Studies published by two independent research teams this past week are extending the scope of sequencing-based analyses applied to cell-free tumor DNA circulating in cancer patients' blood.
Together, the proof-of-principle studies point to the circulating tumor DNA both as a source of information for cataloging the genes and pathways involved in treatment resistance or relapse and as a source of information on metastatic disease in a clinical setting.
In a study published online Sunday in Nature, a University of Cambridge-led team did whole-exome sequencing on circulating tumor DNA in blood samples taken over time from half a dozen individuals with metastatic breast, ovarian, or lung cancer as they received treatment for their disease.
"We could actually look at acquired resistance using a hypothesis-free approach," the study's first author Muhammed Murtaza, a clinically trained graduate student in co-senior author Nitzan Rosenfeld's University of Cambridge lab, told Clinical Sequencing News, "which seemed like the logical next step from where the field of circulating tumor DNA was."
The blood-based analyses unearthed somatic point mutations in the individuals' tumors, including post-treatment alterations known for — or suspected of — helping tumors evade the particular therapies used.
Meanwhile, researchers based in Austria and Germany relied on a shallow whole-genome sequencing method, dubbed plasma-seq, and deeper, targeted gene panel sequencing to see copy number changes and somatic mutations, respectively, in advanced prostate cancer DNA nabbed from patient circulation — work that they described in Genome Medicine on Friday.
From that research, study authors concluded that "molecular analysis of plasma may provide a non-invasive approach for tumor cell genotyping, which can easily be repeated during the course of therapy."
The approaches used by each of the groups differ both in the nature of the genetic information they provide and in their most immediate anticipated applications. For instance, the plasma-seq team hopes that their strategy will make its way into clinical practice in the not-too-distant future, while authors of the whole-exome sequencing study plan to apply their strategy as a research tool for digging into the genetics of treatment resistance and disease progression.
The latter approach could ultimately find a place in the clinic, though the lack of clear criteria for interpreting information in the exomes for routine clinical applications is still something of a hindrance — as is the depth of sequence needed to carefully critique coding sequencing in circulating tumor DNA.
For his part, University of Cambridge's Rosenfeld suspects that exome sequencing on circulating tumor DNA could eventually evolve into a clinical method for doing fairly detailed genetic analyses on tumors at key stages of an individual's treatment rather than a means for tracking tumors throughout the treatment process.
"Exome sequencing is not something you'd want to do for monitoring [treatment] response," Rosenfeld told CSN. "It's not the right tool for that job."
"I think there is potential for this in the not-immediate future, not for standard, routine monitoring but at some junctions where you need to think about a change in treatment," he explained.
In the past, most sequencing-centered analyses of cell-free circulating tumor DNA have either involved targeted sequencing to look at sets of genes or known tumor-specific mutations (CSN 3/20/2013) or else broader, low-coverage sequencing to see relatively large alterations in the tumor genome.
In independent studies published last year, for instance, a Chinese group (CSN 10/17/2012) and a Johns Hopkins-led team (CSN 12/5/2012) described methods for tracking tumor-specific chromosomal changes in multiple cancer types using cell-free circulating DNA.
In hopes of getting a more in-depth, but genome-wide, look at the circulating tumor sequences, Rosenfeld, Murtaza, and their colleagues decided to take a crack at doing whole-exome sequencing on cell-free tumor DNA in the blood.
With this approach, Rosenfeld explained, researchers are "a bit constrained" in that they need to pair samples that have sufficient quantities of tumor DNA with adequate sequencing depths. With that in mind, his team focused their current analysis on advanced, metastatic cancers, which tend to shed more DNA into the bloodstream
Though it's feasible to consider characterizing point mutations across both coding and non-coding portions of the tumor genome using circulating tumor DNA, Rosenfeld noted that the depth of coverage required to do so would considerably bump up the cost of such analyses.
"The idea was that if we could look at the whole exome …we'd be able to detect how different clones were changing in response to treatment," said Murtaza. He added that "the reason we've gone with an exome is that we want a certain amount of depth to be able to detect changes in allele fractions with confidence.
For their new study, Rosenfeld, Murtaza, and colleagues assessed tumor DNA from a total of 19 blood samples collected over anywhere from 109 days to 665 days in individuals being treated for metastatic breast, ovarian, or lung cancer.
In addition to focusing on these more advanced cancer cases, they also stuck to samples known to contain a relatively high representation of tumor DNA based on prior digital PCR and tagged-amplicon deep sequencing analyses.
After capturing coding sequences in cell-free DNA from each sample with the Illumina TruSeq Exome Enrichment kit, the team used Illumina's HiSeq 2500 to sequence each exome to an average coverage of between 31-fold and 160-fold.
Exome coverage for most of the samples hovered around 150-fold, Rosenfeld noted, explaining that the desired depth for a given circulating DNA sample tested in this manner may vary somewhat depending how much tumor DNA is present in the plasma.
"We weren't, at this point, trying to optimize [the depths]," he added. "We were just trying to get sufficient coverage to show that this is possible."
Analyses of the advanced tumor sequences unearthed mutations affecting genes or pathways implicated in treatment resistance and/or cancer progression in the past, including an activating EGFR mutation in a sample from an individual whose non-small cell lung cancer progressed after gefitinib treatment.
But researchers found a few intriguing new mutations, too. In an individual with estrogen receptor-positive and HER2-positive breast cancer, for example, exome sequence data from a sample collected after an initial tamoxifen and trastuzumab treatment turned up MED1 mutations, while circulating tumor DNA collected after subsequent lapatinib and capecitabine-based treatment revealed a splicing mutation in another gene called GAS6.
Genes involved in everything from drug efflux to drug metabolism — pathways that "fit in with our biological picture of what is happening" — also appeared in the team's exome data, Rosenfeld noted.
He cautioned that it's difficult to draw conclusions based on just a handful of cases. But applied to larger groups of patients, he and his co-authors suspect that the exome sequencing strategy could uncover new processes involved in acquired drug resistance and/or disease progression.
"The ideal next step in my mind would be a study of a particular tumor type and a number of patients undergoing a specific set of treatments," Murtaza said, since driver mutations "are defined by what statistically stands out and is recurrent, rather than a functional assay."
And with the ability to sequence tumor DNA from the blood, he explained, it becomes possible to look at a series of blood samples to try to uncover such genomic drivers of acquired resistance.
Until now, it hasn't been feasible to do such studies because they would have required multiple biopsies on many individuals over the course of their treatment. Such re-biopsy material is rarely, if ever obtained, particularly in patients receiving established treatments not being tested by clinical trials.
At the moment, the Rosenfeld/Murtaza group is primarily interested in applying its exome sequencing analyses of circulating tumor DNA for that sort of research. That work will likely involve studies on larger patient cohorts, Murtaza explained, as well as efforts to improve the analytical methods used to assess exome sequence data being generated from circulating tumor DNA.
In the longer term, Rosenfeld noted that there may be clinical opportunities to do exome sequencing on circulating tumor DNA at carefully selected points in cancer patients' treatment — perhaps as a means of finding targeted therapies for individuals whose tumors have progressed after an initial treatment, for instance.
Until researchers come up with a straightforward and standardized scheme for interpreting the information from whole-exome sequencing, though, he argued that it may be just as clinically informative to stick to sequencing smaller collections of genes most apt to contain actionable mutations that can help guide subsequent treatment in some way.
Several groups have already taken the gene panel approach to testing tumors themselves, he added, and the same approach should be applicable to circulating tumor DNA, provided sufficient quantities are present in the blood.
Indeed, targeted gene panel sequencing was one of the strategies that the Austrian and German group used to characterize circulating tumor DNA from individuals with advanced prostate cancer.
For their own proof-of-concept study, the researchers paired gene panel sequencing with shallower whole-genome sequencing to look for driver mutations or fusions in circulating tumor DNA, while also profiling copy number patterns across the tumor genome.
At the moment, some prostate cancer patients are treated using a therapy designed to deprive the tumor of the male hormone androgen. But while that treatment is often effective over the short term, many individuals with prostate cancer eventually go on to develop progressive forms of the disease, known as castration-resistant prostate cancer.
As opposed to the high depths of coverage often used in genome sequencing studies, the group's plasma-seq scheme uses shallow whole-genome sequencing — with mean depths of just 0.1-fold coverage — as a means of profiling copy number patterns in cell-free circulating tumor DNA.
"We cannot do mutation calling with this coverage, but we can establish a copy number profile," the study's co-first author Ellen Heitzer, with the Medical University of Graz Institute of Human Genetics, told CSN.
In principle, the approach is similar to a method that researchers from Cold Spring Harbor Laboratories and elsewhere described in 2011 for finding copy number profiles in breast cancer based on individual tumor cell sequence data (see GWDN 3/14/2011). For this study, though, researchers used tumor DNA that came exclusively from the cell-free fraction of blood.
After demonstrating the feasibility of their approach using blood samples from individuals who hadn't been diagnosed with any malignancies and from pregnant women, (whose blood typically contains fetal DNA in the cell-free fraction), the team went on to test the approach on nine individuals with advanced prostate cancer.
Five were individuals with castration-sensitive prostate cancer and five had castration-resistant forms of the disease.
From plasma-seq data on these patients, researchers turned up a range of copy number changes and rearrangements in the circulating tumor genomes. Among them: gains and losses in parts of chromosome 8 that have been implicated in prostate cancer previously and a characteristic prostate cancer-related fusion between the TMPRSS2 and ERG genes.
In individuals whose prostate cancers had been classified as castration resistant, the team also found amplifications affecting androgen receptor gene locus, a copy number shift that's consistent with tumor resistance to hormone deprivation therapy.
That group of investigators also used the circulating tumor DNA for deeper sequencing experiments (around 50-fold coverage) on a set of 55 genes suspected of containing prostate cancer-related driver mutations. By including a few dozen introns on that panel, they were able to see telltale breakpoints, too, pointing to translocations, such as the TMPRSS2-ERG fusion.
All of the sequencing experiments in that study were done using Illumina's MiSeq instrument, which was selected for its speed and affordability — factors that are expected to be important should the plasma-seq technique find favor in a clinical setting.
By running four samples at a time on the MiSeq using the plasma-seq strategy, the team profiled tumor CNVs for around 300 Euros ($390) per sample in the current study, within two to three days of getting each sample.
Sample preparation steps proved somewhat more time-consuming for the gene panel sequencing experiments, which require targeted enrichment. Those experiments had a turnaround time that was a few days longer than the plasma-seq analyses, Heitzer noted, though more samples can be sequenced simultaneously with the panel approach.
Though they've found that they can get away with 1 million reads per sample in a pinch, Heitzer noted that the team has been aiming for at least 1.5 million MiSeq reads for each sample when profiling circulating tumor DNA by plasma-seq.
For their part, Heitzer and her colleagues are keen to see the sequencing-based analyses become a routine liquid biopsy method in the clinic, both for characterizing tumors and monitoring treatment response or disease progression.
Still, because questions remain about factors affecting tumor DNA concentration in the blood — including the extent to which earlier stage tumors shed DNA — more research is needed to determine if, when, and how often such methods should be used in the clinic.
With that in mind, Heitzer and her colleagues are now starting to try out their techniques on blood samples taken from individuals with earlier-stage prostate, colorectal, and breast cancers. Ultimately, such experiments should offer clues about extending plasma-seq to individuals whose disease has not yet become metastatic.
"Mainly, we're focusing now on earlier stages to try to reproduce this data and see whether we can also find these tumor-specific aberrations in earlier stages," Heitzer said, "because this is probably the patient population that really would benefit from such a liquid biopsy."