Cancer of unknown primary site accounts for 3% to 5% of all cancer diagnoses. This syndrome occurs with equal frequency in men and women and increases in incidence with advancing age. Numerous clinical presentations and histologic tumor characteristics are represented in this heterogeneous group of patients. Common sites of metastases include the liver, bone, lung, lymph nodes, pleura, and brain. The majority of patients have metastases present in two or more sites. Patients are diagnosed with cancer of unknown primary site after an initial standard evaluation fails to identify a primary tumor site.
Historically, patients with an unknown primary tumor have responded poorly to empiric therapy, and median survival from diagnosis has usually been less than 8 months. The limited activity of early chemotherapeutic regimens and short median survival have contributed to clinical nihilism about patients with cancer of unknown primary site.
A number of patient subsets with specific treatment implications have now been described, however. These subsets, identified by specific clinical or pathologic criteria, account for approximately 40% of all patients with unknown primary tumors. In the remaining patients, recent studies of empiric taxane-containing therapies have produced higher response rates, and have probably extended survival. Ongoing evaluations of other new regimens and approaches to the treatment of cancer of unknown primary may result in additional treatment advances.
Approximately 60% of patients with cancer of unknown primary site have well-differentiated or moderately well-differentiated adenocarcinoma. Poorly differentiated carcinomas or poorly differentiated adenocarcinomas are diagnosed in approximately 30% of patients. Squamous cell carcinomas and poorly differentiated neoplasms account for 5% of diagnoses. Clinical presentation, evaluation, prognosis, and treatment options differ for these four histologically distinct groups. The presence of specific clinical features or the use of specialized pathologic studies can identify subsets within these broad categories of patients who may benefit from tumor-specific therapy.
In most patients, the clinical evaluation, including the search for the primary site, should be brief and focused. A careful history should be taken and a complete physical examination should be performed (including pelvic and rectal examinations), with routine laboratory evaluation (hematologic and chemistry profiles), chest radiography, and computed tomography (CT) of the abdomen and pelvis. Computed tomographic scans of the abdomen may identify a primary site in 10% to 35% of patients; CT scans may also locate additional sites of metastatic disease.[3,4] Tumors identified by CT scans of the abdomen and pelvis include those arising from the pancreas, kidney, hepatobiliary tract, and ovary.
Additional procedures (eg, radiologic imaging and/or endoscopic examinations) are warranted to evaluate specific presenting signs or symptoms. Positron emission tomography (PET), performed with fluorine-18-deoxyglucose, is a noninvasive technique that has identified primary sites in a few patients, although its role in the evaluation of patients with cancer of unknown primary site is incompletely defined.[5,6]
Certain additional diagnostic evaluations should be performed in specific patient subsets. For example, mammography should be performed in women with clinical features suggestive of metastatic breast cancer. All men with adenocarcinoma of unknown primary site should have their serum prostate-specific antigen (PSA) levels measured, as well as PSA staining of the biopsy specimen. Serum levels of human chorionic gonadotropin (HCG) and alpha-fetoprotein (AFP) should be measured in young men with poorly differentiated carcinoma.
The diagnostic yield of other additional diagnostic procedures is low[7,8] (Table 1) and can add considerable expense to the initial evaluation. Extensive radiologic or endoscopic investigations of asymptomatic areas are rarely useful in identifying a primary site.
Measurement of various commonly used serum tumor markers (eg, carcinoembryonic antigen [CEA], cancer antigen 15-3 [CA 15-3], cancer antigen 125 [CA-125], and cancer antigen 19-9 [CA 19-9]) also is not useful in providing clues to the location of the primary site. All of these markers are frequently elevated in patients with carcinoma of unknown primary site, and thus, provide no diagnostic or prognostic information. Modest elevations of HCG or AFP (ie, levels of either marker < 100) are also of no diagnostic or prognostic utility in this patient population. If levels of any of these markers are elevated, serial measurement may be useful in assessing response to treatment.
An accurate pathologic evaluation of tumor tissue is an essential part of the diagnostic process. A biopsy of the most accessible lesion should be performed. A fine-needle biopsy is least invasive but may not yield enough tissue for specialized pathologic studies, which are required for all patients with poorly differentiated tumors. Close collaboration with the pathologist is therefore necessary to ensure that optimal diagnostic studies are performed. The pathologist should be informed about all of the clinical information to assist in the selection of specialized pathologic studies.
Light microscopic examination results in the identification of several subgroups, including poorly differentiated neoplasms, well-differentiated and moderately well-differentiated adenocarcinomas, squamous cell carcinomas, and poorly differentiated carcinomas (with or without features of adenocarcinomas). Histochemical staining for mucin may facilitate identification of some adenocarcinomas.
In patients with well-differentiated or moderately differentiated adenocarcinoma, further pathologic studies rarely identify a primary site and are only indicated in selected patient subgroups. Staining for PSA occasionally identifies a prostate primary site in men with metastatic adenocarcinoma, and positive staining for estrogen/progesterone receptors is useful in women with clinical features compatible with metastatic breast cancer.
As opposed to their limited utility in patients with adenocarcinoma, additional pathologic studies frequently add useful information to the evaluation of poorly differentiated tumors. Immunocytochemical stains are now widely available and should be routinely performed. Although definitive diagnoses can be made only in a minority of patients, a number of diagnoses can be suggested with this technique, including lymphoma, melanoma, sarcoma, and neuroendocrine carcinoma (Table 2). Several of these diagnoses have specific therapeutic implications.
A number of tumor-specific antibodies or antibody panels have been evaluated. However, none of these is sufficiently specific to reliably identify a primary site.[12-15]
Electron microscopy is also a valuable tool in the evaluation of poorly differentiated tumors. Since it is less widely available and often requires rebiopsy with special tissue preparation, electron microscopy should be reserved for use in tumors of uncertain lineage following immunoperoxidase studies. Definitive diagnoses of lymphoma, neuroendocrine tumors, melanoma, and a variety of sarcomas is possible with electron microscopy.
When a primary site is definitively identified by either clinical or pathologic evaluation, treatment should proceed according to the guidelines for that type of cancer. In the large majority of patients, however, no primary site can be identified. In these patients, it is probably best to avoid the temptation to guess at a primary site based on either suggestive clinical or pathologic features.
For example, patients with mucin-positive adenocarcinoma involving the liver may have an unidentified gastrointestinal primary, and yet empiric treatment with gastrointestinal regimens (eg, fluorouracil(Drug information on fluorouracil) [5-FU]/leucovorin)
has shown very little benefit in this group. Therefore, empiric treatment for carcinoma of unknown primary site should be employed (see below), rather than relatively ineffective treatment for a presumed insensitive primary site.
While, as mentioned above, it is not particularly helpful to label patients with a presumed primary site, it is important to identify patients who fall into certain treatable subsets, based on either clinical or pathologic features (see below). Specific treatment guidelines are available for these subsets. In the remainder of patients, empiric treatment designed for cancer of unknown primary should be employed.