NEW YORK (GenomeWeb) – A team led by researchers from the Lund University Cancer Center in Sweden have demonstrated that combining sequencing and digital PCR to analyze cell-free circulating tumor DNA (ctDNA) could be a viable way to detect metastatic disease in primary breast cancer patients months before a symptom-based diagnosis might be possible.
The investigators, led by Lund University's Lao Saal, published their work this week in EMBO Molecular Medicine. In the retrospective study, the researchers performed low-pass Illumina sequencing on patient blood samples to identify tumor-specific chromosomal rearrangements in ctDNA, followed by Bio-Rad Droplet Digital PCR assays for the low-frequency mutations.
Using this workflow, they were able to identify 93 percent of the patients who went on to develop metastatic disease and 100 percent of the patients who had long-term disease-free survival. What's more, in 86 percent of patients, ctDNA-based detection preceded clinical detection by an average of about 11 months.
The results are promising enough that the group has begun larger retrospective and prospective studies to validate its findings, Saal told GenomeWeb this week.
"In early breast cancer, there haven't been any studies showing that you can pick up recurrences," Saal said. "Ours shows that we can do it pretty accurately, and we also had a pretty good lead time advantage over clinical detection. In a couple of patients, we had detection about three years before [clinical diagnosis]. We think this kind of early detection could be meaningful with further validation studies to be able to detect recurrence … before you have a clinical metastasis, and it's almost impossible to cure that patient."
Saal said his lab has taken an interest in the potential of analyzing ctDNA for early cancer diagnosis since the first papers were published on the subject several years ago. Further, his group had access to biobanked breast cancer tumor and plasma samples spanning several years — "so we knew how these patients did, whether they had recurrences or not after primary diagnosis."
For their pilot study, Saal and colleagues analyzed tumor samples and corresponding plasma samples from 20 breast cancer patients who had been enrolled in the Breast Cancer and Blood study, a prospective study that has been ongoing at Lund University since 2002. They analyzed a total of 93 blood samples at a variety of time points from these patients.
The researchers adopted a similar approach to the personalized analysis of rearranged ends (PARE) method developed in the laboratory of Victor Velculescu and colleagues at Johns Hopkins University.
"Basically, you first sequence the primary tumor, so we did low-pass, whole-genome sequencing using [the] Illumina method, with some small tweaks to increase the fragment size," Saal said. "We did pretty low coverage because we just wanted to find the rearrangements — we didn't need to find all the point mutations."
Then the group developed software, dubbed SplitSeq, to better identify the rearrangements and reconstruct the exact fusion sequences. "The ctDNA is really fragmented, so if you want to find an assay for something that's less than 150 bases, you really need to have that exact sequence to design against," Saal explained.
Finally, they designed PCR assays against the ctDNA chromosomal rearrangements and ran them on a Bio-Rad QX100 Droplet Digital PCR instrument.
"We thought about qPCR, and some of the other digital PCR methods, and we also thought about doing some kind of direct sequencing of plasma DNA, whether whole-genome or targeted sequencing," Saal said. "For us, the Bio-Rad ddPCR solution was attractive because it seems to have very good performance characteristics. What we liked about it over qPCR, for example, was that there wasn’t really any need for standards. It's very quantitative, and [provides] really good estimates of actual molecule counts in a sample."
For the 20 patients analyzed, 14 of them had an eventual metastasis event less than six years after primary diagnosis, and six of them exhibited long-term disease-free survival. Using their method, Saal and colleagues were able to identify 13 of the 14 metastatic patients from their blood samples. They also did not detect any ctDNA following surgery in the six disease-free patients.
The absolute quantification capabilities of droplet digital PCR also turned out to be important. In addition to early detection, the actual quantity of ctDNA seemed to be predictive of both metastasis and patient death. "Essentially, we did an analysis where we determined that for every doubling of the ctDNA quantity, there was a very significant increase in the odds of metastasis, an odds ratio of about 2. For death, the odds ratio was about 1.3 for every doubling of ctDNA."
The study did have limitations, first and foremost, the small number of patients studied. As such, the researchers are already working on larger retrospective studies to validate their findings, Saal said.
Furthermore, the researchers noted, the availability and quantity of archival frozen plasma, as well as the specific time points of their collection, were limiting factors. For their study, they isolated cell-free DNA from 0.5 ml of plasma, and 4 percent of this was input per replicated ddPCR assay. With these parameters, they estimated their method to be sensitive enough to detect an amplifiable target DNA molecule in 40 µl of plasma.
"The sensitivity of our method to detect exceedingly low counts of target ctDNA could be improved linearly by increasing the amount of input DNA into ddPCRs, by multiplexing, by preamplification, and/or by isolating circulating DNA from a larger volume of plasma," the researchers wrote in their paper.
In the prospective setting, they noted, they would be able to better control the blood plasma collection procedures and time points, and take larger-volume samples. "Therefore, the sensitivity and apparent lead time advantage for occult metastasis detection reported herein may in fact be an underestimation," they noted.
As such, Saal and colleagues have also initiated one prospective study starting with patients being treated in the neo-adjuvant setting. "There is a window where you can administer chemotherapy to try and shrink the tumor," Saal said. "So we … [will] follow these patients during that neo-adjuvant time period. And then even after the surgery we can continue to follow them for many years."
Finally, Saal noted that "chromosomal rearrangements are really good biomarkers because basically all cancer patients have some … but looking at point mutations could also be very useful, so we have other projects focusing on those."
For those projects, the group has developed a method to improve detection of such mutations using ddPCR by "making some fundamental changes to the way it's done to really improve sensitivity and specificity," he said. It has been widely reported that at very low mutant allele fractions, false positives can be a problem with droplet-based digital PCR.
"We have an innovation where we can drastically reduce those false positives, so you can detect lower percentages of ctDNA," Saal said, noting that this work has not yet been published and that his group has filed for patents on the technique.