Tumor Mutational Burden Estimation Research Articles

Leveraging non-coding regions to guarantee the accuracy of small-sized panel-based tumor mutational burden estimates.

Accurate estimation of tumor mutational burden (TMB) as a predictor of responsiveness to immune checkpoint inhibitors in gene panel assays requires an adequate panel size. The current calculations of TMB only consider coding regions, while most of gene panel assays interrogate non-coding regions. Leveraging the non-coding regions is a potential solution to address this panel size limitation. However, the impact of including non-coding regions on the accuracy of TMB estimates remains unclear. This study investigated the validity of leveraging non-coding regions to supplement panel size using the OncoGuide NCC Oncopanel System (NOP). The aim of this study was to evaluate test performance against orthogonal assays and the association with responsiveness to immune checkpoint inhibitors was not included in the evaluation. We compared TMB status and values between TMB calculated only from coding regions (NOP-coding) and from both coding and non-coding regions (NOP-overall) using whole exome sequencing (WES) and FoundationOne®CDx (F1CDx) assay. Our findings revealed that NOP-overall significantly improved the overall percent agreement (OPA) with TMB status compared with NOP-coding for both WES (OPA: 96.7% vs. 73.3%, n = 30) and F1CDx (OPA: 90.0% vs. 73.3%). Additionally, the mean difference in TMB values compared with WES was lower for NOP-overall (3.55 [95% CI: 0.98-6.13]) than for NOP-coding (6.22 [95% CI: 3.73-8.70]). These results exemplify the utility of incorporating non-coding regions to maintain accurate TMB estimates in small-sized panels.

Flushing fluid of post-bronchoscopic biopsy as a novel liquid biopsy specimen for genome profiling in advanced lung cancer.

e15038 Background: Tumour tissue is the primary substrate for molecular testing in advanced NSCLC. However, obtaining sufficient tissue samples from biopsy for molecular testing can often be challenging. Flushing fluid of post-bronchoscopic biopsy, which contains tumor-derived DNA, may serve as a supplement for genotyping. The aim of this study is to assess the efficacy and precision of detecting gene alterations in flushing fluid samples and evaluate their clinical potential. Methods: Flushing fluid, matched tissue (TIS), and whole blood were prospectively collected from 25 patients with stage III-IV NSCLC between June 2022 and December 2023. Ten milliliters of flushing fluid and whole blood were used to prepare flushing fluid supernatant (FFS) and plasma (PLA), respectively. Capture-based targeted sequencing of TIS, FFS and PLA samples was performed using a universal oncology panel. Results: In this cohort of 25 patients, there was a male predominance (22 out of 25) and the mean age was 60.2 years. The majority of patients were diagnosed with lung adenocarcinoma (23 out of 25), and smokers outnumbered non-smokers (16 out of 25). Furthermore, 18 out of 25 patients presented with metastatic disease at diagnosis. The performance of three sample types regarding key quality control indices was evaluated first. FFS exhibited a maximum allelic fraction similar to TIS, while PLA showed a significantly lower than TIS and FFS. Subsequently, we compared the positive detection rates, with TIS, FFS, and PLA achieving respective rates of 100%, 96%, and 80%. A total of 495 mutation events were detected from the three samples (FFP:209, FFS:196, PLA:90), Among these, 161 were concordant between TIS and FFS samples, and 72 were concordant between TIS and PLA samples. The sensitivity and positive predictive value (PPV) of detecting eight established driver alterations in NSCLC were compared using TIS as the gold standard with the two liquid biopsies. FFS demonstrated superior sensitivity and PPV compared to PLA (80.5% vs. 56.1%, 97% vs. 88.5%) and showed equivalent performance for EGFR mutation. Furthermore, estimates of tumor mutational burden (TMB) based on TIS and FFS were highly correlated (Pearson r= 0.97). Conclusions: Our study demonstrated that FFS performs comparably to matched TIS and superiorly to PLA for genotyping advanced NSCLC. FFS emerges as a valuable specimen source, especially in patients with insufficient tumor tissue.

Harmonizing tumor mutational burden analysis: Insights from a multicenter study using in silico reference data sets in clinical whole-exome sequencing (WES).

Tumor mutational burden (TMB) is a significant biomarker for predicting immune checkpoint inhibitor response, but the clinical performance of whole-exome sequencing (WES)-based TMB estimation has received less attention compared to panel-based methods. This study aimed to assess the reliability and comparability of WES-based TMB analysis among laboratories under routine testing conditions. A multicenter study was conducted involving 24 laboratories in China using in silico reference data sets. The accuracy and comparability of TMB estimation were evaluated using matched tumor-normal data sets. Factors such as accuracy of variant calls, limit of detection (LOD) of WES test, size of regions of interest (ROIs) used for TMB calculation, and TMB cutoff points were analyzed. The laboratories consistently underestimated the expected TMB scores in matched tumor-normal samples, with only 50% falling within the ±30% TMB interval. Samples with low TMB score (<2.5) received the consensus interpretation. Accuracy of variant calls, LOD of the WES test, ROI, and TMB cutoff points were important factors causing interlaboratory deviations. This study highlights real-world challenges in WES-based TMB analysis that need to be improved and optimized. This research will aid in the selection of more reasonable analytical procedures to minimize potential methodologic biases in estimating TMB in clinical exome sequencing tests. Harmonizing TMB estimation in clinical testing conditions is crucial for accurately evaluating patients' response to immunotherapy.

Assessments of tumor mutational burden estimation by targeted panel sequencing: A comprehensive simulation analysis.

Tumor mutational burden (TMB), when at a high level, is an emerging indicative factor of sensitivity to immune checkpoint inhibitors. Previous studies have shown that the more affordable and accurate targeted panels can be used to measure TMB as a substitute for whole exome sequencing (WES). However, additional processes, such as hotspot mutations exclusion and TMB adjustment, are usually required to deal with the effect of the limited panel sizes. A comprehensive investigation of the effective factors is needed for accurate TMB estimation by targeted panels. In this study, we quantitatively evaluated the variances of TMB values calculated by WES and targeted panels using 10,000 simulated targeted panels with panel sizes ranging from 0.2 to 3.1 million bases. With The Cancer Genome Atlas (TCGA) cancer samples and mutation profiles, we fixed regressions on WES-TMBs and panel-TMBs to assess the performance of a given targeted panel. Panel size was found as one of the major effective factors of TMB estimation. Meanwhile, by investigating the well-performing small panels that reported TMB values similar to those of WES, we demonstrated the evidence of the cancer type-specific impacts of genes on TMB estimation and identified high-impact gene sets for different cancer types based on the TCGA data. This study revealed the quantitative correlations between TMB variance and panel size, and the potential impacts of individual genes on TMB estimation. Our results suggested that for cancer patients diagnosed using targeted panels, it would be highly beneficial to have the capability to directly measure TMB from the targeted sequencing data. This would greatly assist in making decisions regarding the use of immunotherapies.

Whole genome sequencing for more accurate TMB and mutational signatures in common cancers.

e15173 Background: Due to the dramatic decrease in cost of genome sequencing, we are entering the era of whole genome sequencing (WGS) at $100. Tumor mutational burden (TMB) and mutational signatures have been introduced as potential prognostic/predictive cancer biomarkers and mostly assessed by targeted cancer gene sequencing panels. However, they are truly whole genome-wide events and thus more accurately estimated by whole genome sequencing data. Methods: We analyzed WGS data produced by Genome Insight Inc. that encompass &gt; 1,300 breast, lung, and hepatocellular carcinoma samples, collectively. Our genome-wide TMBs (wgsTMB) were systematically compared with the TMB estimates by targeted sequencing methods, such as conventional whole-exome sequencing (WES) and targeted panel sequencing from FoundationOne CDx (F1CDx). In parallel, we leveraged genome-wide somatic mutations for validating mutational signatures projected from the targeted sequencing methods. Results: Pembrolizumab is FDA approved for solid tumors with high TMB (≥ 10 mutations/Mb) assessed by F1CDx. This criterion classified that 546 of the tumors (41%) in our cohort as high F1CDx TMB (f1TMB), who may benefit from immunotherapy. However, the high f1TMB classification was inconsistent to the genome-wide TMB classification ( p &lt; 2.2e-16). For example, 132 high f1TMB tumors (24.1%; 132/546) showed lower than the median wgsTMB estimate. On the flip side, 32% of the low f1TMB tumors (251/784) demonstrate higher than median wgsTMB estimate (see table). Although there is no approved criterion in WGS that defines hypermutators, our data illustrates that WGS will be helpful for redefining the criteria for selecting patients for immune checkpoint inhibitors. The wgsTMB was also an essential factor for accurate estimation of mutational signatures from targeted sequencing. For example, in high TMB tumors, mutational signature estimated from WES showed on average 0.87 cosine similarity with signatures from WGS. However, the correlation was substantially reduced in low TMB tumors, showing only ~0.64 on average. Particularly, ~12% of breast tumors with considerable levels of homologous recombination deficiency (HRD) signatures by WGS were not captured by whole-exome sequencing based approaches. Further exploration should be performed with WGS to select proper patients for targeted therapies, including PARP inhibitors. Conclusions: Despite the current guidelines depending on targeted sequencing in precision oncology, we showed that whole-genome sequencing can deliver more accurate cancer genome profiles, such as, but not limited to, TMB and mutational signatures. Given its more comprehensive capabilities and affordable cost, we believe that WGS-based precision oncology is medically necessary. [Table: see text]

Comprehensive bioinformatic analysis constructs a CXCL model for predicting survival and immunotherapy effectiveness in ovarian cancer.

Background: Immunotherapy has limited effectiveness in ovarian cancer (OC) patients, highlighting the need for reliable biomarkers to predict the effectiveness of these treatments. The C-X-C motif chemokine ligands (CXCLs) have been shown to be associated with survival outcomes and immunotherapy efficacy in cancer patients. In this study, we aimed to evaluate the predictive value of 16 CXCLs in OC patients. Methods: We analyzed RNA-seq data from The Cancer Genome Atlas, Gene Expression Omnibus, and UCSC Xena database and conducted survival analysis. Consensus cluster analysis was used to group patients into distinct clusters based on their expression patterns. Biological pathway alterations and immune infiltration patterns were examined across these clusters using gene set variation analysis and single-sample gene set enrichment analysis. We also developed a CXCL scoring model using principal component analysis and evaluated its effectiveness in predicting immunotherapy response by assessing tumor microenvironment cell infiltration, tumor mutational burden estimation, PD-L1/CTLA4 expression, and immunophenoscore analysis (IPS). Results: Most CXCL family genes were overexpressed in OC tissues compared to normal ovarian tissues. Patients were grouped into three distinct CXCL clusters based on their CXCL expression pattern. Additionally, using differentially expressed genes among the CXCL clusters, patients could also be grouped into three gene clusters. The CXCL and gene subtypes effectively predicted survival and immune cell infiltration levels for OC patients. Furthermore, patients with high CXCL scores had significantly better survival outcomes, higher levels of immune cell infiltration, higher IPS, and higher expression of PD-L1/CTLA4 than those with low CXCL scores. Conclusion: The CXCL score has the potential to be a promising biomarker to guide immunotherapy in individual OC patients and predict their clinical outcomes and immunotherapy responses.

Fast, accurate, and racially unbiased pan-cancer tumor-only variant calling with tabular machine learning

Accurately identifying somatic mutations is essential for precision oncology and crucial for calculating tumor-mutational burden (TMB), an important predictor of response to immunotherapy. For tumor-only variant calling (i.e., when the cancer biopsy but not the patient’s normal tissue sample is sequenced), accurately distinguishing somatic mutations from germline variants is a challenging problem that, when unaddressed, results in unreliable, biased, and inflated TMB estimates. Here, we apply machine learning to the task of somatic vs germline classification in tumor-only solid tumor samples using TabNet, XGBoost, and LightGBM, three machine-learning models for tabular data. We constructed a training set for supervised classification using features derived exclusively from tumor-only variant calling and drawing somatic and germline truth labels from an independent pipeline using the patient-matched normal samples. All three trained models achieved state-of-the-art performance on two holdout test datasets: a TCGA dataset including sarcoma, breast adenocarcinoma, and endometrial carcinoma samples (AUC > 94%), and a metastatic melanoma dataset (AUC > 85%). Concordance between matched-normal and tumor-only TMB improves from R2 = 0.006 to 0.71–0.76 with the addition of a machine-learning classifier, with LightGBM performing best. Notably, these machine-learning models generalize across cancer subtypes and capture kits with a call rate of 100%. We reproduce the recent finding that tumor-only TMB estimates for Black patients are extremely inflated relative to that of white patients due to the racial biases of germline databases. We show that our approach with XGBoost and LightGBM eliminates this significant racial bias in tumor-only variant calling.

External quality assessment (EQA) for tumor mutational burden: results of an international IQN path feasibility pilot scheme

Tumor mutational burden (TMB) has recently been approved as an agnostic biomarker for immune checkpoint inhibitors. However, methods for TMB testing have not yet been standardized. The International Quality Network for Pathology (IQNPath) organized a pilot external quality assessment (EQA) scheme for TMB testing. The aim of this program was the validation of the materials and the procedures for the EQA of this complex biomarker. Five formalin-fixed paraffin-embedded (FFPE) cell lines were selected to mimic the various TMB values observed in clinical practice. The FFPE samples were tested with the FoundationOne CDx (F1CDx) assay as the reference test and three commercially available targeted sequencing panels. Following this internal validation, the five cell lines were sent to 29 laboratories selected on the basis of a previous survey. Nineteen of the 23 laboratories that submitted results (82.6%) used targeted sequencing for TMB estimation. Only two laboratories performed whole exome sequencing (WES) and two assessed TMB by clinical exome. A high variability in the reported TMB values was observed. The variability was higher for samples with the highest TMB value according to the F1CDx test. However, good reproducibility of the TMB score was shown by laboratories using the same panel. The majority of laboratories did not indicate a TMB cut-off value for clinical interpretation. In conclusion, this pilot EQA scheme suggests that it is feasible to run such an EQA program for TMB assessment. However, the results of our pilot highlight the numerous challenges for the standardization of this test.

Paired tumor/normal sequencing to overcome racial differences in tumor mutational burden (TMB).

3138 Background: TMB is routinely reported in cancer patients tested with broad-panel next generation sequencing and has become a predictive biomarker associated with response to checkpoint inhibitor (CPI) therapy. Sequencing of paired tumor and normal specimens allows correction of TMB estimates with patient-specific germline variants. When a paired normal specimen is unavailable, TMB estimates are corrected using germline variant annotations derived from population-scale germline variant surveys. Germline variants do not generate neoantigens, which is the putative target of the immune response in CPI treated patients. To evaluate TMB differences in paired sequencing (PS) and tumor-only sequencing (TOS), we compared TMB assessments—stratified by race—in two common malignancies. Methods: Using de identified records from the Tempus clinico-genomic database, cohorts of patients with non-small cell lung cancer (NSCLC) and breast cancer sequenced using the Tempus xT NGS platform (DNA-seq of 595-648 genes at 500x coverage, whole exome capture RNA-seq) and noted to not have microsatellite instability, were identified for analyses. The Kruskall-Wallis test was used to compare TMB distributions. Results: Among 4,817 NSCLC patients with race information (13% Black (B), 5% Asian (A), 82% White (W), 3,052 had PS, and 1,765 had TOS performed. Median TMB for B, A, and W patients was 5.8, 2.6, and 4.7 (within group p &lt; 0.0001), respectively in patients with PS, and 9.5, 6, and 7.4 (within group p &lt; 0.0001), in patients with TOS. Comparisons across PS and TOS were highly significant (p &lt; 0.0001). The absolute difference in median TMB was 3.7, 3.4, and 2.5, respectively. Among 3,191 patients with breast cancer (17% B, 4% A, 78% W), 2,220 had PS, and 971 had TOS. Median TMB for B, A, and W patients was 2.6, 2.1, and 2.6 (within group p = 0.11), respectively, in patients with PS, and 6.3, 5.8, and 4.7 (within group p &lt; 0.0001) in patients with TOS. Comparisons across PS and TOS were highly significant (p &lt; 0.0001). The absolute difference in median TMB was 3.7, 3.7, and 2.1, respectively. Conclusions: PS reduces estimated TMB compared to TOS across all racial groups with a pronounced difference in Black and Asian racial groups. This is expected as population databases of germline variation are based on cohorts predominantly from individuals of European ancestry, leading to artifactually high TMB in minorities tested with TOS. As a result, artifactually elevated TMB estimates from TOS may promote treatment with CPI in patients with a low probability of response which could exacerbate known race-based outcome disparities. PS provides a more accurate estimate of TMB regardless of race and could reduce the use of CPI in patients with a low likelihood of response.

Development of a Novel Reference Material for Tumor Mutational Burden Measurement Based on CRISPR/Cas9 Technology.

As a biomarker that affects treatment decisions of immune checkpoint inhibitors, the accuracy, reliability, and comparability of tumor mutational burden (TMB) estimation is of paramount importance. To improve the consistency and reliability of these tests, qualified reference materials providing ground-truth data are crucial. In this study, we developed a set of formalin-fixed and paraffin-embedded (FFPE) samples with different TMB values as the novel reference materials for TMB estimation. By introducing several clinically relevant variants in MutS Homolog 2 (MSH2) gene and DNA polymerase epsilon (POLE) gene into human cell lines using CRISPR/Cas9 technology, we first constructed four typical cell lines which verified with hypermutator or ultramutator phenotype. Followed by cell mixing and paraffin embedding, the novel FFPE samples were prepared. It was confirmed that our novel FFPE samples have sufficient quantity of cells, high reproducibility, and they can provide matched wild type sample as the genetic background. The double-platform whole exome sequencing validation showed that our FFPE samples were also highly flexible as they containing different TMB values spanning a clinically relevant range (2.0–106.1 mut/Mb). Without limitations on production and TMB values, our novel FFPE samples based on CRISPR/Cas9 editing are suitable as candidate reference materials. From a practical point of view, these samples can be used for the validation, verification, internal quality control, and proficiency testing of TMB assessment.

Optimizing the evaluation of gene-targeted panels for tumor mutational burden estimation

Though whole exome sequencing (WES) is the gold-standard for measuring tumor mutational burden (TMB), the development of gene-targeted panels enables cost-effective TMB estimation. With the growing number of panels in clinical trials, developing a statistical method to effectively evaluate and compare the performance of different panels is necessary. The mainstream method uses R-squared value to measure the correlation between the panel-based TMB and WES-based TMB. However, the performance of a panel is usually overestimated via R-squared value based on the long-tailed TMB distribution of the dataset. Herein, we propose angular distance, a measurement used to compute the extent of the estimated bias. Our extensive in silico analysis indicates that the R-squared value reaches a plateau after the panel size reaches 0.5 Mb, which does not adequately characterize the performance of the panels. In contrast, the angular distance is still sensitive to the changes in panel sizes when the panel size reaches 6 Mb. In particular, R-squared values between the hypermutation-included dataset and the non-hypermutation dataset differ widely across many cancer types, whereas the angular distances are highly consistent. Therefore, the angular distance is more objective and logical than R-squared value for evaluating the accuracy of TMB estimation for gene-targeted panels.

Aligning tumor mutational burden (TMB) quantification across diagnostic platforms: phase II of the Friends of Cancer Research TMB Harmonization Project

Tumor mutational burden (TMB) measurements aid in identifying patients who are likely to benefit from immunotherapy; however, there is empirical variability across panel assays and factors contributing to this variability have not been comprehensively investigated. Identifying sources of variability can help facilitate comparability across different panel assays, which may aid in broader adoption of panel assays and development of clinical applications. Twenty-nine tumor samples and 10 human-derived cell lines were processed and distributed to 16 laboratories; each used their own bioinformatics pipelines to calculate TMB and compare to whole exome results. Additionally, theoretical positive percent agreement (PPA) and negative percent agreement (NPA) of TMB were estimated. The impact of filtering pathogenic and germline variants on TMB estimates was assessed. Calibration curves specific to each panel assay were developed to facilitate translation of panel TMB values to whole exome sequencing (WES) TMB values. Panel sizes >667 Kb are necessary to maintain adequate PPA and NPA for calling TMB high versus TMB low across the range of cut-offs used in practice. Failure to filter out pathogenic variants when estimating panel TMB resulted in overestimating TMB relative to WES for all assays. Filtering out potential germline variants at >0% population minor allele frequency resulted in the strongest correlation to WES TMB. Application of a calibration approach derived from The Cancer Genome Atlas data, tailored to each panel assay, reduced the spread of panel TMB values around the WES TMB as reflected in lower root mean squared error (RMSE) for 26/29 (90%) of the clinical samples. Estimation of TMB varies across different panels, with panel size, gene content, and bioinformatics pipelines contributing to empirical variability. Statistical calibration can achieve more consistent results across panels and allows for comparison of TMB values across various panel assays. To promote reproducibility and comparability across assays, a software tool was developed and made publicly available.

Development and Validation of StrataNGS, a Multiplex PCR, Semiconductor Sequencing-Based Comprehensive Genomic Profiling Test

Despite widespread use in targeted tumor testing, multiplex PCR/semiconductor (Ion Torrent) sequencing-based assessment of all comprehensive genomic profiling (CGP) variant classes has been limited. Herein, we describe the development and validation of StrataNGS, a 429-gene, multiplex PCR/semiconductor sequencing-based CGP laboratory-developed test performed on co-isolated DNA and RNA from formalin-fixed, paraffin-embedded tumor specimens with ≥2 mm2 tumor surface area. Validation was performed in accordance with MolDX CGP validation guidelines using 1986 clinical formalin-fixed, paraffin-embedded samples and an in-house developed optimized bioinformatics pipeline. Across CGP variant classes, accuracy ranged from 0.945 for tumor mutational burden (TMB) status to >0.999 for mutations and gene fusions, positive predictive value ranged from 0.915 for TMB status to 1.00 for gene fusions, and reproducibility ranged from 0.998 for copy number alterations to 1.00 for splice variants and insertions/deletions. StrataNGS TMB estimates were highly correlated to those from whole exome- or FoundationOne CDx-determined TMB (Pearson r=0.998 and 0.960, respectively); TMB reproducibility was 0.996 (concordance correlation coefficient). Limit of detection for all variant classes was <20% tumor content. Together, we demonstrate that multiplex PCR/semiconductor sequencing-based tumor tissue CGP is feasible using optimized bioinformatic approaches described herein.

Tumor mutational burden and purity adjustment before and after treatment with temozolomide in 27 paired samples of glioblastoma: a prospective study.

Treatment of glioblastoma (GBM) remains a challenging task, with limited treatment options, none offering a cure. Immune therapy has proven effective across different cancers with remarkable response rates. Tumor mutational burden (TMB) is a marker of response, but technical and methodological differences in TMB estimates have made a proper assessment and comparison challenging. Here, we analyzed a prospective collection of paired samples from 35 patients with newly diagnosed GBM, all of whom were wild‐type (WT) for isocitrate dehydrogenase, before and after treatment with radiotherapy and temozolomide. Seven patients (20%) had O6‐methylguanine‐DNA methyltransferase‐methylated tumors. Six patients (17%) had two relapse surgeries, and tissue from all three surgeries was collected. We found that accurate evaluation of TMB was confounded by high variability in the cancer cell fraction of relapse samples. To ameliorate this, we developed a model to adjust for tumor purity based on the relative density distribution of variant allele frequencies in each primary–relapse pair. Additionally, we examined the mutation spectra of shared and private mutations. After tumor purity adjustment, we found TMB comparison reliable in tumors with tumor purity between 15% and 40%, resulting in 27/35 patients (77.1%). TMB remained unchanged from 0.65 mutations per megabase (Mb) to 0.67/Mb before and after treatment, respectively. Examination of the mutation spectra revealed a dominance of C > T transitions at CpG sites in both shared and relapse‐private mutations, consistent with cytosine deamination and the clock‐like mutational signature 1. We present and apply a cellularity correction approach that enables more accurate assessment of TMB in paired tumor samples. We did not find a significant increase in TMB after correcting for cancer cell fraction. Our study raises significant concerns when determining TMB. Although a small sample size, corrected TMB can have a clinical significance when stratifying patients to experimental treatment, for example, immune checkpoint therapy.

DNA Damage Repair Gene Set as a Potential Biomarker for Stratifying Patients with High Tumor Mutational Burden

Simple SummaryImmunotherapy has been a promising therapeutic approach for cancer treatment in recent years. Although cancer immunotherapy has achieved remarkable success, treatment response is only observed in a small number of patients. As nonresponders need to endure high treatment costs and toxicities with little benefit from treatment, identifying potential predictive biomarkers is critical to optimize the benefits of immunotherapy in patients. The total number of mutations in the tumor genome is a useful biomarker. Patients with a large number of mutations tend to respond better to cancer immunotherapy. However, assessment of the total number of mutations may not be easy. In this study, we identified gene sets with only a small number of genes whose mutations serve as an indicator of the total number of mutations. These cancer-specific gene sets can be used as a cost-effective approach to stratify patients with a large number of mutations in clinical practice.Tumor mutational burden (TMB) is a promising predictive biomarker for cancer immunotherapy. Patients with a high TMB have better responses to immune checkpoint inhibitors. Currently, the gold standard for determining TMB is whole-exome sequencing (WES). However, high cost, long turnaround time, infrastructure requirements, and bioinformatics demands have prevented WES from being implemented in routine clinical practice. Panel-sequencing-based estimates of TMB have gradually replaced WES TMB; however, panel design biases could lead to overestimation of TMB. To stratify TMB-high patients better without sequencing all genes and avoid overestimating TMB, we focused on DNA damage repair (DDR) genes, in which dysfunction may increase somatic mutation rates. We extensively explored the association between the mutation status of DDR genes and TMB in different cancer types. By analyzing the mutation data from The Cancer Genome Atlas, which includes information for 33 different cancer types, we observed no single DDR gene/pathway in which mutation status was significantly associated with high TMB across all 33 cancer types. Therefore, a computational algorithm was proposed to identify a cancer-specific gene set as a surrogate for stratifying patients with high TMB in each cancer. We applied our algorithm to skin cutaneous melanoma and lung adenocarcinoma, demonstrating that the mutation status of the identified cancer-specific DDR gene sets, which included only 9 and 14 genes, respectively, was significantly associated with TMB. The cancer-specific DDR gene set can be used as a cost-effective approach to stratify patients with high TMB in clinical practice.

Sputum supernatant as a viable alternative for liquid biopsy in advanced non-small cell lung cancer.

3053 Background: Sputum has previously attracted attention as a potential medium for detecting genetic and epigenetic alterations in early-stage lung cancer. Recent progress in advanced non-small cell lung cancer (aNSCLC) has highlighted plasma as an option in scenarios where tissue specimens are unavailable. In this study, we investigated the feasibility of mutation profiling with sputum from aNSCLC patients. Methods: Matched tissue (TIS), whole blood, and either induced or spontaneous sputum were collected from 42 prospectively enrolled, previously untreated patients with stage IIIB-IV NSCLC. Eight and two milliliters of whole blood and sputum were respectively used for preparation of plasma (PLA) and sputum supernatant (SPU). Capture-based targeted sequencing of TIS, PLA and SPU samples was performed with a panel of 520 cancer-related genes. Results: The cohort had a median age of 59 (range 26-79) with a majority of men (31/42). Thirty-five patients were diagnosed with lung adenocarcinoma (LUAD) and the remaining seven with squamous cell carcinoma (LUSC). There were more non-smokers (26/42) than smokers. Most patients (34/42) had metastatic disease. First, we assessed the performance of three sample types in terms of key quality control indices. PLA and SPU showed maximum allelic fraction that were both significantly smaller than that of TIS. Subsequently compared was positive detection rate, TIS, PLA, and SPU achieved respective rates of 92.9%, 71.4%, and 59.5%, revealing comparable levels between PLA and SPU samples (p = 0.36). Rates of detection were highly similar between induced (54.6%) and spontaneous sputum (61.3%), but were higher in LUSC (85.7%) than in LUAD patients (54.3%) although this difference was not statistically significant (p = 0.21). Furthermore, detection rates rose in smokers (81.3% for PLA and SPU, 100% for TIS). Next, using TIS as the gold standard, we compared the sensitivity of detecting eight established driver alterations in NSCLC with the two liquid biopsies. SPU and TIS achieved respective levels of 50.0% and 70.0%, which, similar to detection rate, improved to 88.9% (8/9) and 77.8% (7/9) in the smokers subgroup. Importantly, detection of altered KRAS and ALK was more sensitive with SPU (6/6, 2/3, entire cohort) than with PLA (5/6, 1/3). For EGFR mutation, which occurred mainly in nonsmokers, the sensitivity was considerably lower in SPU (4/14, entire cohort) than in PLA (12/14). Detection of mutant MET, however, appeared difficult for both biopsies (0/4). In addition, there was a high level of correlation between estimates of tumor mutational burden based on PLA and on SPU (Pearson r2 = 0.58). Conclusions: Our study demonstrated that for aNSCLC patients, sputum supernatant demonstrates comparable performance as plasma. Sputum is therefore a viable alternative for profiling for established driver alterations in aNSCLC, especially in smokers.

Influence of low tumor content on tumor mutational burden estimation by whole-exome sequencing and targeted panel sequencing.

BackgroundTumor mutational burden (TMB) is a promising biomarker for stratifying patient subpopulation who would benefit from immune checkpoint blockade (ICB) therapies. Although great efforts have been made for standardizing TMB measurement, mutation calling and TMB quantification can be challenging in samples with low tumor content including liquid biopsies. The effect of varying tumor content on TMB estimation by different assay methods has never been systematically investigated.MethodWe established a series of reference standard DNA samples derived from 11 pairs of tumor–normal matched human cell lines across different cancer types. Each tumor cell line was mixed with its matched normal at 0% (control), 1%, 2%, 5%, and 10% mass‐to‐mass ratio to mimic the clinical samples with low tumor content. TMB of these reference standards was evaluated by both ∼1000× whole‐exome sequencing (wesTMB) and targeted panel sequencing (psTMB) at four different vendors. Both regression and classification analyses of TMB were performed for theoretical investigation and clinical practice purposes.ResultsLinear regression model was established that demonstrated in silico psTMB determined by regions of interest (ROI) as a great representative of wesTMB based on TCGA dataset. It was also true in our reference standard samples as the predicted psTMB interval based on the observed wesTMB captured the intended 90% of the in silico psTMB values. Although ∼1000× deep WES was applied, reference standard samples with less than 5% of tumor proportions are below the assay limit of detection (LoD) of wesTMB quantification. However, predicted wesTMB based on observed psTMB accurately classify (>0.97 AUC) for TMB high and low patient stratification even in samples with 2% of tumor content, which is more clinically relevant, as TMB determination should be a qualitative assay for TMB high and low patient classification. One targeted panel sequencing vendor using an optimized blood psTMB pipeline can further classify TMB status accurately (>0.82 AUC) in samples with only 1% of tumor content.ConclusionsWe developed a linear model to establish the quantitative correlation between wesTMB and psTMB. A set of DNA reference standards was produced in aid to standardize TMB measurements in samples with low tumor content across different targeted sequencing panels. This study is a significant contribution aiming to harmonize TMB estimation and extend its future application in clinical samples with low tumor content including liquid biopsy.

Tumor Mutational Burden as a Predictive Biomarker in Solid Tumors.

Tumor mutational burden (TMB), defined as the number of somatic mutations per megabase of interrogated genomic sequence, varies across malignancies. Panel sequencing-based estimates of TMB have largely replaced whole-exome sequencing-derived TMB in the clinic. Retrospective evidence suggests that TMB can predict the efficacy of immune checkpoint inhibitors, and data from KEYNOTE-158 led to the recent FDA approval of pembrolizumab for the TMB-high tumor subgroup. Unmet needs include prospective validation of TMB cutoffs in relationship to tumor type and patient outcomes. Furthermore, standardization and harmonization of TMB measurement across test platforms are important to the successful implementation of TMB in clinical practice. SIGNIFICANCE: Evaluation of TMB as a predictive biomarker creates the need to harmonize panel-based TMB estimation and standardize its reporting. TMB can improve the predictive accuracy for immunotherapy outcomes, and has the potential to expand the candidate pool of patients for treatment with immune checkpoint inhibitors.

A novel tumor mutational burden estimation model as a predictive and prognostic biomarker in NSCLC patients

BackgroundTumor mutational burden (TMB) has both prognostic value in resected non-small cell lung cancer (NSCLC) patients and predictive value for immunotherapy response. However, TMB evaluation by whole-exome sequencing (WES) is expensive and time-consuming, hampering its application in clinical practice. In our study, we aimed to construct a mutational burden estimation model, with a small set of genes, that could precisely estimate WES-TMB and, at the same time, has prognostic and predictive value for NSCLC patients.MethodsTMB estimation model was trained based on genomic data from 1056 NSCLC samples from The Cancer Genome Atlas (TCGA). Validation was performed using three independent cohorts, including Rizvi cohort and our own Asian cohorts, including 89 early-stage and n late-stage Asian NSCLC patients, respectively. TCGA data were obtained on September 3, 2018. The two Asian cohort studies were performed from September 1, 2018, to March 5, 2019. Pearson’s correlation coefficient was used to assess the performance of estimated TMB with WES-TMB. The Kaplan-Meier survival analysis was applied to evaluate the association of estimated TMB with disease-free survival (DFS), overall survival (OS), and response to anti-programmed death-1 (PD-1) and anti-programmed death-ligand 1 (PD-L1) therapy.ResultsThe estimation model, consisted of only 23 genes, correlated well with WES-TMB both in the training set of TCGA cohort and validation set of Rizvi cohort and our own Asian cohort. Estimated TMB by the 23-gene panel was significantly associated with DFS and OS in patients with early-stage NSCLC and could serve as a predictive biomarker for anti-PD-1 and anti-PD-L1 treatment response.ConclusionsThe 23-gene panel, instead of WES or the currently used panel-based methods, could be used to assess the WES-TMB with a high relevance. This customized targeted sequencing panel could be easily applied into clinical practice to predict the immunotherapy response and prognosis of NSCLC.

Abstract 2008: Pan-cancer profiling of the effect of biopsied site on tumor mutational burden measurements in a real-world data cohort

Abstract Background: Tumor mutational burden (TMB) has been associated with efficacy of immune checkpoint therapy and is being investigated as a potential biomarker in several tumor types. The foundational studies evaluating TMB in clinical trials have not examined factors that may affect TMB estimates in real-world practice, such as the site of biopsy relative to the tumor tissue of origin. Methods: We used the Flatiron Health/Foundation Medicine Clinico-Genomic Database (Presley CJ, et al. JAMA 2018; Singal G, et al. JAMA 2019), which contains deidentified patient-level clinical records from Flatiron Health (New York, NY), combined with genomic sequencing test results of tissue TMB from Foundation Medicine (Cambridge, MA). We examined patient records from 7 advanced cancer types: non-small cell lung cancer (NSCLC), breast cancer (BC), gastric/esophageal cancer (G/EC), melanoma (MEL), metastatic castration-resistant prostate cancer (CRPC), head and neck cancer (H&amp;N), and small cell lung cancer (SCLC). We performed a multivariate mixed-model linear regression to test for associations between the site of tissue biopsy and TMB measurement. We used a cutoff of 10 mutations per megabase (mut/Mb) to classify specimens as high TMB. Results: TMB data were available for 11,928 biopsy specimens from 11,567 patients: NSCLC, n = 5556; BC, n = 3381; G/EC, n = 1335; MEL, n = 528; CRPC, n = 506; H&amp;N, n = 342; SCLC, n = 280. In multiple cancer types, we found that TMB estimates were significantly associated with the biopsied site. For example, brain metastasis specimens from patients with NSCLC had higher TMB (median 11.3 mut/Mb) than specimens collected from primary lung tissue (median 7.0 mut/Mb) or from pleura or pleural fluid (median 3.5 mut/Mb). TMB measurements were higher in brain metastases than in samples from the primary tissue across multiple cancer types. In contrast, biopsies from lymph nodes showed different TMB patterns depending on the cancer type. Consistent with those results, the proportion of patients with high TMB differed significantly by biopsied site. Conclusion: We used a real-world database of clinical data linked with genomic tests to evaluate TMB values across tissue biopsy sites for multiple cancer types. Our results highlight the biopsied site as a potential source of variation for TMB results measurements in a cancer type–specific manner. This will be important to consider when performing research in heterogenous data sets. Citation Format: Simon Papillon-Cavanagh, Chuan Gao, G. Celine Han, Alice M. Walsh. Pan-cancer profiling of the effect of biopsied site on tumor mutational burden measurements in a real-world data cohort [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2008.