NEW YORK – Attendees at the annual meeting of the Association for Molecular Pathology gathered last week to review current practices and raise issues that they hoped to resolve before the association's clinical practice committee working group on tumor mutational burden (TMB) releases draft guidelines next year.
Among takeaways from the meeting was an updated read on the extent to which laboratories are already implementing clinical TMB testing — despite unresolved questions about its utility for predicting patients' response to cancer immunotherapy — and more granular detail about the challenges being faced as institutions seek to meet oncologists' demand for the new biomarker.
Working group leader Larissa Furtado, director of molecular pathology at St. Jude Children's Research Hospital, shared results from a survey that her team conducted among AMP members and non-member laboratories on TMB implementation and perceived barriers. Attendees also spoke about how they've implemented TMB and the questions they'd like AMP to address for future standardization and innovation.
Although only a handful of individuals raised their hands during the session when asked if they've implemented TMB testing for oncologists to use at their institutions, or are about to, Furtado reported that about 40 percent of respondents to the working group survey said they already have clinical assessment of the biomarker up and running at their institutions.
The survey received more than 100 responses, but only 57 completed the entire questionnaire, so the analysis shared at the meeting was limited to that subset. Within the group, uptake appears to be strong, and even among the 60 percent who indicated that they had not implemented TMB, the vast majority said that they plan to implement it within the next 12 months.
Most respondents said they were molecular pathology professionals and pathologists, although Furtado said there were also participants from "a variety of other healthcare specialties." Sixty percent of those surveyed marked their practice setting as a university hospital or academic medical center, with the majority in North America.
The survey, along with an extensive literature review that is still being analyzed, will inform a document outlining TMB validation and implementation best practices, which will address questions like whether tumor-only sequencing can support reliable TMB measurement or if labs need to compare tumor and normal sequences, or what format results should be returned in, and with how much contextual information.
"Many laboratories are currently offering TMB testing, and information generated by these tests is being utilized clinically. However, from an analytical standpoint there currently isn't a standardized way … to calculate TMB. In addition there are no published guideline recommendations on approaches for TMB test validation," Furtado said during her presentation.
According to Furtado, the primary barrier that survey respondents cited for TMB clinical implementation was a lack of guidelines and standardized procedures — indicating that it is "good timing" for the working group's effort.
Participants who hadn't yet begun offering TMB testing also said that not having a large enough next-generation sequencing panel in place at their institution also presented a hindrance.
There are a number of large targeted sequencing panels that gauge several hundred genes and report out TMB, including MSK-IMPACT and FoundationOne CDx. On Tuesday NantHealth announced that its whole-exome tumor-normal IVD for measuring overall tumor mutational burden, called Omics Core, had received US Food and Drug Administration clearance.
The survey indicated, however, there is significant diversity in TMB testing, and importantly, in reporting practices in the field. Among the labs that said they are currently offering TMB, about half routinely perform sequencing on tumor-normal paired samples, while the other half are doing somatic-only.
Somatic-only testers said that they use population database resources for filtering out polymorphic and germline variants. And when asked about their TMB calculation practices, approximately 35 percent said that their assays include all exonic, SNV, and indel variants, while 22 percent indicated that only exonic, non-silent SNVs, and indel calls are included. "Other practices include using all SNV and indel variant calls or only exotic SNV variant calls," Furtado said.
In terms of reporting, 43 percent of the respondents indicated that they only provide the number of mutations per megabase of sequence territory, without contextualizing whether this corresponds to a high or low TMB in the particular tumor type tested. Another 13 percent do categorize a result as representing "low or high TMB" based on a tumor-specific threshold, and the same number of respondents answered that they also provide a description in their reports of how TMB is distributed in the tumor type in question.
Although this indicates "high variability," Furtado said that the survey results by no means indicate or imply that there is an issue with the quality of testing being performed by clinical labs.
"This [reporting variability] is not surprising, and it actually indicates that the development of these guidelines is very timely and has the potential to benefit not only those labs that are planning to implement TMB in the near future, but also those that have already implemented it and are now dealing with specific challenges related to their reporting and [test performance]," she said.
Prompted to bring their concerns to the table, several audience members at the session shared where they are in their own TMB journey, and specific concerns they hoped that the AMP group could address in its planned guidelines.
A meeting attendee who identified himself as being from BioReference Laboratories said that his group has already implemented TMB, using a commercial RUO kit from Thermo Fisher Scientific. TMB is a clinical application the company has begun to highlight more actively in recent months.
Overall, testimonials from attendees speaking at the meeting cited a wide variety of test technologies, including custom-designed panels and other commercial products from companies like PGDx.
Asked by Furtado how it is reporting TMB to ordering oncologists, the BioReference attendee said that his group is "deliberately conservative," reporting the TMB as a number, and using a general statement that reflects results published last year from the CheckMate227 trial, which indicated that, in lung cancer at least, TMB of at least 10 mutations per megabase appears associated with a greater response rate to checkpoint inhibition.
"I am well aware that people are using different cutoffs," the attendee added. "But we have deliberately taken a hard line … We'd rather err on the side of giving the patient the opportunity to have access to the therapies."
In the future, he said, BioReference may establish tumor-specific quartiles or tertiles as internal clinical utility evidence accumulates for histology-specific cutoffs. "But honestly, we have been waiting for AMP to establish consensus recommendations," he added.
Another attendee, from Brigham and Women's hospital, said that his lab is also reporting TMB already, though he has reservations. "For example, he said, we do tumor-only testing without a normal control. And I'm worried that my patients of different ethnic backgrounds might have falsely elevated TMB."
Developing best practices for this tumor-only scenario in particular, Furtado answered, is a primary goal for the working group. "That's certainly one of the things we will address," she said.
Asked about his lab's reporting process, the attendee said that Brigham and Women's reports the number of mutations per megabase, as well as a range for where that sample falls. "If we have a lung cancer where TMB is 12 per megabase, we'll say that within our history of non-small cell lung cancers, this cancer is in the 80th percentile or something like that," he explained.
Interestingly, although the AMP committee has limited the scope of their project to technical questions of test validation and implementation, leaving aside clinical utility for other efforts, laboratorians at AMP appeared to be very cognizant of the nascency of evidence for TMB's clinical utility, especially across different cancer types.
According to Furtado, in the working group survey, both respondents who aren't yet offering TMB and those who already are reported "insufficient evidence regarding TMB clinical utility" as a major implementation barrier. This is perhaps unsurprising considering the evolving state of the field, she added.
The issue for labs in this is that because it's not yet clear exactly how TMB should, or will be used to guide treatment decisions, tests may have to evolve and shift in coming months and years.
Jason Rosenbaum, assistant professor of clinical pathology and laboratory medicine at the University of Pennsylvania said that his institution is close to launching its own TMB report, using an amplicon-based, tumor-only method. But what he'd want to see from the planned AMP guidelines is recommendations on what kind of evidence is needed to improve, or iterate TMB testing.
"This is a space in which algorithms can be adjusted fairly readily, filters can be adjusted, and practice patterns [can be changed] in order to improve your sensitivity and specificity … I would like to see recommendations about what kind of evidence you would need in order to move ahead with a perceived improvement," Rosenbaum.
Panel members from the AMP working group asked if he meant improvement in terms of correlation to a gold standard like exome sequencing-derived TMB, or improvement in terms of predicting therapy response.
"That's exactly the example I would have given about the advice I would like back from you," Rosenbaum replied. "Do I need a different level of evidence if I'm saying I'm getting closer to whole-exome sequencing than if I'm saying I do a better job segregating responders from non-responders? We [also] need guidance in that area," he argued.