Portal de Periódicos da CAPES

Analytical Performance of an Immunoprofiling Assay Based on RNA Models

2020; Elsevier BV; Volume: 22; Issue: 4 Linguagem: Inglês

10.1016/j.jmoldx.2020.01.009

1943-7811

Ian Schillebeeckx, Jon R. Armstrong, Jason T. Forys, Jeffrey Hiken, Jon Earls, Kevin C. Flanagan, Tiange Cui, Jarret Glasscock, D. Messina, Eric J. Duncavage,

Immune Cell Function and Interaction

As immuno-oncology drugs grow more popular in the treatment of cancer, better methods are needed to quantify the tumor immune cell component to determine which patients are most likely to benefit from treatment. Methods such as flow cytometry can accurately assess the composition of infiltrating immune cells; however, they show limited use in formalin-fixed, paraffin-embedded (FFPE) specimens. This article describes a novel hybrid-capture RNA sequencing assay, ImmunoPrism, that estimates the relative percentage abundance of eight immune cell types in FFPE solid tumors. Immune health expression models were generated using machine learning methods and used to uniquely identify each immune cell type using the most discriminatively expressed genes. The analytical performance of the assay was assessed using 101 libraries from 40 FFPE and 32 fresh-frozen samples. With defined samples, ImmunoPrism had a precision of ±2.72%, a total error of 2.75%, and a strong correlation (r2 = 0.81; P < 0.001) to flow cytometry. ImmunoPrism had similar performance in dissociated tumor cell samples (total error of 8.12%) and correlated strongly with immunohistochemistry (CD8: r2 = 0.83; P < 0.001) in FFPE samples. Other performance metrics were determined, including limit of detection, reportable range, and reproducibility. The approach used for analytical validation is shared here so that it may serve as a helpful framework for other laboratories when validating future complex RNA-based assays. As immuno-oncology drugs grow more popular in the treatment of cancer, better methods are needed to quantify the tumor immune cell component to determine which patients are most likely to benefit from treatment. Methods such as flow cytometry can accurately assess the composition of infiltrating immune cells; however, they show limited use in formalin-fixed, paraffin-embedded (FFPE) specimens. This article describes a novel hybrid-capture RNA sequencing assay, ImmunoPrism, that estimates the relative percentage abundance of eight immune cell types in FFPE solid tumors. Immune health expression models were generated using machine learning methods and used to uniquely identify each immune cell type using the most discriminatively expressed genes. The analytical performance of the assay was assessed using 101 libraries from 40 FFPE and 32 fresh-frozen samples. With defined samples, ImmunoPrism had a precision of ±2.72%, a total error of 2.75%, and a strong correlation (r2 = 0.81; P < 0.001) to flow cytometry. ImmunoPrism had similar performance in dissociated tumor cell samples (total error of 8.12%) and correlated strongly with immunohistochemistry (CD8: r2 = 0.83; P < 0.001) in FFPE samples. Other performance metrics were determined, including limit of detection, reportable range, and reproducibility. The approach used for analytical validation is shared here so that it may serve as a helpful framework for other laboratories when validating future complex RNA-based assays. Cancer pathogenesis has traditionally been viewed as a multistep process through which normal cells progressively acquire tumorigenic traits, the so-called hallmarks of cancer.1Hanahan D. Weinberg R.A. The hallmarks of cancer.Cell. 2000; 100: 57-70Abstract Full Text Full Text PDF PubMed Scopus (22279) Google Scholar,2Hanahan D. Weinberg R.A. Hallmarks of cancer: the next generation.Cell. 2011; 144: 646-674Abstract Full Text Full Text PDF PubMed Scopus (42766) Google Scholar Specifically, genetic and epigenetic alterations have been considered the predominant drivers of cancer pathogenesis. Increased tumor-infiltrating lymphocytes have been associated with improved outcomes in a broad range of human cancers, including melanoma, colorectal cancer, and triple-negative breast cancer. More recently, the presence of immune cells in the tumor microenvironment has been shown to play a role in the progression of cancer and the response of the patient to therapy.3Gooden M.J.M. de Bock G.H. Leffers N. Daemen T. Nijman H.W. The prognostic influence of tumour-infiltrating lymphocytes in cancer: a systematic review with meta-analysis.Br J Cancer. 2011; 105: 93-103Crossref PubMed Scopus (885) Google Scholar, 4Denkert C. Loibl S. Noske A. Roller M. Müller B.M. Komor M. Budczies J. Darb-Esfahani S. Kronenwett R. Hanusch C. von Törne C. Weichert W. Engels K. Solbach C. Schrader I. Dietel M. von Minckwitz G. Tumor-associated lymphocytes as an independent predictor of response to neoadjuvant chemotherapy in breast cancer.J Clin Oncol. 2010; 28: 105-113Crossref PubMed Scopus (1237) Google Scholar, 5Lee H.J. Seo J.-Y. Ahn J.-H. Ahn S.-H. Gong G. Tumor-associated lymphocytes predict response to neoadjuvant chemotherapy in breast cancer patients.J Breast Cancer. 2013; 16: 32Crossref PubMed Scopus (106) Google Scholar, 6Hendry S. Salgado R. Gevaert T. Russell P.A. John T. Thapa B. et al.Assessing tumor-infiltrating lymphocytes in solid tumors: a practical review for pathologists and proposal for a standardized method from the International Immuno-Oncology Biomarkers Working Group: part 2: TILs in melanoma, gastrointestinal tract carcinomas, non-small cell lung carcinoma and mesothelioma, endometrial and ovarian carcinomas, squamous cell carcinoma of the head and neck, genitourinary carcinomas, and primary brain tumors.Adv Anat Pathol. 2017; 24: 311-335Crossref PubMed Scopus (375) Google Scholar, 7Fortes C. Mastroeni S. Mannooranparampil T.J. Passarelli F. Zappalà A. Annessi G. Marino C. Caggiati A. Russo N. Michelozzi P. Tumor-infiltrating lymphocytes predict cutaneous melanoma survival.Melanoma Res. 2015; 25: 306-311Crossref PubMed Scopus (51) Google Scholar With this increased understanding of the important interplay between cancer development and the immune system, there have been new therapeutic efforts aimed at modulating the immune system to improve patient outcomes. Most salient is the checkpoint inhibitor class of drugs. This class of drugs works by blocking immune checkpoints, which normally act to moderate immune responses by suppressing immune cell activation. By lifting this suppression, checkpoint inhibitors can reverse the blockade of an immune response against cancer cells, which is often imposed in the tumor microenvironment. Checkpoint inhibitors, such as ipilimumab and pembrolizumab, humanized monoclonal antibodies that block activation of cytotoxic T-lymphocyte-associated protein 4 and programmed cell death protein 1, respectively, have revolutionized the treatment of multiple cancer types, including melanoma, non–small-cell lung cancer, and several other indications.8Gong J. Chehrazi-Raffle A. Reddi S. Salgia R. Development of PD-1 and PD-L1 inhibitors as a form of cancer immunotherapy: a comprehensive review of registration trials and future considerations.J Immunother Cancer. 2018; 6: 8Crossref PubMed Scopus (717) Google Scholar,9Darvin P. Toor S.M. Sasidharan Nair V. Elkord E. Immune checkpoint inhibitors: recent progress and potential biomarkers.Exp Mol Med. 2018; 50: 165Crossref PubMed Scopus (992) Google Scholar However, despite durable responses in some patients, the objective response rate for these treatments is typically <30%, demonstrating a clear need to understand their underlying mechanisms and develop tools to identify patients who are likely to benefit from them.9Darvin P. Toor S.M. Sasidharan Nair V. Elkord E. Immune checkpoint inhibitors: recent progress and potential biomarkers.Exp Mol Med. 2018; 50: 165Crossref PubMed Scopus (992) Google Scholar,10Marconcini R. Spagnolo F. Stucci L.S. Ribero S. Marra E. Rosa F De Picasso V. Di Guardo L. Cimminiello C. Cavalieri S. Orgiano L. Tanda E. Spano L. Falcone A. Queirolo P. Italian Melanoma Intergroup (IMI) for the IMICurrent status and perspectives in immunotherapy for metastatic melanoma.Oncotarget. 2018; 9: 12452-12470Crossref PubMed Scopus (56) Google Scholar For example, studies suggest that checkpoint inhibitor therapies are less effective when tumors lack infiltrating lymphocytes in general, and are most effective in patients with preexisting programmed death-ligand 1–inhibited tumor infiltrating T cells.11Tumeh P.C. Harview C.L. Yearley J.H. Shintaku I.P. Taylor E.J.M. Robert L. Chmielowski B. Spasic M. Henry G. Ciobanu V. West A.N. Carmona M. Kivork C. Seja E. Cherry G. Gutierrez A.J. Grogan T.R. Mateus C. Tomasic G. Glaspy J.A. Emerson R.O. Robins H. Pierce R.H. Elashoff D.A. Robert C. Ribas A. PD-1 blockade induces responses by inhibiting adaptive immune resistance.Nature. 2014; 515: 568-571Crossref PubMed Scopus (4367) Google Scholar,12Herbst R.S. Soria J.-C. Kowanetz M. Fine G.D. Hamid O. Gordon M.S. Sosman J.A. McDermott D.F. Powderly J.D. Gettinger S.N. Kohrt H.E.K. Horn L. Lawrence D.P. Rost S. Leabman M. Xiao Y. Mokatrin A. Koeppen H. Hegde P.S. Mellman I. Chen D.S. Hodi F.S. Predictive correlates of response to the anti-PD-L1 antibody MPDL3280A in cancer patients.Nature. 2014; 515: 563-567Crossref PubMed Scopus (3696) Google Scholar The early clinical success of checkpoint inhibitors has led to the development of other immuno-oncology therapy modalities, including adoptive cell transfer13Newick K. O'Brien S. Moon E. Albelda S.M. CAR T cell therapy for solid tumors.Annu Rev Med. 2017; 68: 139-152Crossref PubMed Scopus (473) Google Scholar,14Ö Met Jensen K.M. Chamberlain C.A. Donia M. Svane I.M. Principles of adoptive T cell therapy in cancer.Semin Immunopathol. 2019; 41: 49-58Crossref PubMed Scopus (81) Google Scholar and cancer vaccines.15Guo C. Manjili M.H. Subjeck J.R. Sarkar D. Fisher P.B. Wang X.-Y. Therapeutic cancer vaccines.Adv Cancer Res. 2013; 119: 421-475Crossref PubMed Scopus (373) Google Scholar,16Banchereau J. Palucka K. Immunotherapy: cancer vaccines on the move.Nat Rev Clin Oncol. 2017; 15: 9-10Crossref PubMed Scopus (100) Google Scholar These therapies all share the same intent: to improve the ability of the immune system to detect cancer cells, recruit immune cells to the site of tumors, and ultimately promote the cytotoxic functions of these immune cells to destroy cancer cells. An example of adoptive cell transfer is the use of chimeric antigen receptor T cell therapy. Representative of chimeric antigen receptor T therapies, one US Food and Drug Administration–approved chimeric antigen receptor T therapy demonstrates high overall remission rate of 81%; however, 73% of patients experienced serious adverse events.17Maude S.L. Laetsch T.W. Buechner J. Rives S. Boyer M. Bittencourt H. Bader P. Verneris M.R. Stefanski H.E. Myers G.D. Qayed M. De Moerloose B. Hiramatsu H. Schlis K. Davis K.L. Martin P.L. Nemecek E.R. Yanik G.A. Peters C. Baruchel A. Boissel N. Mechinaud F. Balduzzi A. Krueger J. June C.H. Levine B.L. Wood P. Taran T. Leung M. Mueller K.T. Zhang Y. Sen K. Lebwohl D. Pulsipher M.A. Grupp S.A. Tisagenlecleucel in children and young adults with B-cell lymphoblastic leukemia.N Engl J Med. 2018; 378: 439-448Crossref PubMed Scopus (2609) Google Scholar Because of the high cost of chimeric antigen receptor T cell therapies and the high risk of serious adverse effects, there is a strong need for understanding which patients will benefit from treatment even when overall response rates are high. Given the central role of immune cells in oncology and their connection to positive patient outcomes, a clear need is demonstrated for the quantification of immune cell presence, especially in solid tumor biopsies.8Gong J. Chehrazi-Raffle A. Reddi S. Salgia R. Development of PD-1 and PD-L1 inhibitors as a form of cancer immunotherapy: a comprehensive review of registration trials and future considerations.J Immunother Cancer. 2018; 6: 8Crossref PubMed Scopus (717) Google Scholar The current gold standard method, immunohistochemistry (IHC), has limited throughput and exhibits variability.18Anagnostou V.K. Welsh A.W. Giltnane J.M. Siddiqui S. Liceaga C. Gustavson M. Syrigos K.N. Reiter J.L. Rimm D.L. Analytic variability in immunohistochemistry biomarker studies.Cancer Epidemiol Biomarkers Prev. 2010; 19: 982-991Crossref PubMed Scopus (76) Google Scholar,19Micke P. Johansson A. Westbom-Fremer A. Backman M. Djureinovic D. Patthey A. Isaksson-Mettävainio M. Gulyas M. Brunnstrom H. PD-L1 immunohistochemistry in clinical diagnostics: inter-pathologist variability is as high as assay variability.J Clin Oncol. 2017; 35: e20637Crossref Google Scholar To try to address these weaknesses, clinical methods, such as Omniseq,20Conroy J.M. Pabla S. Glenn S.T. Burgher B. Nesline M. Papanicolau-Sengos A. Andreas J. Giamo V. Lenzo F.L. Hyland F.C.L. Omilian A. Bshara W. Qin M. He J. Puzanov I. Ernstoff M.S. Gardner M. Galluzzi L. Morrison C. Analytical validation of a next-generation sequencing assay to monitor immune responses in solid tumors.J Mol Diagn. 2018; 20: 95-109Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar and some research methods, such as the Nanostring PanCancer IO 360 Gene Panel21Danaher P. Warren S. Dennis L. D'Amico L. White A. Disis M.L. Geller M.A. Odunsi K. Beechem J. Fling S.P. Gene expression markers of tumor infiltrating leukocytes.J Immunother Cancer. 2017; 5: 18Crossref PubMed Scopus (372) Google Scholar and Cibersort,22Newman A.M. Liu C.L. Green M.R. Gentles A.J. Feng W. Xu Y. Hoang C.D. Diehn M. Alizadeh A.A. Robust enumeration of cell subsets from tissue expression profiles.Nat Methods. 2015; 12: 453-457Crossref PubMed Scopus (5009) Google Scholar have used RNA expression to profile the immune response in tumors. However, these qualitative assays rely on rank ordered gene lists or single-gene identifiers to generate cell scores that have not been validated or weakly correlate with immune cell presence. Furthermore, the qualitative cell presence generated by methods such as Nanostring is not comparable within samples. For example, T-cell scores cannot be compared with B-cell scores in the same sample. As such, these methods are not able to accurately or precisely quantify tumor-infiltrating lymphocytes in the tumor microenvironment. Therefore, new quantitative methods and technologies are required to quantify tumor-infiltrating lymphocytes in the tumor microenvironment to provide a platform for diagnostics and, ultimately, help drive drug development. To address this need, a new RNA-based approach was developed to accurately detect the relative amount, or percentage, of immune cells in a heterogeneous cell mixture. The approach uses a database of immune health expression models (iHEMs), which are composed of the most discriminatively expressed genes that identify each of the eight immune cell types. These gene expression models afford several advantages over the single marker or ranked gene list approaches cited previously. Of note, the use of iHEMs enables more robust quantification and is less susceptible to molecular noise. The molecular and informatic pipelines of this approach have been standardized into the ImmunoPrism assay (Cofactor Genomics, Inc., San Francisco, CA). ImmunoPrism quantifies the presence of eight different immune cell types from as little as 40 ng of RNA extracted from a formalin-fixed, paraffin-embedded (FFPE) tumor sample. ImmunoPrism enriches for the small fraction of gene constituents of the iHEMs to maximize the analytical performance of immune characterization (see Materials and Methods). By enriching for and leveraging iHEMs, the ImmunoPrism assay provides quantitative immune profiling of tumor samples, with the goal of facilitating drug development, clinical studies, and patient care in oncology. This article describes the validation employed to characterize ImmunoPrism's analytical performance in our College of American Pathologists–accredited and Clinical Laboratory Improvement Amendments–certified laboratory. Robust assay performance is demonstrated, and an experimental framework is provided that others can use to validate complex RNA-based assays. There is little guidance from the College of American Pathologists on how to measure analytical performance in RNA-based assays, especially those that use machine learning methods to generate RNA models and provide percentages as read outs. Herein, how performance was measured in this new type of assay is explained. ImmunoPrism estimates the relative abundance of eight immune cell types (eg, 10% of the cells in a processed FFPE tumor sample are CD4+ T cells). Central to the analytical validation in this work are samples generated with known immune cell ratios. These samples were processed with ImmunoPrism molecular and analysis pipelines to estimate the abundance of immune cells. These estimations were compared with the known values of these samples to understand the error of ImmunoPrism estimations and derive performance metrics. The assay's error and the subsequent analytical performance metrics are described in the same unit as the assay's estimations: percentage. As such, these metrics consider the absolute error in percentage points (of 100), as opposed to a relative error (depending on the known value). For example, if CD8 cells are known to be 10% of the cells in a sample and ImmunoPrism estimates it to be 9%, a 1% absolute error is measured, not a 10% relative error. This was chosen to simplify evaluation of errors, estimations, and performance metrics, especially across a wide reportable range of values. For the figures in this work, absolute error metrics are presented. However, for select figures, relative error metrics are also provided. The ImmunoPrism assay employs an analysis algorithm that is different than common genetic assays because it does not estimate a binary classification (eg, the presence or absence of a particular single-nucleotide polymorphism). Instead, this assay estimates continuous values. Therefore, diagnostic metrics, such as precision, recall, sensitivity, and specificity, are not appropriate to describe the performance of this assay. Instead, analytical metrics are used, sometimes of the same name (eg, sensitivity or limit of detection). See Saah and Hoover23Saah A.J. Hoover D.R. "Sensitivity" and "specificity" reconsidered: the meaning of these terms in analytical and diagnostic settings.Ann Intern Med. 1997; 126: 91-94Crossref PubMed Scopus (170) Google Scholar for a complete discussion comparing diagnostic and analytic metrics. In lieu of diagnostic metrics, this study adopted analytical performance metrics inspired by International Organization for Standardization 5725:1994 accuracy (trueness and precision) of measurement methods and results—part 1: general principles and definitions. In particular, performance was characterized by trueness, precision, and accuracy. Trueness is measured as the average error across all samples. This metric describes the bias of estimates (ie, "is it consistently too low or too high"?). Precision is measured as the first SD of error across considered samples. This metric gives a sense for the distribution of how much samples may deviate from a known value. Accuracy is defined as the root mean squared error across all samples. This metric gives a single, high-level measure of how well the assay estimates cell presence. In effect, it combines the different ways that trueness and precision describe error into a single metric. Trueness, precision, and accuracy describe the assay's error, and therefore, these should ideally be 0%. The potentially confusing semantics this may cause are noted (eg, a low accuracy value, such as 1.2%, indicates highly accurate performance). Typically, the trueness, precision, and accuracy are calculated across all cell types and all samples. To provide further resolution to the performance of the assay, these metrics are also calculated across all samples per cell type. In addition to these analytical metrics, more traditional correlation statistics are also shown when appropriate. These include the coefficient of determination, denoted as r2, and the two-tailed null hypothesis significance test, denoted as p. The r2 value was calculated by squaring the sample Pearson correlation coefficient r. For reporting, r2 was chosen over r for its ease of interpretation: it is the proportion of the variance in the dependent variable that is predictable from the independent variable. p was calculated via the t-test from r. ImmunoPrism leverages eight iHEMs. Each iHEM is a distinguishing pattern of gene expression that characterizes an immune cell type. For this assay, iHEMs are used that describe the identity of CD4+ T cells, CD8+ T cells, regulatory T cells (Tregs), M1 and M2 macrophages, monocytes, natural killer (NK) cells, and B cells. Briefly, these iHEMs were generated using machine learning methods to mine the RNA expression data of purified immune cells (origin described below) and other databases. This section describes how the iHEMs are generated and ultimately used to estimate immune cell percentages in tumor samples. FASTQ files were preprocessed with trim_galore/cutadapt version 0.4.1 to remove adapter sequences as well as reads with PHRED quality scores <20 and reads that were 25%, within a cell type, were filtered from further analysis.c.Genes displaying expression levels of 2 reads per kilobase of transcript per million mapped reads were removed from further analysis. Taken together, these intraimmune and interimmune filtering approaches yielded 125 genes. These genes are listed in Supplemental Table S1. The function of these genes, pathways, and their associated tissue expression were further investigated using Reactome Pathway Analysis version 3.6 (release 70)27Fabregat A. Sidiropoulos K. Viteri G. Forner O. Marin-Garcia P. Arnau V. D'Eustachio P. Stein L. Hermjakob H. Reactome pathway analysis: a high-performance in-memory approach.BMC Bioinformatics. 2017; 18: 142Crossref PubMed Scopus (322) Google Scholar and Molecular Signatures Database version 7.028Liberzon A. Subramanian A. Pinchback R. Thorvaldsdottir H. Tamayo P. Mesirov J.P. Molecular signatures database (MSigDB) 3.0.Bioinformatics. 2011; 27: 1739-1740Crossref PubMed Scopus (2708) Google Scholar tools. This information can be found in Supplemental Figure S1 and Supplemental Tables S2 and S3. In brief, the Molecular Signatures Database showed 65 genes in gene families consisting of tumor suppressors, oncogenes, translocated cancer genes, protein kinases, cell differentiation markers, homeodomain proteins, transcription factors, cytokines, and growth factors, with >50% of the genes comprising cell differentiation markers. Reactome Pathway Analysis showed 52 genes involved in immune pathways, including adaptive and innate immune pathways. Capture probes were designed to enrich total RNA for the resulting 125 genes, and using this reduced capture, the same pure cell donor samples were resequenced. For the chosen 125 genes, the mean count per million values were calculated, observed with the reduced capture method, across all donors for each cell type. The mean values of these 125 genes define each iHEM. Thus, informatically speaking, each iHEM is a 125-gene vector. RNA sequencing indicates the relative expression level of different genes; however, each cell type has varying levels of RNA content. To translate transcript quantification into the relative amount of cell present for immune profiling, a corrective factor is necessary. In Using iHEMs for Immune Profiling, corrective factors are used that were derived to enable immune profiling in terms of cell percentage. Briefly, pure immune cell types were combined in known quantities to generate several artificial cell mixtures. RNA from these mixtures was sequenced, and immune mRNA content was estimated using the procedure described below. Correction factors were estimated via Powell nonlinear optimization. The optimization minimized the total squared error of known values and corrected values of all cell types and samples. The task of immune profiling is to determine the relative amounts of each cell type in a sample. The expression of immune cells was characterized via iHEMs. However, a heterogeneous tissue sample will have a diverse mix of different immune cell types and nonimmune cells, and therefore a heterogeneous expression for a set of genes. Immune profiling thus seeks to solve for the relationship between the heterogeneous expression that is sequenced and the iHEMs that define immune cells. This relationship can be modeled as a linear combination of the gene expression of each cell type present in the bulk reduced capture RNA sequencing data:B=S∗F(2) where B is a vector representing the gene expression of the 125 genes from a heterogeneous sample, S is a 125 by 8 matrix of iHEMs, and F is a vector of length 8 that represents the estimated mRNA fractions of each immune cell type present in the heterogeneous sample. For every sample, S is known, B is sequenced, and immune profiling thus solves for F. Raw counts of an input sample were normalized to counts per million, and linear epsilon support vector regression was used to solve equation 1, yielding estimated mRNA fractions of the immune cells represented in the iHEMs. Different immune cell types generate differing amounts of mRNA, so a final operation is needed to generate a final cell type percentage estimation. To do this, cell type–specific correction factors were applied to the mRNA fractions, and the resulting corrected fractions were then scaled such that the sum of the corrected coefficients equaled the sum of the coefficients from the original mRNA. The materials and methods that went into the analytical validation are detailed in the following sections. Where applicable, these materials and methods also apply to iHEM generation and definition. Cryopreserved human peripheral blood mononuclear cells (PBMCs) from normal healthy donors and cryopreserved human CD4+ T cells (enriched by negative selection) from normal healthy donors were purchased from StemExpress (Folsom, CA) and Astarte Biologics (Bothwell, WA), and were stored in liquid nitrogen on receipt. Cryopreserved human CD56+ NK cells from normal healthy donors (enriched by negative selection) were obtained from StemExpress. Cryopreserved human Tregs from normal healthy donors (enriched first by negative selection of CD4+ T cells, followed by positive selection of CD25+ cells) were obtained from StemExpress. Fresh human CD14+ peripheral blood monocytes from normal healthy donors were purchased from StemExpress and received within 20 hours of donor apheresis. The PC3 human prostate cancer cell line was purchased from Sigma-Aldrich (St. Louis, MO), and was maintained in RPMI 1640 media supplemented with 10% fetal bovine serum, 10 mmol/L HEPES buffer, 1× GlutaMAX, and 50 μg/mL gentamicin. Cell culture reagents for PC3 maintenance were purchased from Gibco/Thermo Fisher (Waltham, MA). FFPE samples were acquired from Discovery Life Sciences (Huntsville, AL), Cureline (Brisbane, CA), and House of Tissues (Janesville, WI). Cryopreserved PBMCs (50 to 100 million cells) were removed from liquid nitrogen storage and thawed in a 37°C water bath with gentle hand shaking until only a small piece of ice remained. Th