficult questions for both diagnosis and treatment Although the median survival time of patients with cancer of unknown origin has been reported to be 6 to 9 months, subgroups of patients have been defined who have a more favorable outlook with aggressive management. With current therapy modalities, the overall survival of these patients appears to be improving A major responsibility of the clinician is to identify those patients with a characteristic presentation who might benefit from a specific strategy and to identify the increasingly large group of patients who might benefit from a trial of chemotherapy.

Tumors thought to be more amenable to treatment, and thus have a more favorable prognosis, include poorly differentiated cancers with midline distribution, squamous cell carcinoma (SCC) involving cervical lymph nodes, papillary adenocarcinoma of the peritoneal cavity in women, and adenocarcinoma involving only axillary lymph nodes in women. Conversely, adverse prognostic findings include adenocarcinoma metastatic to the liver or other organs, nonpapillary malignant ascites, and multiple metastases to brain, bones, or lungs.


A. Histology and presenting clinical manifestations

Moderately differentiated adenocarcinoma and poorly differentiated carcinoma or adenocarcinoma respectively comprise up to 60% and 30% of all cancers of unknown origin. SCC and poorly differentiated cancers other than adenocarcinoma each account for about 5% of unknown primary tumors. Other histologies that may present as cancer of unknown origin include lymphomas, germ cell tumors, and neuroendocrine carcinomas. These histologies are particularly important to identify because they represent tumors that may be effectively managed with systemic chemotherapy. Nearly half of all patients with unknown primaries and well over half of those with adenocarcinoma present with hepatomegaly, abdominal mass, or other abdominal symptoms. Lymphadenopathy is the presenting clinical manifestation in 15% to 25% of patients. Lower cervical or supraclavicular lymph nodes usually contain adenocarcinoma or undifferentiated carcinoma, and middle to high cervical adenopathy generally represents SCC. Between 10% and 20% of patients present with manifestations of bone, lung, or pleural involvement, whereas fewer than 10% present with evidence of central nervous system disease. Most of the latter group is eventually found to have either lung or gastrointestinal tract primaries.

A few presentations of advanced carcinoma of unknown primary site have been recognized as being more treatable and thus having a more favorable prognosis than others. Poorly differentiated carcinoma or adenocarcinoma, especially with predominant sites of involvement in the mediastinum, retroperitoneum, lymph nodes, or lungs, and adenocarcinoma in women predominantly involving the peritoneal surfaces may respond to platinum-based chemotherapy regimens designed for germ cell or ovarian cancers and result in occasional long-term disease-free survival. Women with axillary adenopathy and adenocarcinoma or undifferentiated carcinoma may occasionally enjoy sustained responses to therapy directed against breast cancer.

B. Sites of origin

It is sometimes possible to predict the most likely primary sites from the histology and location of the metastatic lesion of unknown origin. Pancreas and lung are the most common ultimately determined sites of origin. Together, they represent more than 40% of the adenocarcinomas of unknown origin in which the site can be ultimately determined. Colorectal, gastric, and hepatobiliary carcinoma each represent about another 10%.

In general, adenocarcinomas or undifferentiated carcinomas presenting with hepatic metastases or left supraclavicular adenopathy are eventually demonstrated to be of gastrointestinal origin. SCCs that present in the supraclavicular or low cervical lymph nodes are usually from lung primaries, whereas similar lesions of higher cervical nodes are more likely to have originated from occult primary lesions in the head and neck region.

The pattern of metastatic involvement associated with occult primary tumors may differ from that associated with overt primaries. For example, occult lung cancer rarely involves bone, a common site of metastasis from overt lung cancer; however, bone metastases appear to be more common in patients with gastrointestinal cancer who have occult primaries than in those who have overt primaries. Nonetheless, occult primaries can metastasize to any site and, in general, one should not rely on the pattern of metastatic spread to predict the site of origin.

C. Aims of diagnostic evaluation

The first objective in the management of a patient newly diagnosed with cancer of unknown origin is to plan the appropriate diagnostic evaluation. There are three chief aims of this evaluation:

1. Identify a tumor in which cure or effective disease control is possible.

2. Determine if the tumor is regionally confined or widely metastatic.

3. Identify any complication for which immediate local therapy is indicated.

D. Goals of treatment

In patients with tumors for which effective systemic therapy is available and in patients with disease regionally confined to peripheral lymph nodes alone, the primary goal of treatment is prolongation of life through extended disease control; in some cases, cure should be considered. These patients represent approximately 25% of patients with occult primaries. For the remaining patients, the chance of prolonging life has been less likely, but with the advent of new cytotoxic agents and targeted drugs, prospects for this group are improving. Treatment should also address palliation of symptoms and preservation of the best possible quality of life.


A. Initial workup

The initial evaluation of a patient presenting with a metastatic tumor should include a complete history and physical examination including a breast, pelvic, and rectal examination in women and prostate and testicular examination in men. Routine laboratory studies (complete blood count, electrolytes, creatinine, blood urea nitrogen, and calcium, as well as liver function testing) should also be done early on in the evaluation. The clinical scenario should determine whether more specialized laboratory testing is appropriate. It is wholly reasonable to measure cancer antigen (CA)-125 in the setting of a woman with ascites, for example. Imaging studies should include a chest radiograph at minimum; additional studies such as mammography may be useful depending on the location of the metastases and symptoms. Computed tomography (CT) scanning is now being routinely used to evaluate patients with occult primary tumors and may be responsible for the decreased frequency of cancers that remain of truly unknown origin. The role of [18F]fluorodeoxyglucose positron emission tomography (PET)-CT scanning is not clear, but some series have reported a usefulness of this modality for identifying the primary site, particularly when the presentation is in cervical lymph nodes. Although not recommended as part of the routine evaluation of all patients with occult primary cancers, PET-CT scans may be particularly useful with occult primary tumors with a single site of metastasis when therapy with a curative intent is planned. More invasive testing, such as bronchoscopy, endoscopy, or colonoscopy, should be guided by the patient's symptoms.

B. Analysis of the biopsy specimen

If possible, the pathologist should receive fresh, unfixed material to allow electron microscopy, histochemistry, immunohistology, and hormone receptor studies to be done, if needed, after routine examination. Careful review of the biopsy material should be undertaken to attempt to classify the tumor conclusively as SCC, adenocarcinoma, or other identifiable histology. Up to 40% of cancers of unknown origin are undifferentiated or poorly differentiated tumors based on evaluation of hematoxylin and eosin–stained material. Electron microscopy, when available, may be useful for the further classification of these tumors through the identification of desmosomes and intercellular bridges (SCC); tight junctions, microvilli, and acinar spaces (adenocarcinoma); premelanosomes (amelanotic melanoma); neurosecretory granules (small-cell or neuroendocrine carcinoma); and absence ofjunctions (lymphoma). Immunohistochemistry is an indispensable part of the evaluation of carcinoma of unknown primary site. The expression of cytokeratins, particularly CK7 and CK20, may be useful in narrowing down the origin of a tumor. Immunohistochemical studies on the tumor may be used to demonstrate the presence of prostatic acid phosphatase or prostate-specific antigen (PSA; prostate carcinoma), human chorionic gonadotropin (β-hCG; germ cell tumors), α-fetoprotein (germ cell tumors or hepatocellular carcinoma), or monoclonal immunoglobulin (lymphoma, plasmacytoma). Immunoglobulin or T-cell receptor gene rearrangements may be helpful in identifying tumors of lymphoid origin. Undifferentiated carcinomas or adenocarcinomas in women should be evaluated for estrogen and progesterone receptors. Mucin positivity is helpful in eliminating the possibility of renal cell carcinoma.

Molecular profiling of tumors has emerged as a powerful new tool to aid in determining the site of origin. This methodology relies on the observation that different tumor types exhibit unique qualitative and quantitative patterns of expression of genes, which can be measured using DNA microarray analysis. These analyses can be done on formalin-fixed, paraffin-embedded tissue or fresh frozen specimens. In one large, retrospective study with 547 specimens, the tissue of origin was correctly identified in 88% of the specimens. Although this promising technology is now widely available, prospective trials are needed to determine if these data can determine optimum treatment for cancers with occult primaries and prolong survival.

Clearly, the use of many of these specialized studies must be balanced against their expense. If judiciously applied, they can aid in the identification of some of the undifferentiated or poorly differentiated tumors of unknown origin and help to focus their subsequent diagnostic evaluation and management.

C. Squamous cell carcinoma (SCC)

One exception to the policy of seeking a definitive histologic diagnosis as the first step in evaluating a tumor of unknown origin is when the patient presents with a potentially resectable neck mass (other than supraclavicular adenopathy) and no other apparent lesion. In these patients, a head and neck primary should be sought by detailed head and neck examination, radiographs of the sinuses, and, if necessary, panendoscopy under general anesthesia, including laryngoscopy, bronchoscopy, esophagoscopy, and nasopharyngoscopy with blind biopsy of the base of the tongue, piriform sinuses, nasopharynx, and tonsillar fossae if no gross primary is found. A CT scan of the head and neck may also be of value. PET-CT scanning has also been utilized in this setting. If this workup is not diagnostic, biopsy of the neck mass is undertaken. This order of evaluation is chosen so that if a resectable SCC of the head and neck is found, the neck mass can be removed as part of the curative procedure.

For SCCs with apparent involvement of only one lymph node group, the possibility of long-term survival exists ifproper treatment is carried out. The diagnostic evaluation depends on the lymph node region involved. The most common lymph node presentation for SCCs of unknown origin is in the cervical or supraclavicular region. Cervical lymph node metastases above the supraclavicular region usually originate from head and neck primary lesions. The diagnostic approach to these lesions is discussed in the preceding paragraph. Because surgery, irradiation, or both, with curative intent, are employed if disease is localized to this region, distant metastases should be excluded with a bone scan, a chest radiograph, and, in some instances, a chest CT scan. SCC of supraclavicular lymph nodes is usually of lung or esophageal origin and seldom represents regionally confined disease. Evaluation is the same as that for disease that extends beyond regional lymph nodes.

SCC in axillary or inguinal lymph nodes is rarely associated with an occult primary. Regional skin and lung should be examined as possible primary sites with axillary disease, whereas the skin, anus, and genitalia should be carefully examined when the presentation is SCC in the inguinal nodes.

SCC with generalized lymphadenopathy or, more commonly, with disease that extends beyond the lymph nodes, represents disease that cannot be satisfactorily controlled by present day techniques. The search for the primary lesions should be done mainly by a chest radiograph and careful physical examination of the appropriate organs. Serum chemistries, including the calcium level, should be determined. Further diagnostic studies are needed only if indicated by signs, symptoms, or abnormalities on the initial studies.

D. Adenocarcinoma and poorly differentiated carcinoma

Women with adenocarcinoma or poorly differentiated carcinoma of unknown origin should undergo mammography, careful pelvic examination, and hormone receptor evaluation of the tumor. In men, serum acid phosphatase, PSA, β-hCG, and α-fetoprotein should be determined to help exclude prostate and germ cell tumors, respectively. An elevated CA-27-29 or CA-15-3 level would point toward a breast primary. All patients should have stools and urine examined for occult blood, and the serum should be tested for abnormalities in the liver chemistries, creatinine, and electrolytes. With disease apparently confined to axillary lymph nodes, mammography is particularly important in women and should be considered in men as well. If there is a strong suspicion of breast cancer and the mammogram is negative, magnetic resonance imaging should be considered. Undifferentiated carcinoma found only in middle to high cervical lymph nodes should be evaluated in the same manner as described in Section II.C for cervical node SCC.

Traditional contrast studies such as intravenous pyelogram, barium enema, and upper gastrointestinal series are not indicated unless specifically suggested by signs or symptoms (e.g., occult blood in the stool). Abdominal CT scan with intravenous contrast medium is a reasonable option in view of the frequency with which it detects carcinoma of the pancreas or hepatobiliary cancer in this setting.

E. Malignant melanoma

The finding of malignant melanoma confined to a single lymph node group and without a detectable primary lesion represents stage II disease and is associated with a 30% 5-year survival rate after lymphadenectomy. Evaluation to exclude more extensive disease should include a history, physical examination (emphasizing skin and ophthalmoscopic examination), chest radiograph, liver chemistries, liver scan, and brain CT scan.


A. General strategy

The importance of identifying tumors that may be treated effectively, such as lymphomas, germ cell tumors, trophoblastic tumors, and breast, prostate, ovarian, and neuroendocrine carcinomas, is readily apparent. Once identified, these lesions should be treated as described in their respective chapters. In patients whose primary lesion remains obscure, a therapeutic distinction must be made between those with disease confined to one lymph node region and those with more widespread disease or involvement of visceral organs. In the former, some may be treated with curative intent, whereas in the latter, the aims of treatment are palliative.

B. Squamous cell carcinoma (SCC)

Patients with SCC confined to the cervical lymph nodes above the supraclavicular region should receive full-course radiotherapy to a field extending from the base of the skull to the clavicles. Alternatively, they may be treated with radical lymph node dissection followed by radiation therapy. In either case, the irradiation is designed to include any possible head and neck primary carcinoma. Survival of patients so treated is at least as good as that for patients with known head and neck primaries. More limited lymph node dissection or regional irradiation may also be indicated for SCC confined to unilateral involvement of the axillary or inguinal nodes.

More widespread SCCs of unknown origin are treated with a palliative intent. No treatment except for local radiotherapy to symptomatic lesions is the standard approach. In patients with symptomatic or progressive disease who desire chemotherapy, regimens designed mainly for head and neck (see Chapter 5) or non-small-cell lung cancer (see Chapter 6) should be considered.

C. Adenocarcinoma and poorly differentiated carcinoma

In women, if these carcinomas are confined to the unilateral axillary lymph nodes, they should be considered possible breast cancer and treated accordingly as stage II disease (see Chapter 9). A woman with adenocarcinoma or poorly differentiated carcinoma predominantly confined to the peritoneal surface should be considered for a platinum-based ovarian cancer regimen. Undifferentiated carcinoma confined to the middle or high cervical lymph nodes should be treated actively as SCC (see Section II.C). Men with adenocarcinoma of unknown primary and a positive tumor or serum PSA should have a trial of hormonal therapy.

Patients with more advanced adenocarcinoma or poorly differentiated carcinoma in which the evaluation previously described in Section II. does not suggest breast, prostate, or other highly treatable primary or a primary site should be managed according to the histology. Platinum-based combination chemotherapy may be valuable in the treatment of poorly differentiated carcinoma and poorly differentiated adenocarcinoma. Newer regimens containing paclitaxel, docetaxel, or gemcitabine have shown efficacy in phase II studies of patients with occult primary tumors. In patients with mediastinal or retroperitoneal tumors, the germ cell regimen of bleomycin, etoposide, and cisplatin (BEP) or etoposide and cisplatin (EP) has been studied in a series of 220 patients. This combination may produce more than a 60% objective response rate and more than a 20% complete response rate with up to a 13% long-term survival rate.

bull BEP

bull Cisplatin 20 mg/m2 intravenously (IV) over 30 minutes on days 1 to 5, and

bull Etoposide 100 mg/m2 IV on days 1 to 5, and

bull Bleomycin 30 U IV push weekly on days 1, 8, and 15

bull Repeat cycle every 21 days regardless of blood cell counts for two (adjuvant therapy), three (good-risk patients), or four (intermediate or poor risk) cycles.

Patients with widespread adenocarcinoma that is well or moderately well differentiated may be responsive to systemic therapy. Combinations such as paclitaxel, carboplatin, and oral etoposide (PCE) or gemcitabine and carboplatin followed by paclitaxel yield objective responses in 25% to 45% of patients.

bull PCE

bull Paclitaxel 200 mg/m2 as a 1-hour IV infusion on day 1, and

bull Carboplatin area under the curve (AUC) 6 IV over 30 to 60 minutes on day 1, and

bull Etoposide 50 to 100 mg by mouth on alternate days between days 1 and 10.

bull Gemcitabine and carboplatin

bull Carboplatin AUC of 5 on day 1

bull Gemcitabine 1000 mg/m2 IV on days 1 and 8.

bull Repeat each cycle every 3 weeks.

Both regimens are worthy of consideration in patients with good performance status because occasional durable responses have occurred. Reports on both of these regimens are based on limited accrual phase II studies. Responding patients show improvement within two cycles, and chemotherapy should be stopped after two cycles if no improvement is seen. PCE is also effective in patients with poorly differentiated adenocarcinoma and poorly differentiated carcinoma and may be considered as an alternative to BEP.

D. Malignant melanoma

For disease confined to a single lymph node group, radical lymph node dissection yields long-term survival in 30% of treated patients. Treatment of disseminated melanoma is discussed in Chapter 14.

E. Neuroendocrine carcinoma

Poorly differentiated neuroendocrine carcinoma may represent up to 13% of cases of poorly differentiated carcinoma or adenocarcinoma. The diagnosis is secured by recognition of neurosecretory granules on electron microscopy. Localized lesions are uncommon and should be treated with surgery or radiation therapy. Metastatic disease frequently responds to platinum-based chemotherapy such as EP or newer regimens, such as irinotecan and cisplatin.

bull EP

bull Etoposide 120 mg/m2 IV on days 1 to 3 or 120 mg/m2 by mouth twice a day on days 1 to 3, and

bull Cisplatin 60 mg/m2 IV on day 1.


Advances in our understanding of the molecular events leading to carcinogenesis and metastasis have led to the development of gene profiling for specific types of cancer. Retrospective studies of archival breast tumor specimens, for example, in which genetic profiles are identified and compared to the known treatment outcomes, have led to prospective studies in which gene profiles are being used to guide specific therapy. Similar retrospective analyses have also been done with many other tumor types and have led to the development of DNA microarray assays which are now widely available. It is likely that genetic profiling will be used routinely as an essential step in the characterization of most tumors. This approach should identify the tissue of origin of most tumors, leading to a further decline in the apparent incidence of true occult primary tumors. More importantly, cost-effective, targeted therapies that act at defined points in the cancer cell could then be chosen based on the molecular characteristics of the tumor rather than the purported tissue or organ of origin.


The author is indebted to Dr. Martin Oken who contributed to this chapter in previous editions and whose work has been included in this revision of the handbook.

Selected Readings

Abbruzzese JL, Abbruzzese MC, Hess KR, et al. Unknown primary carcinoma: natural history and prognostic factors in 657 consecutive patients. J Clin Oncol. 1994;12:1272–1284.

Altman E, Cadman E. An analysis of 1539 patients with cancer of unknown primary site. Cancer. 1986;57:120–124.

Ettinger DS, Aguinik M, Cates JMM, et al. Occult primary (cancer of unknown primary [CUP]). NCCN Practice Guidelines in Oncology Version 2.2011; NCCN.org.

Greco FA, Burris HA III, Litchy S, et al. Gemcitabine, carboplatin, and paclitaxel for patients with carcinoma of unknown primary site: a Minnie Pearl Cancer Center Research Network Study. J Clin Oncol. 2002;20:1651–1656.

Greco FA, Vaughn WK, Hainsworth JD. Advanced poorly differentiated carcinoma of unknown primary site: recognition of a treatable syndrome. Ann Intern Med. 1986;104:547–556.

Hainsworth JD, ErlandJB, Kalman LA, et al. Carcinoma of unknown primary site: treatment with one-hour paclitaxel, carboplatin, and extended schedule etoposide. J Clin Oncol. 1997;15:2385–2393.

Hainsworth JD, Greco FA. Treatment of patients with cancer of an unknown primary site. N Engl J Med. 1993;329:257–263.

Hainsworth JD, Johnson DH, Greco FA. Cisplatin-based combination chemotherapy in the treatment of poorly differentiated carcinoma and poorly differentiated adenocarcinoma of unknown primary site: results of a 12-year experience. J Clin Oncol. 1992;10:912–922.

Hainsworth JD, Johnson DH, Greco FA. Poorly differentiated neuroendocrine carcinoma of unknown primary site: a newly recognized clinicopathologic entity. Ann Intern Med. 1988;109:364–371.

Hainsworth JD, Spigel DR, Raefsky EL, et al. Combination chemotherapy with gemcitabine and irinotecan in patients with previously treated carcinoma of an unknown primary site: a Minnie Pearl Cancer Research Network Phase II trial. Cancer. 2005;104(9):1992–1997.

Kolesnikov-Gauthier H, Levy E, Merlet P, et al. FDG PET in patients with cancer of an unknown primary. NuclMed Commun. 2005;26(12):1059–1066.

Lenzi R, Hess KR, Abbruzzese MC. Poorly differentiated carcinoma and poorly differentiated adenocarcinoma of unknown origin: favorable subsets of patients with unknown-primary carcinoma. J Clin Oncol. 1997;15:2056–2066.

Moertel CG. Adenocarcinoma of unknown origin. Ann Intern Med. 1979;91:646–647.

Monzon, FA, Lyons-Weiler M, Buturovic LJ, et al. Multicenter validation of a 1,550-gene expression profile for identification of tumor tissue of origin. j Clin Oncol. 2009;27:2503–2508.

Stokkel MP, Terhoard CH, Hordij KFJ, et al. The detection of unknown primary tumors in patients with cervical metastases by dual-head positron emission tomography. Oral Oncol. 1999;35:390–394.

Strnad CM, Grosh WW, Baxter J, et al. Peritoneal carcinomatosis of unknown primary site in women: a distinctive subset of adenocarcinoma. Ann Intern Med. 1989;11:213–217.