A Comparison of Adult Rhabdomyosarcoma and High Grade Neuroendocrine Carcinoma of the Urinary Bladder Reveals Novel PPP1R12A Fusions in Rhabdomyosarcoma

Abstract

Some rhabdomyosarcomas and sarcomatoid carcinomas with heterologous rhabdomyosarcomatous elements resemble high-grade neuroendocrine carcinoma, creating a diagnostic difficulty. The purpose of this study was to characterize the overlap of adult genitourinary rhabdomyosarcomas, excluding those occurring at paratesticular sites, with high-grade neuroendocrine carcinoma and identify features helpful in their separation. Seventeen cases of rhabdomyosarcoma (eleven from the urinary bladder and three each from kidney and prostate) were compared to ten cases of high-grade neuroendocrine carcinoma from the urinary bladder. These tumors were analyzed by immunohistochemistry for desmin, MyoD1, myogenin, chromogranin, synaptophysin, CD56, TTF1 and ASCL1 and RNAseq was performed on four cases of bladder rhabdomyosarcoma (two rhabdomyosarcomas and two sarcomatoid-rhabdomyosarcoma) and ten cases of bladder high-grade neuroendocrine carcinoma. This was compared to public data from 414 typical urothelial carcinomas from The Cancer Genome Atlas dataset. Morphologic and immunophenotypic overlap with high-grade neuroendocrine carcinoma was seen in half of the bladder tumors, which included four rhabdomyosarcomas and two sarcomatoid-rhabdomyosarcoma. RNAseq confirmed expression of neuroendocrine markers in these cases (two rhabdomyosarcomas and two sarcomatoid-rhabdomyosarcoma). Differential neuroendocrine differentiation was highlighted by ASCL1 protein expression only in high-grade neuroendocrine carcinoma. Moreover, both a pure alveolar rhabdomyosarcoma and sarcomatoid-rhabdomyosarcoma of the urinary bladder demonstrated a fusion involving PPP1R12A. In summary, adult rhabdomyosarcomas of the urinary bladder are molecularly distinct from high-grade neuroendocrine carcinomas based on specific patterns of expression of myogenic and epithelial to mesenchymal transition-related transcription factors as well as the presence of a novel PPP1R12A fusion which is seen in a subset of cases.

1.0. Introduction

Adult rhabdomyosarcomas (RMS) involving the prostate, kidney and urinary bladder are rare entities, and have for the most part been reported as isolated case reports or small series (1–12). The largest report to date of adult (defined as older than 16 years) sarcomas of these organs, diagnosed over a 25 year span at a referral center, consisted of 74 cases, of which RMS comprised only 13 cases (renal: 0; bladder: 4; prostate: 9) (13). Likewise, sarcomatoid carcinomas with heterologous rhabdomyosarcomatous differentiation (S-RMS) are very rare, with no cases being reported in a recent series of 28 cases of the bladder that were diagnosed over a 13 year span (14).

Recent small studies have reported that adult RMS including tumors of the urinary bladder can resemble small cell carcinoma and express neuroendocrine markers (5, 15–17). This includes the documentation of ultrastructural features of both myogenic and neuroendocrine differentiation characterized by the presence of both bundles of filaments containing abortive Z bands and dense core neurosecretory-type granules, in the same cells (5).

Misdiagnosis of RMS of the urinary bladder as small cell carcinoma has significant clinical implications due to the chemosensitive nature of the latter. Indeed, such high-grade neuroendocrine carcinomas (HGNEC) are optimally treated with initial systemic chemotherapy, with recent studies showing significantly improved outcomes following receipt of neoadjuvant chemotherapy followed by definitive local therapy (18–20). On the other hand, due to the limited number of RMS identified in the literature, outcomes are poorly defined, with available data suggesting a poor prognosis (17, 21). The clinical implication of confusing S-RMS with HGNEC is less clear, as the optimal treatment approach for S-RMS remains to be defined, although the chemosensitivity of this entity to the agents which target HGNEC has not been well established. The aim of this study was to further characterize adult RMS of the urologic organs and compare these tumors to HGNEC to identify features helpful in their separation.

2.0. Materials and Methods

2.1. Patient Specimens and Selection of Cases

This study was approved by the institutional review board at Mayo Clinic, Rochester, MN. We identified a total of 17 adult genitourinary RMS, excluding paratesticular tumors, that were diagnosed between 2002 and 2016. Sixteen cases were evaluated at Mayo Clinic, including 10 received in consultation and an additional case was evaluated at Memorial Sloan Kettering Cancer Center. Based on pre-existing diagnostic criteria including immunophenotypic evidence of myogenic differentiation characterized by expression of desmin, myogenin and MyoD1, the original diagnosis of RMS was confirmed for 15 cases. We further included two additional cases for analyses, one originally diagnosed as a leiomyosarcoma which was reclassified following review of renal smooth muscle neoplasms and another case of S-RMS, originally diagnosed as a small cell carcinoma of the bladder, which was identified following review of bladder HGNEC (18, 22, 23).

2.2. Immunohistochemistry

Tumors were immunostained for desmin (Leica, Novocastra, Buffalo Grove, IL, clone DE-R-11, 1:50 – 1:100), myogenin (Dako, Carpinteria, CA, clone F5D, 1:25 – 1:50), MyoD1 (Ventana, Tucson, AZ, clone EP212, 1μg/mL), synaptophysin (Leica, Novocastra, Buffalo Grove, IL, clone 27G12, 1:100 – 1:200), chromogranin A (Ventana, Tucson, AZ, clone LK2H10, 1μg/mL), CD56 (Dako, Carpinteria, CA, clone 123C3, 1:50 – 1:100), TTF1 (Leica, Novocastra, Buffalo Grove, IL, clone SPT24, 1:200) and ASCL1 (BD Pharmingen, San Jose, CA, Clone: 24B72D11.1, 1:100). Immunohistochemical results were dichotomized into cases with absent expression and those with positive expression, defined by staining noted in at least 5% of neoplastic cells. Immunophenotyping of all RMS was performed on representative whole slide sections.

ASCL1 expression in rhabdomyosarcomas was compared to 70 patients with HGNEC and 110 patients with typical urothelial carcinomas (UC) from specimens obtained from patients treated with radical cystectomy at our institution between 1987 and 2014 (18). Tissue microarrays, with four 1.0mm cores representing each HGNEC and typical UC, were immunostained for ASCL1 for this purpose. In addition, whole slide sections of 47 cases of non-urinary bladder rhabdomyosarcomas were immunostained for ASCL1, to assess specificity.

2.3. RNA Sequencing and Data Extraction from The Cancer Genome Atlas Project

Library preparation and gene expression profiling by RNA sequencing (RNAseq) was performed at the Mayo Clinic Genomic Facility following Institutional Review Board approval using formalin fixed paraffin embedded tissue (2 alveolar RMS, 2 S-RMS) as well as archived frozen tissue (10 HGNEC).

All analyses were in the R programing environment. RNA sequence data were mapped on the latest reference genome (HG38) using the STAR aligner and following computational pipeline described in the Genomic Data Commons (GDC) website (https://docs.gdc.cancer.gov/Data/Bioinformatics_Pipelines/Expression_mRNA_Pipeline). This approach allowed us to combine expression count matrices of the RNA-seq data generated at Mayo Clinic and The Cancer Genome Atlas (TCGA) data containing 414 typical UC. Normalized expression matrix for all RNAseq files was generated using edgeR package and then log2 transformed.

Identification of fusion events was by MAP-RSeq pipeline, STAR-Fusion, and FusionInspector a component of STAR-Fusion (Haas B. et al. (2015) STAR-FUSION, https://github.com/STARFusion/STAR-Fusion, 13 April 2016, date last accessed) (24). To validate fusion events, RNA was used in RT-PCR reactions and the PCR amplicons were inserted into a TA cloning vector. The inserted amplicon was then sequenced by pyrosequencing.

2.4. Statistical analysis

Continuous clinicopathological variables were analyzed with frequency counts and percentages. Tests to assess statistical significance were two-sided, with p<0.05 considered to be statistically significant.

3.0. Results

3.1. Clinicopathologic Features

Clinicopathologic features of the 17 cases of RMS and S-RMS are provided in Table 1, including 3 renal, 11 bladder and 3 prostatic tumors. Mean age at diagnosis was 57 years (range, 19–84) (renal: 58 years, range, 54–64; urinary bladder: 60 years, range, 19–84; prostate: 44 years, range, 23–55). No gender predilection was identified in the bladder tumors. The five cases classified as S-RMS had a concurrent UC component (in situ, n=2 and UC not otherwise specified, n=3), and all tumors were positive for cytokeratin. Six cases of pure RMS of the urinary bladder included alveolar (n=4), embryonal (n=1) and spindle cell (n=1) subtypes. Of the renal and prostate RMS, none were associated with a sarcomatoid carcinoma component, and were classified as spindle cell (n=4) and embryonal subtypes (n=2). None of the renal or prostate rhabdomyosarcomas had appreciable morphologic overlap with HGNEC. However, four cases of bladder RMS (all pure alveolar) and two S-RMS exhibited significant morphologic overlap with small cell carcinomas (Figure 1).

Table 1.

Adult Genitourinary Rhabdomyosarcomas: Clinicopathologic Features

	Kidney	Urinary Bladder	Prostate	Summarized Data
Number of Cases	3	11	3	17
Mean Age (years, range)	58 (54–64)	60 (19–84)	44 (23–55)	57 (19–84)
Gender
-Male:	2	6	3	11
-Female:	1	5	0	6
Histologic Subtype
-S-RMS:	0	5*	0	5
-Alveolar:	0	4**	0	4
-Embryonal:	0	1	2	3
-Spindle Cell:	3	1	1	5
Morphologic Overlap with Small Cell Carcinoma
-S-RMS:	0	2/5	0	2/5
-Alveolar:	0	4/4	0	4/4
-Embryonal:	0	0/1	0/2	0/3
-Spindle Cell:	0/3	0/1	0/1	0/5
Neuroendocrine Differentiation
-Synaptophysin:	1/3	8/11	0/3	9/17
-Chromogranin:	0/3	1/11	0/3	1/17
-CD56:	3/3	10/10	2/3	15/16
-TTF1:	0/3	2/10	0/3	2/16

Abbreviations: sarcomatoid carcinomas with heterologous rhabdomyosarcomatous elements (S-RMS).

^*Documented gene rearrangement in 1 case (PPP1R12A, RNAseq).

^**Documented gene rearrangements in 2 cases include FOXO1 (n=1; FISH) and PPP1R12A (n=1; RNAseq).

Figure1: Adult Genitourinary Rhabdomyosarcoma: Histopathology.

Representative H&E stained images of pure rhabdomyosarcomas of the urinary bladder (A, B; 400X magnification) and a sarcomatoid carcinoma with heterologous rhabdomyosarcomatous elements showing morphologic overlap with small cell carcinoma (C; 400x magnification).

Clinical follow up of at least 6 months was available for 3 patients with bladder RMS and 2 patients with bladder S-RMS. Four of these patients had tumors which exhibited morphologic/immunophenotypic overlap with HGNEC. The first, a 74-year-old male with S-RMS, died of disease related complications including metastasis to the brain, at 10 months. The second patient, a 69-year-old male with S-RMS, developed bone metastases at 6 months of follow up. The third, a 72-year-old male with a pure RMS, died of disease at 110 months. This included widely metastatic disease (brain and lung). The fourth patient, a 19-year-old male with a FOXO1 rearranged alveolar RMS, was alive with a biopsy proven left supraclavicular lymph node metastasis at 27 months.

Finally, a 37-year-old female with an embryonal RMS that did not exhibit morphologic/immunophenotypic overlap with HGNEC, was alive without disease at 158 months, following initial management with a partial cystectomy.

3.2. Immunohistochemistry

Immunohistochemistry was performed to assess expression of chromogranin, synaptophysin, CD56, and TTF1. While a combination of desmin, MyoD1, and myogenin confirmed myogenic differentiation for pure RMS or S-RMS, aberrant expression of neuroendocrine markers was relatively common, as we have previously reported (Figure 2, Table 1) (15). CD56 demonstrated the least specificity, being detected in 15 (of 16, 94%) RMS. This was followed by synaptophysin (9 of 17, 53%), while the expression of expression of chromogranin (1 of 17, 6%) and TTF1 (2 of 16, 13%) was infrequent. The morphologic overlap with small cell carcinoma in 6 cases of bladder RMS and S-RMS, coupled with the expression of neuroendocrine markers (CD56: 5 of 5, 100%; synaptophysin: 5 of 6, 83%; TTF1: 2 of 5, 40%; chromogranin: 1 of 6, 17%) highlights the diagnostic pitfall of misclassifying these cases as small cell carcinomas.

Figure2: Adult Genitourinary Rhabdomyosarcoma: Immunohistochemistry.

Representative images detailing the immunophenotype of rhabdomyosarcomas of the urinary bladder, which demonstrate morphologic overlap with small cell carcinoma (Case1: A-H, Case2: I-P, 200 X magnification). H&E stained images (A, I), immunohistochemistry for Desmin (B, J), Myogenin (C, K), MyoD1 (D, L), Synaptophysin (E, M), Chromogranin (F, N), CD56 (G, O), and TTF1 (H, P) is shown.

3.3. Gene Expression Analysis

RNASeq analysis was performed on 4 cases of bladder RMS (pure alveolar: 2 and S-RMS: 2 ) and 10 cases of HGNEC. This analysis was compared to 414 typical UC from the publicly available TCGA database. The dataset was validated by assaying for a gene expression signature characteristic of UC (not shown), that was not enriched in the other tumor types (UC vs HGNEC, p<0.005), including keratin 7 (KRT7), keratin 19 (KRT19), uroplakin 2 (UPK2), claudin 4 (CLDN4), GATA binding protein 3 (GATA3), CD44 molecule (CD44), tumor protein p63 (TP63) and cyclin D1 (CCND1).

Assessment of gene expression for markers of neuroendocrine differentiation mirrored our immunohistochemical results, with increased expression of synaptophysin (SYP), chromogranin A (CHGA), chromogranin B (CHGB) and neural cell adhesion molecule 1 (NCAM1, which encodes for the CD56 protein) showing high levels of expression in both HGNEC and RMS/S-RMS, with no expression detected in UC (Figure 3). Less common genes associated with neuroendocrine differentiation such as proprotein convertase subtilisin/kexin type 1 (PCSK1) and secretogranin II (SCG2) showed a similar pattern, as well. However, at the gene expression level, Insulinoma-associated 1 (INSM1) and achaete-scute family bHLH transcription factor 1 (ASCL1) did not show significant changes. Statistical analysis for RMS, however, was limited by the restricted number of cases analyzed (n=4).

Figure3: Adult Genitourinary Rhabdomyosarcoma: Neuroendocrine Gene Expression.

Relative gene expression (log₂) for urothelial carcinoma (n=414), high-grade neuroendocrine carcinomas (n=10), sarcomatoid carcinoma with rhabdomyosarcomatous elements (n=2) and pure rhabdomyosarcomas (n=2) for PCSK1 (A), SCG2 (B), NCAM1 (C), CHGB (D), CHGA (E), SYP (F), ASCL1 (G), and INSM1 (H) is shown.

Furthermore, gene expression of transcription factors that drive both myogenic differentiation (Figure 4) and epithelial to mesenchymal transitions (Figure 5) was assessed (14, 21). As expected, myogenic transcription factors including Myogenin and myogenic differentiation 1 (MYOD1) showed a significant increase in expression in RMS/S-RMS compared to cases of UC and HGNEC, consistent with results of immunohistochemistry. Meanwhile, expression of the myogenic transcriptional regulator forkhead box O1 (FOXO1) was similar in RMS, while paired box 3 (Pax3), paired box 7 (Pax7) and BCL6 corepressor (BCOR) did not show a difference in expression. The transcription factors zinc finger E-box binding homeobox 1 and 2 (ZEB1/ZEB2) that are known to drive epithelial to mesenchymal transition showed highest expression in S-RMS compared to UC, HGNEC and pure RMS. Other regulators such as snail family transcriptional repressor 2 (SNAI2) and twist family bHLH transcription factor 1 (TWIST1) did not exhibit differential expression.

Figure4: Gene Expression of Transcriptional Regulators of Myogenesis.

Relative gene expression (log₂) for urothelial carcinoma (n=414), high-grade neuroendocrine carcinomas (n=10), carcinoma with rhabdomyosarcomatous elements (n=2) and pure rhabdomyosarcomas (n=2) for MYOD1 (A) and Myogenin (B), FOXO1 (C), PAX7 (D), PAX3 (E) and BCOR (F) is shown.

Figure5: Gene Expression of Transcriptional Regulators of Epithelial to Mesenchymal Transition.

Relative gene expression (log₂) for urothelial carcinoma (n=414), high-grade neuroendocrine carcinomas (n=10), carcinoma with rhabdomyosarcomatous elements (n=2) and pure rhabdomyosarcomas (n=2) for ZEB1 (A) and ZEB2 (B), SNAI2 (C) and TWIST1 (D) is shown.

3.4. Immunohistochemistry, ASCL1

ASCL1 expression has been found to be a specific marker of neuroendocrine differentiation in several tumor types and in our series of lung adenocarcinomas with neuroendocrine differentiation (25, 26). Herein, ASCL1 immunohistochemistry showed a lower sensitivity for the detection of HGNEC relative to other neuroendocrine markers (present in 34 of 70 cases; 49%) (Figure 6, Supplementary Table 1). However, ASCL1 immunohistochemical positivity was highly specific for HGNEC, as expression was not seen in typical UC (0 of 110 cases), pure RMS (0 of 7 cases) and S-RMS (0 of 2 cases). To assess specificity, 47 cases of non-urinary bladder rhabdomyosarcomas (16 embryonal, 1 spindle cell, 8 pleomorphic and 17 alveolar) were tested for ASCL1 expression by immunohistochemistry and all cases were negative.

Figure6: Adult Genitourinary Rhabdomyosarcoma: Immunohistochemistry, ASCL1.

Representative H&E stained images and corresponding ASCL1 expression in a typical urothelial carcinoma (A, B, 200 X magnification), high-grade neuroendocrine carcinoma (C, D, 200 X magnification) and a pure rhabdomyosarcoma (E, F, 200 X magnification) is shown.

3.5. Gene Rearrangements

RNASeq analysis for the 4 cases of RMS/S-RMS did not reveal any structural rearrangements for genes commonly rearranged in alveolar RMS, such as Pax3, Pax7 and FOXO1. Interestingly, we identified fusions involving genes at the 12q21.1 to 12q21.31 locus for 1 case each of pure RMS and S-RMS (Figure 7A). The specific events included a structural rearrangement between exon 2 of protein phosphatase 1 regulatory subunit 12A (PPP1R12A) and exon 2 of lin-7 homolog A (LIN7A) in the first case (Figure 7B, C). In the second case, a fusion between exon 10 of PPP1R12A and exon 4 of protein tyrosine phosphatase, receptor type Q (PTPRQ) was documented (Figure 7D, E).

Figure7: Confirmation of PPP1R12A Gene Fusion in Pure Rhabdomyosarcomas of the Bladder.

A schematic representation of the PPP1R12A-LIN7A and PPP1R12A-PTPRQ structural variants has been depicted in A. Pyrosequencing traces demonstrate the breakpoint between exon 2 of PPP1R12A and exon 2 of LIN7A on forward (B) and reverse (C) traces. Similarly, the breakpoint between exon 10 of PPP1R12A and exon 4 of PTPRQ has been shown on both forward (D) and reverse (E) traces.

4.0. Discussion

Adult genitourinary RMS and S-RMS are rare. A total of 12 cases of pure RMS and 5 cases of S-RMS were identified in our study over a 14-year period, with the majority (11 cases), involving the urinary bladder. Similar to prior studies that documented a striking morphologic overlap with HGNEC, we identified six such cases, all involving the urinary bladder. The RMS seen in the kidney and prostate were classified as embryonal or spindled subtypes and bore no morphologic resemblance to small cell carcinoma. Prior studies of pure RMS of the bladder have documented a neuroendocrine phenotype characterized by the expression of a combination of synaptophysin, chromogranin and neuron specific enolase in 1 case reported by Eusebi et al, in 3 (of 4) cases reported by Paner et al, and in 1 case reported by Bing et al (5, 16, 17). Similar expression of neuroendocrine markers in RMS occurring as heterologous elements in S-RMS of the bladder was reported in 2 cases by Bing et al (16).

Herein, we report a series of six cases of pure RMS and five S-RMS of the bladder. All eleven cases showed expression of neuroendocrine markers, with six of these cases showing significant morphologic overlap with small cell carcinoma, similar to the reports of Eusebi et al, Paner et al and Bing et al (5, 16, 17). Also, consistent with prior studies primarily focused on head and neck alveolar RMS, CD56 was almost ubiquitously expressed in all but one of our cases and half the cases expressed one of the more specific neuroendocrine markers (synaptophysin in 9; chromogranin in 1), compared to our reported rate of 30–40% for synaptophysin/chromogranin in head and neck alveolar RMS (15). Our prior study of 79 HGNEC compared to 122 stage matched UC of the bladder had shown a high diagnostic specificity of chromogranin (100%), synaptophysin (98%), TTF1 (97%) and CD56 (96%) in differentiating HGNEC from UC (18). In this context, this is a noteworthy diagnostic pitfall as six cases of RMS/S-RMS had a significant morphologic overlap with HGNEC. This is highlighted by the identification of a case of S-RMS in our series, originally diagnosed as small cell carcinoma of the urinary bladder (18).

Furthermore, the neuroendocrine immunophenotype was concordant with the underlying gene expression signature, as SYP, CHGA, CHGB and NCAM1 expression was roughly equivalent in both HGNEC and RMS/S-RMS. A similar pattern was seen for PCSK1 and SCG2, as well. ASCL1 is a basic helix-loop-helix transcriptional factor that plays a major role in differentiation during neurogenesis and previous studies have showed that its expression defined a subset of lung adenocarcinomas with neuroendocrine differentiation (25, 26). Although ASCL1 mRNA expression did not reliably discriminate between HGNEC and UC, our results indicate that immunohistochemical detection of the corresponding protein has a high specificity for HGNEC as it was not detected in any of 110 cases of UC, 9 genitourinary RMS or 47 non-bladder RMS that were tested. Although, ASCL-1 is highly specific for neuroendocrine differentiation in the carcinomas, its clinical utility is limited by the lower sensitivity of this marker.

A pure alveolar RMS and a S-RMS profiled by RNA-Seq that resembled HGNEC had novel fusions involving PPP1R12A at the 12q21.1 to 12q21.31 locus. PPP1R12A, also known as the myosin-binding subunit of myosin phosphatase plays an important role in the regulation of actin-myosin contractile dynamics (27). In a physiologic context, this active protein complex dephosphosphorylates the myosin light chain kinase. This inhibits the interaction of actin and myosin subunits and consequently smooth muscle contraction. Both identified structural rearrangements disrupt the full length PPP1R12A gene, potentially leading to the interaction of actin and myosin subunits. Prior studies looking at active phosphatases that were expressed at high baseline levels in primary RMS had identified PPP1R12A as a candidate target gene, however, to the best of our knowledge gene fusions involving PPP1R12A have not been previously described in RMS in the current English language literature (28, 29). Specifically, global fusion transcriptome profiling for recurrent chromosomal rearrangements in RMS in at least two studies have not revealed any structural variants involving the PPP1R12A gene and similar datasets are not available for bladder RMS in the reviewed English language literature (21, 30). Additional studies examining larger numbers of alveolar RMS of the urinary bladder are required to determine the significance of this fusion, and its potential role in diagnosis.

Furthermore, the functional role of this fusion product needs to be determined as does its relationship to neuroendocrine differentiation.

Our current understanding of transcriptional regulation of mammalian myogenesis suggests that PAX7 promotes initial lineage specification, followed by lineage commitment by MYOD1, and terminal differentiation by myogenin (31). All the RMS/S-RMS profiled by RNASeq showed a gene expression profile that had high levels of the myogenic transcription factors myogenin and MyoD1 and this was compatible with the observed immunostaining profile for these cases. Epithelial to mesenchymal transition occurs during embryologic development and similar pathways play an important role in oncogenesis. This involves the upregulation of key transcriptional factors such as SNAI2, ZEB1/ZEB2, and TWIST1 (14, 32). Transcription factors such as ZEB1 and ZEB2 were upregulated in both HGNEC and RMS/S-RMS, relative to UC, with significantly higher levels of ZEB2 expression being noted in S-RMS. This overall pattern of gene expression highlights multiple pathways related to myogenesis as well as epithelial to mesenchymal transition, which influence the genetic makeup of RMS/S-RMS and contribute to its unique phenotype.

Prior work by Eusebi et al had suggested that bladder RMS in adults is similar to HGNEC (5). Our study supports the findings of Paner et al in interpreting these bladder RMS as distinct from HGNEC based on a lack of immunophenotypic evidence of epithelial differentiation and gene expression studies that show expression of a larger number of genes involved in myogenic differentiation (17).

In summary, adult RMS/S-RMS of the urinary bladder showed morphologic and immunophenotypic overlap with HGNEC and neuroendocrine differentiation in the former is supported by gene expression analyses. These tumors, however, have an altered genetic makeup that can be attributed to increased expression of myogenic transcription factors as well as a unique gene fusion event involving PPP1R12A in a subset of cases. Practically, in cases of suspected small cell carcinoma of the urinary bladder that express one or more neuroendocrine markers but lack unequivocal keratin staining, application of desmin, myoD1 and myogenin is necessary to exclude RMS. The role of PPP1R12A fusion events in the diagnosis of RMS of the urinary bladder is yet to be defined.

Highlights.

• Some rhabdomyosarcomas (RMS) resemble high-grade neuroendocrine carcinomas (HGNEC)

• 17 adult genitourinary RMS were evaluated for neuroendocrine features

• Half of all bladder RMS exhibited neuroendocrine features (confirmed by RNAseq)

• Bladder RMS have a distinct molecular profile including PPP1R12A fusions