Genotyping cell-free DNA and circulating tumor cells can provide a useful measure of cancer treatment response up to 16 weeks prior to standard clinical measures, according to a study from a group at the Dana-Farber Cancer Institute.
The study, published last month in Clinical Cancer Research, measured cfDNA in blood plasma from lung cancer patients receiving erlotinib chemotherapy, and was able to detect drug resistance mediated by EGFR T790M mutations in non-responders weeks before radiographic progression.
The work was notable for its use of droplet digital PCR. "Droplet digital PCR is an attractive technology as its speed, cost, and ease of use is similar to other PCR-based assays, yet the sensitivity and quantitative nature of this assay offers broader clinical application," the authors wrote.
The researchers first perfected ddPCR assays measuring EGRF, KRAS, and BRAF mutations in plasma from patients with lung cancer or melanoma of known cancer genotypes. Next, they ascertained reference ranges for levels of these mutations in cfDNA and set thresholds, honing specificity and sensitivity by using non-overlapping genotypes and gold standard samples as positive and negative controls. Then, they assessed serially collected plasma samples in a set of four cases with disease progression or favorable response to treatment, showing a correlation with cfDNA genotyping results. Finally, they assessed response and resistance in a prospective sample of nine EGFR-mutant lung cancer patients in a trial of erlotinib, marketed by Roche as Tarceva.
As leader of blood biomarker development at the translational research laboratory of DFCI's Belfer Institute for Applied Cancer Science, Cloud Paweletz provided technical and analytical leadership, while DFCI's Geoff Oxnard, also a professor of oncology at Harvard, led the clinical side. In an email to PCR Insider Paweletz claimed the group's study is "the first to use plasma genotyping to identify a validated and targetable marker of resistance."
The group used Bio-Rad's QX100 ddPCR platform to quantify mutations, choosing this system over competitors like RainDance Technologies' RainDrop based on "throughput and quality of data we desired," Paweletz said.
"With ddPCR, we are specifically quantifying the DNA that carries the mutation specific to the cancer, and are only studying oncogenic driver mutations," he said. "In that way, we feel that a change in cfDNA genotype levels may be more representative of the specific biology of the cancer being treated."
In the study, they were able to detect mutation prevalence between 0.005 and 0.01 percent with a sensitivity of 5 to 50 mutant copies in a background of 10,000 wild type copies, using serial dilutions of mutant DNA.
Importantly, they found they could improve sensitivity by optimizing cfDNA concentrations. They added a quality control step to their assay, using long interspersed elements (LINE-1) to measure how much DNA was isolated from plasma, and excluding samples without enough cfDNA. Detection of mutant alleles improved with higher LINE-1 levels, but three samples with greater than about 20,000 nanograms per milliliter of DNA also showed no detectable levels of plasma genotype.
Paweletz concluded, "Too little or too much isolated DNA can limit your assay. Too much DNA masks the mutations of interest; too little and you don't [have] enough material. LINE-1 is a good surrogate to measure how much DNA we isolated [and it] has been used as a quality step before."
The researchers reported their ddPCR cfDNA genotyping results in two ways ─ copies per milliliter of plasma and percent reactions that are mutant. Paweletz said that this was done for transparency.
"We have put an enormous amount of thought into the units used for our analysis," he said. "Given that this remains an emerging field and there is no standard unit at this time, we felt it appropriate to report the two most commonly used units in the field so that our results can be compared … We encourage other investigators in the field of plasma genotyping to work towards a standardization of methods and results communication." He added that the situation is reminiscent of the state of affairs in 2004, when the field was standardizing reporting of circulating tumor cells.
Previous work by other groups on genotyping tissue showed lung cancers positive for an EGFR mutation were only detectable by a very sensitive assay, and did not demonstrate the expected benefit from EGFR kinase inhibitors, suggesting detection of false positives or mutations present in minor populations, Paweletz said.
"I would argue that plasma genotype assays should have 100 percent specificity, as even a small false positive rate can significantly impair the clinical utility of an assay," he said. Paweletz's study took into account this potential pitfall. "Our assay used a rigorous definition of 'positive' for a plasma genotype. We used a gold-standard negative population made up of advanced lung cancer patients harboring a contrasting genotype (KRAS) to generate a rigorous definition of positive for an EGFR mutation. In other words, our cut-off values are not analytically but clinically based," he said.
The DFCI study is similar to a recent study published in Science and covered this week in GenomeWeb Daily News, in which Johns Hopkins University researchers measured DNA from circulating tumor cells. Like that research group, Paweletz's team uses the QIAamp circulating nucleic acid kit from Qiagen to prepare its samples. This kit includes a lysis step, and both groups use whole plasma as starting material. Thus, they likely capture freely circulating nucleic acids as well as DNA from circulating tumor cells.
A review in the current issue of the Journal of Clinical Oncology discusses the state of the field for genotyping liquid biopsies and circulating tumor DNA, concluding that digital PCR technologies are now making possible assessment of low-level tumor DNA from early-stage disease and situations where there may be minimal residual disease.
Many groups have been jumping into the sample prep market for cfDNA. For example, Biological Dynamics hopes to use isotachophoresis to pluck cfDNAs from plasma. Meantime, Transgenomic's ScreenCell CTC-capture device is currently being tested on samples from lung cancer patients in collaboration with NYU's Langone Medical Center.
Total cfDNA may also be a useful biomarker. In Oncology Reports, a prospective trial recently concluded that high baseline levels of cfDNA indicate a poor prognosis, and total levels increase significantly at progression. However, Paweletz said that "most studies measuring total cfDNA have demonstrated analytical and clinical validity, but not clinical utility (i.e. does this make a difference to the patient). We feel that measuring actionable and targetable mutations for which genotype-directed therapies exist will ultimately be more beneficial to the patient." He also noted that "with ddPCR we are specifically quantifying the DNA that carries the mutation specific to the cancer, and are only studying oncogenic driver mutations … We feel that a change in cfDNA genotype levels may be more representative of the specific biology of the cancer being treated."
The assay has great potential, according to Paweletz, because it is specific and non-invasive. "It's the application that makes it so intriguing," he said. "Obtaining a biopsy at time of progression (especially in lung cancer) is not always medically feasible and carries some risks."
He added that the group is "very confident" in its assay's validation and plans to further validate it in prospective clinical trials. In terms of commercialization of the assay, Paweletz said the group first plans to continue to "vigorously define the clinical characteristics of our assays so that it can be used to guide clinical treatment for cancer patients in the future." Toward that end, the group has several ongoing and planned prospective studies.