DNA methylation profiling may have potential in unmasking the origin of tumors classified as cancer of unknown primary. A diagnostic assay developed by the researchers, EPICUP, predicted the primary tumor site in 87% of 216 cases, and these predictions were subsequently verified through various tests, including immunohistochemistry.

Every year, more than 80,000 patients in the United States are diagnosed with cancer of unknown primary (CUP), meaning the site where their disease began cannot be determined. Because CUP is the fourth most common cause of cancer deaths worldwide, there is considerable interest in finding ways to unmask the origin of such tumors to better guide therapy and prolong survival.

A recent study led by the Bellvitge Biomedical Research Institute in Barcelona, Spain, indicates that epigenetic profiling may be one potential method (Lancet Oncol 2016;17:1386–95). The researchers designed EPICUP, a diagnostic assay that uses a comprehensive DNA methylation platform to interrogate approximately 450,000 CpG sites in the human genome. In a test run, they analyzed 10,481 tumor samples of known origin and reported that EPICUP's specificity in determining tumor type was 99.6%; its sensitivity was 97.7%.

Next, the team applied their assay to 216 cases of CUP diagnosed between 1998 and 2015 and found that it predicted the tissue of origin in 188 of them. They verified their results in several ways, including an autopsy in one case. EPICUP's predictions were compatible with immunohistochemistry in all of the 31 cases evaluated—for instance, vimentin expression in sarcoma, CDX2 in colon carcinoma, and mammaglobin in a case of male breast carcinoma. The assay's predictions also held up in 174 of 181 cases that were subjected to follow-up histology.

“DNA methylation profiling provides a consistent diagnosis of the primary tumor site in cases of CUP, according to our study,” says senior author Manel Esteller, MD, PhD. “We've created a system that can match an anonymous fingerprint against an immense database of potential tumor culprits.”

When the researchers examined clinical outcomes that were available from 92 of the 188 cases profiled by EPICUP, they found that 31 patients who received site-specific treatments that matched EPICUP's predictions, such as 5-fluorouracil for colon carcinoma, had a median overall survival (OS) of 13.6 months. In contrast, the remaining 61 patients who received empirical chemotherapy not matching the predicted tissue of origin had a median OS of 6 months.

Roel Verhaak, PhD, of The University of Texas MD Anderson Cancer Center in Houston, is unsurprised by the findings, pointing out that “CpG methylation patterns are highly determined by tissue lineage, and lineage footprints are expected to persist in CUP.” To him, Esteller's study “provides a solid basis for the use of epigenetics to classify such tumors.”

Verhaak also notes that CUP classifiers using transcriptome, or RNA-based, profiling instead are already commercially available. However, this approach has limitations. “RNA is more sensitive to degradation, especially in formalin-fixed, paraffin-embedded samples,” he says. Esteller agrees, adding that DNA is stable over time regardless of tissue fixation, “so we don't worry too much about conserving sample quality.”

The next steps for Esteller's team include a prospective randomized clinical trial to show that having the correct primary cancer diagnosis through EPICUP does improve survival. –Alissa Poh

For more news on cancer research, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/content/early/by/section.