The Rise of Tumor Mutation Burden in Immuno-Oncology
August 25, 2021
Near-term trials to watch, key players, and thoughts on future development
Tumor mutation burden (TMB) is one of the fastest growing biomarkers in the immuno-oncology (I/O) space, bringing with it shifts in cancer diagnostic technology. Early predictive markers that characterize the immune inflammation state of a tumor, such as PD-L1 testing via IHC, have proven somewhat useful at guiding treatment decisions, however, the predictive shortcomings of PD-L1 have become increasingly apparent, driving demand for additional and/or better biomarkers. TMB, which serves as a proxy for the presence of immunogenic neoepitopes, and which appears to be orthogonal to PD-L1 expression, has emerged as a complementary marker to PD-L1 for predicting response to checkpoint inhibitors.TMB has yet to receive regulatory approval for guiding therapeutic decision making, however, its inclusion in numerous clinical trials paints a clear picture of what is to come in the I/O industry. Over the past 3 years, TMB inclusion in I/O trials has seen a CAGR of 90%, making it one of the fastest growing biomarkers in this trial space. By evaluating high profile I/O trials and technologies that screen for TMB, we can understand how TMB may ultimately be brought to market and who are the key players to be aware of.
Using our BioMAP, we have identified at least 23 I/O clinical trials exploring TMB as a biomarker that have a primary completion date in the next 18 months. By identifying the small subset of high impact trials based on phase and total enrollment, we can identify key players in this space and analyze early trial results that hint at future trends in TMB development.[table id=4 /]Request a demo of our Immuno-oncology BioMAP
Commercial options of different TMB panel tests
NGS is the primary method employed to measure TMB, either through targeted panels or whole exome sequencing (WES). The interactive tool below gives an overview of how 7 commercially available NGS + TMB panel tests compare. Tests include:
Foundation Medicine’s F1CDx
Caris Molecular Intelligence’s CGP+
Illumina’s Trusight 170
Guardant Health’s Guardant360*
Personal Genome Diagnostic’s Plasma Select 64*
Kew Inc. CANCERPLEX
Memorial Sloan Kettering’s MSK-IMPACT
TMB Panel Test Tool
3 Observations from These Gene Panels
DNA is measured more than RNA
With the exception of fusion genes in the Caris and Illumina tests, only DNA is measured in these panels. Gene expression profiling (GEP) is often measured with RNASeq and is another fast growing I/O biomarker, raising questions about how panel tests may evolve in the future to accommodate both TMB and GEP.
Smaller than expected genomic regions measured
Panel size does not directly correlate with the genomic region covered by these tests. Cancerplex has the 3rd largest panel, but covers a region of 2.8 Mb, almost 2x larger than MSK-IMPACT, 1.5 Mb. A point to note here is that these regions are significantly smaller than what would be expected with gene panels of this size. If the average gene is assumed to be 10 Kb in length, Cancerplex should actually cover a region of ~4.4 Mb. What this seems to indicate is that the only fractional portions of the genes are being sequenced beyond the promoter regions, especially for the larger panels, which further differentiates these tests from one another as they may offer varying levels of coverage from one gene to the next.
Converging TMB cutoffs
The standard TMB metric is mutations/megabase (Mut/Mb), with studies often establishing cutoffs to designate TMB high vs. TMB low. From study results establishing TMB cutoffs for the tests above, the TMB-high vs. low threshold is clustered around 12.3 Mut/Mb on average. Even though these panels generally cover a genomic region of ~1.5 Mb, observed mutation rates are assumed to be indicative of the remaining sequenced and un-sequenced regions.
Thinking About the Future of TMB
As several companies race towards clinical trial finish lines, the establishment of TMB cutoffs, and industry standards for testing, there are several questions our diagnostics market research team looks forward to answering:
Will TMB testing bring the same challenges as PD-L1?
Much has been said about the complexity and confusion caused by the development of multiple PD-L1 tests, which use different antibody clones, stain different cell types, have different scoring cutoffs, and have different regulatory designations in different indications. While, at first glance, the measurement of TMB appears to be more standardized than PD-L1 (i.e., relies on a numerical score (# mutations / megabase)), it remains to be seen how comparable TMB results are across different tests, given varied genomic coverage in these tests and lack of consistent cutoffs levels. It is worth mentioning that the Cancer Genome Atlas (TCGA) has set benchmarks for standard TMB levels across numerous cancer types. However, it is as of now too early to determine if cutoffs set by studies or TCGA will truly be relevant when measured across different platforms and indications, raising the possibility that diagnostics companies may develop proprietary measurement systems that link their platforms to specific therapeutics, as was the case with PD-L1 CDx tests.
Will blood based TMB overrule tissue sampling?
Liquid biopsy testing is an exciting field that we monitor closely (see our Liquid Biopsy White Paper or Liquid Biopsy Competitive Intelligence Tracker for more), with improving analytical capabilities and the decreasing cost of sequencing driving testing forward in this field. While early detection and GEP with liquid biopsy are clear use cases, there are arguments for and against bTMB testing. On the supportive side of bTMB, blood-based testing allows therapy response monitoring and enables correlation with systemic immune responses. Furthermore, bTMB enables non-invasive testing and reduces variation due to tumor heterogeneity. On the other hand, ctDNA is inherently shorter than DNA derived from tissue (78-138 Kb vs 0.79-1.1 Mb), leading to smaller read lengths and difficulty in establishing equivalence between TMB screened from tissue vs. blood samples.
Do driver mutation tests confer TMB detection bias?
While TMB testing measures the overall rate of mutations across a certain reference region, there is a body of research that mutations actually accumulate at non-linear rates across different genes (Nature Paper, HGV Paper). The research indicates that while driver mutations are the initial source of tumor development, passenger mutations occur in less conserved regions and accumulate more rapidly. NGS cancer panels frequently screen for many driver mutations, and in order to capture an accurate representation of overall mutational load, a random sampling of exons should be assessed rather than a panel of genes with known links to cancer pathology and a propensity to resist alterations.We continually update our the I/O BioMAP data tool and are always looking for new trends related to immunotherapy, diagnostics, NGS testing, and big data in the life sciences. Please reach out if there are new trends you find interesting or questions you want answered and we would be would be thrilled to investigate these with you.*Plasma Select 64 and Guardant360 do not currently screen for TMB but research is underway to validate them for this purposeColin Enderlein is a Senior Analyst at DeciBio Consulting.
Author | Colin Enderlein
Colin is a Senior Analyst at DeciBio with a background in industry CDx business development, tech transfer, and preclinical translational research. At DeciBio, Colin has been involved with several projects that incorporate customizable data visualization tools and coordinated campaigns to gain KOL and clinician perspectives on novel biomarkers in immuno-oncology. Connect with him on LinkedIn.Disclaimer: Companies listed above may be DeciBio clients and/or customers.
No items found.
We're still adding posts. Check back later to see something here.