While PD-L1 testing has become commonplace with immuno-oncology treatment, the multitude of available antibodies, cut-offs, and scoring algorithms makes this biomarker anything but uniform and consistent. Developments over this past year in the diagnostic market hint that tumor mutational burden (TMB) may face the same variability challenges as PD-L1: the results from one TMB test may not be interchangeable with the next.
Studies show TMB panel tests provide highly varied results. Earlier this year, Memorial Sloan Kettering (MSK) published data on more than 7000 cancer patients, across indications, who underwent TMB testing with the MSK-IMPACT test.1 Key findings highlight that while TMB-high cancers correlated with improved outcomes for multiple tumor types, the TMB cutoffs to predict this survival benefit varied significantly across indications, making a single pancancer TMB cutoff unlikely.
Despite the PD-L1-like trend toward segmentation of tests, there is a key area where TMB does not experience the same challenges as PD-L1: quantitative analysis. TMB is a separate biomarker that is orthogonal to PD-L1, and because TMB is quantifiable, it enables more objective analysis compared to the subjective interpretation of PD-L1 immunoassays.
TMB testing presents real opportunities for guiding the use of precision therapies for a wide range of cancers. Two independent initiatives—one by Friends of Cancer Research in the United States, the other by the Quality Assurance Initiative Pathology (QuIP) in Germany—are now seeking to harmonize TMB testing, comparing existing assays along the way.2,3 As a collaborator, QIAGEN is working alongside academic centers, pharmaceutical organizations, and diagnostic companies to standardize the way in which TMB scores are calculated.
Burdens to standardization
Commonly defined as the total number of exonic somatic mutations, TMB approximates the amount of neoantigens that potentially are recognized by the immune system. The very definition leaves several factors up to interpretation in the estimation, such as whether or not to include insertions and delection (indels) and what value to use for the allelic fraction cutoff.
In the last year, the list of companies offering TMB-enabled panel tests has become noticeably longer, with over a dozen known tests currently available or under development. Similarly, consumers are presented with a variety of TMB testing solutions. Some companies ask users to send in tumor samples, then they report TMB scores back, often with little explanation. Other companies provide the tools necessary for labs to calculate TMB on their own. Given the growing number of TMB testing options, with their varying degree of transparency, there’s an unavoidable challenge of accurately comparing scores generated by different pipelines.
Before TMB testing can become a routine part of clinical cancer care, assay results must be easily comparable, regardless of which sample preparation, sequencing, or analysis tools are used. That standardization starts at the workflow level.
Calculating TMB score
Current TMB assessment methods generally follow these basic steps:
- Tumor sample extraction. Samples may be cytology samples or liquid biopsies, or (more commonly) formalin-fixed, paraffin-embedded (FFPE) tissue samples. However, FFPE samples present a risk for use with high-throughput genomic assays because preanalytical factors common to next-generation sequencing (NGS) testing can affect the sample integrity or amount of extracted DNA. Insufficient DNA is a common cause of sample attrition in clinical trials, so assays that require less DNA may be advantageous.
- Library preparation. In general, TMB panels include a few hundred genes that are widely acknowledged to have a role in cancer tumorigenesis. To accurately calculate TMB, all relevant variants must be detected, even those at low frequency. Some workflows incorporate unique molecular index (UMI) technology to effectively “tag” each molecule with a digital barcode to facilitate more precise allele fractions during downstream data analysis and TMB estimation.
- Sequencing and primary data analysis. NGS methods deliver high-quality base calls, which is critical for achieving accurate and reproducible read mapping and variant calling.
- Secondary data analysis and variant calling. Reads are aligned with a reference genome and library artifacts, and biases are removed to reduce false positives. To this end, careful attention must be given to detecting and cleaning the alignment from reads originating from homologous genes and pseudogenes.
- Variant filtering and TMB estimation. A candidate set of somatic mutations is selected for TMB estimation. The TMB score is then calculated as the number of remaining somatic variants detected per sequencing length unit (in megabase pairs, Mbps).
What’s next for TMB testing?
TMB has yet to receive regulatory approval for guiding therapeutic decision making; however, its rapid adoption by clinical immunotherapy trials paints a clear picture of what is to come in the immuno-oncology industry. Already, TMB is one of the fastest growing biomarkers in this trial space.
With this exciting momentum, standardization must also be included. There are several sources of variability in TMB testing today. From panel content to analysis pipelines, TMB calculation workflows involve a range of different tools and processes. As we push forward in this field, transparency will be of the utmost importance for ensuring harmonization of TMB calculation and implementation in the clinical setting.
There is a race for TMB test positioning as diagnostic and pharmacological companies together with hospitals and clinics address multiple indications and sample types, as we have seen with PD-L1. Ultimately, adopting TMB calculation in routine clinical cancer care depends on two factors: standardization to ensure actionability and reimbursement to increase access. Together, these factors lay the groundwork for successfully transitioning TMB calculation research into precision medicine practice.