The promise of personalized medicine hinges on the ability of clinicians to understand and follow an individual patient's disease. Recent advances in methods used to isolate and characterize circulating tumor cells and free DNA may help move that promise a bit closer to reality.
The number of circulating tumor cells (CTCs) is both prognostic and predictive for patient outcomes in colorectal cancer, according to studies testing the FDA-approved Veridex CellSearch system. Patients who have more CTCs at baseline have a worse prognosis than patients with fewer CTCs. And patients whose CTC count either remains low or decreases during the first few weeks of chemotherapy have a longer median progression-free survival and overall survival. Similar results have been reported in other tumor types.
Still, the Veridex CTC system is not perfect. To improve on that, David T. Ting, MD, of Massachusetts General Hospital and colleagues developed an alternative method for isolating CTCs called the CTC chip.
Antibody Against Epithelial Cell Surface Marker
As Dr. Ting described at the most recent Gastrointestinal Cancers Symposium, both the Veridex system and the CTC chip use an antibody against the epithelial cell surface marker EpCAM to capture CTCs.
The Veridex system attaches the antibodies to magnetic beads, which can be mixed with blood from patients and then purified using a magnetic field. By contrast, the first iteration the CTC chip was a microfluidic device composed of a silicon chip with 80,000 microscopic posts etched onto its surface and coated with the anti-EpCAM antibody. As blood flows through the device, the antibodies grab hold of the CTCs between the posts.
The CTC yield was significantly higher with the chip device than with the Veridex system—about two to three orders of magnitude better capture and yield, Dr. Ting reported.
In studies using the Veridex system, researchers typically found 25% to 30% of patients to be positive for CTCs (defined as three or more CTCs per 7.5 mL of blood). With the CTC chip, 45 out of 107 patients (42%) with localized prostate cancer had CTCs, as did 74 of 116 patients (64%) with metastatic prostate cancer. (Dr. Ting did not report a head-to-head comparison of the two systems, however, so the comparison should be viewed with caution.)
The chip technology appears to be relatively flexible. The researchers can use other antibodies, including antibodies against HER2 and the epidermal growth factor receptor, to capture CTCs with the chip, which is important since not all CTCs are likely to express any one marker.
Amenable to FISH
And cells isolated with the chip are amenable to fluorescent in situ hybridization (FISH). In one study, Dr. Ting showed HER2 gene amplification in single CTCs using FISH. The researchers have also been able to examine Ki67 expression in isolated CTCs using immunohistochemistry.
In one study, they found that patients with castrate-sensitive prostate cancer had, on average, more CTCs than patients with castrate-resistant disease. However, a higher percentage of the CTCs isolated from patients with resistant disease were Ki67 positive, indicating that more of the circulating cells were proliferating. Remarkably, the team has even been able to isolate and sequence nucleic acids from these cells.
Still, Some Drawbacks
But the CTC chip design has drawbacks. It is opaque and thus doesn't allow morphological study of the cells. Also, it is difficult to remove the cells from the chip, limiting the experiments that can be done after isolation. And finally, the micropost design is difficult to manufacture, limiting the number of devices available.
So the team, led by Dan Haber, MD, PhD, Professor of Medicine at Harvard Medical School and Director of the Cancer Center at Massachusetts General Hospital, went back to the drawing board and came up with a second design.
This time, instead of microposts, the chip is comprised of silicone herringbone-shaped channels on a glass slide. As with the previous system, the herringbone CTC chip uses anti-EpCAM antibodies to grab onto the CTCs as they flow through the chamber.
There appear to be several advantages associated with the second-generation chip. A head-to-head comparison shows that the herringbone CTC chip is more efficient at isolating CTCs. In one test, the new chip yielded 65 CTCs from one milliliter of blood from a patient with metastatic cancer compared with only 4 CTCs with the old chip.
Additionally, the use of a glass slide allows for direct imaging of the cells. With that change, the team has found that CTCs sometimes travel in clusters of cells rather than individually. What that means from a biology standpoint, though, remains unclear.
Purification of nucleic acids is also more efficient from the new chip. Teaming up with Helicos BioSciences in Cambridge, the investigators have been able to perform digital gene expression analysis on purified CTCs. Early experiments show that the gene-expression pattern differs between CTCs isolated from pancreatic patients with a low number of CTCs and CTCs isolated from pancreatic cancer patients who have a large number of CTCs.
Dr. Ting said they had identified some of the individual genes whose expression was altered between the two patient groups and were evaluating whether they were suitable targets for therapy or research.
Finally because the cells are captured from freshly drawn blood, the CTCs remain viable. When the team adds cell culture media to the device, 95% of CTCs isolated from prostate cancer patients start growing within 12 hours, he said, noting that the group has been able to grow the cells for up to 24 days.
Unfortunately the team hasn't been able to find a way to use frozen samples in the herringbone CTC chip.
Therefore, the only center currently working with the technology is the one where it was developed. That will change by the end of the year, though. One stipulation of a Stand Up To Cancer grant that has supported the work is that the researchers need to make the technology available to other groups. Therefore, several major cancer centers will receive the herringbone CTC chip technology, including Memorial Sloan-Kettering Cancer Center, MD Anderson Cancer Center, and Dana-Farber Cancer Center.
Dr. Ting also noted the group's relatively new data showing that some CTCs travel in clusters, while others are single cells. He pointed out that that observation fits with a growing body of data suggesting that not all CTCs are equal.
The implications of that statement are substantial for the development and application of CTC technology, which has largely focused on how many cells an individual has. “Maybe CTC counts themselves are not necessarily the thing we should be focused on,” Dr. Ting said. “CTC subsets are probably more clinically relevant.”
The question, then, is how do researchers find and develop markers that show them which subsets are important. And that remains a big, open question.
Needed: Smarter Development Strategy
Not everyone is pleased with how the field is advancing, however. For example, Emile E. Voest, MD, PhD, from the Department of Medical Oncology at University Medical Center Utrecht, pointed out in an Education Session at this year's AACR Annual Meeting that the first paper reporting a correlation between CTCs and outcome was published back in 2004. Yet, when he checked the clinicaltrials.gov database in March of this year, he found that out of 316 trials that were incorporating CTCs, only one was a prospective test of their clinical utility—”kind of scary,” he said.
Dr. Voest estimated that based on a survey of some of the 316 trials, that on average they would plan to enroll 200 patients. If CTCs were tested at four different time points per study, and each CTC assay costs $50—a low estimate, he said—the cost of the CTC testing alone for the 316 trials would be $12.6 million. “Then we have a large amount of money that is invested just to measure this, and if we don't take it to the next level, that is going to be a waste of money,” he said.
There are basic questions that remain unanswered. For example, he said, there is no consensus on what constitutes a CTC or whether CTCs should be measured as a continuous variable or with a cutoff. Furthermore, health care payers are not convinced of their utility. He showed a policy from Cigna from this year stating that CTCs have not been proven to have a meaningful impact on health outcomes and that their role in patient management is unknown.
“Though I am a true believer of biomarker studies, this is a true statement,” Dr. Voest said. “We need to be aware of so many years of doing all these trials, and currently being at this point.” In his view, the field, including his own laboratory group, is doing too many descriptive discovery studies and not enough driving toward an end-goal of clinical implementation.
“Even at discovery we need to think about implementation,” he said. “We need to have hypothesis-driven prospective studies to move forward in any of the circulating cell fields.
“I cannot emphasize it enough: We need a strategy and prospective evaluation of all the biomarkers we are studying.” Pointing out again that the initial correlative report was published seven years ago and the current dearth of prospective trials, he concluded, “We need to do better than that.”
Using Circulating Free DNA to Personalize Treatment
While circulating tumor cells have been the subject of much discussion, a lesser known serum biomarker is circulating free DNA (cfDNA). Scientists know that DNA is released into the bloodstream when cells are damaged or die, and previous work has shown that the amount of cfDNA in a cancer patient's blood is proportional to their tumor burden.
Additionally, researchers have been able to sequence cfDNA or test it for particular mutations. But just exactly how to use this circulating biomarker has been a question.
In presentations earlier this year at the AACR Annual Meeting and the Gastrointestinal Cancers Symposium, Timothy Yap, MD, a research fellow in the Phase I Unit at Royal Marsden Hospital, described his group's effort to use mutations identified in cfDNA to assign patients in Phase I trials to particular drug treatments.
To determine whether this approach was feasible, the group first had to show that mutations detected in cfDNA are the same as mutations in DNA isolated from tumors — and that such an analysis could be done in real time.
With those goals in place, the team recruited 25 patients with advanced colon cancer, who had been referred to the Phase I clinic there between September 2009 and August 2010. For each patient enrolled, the researchers purified and quantified cfDNA and DNA from formalin-fixed paraffin-embedded tumor samples on day one.
On day two, the team amplified the nucleic acids by PCR. On day 3, they analyzed the DNA using the Sequenom Oncocarta mass spectrometry system to look for 238 known mutations in 18 genes commonly associated with cancer.
The team was able to collect and analyze cfDNA from all 25 patients and tumor DNA from 21 patients. Nine (43%) of the patient tumor samples carried a KRAS mutation, three (14%) carried a BRAF mutation, three (14%) had a PIK3CA mutation, and one (4%) had an AKT1 mutation.
Of the 25 cfDNA samples, nine (36%) were KRAS-mutant, three (12%) were BRAF-mutant, three (12%) were PIK3CA-mutant, and one (4%) was AKT1-mutant.
The concordance between cfDNA samples and tumor samples was good but not perfect. For the KRAS gene, 19 of 21 patients (90%) showed the same genotype in their cfDNA and tumor DNA.
There was 100% concordance for at the BRAF and AKT1 loci. The PIK3CA locus, however, showed substantial variability, with a total of five mutations discovered, two detected only in cfDNA, one in the tumor sample alone, one in both the tumor and cfDNA, and one in a liver metastasis but not in the primary tumor or cfDNA.
The team has subsequently found similar concordance between cfDNA and tumor DNA in breast cancer and melanoma patients. There was 95% concordance for PIK3CA status between cfDNA and the tumor sample in 21 breast cancer patients. Of 11 melanoma patients, there was 82% concordance for the BRAF locus between cfDNA and tumor, and 91% concordance for the NRAS locus.
Although the results were not in perfect agreement between the two DNA sources, serum and tumor, Dr. Yap said he thinks the data demonstrate the utility of cfDNA analysis in Phase I trials.
“We are using this technique in patients referred to the Drug Development Unit [the Phase I unit] at the Royal Marsden Hospital,” Dr. Yap said via e-mail. “I believe we are one of the first units, if not the first, to show data from patients who were screened prospectively.
“We haven't presented any response data in the public domain,” he continued, “but we are certainly cautiously optimistic that matching a patient with a targeted therapy that inhibits their molecular aberration will be of benefit to our patient population.”
Dr. Yap's assertion that cfDNA will have an impact in clinical care in the future was bolstered in May, when Myriad Genetics licensed proprietary cfDNA technology from Chronix Biomedical. Researchers from Chronix presented preliminary evidence at the 2010 ASCO Annual Meeting suggesting that cfDNA can be used for early detection of breast and prostate cancers with 92% sensitivity and 100% specificity.
Financial terms of the deal were not released, but Myriad says they plan to use the technology to develop early detection tests.