NEW YORK (GenomeWeb) – Investigators with the Institute of Cancer Research in London have demonstrated that next-generation sequencing-based assays that analyze circulating tumor DNA can be used to monitor patients' tumors while they are in a clinical trial.
Reporting their results in Clinical Cancer Research, the researchers analyzed 39 patients serially throughout various phase I clinical trials, demonstrating that the assays could detect changes in allele frequency that correlated with drug response and disease progression.
Johann de Bono, senior author of the study and head of the division of clinical studies at ICR, told GenomeWeb that his group has been working on cell-free DNA for about five or six years, so knows it "can be useful as a multi-purpose biomarker." However, he added, this study demonstrates that cell-free DNA can be used within a "drug trial studying specific targets, to evaluate the impact of the drug on different tumor clones."
Prior to enrolling patients in the study, the researchers first sequenced tumor biopsy samples on the Illumina MiSeq, selecting those patients who had at least one mutation and who had already completed at least two courses of a therapy.
The team enrolled 39 patients with a range of tumor types — the most common being colorectal, ovarian and breast cancers — into the study and then ran two ctDNA assays that they had validated previously in their lab: Thermo Fisher's AmpliSeq Hotspot Cancer panel, which covers 2,800 mutations in 50 genes, and the AmpliSeq Colon and Lung Panel, which covers 22 genes.
At baseline, the researchers identified 44 different mutations in 23 patients, with TP53, PIK3CA, and KRAS the top three mutated genes. Of those 23 patients, 13 received a drug matching their molecular profile.
Thirty-five of 39 patients received therapies that targeted the PIK3CA-AKT-mTOR pathway, and five patients received two consecutive therapies.
The median number of mutations per patient at a specific time point was two and the mean allele frequency was 15 percent.
All the ctDNA mutations had been identified previously via tumor sequencing except for one TP53 mutation, which was detected in the blood of an ovarian cancer patient at an allele frequency of 5 percent. Re-analysis of the MiSeq sequencing data confirmed that the mutation was not previously present.
Researchers subsequently sequenced patients' ctDNA every week for the first month of therapy and then monthly until progression, analyzing a total of 159 blood samples at various time points.
Four patients for whom the ctDNA assay did not identify any mutations at baseline did have mutations at subsequent time points. One breast cancer patient developed a PIK3CA mutation, an ovarian cancer patient developed a copy number loss of PTEN, and an endometrial cancer patient developed a TP53 mutation. For one melanoma patient who had discontinued vemurafenib prior to the study because he was progressing, the researchers found a BRAF mutation indicating sensitivity to vemurafenib that was not identified when the patient was initially enrolled. The patient was again enrolled in a RAF/MEK inhibitor trial, but progressed after two cycles.
The researchers identified a number of different trends in patients' tumor molecular profiles throughout treatment. For the 13 patients that received a targeted drug, the team monitored allele frequency of the targeted mutations. For a number of patients, the researchers noted that allele frequency decreased.
For instance, a 76-year old patient with advanced colorectal adenocarcinoma had been found to have KRAS, two TP53 mutations, and APC mutations in both the initial tumor sequencing assay and baseline ctDNA assay. She was placed into a clinical trial of a combination PI3K inhibitor and MEK inhibitor. After 63 days, each of the four ctDNA mutations fell to very low allele frequency levels — 4.1 percent, 2 percent, 1.3 percent, and 2 percent, respectively — "suggesting a clonal regression," the authors wrote. Regression was confirmed via a CT scan and her disease was stable for 119 days, after which progression was seen both in the CT scan and in the ctDNA assay via increased allele frequencies. After she discontinued the therapy, allele frequency of the mutations shot up rapidly.
Another patient, a 38-year old with clear cell ovarian cancer, had a PIK3CA and CCTNB1 mutation and was placed on an AKT inhibitor. After 45 days, the PIK3CA allele frequency dropped to 3.1 percent from 7.8 percent and became undetectable after 66 days. After 87 days of treatment, the patient still had stable disease.
In another set of patients, however, allele frequencies of the mutations increased despite being on therapy. For instance, a 67-year old colorectal cancer patient with PIK3CA, FBXW7, and TP53 mutations, received a PI3K inhibitor. The mutations showed the same pattern of clonal evolution throughout therapy, and 42 days into treatment they continued to increase in frequency. The PIK3CA mutation, for example, increased to 72 percent from 58 percent.
Finally, there were two patients who had discordant responses. One patient, who had breast cancer with bone metastases, ascites, and pleural and pericardial effusions harbored PIK3CA and KRAS mutations and received a PI3K inhibitor. Although the PIK3CA mutation became undetectable after 76 days of therapy, dropping from an allele frequency of 9.1 percent, the patient's pleural effusion continued to progress and she had worsening pericardial effusion and ascites. She stopped the therapy and died shortly afterwards.
De Bono said that for the two patients with discordant results — where the analyzed mutation declined despite disease progression — there were likely other mutations driving that disease. Tumor clones are "emerging and evolving," and metastatic cancer in particular may have more than one driving clonal population.
The team is continuing to develop ctDNA assays using both the PGM and MiSeq systems, as well as droplet digital PCR, de Bono noted. The three techniques have different advantages, he said. For instance, ddPCR can detect very low frequency mutations, although it requires a priori knowledge of the mutation.
In addition, he said, "using multiple platforms gives you the ability to validate the results."
Now, he said, his group is running the assay within large clinical trials that include thousands of patients in order to evaluate "multi-purpose biomarkers," which he said included using ctDNA as predictive biomarker, to understand mechanisms of drug resistance, and to look at DNA changes serially to measure drug response. "One assay can measure all of these," he said.
Currently, he said that turnaround time is about one week, but in order for such an assay to be used in real time to make treatment decisions, that should be reduced to between 24 and 48 hours.
Like other groups developing ctDNA assays, de Bono said that such assays hold a lot of promise in the field of oncology. The first application will likely be to identify lung cancer patients who have EFGR mutations for targeted therapy, which he said could become standard within the next two years.
"But the more exciting thing will be to run the assay serially at multiple time points," he said, and to use the results to monitor patients' response to therapy and detect drug resistance early on in order to switch to more effective therapies.
"It's an exciting additional way to study these diseases in real time," he said, and "I think it is likely to revolutionize how we think about cancer."