Clinical relevance of the molecular landscape in cancers of unknown primary

The development of sequencing techniques has revolutionized the possibilities of deciphering the molecular landscape of cancer genomes (1). By now, testing for selected genomic aberrations has become standard in oncology. Beyond this, panel sequencing of groups of genes and whole-genome sequencing approaches have become feasible. Technically, it is possible to perform these sequencing analyses from formalin-fixed paraffin-embedded (FFPE) tissue. Moreover, newly developed technologies permit us to additionally analyze circulating tumor cells, DNA and RNA.

In cancer of unknown primary (CUP), the diagnosis of cancer is made based on the metastases, but no primary tumor can be identified in spite of thorough immunohistochemical testing and extended clinical work-up with imaging and endoscopies. Sequencing technologies, therefore, hold a particular promise in CUP. They offer the potential to shed light on the pathogenesis of this enigmatic entity, to provide hints for the most likely tissue of origin and to offer starting points for targeted therapy.

Molecular analyses so far have yielded insights into the pathogenetic mechanisms of CUP.(2-5)

Firstly, CUP is typically characterized by a variety of genomic aberrations, including presumed oncogenic drivers. The largest study to date by Ross et al. assessed 251 genes in a series of 200 patients and detected a mean 4.2 genomic aberrations per patient (2). At least one genomic aberration was present in 192/200 cases (96%).

Likewise, the studies by Löffler et al. and Tothill et al., utilizing panels of 50 and 701 genes, respectively, detected at least one genomic aberration in 46/55 (84%) and 16/16 (100%) of patients (3, 4). Regarding the frequencies of individual genomic aberrations, these three studies have consistently shown that TP53 mutations are by far the most abundant, found mutated in 38–-63%  of CUP cases (2–5). Other frequently detectable aberrations included K-RAS (reported in the 13–20% range), CDKN2A (19–22%), ARID1A (11–13%), PIK3CA (4–19%) and SMAD4 (3–13%). Of note, numerous aberrations affect the RTK/RAS signaling pathway, including BRAF, EGFR, ERBB2, FGFR, K-RAS and N-RAS.

Secondly, the broad spectrum of differing aberrations suggests that there is no uniform pathogenetic signature in CUP (1, 3). The interpatient heterogeneity of the molecular landscape implies that the molecular architecture follows that of the presumed primary tumor.

Thirdly, the mutational landscape appears to depend on the histological type. The RTK/RAS signaling mutations of ERBB2, EGFR and BRAF are more prevalent among CUP cases of adenocarcinoma type, whereas MLL2 mutations have only been found in non-adenocarcinoma CUPs (2).

Finally, the study by Löffler et al. also gives first insights into the impact of specific genetic aberrations for the prognosis of CUP patients: K-RAS and CDKN2A mutations confer poor progression-free survival. TP53 mutations led to marginally shorter overall survival, with a better prognosis of TP53 wild-type cases, particularly in females (3).

High hopes have been placed into the role of gene-expression profiling to identify the likely primary tumor. Advances in this field are eagerly awaited, given that the trend in CUP goes towards therapy tailored to the presumed primary tumor. The molecular profiling approaches are particularly warranted for poorly differentiated tumors, where immunohistochemistry fails to suggest a likely primary tumor (1).

Several microarray-based expression profiling tests have been established for the purpose of locating the likely site of tumor origin. In these assays, the molecular profile of a CUP case is assigned to the most likely primary tumor by comparing the molecular signature of the CUP case to control panels with known primaries for similarity (6–8). The accuracy of these techniques has been validated by applying them to cancers with known primary and CUP cases with primary tumors unmasked during the further course of disease.

In a study by Hainsworth et al., expression profiling with a 92-gene cancer classifier assay permitted allocation of the CUP tumor to a likely tissue of origin in 98% of cases (6). Subsequently, treatment was implemented according to the likely primary site. This treatment strategy led to a 2-year overall survival of 49% in patients with treatment-responsive tumor types and high prediction probability, which compared favorably with retrospective results using empiric CUP regimens. The validity of this strategy for primary identification is also supported by a recent study by the same group, where patients predicted to have a likely lung primary by this assay displayed a high frequency of ALK translocations with 4/21 positive patients, reminiscent of lung cancer (9).

In line with these observations, another group observed that patients presumed to have a primary tumor typically responsive to a platinum/taxane chemotherapy combination (breast, lung and ovarian cancer) by a 2000-gene expression microarray assay actually displayed a superior response to platinum/taxane as compared with patients with a predicted primary known to be less responsive to this regimen (response rates 53% vs. 26%, respectively) (10).

In addition to this microarray-based profiling, individual genomic aberrations can give hints towards the most likely primary as well. As discussed above, ALK translocations are judged to be suggestive of a lung primary (9). Further examples are BRCA mutations, which imply a gynecological primary and IDH1 mutations, which suggest a pancreatico-biliary primary tumor (11, 12). Often, however, mutational profiles, which frequently contain mutations in TP53, K-RAS and CDKN2A as described above, are compatible with several possible tissues of origin.

Next-generation sequencing has entered clinical medicine in the last decade and is expected to impact the standard of care in oncology. Beyond guiding treatment decisions by suggesting a primary via molecular profiling, sequencing strategies also open up the perspective for directly targeted therapies with drugs like kinase inhibitors, immune checkpoint inhibitors and antibodies, although the identification of patients who benefit from these approaches is still challenging (1, 13).

A paramount example for biomarker-predicted targeted therapy in CUP is the use of crizotinib for cancers harboring ALK translocations or MET amplifications. Case reports have shown that these mutations are successfully druggable in CUP (2, 14). Activating EGFR mutations offer the possibility of anti-EGFR treatment with erlotinib, afatinib, gefitinib or lapatinib, mimicking the treatment of lung cancer. ERBB amplification or overexpression is known to entail responsiveness to ERBB/Her2 inhibitors like trastuzumab or lapatinib in breast, gastric and salivary gland cancers (15), suggesting that it is druggable in CUP as well.

On the contrary, the potential efficacy of BRAF inhibitors like vemurafenib or dabrafenib is still speculative in CUP, given that the same BRAF V600E mutation typically confers responsiveness to BRAF inhibitors in melanoma, though not in colon cancer. BRCA mutations suggest sensitivity to platinum-based chemotherapy and PARP inhibitors. Additionally, it was recently reported that immune checkpoint inhibitor therapy with pembrolizumab led to a lasting partial remission in a patient with adenocarcinoma CUP and PD-L1 amplification in conjunction with a high mutational load, who was refractory against multi-agent chemotherapy (16). The publication by Ross et al. gives a comprehensive overview of the numerous potentially druggable genomic aberrations in CUP (2).

Targeted therapies are obviously needed in CUP, where empiric standard chemotherapy fails to overcome the typically dismal prognosis. It appears reasonable to test a comprehensive panel of relevant mutations in a single genomic profiling test in all patients without an a priori selection for the likely primary (2). For example, ALK translocations reminiscent of lung cancer have been detected in patients where the immunohistochemical profile and the clinical picture were not really suggestive of lung cancer (9), arguing against limited panels on a case to case basis.

Clonal heterogeneity, evolution and tiding complicate the use of targeted therapies (1).
Obviously, even with the growing establishment of targeted therapies, many questions remain open. How should the molecular drugs be combined with chemotherapy? Could they help to avoid resistance due to clonal heterogeneity and clonal evolution (1)?

Molecular analyses have become a valuable tool in the diagnostic work-up of CUP patients and are increasingly employed at specialized centers. They will likely become part of the routine diagnostic work-up in CUP soon.

However, the lack of clinical trial evidence in CUP in general and in molecular therapies in particular poses significant difficulties both for clinicians to decide on the clinical benefits that can be reached by treating molecular aberrations as well as for insurance companies, which have to decide on financing on the background of little evidence. This situation also accounts for the large discrepancy between frequencies of what is reported to be a druggable mutation in CUP. In the study of Ross et al., 169/200 (85%) cases were deemed to harbor at least one genomic aberration with the potential for targeted treatment, whereas the percentage of druggable mutations was given with 6/55 (11%) or at best 8/55 (15%) by Löffler et al. (2, 3).

Clinical trials should be pursued to offer patients the access to new molecular drugs. Clinical basket trials represent one possible solution, as they permit the inclusion of patients irrespective of cancer entity, like the ongoing trials with IDH1 inhibitors. The German CUP working group is currently planning to establish a clinical trial comparing conventional chemotherapy with treatment driven by mutational profiling for second- line treatment after a first-line treatment comparing checkpoint inhibitor therapy with standard chemotherapy has failed.

  1. Dietel M, Johrens K, Laffert MV, Hummel M, Blaker H, Pfitzner BM et al. A 2015 update on predictive molecular pathology and its role in targeted cancer therapy: a review focussing on clinical relevance. Cancer Gene Ther. 22(9), 417–430 (2015).
  2. Ross JS, Wang K, Gay L, Otto GA, White E, Iwanik K et al. Comprehensive Genomic Profiling of Carcinoma of Unknown Primary Site: New Routes to Targeted Therapies. JAMA Oncol. 1(1), 40–49 (2015).
  3. Löffler H, Pfarr N, Kriegsmann M, Endris V, Hielscher T, Lohneis P et al. Molecular driver alterations and their clinical relevance in cancer of unknown primary site. Oncotarget. DOI: 10.18632/oncotarget.10035 (2016).
  4. Tothill RW, Li J, Mileshkin L, Doig K, Siganakis T, Cowin P et al. Massively-parallel sequencing assists the diagnosis and guided treatment of cancers of unknown primary. J Pathol. 231(4), 413–423 (2013).
  5. Gatalica Z, Millis SZ, Vranic S, Bender R, Basu GD, Voss A et al. Comprehensive tumor profiling identifies numerous biomarkers of drug response in cancers of unknown primary site: analysis of 1806 cases. Oncotarget. 5(23), 12440–12447 (2014).
  6. Hainsworth JD, Rubin MS, Spigel DR, Boccia RV, Raby S, Quinn R et al. Molecular gene expression profiling to predict the tissue of origin and direct site-specific therapy in patients with carcinoma of unknown primary site: a prospective trial of the Sarah Cannon research institute. J Clin Oncol. 31(2), 217–223 (2013).
  7. Hainsworth JD, Greco FA. Gene expression profiling in patients with carcinoma of unknown primary site: from translational research to standard of care. Virchows Arch. 464(4), 393–402 (2014).
  8. Greco FA, Lennington WJ, Spigel DR, Hainsworth JD. Poorly differentiated neoplasms of unknown primary site: diagnostic usefulness of a molecular cancer classifier assay. Mol Diagn Ther. 19(2), 91–97 (2015).
  9. Hainsworth JD, Anthony Greco F. Lung Adenocarcinoma with Anaplastic Lymphoma Kinase (ALK) Rearrangement Presenting as Carcinoma of Unknown Primary Site: Recognition and Treatment Implications. Drugs Real World Outcomes. 3, 115–120 (2016).
  10. Yoon HH, Foster NR, Meyers JP, Steen PD, Visscher DW, Pillai R et al. Gene expression profiling identifies responsive patients with cancer of unknown primary treated with carboplatin, paclitaxel, and everolimus: NCCTG N0871 (alliance). Ann Oncol. 27(2), 339–344 (2016).
  11. Quezado MM, Moskaluk CA, Bryant B, Mills SE, Merino MJ. Incidence of loss of heterozygosity at p53 and BRCA1 loci in serous surface carcinoma. Hum Pathol. 30(2), 203–207 (1999).
  12. Borger DR, Tanabe KK, Fan KC, Lopez HU, Fantin VR, Straley KS et al. Frequent mutation of isocitrate dehydrogenase (IDH)1 and IDH2 in cholangiocarcinoma identified through broad-based tumor genotyping. Oncologist. 17(1), 72–79 (2012).
  13. Roychowdhury S, Chinnaiyan AM. Translating genomics for precision cancer medicine. Annu Rev Genomics Hum Genet. 15, 395–415 (2014).
  14. Palma NA, Ali SM, O’Connor J, Dutta D, Wang K, Soman S et al. Durable Response to Crizotinib in a MET-Amplified, KRAS-Mutated Carcinoma of Unknown Primary. Case Rep Oncol. 7(2), 503–508 (2014).
  15. Chmielecki J, Ross JS, Wang K, Frampton GM, Palmer GA, Ali SM et al. Oncogenic alterations in ERBB2/HER2 represent potential therapeutic targets across tumors from diverse anatomic sites of origin. Oncologist. 20(1), 7–12 (2015).
  16. Gröschel S, Bommer M, Hutter B, Budczies J, Bonekamp D, Heining C et al. Integration of genomics and histology revises diagnosis and enables effective therapy of refractory cancer of unknown primary with PDL1 amplification. Mol Case Studies. 2016; August 24 (Epub ahead of print).
Author affiliations

Authored by Tilmann Bochtler (a,b), Alwin Krämer (a,b)

aGerman Cancer Research Center (DKFZ), Heidelberg, Germany, bClinical Cooperation Unit Molecular Hematology/Oncology, German Cancer Research Center (DKFZ) and Department of Internal Medicine V, University of Heidelberg, Heidelberg, Germany.