Epigenetic Profiling Identifies Origins of Unknown Cancers

Alexander M. Castellino, PhD

September 12, 2016

The unknown part of cancers of unknown origin may have become a little less mysterious.

Epigenetic profiling, which describes DNA methylation status, determined the primary origin of these tumors with a reported accuracy of 87%, according to a study published online August 26 in Lancet Oncology.

The researchers devised the EPICUP test to classify cancer type on the basis of microarray DNA methylation signatures in a training set (2790 samples) of 38 known tumor types. The signatures were validated in an independent set (7691 samples) of known tumor samples and were then applied to predict tumor types for 216 cases of cancers of unknown origin.

"One of the most important ideas from the report is that the correct determination of the tumor site that originated the metastasis gives the patient an increased overall survival if he/she is treated with a specific therapy for that tumor type. The developed test, EPICUP, provided that advantage," corresponding study author Manel Esteller, MD, PhD, of the Cancer Epigenetics and Biology Program (PEBC), Bellvitge Biomedical Research Institute, Barcelona, Spain, told Medscape Medical News.

The technical approach here makes good sense, say a pair of Greek experts in an accompanying commentary.

"[M]any genes have been shown to obtain DNA methylation status in a cancer-specific manner and abnormal DNA methylation landscape are ubiquitous in human cancers," write Panagiota Economopoulou, MD, PhD, of the National Kapodistrian University of Athens and George Pentheroudakis, MD, University of Ionnina.

"[The researchers] used epigenetics as a molecular marker to develop a tissue of origin molecular profiling assay based on DNA methylation status. The results of the generated assay are encouraging," they add.

Leonard Saltz, MD, chief of the Gastrointestinal Oncology Service, Memorial Sloan Kettering Cancer Center, New York City, told Medscape Medical News: "The study is scientifically interesting, but should, I believe, be considered a work in progress.

"It confidently tells us the site of the primary tumor, but, as the authors correctly state in their concluding paragraph, prospective studies are needed to show whether or not this approach truly leads to better outcomes for patients," added Dr Saltz, who was not involved in the study.

Study Details

The researchers first determined the methylation signatures of 692 samples from their PEBC cohort (2011-2015) to establish a reference dataset for DNA methylation profiles of 38 tumor types.

They used a microarray that determined the methylation status of about half a million CpG sites in the human genome. The same platform was used by the Cancer Genome Atlas (TCGA) to analyze the methylation status of a large number of tumor types. The investigators added 2098 tumor samples of known origin from the publicly available database that corresponded to their 38 tumor types. In total, the training set comprised of 2790 tumor samples.

For their validation set, the researchers used 1948 independent samples from the PEBC cohort and 5743 samples from TCGA — all across the same 38 tumor types.

The specificity (true negative rate) of the EPICUP test was 99.6%, with 97.7% sensitivity (true positive rate) for determining the tumor type in the validation set. The positive predictive value was 88.6%, and the negative predictive value was 99.9%.

Applied to a validation cohort of 534 metastases in tumors of known primary origin, the EPICUP assay correctly predicted tumor type for 501 (94%) of the 534 metastases.

Dr Esteller pointed out that tumors of unknown origin have a very enigmatic pathology in which the patient presents with multiple metastases, but the origin of the primary tumor remains unknown. "Our approach using epigenetic profiling uncovered where the primary culprit is located," he said.

The investigators further showed that the type of sample used in EPICUP had no effect on its ability to predict tumor type. Samples from fresh frozen tissue or from formalin-fixed, paraffin-embedded samples behaved alike.

"The use of DNA material in EPICUP might solve practical issues such as tissue availability," comment Dr Economopoulou and Dr Pentheroudakis.

The researchers used the validated EPICUP assay to determine the methylation signatures of 216 patients with tumors of unknown origin diagnosed between 1998 and 2015. In 188 of these samples, the EPICUP assay predicted the tissue of origin with an accuracy of 87%. The results were verified with an accuracy of 96% to 100% using several methods — necroscopy (1 sample), light microscopy evaluation (181 samples), and immunohistochemistry with specific markers (43 samples).

However, Dr Saltz asked: "If 181 samples could be validated by light microscopy, and 43 by specific markers (presumably immunohistochemistry), then why are these considered to be unknown primary?

"Cancer of unknown primary is a diagnosis usually reserved for a patient in whom the primary cannot be identified after routine quality diagnostic imaging and procedures, as well as routine pathologic review," he added.

"Ideally, validation of tumor of origin should be based on patients' autopsies in a prospective study; however, in view of legal and ethical issues, such studies are very difficult to do in current clinical practice," Dr Economopoulou and Dr Pentheroudakis write.

The investigators determined the outcomes for patients who were given site-specific treatment, as suggested by the EPICUP assay (n = 31), and compared them with outcomes from empiric therapies that did not match the profile of the methylation assay (n = 61). With a hazard ratio of 3.24, patients receiving site-specific treatment had a significantly longer overall survival (P = .0029).

However, Dr Saltz indicated that when one looks at the treatment the patients received, it is not clear what criteria were used to designate a therapy as site-specific.

"I have a problem understanding what was considered site-specific treatment and what was not. There appears to be a lot of subjectivity in these designations," he said.

He noted that a patient treated with carboplatin/paclitaxel for breast cancer is listed as having site-specific treatment, as are another breast cancer patient who was treated with cisplatin/taxanes and cisplatin/5FU and a prostate cancer patient who was treated with single-agent capecitabine followed by single-agent paclitaxel.

Dr Saltz commented that this appears to be a retrospective review of results. "It is not a prospective randomization to use or not the EPICUP assay to guide therapy, which is what would be ideally needed," he said.

"I'm not quibbling about whether this assay can identify a primary; I'm saying that I'm not yet seeing data to support a claim that use of this assay improves outcome for patients. That question remains to be addressed," Dr Saltz said.

"Further prospective trials are required before we can declare victory using this approach," he added.

Clinical Management of Cancers of Unknown Origin

Dr Saltz explained that at Memorial Sloan Kettering, patients presenting with tumors of unknown primary origin are first evaluated for clinical presentation and receive a CT scan of the chest, abdomen, and pelvis and undergo other diagnostic procedures as clinically indicated. The biopsy material is evaluated by light microscopy with judicious use of immunohistochemistry.

For cancers of unknown origin that are predominantly below the diaphragm, patients are usually treated in accordance with a gastrointestinal/colorectal paradigm; for tumors predominantly above the diaphragm, a non–small cell lung cancer paradigm is typically used.

In research, Dr Saltz and his colleagues are studying molecular profiling to determine actionable mutations in the biopsy sample and to what degree patients can or cannot benefit from such determinations.

In clinical practice, some clinicians use a variety of diagnostic tests, such as Pathway Diagnostics and BioTheranostics Cancer Type ID, which accurately determine tumor origins in 75% to 92% of patients. These assays are based on mRNA expression, which is unstable, the commentators point out. The EPICUP assay may be more cost-effective, they indicate.

However, Dr Saltz told Medscape Medical News that none of the assays are endorsed by National Cancer Care Network guidelines.

"All tests come at an additional cost, and unless they are shown to offer tangible benefits, such as improved outcomes for patients, I don't believe they should be used in routine clinical practice," he added.

With this viewpoint, he disagreed with the study investigators, who conclude: "[It] is becoming clear that the days of empirical chemotherapy treatment of cancers of unknown primary are reaching their end, and that molecular profiling...will be crucial to the development of tumour type and patient type-specific treatment."

Dr Economopoulou and Dr Pentheroudakis point out that no randomized study has compared outcomes for patients receiving site-specific treatment with empirical therapy. Such a study would identify a subset of patients who may benefit more from targeted therapies, they indicate.

Although two randomized studies are recruiting patients to assess molecular profiling–based treatment in metastatic malignancies, "a question that remains unanswered is whether a cancer of unknown primary with a molecular signature of a specific primary behaves similarly to a typical metastatic cancer," they write.

"A vision of the future would be the creation of a clinically useful diagnostic algorithm, which would incorporate pathological findings and molecular profiling tests along with crucial clinical judgment to maximise clinical benefits and limit costs," Dr Economopoulou and Dr Pentheroudakis conclude.

Several study authors have financial relationships with industry, as indicated in the original article. Dr Economopoulou, Dr Pentheroudakis, and Dr Saltz have disclosed no relevant financial relationships.

Lancet Oncol. Published online August 26, 2016. Abstract, Commentary


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.