Background
There is a strong impetus to personalize the care of cancer patients to deliver selective therapies. One way this is being done clinically is by performing next-generation sequencing (NGS) on panels of cancer-related genes to identify targeted therapy approaches to which patients are most likely to respond. While a number of ongoing targeted therapy trials are utilizing this approach to optimize patient selection, in parallel, new immunotherapies have shown remarkable anti-cancer effects in melanoma and multiple other malignancies [
1‐
6]. Notably, several recent whole exome sequencing (WES) studies have demonstrated a significant correlation between the total mutation load (i.e., the complete set of genes harboring non-synonymous, exonic mutations in a tumor) and clinical benefit with immune checkpoint inhibitors [
7‐
9]. Widespread access to WES remains limited due to infrastructure requirements, cost and bioinformatic demands. The development of a method to accurately estimate total mutation load from widely available NGS gene panels could further personalize the care of cancer patients by improving patient selection for immune-based therapies. In order to address this unmet clinical need, we developed an algorithm to generate a predicted total mutation load (PTML) from the mutation status of a tumor from a small set of cancer-related genes. Several cohorts of patients were used to determine the accuracy of PTML, and to assess its association with clinical outcomes in melanoma and lung cancer patients treated with immunotherapies.
Discussion
In this analysis, we employed WES data from 1104 distinct tumor samples to derive and validate an algorithm that accurately predicts the total mutation load for melanomas or lung cancers using results from a set of 170 genes broadly used in NGS cancer gene panels. We demonstrate that the individual genes were “weighted” differently depending on tumor type, creating melanoma- and lung cancer-specific PTML algorithms. These cancer-specific PTMLs strongly correlated with ATMLs from WES in both cancer types, and with clinical outcomes using three distinct immunotherapies (i.e., anti-CTLA-4, anti-PD-1, and ACT-TIL). These results suggest that small NGS mutation panels used to select targeted therapy approaches may also have utility for the personalized use of immunotherapies.
The relative cost comparison of targeted next generation sequencing of 169 genes versus WES of 20,000 genes depends on the assessment of multiple variables: the depth of sequencing, extra reagent expense, difference in flow cell occupancy, contrast in informatics and data storage needs, and time costs. We estimate this difference to be at least 5- to 10-fold greater to perform targeted next generation sequencing on approximately 200 gene exomes versus WES with current technology and resources.
The clinical benefit observed in melanoma and lung cancer to single-agent immunotherapy is thought to be related to these two tumor types having the highest frequency of gene mutations among common solid cancers, and thus increasing the likelihood of generating neoantigens recognized by the immune system [
7,
8,
19]. Multiple WES studies have now shown a significant correlation between the total tumor mutation load and the predicted neoantigen load [
9,
14,
18,
22]. In addition, recent data from the melanoma and lung adenocarcinoma TCGAs showed a strong association between total tumor mutation load and immune cytolytic activity within samples, with an abrupt increase in the immune cytolytic activity within both tumor types beginning at approximately 100 total mutations within a tumor [
9,
18]. Consistent with this data, a recent study by Van Allen et al. [
9] showed, in an independent melanoma cohort, that both the mutation/neoantigen load and the expression of immune cytolytic enzymes are associated with clinical benefit. Our own analysis of this data indicates that the combination of mutation load and cytolytic score results in an even more significant correlation with clinical benefit (unpublished data).
In this study, we observed that patients with a low PTML (≤ 100) had worse clinical outcomes compared to patients with a high PTML (> 100) in three independent advanced melanoma cohorts treated with ipilimumab. We also report here, for the first time, that low PTML is associated with significantly shorter PFS and OS in a cohort of metastatic melanoma patients treated with ACT-TIL. The correlation between ACT-TIL clinical outcome and the total mutation load is reinforced by prior case reports suggesting that benefit from ACT-TIL may be due to the existence and persistence of clones within the TIL that recognize neoantigens on the matched cancer cells [
23,
24]. Finally, utilizing the PTML ≤ 100 threshold, patients with lung adenocarcinoma and a low PTML failed to achieve partial tumor response, failed to achieve durable clinical benefit, and had a lower PFS. Thus, low PTML represented a clinically worse outcome category in pembrolizumab-treated lung adenocarcinoma.
Despite the association between tumor mutation load and clinical outcome in these and other studies, as with other effective markers in cancer, the relationship between these variables is not deterministic. Patients with melanoma or lung cancer that harbor a “low” tumor mutation load can “respond” to immunotherapy and those with a “high” mutation load may not respond. However, our analysis across five clinical data sets in two cancer types indicates that, below a threshold level, the probability of clinical benefit is significantly lower. Thus, the total mutation load or the PTML may serve as an important variable when assessing the potential benefits of immune-based therapeutics in individual patients. In addition, the PTML demonstrated efficacy across cohorts in which distinct tissue procurement, exome capture/sequencing techniques, mutation calling algorithms, and different definitions of clinical benefit were used. In this study, we were agnostic about the specific location and functionality of a gene mutation and its association with the ATML. Further studies to enhance the use of the PTML could focus on standardizing analytic practices, potentially assigning further weight to functionally relevant genes/mutation sites and combining the PTML with other variables that may have predictive value (i.e., immune infiltrate or cytolytic score) as has been suggested [
9].
There are multiple examples of oncogenic mutations that predict benefit from FDA-approved targeted therapies (CML/imatinib, melanoma/BRAFi, lung/EGFR and ALK inhibitors, breast/HER2 inhibitors), and other mutations that predict resistance (
RAS mutations and EGFR inhibitors). Taken together, these data support the clinical use of molecular testing to guide personalized cancer treatment [
25]. In addition, multiple trials are currently ongoing in which patients are assigned to investigational targeted therapy strategies based on the results of clinical NGS panels. Here, we demonstrate that the mutation status of a small set of genes, that could also be used to select rationale targeted therapies, provides an accurate estimate of the total mutation load (PTML) which significantly correlates with clinical benefit from immunotherapy in melanoma and lung cancer patients. These results provide a new and easily actionable approach to personalize the care of cancer patients and to further optimize the use of immune therapies.
Acknowledgements
Not applicable.
Competing interests
Dr. Roszik is a consultant in Merck KGaA. Dr. Hess is an unpaid consultant in Angiochem, Inc. Dr. Tetzlaff serves on the advisory board for Advisory boards of Myriad Genetics and Seattle Genetics. Dr. Wargo is a speaker or advisory board for Dava Oncology, Roche Genentech, Novartis, GSK, and Illumina. Dr. Chen is an employee of Lion Biotechnologies. Dr. Meric-Bernstam has a consulting/advisory role in Genentech, Novartis, Roche, Inflection Biosciences, Celgene, and receives Honoraria from Genentech, Roche Diagnostics and research support from Novartis, AstraZeneca, Taiho Pharmaceutical, Genentech, Calithera Biosciences, Debiopharm Group, Bayer, Aileron Therapeutics, PUMA Biotechnology, Verastem, and CytomX Therapeutics. Dr. Mills is a consultant in Adventist Health, Allostery, AstraZeneca, Catena Pharmaceuticals, Critical Outcome Technologies, ImmunoMET, Isis Pharmaceuticals, Lilly, Medimmune, Novartis, Precision Medicine, Provista Diagnostics, Signalchem, Lifesciences, Symphogen, Takeda/Millenium Pharmaceuticals, Tau Therapeutics Tarveda, and receives sponsored research from the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, AstraZeneca, Breast Cancer Research Foundation, Critical Outcome Technologies, Illumina, Karus, Komen Research Foundation, Nanostring, and Takeda/Millenium Pharmaceuticals. Dr. Gershenwald has a consulting or advisory role in Navidea, Castle Biosciences, and Merck, and intellectual property in Mercator therapeutics. Dr. Radvanyi is an employee of EMD Serono. Dr. Hwu has a stock or other ownership and consulting or advisory role in Lion Biotechnologies Immatics US, Inc., and receives research funding from Genentech and Bristol-Myers Squibb. Dr. Bernatchez has a consulting or advisory role in Lion Biotechnologies, and receives research funding from Nektar, Idera. Dr. Davies has a consulting or advisory role in GlaxoSmithKline, Genentech/Roche, Novartis, Sanofi, and Vaccinex, and receives research funding from the Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, GlaxoSmithKline, Genentech/Roche, AstraZeneca, Merck, Oncothyreon, Myriad Genetics, and Sanofi. No other disclosures are reported.