The degree of tumor mutational burden can predict whether patients with non–small cell lung cancer will benefit from immunotherapy with a checkpoint inhibitor, but differences in TMB by tumor histology and metastatic sites suggest that clinicians still don’t fully understand the implications of this marker for clinical care, investigators say.
A study of nearly 3,500 unmatched non–small cell lung cancer (NSCLC) samples showed that TMB of 10 mutations per megabase (MB) or higher occurred more frequently in metastases (especially brain metastases) than in primary tumors, in squamous cell carcinomas (SCC), compared with adenocarcinomas, and in poorly differentiated tumors, compared with moderately or well-differentiated tumors, reported Matthew K. Stein, MD, of the University of Tennessee Health Sciences Center in Memphis and colleagues in JCO Precision Oncology.
“Further work is needed to elucidate the mutational differences implied by TMB with the hope of identifying additional ICI [immune checkpoint inhibitor] biomarkers, as well as exploring unique therapeutic approaches for brain metastases and other sites with high TMB,” they wrote.
To get a clearer picture of differences in TMB by biopsy sample location, histology, and other biomarkers, the investigators obtained 3,424 unique, unmatched specimens from multiple U.S. cancer centers. The specimens, including 2,351 adenocarcinomas and 1,073 SCCs, were identified by only age and sex, with histology confirmed by pathology review.
The investigators conducted next-generation sequencing of 592 cancer-related genes on each specimen, as well as programmed death ligand 1 (PD-L1) immunohistochemistry, and TMB. They used a TMB cutoff of 10 mutations/MB, which was established as predictive of response to immune checkpoint inhibitors in the Checkmate 568 and Checkmate 227 trials.
They found that 38% of metastatic adenocarcinoma samples had TMB of 10 mutations/MB or greater, compared with 25% of primary adenocarcinomas (P less than .001), a difference that was greatest for brain metastases, compared with other metastases (61% vs. 35%; P less than .001). The median TMBs were 13 mutations/MB vs. 6 mutations/MB, respectively.
Similarly, TMB above the cutoff was more frequent in SCC metastases than primaries (41% vs. 35%; P = .038).
TMB of 10 mutations/MB occurred significantly more often in SCC primary tumors (35%) than in adenocarcinoma primaries (25%; P less than .001).
They also saw variability in TMB among other sites, with adrenal metastases being the second–most likely site to have high TMB (51%) and bone metastases being least likely (19%).
In an analysis stratified by histology and primary or metastatic sample site, poorly differentiated tumors were significantly more likely to have TMB of 10 mutations/MB or greater both for adenocarcinomas (40% of poorly differentiated primaries vs. 19% for others, and 48% of poorly differentiated metastases vs. 35% for other metastases; P less than .001 for both), and for SCCs (40% vs. 33% of primaries; P = .028; and 42% vs. 31% of metastases; P = .025).
The investigators also detected site-specific differences in PD-L1 positivity, STK11 and KRAS mutation rates, and other markers along the 10 mutations/MB cutoff which they characterized as “clinically informative.”
“Practically, clinicians should consider obtaining metastasis samples when possible if assessing for high TMB. Furthermore, in addition to previously described intratumor heterogeneity, our data highlight significant intertumor heterogeneity in NSCLC. One possible solution to capture tumor heterogeneity is with blood TMB, which theoretically could depict a composite TMB score versus single-site tissue,” Dr. Stein and associates said.
The investigators did not report a study funding source. Dr. Stein reported no relevant disclosures. Several of the coauthors are employees of Caris Life Sciences, which performed the tumor profiling.
SOURCE: Stein MK et al. JCO Precision Oncology. 2019 Jul 26. doi: 10.1200/PO.18.00376.