Tumor mutational burden, also referred to as tumor mutational load, is a measure of the number of mutations in tumor tissue taken from a patient. Recently, several studies have been investigating the utility of TMB as a biomarker to predict response to therapy in patients with non–small cell lung cancer.
Tumor mutational burden (TMB), also referred to as tumor mutational load, is a measure of the number of mutations in tumor tissue taken from a patient. Recently, several studies have been investigating the utility of TMB as a biomarker to predict response to therapy in patients with nonsmall cell lung cancer (NSCLC).
The Lung Adjuvant Cisplatin Evaluation (LACE)- Bio-II trial evaluated the prognostic value of copy number alterations and somatic mutations in tumor tissue resected from patients with NSCLC who had been treated in 3 landmark clinical trials. LACE-Bio-II also looked at the predictive value of these alterations in guiding adjuvant therapy. A targeted panel of 1538 genes was used to sequence 908 NSCLC specimens from 453 patients who had undergone resection alone and 455 patients who had undergone resection plus adjuvant chemotherapy. Gene mutations were correlated to disease-free survival (DFS), overall survival (OS), and lung cancerspecific survival.1
Nonsynonymous mutationsnucleotide mutations that alter the amino acid sequence of a protein—were identified in 1515 genes in the 908 samples. Tumors were categorized into tertiles of low (≤4 mutations per megabase [Mb]), intermediate (>4 to ≤8 mutations/Mb), or high (>8 mutations/Mb) nonsynonymous TMB. High TMB was prognostic for better DFS, OS, and lung cancerspecific survival; low TMB was associated with poor outcomes for DFS (P= .007), OS (P= .016), and lung cancerspecific survival (P= .001). Better lung cancerspecific survival, but not DFS or OS, was associated with low TMB in patients who received adjuvant chemotherapy.1
The presence of mutations in DNA repair pathways, tumor-infiltrating lymphocytes,TP53alteration subtype, and intratumor heterogeneity were not found to predict outcomes. Statistically significant effects of mutations in individual genes could not be determined due to high false-discovery rates. Other limitations of the study include the inference of TMB from a limited gene panel, as well as the underrepresentation of tumor tissue from individuals who never smoked.1As in the CheckMate 026 trial,2no correlation was seen in LACE-Bio-II between TMB and PD-L1 expression. The study authors concluded that further studies should be done to confirm their findings and define the underlying mechanisms.1
In an accompanying editorial,3Yasushi Goto, MD, of the Department of Respiratory Medicine, National Cancer Center Hospital in Tokyo, Japan, said the comprehensiveness and large data set of the LACE-Bio-II study were “noteworthy,” agreeing that validation of the correlation of TMB with response and survival is needed.
Although tumor mutations can increase the immunogenicity of tumors, TMB is not a direct measure of immunogenicity; it can only indirectly predict the efficacy of immunotherapy. Until the correlation between TMB and immunogenicity is understood, Goto said, the cutoff scores described in the LACE-Bio-II study should be considered arbitrary and applied in the clinic with caution. Before TMB can be considered a more reliable predictive marker, additional “biologic guidance or computational assistance,” as well as validated assays and cutoff points, are required.3
Dennis A. Lowenthal, MD, medical director of the Carol G. Simon Cancer Center and section chief of hematology and oncology at Overlook Medical Center in Summit, New Jersey, agreed that the findings from the LACE-Bio-II study are not practice-changing.
“What will be practice-changing are the CheckMate 568, CheckMate 026, and CheckMate 227 multi-arm, prospective, randomized studies in the advanced lung cancer setting. The takeaway point from all of those studies is that a very high TMB clearly impacts the response to immunotherapy [TABLE]. High TMB and high PD-L1 expression are 2 independent variables because you get many patients who have a high TMB and low PD-L1 and vice versa. Those with a high TMB and a high PD-L1 will likely do well with single-agent immunotherapy with a PD-1 or PD-L1 inhibitor. Those with a high TMB but a low PD-L1 will probably do well with combined immunotherapy,” Lowenthal said in an interview with Targeted Therapies in Oncology, noting that combining immunotherapies in NSCLC has not yet been approved.
CheckMate 026 (NCT02041533), a phase III trial, had shown that nivolumab (n = 211) was not associated with significantly longer progression-free survival (PFS) than chemotherapy (n = 212) in patients with untreated stage IV or recurrent NSCLC with a PD-L1 expression level of ≥5%.2In an exploratory analysis, the response rate in patients with a high TMB was better in those treated with nivolumab than with chemotherapy (47% vs 28%), and PFS was longer (9.7 vs 5.8 months; HR, 0.62; 95% CI, 0.38-1.00); however, OS was similar between groups, possibly due to crossover from chemotherapy to nivolumab.2
There was no correlation between TMB and PD-L1 levels; however, among patients treated with nivolumab, those with both high TMB and PD-L1 expression ≥50% had a response rate of 75% compared with those who had only high TMB (32%), high PD-L1 (34%), or neither (16%).
CheckMate 568 (NCT02659059), a phase II study of nivolumab (Opdivo) plus ipilimumab (Yervoy) as first-line NSCLC treatment, used the FoundationOne CDx (Foundation Medicine) assay to assess TMB and identify an appropriate TMB cutoff in 288 patients with chemotherapy-naïve stage IV NSCLC that was not EGFR- or ALK-targetable. At a minimum 3-month follow-up, a higher TMB was associated with higher overall response to nivolumab plus ipilimumab. In patients with a TMB of ≥10 mutations per Mb, the objective response rate was 44% and in patients with a TMB of ≥15 mutations per Mb, 39%.4
The TMB cutoff identified in CheckMate 568 of ≥10 mutations per Mb was subsequently validated in a preplanned analysis in CheckMate 227 (NCT02477826), a multipart phase III trial that evaluated PFS in patients with stage IV or recurrent chemotherapy-naïve NSCLC and high TMB treated with nivolumab plus ipilimumab.
Of the 1004 patients with evaluable TMB, 444 (44.2%) had TMB ≥10 mutations per Mb; among these patients, 139 were assigned to nivolumab plus ipilimumab and 160 to chemotherapy alone.
In a coprimary analysis, PFS was significantly longer in patients with high TMB who received combination immunotherapy (n = 139; 43% progression free at 1 year) than in those treated with chemotherapy (n = 160; 13% progression free at 1 year), irrespective of PD-L1 expression, supporting the independence of TMB and PD-L1 as biomarkers. The authors concluded that the study validates the role of TMB as a biomarker for patient selection for combination immunotherapy.5
Timothy Chan, MD, PhD, director, Immunogenomics and Precision Oncology Program, Memorial Sloan Kettering Cancer Center, said the results from these multiple phase III trials show that TMB is an independent, easy-to-measure predictor. “I think that compared to PD-L1 staining, there is a lot more validation at the present time [for TMB].”
A liquid biopsy (blood-based assay) was developed to assess TMB in patients with NSCLC. Using blood samples is less invasive and easier to repeat than testing tumor biopsy samples and can be performed in patients who do not have sufficient tumor tissue for genomic analysis. This liquid biopsy therefore addresses an unmet need of patients with NSCLC who require testing for PD-L1 expression to support the use of checkpoint inhibitor therapy. The methodology was the same as the FoundationOne CDx assay.6
In this retrospective study, blood samples were analyzed from patients with locally advanced or metastatic NSCLC who had received ≥2 lines of therapy in 2 randomized studies comparing atezolizumab (Tecentriq) with docetaxel (Taxotere). Samples (211 biomarker-evaluable) from POPLAR (NCT01903993), a phase II study, were used to identify a range of blood TMB cutoff points correlated with clinically meaningful outcomes (training set); these were confirmed with samples (583 biomarker-evaluable) from OAK (NCT02008227), a phase III study (validation set).6
TMB levels in blood samples correlated with those in the tissue. The liquid biopsy test results were predictive of benefit, including improved response to and PFS following treatment with the PD-L1 inhibitor atezolizumab.6
The liquid TMB assay requires a minimum amount of circulating tumor DNA (ctDNA) in the blood, and the tumor ctDNA shed into the blood must contain mutations with an allele frequency of ≥1% for a valid blood TMB score.6According to a news release from UC Davis Health, Foundation Medicine is seeking FDA approval to incorporate the TMB liquid assay into its FoundationACT liquid biopsy, which has received a breakthrough therapy designation.7The liquid TMB assay is being validated in a prospective phase III trial in first-line NSCLC therapy.6
Chan said that although FoundationOne CDx has shown clinical utility in large clinical trials, and Foundation Medicine has been very aggressive in bringing this to the clinic, “technologically it is not the best diagnostic.” He predicts the technology in general will improve over time.
Lowenthal said that he is already using TMB in practice. “TMB still needs to be better defined. If you look at LACE-Bio-II, they looked at a genomic panel of over 1500 mutations in calculating their TMB. If you look at the CheckMate studies, all have used FoundationOne CDx, which looks at a 324-gene mutation panel. Going forward what needs to happen is some standardization of PD-L1 testing across the board, number 1. Number 2, are there certain mutations that are more immunogenic than others? Are those the ones that are critical to TMB for us to use as biomarkers for response to therapy?”
“I think going forward, the studies that need to be done, which are being done, are studies in the adjuvant setting using immunotherapy and stratifying for PD-L1 and TMB, and perhaps other biomarkers, and studies also looking in the neoadjuvant setting,” Lowenthal said.
Chan views TMB as a platform of multiple biomarkers that can be extracted from next-generation sequencing (NGS). “In my opinion, [TMB] helped to reinvigorate the whole NGS field. It’s being reported that only about 15% of all tumors have actionable mutations. Look at what’s going on now with mutational load and microsatellite instability, which are related. You suddenly have the utility of looking for something in NGS [that’s] actionable. It is, I think, reinvigorating the NGS utility for these assays.”