NEW YORK (GenomeWeb) – Investigators from the University of Pennsylvania Health System's Center for Personalized Diagnostics have compared tumor profiling by next-generation sequencing with single-gene tests and found that the results agreed most of the time, suggesting that NGS is a reliable platform for routine clinical use.
In total, UPenn CPD has now sequenced over 5,500 tumor samples, according to Jennifer Morrissette, the center's clinical director.
Morrissette told GenomeWeb last week that more than two years into its clinical NGS activities, UPenn CPD has already substantially updated its assays — for example, expanding its hematologic panel to 68 genes — and is currently gearing up to replace its present 47-gene solid tumor panel with a new 155-gene version.
Earlier this month, she and her colleagues published a study in PLoS One in which they compared results of 304 cancer specimens tested by both NGS and one or more targeted single-gene tests.
The study, which began as the CPD implemented its first clinical NGS tests in March 2013, is one of a few to directly compare clinical NGS results with matched single-gene testing as part of the routine pathology practice of a medical center.
The investigators analyzed mutation results obtained during the first year of clinical NGS testing on solid and liquid malignancies carried out at the CPD, during which more than 900 specimens were submitted and processed.
The group was not able to sequence a fraction of solid tumor samples due to low tumor percentage. Of the samples sequenced using NGS, 139 specimens were also tested using single-gene approaches for EGFR mutations, and 138 for KRAS mutations.
Among those shared specimens, 15 EGFR-tested and 13 KRAS-tested samples were excluded from NGS analysis due to insufficient DNA quality or quantity, but in the ones remaining, all mutations detected by the targeted assays were also detected by NGS.
Similarly, of 224 specimens that were tested with targeted BRAF tests by the UPenn pathology lab during the study period, 38 were also analyzed by NGS. About 20 percent of the shared specimens could not be analyzed by NGS due to poor DNA quality or inadequate DNA quantity, but all those remaining showed perfect concordance between NGS and targeted testing.
Of 221 hematologic specimens tested by NGS during the study period, 118 were also tested with FLT3, and 98 with NPM1 targeted tests. In two cases, the targeted FLT3 test detected an internal tandem duplication that was not automatically called by the NGS analysis pipeline, the authors wrote. However, when they manually reviewed the sequencing data, they could see the presence of ITD mutations at allele frequencies of 1.3 percent and 1.6 percent.
In one of these cases, the hematologic malignancy NGS panel additionally detected a pathogenic FLT3 D839G mutation, which was not detected by the single-gene test.
For the remainder of FLT3 shared cases, all of the NPM1 shared cases, and eight cases that had been tested for JAK2 mutations by both modalities, NGS and targeted testing were completely concordant.
As oncologic clinical sequencing programs across the US transition from early NGS pilot programs to more routine use of NGS in pathology, much has been learned and much continues to change in how institutions negotiate what tests to offer, when to use them, and how to best communicate results.
Morrissette said that she hopes the data she and her colleagues collected can be informative to other institutions as they grapple with the practicalities of integrating NGS with more traditional pathology lab services, toward a balance that ensures ideal patient care.
"We're trying to find the appropriate sample flow for any patient with cancer, for when they come in and you don't know what types of mutations you are going to be finding, but also when they are coming back and being followed for relapse or minimal residual disease monitoring," Morrissette said.
"At meetings, people are always saying that molecular pathology is going to disappear, and I've never felt as if that's going to be true," she added. "In our study, we found that even within genes that are easily targetable by NGS, there is still a role for molecular pathology, and on top of that, when it comes to following patients for disease recurrence, running a single-gene test alone can be completely appropriate, and a cost saving mechanism for institutions and insurance companies."
That said, from the results of the study, and UPenn CPD's continuing clinical sequencing results, Morrissette said that it is clear NGS offers a unique ability to discover potentially actionable mutations that otherwise would not be discovered.
"In almost all our tumors, the more you sequence, the more you find targetable mutations that are atypical for that tumor type," she said. For example, BRAF mutations in two cholangiocarcinoma patients led a UPenn clinician to treat them with a BRAF inhibitor, with dramatic positive results.
Unlike some other clinical sequencing centers, Morrissette said UPenn CPD does not include specific therapeutic recommendations in its reports to clinicians, and comfort with broader genomic testing varies from physician to physician.
However, she said that as she and her colleagues have performed more and more NGS, it appears, anecdotally at least, that greater exposure to this type of testing among clinicians is leading to greater comfort, and even dramatic shifts in enthusiasm.
"Some clinicians have moved to 100 percent NGS and they're disappointed if they get anything else, there are some who order both and want the results from both, and then there are some who say 'I can't imagine doing anything more than working with this single gene and that's all the information I want',"Morrissette said.
"But one of our most vocal clinicians, who used to say 'I can't imagine doing more than four genes', is now our biggest advocate for moving to our new 155-gene panel."
According to Morrissette, the areas where their studies support directing patients to single-gene testing over NGS are centered either around urgency and the quality of samples, or around cost savings and practicality in the context of repeat assays to monitor disease progression or remission.
NGS, while getting faster as technologies advance, still represents a longer turnaround time than a more targeted test. Moreover, because clinical labs require their NGS tests to reach a particular allele frequency — in UPenn's case, 4 percent for typical missense mutations, and 1 percent for rare insertions, for example — particularly low-quality or low-tumor samples may be better assayed by more sensitive single-gene tests.
However, Morrissette said that the UPenn CPD has also recently developed a more targeted panel, called the Penn Precision Panel, to increase its ability to do NGS, even on borderline samples.
"We find approximately the same abnormality rate, although we see a smaller variety of mutations," she said. "But most excitingly, it's dropped our frequency of quantity- or quality-not-sufficient results down to less than 1 percent."
"I had the first one of those the other day in so long that I forgot how to generate the report," she said.
Looking to the future, Morrissette said, her lab is also cautiously investigating liquid biopsy in a research context, with an eye to establishing where the noninvasive approach is and is not concordant with tissue sequencing results before they can implement it clinically.
"The concern has been, for people who have been sending this out to some of the outside vendors, that there isn't a perfect correlation, and it seems to depend on tumor type with some doing better than others," she said.
"When we are confident how we can best benefit patients, that's when we'll really start to pull the trigger," she added.