Alterations of RB1 gene in embryonal and alveolar rhabdomyosarcoma: special reference to utility of pRB immunoreactivity in differential diagnosis of rhabdomyosarcoma subtype

Abstract

Purpose

Rhabdomyosarcoma (RMS), which is the most common pediatric soft tissue sarcoma, is classified into two major histologic subtypes, embryonal RMS (ERMS) and alveolar RMS (ARMS). RMS is occasionally reported to be the second neoplasm of hereditary retinoblastoma. Osteosarcoma is known as the most common second neoplasm of hereditary retinoblastoma, and tumorigenesis of osteosarcoma has been proven in previous studies to be related to the RB gene (RB1) alteration. Therefore, there might be a correlation between the tumorigenesis of RMS and RB1 alteration.

Methods

We examined the RB protein (pRB) expression and RB1 alteration such as allelic imbalance (gain or loss) and homozygous deletion, using immunohistochemistry, microsatellite makers, and quantitative real-time PCR in 57 sporadic RMS.

Results

Allelic imbalance was more frequently detected in ERMS (13/27), than in ARMS (3/20) (P = 0.04). Homozygous deletion on the protein-binding pocket domain of RB1 was found in 6 of 27 ERMS and in 2 of 20 ARMS (P = 0.24). Furthermore, immunohistochemical pRB labeling indexes (LI) in 31 ERMS (median value, 31%) were significantly reduced in comparison with those observed in 26 ARMS (median value, 85%) (P < 0.0001).

Conclusions

Our results support the assertion that tumorigenesis of RMS may be associated with RB1 alteration especially in ERMS, as previously reported for osteosarcoma. As for the RB pathway, each subtype of RMS may have a different tumorigenesis. In addition, immunohistochemical pRB LI may have the potential to be a useful ancillary tool in the differential diagnosis of RMS subtypes.

Introduction

Rhabdomyosarcoma (RMS) is the most common pediatric soft tissue sarcoma. Two major histologic subtypes of RMS can be identified: embryonal rhabdomyosarcoma (ERMS) and alveolar rhabdomyosarcoma (ARMS). Most ARMS is characterized cytogenetically by t(2;13)(q35;q14) or its variant, t(1;13)(p36;q14). These translocations create the chimeric genes PAX3-FKHR and PAX7-FKHR, respectively. However, no specific fusion gene has been documented in ERMS (Parham and Barr 2002a, b).

Retinoblastoma is hereditary in all patients with bilateral disease and in 10–15% of those with unilateral disease identified by a family history of retinoblastoma. Although the retinoblastoma gene (RB1) was initially cloned as a result of its frequent mutation in retinoblastoma, it is now thought to play a fundamental role in cellular regulation and is the target of tumorigenic mutations in many cell types. Recently, the cure rate of retinoblastoma has been excellent after enucleation or irradiation. Therefore, survivors of hereditary retinoblastoma are at increased risk of developing recurrent tumors or second primary neoplasms. Osteosarcoma is the most common second primary neoplasm after retinoblastoma, while rhabdomyosarcoma is also common as a secondary neoplasm (Aerts et al. 2004; Raymond et al. 2002; Tateishi et al. 2003). The tumorigenesis of these neoplasms may be related to RB1 alteration.

Therefore, we hypothesized that the tumorigenesis of RMS, especially that of ERMS, may be associated with RB1 alteration as well as osteosarcoma. In the current study of a large series of RMS, we analyzed RB1 alteration, including retinoblastoma protein (pRB) expression, allelic imbalance (gain or loss), and homozygous deletion, and compared the frequency of these alterations between ERMS and ARMS.

Materials and methods

Tumor samples

Fifty-seven formalin-fixed, paraffin-embedded RMS specimens (ERMS: 31 cases, ARMS: 26 cases) registered in the Department of Anatomic Pathology, Kyushu University, between 1970 and 2005 were available, and all the samples were from different patients. In addition, frozen tissue samples were available for 20 of the 57 cases (ERMS: 9 cases, ARMS: 11 cases). The diagnosis of all cases was based on light microscopic examination with hematoxylin-eosin (HE) staining according to the most recent WHO classification and, where necessary, immunoperoxidase procedures using the streptavidin-biotin peroxidase (SAB) method were carried out (Parham and Barr 2002a, b). That histological architecture of ERMS is composed of primitive mesenchymal cells in various stages of myogenesis, that is rhabdomyoblasts, with myxoid mesodermal tissues. In addition, ERMS is never arranged in alveolar patterns. ARMS is a primitive round cell neoplasm that cytologically resembles lymphoma and which shows partial skeletal muscle differentiation. That histological architecture produces fibrovascular septa that separate the round cells into discrete nests. All 20 frozen tissue samples were analyzed for PAX3/7-FKHR fusion transcripts as described previously, and these fusion transcripts were expressed only in ARMS (Jin et al. 2003).

Genomic DNA was isolated from paraffin-embedded or frozen material in 47 cases (ERMS: 27 cases, ARMS: 20 cases) using standard proteinase K digestion and phenol/chloroform extraction, and was used for the following molecular analysis.

Allelic imbalance analysis

Tumor and normal DNA samples were subjected to PCR using primers for the following five dinucleotide microsatellite markers at 13q12-14: D13S175, D13S153, D13S233, D13S1293, and D13S1312. D13S153 is located within intron 2 of the RB1. The primer sequences were obtained from use of the NCBI UniSTS database (http://www.ncbi.nlm.nih.gov/). The forward primer was end-labeled with 6-carboxyfluorescein (6-FAM) at the 5′ end. The procedure was carried out according to a method described previously (Takahira et al. 2005). The data which were processed using GeneScan software (Applied Biosystems) were compared between the tumor and normal DNA for each patient. Informative cases were defined as cases in which the heterozygous alleles were identified within the normal DNA. Allelic imbalance (gain or loss) was considered to occur when the allelic imbalance index (AI index) was calculated to be less than 0.70 as previously described (Bergthorsson et al. 2006; Nishimura et al. 2006). The AI index was calculated with the allelic ratios of the tumor and normal DNA (select indices <1.0), using the peak areas, with the following formula: AI index = (T1/T2)/(N1/N2) or (T2/T1)/(N2/N1), in which T and N represent “tumor” and “normal”, and 1 and 2 indicate shorter and longer alleles (Fig. 1) (Bergthorsson et al. 2006; Nishimura et al. 2006). Reproducibility was confirmed by 2–4 independent PCR amplifications for each sample.

Fig. 1.

Estimation of allelic imbalance (AI) from the output of the ABI 310 genetic analyzer: chromatograms showing the genotypes of two normal-tumor tissue pairs. AI is evident in case A (2 years old, embryonal rhabdomyosarcoma): AI index = (T2/T1)/(N2/N1) <0.7. The tumor in case B (24 years old, alveolar rhabdomyosarcoma) is unaltered: AI index >0.7

Quantitative real-time PCR for homozygous deletion

The quantitative real-time PCR assay was based on original primers that specifically amplify the protein-binding pocket domain (exons 20–24) of RB1 (GenBank accession no. L11910) and GAPDH (GenBank accession no. AY340484) as a reference gene (Table 1). Using the ABI Prism 7700 Sequence Detection System (PE Applied Biosystems, Foster City, CA, USA), the PCR was performed in a total volume of 50 μl in each well, including 25 μl of SYBR Green PCR Master Mix (PE Applied Biosystems), 50 ng of genomic DNA, and 0.35 μl of 10 μM specific forward and reverse primers, which were determined after the analysis of the optimal concentrations of each primer. Each sample was run in triplicate in separate tubes to permit quantification of the RB1 gene normalized to GAPDH. The PCR condition consisted of an initial denaturation step of 95°C for 10 min, followed by 50 cycles of 95°C for 15 s and 60°C for 1 min (Aveyard and Knowles 2004; Berggren et al. 2003; Labuhn et al. 2001).

Table 1.

Oligonucleotides used for quantitative real-time PCR amplification

Primer name	Sense 5′–3′	Antisense 5′–3′	Annealing (°C)	Size (bp)
GAPDH	gggctgctcacatattctgg	gcccaatacgaccaaatctaa	60	80
RB1 exon20	cactttgtgaacgccttctg	ctggaaaagggtccagatga	60	65
RB1 exon21	tggcatatgcaaagtgaagaa	cagcatgaggaagatccttg	60	77
RB1 exon22	gtattctataactcggtcttcatgc	accctggtggaagcatactg	60	68
RB1 exon23	acccttacggattcctggag	tgttggcagaccttctgaaa	60	79
RB1 exon24	tggtatttcatcttaacttgacagaat	gaaaatactcaccccgaatga	60	67

Data analysis was carried out using ABI Prism 7700 Sequence Detection Software, which calculates threshold cycle numbers (Ct). The Ct value is the cycle at which the fluorescent signal from a sample crosses a threshold as it enters the exponential phase in the PCR reaction. A standard curve is formulated from the starting copy number (x-axis) plotted against the Ct value (y-axis). The Ct values of unknown samples can then be used to determine the copy number from the standard curve. Only standard curves with correlation coefficients of 0.98 and higher were used. The gene dosage ratio was calculated as follows: (average copy number of target gene)/(average copy number of the reference gene). All gene dosage ratios were then normalized against the normal diploid control DNA to give a normalized gene dosage ratio (GDR). This procedure was carried out according to a method described previously (Aveyard and Knowles 2004).

Homozygous deletion was assumed when a paraffin-embedded sample had a GDR less than or equal to 0.36. This allows for the presence of up to 30% contaminating normal cell DNA in the template (Aveyard and Knowles 2004).

Immunohistochemistry

Immunohistochemical study was performed using a streptavidin-biotin-peroxidase method (histofine; Nichirei, Tokyo, Japan). The primary monoclonal antibody for the RB protein used in this study was the antibody for both the hyperphosphorylated and underphosphorylated RB gene product (clone G3-245; BD Pharmingen™, San Diego, CA; 1:800; 15 min microwave). We assessed the immunoreactivity with the labeling index (LI). To determine the RB labeling index (LI), the positively stained nuclei in at least 500 tumor cells were counted by three pathologists (KK, OY, HY) independently for each tumor. Furthermore, the protein expression in tumors was classified into two categories by the LI: heterogeneous (±), less than 50% of LI; homogeneous (+), more than 50% of LI.

Results

Allelic imbalance on chromosome 13q12-14

Allelic imbalance analysis at the 13q12-14 locus was performed in 47 cases. None of the cases showed non-informative findings for all five markers. Allelic imbalance for one or more markers was found in 13 of 27 (48%) cases with ERMS and in 3 of 20 (15%) cases with ARMS (Table 2). Representative examples of allelic imbalance are depicted in Fig. 1. The frequency of allelic imbalance in ERMS was significantly higher than that in ARMS (P = 0.04) (Table 2).

Table 2.

Relationship between RMS subtypes and genealterations

	ERMS	ARMS	P
Allelic imbalance
(+)	13	3
(−)	14	17	0.04
Homozygous deleton
(+)	6	2
(−)	21	18	0.24
Allelic imbalance and/or homozygous deletion
(+)	16	5
(−)	11	15	0.04

Homozygous deletion of protein binding pocket domain of RB1 gene

The results of quantitative real-time PCR for homozygous deletion analysis are summarized in Table 2. None of the cases showed non-informative findings for five exons. Homozygous deletion for one or more exons was more frequently found in ERMS (6/27: 22%) than in ARMS (2/20: 10%), but the difference was not statistically significant (P = 0.24) (Table 2). Furthermore, allelic imbalance and/or homozygous deletion in tumors were found significantly more frequently in ERMS (16/27: 59%) than in ARMS (5/20: 25%) (P = 0.04) (Table 2).

Immunohistochemistry of pRB

pRB LI was significantly higher in 26 ARMS cases (median value, 85%) than in 31 ERMS cases (median value, 31%) (P < 0.0001; Figs. 2, 3). Moreover, the relationship between allelic imbalance and pRB expression was also evaluated, and is summarized in Table 3. Among allelic-imbalance-positive cases, heterogeneous staining was observed in 11 out of 16 cases (69%), while among allelic-imbalance-negative cases, heterogeneous staining was recognized in 17 out of 31 cases (55%). There was no significant correlation between allelic imbalance and pRB immunohistochemical expression (P = 0.54).

Fig. 2.

Boxplot of retinoblastoma protein (pRB) expression. The median values of the pRB labeling index in embryonal and alveolar rhabdomyosarcoma are 31 and 85%, respectively (P < 0.0001)

Fig. 3.

Immunohistochemical staining of retinoblastoma protein (pRB) expression. a The numbers of tumor cells with a positive nuclear reaction are reduced; the labeling index is 37% (1 year old, embryonal rhabdomyosarcoma). b Most tumor cells show the immunoreaction in the nuclei; the labeling index is 88% (5 years old, alveolar rhabdomyosarcoma)

Table 3.

Relationship between Allelic imbalance (AI) and pRB expression

	Heterogeneous	Homogeneous	Total
AI (+)	11 (69%)	5 (31%)	16
AI (−)	17 (55%)	14 (45%)	31
			P = 0.54

Discussion

Hereditary retinoblastoma patients have a high risk of second primary neoplasms such as osteosarcoma, rhabdomyosarcoma, meningioma, and others (Aerts et al. 2004; Tateishi et al. 2003). Generally, such an osteosarcoma is likely to show loss of heterozygosity (LOH) at 13q and alteration of the RB1 tumor suppressor gene (Raymond et al. 2002; Wadayama et al. 1994). According to several previous studies, the frequency of RB1 alterations in sporadic osteosarcoma has been found to vary between 30 and 40% (Raymond et al. 2002). Therefore, tumorigenesis of osteosarcoma was indicated in association with RB1 alteration in part. Rhabdomyosarcoma was also occasionally reported as a second neoplasm of hereditary retinoblastoma (Aerts et al. 2004; Tateishi et al. 2003). Therefore, we analyzed the RB1 alteration of sporadic rhabdomyosarcoma.

In molecular analysis, we identified allelic imbalance at 13q12-14 and/or homozygous deletion of the protein binding pocket domain of RB1 gene in 21 of 47 cases (45%). In particular, in ERMS, the frequency of allelic imbalance at 13q12-14 (48%; 13 of 27 ERMS) was found to be significantly higher than in ARMS (15%; three of 20 ARMS) (P = 0.04). In the previous cytogenetic and comparative genomic hybridization studies, ERMS was found to show a complex structure and numerous chromosomal changes, often including extra copies of chromosome 13 and other changes (Bridge et al. 2002; Gordon et al. 2001; Polito et al. 1999; Xia et al. 2002), and those results supported our results.

At the present time, the common diagnostic factors distinguishing ERMS from ARMS are morphological arrangements of the tumor cells, myogenin immunoreactivity, and specific fusion gene transcripts (Parham and Barr 2002a, b; Hostein et al. 2004). However, when molecular genetic analysis was not available, especially in the solid variant of ARMS, differential diagnosis might be very difficult. In the present study, pRB expression in ARMS subtype (median value of LI, 85%) was almost preserved, whereas, pRB expression in ERMS subtype (median value of LI, 31%) was significantly decreased (P < 0.0001). Therefore, pRB LI has the potential to be a useful ancillary tool in the differential diagnosis of RMS subtypes (Fig. 4).

Fig. 4.

Solid variant of alveolar rhabdomyosarcoma (1 month old, orbit). a Although the characteristic alveolar structures are not present, cellular nests are still separated by fibrovascular septa. b PAX3-FKHR fusion gene transcript is detected in this case. c Most tumor cell nuclei show retinoblastoma protein (pRB) immunopositivity

Moreover, our immunohistochemical result suggested that RB1 gene alteration may frequently be present in ERMS, rather than ARMS; may be predominant in second neoplasms of hereditary retinoblastoma. However, the actual subtype of secondary RMS have not been precisely documented in previous studies (Aerts et al. 2004; Tateishi et al. 2003). In addition, no close relationship between pRB expression and allelic imbalance at 13q12-14 was found (P = 0.54). Those results suggested that other alterations such as hypermethylation and mutation may be present.

Although various sarcomas were found to have acquired RB1 mutation in the previous studies (Polito et al. 1999), no such mutations were identified in sporadic cases of rhabdomyosarcoma (Polito et al. 1999). While RB1 mutations were not detected in RMS, genetic changes were found in genes encoding proteins that regulate pRB function (Polito et al. 1999). In particular, RMS cases have been identified with the amplification of the CDK4 gene, and with deletions of the CDKN2A and CDKN2B genes (Polito et al. 1999). However, as pRB expression levels were reduced in our study, the possibility of a non-mutational mechanism causing the reduction of protein expression should also be considered.

Genetically, sarcomas fall into two main categories with regard to tumorigenesis (Helman and Meltzer 2003). One group is characterized by a tumor-specific translocation that seems to be the main pathogenesis of the tumor, and indeed is being incorporated as a diagnostic criterion. Another group is characterized not by a recurring, tumor-specific genetic alteration but by complex karyotypes that are characteristic of severe genetic and chromosomal instability. However, ERMS and ARMS each belong to a different one of those groups despite having the same expression profile of skeletal muscle differentiation; therefore, each subtype of RMS is suggested to have a different tumorigenesis. ARMS is found to have the specific fusion gene transcription of PAX3-FKHR and PAX7-FKHR, and ERMS is found to disclose the complex karyotypes seen in previous studies (Bridge et al. 2002; Gordon et al. 2001; Polito et al. 1999). In particular, in ERMS, allelic loss in chromosomal region 11p15 has also been observed (Xia et al. 2002; Asakura and Rudnicki 2003; Koi et al. 1993; Li et al. 1998; Loh et al. 1992; Scrable et al. 1987). In addition, we found the possibility of a relationship between RB1 gene alteration and tumorigenesis, as is considered to exist in osteosarcoma. However, the possibility that a higher frequency of RB1 alterations is simply related to genetic instability should be considered.

In summary, in each subtype of RMS, we first investigated the RB1 gene alteration, including pRB expression. ERMS significantly showed a higher frequency of RB1 gene and protein alteration than ARMS. Our results support the assertion that the tumorigenesis of rhabdomyosarcoma, especially ERMS, may be associated with RB1 gene alteration, as in osteosarcoma. Each subtype of RMS is suggested to have a different tumorigenesis with respect to the RB pathway. Moreover, pRB LI has the potential to become an ancillary parameter in the differential diagnosis of the RMS subtypes.

Abbreviations

RMS

Rhabdomyosarcoma

ERMS

Embryonal rhabdomyosarcoma

ARMS

Alveolar rhabdomyosarcoma

RB1

Retinoblastoma gene

pRB

Retinoblastoma protein

LI

Labeling index

AI

Allelic imbalance