14/08/2026
Challenges and Future Directions of Treatment for Advanced Solid Tumors with High Tumor Mutation Burden (TMB-H)
Foreword
The standardized assessment of tumor mutation burden (TMB) encounters challenges across different tumor histologies, treatment regimens and detection platforms, and deliberate consideration is required to ensure the consistency and repeatability of assessment results. While clinical trials have verified that patients treated with immune checkpoint inhibitors (ICIs) can achieve favorable response outcomes, not all patients with elevated TMB can obtain clinical benefits from ICIs, and some low-TMB tumors may still respond to this treatment. Therefore, a comprehensive understanding of the complex interactions among TMB, the tumor microenvironment and genomic characteristics is critical for improving the predictive value of TMB.
Progress in bioinformatics provides possibilities for improving the accuracy and cost-effectiveness of TMB assessment and addressing the existing challenges. In a similar vein, combining TMB with other biomarkers and adopting an integrated multi-omics strategy can further optimize its predictive performance. Sustained collaborative efforts across research, standardization and clinical validation are necessary to fully unlock the clinical application potential of TMB as a genomic biomarker.
Tumor mutation burden (TMB) represents a promising genomic biomarker for predicting therapeutic responses to immune checkpoint inhibitors. TMB is defined as the total number of somatic mutations identified within the tumor genome, (quantified as the count of mutations per million base pairs of the genome (abbreviated as mut/Mb), and functions as a surrogate measurement of potential neoantigen load that can elicit an anti-tumor immune response [1]. The molecular features of TMB are heterogeneous, resulting from the combined effects of both exogenous and endogenous factors. These factors include environmental elements that influence DNA mutagenesis, as well as mutations generated by random errors occurring during deoxyribonucleic acid (DNA) replication. Multiple clinical trials have validated that this molecular marker can effectively predict responses to immune checkpoint inhibitors (ICIs). The pathogenic drivers of highly mutated tumors exhibit inter-tumor heterogeneity. Furthermore, TMB varies considerably across distinct cancer types: melanoma and non-small cell lung cancer often display high TMB. Consequently, it is not feasible to establish a unified median reference range that is applicable to all cancer categories. Certain cancers, including renal cell carcinoma (RCC), do not generally present high TMB, yet still achieve favorable responses to ICI treatment.
Extensive research indicates that tumors with elevated tumor mutational burden (TMB) exhibit an increased probability of neoantigen generation and subsequent recognition by the immune system, which renders such tumors responsive to immune checkpoint blockade therapy. Currently, TMB is universally acknowledged as a predictive biomarker for therapeutic responses to immune checkpoint inhibitors (ICI). The U.S. Food and Drug Administration (FDA) has approved multiple TMB detection assays as complementary diagnostic tools for clinical application.
Notwithstanding widespread recognition of the potential clinical value of TMB detection, significant challenges persist in TMB testing protocols and methodological standardization, which currently impede the widespread implementation of TMB as a robust predictive biomarker for ICI therapy. The inherent molecular complexity and heterogeneity of malignant neoplasms constitute the fundamental basis for the inaccuracy observed in current molecular biomarker detection. Specifically, tumor heterogeneity leads to discrepancies in genomic profiles, treatment sensitivity, and drug resistance mechanisms between primary tumors and their corresponding metastatic lesions, as well as within a single lesion.
This review examines the aforementioned challenges and variable factors associated with the clinical translation and implementation of TMB as a biomarker, while also elaborating on its intended clinical applications. Prospective developments in the clinical application of this biomarker will likely exert far-reaching impacts on the broader field of molecular diagnostics and cancer immunotherapy.
PART1
The Challenge of Using TMB as a Biomarker for ICI Response
Whole exon sequencing (WES) covers 32Mb of the coding region, which corresponds to all 22,000 genes, accounting for approximately 1% of the entire genome. As the gold standard for TMB calculation, WES quantifies the total number of somatic mutations. By contrast, FDA-authorized alternative panel-based assays, including Memorial Sloan-Kettering Integrated Actionable Cancer Target Mutation Profiling (MSK-IMPACT, covering 468 genes) and F1CDx assays (covering 324 genes), quantify the density of somatic mutations, targeting approximately 1.14Mb and 0.8Mb of coding regions respectively. Numerous studies have confirmed that WES-derived TMB and panel-derived TMB (pTMB) exhibit a significant correlation; nevertheless, differences such as the systematic overestimation of TMB by panel detection still remain.
Differences in the content of panels in commercial panel inspections may lead to differences between inspections. For example, in some panels, the detection rate of pathogenic driver mutations may be higher than that of tumor background mutations. This may lead to higher variability at low TMB values. Therefore, after considering the variability within the tumor type, in addition to calibrating the threshold value of each panel, it is also important to generate an adjusted score (such as “mutation load”). This helps to better interpret the TMB score. Many of the panel adjustment scores used to determine TMB have a good correlation with the TMB scores generated by WES and WGS, and the TMB calculations of WES and WGS seem to be highly consistent.
A variety of factors will affect the TMB calculation in different panel tests. The first is the change in the size of the panel, for example, F1CDx is 0.8MB, while the TSO500 panel is 1.94MB. Studies have shown that smaller panel sizes may lead to TMB calculation errors, while larger panel sizes are more likely to be misjudged. This includes the difference in threshold value between the smaller panel and the larger panel. In particular, the smaller panel is not precise enough in distinguishing high-mutant cancer from non-high-mutant cancer. Smaller panels also tend to overestimate the TMB value.
Heterogeneity between and within tumors may lead to inaccurate TMB measurements. Due to the different clonal heterogeneity, the TMB of the metastatic site may be higher than that of the primary site, but this difference may not affect the survival benefits of ICI treatment. In addition, there are organ-specific ecological niches such as bone metastasis, and drug resistance mechanisms may exist, resulting in a weakened clinical response to ICI treatment. The location of the sequencing area and the type of mutation contained also vary depending on the detection of different panels. In addition, the tumor is known to undergo clonal evolution during treatment, which may preferentially affect the non-neoantigen pathway, so that the calculation of TMB is variable throughout the tumor development process.
The variation of targeted cytotoxic T lymphocytes in the cell system (CNs) is dynamic and affects the response to ICI treatment. Among them, the downregulation of HLA-1 expression predicts the adverse prognosis of ICI treatment in patients with colorectal cancer.
Another challenge lies in the impact of tumor purity. For example, related to tissue sampling, low tumor cell concentration may lead to low TMB measurement misjudgment. The exclusion of reproductive system changes by different panels is also different.
The composition of panel tests and the variable selection of genomic variants may also lead to variability in TMB calculations. After accounting for artifacts and germline variants, comparisons of panel tests show that panels including synonymous and coding non-synonymous variants achieve good correlation.
Overestimation of TMB may also occur when mutations only include variants with a variant allele frequency (VAF) exceeding a certain threshold value [2]. Therefore, the specific background of potential mutations (synonymous, non-synonymous, or Indels), and whether mutations occur in coding regions or non-coding regions, may lead to slight differences in TMB calculations across different panels. In addition, variant calling tools, particularly when identifying low-frequency changes, may require greater sequencing depth, which may in turn lead to variant omission. Although these tools can assist with excluding clonal hematopoietic and resistance mutations, as understanding of novel primary and secondary resistance mechanisms continues to deepen, these factors must be taken into account in TMB estimation..
PART2
Future direction
The accuracy of patient selection remains a core factor when adopting TMB as a predictive biomarker for cancer patients. Simply raising the TMB threshold may reduce the size of the patient population eligible for relevant treatment. At the same time, the application of cancer-specific thresholds, which varies depending on the treatment context, can affect final clinical outcomes. Therefore, striking an appropriate balance is essential when adjusting cancer-specific TMB thresholds: setting the threshold too high may lead to the exclusion of patients who could potentially benefit from immunotherapy.
This necessitates the standardization of TMB assessment across all workflow steps, including sample collection, DNA extraction and processing, sequencing technology adoption, bioinformatic analysis, and TMB scoring reporting.
Key standardization measures include: covering both synonymous and non-synonymous mutations in panel testing for WES panels; incorporating driver mutations, germline mutations, and genomic features that may affect ICI response; eliminating germline mutations through simultaneous blood testing or optimized bioinformatic processing; and subsequently developing standardized analysis pipelines, quality control (QC) indicators and annotation tools, to ensure consistent TMB identification and interpretation across different laboratories and research projects.
Potential biomarkers for ICI treatment include tumor-intrinsic biomarkers such as TMB, neoantigens, PD-L1 expression and MSI status, as well as immune-specific biomarkers including T cells, Teffector/T regulatory ratio, tertiary lymphoid structures (TLSs), γ-IFN (gamma interferon) and B cell characteristics[3].
Both MSI-H and TMB are characterized by genomic instability and enhanced immunogenicity. The majority of MSI-H tumors are also TMB-high. However, other abnormal biological processes in tumors can also lead to elevated mutation rates, (such as UV exposure and mutations in the POLE and POLD1 genes. Although these scenarios originate from different mechanisms, both ultimately result in increased tumor mutation burden. Therefore, MSI-H tumors such as colorectal cancer and endometrial cancer often present with high TMB. This synergistic effect has been investigated in relation to responses to ICIs. Nevertheless, this correlation appears to be tumor-specific and may not hold across all solid tumors. Considering that the combined MSI-H and TMB-H status is not prevalent in all cancer types, studies have shown that patients with MSS (microsatellite stable) tumors and high TMB levels can still obtain clinical benefits from ICIs. In fact, the inconsistency between these two biomarkers may be a cause of ICI drug resistance.
Incorporation of additional biomarkers, such as PD-L1, may enhance the predictive and prognostic performance of TMB in guiding clinical decision-making for immune checkpoint inhibitor (ICI) therapy. Nevertheless, the correlation between TMB and PD-L1 expression tends to be poor across different tumor types[4].
Other biomarkers associated with ICI sensitivity, including CD8+ T cell abundance, inflammatory tumor microenvironment (TME), and T score, may serve as components of composite biomarkers for future clinical application. However, retrospective analyses have demonstrated that in cancer patients where neoantigen load, a surrogate for TMB, is not positively correlated with CD8+ T cell levels, high TMB fails to act as a valid predictive biomarker for ICI therapy. Additionally, high TMB cannot predict ICI response in certain cancer types, including brain cancer, breast cancer, and prostate cancer. Accordingly, further investigation is required, potentially exploring distinct TMB threshold settings, to clarify the actual clinical utility of TMB in these less immunogenic and other rare tumor types. Assessment of the respective impacts of TME and TMB status on ICI therapeutic outcomes necessitates collinearity assessment and multivariate statistical analysis.
Current research has explored the combined association between TMB and T cell inflammatory gene expression profile (GEP) scores. Findings from key clinical trials of pembrolizumab indicate that both high GEP scores (defined as the top tertile in pan-cancer cohorts) and high TMB scores (calculated via whole-exome sequencing, WES) are correlated with improved objective response rate (ORR) and progression-free survival (PFS) in pan-cancer, head and neck cancer, and melanoma patient cohorts. However, the overall correlation between GEP and TMB scores in pan-cancer populations remains low.
Tertiary lymphoid structures (TLSs) are well-recognized as critical mediators of anti-tumor immunity and lymphocyte responses. Owing to their association with improved survival outcomes following ICI therapy, TLSs have emerged as a potential prognostic biomarker. To date, however, no definitive correlation between TLS presence and TMB status has been established in the context of ICI response[5].
The neutrophil-to-lymphocyte ratio (NLR), along with circulating monocyte and neutrophil counts, represents an additional category of peripheral blood-based biomarkers that may predict ICI response. Pre-treatment NLR in combination with TMB may serve as a candidate composite biomarker. In this framework, a low NLR combined with high TMB is associated with a more favorable response to ICI therapy.
HAlterations (in homologous recombination defect (HRD) and genomic instability may be significantly correlated with high TMB. It has been demonstrated that heterozygous human leukocyte antigen (HLA) genotypes correspond to better responses to ICIs, and studies indicate that HLA loss of heterozygosity (HLA-LOH)—which is associated with immune escape in lung cancer—may also be linked to high TMB. Based on these findings, it has been proposed that HLA-corrected TMB may correlate with survival advantages in patients with non-small cell lung cancer (NSCLC)[6].
Additionally, the baseline clonality and diversity of the T cell receptor (TCR) repertoire have been confirmed to have potential predictive value for treatment response to ICIs.
In future research, it will be necessary to consider how the tumor landscape influences TMB in specific cancer types, and to incorporate this factor when evaluating the neoantigen patterns of these tumors, as TMB serves as a surrogate for neoantigen burden in cancer cells. It is essential to adopt a multi-omics strategy to identify biomarkers capable of predicting immunotherapy response. Through multi-omics analysis, researchers have found that specific alterations such as those in ROS1 (c-ros oncogene 1), SPEN (SPEN family transcriptional repressor) and PTPRT (protein tyrosine phosphatase receptor type T) can predict immunotherapy response across multiple cancer types. Furthermore, response differences are also substantially affected by a range of clinical and demographic factors.
Summary
In summary, although TMB is not a perfect biomarker, combined with the progress of TMB measurement technology, the implementation of standardization measures, the optimization of detection schemes, the characterization of TMB in multiple cancer types, the combination with other ICI response biomarkers, dynamic monitoring based on blood-based TMB (bTMB), and the strict validation of improved detection methods and other relevant factors, TMB has the potential to be better applied in real clinical practice..
References
[1] Sci. Rep. 2022, 12, 20495.
[2] Rev. Immunol. 2019, 37, 173–200
[3] Sci. Rep. 2021, 11, 21072
[4] Lancet Oncol. 2020, 21, 1353–1365
[5] Cancer 2022, 10, e003091.
[6] Ann. Oncol. 2023, 34, 377–388
Statement: This article is only for sharing. If it involves copyright and other issues, please contact us as soon as possible and we will correct it as soon as possible. Thank you!
