BY Gauri R. Varadhachary, MD
Professor, GI Medical Oncology
University of Texas MD Anderson Cancer Center
Background and Definition
Carcinoma of unknown primary (CUP) is a diverse group of cancers and ranks within the top 10 cancer presentations. Broadly, CUP is defined as a biopsy-proven malignancy for which the anatomic origin remains unidentified (or undetermined) after adequate history and physical examination, laboratory studies, imaging, and pathologic evaluation.
To further define the patient population, presentations that have histology and stage-specific therapy even in the absence of primary (melanoma, sarcoma, or lymphoma) are excluded as CUP cancers. The CUP clinical and research efforts concentrate on the vast majority of patients with common epithelial histologies such as adenocarcinoma, carcinoma, squamous carcinoma, and neuroendocrine carcinoma.
The inability to identify a primary carcinoma tends to generate anxiety for the patient, who may feel that the evaluation is incomplete or that the prognosis and therapy would be improved if only the primary site could be established. The management of CUP is evolving and the recently developed sophisticated diagnostic modalities and novel therapies (for known cancers) present both an opportunity and a challenge. In the current economic environment, it is important that we have a clear grasp on how these tests (and treatments) may impact the comprehensive management of our patients.
Initial Clinic Visit
In my practice, almost all patients with CUP get a baseline iodinated contrast CT of the chest, abdomen, and pelvis. I do not routinely repeat a mammogram if it was done within two years (exception being isolated axillary nodal presentation and pathology clearly suggestive of breast cancer profile). In selected women with isolated axillary adenopathy, nonsquamous pathology and a negative mammography and ultrasound, bilateral breast MRI is a sensitive modality and is indicated.
18F-fluorodeoxyglucose positron emission tomography (FDG-PET) scan is commonly overused as part of the diagnostic work up of CUP. I believe that a PET-CT rarely adds to a good quality IV contrast CT scan as the initial diagnostic study. There are some exceptions where a PET-CT may be recommended -- mainly the diagnostic work of patients with squamous cell cancer and cervical adenopathy and a select patient with solitary metastatic disease (prior to definitive locoregional therapy).
One other setting where I prefer to use PET-CT (baseline and follow up) is for patients with predominant osseous metastases on active therapy; here a single study (PET) may replace CT + MRI (bones) or bone scan and may suffice as the (cost-effective) predominant imaging study to evaluate the extent of disease and therapy response.
I preserve the use of invasive studies to patients with symptoms or clear radiographic and pathologic indications (e.g., use of colonoscopy in a patient with CK20+, CDX-2 + adenocarcinoma).
I cannot overemphasize the need for adequate tissue (core biopsy) and communication with the pathologist especially for perplexing diagnosis. On light microscopy, most CUP cancers are identified as adenocarcinoma (60%); poorly differentiated adenocarcinoma, carcinoma, or undifferentiated neoplasm (30%); squamous cell carcinoma (5%) or neuroendocrine cancer (4%).
On occasion, CUPs present as mixed tumors, such as adenosquamous carcinoma, adenocarcinoma with neuroendocrine features, and sarcomatoid carcinoma. Immunohistochemical markers (IHC) form the backbone of pathologic diagnosis, and this is based on the premise that there is concordance in the expression profiles of primary and metastatic cancers.
They perform the best when used in groups that give rise to patterns that are strongly indicative of certain profiles and can assist with treatment planning. TTF-1, CK7+, Ck20+, CDX-2+, ER, Mammoglobin, WT-1, and synaptophysin are markers that I commonly request in my practice based on clues from presentation, radiographic pattern of spread and laboratory studies.
Additionally, molecular markers with a therapeutic intent are considered for select presentations (e.g., use of KRAS, HER2, ALK mutations). IHC is not without its limitations and it is crucial that one understands there may be variability related to processing, interpretation, and just tissue heterogeneity.
About 25 percent of CUP patients in my practice have “diagnostic” IHC patterns – i.e. adequate data to suggest a unique cancer profile (lung, breast, colorectal, neuroendocrine, ovarian, hepatocellular). In most patients, the IHC is not diagnostic.
Tissue of Origin Profiling:
Tissue of origin (ToO) profiling of tumors is a promising technique to improve the site of origin diagnosis in CUP patients. Molecular profiling (MP) methods using various platforms including DNA microarray and RT-PCR and have been used to evaluate the ToO in metastatic samples. The data on known metastases has been validated using independent blinded sets of tumor samples, where the reference diagnosis is known, with an accuracy of 80 to 90 percent. Given the feasibility of utilizing mRNA/microRNA from small biopsy samples available as formalin-fixed and paraffin embedded (FFPE) tissue makes this a practical approach for clinic use.
In my practice I order a ToO test for select good performance status patients with non-diagnostic IHC (in the presence of adequate tissue) with therapy-related questions. In few patients with a diagnostic IHC but a very atypical presentation, the ToO profiling study can help minimize ambiguity related to pathology. The field continues to evolve and ongoing studies will help us better understand which patients are best suited for additional tests beyond basic pathology.
Treatment Paradigm and Future Direction
The treatment of CUP cancers has changed over the last decade. We have noted a shift from anatomically defined favorable subsets to pathologically defined favorable subsets. The era of empiric treatment trials is over and we rarely use broad-based empiric therapies for all-comers CUP. Radiology, pattern of spread and pathology (IHC, molecular profiling when indicated) help guide first and subsequent lines of therapy.
Nevertheless, taxanes, platinum agents, gemcitabine and fluoropyrimidines are commonly used first-line cytotoxics. For example, when I see a CUP patient with a CDX-2 (+) isolated carcinomatosis, FOLFOX is a commonly used first-line regimen.
We now recognize that cholangiocarcinoma (localized and distant) is often mislabeled as a CUP cancer (~15% in practice): imaging, pattern of spread (carcinomatosis, lung, and bone), IHC patterns and serum markers (Ca19-9) help select these patients and they are good candidates for first-line gemcitabine and cisplatin.
I believe that at this time, for therapy, cellular context (tissue of origin) matters, since we have much to learn about actionable targets in known and CUP cancers. As novel therapies are developed for known cancers, it will likely impact appropriate CUP subtypes.
Over the next several years, we also need to focus our efforts in refining CUP subsets (e.g., isolated nodal, osseous, carcinomatosis presentations, etc.) and leverage selective genomics and proteomics techniques to eventually deliver validated new therapeutic approaches to our patients.