BACKGROUND

Section:

ChooseTop of pageAbstractBACKGROUND <<HISTOLOGYTUMOR SITEHISTOLOGY VERSUS FUSION S...PROGNOSTIC SIGNIFICANCECONCLUSIONS AND FUTURE DI...References

The International Classification of Rhabdomyosarcoma (ICR) published in 1995 showed histology to be an important predictor of prognosis, forming the basis for risk stratification strategies adopted in subsequent Children's Oncology Group (COG) studies.¹ In the ICR, ARMS were associated with a poor prognosis, conventional ERMS with an intermediate prognosis, and botryoid and spindle cell variants of ERMS had a superior outcome. Notably, however, the ICR changed the histologic definition of ARMS from that used in prior Intergroup Rhabdomyosarcoma Studies (IRS). First, the ICR included the solid variant of ARMS that had been previously been under-recognized. Second, the criteria for a diagnosis of ARMS decreased from requiring greater than 50% alveolar histology to any amount of alveolar histology.

Following adoption of the ICR criteria, the COG Soft Tissue Sarcoma (STS) Committee saw an increase in both the number of cases classified as ARMS as well as the frequency of fusion-negative ARMS (ARMSn) in subsequent IRSG and COG studies. In the intermediate risk RMS study D9803, which enrolled patients with both ARMS and ERMS diagnosed from 1999 to 2005, total ARMS diagnoses increased from 30% to 41%. Additionally, the frequency of ARMSn doubled from approximately 20% to 25% of all ARMS in IRS-III and -IV to 40% to 45% in later studies. ARMSn comprise a heterogeneous group consisting of tumors with alternate fusions, fusions without production of detectable RNA, tumors with only rare fusion-positive cells, and those with no molecular evidence of fusion.^2,3 The latter category constitutes the bulk of this group, and it is likely that a noteworthy portion of these cases is accounted for by misclassification of histologic variants of ERMS.

Because of these diagnostic challenges and the increase in ARMS and ARMSn following application of ICR criteria to COG central pathology review, the COG STS committee central pathology review again required greater than 50% alveolar pattern for a diagnosis of ARMS beginning around 2007. Emphasis was also added to the presence of strong, diffuse myogenin expression for supporting a diagnosis of ARMS. To address these shifts in histologic criteria of ARMS, we recently completed a re-review of all ARMS cases enrolled on COG intermediate risk RMS study D9803 using current COG central pathology review criteria.

HISTOLOGY

Section:

ChooseTop of pageAbstractBACKGROUNDHISTOLOGY <<TUMOR SITEHISTOLOGY VERSUS FUSION S...PROGNOSTIC SIGNIFICANCECONCLUSIONS AND FUTURE DI...References

Although ARMS and ERMS are readily identified in their classic forms, even the earliest studies of RMS recognized the morphologic variability of these tumors.^4-6 First described by Riopelle and Theriault,⁵ classic ARMS is well recognized and easily distinguished from ERMS, including the typical, botryoid, and spindle cell subtypes. In some cases histologic patterns of ARMS and ERMS may overlap, however. In their study of ARMSn, Parham and colleagues showed that many ERMS were originally classified as ARMS as a result of misinterpretation of dense or sclerosing areas.⁷ We observed a similar phenomenon in our re-review of ARMS in D9803.⁸ On re-review of histology for D9803, 33% of original ARMS diagnoses were reclassified as ERMS. The most common reason for reclassification was the presence of at least focal dense or sclerosing patterns of RMS previously interpreted as small foci of ARMS. As expected, botryoid and typical embryonal morphologies were rare. In contrast, more than 75% of control ERMS cases showed typical or botryoid morphology.

The solid-variant of ARMS poses particular diagnostic difficulty. This pattern was first described by Tsokos and colleagues in 1992⁹ and shows noteworthy overlap with the dense pattern of ERMS. Both tumors consist of sheets of closely packed cells; however, in contrast to solid-variant ARMS, which shows a monomorphic round cell cytology, dense ERMS have variation in cellular and nuclear size and shape within a tumor, although this feature may be subtle.

Myogenin expression is valuable in distinguishing ARMS from its histologic mimics.^10-12 Nearly all ARMS exhibit diffuse 3–4+ myogenin expression, a feature helpful in classifying small RMS samples as 4+ myogenin expression is essentially restricted to ARMS. This differential myogenin expression pattern is useful in distinguishing sclerosing RMS from classic ARMS, which may have marked stromal sclerosis. Sclerosing RMS shows weak, patchy myogenin expression but strong and diffuse expression of MyoD1, in contrast to ARMS, which show strong, diffuse myogenin expression and variable expression of MyoD1. Despite myogenin immunohistochemistry, the most diagnostically difficult category remains the dense pattern of ERMS. This pattern typically has moderate (2–3+) myogenin expression, overlapping that of solid ARMS. For this group, cytology remains the only distinguishing morphological feature, as suggested by Palmer and colleagues.¹³

Central pathology review for the COG STS committee currently recognizes both classic and solid variants of ARMS and the typical, spindled, and botryoid variants recognized in the ICR, as well as more recently described variants including dense, sclerosing,^14,15 and epithelioid¹⁶ patterns (Table 1). A diagnosis of RMS-NOS (not otherwise specified) is reserved for cases where the biopsy sample was too small, crushed, or necrotic for a definitive classification. Rare tumors are also truly mixed histologically, and a diagnosis of mixed RMS is applied only for these rare cases.

TABLE 1. Histologic Classification of Rhabdomyosarcoma

Mixed RMS have separate, discrete ARMS and ERMS foci, although the components otherwise resemble pure tumors of the respective types. The two regions often resemble nested or collision tumors, and differential myogenin expression may help distinguish the ARMS and ERMS components. The alveolar component of mixed RMS cases may be minor or it may predominate. By current criteria, tumors with 50% or more alveolar component are considered ARMS, but fusion-positive tumors were not restricted to those with a predominant alveolar component. Although a minor ARMS component in some mixed RMS may result from sampling artifact, it remains puzzling why tumors with a minority alveolar pattern demonstrate the characteristic ARMS gene fusion.

Although morphologically and genetically heterogeneous, mixed RMS do appear genetically consistent between alveolar and embryonal components of the same tumor. Nishio and colleagues examined eight mixed tumors by FISH and showed no evidence of a PAX-FOXO1 fusion in eithercomponent.¹⁷ Other reports, however, demonstrated a PAX-FOXO1 fusion in both alveolar and embryonal components.^18,19 Hachitanda and colleagues published the only study showing distinct genetic features between the two patterns of mixed RMS with MYCN amplification in the alveolar but not the embryonal component.²⁰ The rarity of mixed RMS to date has precluded in depth studies of genetic characteristics or outcome. Based on available data, we believe fusion status should be determined for all mixed RMS, regardless of the percent alveolar component.

TUMOR SITE

Section:

ChooseTop of pageAbstractBACKGROUNDHISTOLOGYTUMOR SITE <<HISTOLOGY VERSUS FUSION S...PROGNOSTIC SIGNIFICANCECONCLUSIONS AND FUTURE DI...References

Reclassified ERMS cases in D9803 most often arose from typical ERMS sites. Genitourinary (GU) tract tumors not arising from the bladder or prostate and perineum/retroperitoneum/trunk tumors were over-represented among reclassified ERMS. GU (nonbladder/prostate) and retroperitoneal primary sites were more frequent among mixed RMS tumors, and GU (nonbladder/prostate) and perineum/trunk primary sites were more common among sclerosing RMS. This is particularly true of large paratesticular tumor resections, which are often very heterogeneous with small foci resembling ARMS.

HISTOLOGY VERSUS FUSION STATUS

Section:

ChooseTop of pageAbstractBACKGROUNDHISTOLOGYTUMOR SITEHISTOLOGY VERSUS FUSION S... <<PROGNOSTIC SIGNIFICANCECONCLUSIONS AND FUTURE DI...References

Recognizing the histologic variability of ERMS improves our classification of RMS.⁷ Although the diagnosis of ERMS increased on our histology re-review, no reclassified ERMS showed evidence of gene fusion. Correspondingly, the fraction of ARMSn decreased from 37% before reclassification to 18%, following reclassification, approximating historic rates of 22% to 23% ARMSn in previous IRSG studies.^7,11

PROGNOSTIC SIGNIFICANCE

Section:

ChooseTop of pageAbstractBACKGROUNDHISTOLOGYTUMOR SITEHISTOLOGY VERSUS FUSION S...PROGNOSTIC SIGNIFICANCE <<CONCLUSIONS AND FUTURE DI...References

We analyzed the prognostic significance of histology, restricting analysis of outcome for D9803 cases to patients with stage 2/3, Group III disease, as these patients were eligible for the D9803 study irrespective of histology subtype. The estimated 5-year EFS was 77% for original and confirmed ERMS (233 patients), 69% for reclassified ERMS (34 patients) and 55% for confirmed ARMS (79 patients; p < 0.001). The estimated 5-year overall survival (OS) was 81% for original and reclassified ERMS and 68% for confirmed ARMS (p = 0.018). Overall, the outcome for patients with ARMS is poorer than that for those with ERMS, and reclassified ERMS have an outcome more similar to ERMS than ARMS.

Fusion status influenced outcome within distinct clinical subgroups. For the same subset of patients with stage II/III, Group III RMS (351 patients), presence of either P3F or P7F portended worse EFS at 5 years (p < 0.001). OS was significantly worse only in P3F cases (p = 0.015). Focusing on the more favorable stage I or stage II and III and Group I/II subpopulation (94 patients), the 5-year EFS for those with P3F (28 patients) and P7F (12 patients) ARMS were similar (65% and 75%, respectively) with a trend toward being inferior to ARMSn (10 patients) (100%) and reclassified ERMS (44 patients) (80%) (p = 0.15). The 5-year OS for P3F was worse than P7F and ARMSn (p = 0.21).

Based on this data, future COG protocols will incorporate fusion status into their risk stratification guidelines

CONCLUSIONS AND FUTURE DIRECTIONS

Section:

ChooseTop of pageAbstractBACKGROUNDHISTOLOGYTUMOR SITEHISTOLOGY VERSUS FUSION S...PROGNOSTIC SIGNIFICANCECONCLUSIONS AND FUTURE DI... <<References

By using data from the D9803 study after the pathology re-review described here, the COG STS committee showed that FOXO1 fusion status was a strong predictor of outcome.²⁵ Although risk stratification and clinical management may rely on fusion status rather than histopathologic diagnosis in the future, histopathology remains important for triaging molecular testing since there is an extremely low rate of fusion positivity in tumors with typical ERMS histology (the majority of RMS). In general, for patients not enrolled on study, we suggest all ARMS and mixed RMS be tested for P3F and P7F gene fusions. For diagnostically challenging cases, particularly dense ERMS with 3+ myogenin or for extremity lesions, fusion status is recommended as a diagnostic adjunct.

Several questions still remain. First, our study was limited to intermediate-risk RMS patients. Survival is 100% at 5 years for children with stage I or stage II/III, grade 1/2, ARMSn disease. However, this is based on only 10 children and in the context of more intensive intermediate-risk therapy. To add to this cohort of patients we have identified a small group of histologically confirmed ARMSn that were enrolled on low-risk RMS study D9602, before the opening of D9803. We are currently examining fusion data in this group to better examine the effects of low-risk treatment in this patient population.

Additionally, the prognostic relevance of P3F compared with P7F ARMS is not as clear. Sorensen and colleagues found a marked survival difference between PAX7-FOXO1 and PAX3-FOXO1 ARMS cases with metastatic disease.²² ARMSn in this study demonstrated outcomes intermediate between PAX7-FOXO1 and PAX3-FOXO1 cases, however, and more recent studies have reached varying conclusions.^23-25 Both clinical and genomics work suggest that P3F and P7F ARMS may be distinct even despite substantial overlap in global gene expression. We are currently examining histology and fusion status in two completed high-risk RMS trials, D9802 and ARST0431, to better answer this question.

We are moving forward into an era in which molecular features are used for risk stratification and treatment assignment. RT-PCR and FISH remain the gold standards for analyzing fusion gene status. In the future, however, a multianalyte signature with multifeature metagenes^25,26 or the incorporation of noncoding RNAs and/or structural DNA changes will differ from a single molecular analyte-like fusion gene status. Given the aforementioned difficulties with retrospective, convenience cohorts, this issue must be formally addressed in the context of a prospective trial; this work is also ongoing.

The author(s) indicated no potential conflicts of interest.