Genetic analysis of two bladder exstrophy populations of South Asian and North American origin

Summary

Introduction

Bladder exstrophy–epispadias complex (BEEC) is a devastating congenital anomaly of the urinary tract and is associated with an increased risk of bladder cancer. The etiology of the BEEC is unknown, but a clear genetic component has been highlighted. Currently, all genetic studies suffer from small populations, limited to a European background.

Objectives

To identify copy number variations (CNVs) in a South Asian and North American bladder exstrophy cohorts. To identify novel non-European CNVs to expand the literature beyond its European ethnic background.

Study design

Patients from our South Asian and North American cohorts had DNA isolated from peripheral blood samples and biobanked at the Children Hospital of Philadelphia’s (CHOP) Center for Applied Genomics (CAG). DNA Genotyping was performed with various arrays for the North American cohort and the Illumina Global Screening Array for the South Asian cohort.

Controls for both cohorts were identified from CHOP’s CAG and ethnicity matching by principle component analysis was performed. CNV calling and filtering were performed with PennCNV and ParseCNV, respectively.

Results

The North American and South Asian cohorts included 53 and 97 patients, respectively. Fifty-five statistically significant CNVs were identified across three independent analyses, of which 53 (96.4 %) were novel to the BE literature. Thirteen of our CNVs were near (within 100 million base pairs) but not in linkage disequilibrium with 8 previously identified BEEC genome-wide association study (GWAS) loci. One of our CNVs (chr16: 28,635,133–28,636,902) was near a previously reported CNV of clinical significance (chr16:29,645,396–30,168,276). Seventeen CNVs contained 15 distinct genes associated with cancer, of which 10/15 (66.7 %) have reported associations in bladder cancer.

Discussion

We present the first genetic analysis of a non-European cohort of bladder exstrophy patients. Our study identified a high number of novel CNVs containing cancer predisposition genes mostly distinct from those found in European cohorts. Limitations inherent to BEEC research include small sample sizes which required intermixing of all phenotypes of the BEEC spectrum (epispadias to cloacal exstrophy). These data highlight the importance of multi-institutional and international collaboration to create ethnically diverse cohorts and allow for severity-specific genetic analyses of the different phenotypes within the BEEC.

Conclusion

We identified novel CNVs with distinct cancer predisposition genes in the first study to expand the BEEC literature beyond its European ethnic background. Additional studies in other non-European cohorts are needed to expand our understanding of the genetic landscape of BEEC.

Introduction

Bladder exstrophy (BE) is a devastating congenital abnormality of the urinary tract in which infants are born with the urinary bladder extruded through their abdominal wall, a urethra that is open dorsally, and significant secondary abnormalities of the genitalia. It is widely regarded as the most surgically challenging and complex congenital disorder of the urinary tract [1]. BE varies in degree of severity and exists on a continuum described as the bladder exstrophy–epispadias complex (BEEC). The birth prevalence of the entire BEEC spectrum is ~1 in 10,000, with varying birth prevalences estimated by gender-phenotype subtype combinations [2]. Despite major surgical reconstruction, patients born with BEEC suffer from long-term issues with voiding and continence, sexual dysfunction, infertility, mental health, an increased risk of bladder cancer, and an increased risk of renal dysfunction and renal failure [1,3].

While the etiology of BEEC is not known, a clear genetic component has been highlighted. In a Florida population-based study, multiple births conferred a 46 % increased risk of overall birth defects compared to singletons. Within this study, bladder exstrophy was the fifth highest birth defect out of 40 reportable birth defects based on adjusted relative risk [4]. Eighteen cases of BEEC have been reported in twins [5]. Parents who have had one child with BEEC have an approximately 1.4 % risk of having another child with BEEC, 400-fold higher than the general population [6]. Individuals with a history of BEEC have a 1:70 chance of transmission to their progeny [6]. These findings support a multifactorial etiology with evidence for genetic predisposition.

The rarity of BEEC phenotypes limits quality research and study statistics due to small population sizes. Consequentially, experts are still unsure of the embryologic origins and genetic mechanisms that give rise to BEEC. In addition to small populations, all studies to date are also limited to a European background. Ethnically limited studies restricts the discoverability of genetic variants [7], limiting the generalizability of known variants which can negatively impact the performance of genetic risk scores [8].

The development of any genetic risk score or gene panel begins with the identification of disease-specific genetic variants, be it single nucleotide polymorphisms (SNPs) or copy number variations (CNVs). Once identified, either SNPs or CNVs can be leveraged in genome-wide association studies (GWAS) to associate specific genetic aberrations and the affected genes with a disease phenotype [9]. At the minimum, CNVs are 1000 times larger than SNPs which represent a single base pair. Consequently, CNVs harbor the potential to involve multiple genes or regulatory regions with pathogenic CNVs being associated with disorders with Mendelian inheritance and complex multifactorial disease [10]. CNVs can influence disease whether present in coding or non-coding regions of the genome [10], making them excellent targets for identification and further analysis via GWAS to further our understanding of the genetic basis of disease.

The current study is unique in that we evaluated two cohorts of genetically distinct backgrounds: a genetically homogenous population of South Asian origin, and a North American population of predominately European origin. We hypothesized that our genetically unique, South Asian population provided a rare opportunity for gene discovery with significant implications for understanding and potentially modifying the genetic mechanisms responsible for BEEC. These findings will also, when combined with European findings, be generalizable to more geo-ethnic genetic populations.

Materials and methods

Population

Our study population consists of two cohorts: a North American cohort of predominately European genetic origin (Philadelphia, PA) and one of South Asian origin (Ahmedabad, India).

North American cohort

Our North American cohort consists of 53 BEEC patients treated at the Children’s Hospital of Philadelphia (CHOP). These patients all had their DNA bio-banked at CHOP’s Center for Applied Genomics (CAG). All patients had DNA bio-banked following institutional IRB approval and parental consent.

South Asian cohort

Our South Asian population included 98 BEEC patients treated at Civil Hospital in Ahmedabad, India. Following Civil Hospital institutional IRB approval and parental consent, DNA from BEEC patients treated at Civil Hospital was extracted and bio-banked by Neuberg Center for Genomic Medicine.

Controls

Controls for both cohorts were identified from CHOP’s CAG, first by electronic health record mining for BEEC-negative individuals, and then by ethnicity matching by principle component analysis [11,12]. To achieve optimal power, we used 119 matched controls for our North American cohort and 312 for our South Asian cohort. Optimal power is defined here as a 1:4 case:control ratio for rare CNV association testing statistical power in matched race samples.

Genetic analysis

Genotyping and genomic input data

Genomic DNA was extracted from peripheral blood for all patients. The North American cohort was genotyped within the CAG using a variety of different arrays shown in Supplemental Table 1. The use of multiple arrays has been utilized to detect CNVs [13]. The South Asian cohort was genotyped using the Illumina Global Screening Array. The Illumina Global Screening Array contains approximately 660,000 SNPs, detecting CNVs of over 100 kb in length. Related individuals and duplicates were removed using identity-by-state identified by PLINK (http://pngu.mgh.harvard.edu/purcell/plink/) [14].

The Illumina Global Screening Array utilizes a BeadChip technology which contains beads coated with oligonucleotide probes. Complementary sample (patient) DNA hybridizes to the beads creating fluorescent signal intensities reflective of the corresponding sample DNA amount [15]. These signal intensities increase and decrease with duplications or deletions, respectively, and are later utilized in the CNV calling process. Multiple algorithms for calling CNVs exists [16]. Selection of which algorithm to use depends on the origins of the genomic input data which varies from SNP arrays, whole-genome sequencing, or exome sequencing data. In our study, we employed PennCNV for CNV calling of our SNP-based genomic input data originating from Illumina SNP genotyping arrays.

CNV calling

CNV calling refers to the process of identifying and classifying genomic alterations (i.e., CNVs) from genomic data. CNV calls were made with the PennCNV algorithm, which is based on a Hidden Markov Model (HMM) as previously described [17]. Generally speaking, CNV calling relies on genomic metrics which reflect overall genetic variation in a given region (Log R Ratio, LRR) and allelic variation at a SNP locus (B Allele Frequency, BAF) by comparing observed signal intensity to expected signal intensity of diploid genomes and the two allele SNP locus, respectively. Individual sample values are then compared to Population frequency of B allele (PFB) information. In addition, PennCNV uniquely incorporates the distance between adjacent SNPs. These components (LRR, ABF, PFB, distance between neighboring SNPs) are incorporated into the HMM which provides a list of raw CNV calls amenable to filtering.

CNV filtering

CNV filtering refers to the process of converting raw CNV calls (PennCNV output) into high fidelity list of CNV calls through the elimination of false positive results and background noise. Examples of processes within the CNV filtering process include the utilization of confidence thresholds to ensure statistical certainty, segmentation, and GC (guanine-cytosine base composition) content correction. A sample CNV filtering workflow using ParseCNV has been demonstrated [18].

Following CNV Calling with PennCNV, we used a segment-based scoring approach that scans the genome for CNV region containing consecutive probes with more frequent copy number changes in cases compared with controls. All the CNV-regions with nominally significant P-value <0.05 were further Quality Control (QC) filtered based on the following criteria (red flags) which can negatively influence filtering confidence: presence of segmental duplications >10, presence of overlapping multiple database of genomics variants >10, nearness to centromere and telomere proximal regions (any overlap), high or low GC (guanine-cytosine) content (31 > GC > 60), low average number of probes (<10), high variability regions (any overlap), high population frequency (>0.01), CNV peninsula of common CNV, inflated sample, large gap in probe coverage (>50 kb), length <10 kb, HMM confidence score in PennCNV <10, and AB banding of BAF frequency in duplications. See ParseCNV (http://parsecnv.sourceforge.net/) for a full description of red flags used to filter CNVs.

CNV reporting

CNVs are reported from three separate analyses, a combined analysis of the North American and South Asian cohorts, a North American only analysis, and a South Asian only analysis. Genes contained within the CNVs identified in our cohorts were screened for cancer associations.

Statistical analysis

Categorical variables were compared using a Chi-squared test. Mean (SD) and median (IQR) were compared using two-tailed Welch’s T test and two-tailed Mann–Whitney U test, respectively. Significance was set at p < 0.05 for all statistical tests. Demographic statistical analyses were performed using Prism 10.

Results

The North American cohort included 53 patients: 33 with classic exstrophy, 10 with cloacal exstrophy, 10 with epispadias. 44/53 (83.0 %) of these patients were from a European genetic origin. The South Asian cohort included 97 patients: 84 with classic exstrophy, and 13 with epispadias. The North American cohort consisted of 54.7 % (29/53) males with a mean (standard deviation) age of 21 ± 8.4 years, while the South Asian cohort consisted of 70.1 % (68/97) males and had a mean (SD) age of 12 ± 4.1 years. Additional demographic information for each genotyped cohort is shown in Table 1. The North American and South Asian cohorts had a statistically significant difference in respect to age (p < 0.01) and race (p < 0.01) but not in respect to gender distribution (p = 0.060).

Table 1.

Demographic information by cohort.

Demographic Characteristic	Combined cohort (N = 150)	South Asian cohort (N = 97)	North American cohort (N = 53)	P-Valueb
Gender	Patient count			0.060
Male	97	68 (70.1 %)	29 (54.7 %)
Female	53	29 (29.9 %)	24 (45.3 %)
Patient age	Age in years
Mean age (± SD)	15 ± 7.4	12 ± 4.1	21 ± 8.4	<0.01
Median age (IQR)	13 (10,19)	11 (9,16)	21 (15, 30)	<0.01
Racea	Patient count			<0.01
Black/African American	4 (2.7 %)		4 (7.5 %)
Caucasian	38 (25.3 %)		38 (71.7 %)
Caucasian, African American	2 (1.3 %)		2 (3.8 %)
Caucasian, African American, Latino	2 (1.3 %)		2 (3.8 %)
Caucasian, other	1 (0.07%)		1 (1.9 %)
Latino, Caucasian	1 (0.07%)		1 (1.9 %)
Latino, other, Native American	1 (0.07%)		1 (1.9 %)
Other	3 (2.0 %)		3 (5.7 %)
Asian	97 (64.7 %)	97 (100 %)

^aIncreased number of patients with European genetic origin in the manuscript because multiple patients have more than one race reported.

^bA Chi-squared test was used to compare gender and race between the North American and South Asian cohorts. Mean (SD) and median (IQR) were compared using two-tailed Welch’s T test and two-tailed Mann–Whitney U test, respectively. Significance set at p < 0.05 for all statistical tests.

Overall CNV count and combined analysis CNVs

A total of 55 statistically significant CNVs were identified across all three analyses, of which 53 were novel (Tables 2 and 3). 38 statistically significant CNVs (23 duplications and 15 deletions) were identified in the combined analysis, of which 36 were novel (Table 2). We identified two overlapping CNVs: chr9:136,131,539–136,131,651 (encodes ABO blood group system) and chr9:137,294,228–137,308,636 (RXRA gene known to be associated with coronary artery disease) previously reported in a BE cohort of 169 patients by Von Lowtzow et al. [19] 13 CNVs were in close proximity (<100 Mbps) but not in linkage disequilibrium with 8 GWAS significance loci reported by Mingardo et al. (Table 2) [20]. One CNV (chr16: 28,635,133–28,636,902) was near a previously reported CNV of clinical significance (chr16:29,645,396–30,168,276) within the cytogenic loci 16p11.2, which is the known syndromic region of 16p11.2 Deletion Syndrome [21].

Table 2.

Statistically significant CNVs from Combined Analysis.

CNVR(hg19)c	CNV type	P-valued	Odds ratio	Gene(s)e	Bladder cancer association (Bold, underlined genes)f
chr1:45,797,164–45,797,401	Dup	9.11E-18	8.316683	MUTYH	Increased incidence in MUTYH-associated polyposis; increased risk [sup ref 46,47]
chr16:2,134,300–2134,572	Dup	1.22E-17	8.995511	TCRBV20S1,TSC2,PKD1	Mutations identified suggesting mTOR signaling pathway involvement [sup ref 43]
a chr11:64,572,049–64,572,290	Dup	1.34E-16	8.735828	MEN1
a chr11:2,593,338–2869,001	Dup	3.13E-13	5.775824	KCNQ1, KCNQ1OT1	Promotes progression; cell-line studies show decreased proliferation, migration, and invasion when gene expression decreased [sup ref 59]
chr19:1,220,367–1220,587	Dup	2.23E-12	5.205592	STK11	Epigenetics – 96.8 % of NMIBC methylated in this study. Methylation differentiates low-grade and high-grade tumors [sup ref 28]
b chr9:136,131,539–136,131,651	Dup	1.22E-11	5.963542	ABO	ABO genotype and phenotype is not a risk factor for bladder cancer. ABO alleles associated with clinical and histologic features of bladder cancer [28].
chr22:29,981,324–30,000,070	Dup	4.53E-10	9.508696	NF2
chr15:48,784,657–48,784,775	Dup	6.90E-09	40.38462	FBN1
a chr11:67,351,241–67,351,441	Dup	1.81E-08	9.706284	GSTP1	Susceptibility [sup ref 20, 22, 24,25]
chr7:150,644,394–150,644,474	Dup	1.51E-07	19.00763	KCNH2
chr13:32,890,615–32,890,746	Dup	2.01E-07	4.36067	BRCA2	Potential predisposition gene, increased risk, good outcomes/better prognosis [sup ref 61, 62, 63,64]
chr10:43,600,452–43,604,611	Dup	1.12E-05	10.48756	RET
chr21:47,421,870–47,423,823	Dup	2.08E-05	13.73162	COL6A1
b chr9:137,294,228–137,308,636	Dup	4.05E-05	12.72263	RXRA
chr22:42,521,984–42,522,392	Dup	9.99E-05	4.329004	LOC100132273
a chr17:78,085,800–78,087,149	Dup	0.00012	5.321701	GAA
a chr17:41,267,738–41,267,767	Dup	0.000585	3.459721	BRCA1
a chr17:37,821,636–37,822,316	Dup	0.001682	4.131217	TCAP
chr18:48,603,007–48,603,060	Dup	0.002128	4.503676	SMAD4	Inverse correlation with cancer-specific death; high expression associated with longer overall survival [sup ref 37]
chr12:40,716,260–40,716,260	Dup	0.007261	Infinity	LRRK2
a chr20:62,076,031–62,076,124	Dup	0.007958	6.965035	KCNQ2
chr16:85,679,992–85,684,448	Dup	0.019609	Infinity	KIAA0182
chr12:121,176,322–121,177,272	Dup	0.024222	4.624709	ACADS
chr19:41,527,275–41,528,667	Del	2.85E-19	25.092	CYP2A7
chr16:28,635,133–28,636,902	Del	3.00E-09	5.6603	NPIPL1, SULT1A1	Arg213His polymorphism associated with increased risk [sup ref 5]
chr18:14,280,170–14,281,662	Del	2.05E-05	23.551	DQ590126, ANKRD20A5P
a chr11:111,965,528-111,965,693	Del	0.000128	Infinity	SDHD
a chr17:41,245,461–41,245,471	Del	0.000549	3.7209	BRCA1
a chr11:67,354,776–67,355,456	Del	0.000862	15.845	GSTP1	See above
chr13:82,542,883–82,745,589	Del	0.000966	Infinity	SPRY2
a chr11:71,546,291 –71,587,349	Del	0.002637	Infinity	DEFB108B, LOC100133315, FAM86C1, ALG1L9P,RP11–849H4.2
a chr1:110,250,026–110,251,116	Del	0.005261	12.153	GSTM5, GSTM2, GSTM1	Null genotype associated with increased risk [sup ref 25, 53,54,55]
a chr11:51,419,375–51,478,650	Del	0.005944	5.8629	OR4A5, TRNA_lys
chr21:15,248,281 –15,263,760	Del	0.006647	4.8518	DQ579288
chr18:9,241,963–9256,258	Del	0.019414	Infinity	ANKRD12, TWSG1
chr2:234,539,769–234,540,809	Del	0.019414	Infinity	UGT1A8
chr5:177,390,777–177,396,392	Del	0.023323	4.6247	AK126616, LOC728554, RP11–423H2.3
chr13:67,098,855–67,161,779	Del	0.029813	6.0521	PCDH9

Dup = duplication, Del = deletion.

^aCNV loci in close proximity to a known GWAS loci (within 100 million BPs).

^bCNV previously reported in bladder exstrophy literature.

^cCNV region of greatest significance and overlap coordinates.

^dTwo-tailed Fischer’s Exact P-value.

^eGenes in bold and underlined have bladder cancer association. Underlined genes have additional cancer associations.

^fSee Supplemental Table 2 for supplementary references and explanations of additional cancer associations.

Table 3.

Statistically significant CNVs in North American (NA) and South Asian (SA) independent analyses.

CNVR(hg19)a	Cohort analyzed	CNV type	P-valueb	Odds ratio	Gene
chr7:57,738,495–57,785,399	SA	Del	5.97E-16	34.721	L37717
chr4:69,693,293–69,693,453	SA	Del	0.002896	7.8476	UGT2B10
chr10:38,968,525–38,968,525	SA	Del	0.035692	Infinity	ACTR3BP5
chr6:31,340,001–31,341,340	NA	Del	0.005399	8.391813	HLA-B
chr3:151,511,907–151,545,958	NA	Del	0.010001	9.931034	AADAC, MIR548H2
chr19:41,381,311–41,382,012	NA	Del	0.018479	14.69492	CYP2A7
chr11:47,353,433–47,357,561	NA	Del	0.027811	4.152047	MYBPC3
chr15:31,221,493–31,221,493	NA	Del	0.030309	Infinity	FAN1
chr22:50,658,424–50,658,424	NA	Del	0.030611	Infinity	TUBGCP6
chr4:156,967,385–156,967,385	NA	Del	0.030611	Infinity	CTSO
chr5:148,205,927–148,205,927	NA	Del	0.030611	Infinity	SH3TC2
chr20:62,172,269–62,176,133	NA	Del	0.030839	Infinity	SRMS
chr2:228,244,397–228,258,199	NA	Dup	0.020251	6.597701	TM4SF20
chr4:128,851,902–128,861,116	NA	Dup	0.021656	3.776786	MFSD8
chr11:107,671,626–107,671,626	NA	Dup	0.030611	Infinity	SLC35F2
chr5:29,634,509–29,634,509	NA	Dup	0.030611	Infinity	AK098570
chr6:124,436,570–124,436,570	NA	Dup	0.030611	Infinity	NKAIN2

Dup = duplication, Del = deletion.

^aCNV region of greatest significance and overlap coordinates.

^bTwo-tailed Fischer’s Exact P-value for deletion and duplication CNVs, respectively.

North American and South Asian specific CNVs

Three additional CNVs were significant within the South Asian population alone: deletion at chr7:57,738,495–57,785,399 a region that encodes macronuclear mRNA, deletion at chr4:69,693,293–69,693,453 a region that encodes a UDP-glucuronosyltransferase enzyme, and a deletion at chr10:38,968,525–38,968,525 which encodes an actin related protein. Fourteen additional CNVs (9 deletions and 5 duplications) were identified on analysis of the North American population alone (Table 3). CNVs identified in these analyses were mutually exclusive from those identified in the combined analysis.

Cancer-related CNVs

Eleven of the 23 duplication CNVs and 6 of the 15 deletion CNVs from the combined analysis contain genes with reported cancer associations (Supplemental Table 2). 10/15 (66.7 %) genes with cancer associations have been reported in the bladder cancer literature, including SULT1A1, GSTP1, GSTM1, STK11, SMAD4, TSC2, MUTYH, KCNQ1OT1, BRCA2, ABO (Table 2). Each gene was contained within one CNV, expect BRCA1 and GSTP1 which were both contained within a CNV duplication and CNV deletion (Table 2). Additional cancer types associated with the genes located within our CNVs are reported in Supplemental Table 2.

Discussion

We hypothesized that our genetically unique, South Asian population would provide a rare opportunity for gene discovery with significant implications for understanding and potentially modifying the genetic mechanisms responsible for BEEC. Our combined analysis revealed 36 novel CNVs and also identified two CNVs that overlapped with CNVs previously described in the literature. Additionally, similar to a recent publication from Mingardo et al., we found that a high percentage of our regions of interest include genes with known cancer associations [20].

Considerations underlying CNV discoverability

These data represent the first non-European analysis of a BEEC cohort which promises the potential for expanding the current genetic landscape but is not without additional considerations. Genetic diversity of a cohort underscores the presumed detection rate of CNVs regardless of cohort size. In our North American cohort, admixture likely contributes to a higher baseline genetic diversity. Among South Asian medical cohorts, reproductive isolation, endogamy, and consanguinity may increase genetic homogeneity of populations [22], effectively reducing genetic diversity. Within the combined analysis, it is difficult to infer the exact rationale for the number of identified CNVs. Future research is needed to characterize the CNVs identified in this study.

Overlap with prior CNVs

von Lowtzow et al. assessed 169 individuals with BEEC (recruited largely from Europe, geo-ethnic genetic origin not reported) [19]. They identified several individuals with pathogenic CNVs, including one with a previously reported duplication in 22q11. This region is also associated with 22q11 Deletion Syndrome (DiGeorge Syndrome), and CAKUT [23]. As stated in our results section, we identified two overlapping CNVs with their cohort: chr9:136,131,539–136,131,651 (encodes ABO blood group system) and chr9:137,294,228–137,308,636 (RXRA gene known to be associated with coronary artery disease). While the overlap is intriguing, given what is currently known about these regions, at this time it is difficult for us to hypothesize their potential etiologic links to BEEC. Interestingly, ABO blood group genotype and phenotype were not genetic risk factors for bladder cancer, but certain ABO alleles have been associated with clinical and histologic features of bladder cancer [24]. We cannot comment on the effects of the CNV on ABO gene products, but additional studies may uncover ABO alleles in BEEC patients associated with higher grade or recurrent tumors.

We identified a CNV deletion at chr16:28,635, 133–28,636,902 which is near a previously reported CNV (chr16:29,645,396–30,168,276) [21]. Nordenskjold et al. recently published CNVs from their cohort of 140 BE patients of European origin. Our CNV was also located within the 16p11.2 loci which spans the genome from 16:28,500,001–35,300,000 (GRCh38). Our CNV overlaps with three microdeletions reported in children with congenital anomalies of the kidneys and urinary tracts (CAKUT) and Hirschsprung disease [25], and a microdeletion associated with developmental delay [26]. Additional microdeletions overlapping with our region have also been reported [27]. This will be a region of investigation in future studies given its overlap with microdeletions identified in patients with CAKUT and Hirschsprung disease [25].

Prior genome-wide association studies (GWAS)

A recent GWAS meta-analysis combining 568 BE patients and 3241 controls of European origin identified an association with a locus containing the transcriptional enhancer ISL1 (p = 2.13 × 10⁻¹²) [28,29]. Further functional and model studies reinforced a possible causal role for ISL1 in BEEC. For example, developmental biology models were used to clarify the location of ISL1 activity in the forming urinary tract [29]. Genetic lineage analysis of ISL1-expressing cells by a lineage tracer mouse model showed ISL1-expressing cells in the urinary tract of mouse embryos [29]. We did not identify the ISL1 gene loci in any of our CNVs.

In a follow-up to their previous GWAS study that identified ISL1 as a candidate gene, Mingardo et al. published a GWAS meta-analysis of 628 patients with BE and 7352 ethnically matched controls comprising seven independent cohorts [20]. They identified eight GWAS significant loci, none of which were in linkage disequilibrium with any regions encompassed by CNVs identified in our study. This may be a result of the fact that our population is the first to include non-European patients. Additional studies on larger populations with more diversity will be needed to better elucidate the relationship between these loci and BE.

Genes with cancer associations

Multiple studies have linked BEEC with an increased risk of cancer, particularly bladder cancer [30]. Genetic pathways that result in both a congenital anomaly as well as an increased risk of cancer development are not uncommon. RASopathies, for example, are caused by germline mutations (or in rare cases by somatic mosaicism) in genes that alter the Ras subfamily and result in congenital conditions with a predisposition to tumor development.

Mingardo et al. recently proposed candidate genes that may play a possible role in BEEC-associated bladder cancer susceptibility. They postulated that different expressions turn these developmental genes on later in life. Similar to Mingardo et al., we found that a high percentage of our regions of interest include genes with known cancer associations [20]. However, the genes we identified did not overlap with the genes identified by Mingardo et al. Genes associated with bladder cancer contained within the CNVs identified in this study include SULT1A1, GSTP1, GSTM1, STK11, SMAD4, TSC2, MUTYH, KCNQ1OT1, BRCA2, and ABO. These genes portend an increased risk/susceptibility of bladder cancer (SULT1A1, GSTM1, MUTYH, GSTP1, BRCA2) and also serve as drivers of disease progression/invasion (KCNQ1OT1), predictors of tumor grade (STK11), predictors of prognosis (BRCA2, SMAD4), and potential predictors of clinical and histologic features (ABO) (Table 2). Further investigation will be needed to better define the associations between the genes implicated in BEEC and cancer as they could provide insight into methods for cancer prevention in this population.

Thoughts for the future, potential clinical implications

The current understanding of the genetic landscape of BEEC is growing but is limited by small sample sizes and an undiversified ethnic background. For populations such as the BEEC continuum, large studies composed of ethnically diverse populations are fundamental in fostering a great understanding of the genetic underpinnings of BEEC. The importance of expanding our understanding of the BEEC genetics cannot be overstated for a multitude of reasons and begins with studies similar to the current study. BEEC is a quintessential example of a disease spectrum that could significantly benefit from a robust genetic understanding given the potential devastation from the associated birth defects and risk of bladder cancer a child born with bladder exstrophy may face. The results of the current study will serve as the foundation of future studies, including but not limited to genome-wide association, whole-genome comparison to familial controls and meta-analysis with other cohorts that further our understanding and elucidate potential therapeutic targets. One potential application is the creation of genetic risk panels that can be used for in utero genetic screening and prognostication counseling of at-risk families. Additional implications include treatments targeted towards pathogenic gene variants for the development of BEEC and subsequent risk of bladder cancer. This work and similar work establishes the groundwork for the development of currently unfathomable treatment approaches which may revolutionize the prevention and/or management of BEEC.

Limitations

Our study suffers from multiple limitations. Similar to all genetic BEEC studies, we suffer from a small population size. The rarity of BEEC makes this limitation almost unavoidable. However, continued collaboration across institutions and countries will be needed to counteract this limitation in the future. As a result of our small sample size, we included patients on the least severe end of the BEEC spectrum (epispadias) with patients on the most severe end of the spectrum (cloacal exstrophy). While these conditions have similarities, they are also significantly different phenotypically and their genetic etiology could also be disparate.

Conclusion

We present the first genetic analysis of a non-European cohort of BE patients. Within our combined BEEC population, 94.7 % (36/38) of the identified CNV have not been previously reported in the BEEC literature. Forty-five percent (17/38) of the identified CNV from the combined analysis contained genes with cancer associations. 10/15 (66.7 %) cancer-associated genes have been reported in the bladder cancer literature. These findings will serve as the foundation for further analyses including genome-wide association, whole-genome, comparison to familial controls and collaboration for meta-analysis with other cohorts.

Abbreviations

BEEC

Bladder exstrophy–epispadias complex

BE

Bladder exstrophy

CNV

Copy number variations

GWAS

Genome-wide association study

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.jpurol.2025.04.018.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.