CancerSEEK Is 'Good First Step,' But More Work Needed

Alexander M. Castellino, PhD

January 24, 2018

Last week's news about a blood test to detect eight common cancers, dubbed CancerSEEK, generated enthusiastic headlines across the world, but cancer experts contacted by Medscape Medical News emphasize that there is still a lot of work to be done before the novel test can be used to screen for cancer in healthy people.

As already reported by Medscape Medical News, CancerSEEK uses a novel approach to detect cancer by testing for both eight protein biomarkers and for DNA segments from 16 genes, and then uses machine-based learning to analyze the data. Developed by a team at the Johns Hopkins Kimmel Cancer Center in Baltimore, Maryland, details of the test and its initial performance were published in Science.

This test falls under the moniker of "liquid biopsy" in that it detects cancer from a blood sample.

Most "liquid biopsy" approaches to date have focused either on detecting protein biomarkers for cancer, such as prostate-specific antigen (PSA) in prostate cancer, or on cell-free DNA (cfDNA), which is DNA shed by normal or malignant cells into the circulation, explained an expert in the field, Daniel Stover, MD, from Ohio State University, Columbus. In addition, most of these approaches have focused on advanced cancers, where circulating biomarkers are relatively abundant and more easily detectable.

The novelty of the new CancerSEEK test is that the researchers combined cfDNA and protein biomarkers. "Their approach is designed to detect many different types of cancer, attempting to offer a 'one stop' liquid biopsy approach," Dr Stover told Medscape Medical News.

In their Science paper, the John Hopkins team reported using the new test in 1005 patients with nonmetastatic, clinically detected cancers (ovary, liver, stomach, pancreas, esophagus, colorectum, lung, or breast) and 812 healthy controls. 

Dr Stover pointed out that the number of tested individuals allowed the team to demonstrate the high specificity of the test —  over 99% — suggesting that a positive test result is very likely associated with the presence of a cancer. But he also noted that while it is a specific test, its sensitivity is lacking — approximately 70% or less for six of the eight tested cancers, with breast cancer sensitivity less than 40%.

"This suggests that while a positive test is strongly associated with the presence of cancer, a negative test does not conclusively rule out cancer and could be falsely reassuring in the clinical setting," he told Medscape Medical News.

"In addition, this test did not perform particularly well for most stage I cancers. While detecting cancer before distant spread (metastasis) is critical, most data suggests better outcomes for cancers detected early (stage I) rather than stage II or III," Dr Stover said.

David Klimstra, MD, chair in the Department of Pathology at the Memorial Sloan Kettering Cancer Center, New York City, said that there is great interest in liquid biopsies that detect abnormal circulating tumor DNA, as they are expected play an important role in the diagnosis of cancer in the future.

In addition to their promise for screening for the presence of cancer, liquid biopsies can be used to profile the mutations of a known cancer when there is inadequate biopsy tissue for conventional sequencing approaches, as well as to help assess the response of a cancer to precision therapy, to search for new mutations that predict resistance to treatment, and to look for genomic heterogeneity, which occurs as cancers progress.   

"However, it is important to recognize that each of these uses requires extensive validation — both at the level of the test performance (analytical validity) and to ensure that the test accurately predicts the disease condition and its clinical consequences (clinical validity)," he noted.   

"For a test to be used for cancer screening, it must clear a very high bar," Dr Klimstra said. 

He elaborated further on the significance of false-positive test results in the area of cancer screening. "False-positive tests — even if they only occur rarely — can result in potentially risky, and costly, additional interventions attempting to find the cancer," he said.  "Also, some cancer-associated mutations occur in precursors to invasive cancer that, by themselves, are not a cause for medical intervention," he added.  

Dr Klimstra pointed out that the potential to harm patients —physically or emotionally — must be weighed against the benefit of early detection, and it must be shown that earlier detection using more sensitive technologies will ultimately lead to better survival. 

"Although the latter point would appear self-evident, decades of experience with mammographic screening challenge the assumption that cancer mortality is universally reduced by earlier detection," Dr Klimstra said. 

Daniel Hayes MD, a breast cancer oncologist at University of Michigan Comprehensive Cancer Center, Ann Arbor, commented that the study with CancerSEEK was "nicely done," and he described it as "a good first step."

This test "has potential to be an assay for screening cancers in the future," Dr Hayes told Medscape Medical News.

However, he added that "generating data and showing clinical utility for cancer screening generally is a huge undertaking with many pitfalls, especially since both high sensitivities and specificities are important."

 "In screening tests, specificity is especially important — specificity for cancer vs noncancer and specificity for one tissue versus another," he added.

"A criticism of screening programs is one of overdiagnosis," Dr Hayes said. Without a doubt, the Papanicolaou smear and colonoscopy are worthwhile screening tools, he noted. "But as a breast cancer physician, it is apparent that a large number of cancers we pick up in screening mammography would never be evident in a patient's lifetime," he said.

He explained that the significance of mammography as a screening tool has been addressed in nine prospective randomized clinical trials in breast cancer involving hundreds of thousands of women. "We are still arguing over the issue of overdiagnosis with mammography," he said. 

The same is true for screening for prostate cancer with the PSA test, which also has been tested in randomized studies but is still controversial. "Therefore, specificity for the cancers that lead to morbidity and mortality vs those that will never be clinically evident must be established," Dr Hayes said.

Strengths of CancerSEEK

Dr Stover noted that one of the strength of the CancerSEEK study is that it included detection of several cancer types that do not have routine screening tests for average-risk individuals, including cancers of the ovary, liver, stomach, pancreas, and esophagus. "This is to be lauded as it seeks to develop a diagnostic where there is a definite need," he said.

In addition, with its estimated cost of $500 per test, CancerSEEK would be relatively cost-effective, he said. "This improves the likelihood of wide accessibility from a financial perspective and raises the possibility that it could be a better fit in the value-based framework of current medical practice," Dr Stover elaborated.

Limitations of CancerSEEK

A limitation of the study was that the patients with cancer all had clinically detected cancer and many had higher-stage (stage II or III) cancer, Dr Stover pointed out. "While an important testing ground, it needs to be evaluated whether this test detects cancers before clinical detection," he added. 

"This is a hand-picked cancer population and even then the sensitivity is not as nice as one would like," Dr Hayes said. He also questioned how healthy the adults in the control group were. "We know that the general population who is a target for screening is not healthy with respect to benign conditions," he told Medscape Medical News.

"Inflammatory and benign conditions can at the least make serum proteins go up and down, and we really don't know about circulating cell-free DNA. How will the test fare in the general population?" Dr Hayes said.

Dr Hayes was also surprised that the proteins the investigators selected provided tissue specificity. CA-125, for example, was first discovered in patients with ovarian cancer, but it is also elevated in patients with breast cancer, he noted. CEA is typically associated with colon cancer, but it is also elevated in patients with breast cancer, lung cancer, and prostate cancer.

Dr Stover also noted that the authors use a machine learning approach to predict cancer localization in an attempt to guide follow-up imaging/testing. "It is not clear how useful this will be clinically," he said.

Validation Required and Is Ongoing

CancerSEEK is a clever idea and an interesting study that offers promise, Dr Hayes said. "We can be optimistic, but we need to be cautious before taking it into clinical practice. In the past similar assays when taken into the clinic without proper validations met with unfortunate clinical outcomes," he added.

Dr Hayes explained that about a decade ago, OvaCheck (Correlogic Systems) was developed for ovarian cancer based on proteomics. It was very quickly taken to the clinic with preliminary data, where it was associated with an unacceptable rate of false-positive result. 

"I'm fond of saying that 'a bad tumor biomarker test is as bad as a bad drug'," Dr Hayes said. "In the screening arena, this can't be said often enough," he added.

"Before we start using a test such as CancerSEEK to screen otherwise unaffected people, we need to be sure we aren't hurting more people than we are helping, and one cannot just assume that the test has benefits. It has to be proven," Dr Hayes commented.

Dr Klimstra agreed. Understanding how the results of liquid biopsies can be used in cancer detection will require extensive research, such as the validation study being conducted. "Only after assessing all of the risks, benefits, and costs can a new diagnostic technique such as this be considered for broad clinical use," he said.

Dr Klimstra was referring to a validation study highlighted in an accompanying Science news story. In collaboration with Johns Hopkins, the Geisinger Health System in Danville, Pennsylvania, has already begun to use CancerSEEK on blood samples from female volunteers aged 65 to 75 years. The planned $50 million, 5-year study of up to 50,000 women is being funded by a private philanthropic group, The Marcus Foundation.

This validation study will also put to rest another criticism about the CancerSEEK report: that a single dataset was used for internal validation. In addition, it is being done in a population of individuals who have never had a cancer diagnosis. Results will be eagerly awaited.

Follow Medscape Oncology 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.