An Integrative Morphologic and Molecular Approach for Diagnosis and Subclassification of Rhabdomyosarcoma

Although traditionally considered a tumor of skeletal muscle, many RMS originate from sites devoid of musculature. Histologically, tumor cells resemble primitive mesenchyme containing developing or disordered muscle cells and showing a spectrum of differentiation from undifferentiated “blastema” cells to rhabdomyoblasts to terminally differentiated skeletal muscle cells with cross striations. The most primitive rhabdomyoblasts are stellate cells with sparse amphophilic cytoplasm. As rhabdomyoblasts differentiate, they acquire more eosinophilic cytoplasm and become more elongated, forming myotubes and eventually cross-striations. The degree of histopathologic evidence of differentiation varies from case to case. Identification of rhabdomyoblasts may be the key to making a confident diagnosis. Standard immunomarkers include proteins indicative of muscle differentiation, such as desmin and muscle-specific actin, and transcription factors leading to rhabdomyogenic differentiation, such as myogenin and MyoD1. Tumor cells are often diffusely positive for desmin, while myogenin and MyoD1 staining are heterogeneous in different subtypes.¹¹ 

The most well-defined categories of rhabdomyosarcomas are those associated with recurrent molecular alterations. The majority of RMS with ARMS histology contain FOXO1 fusions, and SSRMS contains either MYOD1 mutations or NCOA2- and VGLL2-fusions. ERMS shows diverse genetic alterations, including loss of heterozygosity of chromosome 11p15.5 and RASopathy-related mutations that are often accompanied by complex structural and numeric chromosomal changes.³  Molecular genetic information and phenotype-genotype correlation are essential for confirming the diagnosis, and along with tumor site, patient age, histologic subtype, completeness of excision, and evidence of tumor spread are used to stratify RMS patients into low-risk, intermediate-risk, and high-risk groups, provide therapeutic strategies, and guide potential therapeutic target development.

The most important test in ARMS diagnosis is determination of PAX-FOXO1 gene fusion status. Although historically the names and definitions of ARMS and ERMS refer to their histologic features, the presence or absence of PAX-FOXO1 fusion positivity has become a key question that determines risk stratification, therapeutic protocol assignment, and prognosis.^12–17  Fusion-positive RMS receives more aggressive therapy than fusion-negative RMS, and the prognosis of patients with fusion-positive ARMS is worse than that of those with either so-called fusion-negative ARMS or ERMS.^13–17  Biological studies indicate that there are no molecular differences between fusion-negative ARMS and ERMS. For these reasons, modern practice has equated the diagnosis of ARMS with fusion-positive RMS, and fusion-negative ARMS, accounting for approximately 15% of histologic ARMS, is considered a primitive form of ERMS, as recommended by the World Health Organization.^3,4 

ARMS has distinct clinicopathologic features. It is the second most common subtype of RMS, constituting approximately 25% of RMS.^18,19  It mainly occurs in older children and adolescents, although a subset has been reported in adults older than 40 years.^3,20  ARMS is the most histologically undifferentiated subtype of RMS, composed of a monomorphic population of primitive round cells often showing no morphologic evidence of skeletal muscle differentiation. Clinically, ARMS behaves aggressively, with 25% to 30% of patients presenting with distant or local metastasis via lymphatic or hematogenous spread.^3,19 

There are 2 histologic growth patterns in this subtype as follows: classical alveolar and solid variant. Classical ARMS, as the name suggests, shows central loss of cellular cohesion and formation of irregular alveolar spaces^3,4,9  (Figure 1, A). Historically, recognition of ARMS histology, either in pure or “mixed” form, was sufficient for the diagnosis of ARMS; in mixed tumors, at least 50% ARMS histology was required according to the Children Oncology Group and Intergroup Rhabdomyosarcoma Study Group studies.⁹  On the other hand, the solid variant shows highly cellular and uniform round cells but lacks fibrovascular septa and cellular discohesion. It has become apparent that many solid variant cases are fusion-negative ERMS, although a significant subset of these tumors contains genetic features akin to classical ARMS.^21,22  Sporadic wreath-like tumor giant cells are peculiar features of ARMS, although their presence is not unique.

ARMS fusion genes include common and variant types, and variant fusion genes may not be akin to the common fusion genes. Common fusion genes include either a PAX3-FOXO1 or a PAX7-FOXO1, which occur in the majority (>95%) of fusion-positive cases^16,23  (Figure 1, B). These fusions link paired box (PAX) genes with a forkhead domain gene (FOXO1). Very rare fusion variants have been reported, including PAX3-NCOA1, PAX3-FOXO4, PAX3-INO80D, and FOXO1-FGFR1.^24–26  However, some of these fusions form different molecular species from the PAX-FOXO1 types, so that the nature of these rare tumors is uncertain.²⁵  For example, NCOA1 encodes a protein that acts as a transcriptional coactivator for corticosteroid and nuclear protein hormones.²⁶ 

ERMS, the most common subtype of RMS, occurs in relatively young children, with one third in children younger than 5 years. ERMS accounts for approximately 50% to 70% of RMS in ages younger than 20 years and also comprises approximately 20% of adult RMS.²  ERMS contains a mixture of rhabdomyoblastic cells in different stages of myogenesis. The tumor has a hybrid morphology combining clusters of round, spindled, and stellate cells with scattered rhabdomyoblasts in a loose/myxoid stroma/background (Figure 1, C and D). This alternating loose and dense morphology is one of the most characteristic features of ERMS, although some are predominately myxoid and others predominately dense. The dense form of ERMS likely accounts for most so-called “fusion-negative” ARMS.

PAX-FOXO1 fusion negativity is characteristic for this subtype of RMS, although recurrent genomic alterations have been observed,²⁷  and fusion positivity is rarely seen in cases with ERMS histology or mixed histology.²⁵  Genomic alterations in ERMS include recurring patterns of chromosome gains (8, 2, 11, 12, 13, and/or 20) and losses (10 and 15). There is profound loss of heterozygosity on chromosome 11p.²⁸  Loss of function mutations of TP53 occurs in approximately 10% of ERMS and may be associated with anaplasia.⁹  A recent international study on the molecular/genetic role in rhabdomyosarcoma based on an extensive analysis of more than 600 patients with rhabdomyosarcoma revealed that TP53 mutations were associated with worse outcomes in both fusion-negative and FOXO1 fusion-positive cases.²⁹ RAS (NRAS, KRAS, and HRAS) mutations occur in approximately 50% of ERMS; recurring mutations of effectors of P13K (PTEN, PIK3CA) are also seen. The mutation of a RAS isoform occurs in one third of ERMS. NRAS mutations are common in adolescent ERMS, whereas HRAS and KRAS mutations are encountered in 70% of infantile ERMS.^2,3,26 

The botryoid variant of ERMS is a morphologic variant found in mucosal-lined organs, most notably in the common bile duct, urinary bladder, and the vagina in females, and typically presenting as polypoid grape-like masses grossly. The etymology denotes the gross appearance of “grape clusters” (from botryoid in Greek). On histologic examination, the conspicuous cambium layer, a densely linear cellular area tightly abutting an epithelial surface, serves as a diagnostic clue.³⁰  However, the cambium layer may be narrow or absent in some cases, a notorious diagnostic pitfall. Children with botryoid RMS have historically enjoyed a better prognosis than those with other ERMS.⁹  Although botryoid RMS was a distinct subtype in the Horn-Enterline classification, it is now considered a subtype of ERMS, and its better clinical outcome may be related to tumor site rather than histology.

Anaplasia in ERMS is defined by the presence of markedly enlarged hyperchromatic nuclei (at least 3 times the size of neighboring nuclei) with or without large atypical bizarre, multipolar mitotic figures³⁰  and is present in 3% to 13% of ERMS and correlates with a poorer outcome in single variant analysis but not in multivariate studies.³¹  Anaplasia in childhood RMS should not be confused with pleomorphic subtype of RMS in adults, which is a totally distinct subtype. Genomic amplification in the form of double minutes is frequent in ERMS with anaplasia.^31,32  Anaplasia in ERMS may be focal or diffuse (Figure 1, E). Patient age can help in distinguishing between these 2 diagnoses. The anaplastic phenomenon in ERMS may indicate inherited mutation of TP53.³⁰ 

Some ERMS occur in the setting of inherited mutation syndromes, including Li-Fraumeni syndrome, Beckwith-Wiedemann syndrome, neurofibromatosis type 1, Noonan syndrome, and Costello syndrome. Noonan syndrome is a RASopathy that involves mutations of PTPN11, SOS1, RAF1, and KRAS, whereas Costello syndrome is a RASopathy caused by germline mutations of HRAS predisposing to RMS.³³  Neurofibromatosis 1 can also be considered a RASopathy, because functioning NF1 controls the RAS pathway. DICER1 syndrome is caused by germline inactivating mutation of DICER1 and is associated with ERMS of the gynecologic tract and genitourinary system. ERMS occurring in the inherited genetic setting is morphologically indistinguishable from sporadic ERMS and most commonly present as conventional or botryoid variants.^2,34  One exception to this rule is frequent occurrence of cartilage in DICER1-associated ERMS, whereas it rarely if ever occurs in other settings.

As an integral part of RMS diagnosis, immunohistochemical expression of muscular markers (eg, desmin, myogenin, and MyoD1) plays an important role, not only in making a diagnosis but also helping to predict fusion status. As a rule, myogenin is usually strongly and diffusely expressed in ARMS,^35–37  whereas its expression tends to be less diffuse or focal (staining from 10% to 90% of tumor cells) in ERMS.¹¹  On the other hand, MyoD1 mutant SSRMS typically appears to be more diffusely MyoD1 positive than myogenin, which may even be negative.

Though the terminology of “alveolar” or “embryonal” RMS based on histologic features persist as part of our pathology lexicon, the management and prognosis of ARMS and ERMS are currently based on PAX-FOXO1 fusion gene status. Though strong opinions and good scientific evidence indicate that the diagnosis of ARMS should be based on genetic tests and not histologic features,¹¹  genetic tests are not universally available, so some pathologists must continue to rely on histologic distinction for treatment stratification. However, some immunomarkers appear to correlate with fusion gene status and may predict fusion positivity from a relatively convenient and economic perspective. Diffuse myogenin (>90%) staining favors a diagnosis of ARMS,^7,38  but there is some overlap in results with ERMS. Gene expression studies have demonstrated that some proteins might act as surrogates to PAX3/7-FOXO1 fusion testing; these “surrogate markers” include AP-2beta, NOS1, and P-cadherin overexpression in FOXO1 fusion-positive RMS³⁹  and HMGA2 and EGFR overexpression in FOXO1 fusion-negative RMS.^39,40  Combination of these markers increases specificity up to more than 90% and provide a sensitivity of more than 60%.⁹  When combined with molecular testing, such as reverse transcription polymerase chain reaction or fluorescence in situ hybridization, the abovementioned surrogate markers can be used to confirm PAX3-FOXO1 or PAX7-FOXO1 positivity or suggest a rare genetic fusion variety. However, the rationale for their usage remains to be proven in laboratories with access to molecular testing.

The spindle cell and sclerosing subtypes of RMS were once viewed as variants of other RMS. In 2013, the World Health Organization Classification of Tumors of Soft Tissue and Bone proposed it to be a unified separate entity, SSRMS.³  Molecular data emerging since that time clearly supports its separate subtype status, although it continues to show heterogeneous morphology and disparate prognosis. Of note, the recent wealth of information even suffices to suggest the existence of at least three subtypes in SSRMS. One subtype is defined as congenital/infantile SSRMS with VGLL2/NCOA2/CITED2 rearrangements.^41,42  Another subtype is known as SSRMS with MYOD1-mutant p.Leu122Arg.^43–46  A third rare subtype is the predominantly intraosseous RMS with EWSR1-TFCP2, FUS-TFCP2, or MEISI-NCOA2 rearrangements.^47,48  SSRMS accounts for 3% to 10% of RMS and affects infants, children, and adults.^46,49–51 

SSRMS has a fascicular spindle cell and/or sclerosing morphology. Spindle cell tumors contain cells that show elongated borders and form a fascicular or herringbone growth pattern^51,52  (Figure 1, F). Sclerosing RMS shows prominent hyalinization and sclerosis with tumor cells arranged in cords, nests, or trabeculae (Figure 1, G), sometimes with a “microalveolar” appearance. Immunohistochemically, SSRMS is characterized by diffuse expression of desmin and MyoD1 (Figure 1, H) but only focal expression of myogenin.³ 

The infantile form of SSRMS occurs almost exclusively in children younger than 5 years, often younger than 3 years, and only involves soft tissue. Morphologically, most tumors have mixed spindled cell and sclerotic features, and usually form 2 different patterns, sclerosing or cellular. The congenital/infantile subtype of SSRMS is a low-grade tumor with local recurrence and favorable outcome.⁴²  However, a recent study showed that in some patients, this tumor may have a propensity for high-grade transformation that results in local progression and distant metastasis.⁵³ 

MyoD1 mutant RMS was first described in 2013 and is associated with somatic activating mutations of the MYOD1 gene. The tumor occurs at any age (2 to 94 years) but is more frequent in adults and females. The most common sites include the head and neck (one third), extremities, and trunk. Most tumors show a hybrid growth pattern, but occasionally one sees a purely sclerosing pattern or spindle cell pattern. Rhabdomyoblastic differentiation is focally seen in some cases. Desmin and MyoD1 show diffuse positivity with patchy myogenin positivity. MyoD1 mutant SSRMS is a high-grade tumor with poor clinical outcome.^5,38,43,44 

Intraosseous RMS was first described in 2018 and occurs in bone, especially craniofacial bones.^3,49,50  Ages of patients have ranged between 11 and 86 years. Three associated gene fusions, EWSR1-TFCP2, FUS-TFCP2, and less commonly MEIS1-NCOA2 rearrangements have been reported, although not all RMS with TFCP2 gene fusions are intraosseous.^9,49,50  The tumor contains complex karyotypes and shows constant homozygous deletions of CDKN2A. There is small or internal deletion of ALK1 correlated with ALK overexpression in two‐thirds of cases. Morphologically, intraosseous SSRMS shows a mixed spindled and epithelioid cell pattern in most cases, but one third of cases show a purely epithelioid or rarely fascicular pattern^48,49 ; obvious rhabdomyoblastic differentiation is mostly absent. However, tumor cells usually express desmin, MyoD1, and myogenin, although the expression of the latter may be focal. Most cases also express epithelial markers, such as pan-cytokeratin (clone AE1/AE3) and/or epithelial membrane antigen. Similar to other subtypes of RMS, especially ARMS, RMS with TFCP2 fusion has the potential to spread both lymphatically to regional lymph nodes and hematogenously to distal sites (lung or other bones).⁵⁰  This subtype of RMS is a high-grade tumor with a median survival of 8 months.^3,4 

In children, tumors with SSRMS histology originate from the paratesticular region more often than from other regions.^{1,47,49,50,54,55}  It is noteworthy to mention that characteristic mutations have not been reported in this largest subgroup of SSRMS. Many of these cases show mixed embryonal and spindle cell histology, so they may be a subset of ERMS. Whether they are yet another molecular species of SSRMS or variants of ERMS remains to be proven.

The histology of SSRMS is heterogeneous, and diagnosis may require the presence of predominant (at least 80%) spindle cells in tumors to distinguish it from some ERMS that have both round and spindle tumor cells. This particularly occurs in paratesticular tumors. Because morphology of SSRMS mimics leiomyosarcoma and fibrosarcoma, these differential diagnoses should be kept in mind. Myogenin and MyoD1 staining help to distinguish them from leiomyosarcoma, and desmin, myogenin, and MyoD1 separate fibrosarcoma from RMS. Molecular confirmation of MYOD1 mutations or VGLL2/NCOA2/CITED2 gene fusions may be required in selected cases.

Pleomorphic RMS is a high-grade pleomorphic tumor that features sheets of large, atypical, or bizarre, polygonal tumor cells, mimicking undifferentiated pleomorphic sarcoma. It occurs nearly exclusively in adult patients aged 21 to 81 years, with a median of approximately 72 years,⁵  but it also occurs very rarely in children with an underlying genetic predisposition (such as Li-Fraumeni syndrome).⁴  Tumors mostly arise in the deep soft tissue and are seen at any site, but most often occur in the lower extremities.⁵⁶  Pathogenesis and molecular driver alterations are unclear. The pleomorphic subtype of RMS contains complex genomic alterations with numerous copy number changes (similar to osteosarcomas or undifferentiated pleomorphic sarcomas).⁵⁶  Morphologically, pleomorphism is diffuse and rhabdomyoblastic differentiation is scant. Immunostaining shows that tumor cells are strongly positive for desmin with limited MyoD1 or myogenin expression. Clinical prognosis is very poor, and 80% of patients die, with mean survival of approximately 7.3 months.⁵⁶ 

It is worthwhile to mention that rhabdomyoblastic differentiation can be seen in many tumors, including epithelial and mesenchymal malignancies, such as biphenotypic sinonasal sarcoma, carcinosarcoma, Wilms tumor, pleuropulmonary blastoma, Sertoli-Leydig cell tumor, malignant peripheral nerve sheath tumor, dedifferentiated chondrosarcoma, dedifferentiated liposarcoma, and ectomesenchymoma. Rhabdomyoblastic differentiation in these lesions sometimes causes confusion with RMS, particularly in biopsy specimens.

Recent molecular genetic data have provided substantial information related to the etiology and pathogenesis of RMS, and its incorporation into standard testing has set up new diagnostic criteria for RMS (Table). Besides its important role in diagnosis, genetic testing has become essential for the proper management and prognosis of RMS patients. Foremost among these tests is presence of PAX-FOXO1 fusion genes, a predictor of poor outcome that has become widely employed in pathological classification of RMS and incorporated into standard risk stratification and therapy assignment. Surrogate immunohistochemical markers may be useful tools in determining fusion gene status when molecular testing is unavailable, but a rationale for their standard usage is lacking. Fusion-negative ARMS is now associated with ERMS in terms of molecular biology, clinical presentation, outcome, and prognosis. Standard morphologic evaluation to make a precise diagnosis of RMS now includes routine and ancillary studies, including molecular testing and immunohistochemistry. In many cases, routine hematoxylin-eosin evaluation and desmin, MyoD, and myogenin expression pattern are sufficient to both confirm a diagnosis of RMS and assign a histologic subtype, although all ARMS (particularly solid variants) should have PAX fusion testing to corroborate the diagnosis. The new subclassification of SSRMS includes clinically and biologically heterogeneous lesions and appears to be more clinically useful when combined with genetic tests (Figure 2). Knowledge of molecular alterations will lead to a diagnosis that better approximates the pathogenetic state and malignant potential of RMS and molecular testing is indicated as a routine practice in these tumors.