Honing Liquid Biopsies for Application in Cancer Detection, Monitoring

Geoffrey R. Oxnard, MD

Liquid biopsies have evolved beyond finding genotypes and have moved into the cancer detection and monitoring spaces, according to Geoffrey R. Oxnard, MD, medical oncologist at the Dana-Farber Cancer Institute and associate professor of medicine at Harvard Medical School.¹

“We want to use blood tests...to help patients find early cancer, determine [whether] cancer is left over, and figure out what is going on with their treatment,” Oxnard said to an audience at the 2nd Annual Precision Medicine Through Plasma: Using Liquid Biopsies in Contemporary Oncology Care symposium hosted by Physicians’ Education Resource®, LLC, in New York, New York.

Pivoting away from genotyping assays to other indications requires that current tests be tweaked, said Oxnard. Clonal hematopoiesis mutations, benign mutations that are present in blood samples, show up in liquid biopsies routinely, and these results can confuse diagnosing clinicians.

For example, a low-level p53 mutation that shows up on a sample from a liquid biopsy assay that is currently used in clinical practice can most likely be attributed to a benign mutation present in the white blood cells (WBCs) and not cancer, Oxnard explained.

Cancer Detection Assays

For detection assays, the most important factors are including common variants in the test and allowing test sensitivity at low allele fraction; currently there are no assays available in the clinical setting that can be used for these purposes. Oxnard said next-generation sequencing (NGS) assays are broad enough, but they do not take into consideration WBC mutations, which can confuse the test results if they are being used for detection.

He went on to explain the 3 categories of cancer detection liquid biopsies: prebuilt assays that are tumor agnostic; prebuilt assays that are informed by the tumor; and personalized, or bespoke, assays.

For bespoke assays, he used the example of the TRACERx study, in which 100 patients’ tissue samples were prospectively analyzed by a pathologist to create personalized multiplex—polymerase chain reaction (PCR) assays. These were used to detect postoperative relapses in patients with non–small cell lung cancer (NSCLC), and the results from this small cohort demonstrated that these assays showed promise in this setting.²

Further data presented at the European Society of Medical Oncology 2018 Annual Meeting used a personalized multiplex-PCR assay from 16 somatic mutations and an NGS platform in 130 patients with stage I to IV colorectal cancer.³Relapse was observed in 75% of patients with posi- tive circulating tumor DNA (ctDNA) results and only 11.8% of patients who had negative ctDNA results, indicating that positive results were a strong indicator of disease relapse.¹

Oxnard said this technology is not yet accurate enough to make it clinically relevant, “but the potential is very compelling.” He added that once these assays are created, they are cheap to run again and again, making them an attractive clinical tool.

Bespoke assays draw information from the individual, but prebuilt assays use a number of predetermined genes that are often mutated in cancer for detection. Oxnard says this is helpful for patients whose tumor samples are not suitable for analysis.

He cited data from the CCGA study, in which the following prototype sequencing assays were performed: paired cell-free DNA (cfDNA) and WBC targeted sequencing of 507 genes for single nucleotide variants, paired cfDNA and WBC whole genome sequencing for copy number variation, and cfDNA whole genome bisulfite sequencing for methylation.⁴

Data presented at the 2018 American Society of Clinical Oncology Annual Meeting showed that 54% of the variants detected by the test in patients with cancer were attributable to the WBCs versus 98% in patients without cancer.¹Removing the mutations signaled by the WBCs significantly increased the accuracy of the test.

Additionally, less than 1% of patients (n = 5) in the non-cancer group had a dramatic cancer signal; subsequently, 2 of those patients (40%) were given a diagnosis of cancer on follow-up.

“The idea that they could find a cancer signal in the blood before the cancer appears is amazing,” Oxnard said. “That needs more clinical development, but that is the promise of this general approach.”

Cancer Monitoring Assays

For cancer monitoring, Oxnard said testing is dependent on detecting the variable shed of tumor DNA over time and identifying changes that can be informative about the disease. He added that patients’ whose tumors shed DNA generally have worse outcomes than patients whose tumors do not; therefore, shedding is associated with prognosis.

He used the example of a patient with known EGFR-mutated lung cancer whose liquid biopsy test for the mutation came back negative. He said this kind of result is indicative of a good prognosis because the evidence points to the fact that the patient’s tumor is no longer shedding DNA.

Determining the level of DNA in the plasma that is clinically meaningful to a patient’s response needs to be determined to use this as a measure of treatment success.

In the phase I study of the novel EGFR inhibitor ASP8273, a dramatic cfDNA response was observed in the plasma,⁵“but in a randomized trial, this drug was not as good as erlotinib [Tarceva]. This is not a good EGFR inhibitor, but [the data] look great,” Oxnard said.

Similar results were observed in patients who had a complete plasma response for the RET inhibitor LOXO-292. After 2 weeks, some patients had 100% plasma clearance, but their imaging showed stable disease. Unlike the results observed with ASP8273, data from a clinical trial supported these plasma results with a confirmed overall response rate of 74% in patients with RET fusion—positive NSCLC.⁶

Oxnard said the idea that plasma could detect effects not apparent on imaging is very promising and that this technology could be used to supplement imaging.

Monitoring assays need to be fast and cost-effective to be used for repetitive screenings during cancer treatment as well as highly quantitative, reproducible, and accurate. Until those criteria are met, Oxnard said, it is necessary to determine “that key decision point where it is worth sending an [expensive] assay.”

Further optimization and validation of these assay methods are needed for cfDNA genomics to be used as a detection tool, but the prospect of its use in the years ahead is promising, Oxnard concluded.

References:

1. Oxnard GR. Beyond genotyping: cfDNA genomics for detection of cancer and assess- ment of treatment response. Presented at: 2nd Annual Physicians’ Education Resource®, LLC, Precision Medicine Through Plasma: Using Liquid Biopsies in Contemporary Oncol- ogy Care; December 2, 2017; New York, NY.
2. Abbosh C, Birkbak NJ, Wilson GA, et al. Phylogenetic ctDNA analysis depicts ear- ly-stage lung cancer evolution. Nature. 2017;545(7655):446-451. doi: 10.1038/na- ture22364.
3. Reinert T, Henriksen TV, Rasmussen MH, et al. Serial circulating tumor DNA analysis for detection of residual disease, assessment of adjuvant therapy efficacy and for early recurrence detection in colorectal cancer [abstract 456PD]. Ann Oncol. 2018;29(suppl 8). doi: 10.1093/annonc/mdy281.004.
4. Oxnard GR, Maddala T, Hubbell E, et al. Genome-wide sequencing for early stage lung cancer detection from plasma cell-free DNA (cfDNA): the Circulating Cancer Genome Atlas (CCGA) study [published online June 7, 2018]. J Clin Oncol. 2018;36(suppl 18). doi: 10.1200/JCO.2018.36.18_suppl.LBA8501.
5. Yu HA, Spira A, Horn L, et al. A phase I, dose escalation study of oral ASP8273 in pa- tients with non—small cell lung cancers with epidermal growth factor receptor mutations. Clin Cancer Res. 2017;23(24):7467-7473. doi: 10.1158/1078-0432.CCR-17-1447.
6. Oxnard GR, Subbiah V, Park K, et al. Clinical activity of LOXO-292, a highly selective RET inhibitor, in patients with RET fusion+ non—small cell lung cancer. Presented at: International Association for the Study of Lung Cancer 19th World Conference on Lung Cancer; September 23-26, 2018; Toronto, Canada. Abstract OA12.07.