Twists and turns in biomarker exploration

Karen Titus

August 2015—Forget the lone tree falling down, unnoticed and thus possibly soundless, in the forest. For pathologists and medical oncologists, the more meaningful philosophical question involves breast cancer biomarkers. If a biomarker looks promising in research, will its impact be felt in clinical practice?

That’s no idle question. Markers new and old continue to rise up and move out, like waves of soldiers at Pickett’s Charge. Some, such as proneurotensin and proenkephalin, might help identify women’s risk of developing the disease, along with possible links to diet, cardiovascular disease, and diabetes. At the other end of the spectrum, circulating tumor cells and cell-free DNA might guide treatment of metastatic disease. But as David Hicks, MD, says, navigating the biomarker pipeline is a hard journey. “It’s difficult to get from concept and theory to routine clinical practice,” says Dr. Hicks, professor of pathology and laboratory medicine and director of surgical pathology, University of Rochester (NY) Medical Center.

Breast biomarkers are as plentiful as finger-pointing after a financial collapse. “If you think about it, there’s all this stuff floating around,” says Dr. Hicks, yet the time-tested trio of ER/PR/HER2 continues to hold sway. “They’re still fighting about Ki-67,” he says, while Oncotype DX is probably the next closest entity to being widely accepted. “But we could do sequencing, mRNA—lots of stuff. We could spend hundreds of thousands of dollars. But what’s
the information that’s going to help guide treatment?”

Nevertheless, the hunt for new biomarkers hums along, steady as a tugboat. “I wish I were 20 years younger, because this field is about to explode,” says Martin Fleisher, PhD, attending clinical chemist and director, Biomarker Discovery Laboratory, Clinical Chemistry Service, Department of Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York. “I really want to be a part of it. It’s exciting.”

Dr. Fleisher and his colleagues at MSKCC are involved in two primary endeavors. The first involves circulating biomarkers, including proteins, cells, and nucleic acids. The second has to do with CTCs—not only in blood but also in cerebrospinal fluid. As Dr. Fleisher himself points out, measuring CTCs in blood has been a dynamic field for some 15 years. But it may be getting closer to clinical implementation. CTCs have, in his words, “taken on a new look.”

Particularly captivating, says Dr. Fleisher, is the role CTCs might have in diagnosing leptomeningeal infiltration, given that approximately 10 to 12 percent of patients with primary tumors, including breast, will develop brain metastases. Leptomeningeal metastases can be difficult to identify in their early stages, and current detection methods have drawbacks, Dr. Fleisher notes: CSF cytology often requires multiple lumbar punctures, and unequivocal MRI findings often appear only in late-stage disease.

In a poster at the 2015 ASCO Annual Meeting, Dr. Fleisher and colleagues presented data demonstrating the prospective validation of CSF CTCs to diagnose such metastases from epithelial tumors in 62 patients (including 27 cases of breast cancer) using rare cell capture technology. (Dr. Fleisher and colleagues previously evaluated this methodology to detect leptomeningeal metastases and published their findings: Nayak L, et al. Neurology. 2013;80:1598–1605.)

The researchers considered a sample to be positive if they found at least one CSF CTC in a 3-mL sample. Sensitivity with this approach was 95 percent, compared with 81 percent for CSF cytology alone and 62 percent for MRI alone. Specificity was 83 percent, though the researchers noted that results of one or less CSF CTCs/mL are of uncertain significance and might be linked to false-positive results. (There were seven false-positive CSF CTCs, all of which had one or less CSF CTCs/mL. The one false-negative CSF CTC sample was also negative on CSF cytology, but leptomeningeal metastases were seen on MRI in this case.)

In addition to its potential importance from a diagnostic standpoint, Dr. Fleisher is equally excited about the possibility of using a CSF biomarker to assess and possibly alter treatment. If a woman whose breast cancer is being treated with Herceptin develops metastatic disease, for example, CSF CTCs can be tested for HER2 positivity. “If it’s HER2 positive in the brain, the oncologist can treat that with Herceptin as well. That type of therapy has become very important,” he says.

It’s also becoming apparent, Dr. Fleisher continues, that if a patient with metastatic disease has CTCs in the blood, “that tells you something about that metastatic cancer.” If a patient with metastatic disease has a drop in CTCs, it might indicate the treatment is effective. But if the number of CTCs increases, perhaps a different therapy would be more effective. Only recently have researchers begun to pay attention to longitudinal counts of CTCs; early on, Dr. Fleisher says, most effort was spent on merely counting the cells. “We’re still learning,” he says.

“Now the other thing that’s quite interesting about circulating tumor cells in the blood—we’ve learned that you don’t have to capture the [CTC] to know about its genomic content.” In Dr. Fleisher’s lab, he and his colleagues have developed a method to collect blood in a PAXGene tube. The cells are lysed, and the tube can be easily stored for later analysis. “Then we take it out of the freezer and defrost it,” he explains, sounding like he’s reading from a 1960s cookbook. “And we process the PAXGene tube for the DNA by getting the messenger RNA and converting it back to cDNA. Then we can measure the genes that are upregulated in that patient sample.”

Dr. Fleisher would like to see more standardization brought to DNA analysis in tissue, blood, and cell-free DNA that’s extracted from plasma, especially given the interest in so-called liquid biopsies. He’s concerned about studies done in different labs whose results he’s been unable to duplicate. “There’s no standardization for cell-free DNA extraction, preparation, and storage,” he says. “That’s a major failing. Everybody has their own little method, which is not good.” Dr. Fleisher says he and his colleagues have developed SOPs for specimen handling, and he has submitted a proposal to present a course at the CAP ’16 meeting on standardizing DNA preparation. “It’s critical that everybody get onboard with handling nucleotide specimens in the same fashion, especially if they’re going to be used for clinical analysis.”

Dr. Fleisher also has his eye on another type of biomarker, one that could possibly be harnessed to treat cancer with immunotherapy.

The focus here is the PD (for programmed death) ligand, or PD-L1. As Dr. Fleisher explains, PD-1 is an inhibitory receptor that belongs to the CD28 family; when normal cells produce PD-L1, it prevents CD8+ T cells from attacking them. When cancer cells also learn how to produce PD-L1—and they do—they then can also rebuff the killing action of T cells. “The cancer cell is now safe from the destructive action of the T cell because it makes PD-L1. So how do we reprogram this cancer?” Dr. Fleisher asks.

One possibility is to immunosuppress the microenvironment that the inhibitory PD-L1 molecule is using, thus preventing malignant cells from being attacked. But that means researchers also need to figure out a way to keep T cells from attacking normal tissue while still attacking the cancer-related PD-L1 cell. And they need to figure out what to measure, and how.

It’s relatively easy, Dr. Fleisher says, to quantitatively measure PD-L1 in blood using an ELISA. When PD-L1 receptor is expressed on T cells, those cells produce cytokines. Dr. Fleisher and his colleagues have developed a panel of assays for measuring cytokines in peripheral circulation. They’re also working to develop a method for measuring PD-L1+ cells and looking for biomarkers that will differentiate between tumor-related PD-L1 and that which is found on normal tissue.

Dr. Fleisher is well aware of the futility and frustrations that accompany biomarker exploration. (He also knows the questions that lie in wait for every researcher who develops a promising biomarker or assay: “What are you going to do with the data once you get it? Will doing this assay improve patient outcome?”) Developing proteomic fingerprints for cancer, for example, has turned out to be more complicated than first thought.

Ditto for other markers and approaches. With cell-free DNA, for example, “You have to know what you’re looking for,” says Dr. Fleisher. “You can’t do exploratory work on cell-free DNA, because it’s only pieces of DNA, not the whole molecule. So you have to know specifically what mutation you’re looking for, and assay for that mutation.”

Exosomes are also intriguing, though he notes they’re hard to isolate. One-tenth the size of a cell, exosomes contain cellular information that is released into the circulation when a cell dies. Their value, says Dr. Fleisher, lies in their rather protected status—the DNA content and proteins are protected from proteases and other enzymes that can break down proteins and nucleotides in the blood. “These little packets contain very good information about what cell [tumor] they came from. Once we figure out how to collect exosomes efficiently from blood, that will be an important technique,” he predicts.

Dr. Fleisher speaks in the tone of an energetic inventor. The “gold ring,” as he calls it, would be to remove live CTCs from blood. “We want to have a live cell,” he says. “The excitement of this is getting live cells out of a patient with metastatic cancer and growing these cells in tissue culture or in a mouse, to get a xenograft tumor to study, both for biology and the therapeutic potential.”

Five years ago, he says, “We wouldn’t be talking about a lot of this. Now we’re actually doing it.”

There’s obvious clinical need for biomarkers that can help intervene in cancer treatment. But Alan Maisel, MD, is excited about the other end of the spectrum. What if biomarkers could be used to identify risk of breast cancer, and perhaps help patients avoid the disease before it begins?

Dr. Maisel, professor of medicine, University of California, San Diego, and director of the Coronary Care Unit and Heart Failure Program, Veterans Affairs San Diego Healthcare System, has fastened his hopes on two biomarkers that might represent, as he puts it, the missing link between diet and cancer: proneurotensin and proenkephalin. Both have moved into the spotlight as a result of the Malmö Diet and Cancer Study (Melander O, et al. JAMA. 2012;308:1469–1475); Dr. Maisel moderated a panel on breast biomarkers at the 2015 AACC Annual Meeting in July, which included discussion of both. The epidemiological study looked at fasting concentration of proneurotensin in 4,632 participants and found an association with breast cancer, diabetes, and both total and cardiovascular mortality. These findings were supported in another study, the Malmö Preventive Project (Melander O, et al. Cancer Epidemiol Biomarkers Prev. 2014;23: 1672–1676).

“With breast cancer,” says Dr. Maisel, who was involved with both studies, “there is an unbelievable unmet clinical need.” Apart from BRCA mutations—responsible for about five to 10 percent of breast cancers—physicians lack a means to identify a patient’s risk of developing cancer. Such a marker would serve—and here Dr. Maisel gives credit to UCSD colleague Barbara Parker, MD—as a “cholesterol for breast cancer.”

Proneurotensin is a gut hormone that affects how fat is absorbed in the body. It works, as Dr. Maisel explains, through three receptors: neurotensin 1 and 2, which are G-protein–coupled receptors, and neurotensin receptor 3, also known as Sortilin-1, which is non-G–protein coupled. Neurotensin receptors 1 and 2 are very strongly expressed in malignant ductal breast cancer tumors.

But again the question is raised: Even if a biomarker can identify increased risk, what will physicians do with that information?

Women with abnormal levels of proneurotensin and proenkephalin, Dr. Maisel says, should be educated and made aware they are at a higher risk for near-term breast cancer. They should also formulate a plan with their physician to reduce risk factors of breast cancer, which may include lifestyle, treatment therapy, or a combination of both. For the highest-risk group (10 percent), the clinician should use risk-reduction strategies as recommended by the NCCN guidelines, which may be supplemental screening with mammography—for example, MRI, tomosynthesis—and more frequent screening and lifestyle changes. “With this the No. 1 cancer in women, it gives a woman a chance to be empowered to proactively take charge of her health instead of reacting to a disease. The goal with these biomarkers is to prevent disease,” Dr. Maisel says.

The potential value of proneurotensin might be increased by measuring plasma proenkephalin (which was also studied by the Malmö researchers). Enkephalins are related to the opioid growth factor. “So we predicted in the Malmö study that if you had high enkephalins, you would inhibit cancer growth, and low levels would indicate a faulty surveillance mechanism that could lead to cancer growth,” says Dr. Maisel. The next step was to combine information from both markers. Dr. Melander and colleagues, again using fasting plasma from women who participated in the Malmö Diet and Cancer Study, showed that “if you take the highest quartile for proneurotensin and the lowest quartile for PENK, we had a hazard ratio of about 21-fold,” says Dr. Maisel.

There might also be a link between elevated proneurotensin levels, lowered proenkephalin levels, and risk associated with developing breast cancer as a result of hormone replacement therapy, suggesting that the two could be used as biomarkers to help guide HRT, Dr. Maisel says.

The diet link intrigues Dr. Maisel. It’s been shown that proneurotensin levels can rise or fall depending on diet; specifically, lowering fat in the diet lowers levels of the biomarker. “Now, we don’t know if that means you lower your risk,” he says, though that’s being studied as part of the Women’s Health Initiative. Some studies suggest yes, some say no. “But I think this is likely to be one of the major missing links between diet and cancer.” He says he and colleagues have tentative acceptance for three projects related to the Women’s Health Initiative; the plan is to use the study’s data to examine breast cancer and neurotensin levels as they relate to specific interventions in diet and exercise.

Dr. Maisel wants to be clear: These are not potential diagnostic markers. Their strength, instead, might be to predict risk of developing breast cancer. “I’ve spent a lot of my professional life working on people who are already sick, so many end-stage patients. They’re so sick,” he says. “That’s why I’m excited about this. It might be a chance to help people before they get sick.”

For every explorer who gazes beyond the trail and points to its bright destination, like modern-day Lewis and Clarks, there are just as many who gaze, with equal intensity, at the trail itself.

Allan Jaffe, MD, a clinical cardiologist and professor of medicine and laboratory medicine and pathology, Mayo Clinic, Rochester, Minn., falls into that latter group. Speaking of proneurotensin and PENK, he says, bluntly, “It’s a terribly complicated area. These are ubiquitous compounds.”

And just like that, he delivers a stark reminder that a ride in the biomarker pipeline comes with no guarantee of delivery.

Dr. Jaffe (who has not been involved in the Malmö work) isn’t dismissing either marker, and, in fact, he appeared on the AACC panel led by Dr. Maisel to talk about the underlying biology and physiology of proneurotensin and proenkephalin. He says there are experimental data suggesting there are pathways involving the neurotensin 1 and 2 receptors that could be involved in increased risk for tumor development, since they’re related to regulation of apoptosis and other factors that might facilitate cancer progression. “This might give you an early signal,” says Dr. Jaffe, though it’s not clear whether they might also be implicated in initiation of cancer.

Likewise, Sortilin-1 may be linked to cardiovascular disease through an LDL receptor. “So it has the potential to provide a parsimonious explanation as to why markers like proneurotensin may be of value in cardiovascular disease as well as breast cancer,” Dr. Jaffe says. This receptor also appears to be involved in the egress of fat in the cells, thus possibly tying it into the lipid levels and risk of developing atherosclerosis.

Moreover, he says, the breast cancer link likely has to do with proneurotensin’s origins in the hypothalamus, given the relationship between many neuroendocrine tumors and the pituitary gland.

“Now, all this is pretty speculative,” Dr. Jaffe cautions. “But it has been solidified by the data that came from the Malmö study.”

The proenkephalins do something different, “as best we know,” Dr. Jaffe continues. When opiates are high, the theory goes, they inhibit tumor progression and enhance normal killer cell activity. “If that’s the case, opiates would kill the cells, and improve prognosis, predominantly by inducing apoptosis.” It’s a provocative concept, he says.

Then the clinical utility issue rears its head: If all this is true, might these biomarkers be a good way to screen women? All of them? If not, what subsets?

“One of the tensions that’s going to exist is, How good are these markers? And what do you do about it?” asks Dr. Jaffe. If a neurotensin level is high in an asymptomatic patient, what’s the next step?

Quite possibly, as Dr. Maisel suggests, using proneurotensin and proenkephalin together might confer both high positive predictive value and also high negative predictive value. Should that be the case, says Dr. Jaffe, researchers will be in position to do the studies to answer the question, What now? “Maybe this will turn out to have the accuracy of BRCA,” Dr. Jaffe says. “That predictive accuracy is sufficiently high that people like Angelina Jolie are willing to make a decision based on it. These markers may or may not have that level of precision.”

Assuming the markers move forward toward clinical use, moreover, “There’s going to be a real issue in regard to specificity,” Dr. Jaffe says, reiterating the links to cardiovascular disease and diabetes. “We’ll eventually need to figure out the best way to triage specific elevations into the right bin.”

“I think what you’ve got,” Dr. Jaffe sums up, “are some very interesting hypotheses generating information from the Malmö study that need further exploration.”

Older breast biomarkers are having a new day in the sun as well. Like Atticus Finch, they may be worth fresh consideration even after decades of use, particularly as molecular analysis settles in alongside ER, PR, and HER2.

Dr. Hicks and his colleagues at Rochester have been using the modified Magee equations from the Department of Pathology, University of Pittsburgh Medical Center (http://path.upmc.edu/onlineTools/mageeequations.html), to obtain information that is similar to that provided by the Oncotype DX recurrence score. In a study of 283 cases, they found that information derived from the three new equations (which use traditional markers), combined with standard histopathologic and immunohistochemical variables, closely tracked the information provided by the Oncotype test (Turner BM, et al. Mod Pathol. 2015;28:921–931).

Dr. Hicks and his pathologist colleagues found the results impressive—“It did an amazing job,” he says—and they presented them at a weekly multidisciplinary breast conference devoted to CME and new information. The clinicians in the group thought the results were interesting, he says, but that didn’t mean they wanted to stop using Oncotype DX. They now receive information from the Magee equations prospectively, however, and Dr. Hicks says this has helped his colleagues more carefully select cases for Oncotype testing. “We’ve seen a decrease in the number of cases sent out.”
The IHC4 score is another example of using traditional markers in a relatively fresh way. It looks at ER, PR, HER2, and Ki-67 results to calculate a risk score using weighting factors and an algorithm. Like the Oncotype DX, says Dr. Hicks, it provides prognostic information.

Dr. Hicks also points to the PAM50, a quantitative RT-PCR assay for use on formalin-fixed, paraffin-embedded tissue. It uses a 50-gene set for standardizing the intrinsic subtype classification of breast cancers—luminal A, luminal B, HER2, and basal phenotypes—and provides a risk of recurrence score. There’s growing interest in androgen receptor (which has shown promise in triple-negative breast tumors), too, and in PARP inhibitors, which might bring new focus to BRCA mutations.

But from Dr. Hicks’ perspective, technical innovation has burst ahead of just about everything else. “I see the technology being way ahead of our ability to know how to use it,” says Dr. Hicks. (To that point, Dr. Fleisher notes that within the past several years, methods for measuring CTCs have exploded from about four to 25, though there remains only one FDA-approved method.)

Dr. Hicks notes that medical oncologists typically are a tough sell. When, early on, the Oncotype DX test showed that patients with a high recurrence score had a worse prognosis, the response, initially, was sort of a collective shrug: “So what?” as Dr. Hicks puts it. But when the test showed that patients would benefit from chemotherapy if they had a high recurrence score, but those with a low score would not, enthusiasm for using the test grew. “As soon as it guides their treatment decisions, then they get excited,” Dr. Hicks says. Only at that point, it would seem, will promising markers start making a noise everyone can hear.
[hr]

Karen Titus is CAP TODAY contributing editor and co-managing editor.