Genomic Profiling of Two Histologically Distinct Rare Urothelial Cancers in a Clinical Setting to Identify Potential Therapeutic Options for Treatment and Management of Disease

Bladder cancer is the most common malignancy involving the urinary system and the ninth most common malignancy worldwide [1], responsible for ∼150,000 deaths per year worldwide and about 5% of all new cancers in the US [2]. Invasive urothelial carcinoma involving the urinary bladder is very well known for divergent differentiation [3], with transitional cell carcinoma being the most common histological type accounting for 90% of cases. Sarcomatoid carcinomas of the bladder are rare (0.6% of bladder cancers – SEER Statistics) with only about 24 cases reported so far [4]. Similarly, neuroendocrine tumors of the bladder are equally rare (< 0.5% of all bladder cancer malignancies) with ∼12 cases being reported [5]. Prognosis with both variant morphologies is poor with an aggressive clinical course and rapid onset of metastatic disease. Given the rarity of the 2 variant morphotypes presented in the current analysis, it is difficult to draw any conclusions about the optimal treatment strategy or prognostic markers due to the small number of reported cases.

Here we present 2 cases (initial pathology, genomic findings, and clinical follow-up) of the rare pathological subtypes of urothelial cancer, a sarcomatoid tumor and a poorly differentiated urothelial carcinoma (PDUC) with some neuroendocrine differentiation. To our knowledge, this is the first report of these rare variant morphologies of urothelial carcinomas being described with clinical genomic profiling to identify potential targeted therapeutic approaches.

A 70-year-old man with a diagnosis of invasive PDUC of the right lateral wall of the bladder was initially noted by his primary care provider to have persistent microscopic hematuria. However, cystoscopy conducted 2 months later did not identify any tumors. CT scans of the abdomen and pelvis with contrast were negative for metastatic disease. Three months later, he developed macroscopic hematuria and anemia and a new 3–4 cm broad-based nodular lesion was identified in the right lateral wall of the bladder on repeat cystoscopy. Transurethral resection of these lesions and several clots in the bladder were noted during that procedure and were removed, resulting in resolution of his gross hematuria. Pathology of these lesions noted a high-grade, muscle-invasive PDUC with some neuroendocrine differentiation of the right lateral wall of the bladder (Fig. 1a, b). He was clinically staged as stage II, cT2 cN0 cM0. A PET/CT scan done at the same time showed no evidence of regional or distant metastatic disease.

A 69-year-old man with a family history of pancreatic cancer in his mother and a brother initially underwent an evaluation for hematuria followed by a partial cystectomy when a cystoscopy identified a bladder diverticulum. Histological examination identified a high-grade bladder tumor with sarcomatoid features (Fig. 1c, d). Associated lymph nodes were negative for malignancy. The tumor had been resected with negative margins and no further treatment was advised at that time. After 9 months of surveillance, a repeat cystoscopy revealed the possibility of recurrent disease and restaging MRI scan of the abdomen performed at the same time revealed the presence of bulky pelvic and retroperitoneal adenopathy. A transurethral resection of a recurrent tumor in the trigone area showed invasive sarcomatoid urothelial carcinoma, which locally progressed as confirmed on CT scans 1 month later. Chemotherapy with gemcitabine and cisplatin was initiated, but after a brief period of clinical improvement, a repeat staging CT scan after about 2.5 months of treatment revealed progression of his pelvic and retroperitoneal disease. Given a high PD-L1 score of 95%, treatment with atezolizumab was initiated but the patient progressed clinically after 2 cycles of treatment and then expired.

Ultrasound-guided core-needle biopsies were histologically evaluated for diagnosis of malignancy (Fig. 2) and formalin-fixed paraffin-embedded (FFPE) samples were shipped to The Jackson Laboratory for Genomic Medicine (JAX) CLIA lab by the ordering physicians (W.F. and C.A.T.) to be processed using a clinically validated (CLIA/CAP) assay called the ActionSeq Plus^TM. The assay included a DNA-based panel comprising 212 cancer-related genes for which all coding exons are sequenced (to allow for detection of somatic single-nucleotide variants, indels, and copy-number variants) and a RNA-based panel for 53 genes known to form fusions. Data were analyzed to identify variants of actionability, i.e., those with therapeutic, prognostic, or diagnostic relevance.

Two genes were mutated in both cases; actionable variants were identified in ARID1A (p.Q1172* – PDUC and p. H782Tfs*51 – sarcomatoid) and CDKN2A (p. N42Kfs*77 – PDUC and p. W110* – sarcomatoid). Additionally, we identified an actionable KRAS (p.G12D) variant in the PDUC and 2 actionable variants in the tumor suppressors TP53 (p. R273C) and FBXW7 (p. R465C) in the sarcomatoid cancer. While no variants of uncertain significance were identified in the PDUC, we did identify 1 variant of uncertain significance in NFE2L2 (p.E82V) in the sarcomatoid tumor (Table 1). No actionable copy-number variants or fusions were identified in either case. A comprehensive summary of the potential therapies, variant classification, FDA-approved therapies and potential clinical trials identified for the patient-specific mutational profile is listed in Table 2. While there were no FDA-approved targeted therapies identified for urothelial tumors in any of the relevant drug classes, FDA-approved drugs for these mutations in other cancers (metastatic or locally advanced solid tumors) were listed in the genomic testing report as potential options for off-label use, consistent with the AMP/ASCO guidelines for reporting of somatic variants [6].

The patient was initially started on neoadjuvant chemotherapy with cisplatin-etoposide (i.v. cisplatin 80 mg/m² on day 1, i.v. etoposide 80 mg/m² on days 1, 2, and 3). This was planned for 4 cycles. During the completion of this first cycle, a revision of his pathology stated that the neuroendocrine component of his tumor was minimal, with scant evidence of neuroendocrine cells and very weak stain synaptophysin, thus favoring the diagnosis of PDUC. After a multidisciplinary discussion, the patient’s neoadjuvant chemotherapy regimen was changed to dose-dense methotrexate-vinblastin-doxorubicin-cisplatin (ddMVAC) (i.v. cisplatin 70 mg/m², i.v. methotrexate 30 mg/m², i.v. vinblastin 3 mg/m², I.V. doxorubicin 30 mg/m² on day 1 every 2 weeks), a category 1 NCCN guideline recommendation.

The patient completed 3 cycles of neoadjuvant ddMVAC without serious complications except for moderate cytopenia, fatigue, poor appetite, and nausea. Two weeks after completion of that regimen, restaging CT scans of the abdomen, pelvis, and chest showed no radiological evidence of disease. The plan is now to perform a radical cystectomy, to be followed likely by active monitoring for development of metastatic disease. There remains a significant risk for metastatic disease recurrence and having the next-generation sequencing test results may help identify future treatment options or clinical trials if the patient develops metastatic disease.

The patient passed away in the interim during genomic testing and was unable to benefit from the results of genomic testing.

Precision medicine aims to identify novel targets for treating rare and difficult-to-treat cancers. To advance the field and identify novel treatment options for patients, broad molecular profiling is needed. Here we describe the molecular profiling results of 2 cases with urothelial cancers that identified potential treatment options for the patients. We identified some interesting profiles which offer the opportunity to identify new treatment options and may support novel molecular disease classifications in the future.

Mutations in 2 genes, ARID1A and CDKN2A, were identified in both our cases and they have been shown to occur in 25 and 5% of urothelial carcinomas, respectively, indicative of the role of dysregulation of chromatin remodeling and cell cycle control in these rare, highly aggressive pathological subtypes. Evaluation of 347 cases across a broad complement of cancers demonstrated that CDKN2A was more likely to be independently associated with ARID1A in a multivariate analysis (OR D 3.98, 95% CI: 1.36–11.8, p D 0.01, p = 0.01).

In addition, we identified a mutation in KRAS in 1 of the 2 cases (PDUC). Mutations in the RAS oncogenes (HRAS, KRAS, and NRAS) are found in ∼11–13% of bladder tumors and occur in all stages and grades, with mutations in KRAS being the least frequent [7]. Profiling of urothelial cancers across multiple studies have not identified co-occurrence of KRAS with CDKN2A mutations [8], however, a single study has reported the co-occurrence of a KRAS substitution with an ARID1A truncation mutation in a stage IV high-grade urothelial carcinoma in a 51-year-old male [9] indicative of the rarity of this genomic profile. The co-occurrence of the ARID1A and KRAS mutation in the PDUC could therefore possibly account for the aggressive and poorly differentiated nature of the tumor. In contrast to the PDUC, the sarcomatoid urothelial carcinoma had actionable mutations in 3 other genes (TP53, FBXW7, and NFE2L2) known to harbor aberration in bladder cancer. TP53 is the most commonly mutated gene in bladder cancer observed at a frequency of ∼15% in non-muscle-invasive tumors as is the sarcomatoid tumor. Mutations in TP53 have been shown to co-occur with either ARID1A or CDKN2A [9] with aberration in the CDK pathway genes being the second most common mutation across most cancers next to TP53 [10]. Mutations in FBXW7 are observed in 10% of bladder cancers across all stages [7] and mutations in NFE2L2 are observed in ∼8–11% of bladder tumors [11], particularly the basal subtype [12], based on samples in TCGA suggesting a potential role for the oxidative stress pathway in progression of bladder cancer. While co-occurrence of FBXW7 mutations with TP53 has been reported in a single case report, of a high-grade, advanced-stage tumor from a 71-year-old woman, from the lymph node metastasis specimen which had a micropapillary architecture [9], mutations in NFE2L2 have not been shown to co-occur with TP53 mutations [12] across a comprehensive study evaluating ∼400 bladder cancers, contributing to the novelty of this genomic profile and providing further insight into the possible genomic changes involved in the development and evolution of sarcomatoid urothelial carcinoma.

There are currently no molecularly targeted agents approved for the treatment of bladder cancer despite a significant amount of effort in recent times to profile bladder tumors. By following the AMP/CAP/ASCO guidelines on reporting actionability of cancer-specific genes, we were able to identify potential drug classes for clinical consideration (Table 2). These included FDA-approved drugs (for a different indication) as well as some experimental therapies. In addition, we were able to identify a number of clinical trials for each one of the patients based on their genomic profile, although all of the trials were recruiting patients broadly across solid tumors. Most noteworthy is a trial listed for the ARID1A and KRAS mutations (NCT02576444) which is a phase II study of the PARP inhibitor olaparib (AZD2281) alone and in combination with AZD1775, AZD5363, or AZD2014 in advanced solid tumors across different molecular profiles including any mutation in ARID1A, as well as protocol-defined mutations in KRAS and TP53. The TP53 mutation identified in the sarcomatoid case did not fit inclusion criteria, but the mutations in ARID1A and KRAS were eligible, resulting in the inclusion of this trial in the genomic testing report. All in all, molecular profiling has identified a number of genomic-marker-driven treatment approaches for patients with urothelial cancers, but these treatments should be accessed through clinical trial enrollment rather than off-label use of drugs.

Though the ARID1A and CDKN2A genes identified in our cases are amongst those known to be commonly associated with urothelial carcinoma, to our knowledge this study is the first clinical case report comparing and identifying unique genomic profiles in 2 rare pathological subtypes of urothelial cancer. The identification of the co-occurring combinations of ARID1A/KRAS in the PDUC and the TP53/FBXW7 in the sarcomatoid subtype adds a layer of potential stratification across these two histological subtypes. Further, our documentation of these rare co-occurrences in high-grade urothelial carcinomas adds an additional case report to the existing literature of single cases currently reported for each combination, amplifying the potential of broad-based targeted clinical trials. In addition, given the rarity of these tumors, with only 24 clinical cases of sarcomatoid tumors [4] and ∼12 cases of urothelial carcinoma with neuroendocrine component [5] being reported, our evaluation also adds to the knowledgebase of these rare urothelial carcinoma morphologies, both from a genomic profile perspective as well as identification of potential targeted therapies and trials since there are currently no targeted FDA-approved therapies specific for urothelial cancer.

We gratefully acknowledge the participation of the patients. The authors are grateful for the funding support from the Harold Alfond® Foundation that facilitated the genomic testing in collaboration with The Jackson Laboratory as part of the Maine Cancer Genomics Initiative (MCGI) led by J.R.

Patient specimens were sent to the CLIA/CAP-accredited laboratory at The Jackson Laboratory for Genomic Medicine as part of the patients’ routine care and management for genomic testing. Informed written consent was obtained by the ordering physician from the patient for tumor profiling using the JAX ActionSeq Plus^TM assay. The use of data sets is allowed for research reports and scientific publications.

The authors declare that there are no conflicts of interest.

A.N. Hesse, W. Fabricius, C.A. Thomas, R. Gaindh, and R. Christman contributed equally to this work. J. Rueter and H.V. Reddi: equal contribution.