Large, Nested Variant of Urothelial Carcinoma Is Enriched with Activating Mutations in Fibroblast Growth Factor Receptor-3 among Other Targetable Mutations

Simple Summary Urothelial carcinoma of the large, nested variant is a specific histological morphology subtype of urothelial carcinoma. Although it is a rare variant, it requires specific attention due to its bland histology and the fact that it may potentially be missed in routine biopsies. In this study, we identify Fibroblast Growth Factor Receptor-3 (FGFR-3) as the most common mutation present in this subtype among other potential targetable mutations. All our cases of this variant also harbored other potentially actionable mutations in other genes, which could also be amenable to novel targeted therapy. Patients with this variant may benefit from additional molecular screening to identify potential therapeutic targets that could improve the clinical outcome of such patients. Abstract The large, nested variant of urothelial carcinoma (LNVUC) is characterized by bland histomorphology mimicking that of benign von Brunn nests. In the current study, we aimed to investigate the Fibroblast Growth Factor Receptor-3 (FGFR-3) activation and missense mutation in 38 cases, including 6 cases diagnosed with LNVUC and 32 with metastatic invasive urothelial carcinoma (UC). Initially, six formalin-fixed paraffin-embedded (FFPE) tissue samples of the LNVUC were subjected to whole-exome sequencing (WES), and then we performed targeted sequencing on 32 cases of metastatic invasive UC of various morphological subtypes, which were interrogated for the FGFR3. Our results revealed 3/6 (50%) LNVUC cases evaluated by WES in our study showed an activating mutation in FGFR-3, 33% showed an activating mutation in PIK3CA, and 17% showed activating mutation in GNAS or MRE11. Additionally, 33% of cases showed a truncating mutation in CDKN1B. All LNVUC in our study that harbored the FGFR-3 mutation showed additional activating or truncating mutations in other genes. Overall, 6/32 (18.75%) cases of random metastatic invasive UC showed missense mutations of the FGFR-3 gene. The LNVUC variant showed the higher incidence of FGFR-3 mutations compared to other types of mutations. Additionally, all LNVUC cases show additional activating or truncating mutations in other genes, thus being amenable to novel targeted therapy.


Introduction
Urothelial carcinoma is the most common pathological (histological) type of bladder cancer [1]. Approximately 75% of cases are classified as pure urothelial carcinoma and 25% as variants of mixed or less common variants such as squamous and adenocarcinoma [2]. The recent 2022 edition of WHO recognized several of these urothelial carcinoma variants as having important prognostic and predictive value [3].
The nested variant of urothelial carcinoma is a very rare subtype that is characterized by bland nests of neoplastic urothelial cells that are often seen invading muscularis propria [2]. The large, nested variant (LNVUC) is defined as a spectrum of the nested variant and is characterized by larger nests of invasive neoplastic urothelial cells [4]. The nested variant of urothelial carcinoma, and the large, nested variant as a spectrum, was first described in 1979 but was not recognized by the WHO until 2004. The bland histomorphology of this variant can mimic those of a benign von Brunn nests, leading to misdiagnosis, particularly in small biopsies, causing a delay in recognizing this lesion and therefore the treatment of patients [5]. Studies have shown that despite bland histomorphology, this variant is associated with poor outcome [6,7]. Moreover, at the time of diagnosis, approximately 69% of nested variants are in advanced stages (pT3/4), and 19% have nodal involvement. As with all variants subtypes of urothelial carcinoma, the risk of recurrence and progression of the nested variant is increased based on many factors including the presence of a residual tumor and in situ component [8], which, understandably, can be easily missed in this variant due to the bland morphological appearance.
Fibroblast growth factor receptor-3 (FGFR-3) is one of the four highly conserved fibroblast growth factor receptor tyrosine kinases (RTKs) and has been shown to have a role in tumor growth and survival regulations [9][10][11]. FGFR-3 has been implicated in multiple malignant neoplasms including multiple myeloma, urothelial and cervical carcinoma [12][13][14]. Despite this, several studies have focused on FGFR-3 as therapeutic and prognostic marker in urothelial carcinoma [10,15]. The clinical significance of FGFR mutations is highlighted with the development of the pan-FGFR inhibitor, Erdafitinib, and Pembrolizumab, which was approved by the Food and Drug Administration [13,16,17]. This has provided new hope in the management and treatment of malignancies with FGFR-3 mutations [18].
Studies have shown FGFR-3 mutation to be among the most common mutated oncogene in urothelial carcinoma overall [19]. FGFR-3 mutations have been reported in about 75% of noninvasive urothelial carcinoma and about 15-20% of high-grade tumors [20]. However, most studies have focused on the presence of this mutation in the plasmacytoid variant of urothelial carcinoma, another rare and aggressive variant of urothelial carcinoma [21][22][23], but only a few have focused on the nested variant.
This study is designed to characterize the molecular backgrounds of the large, nested variant of urothelial carcinoma, in the hope of identifying potential targetable mutations linking this specific histomorphology with specific genetic mutations.

Patient Samples
In the current study, we recruited 38 metastatic UC samples of which 6 cases were diagnosed with LNVUC at Alberta Precision labs/University of Calgary Cumming School of Medicine between 2015 and 2019. Of these 6 cases, 2 belonged to the same patient (cases 2 and 3), who developed a primary LNVUC in the lower urothelial system (bladder) which was treated with partial cystectomy, and subsequently, after this diagnosis, he developed another primary LNVUC in the upper urothelial system (ureter) which was treated with bilateral nephroureterectomy. These cases were re-reviewed by an experienced genitourinary pathologist to confirm the diagnosis of this variant and select the most suitable areas for sequencing. The remaining 32 cases of metastatic invasive urothelial carcinoma, including 3 cases of the LNVUC, were also analyzed for FGFR-3 target seq.
All cases were classified based on the 2016 edition of the World Health Organization Classification of Tumors of the Urinary System. LNVUC cases demographic data, stage at the time of diagnosis, treatments and outcomes up to date were documented for each case (Table 1), and further demographics about the 32 cases of metastatic invasive urothelial carcinoma cases are provided in Table 2.

DNA Extraction and Whole-Exome Sequencing
Pathologically reviewed samples confirmed the diagnosis and determined tumor content, and marked tumor areas on hematoxylin and eosin (H&E) slides used to accurately obtain tissue samples from formalin-fixed paraffin-embedded (FFPE) samples. Briefly, the study pathologists cored or scrolled the selected tumor areas. Then tumor DNA was extracted using the QIAamp DNA FFPE Tissue kit (Qiagen, Cat # 56404, Hilden, Germany). Germline DNA was extracted from normal kidney tissue adjacent to the tumor. Qubit used to quantify the DNA using Qubit DNA HS assay (Life Technologies, Carlsbad, CA, USA). Whole-exome sequencing (WES) was carried out by The Centre for Applied Genomics (TCAG), Toronto, ON, Canada. Briefly, 750 ng of DNA was used for WES exome library preparation and sequencing using SureSelect XT Human All Exon V5 Kit (Agilent Technologies, Santa Clara, CA, USA). Genomic DNA was fragmented to 200-bp on average using a Covaris LE220 instrument. Prior to ligation, sheared DNA was endrepaired, and the 3 ends were adenylated on adapters with overhang-T. Then, the genomic library was amplified by PCR using 10 cycles and hybridized with biotinylated probes that target exonic regions; the enriched exome libraries were amplified by an additional 8 cycles of PCR. The exomic libraries were validated using DNA High-Sensitivity chips on a Bioanalyzer 2100 (Agilent Technologies) for size and by qPCR using the Kapa Library Quantification Illumina/ABI Prism Kit protocol (KAPA Biosystems) for quantities. Exome libraries were pooled and sequenced with the TruSeq SBS sequencing chemistry using a V4 high throughput flowcell on a HiSeq 2500 platform (Illumina Inc., San Diego, CA, USA), as per Illumina's recommended protocol.

Data Alignment and Validation
Around 6-8 gigabases of raw paired end data of 126-bases were generated per exome library. Reads were aligned to the hg19 build human reference genome using BWA (version 0.5.9). PCR duplicates were marked using picard-tools-1.108, and local re-alignment and base recalibration were performed using GATK 1.1-28. Variants (SNV, indel) were called using GATK UnifiedGenotyper 1.1-28. An Annovar-based pipeline was used for adding gene-based, feature-based and frequency-based annotations for variant filtering and prioritization [24].

Results
Whole-exome sequencing results (Table 3) for the six cases of the large, nested variant of urothelial carcinoma showed 3 cases (50%) to harbor positive activating mutation in FGFR-3. Two cases showed an activating mutation in PIK3CA (33%), and one showed activating mutations in GNAS (17%), and another case showed mutations in MRE11 (17%).
A truncating mutation in CDKN1B was seen in two cases (33%), and truncating mutations in CDKN2A, ARID1B, ARID1A and KDM6A was seen only in one case each (17%). It was interesting to note that cases that showed activating mutations in FGFR-3 also showed additional activating and truncating mutations in other genes including PIK3CA. Detailed genetic mutations detected in the six cases studied are presented in Table 3.
All our cases that harbored FGFR-3 mutations showed additional activating or truncating mutations. One case showed simultaneous mutations in FGFR-3, PIK3CA, CDKN1B, ARID1B and PPP2R1A. Another case showed mutations in FGFR-3 as well as MRE11 and KDM66, and an additional case had simultaneous mutations in FGFR-3, CDKN1B and CDKN2A. Of the cases that did not harbor any FGFR-3 mutations, one case showed simultaneous mutations in PIK3CA, ARID1A and GNAS (Table 3). As exhibited in Table 2, FGFR-3 mutation fusions by targeted sequencing were assessed in the 32 cases, including three cases of the six WES-performed samples series, which did not show FGFR-3 mutations by WES. Of those, 6/32 cases (18%) were positive for FGFR-3 missense mutations (two S249C, three Y373C and one G370C mutations). Three out of six cases exhibited dedifferentiated histology (poorly differentiated or squamous/sarcomatoid differentiation), five of six were from metastatic sites, and one was from the upper urothelial tract. However, no significant association was noted to specific histopathological morphology or metastatic site.

Discussion
Our results indicates that the FGFR-3 mutation is among the most common mutated oncogene in urothelial carcinoma, and it is even more common in the LNVUC. In our small study, we report 50% of cases as having a positive activating mutation in FGFR-3.
Additionally, we observed that many cases of LNVUC harbor simultaneous multiple activating mutations. In our series, four out of six (66%) LNVUC cases showed multiple activating mutations in oncogenes and truncating mutations in tumor suppressors, simultaneously. Of note, one patient who had two separate primaries of bladder and renal pelvis LNVUC demonstrated different mutational landscape between the two tumors, thus supporting different mutational landscape even in same-patient tumors, based on location.
Based on the targeted FGFR-3 sequencing of 32 cases of metastatic invasive urothelial carcinoma, 16% of cases showed FGFR-3 missense mutations. None of the cases that did not harbor FGFR-3 mutations by whole-exome sequencing showed FGFR-3 mutations, raising the possibility that point mutations in FGFR-3 gene are likely more frequent in LNVUC.
To illustrate our results in relation to published reports using genomic sequencing data, we added a review table for FGFR-3 analysis and comparison table analysis of two studies, including a study by Pietzak et al. [25] and the TCGA data, to compare the rate of specific mutations across urothelial carcinomas.
As demonstrated in our data, in non-muscle-invasive urothelial carcinoma, our results showed a slight but non-significantly higher incidence of FGFR-3 mutation, whereas compared to data provided by TCGA, the incident of FGFR-3 mutations in our variant was significantly higher compared to unselected variants of urothelial carcinoma. The incidence of CDKN1B, GNAS, MRE11 and PPP2R1A mutations was also significantly higher in our cases compared to both studies.
Similarly, other studies characterizing non-muscle-invasive and muscle-invasive highgrade urothelial carcinoma reported similar incidence of FGFR-3 mutations, ranging from 11% to 52% suggesting that the rate of FGFR-3 mutations may vary significantly depending on the methods used, site of assessment and variants of urothelial carcinoma included in the studies [26][27][28][29] (Table 4).
As in our study, Weyerer et al. [5] focused mainly on the large, nested variant of urothelial carcinoma, but they reported that 97% of their pure nested variants showed FGFR-3 mutation, whereas only 13% of the mixed tumor variant harbored this mutation. Their finding raises the possibility of different neoplastic pathways for mixed and pure LNVUC in their study.
It is also important to note that LNVUC and advanced UC shows a response to pembrolizumab especially with the recurrent LNVUC; however, our study does not focus on therapeutic strategies but the incidence associated with the presence of FGFR3 mutations [17,18].
Overall, all of these studies document a high incidence of FGFR-3 mutations in urothelial carcinoma, and they support that the LNVUC variant may exhibit an even higher incidence of FGFR-3 mutations especially in more pure histological type. Additionally, the rate of FGFR-3 mutations varies depending on the samples analyzed, whether it is resection or TURBT as well as histological grade, patient's demographics, and underlying risk factors such as history of smoking. Finally, the studies document that FGFR-3 mutations usually occur in association with other activating or truncating mutations, especially in the PI3CKA pathway, and that LUCNV may harbor simultaneous activating and truncating mutations making them amenable for targeted therapy. Large, nested urothelial carcinoma (LNVUC), urothelial carcinoma (UC), lymph node (LN), transurethral resection of the bladder (TURB), polymerase chain reaction (PCR).

Conclusions
Our study provides further evidence of the promising role for FGFR-3 in the diagnosis and treatment of the large, nested variant of urothelial carcinoma, possibly implicating other targetable pathways compared to random unselected variants of urothelial carcinoma. We do acknowledge the limitations of our study, including small sample size and the fact that most cases were those of TURB. To better evaluate the role of these mutations in this rare variant of urothelial carcinoma, more studies, with a larger number of cases, focused on histomorphology, grade, stage as well as patient demographics and prognosis should be designed.