Clinical Exome Gene Panel Analysis of a Cohort of Urothelial Bladder Cancer Patients from Sri Lanka

Background: Bladder cancer has a high rate of recurrence and high mortality rates in those who progress to muscle invasive disease. Biomarkers and molecular sub classification of tumours beyond standard histopathology has been proposed to address therapeutic dilemmas. The Cancer Genome Atlas project and other studies have contributed to the enhanced knowledge base of the mutational landscape of urothelial bladder cancer. Once again, these are mostly from Caucasian and Chinese patients, with data from the rest of Asia and Sri Lanka being sparse. The objective of this study was to assess the genomic variations of a cohort of urothelial bladder cancer patients in Sri Lanka. Methods: The molecular genetic study was conducted on formalin fixed paraffin embedded tumour samples of 24 patients, prospectively enrolled from 2013 to 2017. The samples were sequenced and variant distribution performed based on a 70-gene panel. Results: Total number of filtered mutations in the 24 patients was 10453. Median mutations per patient were 450 (range 22-987). The predominant mutational change was C>T and G>A. The top 5 mutated genes in our cohort were SYNE1, SYNE2, KMT2C, LRP2, and ANK2. The genes were clustered into 3 groups dependent on the number of mutations per patient per gene. The genes of cluster 1 and 2 mapped to Chromatin modifying enzymes and Generic Transcription Pathway. The chromatin remodelling pathway accounted for the largest proportion (22%) of mutations. Conclusions: Clinical exome sequencing utilising a gene panel yielded a high mutation rate in our patients. The predominant mutational change was C>T and G>A. Three clusters of genes were identified. SYNE1 was the gene with the most mutations. The mutations comprised predominantly of genes of the chromatin remodelling pathway.


Introduction
have progressed while on platinum based chemotherapy (Thomas and Sonpavde, 2022).
During the recent past several landmark papers have attempted to clarify the molecular genetic profile of urothelial bladder cancer. The Cancer Genome Atlas Project, Pietzak et al (Weinstein et al., 2014;Pietzak et al., 2017;Robertson et al., 2017) shed light on the molecular genetics of Caucasian populations, while the data of Asian populations is mainly from Chinese populations. The findings of the above studies and Yaoting Gui's team from Shenzhen, China (Gui et al., 2011) andZhiming Cai's (Guo et al., 2013) are summarised in supplementary material 1.
Our objective was to investigate the mutational profile of a Sri Lankan cohort with urothelial bladder cancer, to identify the bladder cancer associated genes in our population and to assess the differences from Caucasian and other Asian studies.

Study sample
A urothelial bladder cancer cohort who underwent TURBT, were prospectively enrolled, by convenient sampling, from June 2013 to January 2017, from two teaching hospital urology departments of Sri Lanka. All patients provided written informed consent. The FFPE blocks of patients who had undergone their first intervention (transurethral resection of bladder tumours i.e. TURBT) for bladder tumours was obtained from the relevant pathology departments.

Genetic analysis
Genomic DNA was isolated from FFPE blocks by using Phenol:chloroform isolation method. The paired-end (PE) library was prepared from the samples using TruSeq Exome Library Preparation Kit. The pool library was sequenced on NextSeq 500. High quality reads were aligned against the Human reference genome Hg19 (Build -GRCh37, obtained from https://genome.ucsc. edu/). GATK best practices protocol was used to perform the somatic mutation calling. Using different flavour of GATK realignment, recalibration and variant calling using VarDict (v1.5.1) was performed with default parameters.
Variant classification adhered to the guidelines of next generation sequencing from American College of Medical Genetics and Genomics (ACMG) and American Congress of Obstetricians and Gynecologists (ACOG). Variants were annotated using INGENUITY® toolkit.
Variant distribution of the samples based on a 70-gene panel was done (Supplementary material 2).
The variants were filtered using the following criteria: • All germline variants with a mean allele frequency (MAF) > 0.01 identified in 1000genome All, 1000 Genomes SAS, GenoAdd exome_ALL and GenoAdd exome _SAS databases were removed.
• All variants with a COSMIC ID were retained while all variants with an rsID were removed.
• As per CLINVAR SIG, variants that are likely benign/ benign were filtered out.
• Variants in exonic regions and splice sites were retained.
• Of the 70-gene panel, 60 genes were selected as per the previously identified mutated genes in urothelial cancer from literature.
• From the Sri Lanka Genetic Variation Database variants and allele counts were extracted from the Inherited cancer patient list (n=187) to identify common germline variants in the Sri Lankan population. In the 60 genes selected as per above, variants were selected with a MAF > 0.01. The duplicated variants were removed from our variant list. To check if its non-silent mutation rate was significantly higher (P < 0.05) than the background we used the chi square test to remove the non-significantly mutated genes.

Results
A total number of 11033 mutations were identified. As per the variant type it comprised of 10384 single nucleotide variants, 559 deletions, 86 insertions and 4 substitutions. The functional classification of the mutations is shown in Table 1. The two commonest mutations documented in our patients was C>T and G>A. The SNV class is shown in Table 2.
From the above mutations, 60 genes were selected from our panel of 70 as per the genes identified in previous literature. To check if its non-silent mutation rate was significantly higher (P < 0.05) than the background we used the chi square test to remove the non-significantly  (Hadd et al., 2013). In a whole exome sequencing of 20 primary bladder cancers from South Korea, they identified 14,864 exonic mutations, with a median of 231 (range 20-1515) (Kim et al., 2020). A whole exome sequencing of a 19-year-old male from India with high grade urothelial cancer with lymph node metastasis, they identified 558 exonic somatic mutations (Sharma et al., 2019).
The predominant mutational change in our 24 patients was C>T and G>A which is similar to mutations reported in other cancers (Greenman et al., 2007). In urothelial tumours an analysis by Gui et al in 9 patients with transitional cell cancers, identified that the SNV class of the mutations was dominated by a C:G>T:A pattern (Gui et al., 2011). This is also the predominant SNV class in our mutations.
In total, we identified 51 significantly mutated genes in our patients. The mutation frequency as per number of patients is very high for individual genes, when compared with the literature. E.g. SYNE1 mutations in 95% of the patients; SYNE2 100%, EP300 91,67%, TP53 70.83%, mutated genes (Table 3).
Number of mutations per patient are detailed in Table 4. Median mutations per patient were 450 (range 22-987). There were three patients with over 900 mutations each. Two were high grade MIBC; the third was an 80-year-old male with pTa LG disease.
Total mutations in the genes, with significant mutations compared to background mutations, were 10453. The number of mutations in genes as per patient, are tabulated in table 5. The top 5 mutated genes were SYNE1, SYNE2, KMT2C, LRP2, ANK2.
Clustering of the genes was done dependent on the number of mutations and is displayed in Figure 1. There were 3 main clusters identified in the 51-gene list. Utilising the DAVID Gene Functional Classification Tool (https://david.ncifcrf.gov/tools.jsp updated December 2021) (Huang da et al., 2009;Sherman et al., 2022) we identified two main biological processes linked to these clusters (Supplementary material 3). The genes of cluster 1 and 2 mapped to chromatin modifying enzymes and generic transcription pathway respectively as per the Reactome pathway database.
The chromatin remodelling pathway accounted for the largest proportion (22%) of mutations of pathways implicated in urothelial cancer oncogenesis ( Table 6). The following 9 genes of this pathway were present in our gene list ARID1A, CHD6, CREBBP, EP300, KDM6A, KMT2A, KMT2C, KMT2D, and NCOR1. The details of the mutations within these genes are given in Supplementary material 4.

Discussion
Our study population shows a high number of somatic tumour mutations with median mutations per patient of 450 (range 22-987). DNA extraction and          Table 5. Continued 1868 samples in 10 studies of urothelial bladder cancer patients, the somatic mutation frequency was 10.8% (Cerami et al., 2012).
SYNE1 encodes for synaptic nuclear envelope protein 1 or nesprin-1 or enaptin which is an actin binding protein in humans. It has been implicated in autosomal recessive spinocerebellar ataxia. SYNE1 encodes for several protein isoforms, which are constituents of many cellular processes including the organisation, integrity and positioning of the nucleus, and Golgi function (Doherty et al., 2010).
In a study by Shaglouf et al evaluating the expression profile of several proteins and to quantify the expressed transcripts in rats with hepatocellular carcinoma (HCC), they found SYNE1 protein was up regulated along with two other proteins. The transcripts of these proteins showed elevated expression, at different stages of hepatocellular carcinoma from initiation to established tumour stage. It indicates the involvement of SYNE1 proteins in HCC tumorigenesis (Faraj Shaglouf et al., 2020). They also hypothesise that the genes coding for the up regulated proteins probably act via interaction with other genes (e.g. CUL7, CUL9, p53 and VEGFR) involved in cancer pathways.
The Exonic Variant rs9479297 in the SYNE1 gene has been implicated in double primary cancer oncogenesis in urothelial carcinoma and hepatocellular carcinoma (Chu et al., 2021). SYNE1 is a known driver gene in breast cancer. It has also been identified as one among 14 genes useful in identifying chemosensitivity in breast cancer (Al Amri et al., 2020).
In an analysis of somatic mutations in clear cell renal cell cancers from the TCGA databases Li and others identified that mutations of SYNE1 were associated with a higher tumour mutational burden and a poorer prognosis. However SYNE1 correlated with a better response to immune checkpoint inhibitors (Li et al., 2020).
A study utilizing the urothelial cancer datasets from the Cancer Genome Atlas (TCGA) database were evaluated using the CIBERSORT algorithm. This algorithm was used to identify signatures based on tumour immune cell infiltration patterns. Four immune subtype clusters were identified with specific genes among them, as identified by mRNA expression profiling. SYNE1 was a gene identified in cluster 2 which had the best outcome with patients having a longer overall survival (Wu et al., 2020).
The high mutation rate in the SYNE1 gene in our population warrants further confirmatory studies, as well as functional evaluation studies to assess its role in bladder cancer oncogenesis.

Chromatin remodelling pathway
One or more of the genes involved in the chromatin remodelling pathways were mutated in 76% of tumours in the TCGA 2014 cohort (Weinstein et al., 2014). Of the 39 significantly mutated genes in the 2017 TCGA study, 10 were from the chromatin-remodelling pathway (Robertson et al., 2017). In a Chinese cohort of 112 patients the mostly mutated pathway was the chromatin remodelling (73.21%) (Yang et al., 2021). In our study the chromatin remodelling pathway accounted for 26.21% of the total mutations seen in our cohort. The mutations of this pathway belonged to the histone acetyltransferase genes, CREBBP and EP300; SWI/SNF-related chromatin remodeling gene ARID1A; histone methyltransferase (HMT) genes KMT2A (MLL) and KMT2C (MLL3); NCOR1 (which codes for a subunit possessing histone deacetylation (HDAT) activity; CHD6 (encodes a component which remodels chromatin) (Gui et al., 2011).
Three main pathways have been implicated in urothelial cancer oncogensis. Cell proliferation and survival is altered by mutations in e.g. FGFR3, RAS, PIK3CA, PTEN, which cause activation of MAPK and PI3K/AKT signal transduction pathways. TP53, RB1, CDKN2A mutations affect p53-cell cycle pathway and cell cycle checkpoints. Also disruption of cell senescence via mutations in TP53, RB1 and TERT promoter gene, are also primarily implicated (Hoffmann and Schulz, 2021;Tran et al., 2021). Recent studies including TCGA (Robertson et al., 2017), and Yang et al in their study of a Chinese cohort with urothelial cancer (Yang et al., 2021), have identified consistent mutations in the chromatin remodelling pathway being implicated in bladder cancer. Urothelial bladder cancer has one of the highest mutations among all cancers within the chromatin remodelling pathway. Chromatin structure regulation is required for genome replication, transcription, mitosis, DNA repair, heterochromatin formation, and other nuclear processes.
In a recent study by Lawson et al utilising microdissected samples from 15 cadaveric transplant patients and 5 bladder cancer patients, they performed targeted whole exome and genome scanning. In this they identified that normal urothelium contains a median number of 40 mutations per exome and 1879 genomic mutations. Of the frequently mutated genes in the normal urothelium, 5 (KMT2D, KDM6A, ARID1A, KMT2C, EP300) of the top six genes were from the chromatin remodelling pathway (Lawson et al., 2020). This was in contrast to the genes of the PI3K/AKT and p53-cell cycle pathways which are more commoner in bladder cancer urothelium in comparison to normal urothelium. This fact and the observation that chromatin remodelling pathway genes are mutated across all tumour stages and molecular  Table 6. Common Pathways Involved in Mutations subtypes of urothelial cancer, support a hypothesis that mutations of chromatin regulators is an early process during bladder cancer oncogenesis (Lawson et al., 2020;Hoffmann and Schulz, 2021). They further expound that mutations may enhance the capability of precursor urothelial cells to self-replicate. Our study adds to this evidence of the significant involvement of the chromatin remodelling pathway in bladder cancer tumorigenesis, and provides the first such evidence in a South Asian population.
Limitations of this study include the small patient numbers which impact the power of the study. Funding restrictions and non-availability of tumour tissue for further validation studies have restricted us from confirming or refuting the findings of this study. The gene panel based next generation sequencing of the samples were only performed on the tumour tissues. The absence of non-tumour samples for tumour-normal comparison could also be considered as a limitation in discovering the true somatic variants from the background variants. Nonetheless, we used optimized bioinformatics workflows and stringent filtering criteria to identify true somatic variants from the tumour samples. In the genetic data analysis, comparisons with several genomic databases (1,000 genome All, 1,000 Genomes SAS, GenoAdd exome_ALL and GenoAdd exome_SAS databases) were carried out including the Sri Lanka Genetic Variation Database. All variants with a minor allele frequency (MAF) > 0.01 were removed to retain the true somatic variants.
In conclusion, the clinical exome sequencing utilising a panel of 70 genes yielded a high mutation rate in our patients with a median mutation rate of 450. The predominant mutational change was C>T and G>A. Three clusters of genes were identified dependent on the number of mutations. The genes of cluster 1 and 2 mapped to Chromatin modifying enzymes and Generic Transcription Pathway. SYNE1 was the gene with the most mutations. The mutations comprise predominantly of genes of the chromatin remodelling pathway.

Author Contribution Statement
APM, RWJ, VHWD, MDSL were involved in the protocol and project development. Data collection was conducted by APM, SASG, NDP, AA, SP. Data analysis and interpretation was by APM, NN, PS, SP, CE, DE. The initial manuscript writing was done by APM, NN, while all authors did the critical revision of the article. All authors approved the final version to be published.