Researchers report that circulating tumor DNA in patients’ blood can be used to predict the likelihood of early-stage breast cancer recurring after apparently curative treatment. [Credit: C. Bickel / Science Translational Medicine]A highly accurate blood test, which fishes the blood of breast cancer patients for circulating tumor DNA fragments, appears to predict who will relapse, according to a new prospective study in Science Translational Medicine.

By probing, via massively parallel sequencing (MPS), patients’ tumors for DNA mutations pre-therapy, then looking for those mutations at many post-therapy time-points in circulating tumor DNA in blood (ctDNA), researchers found patients testing positive were 12 times likelier to relapse. Some feel this is a far more accurate way of predicting relapse in early cancer than even the promising analysis of circulating whole cancer cells. Others say the two approaches may complement each other.

“This is the first study to show we can potentially predict which patients are at risk of relapse using blood tests for tumor DNA,” senior author, and Institute of Cancer Research (ICR) medical oncologist, Nicholas Turner, Ph.D., told Drug Discovery & Development.

Harvard University medical oncologist Tilak Sundaresan, M.D., told Drug Discovery & Development the study was “compelling.” Sundaresan, uninvolved with the work, wrote accompanying commentary. “A number of other groups  have analyzed ctDNA in early-stage cancer,” he said. “But this study is unique in its combination of prospective and serial sampling in this population.”

University of Michigan Breast Oncology Program Clinical Director Daniel Hayes, M.D., called the results “exciting.” Also uninvolved with the new research, Hayes told Drug Discovery & Development a study in a 2013 New England Journal of Medicine “reported the first clinical results related to detection of ct-DNA. The current paper is, however, interesting in that they prospectively followed patients without metastases treated with neo-adjuvant chemotherapy, and then rendered disease-free with local therapy. While this paper does not establish monitoring of ctDNA for routine care, it does push the field out further.”

“The largest of its kind, this study strongly supports the potential for therapeutic advances by monitoring mutations using ctDNA,” Steffi Oesterreich, Ph.D., told Drug Discovery & Development. Uninvolved with the study, she leads the University of Pittsburgh’s Women’s Cancer Research Center. “This is critical. There is increasing evidence the genetic make-up of minimal residual disease, and not of the primary tumor, defines the lethal clone” driving relapse.

12 times more likely to relapse

Researchers with the ICR and the Biomedical Research Centre at The Royal Marsden took tumor and blood samples from 55 supposedly cured patients with early-stage breast cancer undergoing chemotherapy and surgery. The researchers analyzed blood after surgery and every six months thereafter. The result: the team predicted relapses with surprising accuracy.

Patients testing positive for ctDNA were 12 times more likely to relapse than those without. The return of their cancers was detected an average of 7.9 months before scans did.  

The teams pulled this off via a personalized digital PCR test called “droplet-digital PCR,” designed to pick up, in blood, mutation-ridden DNA bits shed from tumors. They did not look for all mutations found solely in each patient’s tumors, but mutations common to their cancers and others.

The teams also demonstrated the way new DNA mutations can form and accumulate in developing cancer, making it clear recurrence needs to be detected early.

Larger clinical trials  of this cost-effective approach start next year.

Minute DNA bits bring major insights

Turner told Drug Discovery & Development one disappointment was that his teams couldn’t detect relapse in women “at risk of developing relapse only in the brain. This is probably because the blood/brain barrier blocks release of DNA from cancer cells into the blood.”

He noted, however, that detecting relapse via DNA is likely more accurate than via whole cancer cells in blood as “there is relatively more free DNA compared to cells, and that allows more sensitive assays. The main technology development is digital PCR, which allows us with high confidence and accuracy to detect very rare mutations in the blood.”

READ MORE: Advance in Capturing Elusive Circulating Tumor Cells

Nicholas Turner, Ph.D,, Institute of Cancer Research (Credit: ICR)He added that a key challenge for oncologists has long been “working out who is at risk of developing future secondary cancer after they have been treated for cancer in the breast. If we can identify which women are at risk of relapse, we can then focus further treatments on them to try and prevent relapse. In this study, we have shown we can potentially predict who is at risk of relapse by detecting minute amounts of tumor DNA in the blood of women after surgery. Those women who had tumor DNA detected in blood after surgery and chemotherapy had a high risk of future relapse.”

Turner said a simple blood test post-treatment “could identify which patients are at risk of relapse. The hope is we can use this blood test to direct further treatments to prevent or substantially defer relapse. We are at an early stage in this research, and further studies are required before the test could enter the clinic.”

In clinic at last

Sundaresan told Drug Discovery & Development the paper has “three particularly compelling parts.” First, despite substantial interest in ctDNA, there has been a relative dearth of studies translating that interest to pressing clinical questions. The new paper does this “nicely, by beginning to answer the question of whether the detection of ctDNA can be used as a predictor of relapse in women treated with curative intent for localized breast cancer.”

Second, Sundarensan said, the study expands on other work showing “feasibility of detecting ctDNA in localized cancer, whereas the majority of attention has  been in the metastatic setting, where ctDNA tends to be more plentiful.”

Serial ctDNA monitoring

Finally, he said, Turner’s teams show that “monitoring ctDNA serially, over multiple time-points, may be superior to a single time-point at judging risk of relapse. This may have an impact on how future studies of ctDNA are designed… Overall, given the high likelihood of relapse when ctDNA is detected, these findings lay the groundwork for future clinical studies that can be designed to test the intensification of adjuvant therapy options in women with detectable ctDNA at the completion of definitive treatment for non-metastatic disease. This may be complementary to other prognostic features , such as residual cancer following neoadjuvant chemotherapy in the resection specimen.”

Future studies will “include much larger numbers of women with different types of non-metastatic breast cancer—for example, women who receive primary surgery without neoadjuvant chemotherapy—and follow them for longer periods of time.” It will likely be, he said, decades before researchers see how well ctDNA predicts relapse, and “before we can be confident in clinical utility.”  

More fully personalizing screens so each patient’s unique tumor signature is represented “would have been better,” he said. This may occur later. “Their initially limited mutation panel also meant that 20 percent of the women studied did not have mutations that could be used to design their ctDNA detection assay, so they were excluded from analysis. Using a more comprehensive approach would prevent that exclusion.”

For best predictions: both ctDNA and CTC?

However, Hayes said, “As exciting as these findings are, many of us believe that CTC (whole circulating tumor cell) and ctDNA may be complementary.” Hayes is a named inventor on a CTC patent licensed to Janssen. He receives royalties, he told Drug Discovery & Development.

“While the latter may be more sensitive—as, or more, specific—for detection and monitoring of cancer than CTC, that remains to be proven. Further, one cannot phenotype [analyze cell surface markers associated with] ctDNA, and therefore you can only detect and monitor `fixed’ genetic defects. Since cancer cells are remarkably plastic, the ability to phenotype CTC over time, reflecting response to environmental changes such as therapy, may provide information ctDNA cannot.”

Further, Hayes said, at least for now, ctDNA offers no insight into the heterogeneity of tumors in each patient, “whereas analysis of individual CTC does.

CtDNA from dead tumor cells

Finally, Hayes told Drug Discovery & Development, “Although I cannot document this, one wonders if cell-free tumor DNA may not be the same as cell-bound DNA. Cell-free DNA must come from dead cells, unless cancer cells secrete DNA. But if so, I’m unaware of it. Therefore, cell-free DNA may or may not represent the living, viable tumor cells causing the clinical problem. In contrast, CTC represent a broad spectrum of viability, depending on how they are captured and evaluated.”

Hayes noted the field will be “evolving as time goes on. Much needs to be done. But in the long run, I foresee monitoring both ctDNA and CTC phenotypes, and even genotypes.”

Oesterreich concluded the “well-designed and executed” study involved a small group of patients representing “a wide range of early-stage disease.” Future priorities should include, she said: deciding if mutation losses seen were the result of detection flaws; hiking linear DNA amplification yield; defining methods for locating copy number/structural variations; and improving assay robustness, MPS hybrid capture, ctDNA stability, and processing plasma extraction processing speed.