Renee Caruthers
March 2020—DNA and RNA sequencing, when used together, can improve detection of MET exon 14 skipping mutations in lung adenocarcinoma compared with DNA testing alone, according to a study reported last November at the Association for Molecular Pathology meeting.
While RNA analysis can play an important complementary role to DNA analysis in detecting the mutation, it can also pick up false-positives if RNA analysis data are not adjusted properly, David Manthei, MD, PhD, a fellow in the University of Michigan Department of Pathology, explained in presenting the data.
In the study, 482 cases of non-small cell lung cancer were sequenced using the Oncomine Focus Assay, a Thermo Fisher Scientific next-generation sequencing assay that conducts DNA and RNA analysis in a single workflow.
“You can examine DNA, and often that is the standard for how we look at things,” Dr. Manthei said. But while DNA sequencing offers advantages in terms of its ability to identify well-defined mutations, its coverage can be limited based on panel design. “Due to assay constraints, you might only be looking in areas that you are able, and not agnostically across the entire region,” he said.
By contrast, RNA analysis allows detection of all MET exon 14 skipping mutations in a single assay. This is important, Dr. Manthei said, because these mutations vary in size and in where they occur. About one-third of the mutations occur between exons 13 and 14, in the area known as the acceptor site of exon 14, and two-thirds occur between exons 14 and 15, in the area known as the donor site. In addition, MET exon 14 skipping can result from large deletions that can span not only all of exon 14 but large portions of the intronic sequence. These types of mutations would not be detectable using DNA-based methods unless a large portion of the introns was regularly sequenced.
The Michigan Medicine molecular diagnostics laboratory team—led by Noah Brown, MD, and Bryan Betz, PhD—chose to study the MET exon 14 skipping mutations because they occur in a sizable minority of lung adenocarcinoma cases—two to five percent, depending on the series—and because they can be treated with targeted therapies, Dr. Manthei said. The mutations can cause the tyrosine kinase protein to have prolonged kinase signaling, which can contribute to oncogenesis. In 2018, crizotinib, marketed as Xalkori, received the FDA’s breakthrough therapy designation for use in treating patients with MET exon 14 alterations with disease progression on or after platinum-based chemotherapy. Some patients have seen dramatic responses, Dr. Manthei said.
Initially, the study’s Oncomine Focus Assay analysis found seven MET exon 14 skipping DNA mutations, representing 1.5 percent of the study group. The RNA portion of the Oncomine Focus Assay analysis showed that those seven cases also had a large range of MET exon 13–15 fusion reads. In those seven cases, there were no alternative driver mutations, consistent with the typical mutual exclusivity of driver mutations in lung adenocarcinoma.
However, there were an additional 61 cases in the study group in which the DNA portion of the assay did not identify MET exon 14 mutations but in which the RNA portion of the assay detected a wide range of MET exon 13–15 fusion reads. Over half of those 61 cases had alternative driver mutations detected by the Oncomine Focus Assay, such as KRAS and EGFR exon 19 mutations, which is not expected and raised questions about the specificity of the RNA assay for MET exon 14 skipping mutations, Dr. Manthei said.
“The question was what do you do with those, and how do you interpret them, especially when you can see that the range of raw fusion reads can get up to a substantial number.”
Because the DNA portion of the Oncomine Focus Assay focuses on the donor site of MET exon 14, where the majority of mutations tend to occur, the team decided to use Sanger sequencing to conduct DNA analysis of the acceptor site of MET exon 14, where mutations are less frequent, Dr. Manthei said. The Sanger sequencing was structured to look for underlying DNA mutations that might correlate with the RNA fusion reads.
In all, after performing the additional Sanger sequencing analysis, the team identified 11 total cases with MET exon 14 mutations, eight involving the splice donor site (3′ end of exon 14) represented by the DNA findings of the Oncomine Focus Assay, and an additional three cases with mutations involving the splice acceptor site (5′ end). There were still 58 cases that had MET exon 13–15 fusion reads but lacked DNA mutations. “This observation reflects the fact that RNA fusion reads can be detected in nonmutated samples due to basal levels of alternative splicing,” Dr. Manthei said.
To refine the process, the researchers used those 11 DNA mutations as a “source of truth,” he said, to stratify the assay results in a way that would better inform clinical decisions. It was determined that DNA and RNA analysis, when used together, had high sensitivity in detecting the MET exon 14 skipping mutation, but a system was needed to better rule out false-positive results.
“If you set a threshold based on fusion read number where you are going to detect everything, 100 percent sensitivity, the specificity for that [was] about 50 percent,” Dr. Manthei said.
The team noticed that more fusion reads were observed in higher quality specimens (with more total reads). In addition, cases with MET gene amplification consistently showed higher numbers of MET fusion reads due to the higher overall expression of MET. To adjust for these factors, the team conducted calculations that “normalized” RNA data by comparing RNA MET exon 13–15 fusion reads to total RNA fusion reads and then to the total number of copies of the MET gene. The new calculation thresholds improved the test’s specificity from 48 percent to 97 percent, enabling the team to minimize false-positive results. “This is what we are going to use to help us know when we might want to reflex a case, or borderline case, to something like Sanger sequencing to inform whether a non-targeted mutation is present and to distinguish what is likely just basal alternative splicing.”
Dr. Manthei shared a case that shows the additional utility of combined testing when evaluating MET variants. The patient was transferred to Michigan Medicine from another institution. Sequencing performed before the transfer, which didn’t include RNA analysis, found a MET variant and characterized it as a MET exon 14 skipping mutation. Also present was an EGFR exon 19 deletion mutation. “It didn’t make sense,” he said, “because usually they are mutually exclusive events as far as drivers, so we dug into this a little more.” The Michigan Medicine laboratory team confirmed the EGFR deletion and used Sanger sequencing to look at the region to see if it too saw the variant. “Yes, we did see the variant present in the tumor sample, which is reassuring that the actual sequencing worked.”
But when they dug further, they found that the variant is reported at low frequency in germline databases. “And furthermore, in our RNA assay,” Dr. Manthei said, “we saw zero reads of the MET exon 13–15 fusion, and this is on a high-quality sample of well over 100,000 RNA reads where our threshold for adequacy is 20,000.” He and colleagues then separately isolated normal lymphocytes from this patient, which also demonstrated this variant. “So nailing it together, this is a germline MET polymorphism that was previously mischaracterized as a MET exon 14 skipping mutation,” he said.
Thus, the team found combined DNA and RNA methods can be used to help detect MET exon 14 skipping mutations that would be missed by DNA-only analysis, and RNA-based methods can provide functional information concerning intronic variants whose effect on splicing may be difficult to discern.
Renee Caruthers is a writer in Mountain Lakes, NJ.