Bone- and Cartilage-Forming Tumors and Ewing Sarcoma: An Update with a Gnathic Emphasis

Abstract

Over the past decade, there have been remarkable advances in bone tumor pathology. Insights into the genetic basis and pathobiology of many tumor types have impacted diagnosis, classification, and treatment. However, because gnathic lesions may comprise only a small proportion of cases overall for many tumors, clinicopathologic features and management considerations specific to this subset may be overlooked. Here we provide a summary of recent developments in the following tumor types: osteosarcoma (OS), chondrosarcoma (CS), osteoid osteoma (OO), osteoblastoma (OB), and Ewing sarcoma (ES). In particular, we will give special consideration to cases arising in the jaws.

Osteosarcoma

The jaws represent the 4th most common site for osteosarcoma (OS), with about 6 % of all cases arising in this location. The most common gnathic sites of involvement are the mandibular body and the canine-premolar area of the maxilla. Most patients present with swelling and pain; tooth displacement and overlying ulceration are also possible [1, 2].

Gnathic OS recently has garnered attention as an entity distinct from extragnathic OS for several reasons [2–4]. First, extragnathic OS exhibits a bimodal age distribution, with a peak in the 1^st and 2^nd decades and a lesser peak among individuals older than 60 years. In contrast, jaw lesions exhibit a peak in the 4th decade, with a mean age of 36.9 years (approximately 1–2 decades older than the mean age for extragnathic lesions) [1, 2]. Secondly, extragnathic OS is highly aggressive and often presents with metastasis. Five-year survival with surgery alone is <50 % but has improved to about 70 % with current combined treatment protocols [e.g., for high-grade, resectable OS: neoadjuvant chemotherapy, primary tumor resection, metastectomy (if appropriate), and adjuvant chemotherapy]. OS of the jaws, however, metastasizes infrequently, with local progression accounting for most morbidity and mortality. Surgery is the mainstay of therapy, with approximately 70 % overall 5-year survival [3]. Reported outcomes for combined treatment of jaw tumors are variable; further studies are needed to determine whether there is a significant survival advantage for patients receiving chemotherapy and surgery versus surgery alone [2, 5]. In the largest series (n = 214) of gnathic OS reported to date, Baumhoer et al. [4] found that patients receiving chemotherapy did not exhibit significantly better outcomes compared to those not receiving chemotherapy. In this study, the only independent prognostic factors were resection margin status and metastasis. In particular, due to anatomic complexity, it is often more difficult to achieve complete surgical removal for maxillary compared to mandibular lesions. One small study noted greater Ki-67 proliferation indices and vascular endothelial growth factor (VEGF) expression in long bone tumors compared to jaw lesions, thereby giving support to the more aggressive nature of extragnathic OS [6]. However, more research is needed to elucidate the biology underlying differences in behavior.

A classification scheme for OS is provided in Table 1. Among conventional tumors, the osteoblastic variant is most common in the appendicular bones, whereas the chondroblastic variant is most common in the jaws (Fig. 1a, b).

Table 1.

Osteosarcoma classification scheme

Type	Variants	Grade
Central (intramedullary)	Conventional variants	High
	Osteoblastic
	Chondroblastic
	Fibroblastic
	Rare variants	High
	Telangiectatic
	Small cell
	Epithelioid
	Giant cell-rich
	Osteoblastoma-like
	Chondroblastoma-like
	Low-grade central	Low
Surface (juxtacortical)	Parosteal	Low
	Periosteal	Intermediate
	High-grade surface	High
Extraskeletal		Low to high

Source: Klein MJ, Siegal GP. Osteosarcoma: anatomic and histologic variants. Am J Clin Pathol. 2006;125:555–581

Fig. 1.

a Gross photograph of a large osteosarcoma involving the ramus (left) and body (right) of the mandible. This osteosarcoma, which contains a prominent chondroblastic component, extends out of the mandible into adjacent soft tissues circumferentially. b Classic chondroblastic osteosarcoma of the jaw, containing areas of cartilaginous differentiation and abundant osteoid matrix. (hematoxylin and eosin, original magnification ×200)

Some gnathic tumors are very low grade, with radiographic and microscopic features that mimic a benign fibro-osseous lesion. Genetically, most low-grade central and parosteal OS exhibit supernumerary ring and giant marker chromosomes with amplification of 12q13–15—a feature typically not found in benign fibro-osseous lesions [7]. Within this region are the cell cycle-related genes MDM2 and CDK4. Recent studies have found immunohistochemical expression of MDM2 and/or CDK4 to be present in 89–100 % of low-grade OS but absent in most benign mimics, including fibro-osseous lesions; these findings were confirmed by comparative genomic hybridization and quantitative real-time polymerase chain reaction (qRT-PCR) [7, 8]. However, Tabareau-Dalanlande et al. [9] noted discordant results, with 33 % of ossifying fibromas and 12 % of fibrous dysplasias exhibiting MDM2 amplification by qRT-PCR but no cases exhibiting MDM2 overexpression by immunohistochemistry. These investigators also showed amplification of an MDM2 neighbor, RASAL1, in all the fibro-osseous lesions with MDM2 amplification but in none of the low-grade OS studied. Interestingly, there are a few reported cases of low-grade OS undergoing dedifferentiation with retained immunoexpression of MDM2 and CDK4 [7, 8].

Chondrosarcoma

Compared to OS, chondrosarcoma (CS) is much rarer in the jaws. In the Mayo Clinic series, about 4 % of all conventional CS originated from the jaw and facial bones, with about 30 % of this subset involving the mandible or maxilla [10]. Gnathic CS tends to arise in the mandibular symphysis, coronoid process, and condyle; these sites correspond to regions of endochondral bone formation. The peak age at presentation is in the 4th decade.

Conventional CS of the jaws and extragnathic sites exhibit similar histopathologic features (e.g., lobulated growth pattern, varying degrees of cellularity and atypia). Secondary features may include spindling of tumor cells, ossification, calcification, and myxoid matrix. The primary differential diagnosis for gnathic CS is chondroblastic OS. Chondroblastic OS arises more commonly in the jaws and is more responsive to chemotherapy compared to CS. Conventional CS produces hyaline or myxoid cartilage only; in contrast, chondroblastic OS produces both cartilage and osteoid matrix, although very careful histologic evaluation may be needed to identify only focal osteoid in some cases.

Immunohistochemical studies have noted more frequent expression of ezrin and galectin-1 in chondroblastic OS compared to CS [11, 12]. More recently, IDH1 and IDH2 mutations have emerged as the most useful markers for distinguishing cartilaginous lesions from their mimics. In an analysis of approximately 1,200 mesenchymal tumors by Sequenom^® spectrometry, Amary et al. detected IDH1 (codon R132) mutations or IDH2 (codon R172 or R140) mutations in the majority of central and periosteal cartilaginous tumors (ranging from chondroma to high-grade CS) but not in OS, peripheral CS, or osteochondroma [13]. Similarly, Kerr et al. [14] found somatic mutations in IDH1 (codon 132) or IDH2 (exon 4) in 61 % of CS versus 0 % of OS.

Mesenchymal Chondrosarcoma

This CS variant is unusual in that it most often (22–27 % of cases) arises in the jaws. There is an even distribution between the mandible and maxilla. The tumor exhibits a distinctive biphasic microscopic pattern, characterized by cellular hyaline cartilage and small, undifferentiated round to spindled cells, often with staghorn-shaped vessels (Fig. 2a, b). Immunohistochemically, the chondrocytes are reactive for S-100 protein, and the small cells express CD99. Immunoreactivity for Sox9 (a master regulator of chondrogenesis) may help to distinguish mesenchymal chondrosarcoma (MC) from other small round cell tumors [15].

Fig. 2.

a Gross photograph of a large mesenchymal chondrosarcoma arising within the ramus of the mandible. The biphasic nature of this tumor is evident, with the heavily calcified chondroid component centrally and the peripheral, fleshy-appearing areas which correspond to the round cell component. (Courtesy of Dr. Thomas W. Bauer, Cleveland Clinic Foundation). b Mesenchymal chondrosarcomas have a biphasic microscopic appearance, with a chondroid component as well as an undifferentiated component composed of small, round cells and prominent staghorn-shaped vessels. (hematoxylin and eosin, original magnification ×200)

In 2012, Wang et al. [16] identified a novel, recurrent fusion involving two genes on chromosome 8: HEY1 and NCOA2. HEY1 is a downstream effector of Notch signaling, interacts with Runx2 in mice, and plays an important role in osteoblastic differentiation. NCOA2 is a member of the p160 nuclear hormone receptor transcriptional co-activator family; it facilitates chromatin remodeling and the transcription of nuclear receptor target genes. Wang et al. hypothesize that the pathogenesis of MC may be due in part to aberrant activation of Notch target genes by the HEY1-NCOA2 fusion protein. Subsequent small-scale studies have identified HEY1-NCOA2 fusion in 75–80 % of cases by FISH, alone or combined with whole genome sequencing [17, 18]. Furthermore, a novel IRF2BP2-CDX1 fusion due to t(1;5) (q42;q32) has been noted in one case of extra-osseous MC [18]. Thus, although the HEY1-NCOA2 fusion appears to be a sensitive marker for MC, there may be some genetic heterogeneity.

MC generally is aggressive. Some studies suggest that the prognosis is more favorable for lesions arising in the jaws compared to other sites. Nevertheless, one recent study noted no significant difference in prognosis for maxillary versus extragnathic tumors [19].

Osteoid Osteoma

Combined analysis of recent reviews shows a total of 26 well-documented gnathic osteoid osteomas (OO) [20, 21]. Similar to extragnathic lesions, jaw lesions typically arise in young individuals (average age: 27 years, peak: 2nd and 3rd decades), albeit with a broad reported age range (4–77 years). There is an approximately even gender distribution, in contrast to a male predilection for OO overall. The mandible is affected more often than the maxilla (81 vs. 19 %). Most patients exhibit painful swelling. Lesions involving the mandibular condyle may cause limited mouth opening [22].

Classically, plain radiography shows either a radiolucent nidus with surrounding dense cortical sclerosis or, less commonly, a targetoid appearance (central sclerosis within a circumscribed radiolucency). However, among reported gnathic lesions, the nidus more often appears radiopaque or mixed than radiolucent. Interestingly, a jaw tumor recently described by An et al. [20] exhibited more than one nidus; this feature also has been noted in a few extragnathic lesions. A periosteal reaction also may be evident. CT is the best imaging method for OO but has been performed in only a few reported gnathic cases.

Because OO and osteoblastoma (OB) share similar microscopic features, diagnosis requires correlation with clinical and radiographic features, such as OO’s tendency for cortical origin and limited growth potential (nidus <1–2 cm). However, some authors question whether size conventions based on extragnathic tumors apply to gnathic lesions [23]. Also, although nocturnally exacerbated pain relieved by nonsteroidal anti-inflammatory drugs (NSAIDs) is considered highly characteristic of OO, this feature has been described in only a few gnathic cases [22, 24–27].

A recent immunohistochemical study by Dancer et al. [28] showed strong nuclear expression of Osterix and Runx2 in OO and OB; these transcription factors play important roles in skeletal development and osteoblastic differentiation. Furthermore, Barlow et al. [29] demonstrated that OO and OB are similar with respect to the following: innervation (as evidenced by immunoreactivity pattern for S-100 protein, PGP9.5, and neurofilament and lack of reactivity for GFAP); stromal immunophenotype (positive for EMA and NSE, negative for SMA); and the presence of large, desmin-positive “smudge” (degenerating) cells. These authors propose that OO and OB should be regarded as a single entity.

OO of the jaws typically is treated by conservative surgical removal, although condylar lesions may require resection. Biopsies obtained by traditional open surgery facilitate identification of characteristic microscopic features (i.e., abrupt interface between the nidus and surrounding reactive, sclerotic bone; central nidus with osteoid, bone, and differentiated osteoblasts within a vascularized fibrous stroma).

For extragnathic lesions, the trend toward minimally invasive treatment (e.g., image-guided percutaneous radiofrequency ablation, laser ablation, cryoablation, or trephine resection) can present challenges for the pathologist, such as limited samples with prominent artifact. Furthermore, successful treatment by noninvasive MR-guided focused ultrasound recently has been reported in a few patients [30]. The most conservative option is monitoring for spontaneous regression (typically over a period of months to years); NSAID administration might accelerate healing [31].

The prognosis is excellent. Recurrence or persistence after incomplete removal has been reported in only one gnathic case [20]. Sporadic reports suggest progression of OO to OB, although a recent critical appraisal by Chotel et al. [32] questions this concept.

Osteoblastoma

A review of the literature reveals at least 135 well-documented cases of gnathic OB [23, 33–52]. The average age is 23 years (range 3–78 years), with a 1.1:1 female-to-male ratio and 72 % involving the mandible. Most patients present with swelling accompanied by pain, tenderness, or discomfort, although some are asymptomatic. Radiographic findings are variable, ranging from radiolucent to mixed or radiopaque, with ill- to well-defined borders (Fig. 3a). Most gnathic lesions are treated by curettage or resection. For extragnathic lesions, percutaneous image-guided radiofrequency ablation has emerged as an alternative treatment. Persistence (following incomplete removal) or recurrence has been reported in 11 gnathic cases (6 mandibular, 5 maxillary) [35, 45, 48, 52]. Notably, Woźniak et al. [40] recently reported a mandibular OB that recurred twice, transformed into OS, and resulted in death.

Fig. 3.

a: Computed tomography (CT) image of a maxillary osteoblastoma shows a well-defined, expansile radiolucency with scattered opacification. (Courtesy of Dr. Michael Zetz). b Biopsy of the lesion depicted in a showed conventional osteoblastoma, with plump osteoblasts lining the bony trabeculae. (hematoxylin and eosin, original magnification ×400). c Epithelioid osteoblastoma with numerous enlarged, epithelioid osteoblasts. (hematoxylin and eosin, original magnification ×200). Current evidence suggests that epithelioid histomorphology alone is not indicative of aggressive clinical behavior

There are several recent reports of “aggressive osteoblastoma” or “osteoblastoma with aggressive features” involving the jaws [43–45, 52]. Unfortunately, there is confusion regarding such terminology, with some authors describing lesions with truly aggressive clinical behavior and others describing lesions with merely epithelioid histomorphology. Nevertheless, there is no evidence that microscopic features alone are indicative of increased recurrence potential [53]. Instead, size >4 cm and an anatomic location that makes complete removal difficult may be of greater importance than histomorphology in predicting aggressive behavior [43]. Microscopic features characteristic of the epithelioid variant include the following: large epithelioid osteoblasts that rim the bony trabeculae or form solid sheets (Fig. 3c); wide, irregular bony trabeculae with solid to lacelike osteoid deposition; occasional typical mitotic figures; lack of stromal pleomorphism and necrosis; and peripheral maturation of lesional bone without entrapment or permeation of normal host bone [43]. The latter features aid in distinction from OS.

Recent cytogenetic findings include t(4;14;17) (q23 ~ 25;q31;q31) in an aggressive femoral OB and t(1;2;14) (q42;q13;q24) in a conventional femoral tumor; similar abnormalities in chromosomes 1 and 14 have been described previously [54, 55]. Nord et al. [56] found more heavily rearranged genomes in 2 aggressive OBs compared to 9 conventional lesions. Furthermore, these authors found recurrent 22q deletions in one conventional and 2 aggressive OBs; the affected genes included MN1 and NF2 (both implicated in leukemia and solid tumor development) and ZNRF3 and KREMEN1 (inhibitors of the Wnt/beta-catenin signaling pathway). Aberrations in the Wnt/beta-catenin pathway have been described in osteoarthritis, osteoma, and OS as well.

Ewing Sarcoma

In the current WHO classification, the term “Ewing sarcoma” (ES) encompasses the following entities: classical ES, Askin tumor (small round cell tumor of the chest wall), primitive neuroectodermal tumor (PNET), atypical ES, and extraskeletal ES [57]. This recent simplification of terminology is based on genetic and clinicopathologic similarities. The histogenesis remains uncertain and controversial, although many authorities favor an origin from mesenchymal or neural crest-derived stem cells.

ES mainly affects pediatric patients; however, increased availability of molecular diagnosis has led to increased detection among adults [58]. Greater ES susceptibility among individuals of European compared to African descent correlates with the ethnic distribution of recently identified single nucleotide polymorphisms in the vicinity of TARDBP and EGR2 [59]. Additional risk polymorphisms have been identified in NR0B1 microsatellites and CTLA-4 [60, 61].

Recent sporadic reports demonstrate the potential for gnathic ES to mimic odontogenic or periodontal infection [62–66]. Clinical findings may include painful swelling, nonvital teeth, dental percussion tenderness, tooth mobility, increased periodontal probing depths, and/or fever. Radiographs may show nonspecific, ill-defined radiolucency. However, lack of improvement after endodontic therapy, antibiotics, and/or tooth extraction should prompt further evaluation. In some cases, paresthesia may portend malignancy.

Review of the English-language literature reveals only 11 cases of primary gnathic ES with genetic confirmation [65–70]. Most lesions exhibited conventional microscopic features and EWSR1 or, more specifically, EWSR1-FLI1 rearrangement. However, the case described by Makary et al. [69] exhibited spindle cell features. Also, Steyn et al. [71] described an unusual tumor with osteoid production, for which they considered classification as either atypical ES with osteoid production or small cell OS with EWSR1 translocation. There is little data regarding uncommon chromosomal translocations (Table 2) in gnathic ES, although at least 2 cases with EWSR1-ERG translocation have been described [70, 72]. For extragnathic ES, studies with genetic confirmation have demonstrated remarkable histopathologic diversity—including not only enlarged pleomorphic cells in atypical ES but also lesser-known variants (e.g., sclerosing, adamantinoma-like, spindle cell, clear cell, and vascular-like types) [73, 74].

Table 2.

Chromosomal abnormalities in Ewing sarcoma

Chromosomal abnormality	TET-ETS family fusion product	Approximate percentage of Ewing sarcoma cases (%)
t(11;22) (q24;q12)	EWS-FLI1	85
t(21;22) (q22;q12)	EWS-ERG	10
t(7;22) (p22;q12)	EWS-ETV1	<1
t(17;22) (q21;q12)	EWS-ETV4	<1
t(2;22) (q35;q12)	EWS-FEV	<1
t(16;21) (p11;q22)	FUS(TLS)-ERG	<1
t(2;16) (q35;p11)	FUS(TLS)-FEV	<1

Source: de Alava et al. [57]

As for immunophenotype, CD99 and FLI1 remain the most widely recognized markers for ES, albeit with low specificity. Recent investigations also have demonstrated NKX2.2 immunoreactivity in 80–93 % of ES but only 11–16 % of other small round cell tumors studied; frequent galectin-1 expression in small cell OS but not ES; ERG expression in not only ERG-rearranged ES but also EWSR1-FLI1-associated ES and other tumor types; and variable ERG reactivity influenced by clone and antibody selection [75–78].

Importantly, detection of EWSR1 rearrangement by FISH is not specific for ES. Indeed, other small round cell tumors—such as desmoplastic round cell tumor–may exhibit EWSR1 rearrangement. Furthermore, because of variability in fusion partners and breakpoints (Table 2), a negative RT-PCR result does not entirely exclude ES [79].

Advances in molecular diagnosis have enabled the discovery of various rare fusions in so-called “Ewing-like sarcomas” (i.e., undifferentiated small round cell sarcomas with histologic and immunophenotypic similarities to ES) (Table 3). It is uncertain whether Ewing-like sarcomas represent ES variants or distinct entities. There is an increased tendency among Ewing-like sarcomas (and also ES with rare translocation types) to arise within soft tissue and to exhibit atypical histopathologic features. To the best of our knowledge, no well-documented cases of gnathic Ewing-like sarcoma have been reported, although a recent abstract by Flaitz et al. [80] describes an infant with a congenital mandibular lesion exhibiting EWSR1-NFATc2 translocation. Currently, patients with Ewing-like sarcomas and rare ES variants are treated in a manner similar to those with typical EWSR1-FLI1-associated ES. However, improved understanding of tumor biology may lead to tailored treatment in the future.

Table 3.

Chromosomal abnormalities in Ewing-like sarcoma

Chromosomal abnormality	Corresponding fusion gene	Year first described [citation]
t(20;22) (q13;q12)	EWSR1-NFATc2	2009 [88]
t(6;22) (p21;q12)	EWSR1-POU5F1	2005 [89]
t(4;22) (p31;q12)	EWSR1-SMARCA5	2011 [90]
Paracentric inv(22) in t(1;22) (p36.1;q12)	EWSR1-PATZ1	2000 [91]
t(2;22) (q31;q12)	EWSR1-SP3	2007 [92]
t(4;19) (q35;q13)	CIC-DUX4	1996 [93]
t(X;19) (q13;q13.3)	CIC-FOXO4	2014 [94]
t(18;19) (q23;q13.2)	(CIC?-unknown partner)	2010 [95]
inv(X) (p11.4p11.22)	BCOR-CCNB3	2012 [96]
t(16;20)	FUS-NFATc2	2014 [97]
t(12;15)	ETV6-NTRK3	2014 [97]

Although multimodal treatment protocols have been instrumental in improving patient outcomes, ES cure rates have plateaued for more than a decade [81]. The Children’s Oncology Group (COG) recently reported that multimodal therapy with intensified chemotherapy achieved 73 % event-free 5-year survival for patients with apparently localized disease [82]. In contrast, despite aggressive chemotherapy, patients with detectable metastatic disease at diagnosis continue to exhibit poor outcomes (5-year survival <35 %) [58, 83]. Likewise, the prognosis is poor for patients with refractory or recurrent disease. Hence, there is intense interest in developing novel treatment strategies.

Current ES translational research trends are reviewed in detail elsewhere [84, 85]. In brief, using gene expression-based, high-throughput screening methods, investigators have identified candidate compounds that may inhibit the aberrant transcription factor EWS-FLI1 directly or may modulate the expression of downstream targets of EWS-FLI1. Other experimental strategies for developing novel therapies include disrupting epigenetic mechanisms of EWS-FLI1-mediated gene repression, PARP1 inhibition, p53 activation, overcoming tumor resistance to IGF-1R inhibition, and immunotherapy.

Evidence of metastasis at diagnosis is the most important prognostic factor. Additional adverse factors include large tumor size, poor histologic response to neoadjuvant chemotherapy, and axial location. The prognosis may be more favorable for ES of the jaws compared to the long bones and pelvis, although further study is needed [63].

Recent European-E.W.I.N.G. 99 and COG trials found no significant relationship between EWSR1 fusion type and outcome [58, 86]. Apparently, the survival advantage previously reported for patients with fusion between exon 7 of EWSR1 and exon 6 of FLI1 no longer holds true with modern intensified treatment regimens [86]. Several retrospective studies suggest that CDKN2A (INK4A/ARF) and TP53 abnormalities correlate with poor prognosis, and further evaluation of these cell cycle gene alterations is underway in a large-scale analysis by COG and the ES Biology Committee [87]. Also of potential prognostic significance but requiring validation are various chromosomal copy number alterations and subclinical disease detection by RT-PCR or flow cytometry. The search for early markers of treatment response has become especially urgent with the growing number of clinical trials for metastatic and/or refractory ES [87].