Mutational Analysis of Field Cancerization in Bladder Cancer
U n c o r r e c t e d A u t h o r P r o o f Bladder Cancer xx (20xx) x-xx Abstract. 11 BACKGROUND: Morphologically normal tissue, adjacent to tumors, contains multiple molecular changes, the so-called field cancerization. The multifocal and recurrent nature of bladder cancer has been hypothesized to originate from this. However, further studies are required to explore the mutational composition of normal tissue adjacent to tumors. 12 13 14 OBJECTIVE: To analyze field cancerization in bladder
... cer patients using a non-tumor guided approach. 15 METHODS: We investigated the mutational landscape of normal appearing urothelium and paired bladder tumors from four patients by applying deep-targeted sequencing. 16 17 RESULTS: Sequencing of 509 cancer driver genes revealed the presence of 2-13 mutations exclusively localized in normal tissue (average target read depth 634×). Furthermore, 6-13 mutations were shared between tumor and normal samples and 8-75 mutations were exclusively detected in tumor samples. More mutations were observed in normal samples from patients with multifocal disease compared to patients with unifocal disease. Mutations in normal samples had lower variant allele fractions (VAF) compared to tumor mutations (p < 2.2*10 -16 ). Furthermore, significant differences in the type of nucleotide changes between tumor, normal and shared mutations (p = 2.2*10 -5 ) were observed, and mutations in APOBEC context were observed primarily among tumor mutations (p = 0.02). No differences in functional impact between normal, shared and tumor mutations were observed (p = 0.61). 18 19 20 21 22 23 24 25 CONCLUSION: Overall, these findings support the presence of more than one field in the bladder, and document non-tumor specific driver mutations to be present in normal appearing bladder tissue. 26 27 in the same patients share multiple mutations and 34 hence are clonally related [1-3]. Furthermore, appar-35 ently normal urothelium has been documented to 36 contain mutations with low variant allele frequencies 37 (∼3%) that are typically observed at high frequencies 38 in tumors (clonal mutations) [1-3]. Multiple stud-39 ies have investigated genomic alterations in normal 40 appearing bladder tissue from cystectomy speci-41 mens, however using technologies that do not allow 42 U n c o r r e c t e d A u t h o r P r o o f 2 T. Strandgaard et al. / Field Cancerization in Bladder Cancer detection of low-frequency mutations. The genomic 43 alterations observed in these studies include recurrent 44 copy number alterations of chromosomes 5, 9, 13, 16, 45 and 17 as well as recurrent mutations or loss of RB1 46 and TP53 [4-9]. These findings corroborate the sug-47 gestions of the presence of field cancerization in the 48 bladder. In addition, several studies in other tissue 49 types have also revealed the presence of mutations 50 in known cancer driver genes in apparently healthy 51 tissue and pre-cancer lesions [10-13]. 52 Approximately 50% of both primary and recurrent 53 bladder tumors in non-muscle invasive bladder cancer 54 (NMIBC) are multifocal . Moreover, recurrent 55 bladder cancer (BC) is common as the majority of the 56 patients with NMIBC relapse within five years [15, 57 16]. Approximately 75% of patients with BC present 58 with NMIBC, and 5-25% of these will progress 59 to muscle-invasive bladder cancer (MIBC) [16, 17]. 60 Multifocality and the frequent recurrences of BC are 61 hypothesized to originate from field cancerization of 62 the bladder urothelium . This concept was first 63 described in oral squamous epithelium in 1953 by 64 Slaughter et al. as an explanation of the high local 65 recurrence rate of oral cancers . More recent, 66 field cancerization has been described as an under-67 lying mechanism for tumor development in various 68 cancer types, including BC . 69 Field cancerization is understood as one or more 70 areas, or fields, with mutated cells. The transformed 71 cells may appear normal or dysplastic [20, 21]. Dif-72 ferent paths for the development of field cancerization 73 have been proposed. These include intraepithelial 74 migration and/or luminal seeding of carcinoma cells 75 from existing tumors followed by implantation of 76 these cells in the urothelium. Fields arising from these 77 cells will hence resemble the tumor . Another 78 explanation is that transformed stem cells, embedded 79 in the urothelium, clonally expand and lead to the 80 formation of fields of transformed cells, which may 81 develop into a tumor [18, 23]. In our previous study of 82 bladder cancer field cancerization, we proposed a the-83 ory of multiple fields being present in the bladder  84 where parallel expansion of different mutated stem 85 cells might lead to multiple transformed fields inter-86 mixed in the bladder urothelium. Tumors will mirror 87 the genetic alterations from the field from which it 88 arose . 89 Previously, we analyzed mutations in adjacent nor-90 mal tissue restricted to mutations observed in the 91 tumor samples, and consequently, non-tumor specific 92 mutations were not investigated . In this study, we 93 characterized mutations in normal appearing urothe-94 lium adjacent to tumors by deep-targeted sequencing. 95 We detected high-impact mutations in known driver 96 genes that were not observed in the tumor. Further-97 more, we observed mutations shared between tumor 98 and normal samples (tumor field effect) as well as 99 mutations specific to the tumors (mutations acquired 100 later in development). 101 RESULTS 102 We performed deep-targeted sequencing of DNA 103 obtained from four patients (patients 1 to 4) with 104 advanced bladder cancer, treated with radical cystec-105 tomy (see Supplementary Fig. S1 and Supplementary 106 Table S1 for detailed disease courses). From each 107 patient, DNA was procured from bulk tumor biopsies 108 (n = 2-7) and laser microdissected (LMD) biopsies 109 of normal appearing urothelium (n = 6-11) (See Sup-110 plementary Table S2 for overview of samples and 111 sequencing information). Individual bulk tumor sam-112 ples were previously analyzed by whole exome 113 sequencing (WES) followed by deep-targeted ampli-114 con sequencing of LMD tumor and normal samples 115 guided by the original WES of bulk tumor . In 116 this present study, we expand on our previous study 117 to include the analysis of mutations uniquely present 118 in normal appearing adjacent tissue by deep-targeted 119 sequencing (Fig. 1). 120 Deep targeted sequencing 121 Extracted DNA from tumors and remaining DNA 122 from LMD normal samples were pooled resulting 123 in one pool of tumor DNA (tumor pool) and one 124 pool of normal DNA (normal pool) for each of the 125 four patients. We performed deep-targeted ampli-126 con sequencing of 509 cancer genes on both pools 127 and on matched leukocyte DNA as reference. We 128 obtained a raw average target read depth of 634× 129 (range: 360-1073). To ensure low error-rates and 130 thereby facilitate robust mutation calls, we employed 131 unique molecular identifiers (UMI), resulting in a 132 consolidated average target read depth of 69× (range: 133 36-129). The coverage reduction is in accordance 134 with previous studies [24, 25]. In total, MuTect2 135 identified 131-283 point mutations and indels in the 136 pools from the four patients. Mutations were classi-137 fied as unique to normal (N-Mutations), unique to 138 tumors (T-Mutations) or shared (S-Mutations). N-139 and T-Mutations were subjected to read-counting 140 in associated sequencing data to assess presence of 141 mutations below the detection threshold of MuTect2, 157 Supplementary Fig. S1: Disease courses of the four 595 patients.