New Approaches Lessen Mystery of Cancer of Unknown Primary

But Clinical Benefits Not Established

Liam Davenport

June 11, 2020

Next-generation sequencing (NGS) of tumor tissue and cell-free DNA (cfDNA) analysis of blood samples are identifying potentially actionable targets in patients with cancer of unknown primary (CUP), although it is not yet clear whether the use of drugs to exploit those will improve outcomes, warn experts.

During a special session of American Society of Clinical Oncology (ASCO) 2020 annual meeting, F. Anthony Greco, MD, Tennessee Oncology, Nashville, and Karim Fizazi, MD, PhD, Institut Gustave Roussy, Villejuif, France, led the audience through a series of presentations on the state of the art for genomics in CUP.

They showed that decades of improvements in molecular and genetic profiling are beginning to bear fruit in CUP, a relatively rare disease that affects 50,000 to 60,000 patients a year in the United States.

Among the included studies was NOMINATOR, which investigated the feasibility of NGS in more than 120 patients with rare cancers. In that study, 56% of patients had at least one actionable finding that could be subject to targeted treatment.

Another study investigated cfDNA. That study, which was led by Shumei Kato, MD, assistant clinical professor, Moores Cancer Center of the University of California, San Diego, found that in almost 2000 patients with CUP, 46% had genomic alterations associated with either response or resistance to therapy.

These patients included a 62-year-old man with poorly differentiated CUP who underwent cfDNA because of a lack of tissue for NGS.

This revealed KRAS and ARID1A gene alterations, which the researchers targeted with trametinib (Mekinist, GlaxoSmithKline) and olaparib (Lynparza, AstraZeneca), respectively.

"After 16 weeks of therapy, the tumor was nearly gone, and the patient is doing well to date," Kato said.

Social Media Responds

Following the presentations, in a live discussion at ASCO 2020, Greco said that molecular assays such as NGS and cfDNA have already made their way into the clinic, particularly for patients for whom classic immunohistochemical labels are not available.

Fizazi pointed out that although the data for NGS, for example, are "quite compelling," it is still not clear "whether we're doing some good to the patients, and we need randomized data to make sure" that the approach improves outcomes more than just on an individual patient basis.

He said that he would still probably perform NGS for his next patient who has a CUP, because it has the "potential to identify a target for which we can indeed use a drug that we would not use in the natural management of a patient with CUP."

The session generated a great deal of interest among clinicians. Pashtoon Kasi, MD, from the University of Iowa Hospitals and Clinics, in Iowa City, said on Twitter that CUP "should have mandatory tumor-based genetic testing as part of [the] upfront algorithm".

He added that targeted therapies and the potential of immunotherapy underline that it is important "to hone in on the tissue of origin."

Responding to NOMINATOR, Stephanie Graff, MD, director of the Breast Cancer Program at Sarah Cannon Cancer Institute, Nashville, Tennessee, described the results as "precision medicine at its finest."

However, one physician who watched the session live on the ASCO website commented that although the presentations were "excellent," the "great difficulty is how to access these diagnostic methods."

Opening the session, Greco said that in more than 800 autopsy studies published over the past 50 or 60 years, 75% of CUPs were found to have a small anatomic primary, often less than a centimeter in diameter. These primaries occurred in more than 25 locations and included diverse tissues, such as the intestinal tract, lung, breast, and urinary tract. It is therefore "clear that CUP is not one single cancer type.

"It represents a clinical syndrome with more than 30 types and subtypes, but all present with clinically undetectable anatomical primary sites," Greco said, adding: "If one finds a primary site, it's not CUP."

Although the etiology remains "obscure and unknown," he said it is "very likely" that genetic and/or epigenetic mechanisms lead to the development of detectable metastatic cancers from undetectable occult primaries.

Until 2008, during the "era of empiric chemotherapy," the survival of the most "unfavorable" groups of CUP patients was around 9 months. These accounted for 80% of cases.

Since then, clinicians have entered "the era of increased sophistication in diagnostic techniques, mainly immunohistochemistry and molecular cancer classifier assays."

As a result, the tissue of origin "became obvious in many of these patients with CUP." Up to 90% of tissue types are now identifiable. The majority are found to be carcinomas, primarily adenocarcinomas, as well as sarcomas and melanomas.

This has allowed patients to receive site-specific therapies based on molecular assay diagnosis, leading to improvements in survival.

"The era of involving improved therapies for many metastatic cancer has really been in the last decade, including precision-targeted therapies and immune checkpoint blockers," Greco said, "so the recognition of the specific cancer type in CUP is becoming more and more important."

The recognition of the specific cancer type in CUP is becoming more and more important. Dr Anthony Greco

This is best achieved, in his opinion, with comprehensive NGS or liquid biopsy, which in many cancer types is "necessary to plan appropriate therapies."

He cautioned, however, that "without knowledge of the cancer type or cellular context," NGS is "often misleading and may suggest inappropriate first-line therapy for several cancer types" that are "otherwise quite treatable with standard approaches."

The most recent NCCN guidelines recommended the "judicious use" of molecular cancer classifier assays. Greco said that "until more robust outcomes and comparative effectiveness data are available, pathologists and oncologists must collaborate on the judicious use of these modalities on a case-by-case basis."

NGS in Rare Cancers

Investigations into the effectiveness of these techniques continues apace. The first study of the session investigated the feasibility of NGS in rare cancers to match the cancer to its most appropriate treatment.

The study was not conducted with CUP patients. The session moderator, Melissa L. Johnson, MD, chair of ASCO's scientific program, explained that it was selected because it "highlighted some of the difficulty that patients with rare cancers and cancers of unknown primary may have gaining access to drugs that may be appropriate based on their molecular profile."

Lead author Damian Kee, MD, Peter MacCallum Cancer Center, Melbourne, Australia, noted, "Collectively, rare cancers account for up one quarter of all cancer diagnoses."

Their low prevalence means that access to research and clinical trials is limited, and patients with rare cancers "often lack evidence-based treatment and have worse outcomes than patients diagnosed with more common cancers."

Kee described the shift to treatment guided by molecular biomarkers as "opportune" for patients with rare cancers. He said that for NOMINATOR, he and his colleagues recruited patients with a "compelling" rare cancer.

These were defined as cancers in which the incidence was less than six per 1000,000 population, that are associated with a poor prognosis, and for which there is no evidence-based treatment.

In all, 121 patients were included. The vast majority (91%) had metastatic disease, and 63% had received at least two prior lines of therapy.

A comprehensive cancer panel revealed that of 100 tumors successfully sequenced, the most common aberrant genes were TP53, in 44%; CDKN2A/B, in 14%; RB1, in 14%; PTEN, in 13%; and NF1, in 12%.

The team used the OncoKB database levels of actionability to assess the genomic results. These range from recognized biomarkers predictive of response to an approved drug in a specific tumor type to more hypothetical associations. In addition, two levels relate to biomarkers associated with drug resistance.

With this approach, the researchers identified 93 actionable events across 27 genetic markers, with 18 matched drug classes. The result was that for 56% of the patients, there was at least one actionable finding, and for 27%, there was a potentially "high-impact" finding.

The genomic testing led to a revised diagnosis in six patients and the identification of germ-line variants in 13.

Eight patients died before the NGS results were available. Kee said that 24% of patients now have access to a matched drug treatment, suggesting that NGS is both "feasible" and should be used "earlier to maximize the potential benefits, as well as minimizing patients attrition".

cfDNA in Patients With CUP

But what should clinicians do for a patient for whom NGS of tumor tissue samples is not feasible?

The next study, presented by Kato and colleagues, investigated at the utility of cfDNA in patients with CUP.

Kato explained that cfDNA relies on detecting in the bloodstream low levels of DNA shed by the tumor. He believes this is "especially relevant for CUP patients, theoretically does not matter where the primary sites are."

Kato's team had previously conducted a preliminary study to demonstrate feasibility. In that study, which involved 442 CUP patients, the team examined 2022 samples from 1937 patients with CUP that were sent for commercially available cfDNA assay over a 3-year period.

The assay comprised a 73- or 74-gene NGS panel that assessed for sequence alterations, fusions, and applications. In addition, microsatellite instability analysis was available for a proportion of patients after its inclusion in the assay.

The assay was able to detect at least one cfDNA alteration on 1819 tests from 1743 patients, giving a detection rate of 90%.

Using the OncoKB database levels of actionability, the team found that 46% of patients had either a level 1 or 2 biomarker association with drug response or a level R1 association with drug resistance.

These included 396 (22%) patients with at least one level 1 cfDNA alteration, indicating the presence of a US Food and Drug Administration (FDA)–approved biomarker predictive of response to an FDA-approved drug in this indication.

In addition, 381 patients (21%) had at least one level R1 cfDNA alteration, suggesting a standard care biomarker predictive of resistance to an FDA-approved drug in this indication.

However, Kato pointed out that this means that approximately 50% of patients "still have challenging alterations without ideal targets, and this requires further clinical investigation."

Checkpoint Inhibition in CUP

As an example of how the latest treatments are beginning to make their way into the management of CUP, Junko Tanizaki, MD, PhD, Kishiwada City Hospital, Kishiwada, Japan, presented findings from NivoCUP.

This is a phase 2 study on the impact of the programmed cell death–ligand-1 (PD-L1) inhibitor nivolumab (Opdivo, Bristol-Myers Squibb) in CUP.

It was undertaken because the researchers had "recently found that the immune profile of a CUP is similar" to that in malignancies responsive to immune checkpoint inhibitors, and so CUP patients "might receive clinical benefit" from the drugs.

To be included in the study, patients had to have undergone a range of mandatory examinations, and tumor samples had to have been obtained less than a year previously. In addition, patients were required to have measurable disease on RECIST 1.1 criteria. Patients who were found to have melanoma, sarcoma, or lymphoma on pathologic examination were excluded.

In all, 56 patients, of whom 45 had been previously treated, were assigned to receive nivolumab twice weekly for a maximum of 52 cycles over 2 years or until disease progression or toxicity.

Patients who had been previously treated were slightly older than untreated patients (median age, 66 years, vs 64 years), and they were more likely to be women (62% vs 55%).

They were more likely to have ECOG performance status 1 rather than 2 (78% vs 46%) and were more likely to have three or more metastatic sites (38% vs 9%).

Of previously treated patients, 42% had received at least two prior lines of chemotherapy, and 22% had undergone radiotherapy.

During a median follow-up of 8.4 months, the objective response rate (ORR) among previously treated patients was 22%. Median progression-free survival (PFS) was 4.0 months, and median overall survival was 15.9 months.

Previously untreated patients were followed for a median of 17.2 months. Among these patients, the ORR was 18.2%, and the median PFS was 2.8 months. Median overall survival was not reached.

There was a trend for association between response rates and PD-L1 expression levels, although Karim Fizazi, MD, PhD, Institut Gustave Roussy, Villejuif, France, noted in his discussion of the presentations that it was not strong.

Although he said that nivolumab showed a clinical benefit, he disagreed with the researchers' assertion that it could become a new standard of care in CUP.

More Trials and the Frustrations of Being "Outside the Box"

Fizazi said that overall, the questions asked by the presentations regarded whether genomics are the "solution" to treatment in CUP and whether they can improve outcomes.

He said that the study by Kato and colleagues was "impressive" and that it was the largest to date to examine the feasibility and potential impact of cfDNA in CUP.

However, Fizazi pointed out that in NOMINATOR, NGS could not be performed in approximately 20% of patients and that the claims for actionable findings made by Kee and colleagues were "optimistic" and underline that the strategy is "challenging" in the real world.

He referred to his own GEFCAPI 04 randomized phase 3 trial, in which a diagnostic and therapeutic strategy that was based on molecular analysis and tailored therapy was compared to an empiric strategy in patients with CUP.

Fizazi underlined that this showed that overall survival was not improved with the tailored approach, even though some subgroups appeared to benefit.

This trial was nonrandomized, but he referenced his own GEFCAPI 04 randomized trial, in which overall survival was not improved. Some subgroups did improve, but the numbers were small.

However, he pointed to the ongoing phase 2 CUPISCO study, which is comparing molecularly guided chemotherapy with standard platinum-based chemotherapy for patients with CUP whose prognosis is poor. He said that that trial could provide clearer direction.

Fizazi summarized that the NGS techniques employed in the studies led by Kee and Kato are "promising," but that currently, there is no level 1 evidence to support their use.

In the subsequent live discussion, Johnson asked about difficulties in accessing drugs that may be appropriate for a CUP patient, as determined on the basis of their molecular profile.

Fizazi replied that although there are benefits to working in a social health system such as that in France, it is not possible to be reimbursed for a drug if you go "outside the box" of its approved indication.

"Sometimes it's very frustrating, because we know that the disease behaves pretty much like something we know with a drug that works, but it's very hard to access to that."

Referring to the NOMINATOR trial, he added that "anything that can help better provide data and strong enough data to support access to treatments in a rational way for rare tumors ― and this also applies to CUP ― is obviously welcome."

NOMINATOR was funded by Melbourne Health. NivoCUP was funded by Ono Pharmaceutical Co, Ltd. The presenters have disclosed multiple relationships with industry.

American Society of Clinical Oncology (ASCO) 2020: Presentation: Redefining Cancer of Unknown Primary: Is Genomics the Answer?

For more from Medscape Oncology, follow us on Twitter: @MedscapeOnc .


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.