Evaluation of tissue- and plasma-derived tumor mutational burden (TMB) and genomic alterations of interest in CheckMate 848, a study of nivolumab combined with ipilimumab and nivolumab alone in patients with advanced or metastatic solid tumors with high TMB

Abstract

Background: An accumulation of somatic mutations in tumors leads to increased neoantigen levels and antitumor immune response. Tumor mutational burden (TMB) reflects the rate of somatic mutations in the tumor genome, as determined from tumor tissue (tTMB) or blood (bTMB). While high tTMB is a biomarker of immune checkpoint inhibitor (ICI) treatment efficacy, few studies have explored the clinical utility of bTMB, a less invasive alternative for TMB assessment. Establishing the correlation between tTMB and bTMB would provide insight into whether bTMB is a potential substitute for tTMB. We explored the tumor genomes of patients enrolled in CheckMate 848 with measurable TMB. The correlation between tTMB and bTMB, and the factors affecting it, were evaluated.

Methods: In the phase 2 CheckMate 848 (NCT03668119) study, immuno-oncology-naïve patients with advanced, metastatic, or unresectable solid tumors and tTMB-high or bTMB-high (≥10 mut/Mb) were prospectively randomized 2:1 to receive nivolumab plus ipilimumab or nivolumab monotherapy. Tissue and plasma DNA sequencing was performed using the Foundation Medicine FoundationOne CDx and bTMB Clinical Trial Assays, respectively. tTMB was quantified from coding variants, insertions, and deletions, and bTMB from somatic base substitutions. Correlations between tTMB and bTMB were determined across samples and with respect to maximum somatic allele frequency (MSAF). Assay agreement and variant composition were also evaluated.

Results: A total of 1,438 and 1,720 unique tissue and blood samples, respectively, were obtained from 1,954 patients and included >100 screened disease ontologies, with 1,017 unique pairs of tTMB and bTMB measurements available for assessment. Median tTMB and bTMB were 3.8 and 3.5 mut/Mb, respectively. A significant correlation between tTMB and bTMB (r=0.48, p<0.0001) was observed across all sample pairs, which increased to r=0.54 (p<0.0001) for samples with MSAF≥1%. Assay concordance was highest for samples with MSAF≥10% across multiple disease ontologies and observed for both responders and non-responders to ICI therapy. The variants contributing to tTMB and bTMB were similar.

Conclusions: We observed that tTMB and bTMB had a statistically significant correlation, particularly for samples with high MSAF, and that this correlation applied across disease ontologies. Further investigation into the clinical utility of bTMB is warranted.

Keywords: Biomarkers, Tumor; Immunotherapy; Ipilimumab; Nivolumab.

Figure 1

Distribution of tTMB and bTMB. Each box shows the IQR, and horizontal lines represent median values. Whiskers represent values 1.5× the upper and lower limits of the IQR. ^aTo accommodate TMB values of 0 for taking logarithms, 0.5 was added to all TMB values prior to taking the logarithm, for plotting purposes. bTMB, blood tumor mutational burden; mut/Mb, mutations/megabase; TMB, tumor mutational burden; tTMB, tissue tumor mutational burden.

Figure 2

Correlation of bTMB and tTMB for (A) all sample pairs; (B) sample pairs with MSAF≥1%; or (C) sample pairs with MSAF<1%. ^aRight-hand panel shows enlargement of the left-hand panel. Shaded area around the regression line in the right-hand panel shows the 95% CIs for Passing–Bablok regression analysis. bTMB, blood tumor mutational burden; MSAF, maximum somatic allele frequency; r, Spearman’s correlation coefficient; tTMB, tissue tumor mutational burden.

Figure 3

Correlations between tTMB and bTMB by disease ontology for (A) samples with MSAF≥1% and for (B) samples with MSAF≥1% to <10% versus ≥10%. Number of sample pairs is shown inside each bar. ^aData are shown for the top five most frequent disease ontologies among samples with MSAF≥1%. bTMB, blood tumor mutational burden; CUP, carcinoma of unknown primary; MSAF, maximum somatic allele frequency; tTMB, tissue tumor mutational burden.

Figure 4

Assessment of TMB status between tissue-based and blood-based assays by disease ontology, for (A) samples with MSAF≥1% and (B) samples with MSAF≥1% to <10% versus ≥10%. Number of sample pairs is shown inside each bar. ^aData are shown for the top five most frequent disease ontologies among samples with MSAF≥1%. CUP, carcinoma of unknown primary; MSAF, maximum somatic allele frequency; PPA, positive percentage agreement; TMB, tumor mutational burden.

Figure 5

Concordance between tTMB and bTMB status among responders and non-responders to ICI therapy.^{a a}Patients with either tTMB-H or bTMB-H status who were randomized in the CheckMate 848 trial. Samples were pooled across the nivolumab+ipilimumab and nivolumab monotherapy treatment arms. BOR, best overall response; bTMB, blood tumor mutational burden; bTMB-H, high blood tumor mutational burden; CR, complete response; ICI, immune checkpoint inhibitor; MSAF, maximum somatic allele frequency; NE, not evaluable (for response); PD, progressive disease; PR, partial response; SD, stable disease; tTMB, tissue tumor mutational burden; tTMB-H, high tissue tumor mutational burden.