Ever since the first description of circulating, non-encapsulated DNA in the bloodstream, the detection in body fluids of DNA that originates from malignant tumors has attracted considerable interest for diagnostic and therapeutic purposes.
Over the years, some of the challenges surrounding circulating DNA have revolved around developing methods to allow for its sensitive detection, interrogating its composition, and translating research findings into clinical applications.
“Cell-free DNA (cfDNA) is stable, easy to collect, and relatively easy to analyze, but we are still learning about what it contains,” says Geoffrey R. Oxnard, M.D., assistant professor of medicine at Harvard Medical School. Dr. Oxnard and colleagues have developed a strategy to discriminate germline and cancer-derived variants of the EGFR T790M mutations from cfDNA in patients with non-small cell lung cancer.
As part of this work, investigators in Dr. Oxnard’s lab interrogated a database of plasma next-generation sequencing (NGS) data originating from over 31,000 patients using a bioinformatics algorithm that they developed and validated—illustrating the possibility of collecting data about germline and somatic cancer mutations and discriminating between the two in a single assay.
“The fact that we are finding germline mutations when we study cfDNA reminds us that the vast majority of this DNA is germline,” says Dr. Oxnard. cfDNA is a complex mixture that contains nucleic acid from benign cells, white blood cells, and viruses, with tumor DNA appearing to represent only a fraction. “The most important lesson from our [work] is that we need to remember that we still aren’t quite sure yet of what exactly is in a liquid biopsy, and a lot of work needs to be done before we can advance this further into a test that is detecting subclinical cancer and impact cure rates.”
The study of molecular changes in cfDNA is intimately linked to technologies that are best positioned to address specific biological questions. “For focused questions, such as looking for EGFR resistance mutations, digital PCR is a very cost-effective and scalable method that allows us to study many specimens,” says Dr. Oxnard. Other questions, particularly if they are exploratory in nature, require broader panels, such as NGS.
“As people talk about bigger and bigger assays to study plasma, like whole-exome sequencing, it is important to benchmark our newer assays against existing assays,” continues Dr. Oxnard. Often, several analyses need to be conducted using the same sample of biological material. This is easier and commonly performed for solid tumors, for which the biological material can be divided for parallel tests.
“In comparison, cfDNA is not a resource that can be divided into multiple ways, and the inability to line up multiple assays and test them against each other is a huge challenge simply because of the scarcity of specimens,” he points out.
One of the efforts in his lab focuses on developing and validating standards that are reliable, highly quantitative, and can be used as established standards that can be tested against multiple assays. “This will be a lot more difficult for cfDNA than for solid tumors,” explains Dr. Oxnard.
New Detection Approaches
Researchers at the Johns Hopkins School of Medicine have developed a different approach for early detection using cfDNA, in a way that is unbiased with respect to the position of mutations that exist in an individual tumor, according to Victor E. Velculescu, M.D., Ph.D., professor of oncology and pathology and codirector of cancer biology.
Historically, a substantial body of work exploring biomarkers has relied on late-stage tumors or used information obtained about patients’ primary tumor to guide subsequent efforts to identify blood-based biomarkers in the same individuals. The scarcity of noninvasive methods for early-stage detection has been an ongoing and acute challenge in cancer management.
To address the much-needed development of noninvasive early-stage biomarkers, Dr. Velculescu and colleagues recently developed a panel of 58 cancer-related genes that are frequently mutated in several common tumor types, and used it to interrogate mutations in the cfDNA from the plasma. “We evaluated this approach using plasma from breast, colorectal, lung, and ovarian cancer patients, and we were encouraged to see that we detected somatic mutations in two-thirds to three-quarters of the patients even with early-stage disease, while no alterations were seen in healthy individuals,” says Dr. Velculescu.
The challenge with detecting mutations in circulating tumor DNA is that mutations are usually infrequent and the DNA is diluted, which makes regular sequencing approaches less informative. “In order to detect alterations at such low levels, we had to develop a methodology that would permit sensitive and specific detection,” explains Dr. Velculescu.
The new methodology developed by his team, called targeted error correction sequencing (TEC-seq), relies on purifying free DNA from the plasma and generating a library that is used for very deep sequencing. “The sequencing of many molecules multiple times allowed the identification of true alterations,” says Dr. Velculescu.
Algorithms used as part of this methodology can distinguish DNA in the blood originating from multiple sources, including the tumor, the germline, and blood cells. “The ability to detect these mutations during early-stage disease could permit a substantial change in the way patients are treated,” maintains Dr. Velculescu.
In the analysis of cfDNA, substantial efforts and hopes are focusing on extracting clinical and prognostic information from its qualitative and quantitative characterization. “The presence of circulating tumor DNA is very important at the time of diagnosis, but may also be informative about a worse type of disease depending on its levels,” according to Dr. Velculescu. For example, cell-free circulating tumor DNA may originate not only from the primary tumor, but also from undetected metastatic lesions.
In a subset of patients with resectable colorectal cancer, higher preoperative levels of circulating tumor DNA were associated with disease recurrence and decreased overall survival. “We need to conduct larger trials for patients with each of these tumor types in various populations to show that this is something that has the sensitivity and specificity that is consistent with what we observed and could be useful in a clinical setting,” says Dr. Velculescu.
Role for Digital PCR
Digital PCR in a well-designed assay gives high specificity and sensitivity, particularly for some of the most difficult-to-distinguish mutations, such as single nucleotide polymorphisms, which are an important source of drivers in tumor oncogenes, says George Karlin-Neumann, Ph.D., director of scientific affairs at Bio-Rad’s digital biology center.
At the recent Next Generation Summit held in Washington, DC, Dr. Karlin-Neumann talked about Bio-Rad’s Droplet Digital PCR (ddPCR™), which combines microfluidics with water-oil emulsion droplet technology and allows the absolute quantitation of nucleic acids (Figure 1).
In ddPCR, the amplification reaction is partitioned into 20,000 nanoliter-size droplets that contain randomly distributed nucleic acid molecules. “As a result, the target molecules of interest from the original sample are subdivided into 20,000 individual microreactions,” says Dr. Karlin-Neumann. The reactions are performed in a 96-well PCR plate, and after thermocycling, they are transferred to a droplet reader. Positive droplets are identified based on their increased fluorescence. Poisson statistics corrects for multiple occupancies, and software is used to back-calculate the target concentration in the initial volume.
Unlike real-time PCR, ddPCR does not require the use of standards and measures absolute concentrations of the targets of interest. “Inherently end-users can get more precision from the measurements,” notes Dr. Karlin-Neumann. While in real-time PCR, the presence of inhibitors or unknown constituents may compromise cycling efficiency and distort data interpretation, this is not a concern with digital PCR. “Another gratifying thing about digital PCR in general is that it is much more tolerant of assays that do not have 100% amplification efficiency,” he points out.
The ddPCR platform provides the first results within 3–4 hours, and an entire 96-well plate can be read within 5 hours. It excels particularly for biomarkers that are known, or for cancers where relatively few recurring mutations drive the malignant process, and it is much more affordable that other complex technologies.
“For liquid biopsies, a challenge that needs to be faced and implemented in the clinic is the establishment of clinical utility,” stresses Dr. Karlin-Neumann. Historically, the response to therapy has been assessed using investigations such as CT scans or other imaging approaches, as well as the levels of imperfect protein biomarkers in the blood.
However, as it has been shown for several cancer types, monitoring tumor DNA in the blood also appears to be informative of the relative volume or size of the tumor of origin. The critical question is whether information obtained from changes in tumor DNA from the blood, months before a CT scan, could improve clinical outcomes.
“Intuitively, we tend to hope and imagine that this is likely to be the case, and there are reasons to believe that clinical utility will be shown both for therapeutic and for reimbursement endpoints,” says Dr. Karlin-Neumann.
Another critical area is to identify biomarkers that are best positioned to represent different cancers. “Some tumors appear not to shed as much as other tumors or when compared to the same tumor in a different location,” according to Dr. Karlin-Neumann. For example, intrathoracic lung tumors appear to shed less and, in these instances, even though a mutation can be detected in tissue samples, tumor DNA measurements in the plasma may not be as informative. “While we need to learn more about the biology of different cancers, we anticipate that circulating tumor DNA levels are likely to be a very good surrogate for clinical outcomes and may be less costly and provide more immediate real-time changes than current clinical measures such as CT scans,” he says.
Focusing on the First Step
Chen-Hsiung Yeh, Ph.D., CSO, Circulogene Theranostics, believes that what makes his company different is that the scientists try to make sure that they capture and target every single cfDNA as much as possible at the very first step. Circulogene launched cfDNA liquid biopsy clinical testing services in 2015. “For everybody in the liquid biopsy industry, isolating the DNA or RNA from the specimen is the universal common first step and the most overlooked step,” points out Dr. Yeh.
While extracting nucleic acids from bacterial or mammalian cells has become routine in every molecular biology research lab, extracting circulating DNA for clinical application presents multiple challenges, including the fact that it is highly fragmented, its concentration is low, and a large percentage of the starting material is lost during isolation.
“In our approach, we don’t need to isolate, purify, or extract the DNA, but we do the genetic manipulation directly on the specimen to enrich DNA,” says Dr. Yeh. Because the current industry standard for DNA extraction/concentration leads to substantial sample loss, a large amount of sample would be needed for liquid biopsies to obtain sufficient cfDNA for analyses. “If a large amount of the starting material is lost in the first step, no matter how good the downstream detection technology is, the testing result and accuracy will be greatly jeopardized,” notes Dr. Yeh.
Using a proprietary enrichment technology, scientists at Circulogene have been able to use low-input volumes, such as 50 µL of unprocessed blood, to recover and yield over 1,000 ng/mL of cfDNA (Figure 2). A key challenge with respect to liquid biopsies has been its relatively slow adoption in the medical community. “We have to continuously communicate with the medical community to point out that the liquid biopsy testing is never intended to replace the current gold standard of tissue biopsy, but that it only tries to fill a gap and complement the limitations of tissue biopsy,” says Dr. Yeh.