Applicability of Routine Targeted Next-generation Sequencing to Estimate Tumor Mutational Burden (TMB) in Patients Treated With Immune Checkpoint Inhibitor Therapy David J. Pinato,*† Heather Urus,*† Thomas Newsom-Davis,‡ Persephone Du Parcq,§ Katherine Belessiotis,† Leah Mapara,† Nandita Gupta,∥ Danielle Power,† Justin Weir,∥ Ching Ngar Wong,* Ragu P. Ratnakumaran,† Kathy Dominy,§ Jamshid Khorashad,§ and Mark Bower‡ Summary: It remains unclear whether targeted next-generation sequencing (tNGS) conveys a reliable estimate of tumor mutational burden (TMB). We sequenced 79 archival samples of immune checkpoint inhibitors (ICPIs) recipients (57% lung cancer, 43% melanoma) using Ion Ampliseq Cancer Hotspot Panel. Employing multiple cutoff values, we verified that TMB by tNGS did not correlate with response or survival following ICPI. We found enrichment of ATM mutations in ICPI-refractory tumors (P = 0.01) to correlate with worse survival (4.2 vs. 10 mo, P = 0.03). Limited- coverage tNGS delivers an imprecise estimate of patients’ TMB but may aid identification of candidate somatic variants of predictive/ prognostic significance. Key Words: TMB, immunotherapy, targeted, NGS, panel (J Immunother 2019;00:000–000) H igher levels of somatic nonsynonymous mutations promote clonal diversity of T-cell responses and asso- ciate with improved responsiveness to immune checkpoint inhibitors (ICPIs). Tumor mutational burden (TMB), an emerging biomarker of response to ICPI, has been qualified using whole-exome sequencing (WES) in clinical trial participants. 1 WES is however largely unapplicable in rou- tine clinical practice due to high costs and the need for bioinformatics support for data interpretation, which is lacking in most of the molecular pathology laboratories processing samples outside of highly specialized academic centers. Unlike WES, targeted next-generation sequencing (tNGS) is widely available, being routinely employed for testing of actionable mutations in solid tumors. Although tNGS platforms have been primarily opti- mized for the detection of somatic mutations occurring in a restricted panel of loci, their ability to reflect the overall mutational status of the corresponding tumor has never been prospectively validated. Recent studies have evaluated the effect of coverage and depth of sequencing as important determinants influencing the performance of TMB as a predictive biomarker by means of an in silico comparison between WES and panel-based tNGS using The Cancer Genome Atlas (TCGA) data. 2 Although hypothesis-generating, the correlation analyses were per- formed on TCGA WES data from 8371 specimens repre- sentative of 25 tumors sites without any linkage to response or survival following ICPI treatment. Moreover, a large pro- portion of the tumor sites included in these experiments are intrinsically resistant to immunotherapy irrespective of TMB, casting doubt upon the applicability of these results to immune-sensitive malignancies where ICPIs are approved for use in routine clinical care. Although a growing body of evi- dence supports the concept that limited-coverage NGS might not adequately scale with overall mutation rates, the capacity of panel-based NGS to provide a reliable estimate of patients’ TMB has not been prospectively tested in ICPI recipients. In this pilot study, we intended to overcome this gap in knowl- edge by evaluating whether TMB, estimated through tNGS, is a reliable predictor and outcome from ICPI therapy in routine practice. PATIENTS AND METHODS From a multicenter dataset of patients treated with ICPI (n = 196), 3 we analyzed 79 patients samples whose diagnostic tissue had been processed for tNGS within Imperial Molecular Pathology between 2016 and 2017. Clinical data were collected in a prospectively maintained registry of patients treated with ICPI outside of clinical trials in 2 academic centers between January 01, 2016 and April 01, 2018. Following DNA purification (QIAamp DNA Kit; Qiagen, UK), tNGS on 10 ng of DNA was performed on an Ion PGM sequencer using the Ion AmpliSeq Cancer Hot- spot Panel v2, designed to amplify 207 amplicons covering 2800 COSMIC mutations from 50 loci (0.226 Mbp cover- age). 4 The Ion Reporter suite (Life Technologies, Waltham, Received for publication May 8, 2019; accepted July 30, 2019. From the *Department of Surgery & Cancer, Imperial College London; †Department of Oncology, Imperial College NHS Trust, Charing Cross Hospital; ‡Department of Oncology, Chelsea & Westminster Hospital; §Molecular Pathology Laboratory, Hammersmith Hospi- tal; and ∥Department of Histopathology, Imperial College NHS Trust, Hammersmith Hospital, London, UK. D.J.P. and H.U. contributed equally. Primary research data are presented in a summative manner in the manuscript. No publicly available dataset has been generated as part of this work. D.J.P.: study concepts, study design, statistical analysis, manuscript preparation, and manuscript editing. H.U., T.N.-D., P.D.P., K.B., L.M., N.G., D.P., J.W., C.N.W., R.P.R., K.D., J.K., M.B.: data acquisition. D.J.P., J.K., P.D.P., K.D.: quality control of data and algorithms. D.J.P., J.K., M.B.: data analysis and interpretation. All the authors: manuscript review. Reprints: David J. Pinato, Department of Medical Oncology, Imperial College London, Hammersmith Campus, Du Cane Road, London W12 0HS, UK (e-mail: david.pinato@imperial.ac.uk). Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved. BRIEF COMMUNICATION J Immunother  Volume 00, Number 00, ’’ 2019 www.immunotherapy-journal.com | 1 Copyright r 2019 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.