Customize your JAMA Network experience by selecting one or more topics from the list below.
In this issue of JAMA Oncology, Ross and colleagues1 present results from a retrospective study of 200 consecutive carcinoma of unknown primary site (CUP) tumor specimens that underwent comprehensive genomic profiling (CGP) using the hybrid-capture–based FoundationOne assay. The DNA extracted from these CUP tumor specimens was analyzed after hybridization capture of 3769 exons from 236 cancer-related genes and 47 introns of 19 genes commonly rearranged in cancer. There were 125 adenocarcinomas of unknown primary site (ACUPs) and 75 nonadenocarcinomas (non-ACUPs). The authors reported that a large number of CUP samples (85%) harbored at least 1 clinically relevant genomic alteration (GA) with the potential to influence and personalize therapy. The mean number of GAs was 4.2 per tumor. The ACUP tumors were more frequently driven by GAs in the receptor tyrosine kinase (RTK)/Ras/mitogen-activated protein kinase (MAPK) signaling pathway than non-ACUP tumors. The authors concluded that CGP can identify novel treatment paradigms and suggested that early testing may have utility in CUP management. This study illustrates some important considerations in the diagnostic workup and management of patients with CUP.
In many ways, our viewpoints on CUP management recapitulate the broader evolution of how we approach cancer therapy. Historically, CUP was viewed as a distinct form of cancer, thereby meriting a search for unique CUP-specific therapies. In recent decades, the observation of latent primaries, immunohistochemistry (IHC) profiles, and tissue-of-origin (ToO) molecular assays provide growing evidence that CUP cancers are likely unrelated groups of site-specific tumors that happen to share the property of having a diminutive primary tumor that escapes detection. This led to the suggestion that CUP cancers with profiles concordant with those of site-specific cancers should be treated with therapies known to be effective in that cancer. The data suggest that this approach may be effective in at least some CUP subsets. Most recently, there has been a broader push toward personalized medicine across all cancer types with the goal of identifying driver mutation(s) in an individual patient that can be treated with targeted agents independent of the site of origin. Should this approach eventually be appropriately validated, CUP would be a natural fit for GA-based targeted therapy independent of tumor site.
To label a cancer as CUP, a focused search for the primary tumor with imaging and pathologic analysis is obligatory.2 In the absence of a primary, the current emphasis then is on identifying the primary profile and directing therapy at that putative tumor type. Immunohistochemical analysis is considered a useful tool in the CUP pathologic evaluation, based on the premise that there is concordance in the expression profiles of primary and metastatic cancers. The predictive value of IHC improves with grouping and recognition of patterns that are strongly indicative of specific tumors.3 Although not studied prospectively, patients with CUP with site-specific IHC and select mutations or actionable fluorescence in situ hybridization–based molecular markers are believed to have well-defined therapy options, as also noted in the study by Ross et al,1 which excluded such patients. For example, TTF-1+/CK7+ and Ck20+/CDX-2+/CK7− phenotypes are viewed as being diagnostic for lung and lower gastrointestinal cancer, respectively, and these patients are treated with the armamentarium of regimens used for these cancers.4 Therefore, most would agree that some basic IHC analysis forms the backbone of the diagnostic evaluation and cannot be abandoned until site-specific therapies for non-CUP (known) cancers are replaced with GA-based targeted therapies. However, we urgently need a tiered, uniform set of IHC markers to promptly determine patterns that are diagnostic of putative primaries and prevent exhaustive testing that depletes the tissue sample, especially the subsets of poorly differentiated and undifferentiated carcinomas. The real question, also teed up by Ross et al,1 is what we do with the roughly 75% of patients for whom the first-level IHC analysis does not strongly indicate the putative primary site. The alternatives are additional IHC or molecular ToO profiling to more comprehensively seek the putative primary or to use next-generation sequencing (NGS) molecular tools to identify potential molecular targets.
In the current era of targeted therapies, the focus is toward individualizing cancer management through the use of NGS molecular tools to identify driver mutations in individual tumors. In that context, the role of tissue specificity and putative primary designation may eventually become less relevant in CUP (as in known primary cancers). However, for now the initial steps in CUP management are clear—first definitively confirm that it is CUP, then use a standard IHC panel to look for a putative primary, and if a match is found, treat appropriately. Ross et al1 show that NGS of CUP samples also identifies potentially druggable mutations, and they suggest that NGS should be the next step in the management of these patients. However, this brings up some issues, some that relate to personalized oncology in general (such as who should pay for NGS and under what circumstances) and others, discussed herein, that are more specific to CUP.
It is worth noting that the imaging and pathologic framework remains essential in better understanding the results of genomic tests in patients with CUP—for example, in Ross et al,1 the most common genomic alterations including TP53 (55%), KRAS (20%), CDKN2A (19%), and ARID1A (11%) are perhaps the predicted alterations in a patient with CUP who presents with a pancreaticobiliary or upper gastrointestinal tract profile. Given the marked heterogeneity of the CUP presentations, a consecutive group of osseous or nodal-dominant cancers may have different frequencies of GA compared with those found by Ross et al.1 In addition, the differing frequencies of GA noted in ACUP vs non-ACUP, as mentioned by Ross et al,1 require larger validation studies—mutation or amplification of ERBB2, and alterations of EGFR and BRAF are common in the known adenocarcinomas compared with the nonadenocarcinomas (excluding BRAF in melanoma). MLL2 alteration was seen in 8 patients with non-ACUP vs 4 patients with ACUP, which is a small sample size with which to make robust biologic or therapeutic judgments. Whether the GA is a function of the histologic subtype, regardless of CUP designation, or unique to CUP cancers remains a key question.
Also, most CUP tumors, as with cancers in general, likely have multiple mutations (4.2 GAs per tumor in this study) and determining oncogenic drivers and which to target is perhaps, today, best understood in the cellular context. For example, a KRAS mutation in a patient with “lung profile” with hilar nodes plus brain metastases may have different implications compared with a patient with a “colorectal profile” with carcinomatosis and ascites, as may a BRAF V600E mutation in a patient with “intestinal profile” ACUP vs an undifferentiated cancer.
Attempts to demonstrate the independent value of NGS by measuring its impact on management decisions or survival are challenging, relating to trial design and CUP heterogeneity. The traditional prospective randomized trial design (comparing biomarker-based therapy with educated-guess first-line cytotoxic therapy) is difficult because an adequately powered trial would require more than 500 patients and still run the risk of ambiguous results. Given the limited number of CUP cases and the lack of a well-defined regulatory path to study new CUP therapies, piggybacking on established early trials is probably the fastest way forward until certain patterns emerge for CUP subtypes. One such trial is the Molecular Profiling–Based Assignment of Cancer Therapeutics (M-PACT) trial developed at the National Cancer Institute (NCI) and expanded to the NCI-supported Early Therapeutics Clinical Trials Network, which uses a randomized trial design to assess whether assigning treatment on the basis of specific genetic mutations can improve the overall response rate and/or 4-month progression-free survival in patients with advanced solid tumors.5 Tumors are genetically sequenced to look for a total of 391 different mutations in 20 genes that are known to have a match with targeted therapies. Patients are assigned to arm A (receive a treatment prospectively identified to target their specific mutation or relevant pathway) or arm B (receive a treatment not prospectively identified to target their specific mutation or relevant pathway). Another planned randomized study is IMPACT2 (Initiative for Molecular Profiling and Advanced Cancer Therapy), which is a collaboration of Foundation Medicine and the University of Texas MD Anderson Cancer Center. These 2 studies and others will inform creative trial design in CUP subtypes.
One could envision using an algorithm that integrates IHC, select ToO profiling, and CGP in some combination to maximize clinically meaningful benefit and minimize costs, especially in patients with limited or difficult-to-access tumors and poorly or undifferentiated neoplasms. An algorithm such as this would then continue to evolve as additional experience is gained with matching the right patient to the right drug and as the trade-off in costs, accuracy, and benefits became clearer.
From a development perspective, it is especially encouraging that several molecular targets may be independent of the tumor site, making it possible to include patients with CUP in new studies of targeted therapies and allowing us to piggyback on the broader advances in personalized cancer therapy. We will require creative approaches to clinical studies and learning from the current trends. These trends can then perhaps help in establishment of an international CUP mutation consortium that groups CUP subtypes (eg, liver, osseous, nodal, carcinomatosis dominant presentations) and mutations to plan innovative smaller trials. Just as we need to be selective in our diagnostic approach using an effective algorithm that leverages the proteomics and genomics techniques, we need to be selective in our research efforts to deliver validated new approaches to our patients with CUP.
Corresponding Author: Gauri Varadhachary, MD, Department of Gastrointestinal Medical Oncology, University of Texas MD Anderson Cancer Center, 1515 Holcombe Blvd, Unit 426, Houston, TX 77030 (email@example.com).
Published Online: February 12, 2015. doi:10.1001/jamaoncol.2014.277.
Conflict of Interest Disclosures: None reported.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Varadhachary G. Carcinoma of Unknown Primary Site: The Poster Child for Personalized Medicine? JAMA Oncol. 2015;1(1):19–21. doi:10.1001/jamaoncol.2014.277
Browse and subscribe to JAMA Network podcasts!
Create a personal account or sign in to: