Patients who are classified within several recognized favorable CUP subsets (about 20 % of all CUP) are treated with specific regimens based upon their presumed tissues of origin (see Chap. 10). In general, these treatments are site specific, based on the probable site of tumor origin in these patient subsets. Examples include treatment of women with adenocarcinoma involving axillary lymph nodes as breast cancer, and treatment of patients with squamous carcinoma in cervical lymph nodes as head/neck cancer. Such treatments have resulted in improved survival in these patient subsets.

The majority of the remaining CUP patients (approximately 80 % of all CUP) have received empiric chemotherapy regimens; prognosis for most of these patients has remained poor. Empiric chemotherapy regimens (i.e., paclitaxel/carboplatin, gemcitabine/cisplatin, others) were developed because the specific cancer type or tissue of origin was not known and for most patients could not be determined. Empiric chemotherapy became the standard of care in the mid-1990s after clinical trials showed benefit at least for some patients [34]. At that time, CUP was considered by many as a single entity with a similar clinical biology, rather than a heterogeneous group comprising many types of cancer with varying behavior. Large trials (100 or more patients) of empiric chemotherapy for the unfavorable group of CUP patients have invariably resulted in median survivals of about 9 months and 2-year survivals of 10–20 % [34]. Those patients with benefit likely harbored more sensitive cancer types (such as breast or ovarian cancer), while treatment was ineffective for those with unresponsive cancers types (such as biliary tract, pancreas).

A few clinical trials several years [1, 2] ago incorporated a few targeted type drugs (first-generation drugs) alone (bevacizumab, erlotinib) or combined with chemotherapy (paclitaxel/carboplatin) but without any knowledge at that time of specific targets within the cancer cells. These endeavors, although producing some tumor responses, were essentially empirical approaches based on activity and benefit seen in a few known advanced solid tumors (non-small cell lung, colorectal).

Recent improvements in therapy have resulted in improved survival for patients with many types of known advanced cancers. These therapies have also become more individualized or specific for each particular cancer type. Furthermore, genetic alterations are now recognized in many cancers and provide the basis for effective precision or targeted therapies for selected patients. Therefore, the recognition or diagnosis of the specific cancer type or tissue of origin in each CUP patient may have critical therapeutic implications. The optimal treatment today for patients with advanced cancers requires the identification of the site of origin or specific type of cancer.

The ability to diagnose the tissue of origin in the majority of CUP patients has inaugurated an era of site-specific therapies rather than the continued use of empiric chemotherapy. Three critical questions needed to be addressed and resolved in order to confidently change the standard of care. First, can molecular cancer classifier assays accurately provide a specific tissue of origin diagnosis? Second, can molecular assays complement standard pathology? Finally, do patients have a better survival if they are treated with site-specific therapies based on a tissue of origin diagnosis versus empiric chemotherapy?

As previously reviewed in this chapter, the accuracy of molecular cancer classifier assays and their ability to complement standard pathology are firmly established. Molecular cancer classifier assays, when used with IHC and the clinical features, facilitate a specific/single tissue of origin diagnosis in 95 % of patients, and about 80 % of these diagnoses are accurate [1–3, 17]. The molecular assays complement IHC staining and frequently make a single diagnosis when IHC is inconclusive [3, 19–21, 25, 26, 31].

Conclusive demonstration of the superiority of site-specific versus empiric therapy in CUP patients is difficult for a number of reasons. There are at least 30 different types/subtypes of metastatic cancers within the CUP population. About half of these are relatively resistant to therapy (e.g., pancreas, biliary tract, liver, others) and can be expected to do poorly regardless of the treatment administered. Many of the CUP patients with more responsive cancers (e.g., breast, ovary, lung) would derive some benefit from empiric taxane/platinum regimens (although without the benefit of site-specific targeted agents and sequential useful therapies), while in only a small percentage (e.g., colorectal, renal) would site-specific and empiric CUP therapy diverge markedly. Given these considerations, a randomized trial comparing empiric chemotherapy versus assay-directed site-specific therapy, as advocated by some, would require a minimum of several hundred patients. Fewer patients would be required if enrollment was limited to potentially responsive subtypes, but enrolling these patients may pose an ethical dilemma for some physicians.