Rhabdomyosarcoma: Value of myogenin expression analysis and molecular testing in diagnosing the alveolar subtype

An analysis of 109 paraffin-embedded specimens




Identification of the alveolar subtype of rhabdomyosarcoma (ARMS) is important, because the poor prognosis associated with this subtype necessitates a modified therapeutic regimen. At present, ARMS diagnoses are made on the basis of histologic findings and the extent of myogenin immunopositivity. Nonetheless, the absence of an alveolar pattern in the solid variant, the low degree of differentiation in certain embryonal rhabdomyosarcomas (ERMS), and the increasing use of microbiopsy samples make the diagnosis of ARMS somewhat difficult. Two specific translocations have been found in ARMS, and fusion transcripts can be detected by reverse transcriptase–polymerase chain reaction (RT-PCR) analysis of paraffin-embedded tissue (PET).


To assess the value of myogenin staining and molecular testing in the diagnosis of rhabdomyosarcoma, the authors examined 109 rhabdomyosarcoma samples (45 ARMS samples and 64 ERMS samples). Real-time RT-PCR analysis of PET was performed in all 109 rhabdomyosarcomas, and RT-PCR analysis of frozen material was performed in 24 cases.


PAX fusion transcripts were present in 44 cases (39 ARMS and 5 ERMS) and absent in 52 cases (2 ARMS and 50 ERMS). In 13 cases (4 ARMS and 9 ERMS), the results were not interpretable. Results were concordant between paired frozen and fixed tumor samples. All 35 interpretable ERMS samples that contained < 50% myogenin-positive cells failed to yield detectable PAX fusion transcripts. Of the 61 interpretable tumor samples (41 ARMS and 20 ERMS) that contained > 50% myogenin-positive cells, 44 (39 ARMS and 5 ERMS) yielded detectable PAX fusion transcripts.


The current study demonstrates that molecular detection of PAX fusion transcripts via real-time RT-PCR analysis of PET is a sensitive and specific method for the diagnosis of ARMS and that immunohistochemical analysis of myogenin expression can be used to select cases for such molecular testing. Although RT-PCR analysis appears not to possess diagnostic value in tumors with < 50% tumor cell immunopositivity, it is strongly recommended for the diagnosis of tumors containing > 50% myogenin-positive cells. Cancer 2004. © 2004 American Cancer Society.

Rhabdomyosarcomas are tumors that exhibit skeletal muscle differentiation. These tumors are roughly divided into three major subtypes: embryonal, alveolar, and pleomorphic. The embryonal and alveolar subtypes represent the most common soft tissue sarcomas observed in children, but these tumor subtypes can also be found in adults.1 In recent years, the botryoid and spindle cell subtypes of rhabdomyosarcoma have been added to the embryonal rhabdomyosarcoma (ERMS) category.2, 3 Histologic subtype is one of the most useful prognostic parameters for patients with rhabdomyosarcoma, with the botryoid and spindle cell subtypes being associated with a low level of risk, the alveolar subtype being associated with a high level of risk, and the embryonal subtype being associated with an intermediate level of risk.2, 4

Alveolar rhabdomyosarcoma (ARMS), which is associated with a relatively high frequency of disseminated metastases and necessitates a modified therapeutic regimen, is the rhabdomyosarcoma subtype that carries the poorest prognosis; therefore, the distinction between ARMS and ERMS is an important one. Nonetheless, the drawing of this distinction in the clinical setting can be difficult, because the classic alveolar pattern may not be present in the solid variant of ARMS, and because ERMS can be cellular and poorly differentiated in some cases. Furthermore, the use of microbiopsy to diagnose soft tissue sarcomas is becoming increasingly common, and this shift also makes the subtyping of rhabdomyosarcomas more challenging.

Immunohistochemical analysis is highly useful with respect to diagnosing rhabdomyosarcomas5 and distinguishing ERMS from ARMS. Extensive nuclear staining for myogenin is reported in most ARMS (tumor cell positivity > 75% in most cases), whereas immunopositivity for myogenin is less uniform (tumor cell positivity < 25% in most cases) and typically focal in ERMS.6–8

Two specific translocations are documented in most ARMS: t(2;13) to fuse PAX3 and FKHR and t(1;13) to fuse PAX7 and FKHR. Gene fusion transcripts can be detected by molecular methods in both frozen and fixed paraffin-embedded tissue samples,9–11 and the use of reverse transcriptase–polymerase chain reaction (RT-PCR) analysis in the diagnosis of ARMS has become increasingly common.

In the current study, we examined a series of 109 paraffin-embedded rhabdomyosarcoma samples with the goal of assessing the value of myogenin staining and real-time RT-PCR in the identification of ARMS.


Case Selection

One hundred nine cases of ARMS and ERMS (including the botryoid and spindle cell subtypes) were included in the study. All cases were positive for desmin and myogenin expression. Cases were collected from the pathology departments at the Institut Bergonié (Bordeaux, France), the Hôpital Pellegrin (Bordeaux, France), the Institut Gustave Roussy (Villejuif, France), and the Centre Léon Bérard (Lyon, France), as well as from the University of Lausanne Institute of Pathology (Lausanne, Switzerland).

The study cohort included 74 males and 35 females, who ranged in age from 0 to 76 years (0–10 years, n = 53; 11–20 years, n = 36; 21–40 years, n = 12; and > 40 years, n = 8). Tumor sites included the head and neck region (n = 39), internal trunk (n = 23), limbs (n = 22), urogenital tract (n = 15), and trunk wall (n = 10). Tumors were histologically classified as alveolar, embryonal, botryoid, or spindle cell rhabdomyosarcomas in accordance with the guidelines of Tsokos et al.4 and Newton et al.12 Forty-six tumors were fixed in Holland Bouin fluid, 2 were fixed in Bouin fluid, 41 were fixed in formalin, and 20 were fixed in a mixture of acetic acid, formaldehyde, and alcohol (AFA). Frozen tissue samples were available (in addition to fixed tissue samples) for 24 tumors.


Aside from sections stained with hematoxylin and eosin, paraffin-embedded sections were deparaffinized and incubated in the presence of antibodies against desmin (1:100 dilution, clone D33; Dako, Glostrup, Denmark) and myogenin (1:40 dilution, clone LO26; Novocastra, Newcastle, United Kingdom). Immunostaining was performed in accordance with the streptavidin-biotin-peroxidase method described by Hsu et al.13 Tissue sections were subjected to microwave heating (20 minutes in 0.1 M citrate buffer, pH 6) before staining. Sections were immunostained using the LSAB kit (Dako) in conjunction with an automated immunostainer (Techmate Horizon; Dako). All steps were performed at room temperature, and signals were visualized using a diaminobenzidine substrate (Dako). Appropriate positive and negative control experiments were performed throughout. Rates of nuclear positivity for myogenin in tumor cells were estimated by viewing stained slides at medium power.

Molecular Analysis

For molecular analysis, 1 paraffin block from each tumor was selected, and 15–30 sections (depending on the tumor cell density) measuring 10 μm in thickness were cut from that block. Special attention was paid to this technical step so as to avoid contamination.

RNA Extraction from PET

Tissue sections were deparaffinized twice in toluene and then washed twice with absolute ethanol. Sections from tissue samples fixed with Holland Bouin or Bouin fluid were also washed with lithium carbonate to remove any picric acid present. Next, the sections were washed with 1X TNE (10 mM Tris, pH 8; 1 mM ethylenediamine tetraacetic acid, pH 8; and 100 mM NaCl, pH 8). Finally, sections were resuspended in 250 μL Qiagen ATL buffer (Qiagen, Chatsworth, CA) containing proteinase K (final concentration, 4 mg/mL) and incubated for 3 days at 55 °C.

RNA was extracted using 750 μL Trizol-LS reagent (Invitrogen, La Jolla, CA) for every 250 μL of cellular lysate. The solution was placed under moderate shaking for 30 minutes to 1 hour, after which RNA extraction was performed in accordance with the instructions provided by the manufacturer of the extraction kit that was used (Qiagen). To remove trace DNA contamination, the RNA pellet, after being resuspended in 10 μL RNase-free water, was incubated with DNAse 1X Mix (Promega, Madison, WI), 1 mM dithiothreitol (Roche, Basel, Switzerland), 40 units (U) RNasin (Promega), and 2 U DNase (Promega) for 1 hour at 42 °C. RNA subsequently was purified once more using Trizol-LS.


Reverse transcription of 5 μg RNA extracted from fixed tissue was performed in a total volume of 20 μL, which contained 50 mM Tris-HCl, pH 8.3; 40 mM KCl; 5 mM MgCl2; 0.5% Tween (Roche); deoxynucleotides at a concentration of 0.5 mM; 10 mM dithiothreitol (Roche); specific reverse primers (75 ng FKHR reverse primer or 20 ng β-2-microglobulin reverse primer; Table 1); 12 U RNAse inhibitor (Promega); and 10 U Expand reverse transcriptase (Roche). Samples were incubated at 42 °C for 1 hour and then at 95 °C for 5 minutes.

Table I. Probes and Primer Sequences
PAX3 forward primer5′-TTG GCA ATG GCC TCT CAC C-3′
PAX7 forward primer5′-CAA CCA CAT GAA CCC GGT C-3′
FKHR reverse primer5′-ATC CAC CAA GAA CTT TTT CCA G-3′
β-2-microglobulin forward primer5′-TGA CTT TGT CAC AGC CCA AGA TA-3′
β-2-microglobulin reverse primer5′-AAT CCA AAT GCG GCA TCT TC-3′
β-2-microglobulin probe5′-6-FAM-TGA TGC TGC TTA CAT GTC TCG ATC CCA TAMRA-3′

PCR amplification was performed in duplicate in a 96-well plate (Applied Biosystems, Foster City, CA) using 25 μL of a reaction mixture containing the following: each primer at a concentration of 300 nM; 100 nM probe PAX or 50 nM probe β-2-microglobulin; equal proportions of deoxyadenosine triphosphate, deoxycytidine triphosphate, deoxyguanosine triphosphate, deoxythymidine triphosphate, and deoxyuridine triphosphate at a total concentration of 200 μM; MgCl2 at a concentration of 6 mM (for PAX-FKHR amplification) or 5 mM (for β-2-microglobulin amplification); 1.25 U Taq polymerase; and 0.25 U AmpErase uracil-N-glycosylase (UNG; Applied Biosystems) in 1X real-time PCR buffer containing a passive reference (Rox fluorochrome). Thermal cycling began with 2 minutes at 50 °C for AmpErase activation and 10 minutes at 95 °C for Taq polymerase activation, followed by 50 PCR cycles of 30 seconds at 95 °C and 1 minute at 60 °C. All reactions were performed using an ABI Prism 5700 sequence detection system (Applied Biosystems).

Primers and probe sequences (Table 1) were chosen with the assistance of Primer Express software (Applied Biosystems). Primers were purchased from Eurobio (Les Ulis, France), and probes and a qPCR Core Kit, which contained real-time PCR buffer, deoxynucleotides, MgCl2, AmpErase UNG, and Taq polymerase, were obtained from Eurogentec (Herstal, Belgium).

In the 24 frozen samples that were available, the presence of PAX3/PAX7-FKHR fusions was ascertained using a previously described method.14

Interpretation of Molecular Results

Amplification of β-2-microglobulin was used as a reference in the evaluation of cDNA amplification levels. PCR products yielded by β-2-microglobulin amplification (85 base pairs [bp]), PAX3-FKHR amplification (106 bp), and PAX7-FKHR amplification (89 bp) were similar in size.

A positive result was defined as the concomitant presence of β-2-microglobulin and PAX3/PAX7-FKHR transcript amplification products, whereas amplification of only β-2-microglobulin at the maximum threshold cycle value of 27 was considered a negative result. All other outcomes were considered to be uninterpretable.


Molecular Analysis of Paraffin-Embedded and Frozen Tissue

Molecular analysis of PET led to the detection of PAX3/PAX7-FKHR fusion transcripts in 43 cases (39.4%; PAX3, n = 37; PAX7, n = 6). Negative findings were documented in 50 cases (45.9%), and results were uninterpretable in 16 cases (14.7%). Due to the limited number of cases processed with each fixative, no correlation could be established between uninterpretable findings and the type of fixative used, although all samples fixed with AFA yielded interpretable results (Table 2). In addition, no statistical association was documented between uninterpretable findings and tissue block age (range, 1–29 years).

Table II. cDNA Polymerase Chain Reaction Amplification Results According to Fixative Type
FixativeNo. of interpretable PCR results (%)No. of uninterpretable PCR results (%)Total
  1. PCR: polymerase chain reaction; AFA: acetic acid, formaldehyde, and alcohol; HB: Holland Bouin fluid.

AFA20 (100.0) 020
HB39 (84.8) 7 (15.2)46
Bouin fluid 1 12
Formalin33 (80.5) 8 (19.5)41
Total93 (85.3)16 (14.7)109

Aside from PET samples, 24 frozen tumor samples also were tested. These frozen samples corresponded to a group of PET samples that yielded the following distribution of results: 10 PAX3-FKHR fusions, 1 PAX7-FKHR fusion, 9 negative results, and 3 uninterpretable results. In the remaining PET sample, a positive result was obtained, but the type of transcript (PAX3 vs. PAX7) could not be identified. Findings in frozen samples were concordant with findings in PET samples in all cases. The three PET samples that had yielded uninterpretable results corresponded to two negative results and one positive result in frozen tissue. Thus, overall, there were 44 positive results (PAX3, n = 37; PAX7, n = 6; transcript type not identified, n = 1), 52 negative results, and 13 uninterpretable results.


On histologic review, 45 tumors were classified as ARMS. There were 43 cases of classic ARMS, and in the remaining 2 cases, the solid variant of ARMS was detected. The cohort of patients with ARMS consisted of 22 males and 23 females, who ranged in age from 2 to 76 years (0–10 years, n = 10; 11–20 years, n = 23; 21–40 years, n = 8; and > 40 years, n = 4). Tumor sites included the head and neck (n = 15), limbs (n = 13), internal trunk (n = 10), trunk wall (n = 6), and vulva (n = 1).

Immunohistochemical analysis revealed that > 50% of tumor cells were positive for myogenin expression in all 45 ARMS samples (mean, 81%). Molecular analysis of PET samples resulted in the detection of PAX3-FKHR transcripts in 32 cases and PAX7-FKHR transcripts in 6 cases; findings were negative in 2 cases and uninterpretable in 6. Two of the six uninterpretable cases were subsequently examined using frozen material: one of these frozen samples was found to be positive for a PAX-FKHR transcript (PAX subtype not specified), and the other yielded negative findings. Therefore, overall, there were 41 interpretable cases, 39 of which yielded positive results (95%) (Table 3). Both negative results were found in ARMS with classic histology; the two ARMS in question were located in the thigh and in the pelvis, respectively, and approximately 70% of all tumor cells were found to express myogenin in both cases.

Table III. PAX-FKHR Fusion Transcript Expression According to Myogenin Immunopositivity
 % Myogenin-positive cellsTotal
  1. +: positive for PAX-FKHR fusion transcript; −: negative for PAX-FKHR fusion transcript; NI: not interpretable; ARMS: alveolar rhabdomyosarcoma; ERMS: embryonal rhabdomyosarcoma.



Sixty-four tumors, including 58 classic ERMS, 4 spindle cell rhabdomyosarcomas, and 2 botryoid RMS, were classified as ERMS. Among the patients with ERMS were 52 males (81.3%) and 12 females, who ranged in age from 0 to 78 years (0–10 years, n = 43; 11–20 years, n = 13; 21–40 years, n = 4; and > 40 years, n = 4). Tumor locations included the head and neck (n = 24), urogenital tract (n = 14), internal trunk (n = 13), limbs (n = 9), and trunk wall (n = 4).

In 37 of 64 cases (57.8%), myogenin was expressed in < 50% of tumor cell nuclei (< 10% positive nuclei, n = 10; 10–29% positive nuclei, n = 12; 30–49% positive nuclei, n = 15; 50–79% positive nuclei, n = 16; and 80–100% positive nuclei, n = 11) (Table 3). Molecular analysis of PET samples revealed the presence of PAX3-FKHR transcripts in 5 cases (Figs. 1, 2); among the remaining 59 cases, results were negative in 49 (Fig. 3) and uninterpretable in 10. One uninterpretable case was subsequently examined in frozen material and was found to be negative for the presence of fusion transcripts. Thus, overall, 55 cases yielded interpretable results, 5 of which were positive (9%) (Table 3). The clinicopathologic characteristics associated with the five PAX-FKHR-positive tumors are summarized in Table 4. All ERMS exhibited high levels of myogenin expression (Figs. 1, 2).

Figure 1.

Histologic features of embryonal rhabdomyosarcoma at a retroperitoneal site in a 16-year-old male. (A) Microbiopsy showing an infiltrative tumor lacking an alveolar pattern. (B) Poorly differentiated malignant round cell tumor. (C) Myogenin positivity is evident in the majority of tumor cells. (D) Detection of PAX3-FKHR transcripts using a real-time polymerase chain reaction (PCR) assay. Graph shows fluorescence emission intensity (which is directly related to the amount of PCR product present) after each PCR cycle. β2M: β-2-microglobulin (internal control); PAX3: PAX3-FKHR fusion transcript.

Figure 2.

Inguinal lymph node metastases from a leg tumor in a 6-year-old female. (A) Tumor is composed of irregular round cells arranged in a solid pattern. (B) Myogenin positivity is evident in the majority of tumor cells. (C) Detection of PAX3-FKHR transcripts using a real-time polymerase chain reaction (PCR) assay. β2M: β-2-microglobulin (internal control); PAX3: PAX3-FKHR fusion transcript.

Figure 3.

Paranasal sinus tumor in a 7-year-old male. (A) Tumor is composed of poorly differentiated round cells. (B) Myogenin positivity is evident in the majority of tumor cells. (C) Real-time polymerase chain reaction analysis shows evidence of β-2-microglobulin (β2M) transcripts, but not PAX3-FKHR fusion transcripts.

Table IV. Characteristics of PAX-FKHR-Positive Tumors with Embryonal Rhabdomyosarcoma Morphology
GenderAge (yrs)LocalizationSampling methodHistologyMyogenin-positive nuclei (%)
  1. M: male; F: female; mets: metastases.

M16RetroperitoneumMicrobiopsyRound cell90
F2HeadMicrobiopsyRound cell100
F6Lymph node metsIncisional biopsyPoorly differentiated90
M23PelvisMicrobiopsyRound cell60
F35PelvisIncisional biopsyPoorly differentiated95

Correlation between Myogenin Staining and PAX3/PAX7-FKHR Amplification

PAX3/PAX7 fusion transcripts were not present in any of the cases in which < 50% of tumor cell nuclei were positive for myogenin expression. In contrast, 44 of the 61 molecularly interpretable cases with > 50% myogenin immunopositivity (72.1%) were positive for PAX3/PAX7 fusion amplification.


The classification of rhabdomyosarcomas is well established, with three distinctive subtypes—embryonal, alveolar, and pleomorphic.1 For patients with rhabdomyosarcoma, histologic subtype is one of the most useful factors for assessing prognosis and making treatment-related decisions; specifically, the alveolar subtype carries an elevated risk of distant metastases and thus requires intensified therapy. The t(2;13) and t(1;13) translocations are specific markers for ARMS,1, 15, 16 and the resulting fusion transcripts (PAX3-FKHR and PAX7-FKHR, respectively) can be detected in frozen material or in PET.9–11, 14, 17, 18 Nonetheless, according to Barr et al.,19 10–30% of all cases histologically classified as ARMS fail to exhibit either of these translocations. Furthermore, the technology required to detect these molecular phenomena is not universally available, and consequently, ARMS generally is diagnosed on the basis of histologic and immunohistochemical findings. The primary objective of the current study was to evaluate the usefulness of molecular testing as compared with the classic diagnostic criteria for ARMS (i.e., histologic appearance and myogenin immunoreactivity). We were able to develop a real-time RT-PCR method for detecting the relevant t(1;13) and t(2;13) fusion products in PET samples. This technique is rapid and easy to perform and involves PCR amplification and hybridization with a molecular probe, thereby confirming the specificity of the PCR product in a single step. Nucleic acids extracted from PET are degraded to varying degrees,20 and in the current series, 14.5% of all fixed tumor samples yielded uninformative results. This uninterpretability does not appear to be related to the type of fixative used or to the age of the paraffin block from which the section was obtained. The specificity of real-time RT-PCR analysis of PET was confirmed by the finding of 100% concordance between paired frozen and fixed tumor samples with respect to PAX-FKHR fusion detection. In previous studies, fusion transcripts have been detected in 70–90% of ARMS and 0–20% of ERMS.7, 15, 16, 19, 21 In the current study, PAX-FKHR fusion transcripts were detected in 95% of all ARMS and, more significantly, in only 9% of all ERMS following careful histologic examination. This discrepancy between the current series and previous reports in the literature may be attributable to the use of different criteria for the identification of ARMS. The absence of the classic alveolar pattern in the solid variant of ARMS, the cellularity and poor differentiation observed in certain cases of ERMS, and the extensive use of microbiopsy samples make the subclassification of rhabdomyosarcomas particularly challenging. In the current series, five cases of rhabdomyosarcoma that initially were classified as ERMS were ultimately reassigned to the ARMS group due to the finding of PAX-FKHR fusion transcripts. Three of those cases involved microbiopsy samples, one involved a cellular form of the disease, and one involved a metastatic lymph node, a finding commonly associated with ARMS. Following reclassification, PAX-FKHR transcripts were detected in 44 of 46 interpretable cases of ARMS (95.6%) and in no cases of ERMS. The remaining two tumors in the ARMS group were negative for PAX-FKHR fusion transcripts. Barr et al.19 previously reported the absence of detectable fusion transcripts on RT-PCR analysis in 20% of ARMS cases; alternative detection methods revealed that some of these cases exhibited low-level positivity for fusion transcripts, and in other cases, evidence of PAX rearrangement involving genes other than FKHR or of PAX-FKHR rearrangement unaccompanied by fusion transcript expression was found. Thus, truly fusion transcript-negative cases of ARMS appear to be relatively rare. At present, any rhabdomyosarcoma in which PAX-FKHR fusion transcripts can be detected is considered an ARMS, and cases in which PAX-FKHR fusion transcripts are not detectable despite the presence of typical ARMS histology are rare.

The myogenic nuclear regulatory proteins myoD1 and myogenin function as transcription factors and stimulate myogenesis. These two proteins are expressed in poorly differentiated rhabdomyoblasts, but not in differentiated skeletal muscle. Several reports have demonstrated that myogenin is expressed in rhabdomyosarcomatous tumors but not in other pediatric tumors.7, 8, 22 Analysis of myogenin expression is routinely used in the subtyping of RMS, and this technique also proved to be useful in the current setting, as most ARMS exhibited extensive nuclear staining for myogenin, whereas ERMS generally exhibited focal positivity. Dias et al.7 analyzed a series of 26 rhabdomyosarcomas and reported a strong correlation between high levels of myogenin expression and PAX-FKHR transcript detection in ARMS. In the current study, all cases in which < 50% positive nuclei were found were classified as ERMS, and no tumor meeting this criterion showed evidence of PAX-FKHR fusion transcripts. Thus, we conclude that molecular testing can be avoided in this subset of cases. Of the 15 tumors that had > 50% nuclear positivity for myogenin but were histologically classified as ERMS, 5 had detectable PAX3/PAX7-FKHR fusion transcripts; in cases such as these 15, it is strongly recommended that molecular testing be performed before a therapeutic decision is made. Because a significant number of patients in the current cohort were referred for a second opinion, data regarding subsequent treatment and outcome were unavailable. Follow-up information was not collected, and prognosis was not assessed. Thus, we are unable to comment on whether there was any difference in biologic behavior between fusion-positive and fusion-negative cases of ERMS. A specific fusion transcript was observed in 95% of all tumors that were histologically diagnosed as ARMS and contained > 50% myogenin-positive nuclei; consequently, molecular testing could be considered optional in such cases. Recent studies16, 21, 23 suggest that patients with PAX3 fusion transcripts have a poorer prognosis than do patients with PAX7 fusion transcripts, with this being particularly true of patients with metastatic disease. Thus, for prognostic (rather than diagnostic) purposes, molecular testing of all ARMS might be indicated.

In conclusion, the current study demonstrates that real-time RT-PCR analysis of PAX3/PAX7-FKHR fusion transcripts represents a sensitive and specific test for the diagnosis of ARMS in fixed, paraffin-embedded material. Furthermore, assessment of myogenin immunoreactivity was useful in the selection of candidates for molecular testing; specifically, RT-PCR held no apparent diagnostic value for tumors containing < 50% myogenin-positive cells, whereas it was quite useful in the diagnosis of tumors in which > 50% of all cell nuclei were positive for myogenin expression.