NEW YORK – Researchers at Ohio State University are hoping their ongoing rapid autopsy program may be able to extend findings from a recent case study that used tumors from an autopsy to characterize the benefits and limitations of circulating tumor DNA or liquid biopsy testing.
In the case report of a deceased patient with a metastatic GI stromal tumor (GIST), published in JCO Precision Oncology in March, the authors concluded that their comparison of tumor samples and liquid biopsy suggests that ctDNA can capture potentially impactful subclonal mutations, but that its scope can be limited by biological factors like tumor burden.
"Moving forward," they wrote, "we recommend combined analysis of both tumor biopsies and ctDNA, particularly at times of cancer progression, to capture dominant or emerging tumor subclones that underlie acquired resistance."
OSU Comprehensive Cancer Center investigator Sameek Roychowdhury, who has been leading the rapid autopsy program, said that circulating cell-free DNA wasn’t at the forefront when he and his team conceived their project, but postmortem analysis of patients' entire tumor burden across multiple metastatic sites has turned out to be a ripe vein for comparing how blood-based genomic assays do or do not reflect the full heterogeneity of a patient's cancer.
"We started the autopsy study to look at drug resistance and acquired resistance … but as we started doing parallel work with cell-free DNA, we thought, hey, this could also work [to examine] what the real limits are: what is actually in the blood [while the patient is alive] compared to autopsy, which theoretically shows you everything," he said.
The initial case report described an early participant in the ongoing autopsy program, a 46-year-old man with GIST, who passed away after progressing on several targeted and experimental therapies.
Investigators compared results from liquid biopsy testing performed at different time points with tumor biopsy samples obtained in real time, and then finally with tissue from the patient's rapid research autopsy, performed after he died. And although the group concluded that liquid biopsy seems to be able to capture complexity that single tumor biopsies might not, it was not a panacea for detecting all emerging mechanisms of drug resistance.
The patient with GIST had initially been treated with the tyrosine kinase inhibitor (TKI) imatinib (Novartis' Gleevec), which is standard of care for this cancer type, in which a majority of tumors harbor activating KIT alterations.
"Unfortunately, most patients with GIST develop TKI resistance as a result of the emergence of secondary KIT mutations that interfere with the mechanism of action of imatinib," Roychowdhury and his coauthors wrote in their report. But when this resistance does emerge, patients can go on to second-line treatments that include various other TKIs that are differentially effective, depending on the type of acquired KIT mutations present. Second-line drugs then induce further resistance mutations in KIT, which in turn require new treatment strategies.
The OSU autopsy program recruits advanced cancer patients with their own or their families' consent, allowing for rapid autopsies after a patient's death to collect samples from multiple primary and metastatic sites. In the GIST patient's case, the samples collected via rapid autopsy enabled researchers to evaluate the polyclonal, heterogeneous nature of his acquired KIT mutations.
But it also allowed the team to perform an unusually comprehensive comparison of tissue and liquid biopsy results. Roychowdhury and his colleagues compared blood samples collected at different timepoints against individual tumor biopsies taken contemporaneously, as well as against the comprehensive picture of the patient's tumor clonality and evolution obtained postmortem.
Liquid biopsy is already being embraced for cancer genotyping, though professional guidelines and regulatory bodies still largely recommend that it be used only for cases where obtaining tissue is either impossible or unsafe for a patient.
But because of the technology's noninvasive nature, there has been enthusiasm that circulating tumor DNA assays might allow more real-time, longitudinal tracking of patients, to monitor disease progression and genomic changes that could inform treatment decisions as a cancer evolves. In contrast, tumor biopsies pose both ethical and practical challenges.
Another benefit that investigators have hoped circulating tumor DNA can offer is the ability to glean genetic information in a more unbiased manner than is possible using a single sample of tumor tissue, which can miss mutations present in unsampled sections of the tumor, or in tumors in other sites of the body that represent clonal offshoots with different genetic profiles.
But most studies comparing liquid biopsy and tumor tissue genomics have used single samples, usually from a primary tumor, limiting the ability to assess whether ctDNA fully captures the genetic heterogeneity in advanced metastatic cancer.
"So far, ctDNA work has [mostly] compared findings in the blood to either a primary tumor ... or a single metastatic tumor or sample that was biopsied. But we know that biopsies are subject to sampling error ... so autopsy is really the only way to ask that question [of concordance] fully," Roychowdhury said.
In their GIST patient, the OSU group analyzed five ctDNA samples using an in-house targeted sequencing assay, alongside multiple tissue biopsies and postmortem samples using exome sequencing.
Comparing a set of five contemporaneous tissue and liquid biopsy samples, the researchers saw that liquid biopsy missed some, potentially important tissue-borne KIT mutations at the earliest sample timepoint. But as the patient's disease progressed, the reverse also occurred, with ctDNA picking up alterations that biopsy samples missed.
In the first biopsy/blood comparison, sequencing of DNA from the biopsy tissue revealed two KIT mutations, neither of which were detected in matching ctDNA. "This could be explained by lower overall tumor burden and less shedding of ctDNA into the circulation at this time point, as well as by differences in the detection limits of WES versus targeted sequencing," the authors wrote.
One of these two tissue-detected mutations, KIT Y823D, has shown sensitivity to ponatinib (Takeda's Iclusig) in vitro. Based on this, and on the results of clinical studies of ponatinib in patients with advanced GIST, the patient was started on the drug, to which he responded by having stable disease for nearly eight months.
At the next time point, after the patient stopped ponatinib as a result of cancer progression, a second tissue biopsy and contemporaneous liquid biopsy showed 100 percent concordance in detected KIT mutations.
In a third set of samples, liquid biopsy not only recapitulated the KIT mutations seen in the paired tumor but also showed a third mutation, D820Y, that the tissue sample failed to capture.
And at a final timepoint, tissue and blood each yielded a single acquired KIT mutation — N822K in the ctDNA and Y823D in the biopsy sample — neither of which showed up in the other sample.
According to the authors, the discordance in these samples, taken approximately one month before the patient’s death, "likely reflects the growing heterogeneity of his disease," and illustrates "how tissue sampling in the advanced metastatic setting may not generate all the information needed to guide therapy."
The investigators also compared KIT mutations identified in the 28 tumor samples obtained in the patient's rapid autopsy against those detected in ctDNA.
They found six unique KIT alterations across all autopsy tumors, with the primary mutation K550_ V559delinsI present in every lesion. The next most prevalent mutation, D820V, was present in 17 of 28 tumor samples, followed by N822K detected in 8 of 28 samples. Two additional KIT mutations were detected in only one tumor.
According to the authors, it appeared, overall, that the KIT mutations that were more prevalent in autopsy tumors were the ones that also showed up in ctDNA, whereas mutations with lower prevalence did not. For example, the patient's fifth ctDNA sample showed the ubiquitous K550_V559delinsI and the shared D820V alteration, but not the private mutations present in just a single tumor.
"These results are consistent with the expectation that tumor burden can affect the ability to detect mutations of interest in ctDNA," the group wrote.
As the team seeks to extend their research to additional cases, the comprehensiveness of having a full autopsy sample set to evaluate could be particularly helpful for better understanding how these biological factors truly affect ctDNA sensitivity, Roychowdhury added.
"We've always wondered about … how much we could miss versus capture. We have these [hypotheses] about what contributes to cell-free DNA content — [including] tumor burden, organ site, how much blood vessel flow goes to that tumor site … the timing of cell-free DNA collection [either] before a therapy has been given [or] after a patient has gotten therapy," he said. "It's a lot of factors. It's not a static thing."
His group has now expanded from this initial case to studying cell-free DNA against autopsies of patients with different cancers to better understand how well cell-free DNA captures clonal heterogeneity.
Clonal heterogeneity is often analogized as a tree with branches, representing truncal mutations versus subclones that branch out with different genomic profiles. "Does it capture the trunk? To what extent does it capture or miss the branches and what are the consequences of that? Are the branches important? These are the questions we are asking … [and] I think this paper is starting to show you that some branches are relevant," Roychowdhury said.
"Exactly how those branches are going to play into patient care or decision-making is to be determined. But if we look at clinical assays or research assays, acknowledging that we're probably missing some of the tree is important," he added.