Cancer is a genetic disease with mutations that cause a change in regulatory processes causing cells to grow out of control and become invasive (1). Some of these somatic mutations in tumor DNA result in neoantigens that can be recognized by our immune system, which then allows our own body to recognize and target the tumor. It stands to reason that the mutational landscape of the tumor may correlate to the number of neoantigens that are likely to form. The greater the number of mutations, the more likely neoantigens will form. A greater proportion of non-synonymous mutations would also result in abnormal peptides that may be recognized as neoantigens. Non-synonymous mutations may also inactivate antigen presenting pathways, making this an alternative method for tumor cells to evade detection from the immune system (2).

Tumor mutation load (TML), also known as tumor mutation burden (TMB), is a way to identify and quantify the number of non-synonymous, somatic mutations in cancer cells that occur per megabase of genetic regions of interest. TML has emerged as a predictor for patient stratification for response to immunotherapy. For example, melanoma studies have correlated high TML to response to anti-CTLA-4 checkpoint inhibitor (3, 4), bolstering T-cell response to target tumor cells. High TML is also associated with high efficacy of anti-PD-1 in non-small cell lung cancer (NSCLC) (5).

Measuring TML with NGS

Over the past several years, TML has rapidly evolved as an immunotherapy biomarker, such that TML assays are now being considered by the FDA as potential companion diagnostics for treatment with immune checkpoint blockade (ICB). TML is assessed using next-generation sequencing (NGS), as this is the only method that provides the high throughput and sensitivity needed to identify non-synonymous mutations across coding regions.

Initial TML studies involved whole exome sequencing (WES), comparing tumor DNA and matching normal DNA across all coding regions. Sequencing of normal/tumor pairs is typically performed to identify the germline mutations, which the immune system would recognize as being normal. Due to the greater cost and complexity of whole exome sequencing, researchers are turning to a targeted sequencing approach that focuses on cancer-driven genes and thus a subset of the exome.

In silico analysis using a targeted assay, such as the Ion Torrent Oncomine TML assay, shows a high correlation with exome mutation counts across cancer types (Figure 1). Efforts are ongoing to validate the targeted approach against WES data (2).

Using NGS to measure TML requires researchers to consider several variables, such as the workflow requirements and the quantity and quality of DNA that can be obtained. To learn more about how these variables impact TML and other applications, explore the basics of NGS.


Figure 1. In silico comparison of Oncomine TML assay with whole exome sequencing (WES). WES data of 21,056 samples were downloaded from COSMIC v80. Mutations were restricted to Oncomine TML targets. Mutation counts by WES strongly correlated (r2 = 0.968) with that of the Oncomine TML assay.

Current limitations of TML

TML is a recent development and as such it has limitations as a potential biomarker for immunotherapy. Because it is a new biomarker there are no defined standards for determining and reporting TML. As noted previously, current TML assays examine non-synonymous mutations in DNA. The measurement treats each mutation as equivalent, yet we know that this is not the case. Some mutations could induce a more robust immune response. There is also the potential to miss modifications that would contribute to neoantigenic load, such as post-translational modifications. Regardless, continued research of both tumor and immune cells will help us further understand the cancer-immunity cycle and improve how we can use TML to identify patients that would benefit from immunotherapy.


  2. Chan TA, et al. Annals of Oncology 30:44 (2019)
  3. Snyder A, et al. N Engl J Med 371:2189 (2014)
  4. Van Allen EM, et al. Science 350:207 (2015)
  5. Rizvi NA, et al. Science 384:124 (2015)