Circulating tumor cell
  tests boon or bust?

January 2005
Cover Story

William Check, PhD

In Fyodor Dostoyevsky’s classic novel The Brothers Karamazov, there is a point at which Ivan Karamazov, suffering from “brain fever,” hallucinates that he is conversing with the devil in the form of a down-at-heels gentleman. The gentleman-devil is complaining of his ailments: “I’ve tried many doctors; they diagnose beautifully, they have the whole of your disease at their fingertips, but they’ve no idea how to cure you. There was an enthusiastic little student: ’You may die,’ he said, ’but you’ll know what disease you are dying of!’”

Although medicine has advanced a long way in the 124 years since Dostoyevsky wrote his novel, it is still plagued by the fact that diagnosis often outruns therapeutics. This flaw particularly bedevils tumor marker assays, including a recent version—detecting circulating tumor cells in the blood. Being able to detect such cells would be very useful, says Karen L. Kaul, MD, PhD, director of molecular diagnostics at Evanston (Ill.) Northwestern Heathcare and professor of pathology and urology at the Northwestern University Feinberg School of Medicine. “We could potentially use it for screening, staging, prognosis, detecting relapse, monitoring therapy, finding contamination of stem cell harvests, and evaluating lymph nodes and surgical margins,” Dr. Kaul says. In her view, detecting circulating tumor cells in the blood would be “a natural extension of what we do in surgical pathology.”

Interest in detecting circulating tumor cells, or CTCs, grew strong this summer with the publication of an article showing that breast cancer patients with measurable metastatic disease who had CTCs had a significantly worse prognosis (Cristofanilli M, et al. N Engl J Med. 2004;351:781-791). Based on the data from this trial, the Food and Drug Administration cleared the assay, called CellSearch Epithelial Cell Kit, used in the trial. “It’s exciting to see these tools used in well-done clinical studies,” Dr. Kaul says. Nonetheless, she believes this method is not yet ready for routine clinical use. “Does showing correlation with outcome mean we are ready to go live with these assays?” she asks. “Probably not. We don’t know enough about, and the clinical relevance of, circulating tumor cells to know what to tell patients about the results.”

Other molecular pathologists express similar views. “These results are a step in the right direction,” says Wayne W. Grody, MD, PhD, professor in the Divisions of Medical Genetics and Molecular Pathology in the Departments of Pathology and Laboratory Medicine and of Pediatrics and Human Genetics, UCLA School of Medicine. “I still would like to see more than one paper demonstrating efficacy before we brought this into clinical use.”

One of the responses that Kevin C. Halling, MD, PhD, gets from clinicians when he talks to them about this type of assay is, How are we going to use the information? “Say a patient has circulating breast cancer cells,” says Dr. Halling, co-director of the clinical molecular genetics laboratory, Department of Laboratory Medicine and Pathology, Mayo Clinic. “Will you be able to do something about it? Or is it just bad news for the patient? The big question for assays that detect CTCs is whether they will change clinical management. People who try to convince you this is a useful assay will tell you that it helps make more-informed decisions about treatment options. They may be right, but we need more studies to know for sure.”

More-informed decisions are precisely what Massimo Cristofanilli, MD, who was first author on the article reporting results with the CellSearch assay, says the test makes possible. “This test provides an extremely important new prognostic factor,” says Dr. Cristofanilli, associate professor in the Department of Breast Medical Oncology, University of Texas M.D. Anderson Cancer Center. He is using the test now in all patients with metastatic breast disease “to be more appropriate in my estimate of prognosis and my treatment plan.” In his view, all women with metastatic breast cancer about to start a new line of therapy should be offered the test. “We already order tumor markers with much less clinical value,” he says, “so why not use this test also?”

Daniel F. Hayes, MD, was also an investigator in the CellSearch trial. Dr. Hayes, who is clinical director of the breast oncology program at the University of Michigan Comprehensive Cancer Center, has worked for many years on breast tumor markers and is one of the discoverers of CA 15-3. “I am, frankly, intensely proud of this trial,” says Dr. Hayes. “Unlike almost any other tumor marker study, this was a prospective study with carefully defined endpoints and followup times. So unlike most tumor markers studies, it is very likely these data are correct.” Asked whether the assay is appropriate for all women with metastatic breast cancer, he answers, “With the qualifier ’every,’ the answer is ’no.’ I would use it in selected patients.” Then he adds: “This test does not replace the human brain. It doesn’t direct therapy, but it can help.”

Why such varied interpretations of the data? As usual with such a complex issue, the devil is in the details.

Looking for the source of metastases in the bloodstream reflects a view of cancer spread that derives from the “seed and soil” hypothesis of 19th-century physician-scientist Steven Paget, who postulated that tumor cells must be adapted to a particular tissue to land, take root, and grow there. Contemporary science has verified that specific genetic alterations are required for secondary growth, so that not all tumor cells in the blood are capable of forming a metastasis, which is one possible source of false-positive results.

It is also possible that the presence of tumor cells in the blood is an “episodic” event, says Dr. Grody. “I have a sense that it is not a constant, that you don’t always have tumor cells in the blood, even when you have metastases,” he says. He speculates that, as a tumor grows, it can break through a major vessel and spill “seeds.” Then it may be contained again. “So that’s one reason why I think that circulating tumor cell assays are not yet ready for prime time,” he says.

Another source of misleading results, Dr. Kaul says, is that “unfortunately, most markers are tissue-specific, not tumor-specific.” PSA, tyrosinase, cytokeratins, HER2, and mammaglobin are all examples. “Early research fell on its face because of inadequate assays,” she says. “There were a lot of poorly done studies that did not define sensitivity and specificity.”“

Dr. Kaul comes by her caution honestly—for the past decade she has been following up a report on the value of molecularly identifying circulating cancer cells to stage localized prostate cancer preoperatively, a report that turned out to be, like Ivan Karamazov’s vision, less than solid. “We got into this in 1994,” she says, “naively thinking we were validating a test that was moving from research to clinical application in short order. We are now in our fifth-generation assay and still in research mode.”

In the original paper, RT-PCR was used to identify circulating cells that made mRNA for PSA. The authors interpreted their data as showing that patients who had such cells would have positive margins after radical prostatectomy, and so shouldn’t have surgery, while those without circulating cells would have negative margins and could be cured (Katz AE, et al. Urology. 1994;43:765-775). “Patients jumped all over this,” Dr. Kaul recalls. Many called urologists and asked how to get the test done. “Patients believed they would no longer have to flip coins whether to have surgery or go straight to radiation oncology,” Dr. Kaul says. One of the urologists at Evanston Northwestern Healthcare came to her for help in implementing an assay and validating these results. In their prospective study, they found positive results about equally often in margin-negative and margin-positive patients. “We concluded that at this level you couldn’t molecularly stage prostate cancer,” Dr. Kaul says (Ignatoff JM, et al. J Urol. 1997;158:1870-1874). Several other laboratories reached the same conclusion. Subsequently, the original investigators were unable to validate their results and softened their claims.

In the meantime, the test was licensed to a reference laboratory, which later had to stop offering it, but not before many patients had the test done. “I don’t know that we were doing a service” by offering the test so quickly, Dr. Kaul says. “Patients could be devastated by a finding of circulating tumor cells—and we don’t know its meaning.”

Dr. Hayes is well aware of the sins committed in the tumor marker field. In what he calls his “diatribe” on this topic, he says, “Tumor marker studies are often not done well. I’ve been responsible for some of these myself.” One source of the problem: “FDA has much stricter criteria for new drugs than new markers.” Tumor marker studies are often retrospective, using archived sera—“whatever you’ve got in the freezer,” as he puts it.

Dr. Hayes and his colleagues set out to design the CellSearch trial like a new drug study, prospectively done with defined clinical and radiographic examinations. The cutoff for positivity—5 CTCs/7.5 mL of blood—was derived from a training set of 100 samples and verified in an independent validation set of 77 samples. To reduce the effect of non-specificity of the tumor marker used, CK19, immunomagnetic separation was done to enrich for epithelial cells. Patients represented a broad cross-section of those with metastatic disease about to start a new line of therapy, ranging from women with new estrogen receptor-positive recurrences for which they would get hormone therapy to those scheduled for third-line chemotherapy, which might be their last treatment.

One significant finding was that the 50 percent of women who had elevated CTCs at baseline (pre-therapy) had worse overall and disease-free survival. “To me as a clinician, that is not so important,” Dr. Hayes says. “I don’t think I am going to select initial therapy from knowing whether a woman has a worse prognosis. Unfortunately, metastatic breast cancer patients almost all die of their disease, so I am trying to choose a therapy most likely to palliate that patient. And baseline CTCs don’t tell me which is most likely to work.”

A more important finding clinically was that the CTC result at first followup—three to four weeks after initiating therapy—was equally prognostic. Dr. Hayes explains how he would use the followup CTC value in practice. “Each time I see the patient after starting therapy I ask, Are we accomplishing the goal of palliation—making the tumor shrink with fewest side effects?” If the cancer is obviously progressing, or the patient has intolerable toxicity, or the tumor is responding well, his next step is clear. But in about half of patients the response to therapy is unclear. Clinical examinations and radiography—bone scan, CT, and MRI—sometimes clarify the situation. However, Dr. Hayes says, scans are time-consuming and expensive. A patient can spend a day in radiology. In addition, “Usually it doesn’t give you an indication of where you are for two to three months. And although they are the gold standard, radiologic scans are not terribly accurate. So after three cycles I often find myself saying, ’Mrs. Jones, I just can’t figure out your scans. Let’s go another three cycles and see what happens.’”

Classic tumor markers—CEA, CA 15-3, CA 27.29—are “reasonably good,” he notes, at aiding decision-making. But very early in the patient’s course they frequently go up before they go down. This tumor marker spike or flare can last two to eight weeks. “So maybe 25 percent of the time we see a false-positive spike,” Dr. Hayes says.

It is here that the CTC assay can help—in patients on their third or fourth cycles of treatment with indeterminate responses by standard evaluations in whom the clinician would like to try a few more cycles of the same therapy. “If she has elevated CTCs, you are probably kidding yourself and you should change therapy,” Dr. Hayes says. “You still have to be a physician,” he emphasizes. “For patients with a clear clinical course, you do what’s reasonable, regardless of what her cells are. But if you are on the fence, you can use the test to help make the decision.

”We hoped that measuring CTCs would be as good as and cheaper than radiographs,“ Dr. Hayes continues. Unpublished data from the trial suggest that it may be a better predictor of survival than classical measures, though Dr. Hayes would like to see that tested prospectively. As for relative price, a bone scan costs $700 to $800 and a CT scan $1,000 to $1,500. Quest Diagnostics, which has licensed the test, charges $600. Of course, saving money requires that physicians substitute the CTC assay for radiography, rather than add it. But will they? ”We can’t answer that question at present,“ Dr. Hayes says. ”Only time and data will tell.“

Dr. Cristofanilli points out other important outcomes of the trial. First, CTCs were prognostic only for patients beginning chemotherapy, not those on hormone therapy alone, possibly because response to hormone therapy is slower. (Dr. Hayes points out that the assay was evaluated in various subgroups, such as those starting a new hormone therapy, in unplanned and retrospective analyses. The assay’s performance in such subgroups should be prospectively tested, he says, in well-designed and properly powered future clinical trials.) Second, the percentage of patients with CTCs dropped from 49 percent at baseline to 28 percent at first followup, correlating with the benefit of treatment. And those who went from positive to negative CTCs at three to four weeks approached the prognosis of patients without cells at baseline. ”So we could use this assay as a measure of treatment efficacy in patients with metastatic disease,“ he says.

Clinically, Dr. Cristofanilli is using the baseline CTC result to influence initial therapy for women with metastatic disease about to start a new line of treatment. He gives two examples. If a patient with estrogen receptor-positive disease has CTCs, he would probably start with chemotherapy rather than hormone therapy. And for a patient with estrogen receptor-negative minimal disease with no CTCs, he would use single-agent chemotherapy instead of a combination.

Outside the university setting, in routine clinical practice, Dr. Cristofanilli suggests, the CTC assay can be used to help patients and physicians make good decisions. “Not necessarily treatment decisions,” he cautions, “because investigations should come first.” That is, studies need to be done to see whether basing treatment on the CTC assay will improve outcomes.

Dr. Hayes agrees: “What we don’t know in patients with CTCs is whether changing therapy will do them any good. Are CTCs just detecting those who are not going to respond to anything?” He suggests that the fact that 20 percent of patients converted from CTC-positive to CTC-negative by first followup indicates that some patients with elevated cells respond to therapy. But that doesn’t say whether therapy will help patients who still have elevated cells at followup.

To address this question, Dr. Hayes and his colleagues in the Southwest Oncology Group are designing a randomized trial in which they hope to prospectively test the value of changing to an alternative chemotherapy versus staying on the originally chosen first-line chemotherapy in patients with metastatic disease. The precise trial design is not set, but they hope to open this validation study this year.

CellSearch Epithelial Cell Kit is marketed by Veridex, a subsidiary of Johnson & Johnson. “We think this is a major step forward in being able to measure the biological activity of tumor cells,” says Mark Myslinski, general manager of Veridex. “This can help the physician and patient make a decision as to what course of treatment to pursue.” The sample is preserved in a CellSave tube for up to 72 hours. To run the assay, a laboratory must purchase AutoPrep, which prepares cells, and CellSpotter Analyzer, which visualizes fluorescently tagged cells. These instruments come from a separate company, Immunicon, and cost $120,000.

The automated system developed by Immunicon scientists uses antibody-coated ferrous particles to separate EpCAM-expressing epithelial cells from whole blood specimens. Confirmation that these are indeed epithelial cells is provided by automated fluorescent microscopy after staining with fluorescently labeled monoclonal antibodies against cytokeratins and CD45 (a pan-leukocyte marker) as well as DAPI. A trained observer reviews computer-generated composite images of the detected events. Cytokeratin-positive, DAPI-positive events are tallied by the computer as epithelial cells and expressed as “CTC/mL of blood collected” (usually 7.5 mL into a 10-mL Vacutainer), while “events” that fail to stain for cytokeratin or DAPI, or that stain for CD45, are not. A proprietary fixative has been identified that makes it possible to mail samples to central laboratories for processing and assay within 48 to 72 hours.

Quest Diagnostics decided to offer the assay because it wanted to “provide patient care at the highest level possible,” says Gary Milburn, PhD, national director of genomics and esoteric testing services at Quest, Chantilly, Va. “We felt that this assay, having been FDA cleared for breast cancer recurrence, gives us an opportunity to impact patient care.” Dr. Milburn predicts clinicians will use it as a more sensitive alternative to traditional followup, such as physical examination and bone scan. “Obviously more studies need to be done to demonstrate what to do with patients who have increased CTCs,” Dr. Milburn says. “But from my discussions with physicians, they seem willing to react exactly as if a patient had an additional lump or an indication from bone scan that there may be recurrence.” Dr. Milburn heads a team of a dozen professionals whose job is to give talks at hospitals to help educate physicians about Quest’s new tests.

Dr. Kaul takes issue with the interpretation of the FDA’s approval wording—“for the enumeration of circulating tumor cells of epithelial origin . . . in whole blood”—as a clinical indication. “It is no more clear how to use it than tests that are not FDA approved,” she says. “It is excellent that we are getting reagents in a standardized format. But until we have more information about the clinical utility of these assays, I don’t think we should be offering them for clinical care. Until we have proof that patients should be treated differently based on CTCs, I don’t think we should be using this assay for clinical decision-making.”

Dr. Kaul’s comments are consonant with those of the authors of the editorial that accompanied the CellSearch article, Stephan Braun, MD, and Christian Marth, MD, both of Innsbruck Medical University, Austria. They wrote: “[A] cautionary note is warranted with respect to uncritical, immediate adoption of this assay for routine use. . . . We still need to know, for example, whether any change in the treatment based on the number of circulating tumor cells alone will translate into a benefit in progression-free survival” (N Engl J Med. 2004;351: 824-826). On the positive side, they said the data “allow us to speculate that this assay will soon realize its potential to change the standard of care for patients with metastatic breast cancer.”

Other assays for CTCs are available for research use only. One from Abbott/Vysis uses real-time PCR to look for mRNA for markers of breast cancer plus β-2-microglobulin, an internal control. Abbott holds intellectual property rights to the two breast-specific markers, BU101 and BS106, as well as proprietary primers for CK19, which were designed to avoid amplification of any of the identified pseudogenes, says Timothy Stenzel, MD, PhD, medical director of Abbott Molecular Diagnostics, who, before moving to Abbott, worked on these markers as part of a grant that Abbott and Duke University received from the NCI-funded Early Detection Research Network. In collaboration with Dr. Kaul, lymph nodes were tested using BU101, CK19, and mammaglobin, another breast-specific marker. And in collaboration with Lyndsay Harris, MD, of Dana Farber Cancer Institute, the assays were used to detect CTCs in HER2-positive breast cancer patients. Data have been submitted for publication.

“We are at a preliminary stage,” Dr. Stenzel says. “All our analytical data are very good. Now the question is whether it has clinical utility. That has not yet been established. We are looking to develop collaborations with individuals who will help us show the clinical utility of our assays. We hope to make our current assay available for clinical purposes in the not-too-distant future.” Dr. Stenzel sees the first application as monitoring response to therapy in advanced or metastatic patients.

Another application would be to see if detecting tumor cells in lymph nodes as well as blood or bone marrow is a more accurate way of staging than standard pathological methods. Dr. Stenzel notes that 20 percent to 30 percent of node-negative women recur even though it looks like they have been cured. “There must have been cancer cells remaining,” he says, “which is the rationale for giving adjuvant chemotherapy to some node-negative women.” Dr. Stenzel suggests that histological examination of lymph nodes with H&E may not be sufficient to identify patients who will recur. “We need to find out whether molecular detection will be more useful.”

ChromaVision offers a research-use-only kit for detecting CTCs using immunomagnetic enrichment combined with an immunocytochemical assay for CK8, 18, and 19. Visualization is done by image analysis with an automated digital microscopy system and review by a pathologist, which introduces a professional component. Early data have been published (Witzig TE, et al. Clin Cancer Res. 2002;8:1085-1091).

In newly diagnosed breast cancer, cytokeratin 19-positive cells visualized in bone marrow by immunohistochemistry are a highly prognostic finding, according to Dr. Braun and his colleagues in Innsbruck. In a prospective study with 552 patients, they showed that the presence of occult metastatic cells in bone marrow of patients with stage I, II, or III disease was an independent prognostic indicator of the risk of death from cancer (Braun S, et al. N Engl J Med. 2000;342:525-533). Only one percent of 191 patients with nonmalignant disease had CK19-positive cells in bone marrow, compared with 36 percent of breast cancer patients. Most important for staging purposes, this finding was highly significant among 301 women without lymph-node metastases at primary diagnosis: 14 of the 100 with CK19-positive cells in bone marrow died of cancer-related causes, but only two of the 201 without such cells. Unfortunately, it is difficult to apply this finding, since bone marrow sampling is not a routine part of the clinical workup of breast cancer patients.

For just over a year a qRT-PCR assay for CTCs in the blood of people with colorectal cancer has been available from Targeted Diagnostics & Therapeutics Inc., of West Chester, Pa., a company formed in 1994 specifically to license the guanylyl cyclase C (GCC) marker from Thomas Jefferson University, where it was discovered by Scott Waldman, MD, PhD, the Samuel M. V. Hamilton professor of medicine and professor of biochemistry and molecular pharmacology and director of the Division of Clinical Pharmacology. “GCC is very specific for the colon,” says Daniel J. O’Shannessy, PhD, chief operating officer of TDT. Colonic cancer cells continue to express the protein and its mRNA. So any GCC-positive cells found in the blood automatically represent potential metastatic disease. Dr. O’Shannessy estimates the sensitivity and specificity of GCC as both greater than 95 percent and superior to that of CEA or CK20, other commonly used markers for CRC. Because of GCC’s high specificity, no enrichment step is necessary, according to Dr. O’Shannessy.

In a small (21 samples) retrospective study, GCC mRNA was expressed in lymph nodes from all stage II colorectal cancer patients with recurrent disease at five years but not in those from patients without recurrent disease (Cagir B, et al. Ann Intern Med. 1999;131:805-812). Dr. O’Shannessy notes that 35 to 40 percent of patients with Dukes B (nonmetastatic) colorectal cancer have recurrence by three to five years and die of their disease. Use of GCC may provide more definitive staging in these patients, he suggests. Dr. Waldman has received a five-year grant from the National Cancer Institute to compare staging with and without the GCC qRT-PCR assay. The NCI has also awarded Dr. Waldman a five-year grant prospectively to monitor therapy in 2,000 to 3,000 colorectal cancer patients using the GCC blood test.

Looking ahead to future developments

Markers have been used not only to identify CTCs, but also to detect free circulating nucleic acids—both DNA and RNA—in serum and plasma, as well as sputum, stool, and nipple aspirates. “To me it’s more of a curiosity,” says Dr. Kaul. “It amazes me that you can find it there, particularly RNA.” However, free nucleic acid detection is an area of active research and one that she predicts “we will be hearing more about.” (Related article: Detecting free DNA)

Dr. Hayes sees improvement in the value of CTC assays coming from methods that do “not just counting, but actually phenotype and genotype cells.” He points to the “tail” on the survival curves of patients who had CTCs in the CellSearch study. “Between 10 to 15 percent of patients who have elevated cells still did well,” he says. “So they have a ’cancer’ cell that may not be so bad to have.” Many researchers have been trying to genotype and phenotype cells, not just circulating but also in the primary tumor and in bone marrow, to find those that actually predict metastases.

Dr. Kaul also stresses the need for new markers to make these assays more clinically useful. “Most markers we use now are rooted in decades-old immunohistochemistry studies done in surgical pathology,” she says. In particular, markers are needed that reflect progression, since “most tumor cells in the circulation are probably not capable of forming metastases.” As an example of one way to find new, more specific markers, she points to work using serial analysis of gene expression to identify genes that were expressed in breast cancer but absent in the expression profiles of blood and bone marrow cells. Four markers were selected that identified 29 percent of samples from patients with breast cancer who had minimal residual disease, compared with none in non-breast cancer patients (Bosma AJ, et al. Clin Canc Res. 2002;8:1871-1877). The ultimate assay, Dr. Kaul predicts, will use new multimarkers to identify cells and their biological relevance.

William Check is a medical writer in Wilmette, Ill.