Perspective Chapter: Osteosarcomas of the Head and Neck

Abstract

Osteosarcomas of the head and neck (HNOS) are an infrequent disease, representing less than 10% of all osteosarcomas and 1% of all head and neck cancers. However, they exhibit a different clinical behavior and natural history than extremities osteosarcomas (OS), therefore requiring a specific study and analysis. Specifically, in head and neck sites, OS have shown a different presentation age, reduced likelihood of distant metastases, and a severely higher local recurrence rates. This may be due to the difficulties in ensuring wide negative margins, given the multiple vital structures surrounding tumors in this particular region. This singular features render HNOS a different prognosis and prognostic factors, becoming a topic that should be assessed independently, as they may need a different treatment approach than osteosarcomas of the trunk or extremities.

1. Introduction

Osteosarcomas (OS) are the most frequent primary malignant tumor arising in the bone, formed by neoplastic cells that synthesize and secrete organic components of the bone matrix, which may or may not be mineralized [1]. It usually affects long bones, occurring in the craniofacial area in only 6–10% of cases and accounting for less than 1% of all head and neck cancers [2, 3, 4, 5]. However, it has a unique clinical behavior, different from OS in other parts of the body, which grants head and neck osteosarcoma (HNOS) a distinctive prognosis. This makes it a subject that should be reviewed separately. However, given its infrequency, the evidence we have on the subject is scarce, mostly based on retrospective studies and case series, and most of its management approaches are extrapolations of the treatment established for long bones OS.

2. Epidemiology

OS itself is a rare tumor, accounting for less than 1% of all cancers diagnosed annually in the United States [4], and osteosarcomas presenting on the head and neck region are even more uncommon, with incidence approximately of 2–3 per million persons per year [6].

In addition, OSHN has specific demographic characteristics at diagnosis.

While the peak incidence of extremity osteosarcomas occurs during adolescence, HNOS generally presents at a later age, albeit with significant variability. According to different series, HNOS usually presents between the third and fourth decade of life, with a wide range. Kassier et al. [7], for example, in a meta-analysis of non-randomized studies between 1980 and 1994 with 173 patients, report a median age of presentation of 36 years, with a range from 5 to 78 years. Smith et al. [8], in a review of the US National Cancer Database (NCDB) cancer registry, with 496 patients diagnosed with HNOS between 1985 and 1996, describes a median age at presentation of 38 years, with 41% of patients aged 30–60 years, 35% younger than 30 years, and 24% older than 60 years. In this study, it is also noteworthy that the age at presentation was strikingly lower in men than in women (34 vs. 44 years respectively, p < 0.001). Finally, in a study conducted by Lee et al. [9], published in 2015, using the Surveillance, Epidemiology and End Results (SEER) cancer registry database to determine the epidemiology and prognostic factors associated with osteosarcoma of the jaw (OSJ), with 541 patients (1973–2011), reported an age at presentation with a median age of 41 years and a range of 0–91 years [9]. Furthermore, the demographic distribution of this study showed 75% where white, 17% African-American, and 8% other races, and in terms of gender as a risk factor, in these two large cohorts, the sex distribution was equal, with a 1:1 ratio [8, 9].

Regarding the clinical subsite, HNOS affects the jaw in more than 80% of cases, with the mandible usually being the most common site [10, 11]. In the mandibular region, it frequently involves the mandibular body and ramus, and in the maxilla, the upper alveolar ridge, maxillary sinus floor, or hard palate. In fact, in a study by Guadagnolo et al. [10] from MD Anderson, with 119 cases of craniofacial OS, they observed 45% mandibular OS, disease in the mandible, 40% maxilla, calvarium 5%; paranasal sinuses 2%, hard palate, 2%; mastoid, 2%; skull base, 1%; zygoma, 1%; infratemporal fossa, 1%; and cervical soft tissues 1%. However, some series report a slightly higher percentage in the skull and facial bones, as in the Smith et al. [8] analysis of the NCDB where the majority of patients (55.6%) had HNOS of the skull and facial bones, and HNOS of the mandible accounted for 38.9%. Approximately 5% of patients had HNOS tumors in the other subsites, which included the soft tissues of the head and neck, parotid gland, the nasopharynx, and the tongue [8]. Likewise, in the SEER series by Lee et al. [9], the distribution of HNOS was 55.6% in the skull or facial bones and 44.4% in the mandible [9].

3. Risk factors

As for the risk factors described, genetic predisposition of young patients with some specific genetic syndromes has been evidenced:

3.1 Syndromes associated with germline mutations in tumor suppressor genes

• Sd. Li-Fraumeni (p53): autosomal dominant disorder involving a germline mutation of the p53 tumor suppressor gene. Affected individuals may suffer from breast cancers, soft-tissue sarcomas, central nervous system malignancies, leukemia, and adrenocortical carcinomas [11]

• Retinoblastoma (Rb1, 13q14 deletion) [11]. Huber et al. in a retrospective analysis of 14 patients between 1974 and 1999, reported four patients (28.6%) with this history, with an average latency of 9 years (range 3–15 years) for OSHN [12].

3.2 Syndromes associated with germline DNA helicase mutations

• Rothmund Thomas Syndrome: a recessive autonomic genodermatoses presenting with characteristic facial erythema (poikiloderma), short stature due to intrauterine and postnatal growth retardation, sparse hair, eyebrows and eyelashes, juvenile cataracts, skeletal anomalies, radial axis defects, premature aging, and predisposition to certain cancers, including OSHN.

• Werner syndrome: a rare autosomal disorder characterized by features of premature aging that appears in the third decade of life. This disorder is known to present with bilateral cataracts, short stature, graying and thinning of scalp hair, characteristic skin disorders, and increased incidence of specific tumors, including HNOS [13, 14].

• Sd. Bloom: a rare disorder associated with prenatal and postnatal growth deficiency, an erythematous telangiectatic rash on the face and other sun-exposed areas, insulin resistance, and predisposition to early-onset and recurrent cancer in multiple organ systems.

3.3 Other risk factors

Association with different bone dysplasias has also been documented: fibrous dysplasia, Paget’s dysplasia, and enchondromatosis [11].

• Fibrous dysplasia: a bone embryonic disorder in which normal bone is replaced with a mixture of immature fibrous tissue and small fragments of immature trabecular bone. The fibrous tissue proliferates within the bone marrow, compresses the cortex from the inside, and produces the expansion that characterizes the disease.

• Paget Disease: relatively common metabolic bone disorder characterized by increased rate of bone turnover, with increased bone resorption and deposition, resulting in cortical and trabecular thickening. Clinically it presents as progressive bone deformities, growth problems, fractures, vertebral collapse, increased skull size, and sensorineural hearing loss. The incidence of osteosarcoma secondary to Paget’s disease is not known, but it is estimated to be about 1%. This association accounts for about half of the osteosarcomas reported in elderly patients [13].

• Enchondromatosis: also known as Ollier disease, it is a rare sporadic nonhereditary skeletal disorder with development of multiple enchondromas distributed predominantly unilaterally or asymmetrically in the metaphyses of the long bones.

Additionally, HNOS has also been associated with trauma, bone infarcts, and chronic osteomyelitis [13].

Finally, one of the most strongly associated factors is a history of previous radiotherapy. In fact, Patel et al. [15] reported on 44 patients treated at Memorial Sloan-Kettering Cancer Center between 1981 and 1998: six patients (15%) had a history of previous radiotherapy. Different authors describe that this would be mainly for patients who received radiotherapy for leukemia or lymphoma, but no correlation has been found with respect to low-dose radiation received for diagnostic medical tests [13]. In a cohort study from Massachusetts general hospital [16], with 47 patients, prior radiation to the head or neck was documented in 27% of subjects and was statistically associated with decreased overall survival on univariate analysis (p = .01).

In parallel, it is important to consider the epidemiology of the place where we are observing the OS cases. For example, in a Chinese series reported by Luo et al. [17] in 2019, with 37 patients with HNOS, 43% of them had a history of previous radiotherapy for Nasopharyngeal Ca, given that the latter is an endemic disease in that country.

4. Clinical presentation

Clinical signs and symptoms will depend on the location of the tumor, its size, and growth rate. The vast majority of patients, up to 70–75%, present with local swelling, associated with pain in more than one-third of patients. This may be followed by facial dysesthesia in about 30% and loose teeth in about 15% [13, 16]. In fact, in Patel’s et al. [15] series, up to 27% had infraorbital nerve paresis (V2). Other manifestations include trismus, nasal obstruction or epistaxis, and/or headache, depending on their subsite.

HNOS are tumors that usually report rapid growth, and on average the time of presentation at consultation is 2–6 months [15].

Physical examination usually identifies a non-painful mass, fixed to the underlying bone, and it is important to note that the mucosa and superficial soft tissue may be normal or with some very slight alteration. The size can be variable, from an initial tumor, as seen in Figure 1, showing a patient from our institution who consulted with a relatively small swelling in right maxillary, hard, with bony consistency, in which we can see a slight irritation of the mucosa, to an obvious enlargement and bone destruction, as we see on Figure 2, with a patient with significant involvement of the maxillary bone and adjacent soft tissue.

Figure 1.

Initial presentation of a 35-year-old female patient with a maxillary osteosarcoma.

Figure 2.

Locally advanced maxillary osteosarcoma on a 48-year-old male.

In fact, in Guadagnolo’s et al. cohort [10], the median tumor size at presentation was 5.5 cm with a range between 1 and 15 cm. However, large series [8] report tumors smaller than 6 cm in 79%, with the majority being between 3 and 6 cm (41–44%), with no significant differences according to anatomic site. According to Granowski-LeCornu et al. [16], increasing tumor size was associated with decreased overall survival (P = .0167).

In addition, although HNOS patients present normally with tumors in the described size ranges, it is necessary to take into account that they are fast-growing tumors, as we see in Figure 3 showing mandibular OS of the left retromolar trigone at presentation (Figure 3a) and its growth in 3 weeks from the first consultation (Figure 3b).

Figure 3.

A 27-year-old male patient with osteosarcoma of the left mandible. (a) Patient on first consult and (b) same patient 3 weeks later.

As part of the physical examination, the evaluation of possible cervical adenopathies is mandatory; however, locoregional lymph node involvement in these patients is very unusual [11].

5. Diagnosis

The diagnosis of HNOS is based on radiological and pathological findings. Most tumors arise within the medullary cavity of the affected bone, with rare examples developing on the bone surface [1].

It constitutes a diagnostic challenge due to its rare occurrence, diverse clinical presentation, and the fact that typical radiological features may not always be present, which can often lead to misinterpretation of lesions or delay in diagnosis.

5.1 Imaging tests

Specific features have been described for these bone tumors on radiographs, CT, and MRI scans.

The effort in the evaluation of HNOS imaging should focus on searching for specific radiological features that may point to the diagnosis of HNOS, assessing bone involvement and destruction, evaluating the extent of adjacent soft tissue involvement, and ensuring the resectability of the tumor. And finally, but most importantly, looking for possible distant disease, especially pulmonary metastases. An initial diagnosis of HNOS should be considered when tumors with matrix mineralization are present early in the fourth decade of life.

The radiologic appearance of HNOS depends on the interplay of three processes: bone formation and mineralization, bone destruction, and periosteal bone formation. On plain radiography, an ill-defined radiolucent lesion is usually seen. Early tumors may show a symmetrical widening of the periodontal membrane space about one or more teeth (Figure 4). Indeed, Lindquist et al. reported that the widening of periodontal ligament space and inferior dental canal, together with sunburst effect, is almost pathognomonic of osteosarcoma of jaw bone [17]. Figure 4 shows X-rays of the patient presented in Figure 1.

Figure 4.

Radiographic imaging of patient showed in Figure 1, a 35-year-old female with a maxillary osteosarcoma.

CT and MRI both have their own superiorities in detecting osteosarcoma, and the combination of CT and MRI has proven to improve the diagnostic accuracy for patients suffering from HNOS. Key points that are important when analyzing a CT and MRI scan for a possible HNOS are summarized below.

On CT, key points include assessing:

• The extent of bone involvement and pattern of bone destruction (lytic/mixed or sclerotic lesions).

• Cortical evaluation and periosteal reaction, which can be aggressive (e.g., lamellar or spiculated) vs. non-aggressive or none.

• Presence of matrix mineralization (identification of high density osteoid matrix).

• Tumor size and tumor margins (well or ill-defined)

• Possible presence of prior bone disease.

• Evaluation of possible distant metastasis (mainly lung).

HNOS primarily exhibits osteolysis and/or osteoblastic destruction, as well as having an irregular tumor margin on CT imaging. According to Luo et al. [18], in CT, more than 97% of patients have some degree of bone destruction, presenting lytic (43%), sclerotic (19%), or mixed lytic-sclerotic (35%) lesions, with or without soft tissue involvement (Figure 5). The mixed and sclerotic radiological pattern in the head and neck region is highly suggestive of osteosarcoma, with differential diagnosis of metastasis, lymphoma, and chondrosarcoma (Figure 5). In purely lytic lesions, the diagnosis can be difficult, as osteosarcomas that mimic hollow areas without new bone formation cannot be differentiated from metastatic disease radiographically. For HNOS, primary features are local or patchy high-density shadows in the medullary cavity with varying degrees of bone destruction and matrix mineralization. In the series of Luo et al., matrix mineralization was present in (86.5%), and high-density osteoid matrix is found in 86% of lesions [15].

Figure 5.

Computed tomography of patient showed in Figure 1, a 35-year-old female with a maxillary osteosarcoma.

It is important to evaluate the cortical, as it can be invaded and eroded by the tumor, which extends into the soft tissues, frequently eliciting a periosteal reaction. The pattern of periosteal reaction can be classified as aggressive or non-aggressive according to Rana et al. [19]. Aggressive reactions include laminated, spiculated (hair-on end, sunburst), disorganized, or Codman triangle reaction patterns, while non-aggressive periosteal reactions include thin, solid, thickly irregular, or septated patterns. Up to 70–87% of the cases have an aggressive periostic reaction [1, 15]. However, sometimes, the tumor grows expanding the bone but without violating the cortex, or it can have a homogeneously radiodense surface, well demarcated from the soft tissues, resembling an osteoma, which may hinder diagnostic suspicion. In the extremities, the Codman triangle signifies subperiosteal bone formation. This feature is less frequent in the head and neck, where the classic “sunburst” appearance of malignant osteoid formation is observed, forming radiopaque striations arising from the tumor.

The rest of the regional bone structure should be examined as previous bone diseases are found in up to 8% cases [15].

Figure 5 shows CT imaging of the patient presented in Figure 1.

On MRI, key points include assessing:

• Presence of soft tissue involvement.

• Size of the mass (measurement of longest diameter).

• Signal intensities of the mass on T1- and T2-weighted images (classified as low, isointense, or high) compared with those of normal bone marrow

• Contrast-enhanced images (classified as homogeneous, heterogeneous, or with peripheral enhancement).

MRI allows a better evaluation of possible soft tissue involvement and relationship with anatomical structures, including the skull base, being crucial to determine the resectability of a tumor in some subsites.

MRI depicts soft tissues and bone marrow infiltration (medulla) better than CT imaging, showing cortical destruction and expansive masses. HNOS tumors may present with low or heterogeneous signal intensities on T1-weighted images and high or heterogeneous signal intensities on T2-weighted images. However, features of osteoblastic HNOS on MRI scans are nonspecific and often indistinguishable from those of other types of sarcoma with T2 hyperintense signals and heterogeneous post-contrast enhancement. Nevertheless, the peripheral rim enhancement observed on Gd-enhanced MR images supports the diagnosis of chondroblastic HNOS [16].

Also, non-enhanced and Gd-DTPA-enhanced MR also allows to evaluate intramedullary involvement and to differentiate osteoid matrix and necrotic, hemorrhagic or mucosal content, especially useful in sinonasal subsites. And of course, it also allows to determine the possible neural invasion.

All these features make MRI an important tool that should be considered for the assessment of biopsy taking, preoperative surgical planning, and eventually for adjuvant radiotherapy planning.

Imaging also plays an important role in the evaluation of possible distant metastases, for which the best tool remains PET CT, followed perhaps by a combination of CT + bone scintigraphy. It is important to note that in HNOS distant disease is less frequent than that observed for OS of long bones, occurring in about 5% of patients at diagnosis and affecting mainly the lung [11].

5.2 Histology, subtypes, and histological grade

The varied radiographic appearance of this lesion highlights the importance of histopathologic analysis in the diagnosis of osteosarcomas.

5.2.1 Cytology

As stated before, diagnosis is based on imaging and tissue histology. However, cytology obtained through a fine needle aspiration (FNA) has been described and could be useful to make a first diagnostic approach to a high-grade sarcoma. On a series reported by Fleshman et al. [20], with 91 patients who had an FNA reporting a high-grade sarcoma, despite only 4% were head and neck tumors, the diagnosis was confirmed by core needle biopsy, open biopsy, or excision with 8% of them being osteosarcoma and an overall diagnostic accuracy of FNA of 91%, an VPP 97%, and sensitivity of 94%. Nonetheless, current practice normally states that FNA biopsy could be used to confirm or rule out local disease recurrence or metastasis in a known sarcoma patient, but never for an initial sarcoma diagnosis or to perform a major resection based on FNA diagnosis.

5.2.2 Histology

The gold standard in the diagnosis of osteosarcoma is tissue histology, from an incisional or open biopsy.

The diagnosis of osteosarcoma is based on recognition of osteoid production by tumor cells [13, 19]. Besides the production of osteoid and immature bone, histological features are the presence of neoplastic cells showing anaplasia with epithelioid, plasmacytoid, or spindle aspects and the growth with a permeative pattern, filling the marrow space surrounding and eroding preexisting trabeculae.

Of note, it is important to highlight that osteoid and immature bone can generate confusion in the differential diagnosis of low-grade carcinomas and other conditions of fibro-osseous lesions that may contain osteoid, such as ossifying fibromas, especially in pediatric patients.

This allows solving sampling errors, histologic heterogeneity, and necrosis that can be often found in a sarcoma sample. Also, it often grants a reliable mitotic count and estimation of the percentage of necrosis, thus permitting accurate grading.

Grossly the tumors are gritty, tan-white, and sometimes myxoid. They destroy the underlying bone with or without soft tissue extension [1].

Histologically, osteosarcomas are matrix-producing tumors that contain neoplastic osteoblasts that produce bone. These osteoblasts are highly pleomorphic and/or may be spindled, epitheloid, plasmacytoid, round, or a mixture of all the above (Figure 6). Approximately, half of all osteosarcomas present as high-grade lesions [1].

Figure 6.

Histology of patient shown in Figure 1, a 35-year-old female with a maxillary chondroblastic osteosarcoma. Courtesy of Dr. Cristobal Araya.

On small biopsies, it can sometimes be difficult to distinguish osteosarcoma from a fibro-osseous lesion. In those instances, the presence of an infiltrative growth pattern can be helpful as it is seen in osteosarcoma, but not in benign lesions [13]. According to the WHO 2017, immunohistochemistry such as Ki67, Mdm2, and cdk-4 is useful in diagnostic confirmation for inconclusive cases.

5.2.3 Histological subtypes

Histologically, osteosarcoma is divided into the central (intramedullary) and peripheral (surface) subtypes.

5.2.3.1 Central sarcomas

The main type of central osteosarcoma is the conventional osteosarcoma. Conventional osteosarcoma arising in the head and neck region has the same histologic appearance seen in other locations, being composed of malignant neoplastic cells and lace-like deposition of bone.

Depending upon the predominant-type extracellular matrix present (osteoid, cartilage, or collagen fibers produced by the tumor), conventional OSs are classified into osteoblastic, chondroblastic, and fibroblastic subtypes [13]. All of these subtypes of osteosarcomas can occur in the jaw bones [1].

• Osteoblastic OS is microscopically characterized by malignant-appearing osteoblasts arranged on a matrix consisting mainly of compact bone. Large amount of osteoid is identified on the sample.

• Chondroblastic OS has a lobular architecture consisting of a cartilaginous matrix with lacunae surrounded by hypercellular regions in which malignant spindle cells are arranged. It has minimal osteoid component, thus in many chondroblastic osteosarcomas, with minimal bone production, it may be difficult to distinguish between an osteosarcoma and a chondrosarcoma (Figure 6).

• Fibroblastic OS is the least common variant. It shows the morphology of a malignant spindle cell neoplasm, in which the only indicator that it is an osteosarcoma is the scarce identifiable osteoid. The tumor usually has a mixed morphology that becomes hard to differentiate from fibrosarcoma [19].

Traditionally, it has been reported that osteoblastic and chondroblastic subtypes are the most common subtypes. Osteoblastic HNOS have been reported to account for up to 75% of the cases, leaving fibroblastic subtype as the least frequent representing 3–15% of the cases [8, 9, 15].

5.2.3.2 Peripheral sarcomas

Peripheral osteosarcomas are represented by parosteal, periosteal, and high-grade surface.

Osteosarcomas occasionally affect the jaw. The most frequent is parosteal (or juxtacortical) osteosarcoma, which represents less than 5% of all osteosarcomas. It is well differentiated and characterized by spindle cell stroma with minimal atypia and rare mitotic figures separating irregular trabeculae of woven bone, arranged in a parallel manner. Even in histopathology, peripheral osteosarcoma could have osteoblastic, fibroblastic, or chondroblastic differentiation.

Approximately 10–25% of parosteal osteosarcomas dedifferentiate into high-grade osteosarcoma with a corresponding worsening of prognosis [13].

Figure 6 shows the histology of the patient presented in Figure 1.

5.2.4 Histological grading

Histologic grade is a key part of the microscopic description of a HNOS, as it has been shown to be an independent prognostic factor. Its importance is such that nowadays, it constitutes part of the information required for staging. However, there are still substantial differences in the various expert groups on how to measure and report it.

Some authors report tumors as classified into high and low grade, as others use three categories: high, intermediate, and low. The large series to date report the histologic grade in four levels, with low grade being levels 1 and 2 and high grade, levels 3 and 4 [8, 9]. In addition, apart from different classification groups in terms of grade, there is high interobserver variability, thus making the grading reproducibility poor.

As far as consensus is concerned, cellularity and mitosis are the most important criteria used for histological grading. In general, the more cellular a tumor is, the higher is the grade. With increased cellularity, tumor loses the trabecular bone architecture and more nuclear atypia appears. Irregularity of the nuclear contour, enlargement, and hyperchromasia of the nuclei are correlated with grade.

Some authors state that the majority of HNOS are high grade [6]. Ha et al. reported up to 76.9% high-grade HNOS in a series of 27 patients [21]. Similarly grade was reported on 60 of 119 patients treated at MD Anderson Cancer Center, informing low grade on 22%, intermediate on 15%, and high grade on 63%. However, on larger series, high-grade tumors represent about 30–40% of the cases. On the.

series by Lee et al., analyzing the SEER database on 2011, with 541 HNOS patients, 40.9% of all tumors were diagnosed to be high grade at presentation, 19.6% low grade, and 39.5% were unknown [9]. In Smith et al. analysis of the NCDB database on 1996, with 496 cases, tumor grade was reported for 47.4% of the patients, a proportion that increased from 39.6% in the first years (1985–1991) to 55.9% in the later years (1992–1996), probably as the importance of histologic grading on treatment planning and prognosis became evident. In this study, of the patients with a reported grade, 38.3% had well-differentiated or moderately well-differentiated tumors, 35.7% had poorly differentiated tumors, and 26.0% had undifferentiated tumors. Interestingly, authors also report that the percentage of high-grade tumors increased as tumor size increased. Also, although mandibular tumors were distributed evenly, with 46.9% low-grade tumors (Grade 1–2) and 53.1% high-grade tumors (Grade 3–4), a greater percentage of skull and facial bone tumors were high-grade lesions (67.4% and 70%, respectively) [8].

5.2.5 Immunohistochemistry

Immunohistochemistry (IHC) plays an important role in clarifying the differential diagnoses between low grade sarcomas and fibro-osseous lesions and between chondrosarcoma and chondroblastic OS.

Focal positivity with CD68 suggests fibrohistiocytic nature of the tumors to be one of the variants of OS. Previous studies have analyzed the clinicopathological features and immunohistochemical expression of p53, MDM2, CDK4, PCNA, and Ki67 proteins in head-and-neck OS and found PCNA as one of the most favorable prognostic markers [1].

The immunohistochemistry such as Ki67, MDM2, and CDK4, is useful in diagnostic confirmation for inconclusive cases. Yoshida et al. reported that the combination of MDM2 and CDK4 by immunohistochemical analysis shows 100% sensitivity and 97.5% specificity for the diagnosis of low-grade OS and reliably distinguishes low-grade osteosarcoma from benign lesions [22].

IHC will show chondrosarcoma to be positive for S100 and vimentin and negative for cytokeratin and epithelial membrane antigen (EMA). Chondroblastic OS will be positive for vimentin, EMA, S100, and rarely cytokeratin.

Fibroblastic OS will be positive for vimentin and S100 negative, thus ruling out the neural tumors [1]. Osteonectin and osteocalcin have been widely used to study OS. Osteocalcin is specific for osteoblasts, whereas the osteonectin is not specific for osteoblasts, but consistently immunostained other cell types such as fibroblasts, pericytes, endothelial cells, chondrocytes, basal layer of the skin epithelium, nerves, and osteoclastic giant cells [23].

6. Staging

Unlike the vast majority of cancers, in OS, the staging system must incorporate not only local and distant spread, but also the degree of differentiation, in order to estimate the prognosis of the patient.

The commonly used lymph node metastasis (TNM) staging system is not commonly used for HNOS because they are unlikely to metastasize to lymph nodes. Also, the current version has been tailored for OS long bones, so it is not entirely applicable to the head and neck region, for example, the tumor size for T1 is up to 8 cm, which in this anatomical subsites generally represents a very locally advanced tumor (Table 1).

Definition of primary tumor (T)		Definition of regional lymph node (N)
T category	T criteria	N category	N criteria
Tx	Primary tumor cannot be assessed	Nx	Regional lymph node cannot be assessed
T1	No evidence of primary tumor	N0	No regional lymph node metastasis
T2	Tumor ≤8 cm in greatest dimension	N1	No regional lymph node metastasis
T3	Tumor >8 cm in greatest dimension
T4	Discontinuous tumors in the primary bone site
Histologic grade (G)		Definition of distant metastasis (M)
G	G definition	M category	M criteria
Gx	Grade cannot be assessed	M0	No distant metastasis
G1	Well differentiated, low grade	M1	Distant Metastasis
G2	Moderately differentiated, high grade	M1a	Lung metastasis
G3	Poorly differentiated, high grade	M1b	Bone or other distant sites metastasis

Table 1.

AJCC 8th edition TNM staging for bone sarcomas [24].

The most commonly system used most often to formally stage bone sarcomas is known as the Musculo-skeletal Tumor Society (MSTS) or Enneking system [13]. It is based on the grade (G) of the tumor, the local extent of the primary tumor (T), and whether or not it has metastasized to regional lymph nodes or other organs (M). The extent of the primary tumor is classified as either intra-compartmental (T1), which refers to the tumor remaining confined to the subsite in which it originated, or extra-compartmental (T2), meaning it has extended into other nearby structures. Tumors that have not spread to the lymph nodes or other organs are considered M0, while those that have spread are M1 (Table 1) [25].

In summary, with this staging system, low-grade tumors are defined as stage I, regardless of extend of primary tumor, high-grade tumors as stage II, and metastatic tumors (regardless of grade) as stage III (Table 2).

Stage	Primary tumor (T)	Regional lymph node (N)	Distant metastasis (M)	Histologic grade (G)
IA	T1	N0	M0	G1 or Gx
IB	T2 or T3	N0	M0	G1 or Gx
IIA	T1	N0	M0	G2 or G3
IIB	T2	N0	M0	G2 or G3
III	T3	N0	M0	G2 or G3
IVA	Any T	N0	M1a	Any G
IVB	Any T	N1	Any M	Any G
	Any T	Any N	M1b	Any G

Table 2.

AJCC 8th edition prognostic stage groups for bone stage primary tumor (T) sarcoma in the appendicular skeleton, trunk, skull, and facial bones.

At presentation, Lee et al. describe 18.5% of patients with stage IA disease; 0.7%, stage IB; 24.4%, stage IIA; 2.2%, stage IIB; 10.7%, stage III, stage IVA, or stage IVB (advanced disease); and 43.5%, unknown stage [9]. In Smith’s et al. analysis of the NCDB, of the 487 patients with tumors that could be staged, an AJCC stage was recorded for only 56.1%, of which 90.1% of patients with locally confined (Stage I 45%, stage II 38.9%, stage III 6.2%, and stage IV 9.9%). Interestingly, in this study, a difference was noted with regard to stage distribution by tumor location, with mandibular tumors being more likely to remain localized than skull/facial bones (92.7 vs. 82.8%, p = .032) and the other craniofacial bony sites showing higher rate of metastases of metastases than mandible (10.5% vs. 3.3) [8].

7. Treatment alternatives

Overall, HNOS are rare tumors that present unique treatment challenges. Due to its infrequency, most studies on the subject are retrospective analyses of small cohorts that utilize multiple treatment modalities, thus most treatment strategies are dictated by the existing knowledge of OS in long bones, and a variety of approaches are being applied without a standardized method of comparing relative outcomes and an answer for which the optimal treatment modality remains inconclusive.

Moreover, as stated before, HNOS has major differences from OS in the rest of the body, which could mean the need for a different treatment approach. HNOS presents at an older age than OS of long bones, has a lower metastatic potential, but a markedly higher rate of local recurrence.

Nevertheless, surgery remains the cornerstone of treatment, and negative margins the main prognostic factor and the only way to ensure locoregional control.

7.1 Surgery

The impact of impact of surgical treatment on 5-year disease-specific survival is dramatic and was evidenced on the analysis of the NCDB [8], where patients who did not undergo surgical therapy had a markedly worse survival. Patients who underwent surgery alone and surgery plus chemotherapy demonstrated similar 5-year survival rates (74.7% and 71.3%, respectively). In comparison, nonsurgical therapy resulted in a 21.7% 5-year survival [8].

However, surgical success in these patients represents a real challenge, as ensuring negative margins can be difficult, because of the anatomical complexity of the region, tumor resections are occasionally incomplete. Local recurrences and intracranial invasion have long been reported as the major causes of treatment failure due to incomplete neoplasm resection.

Furthermore, in addition to the presence of noble structures surrounding the tumor and the significant rate of irresectability that we can find in these fast-growing tumors, the lack of consensus on what we define as “adequate margins,” “close margins,” or “insufficient margins” also poses a problem. Obtaining disease-free resection margins is of course imperative, to avoid the risk of local recurrence; however, adequate margins of several centimeters, usually required for long bone OS, are often not achievable on HNOS since resecting few millimeters more often means endangering pivotal functional structures, with a noticeable decrease in the patients’ quality of life.

Of course, this may vary by tumor subsite, and the rationale and surgical treatment planning will depend on the location. It has been documented that mandibular tumors have a significantly higher chance of achieving a wide negative margin, probably because these tumors are detected earlier than in other bones of the skull/face and because larger resections can be performed with less damage to surrounding noble structures and with better chances of functional reconstruction [7, 16].

On the other hand, while intraoperative determination of resection margins might represent a useful tool in other head and neck malignancies, osteosarcomas do often pose a significant challenge for the surgeon as intraoperative pathological examination does not indeed allow for the assessment of bone margins. Only soft tissue margins can be assessed through the intraoperative consultation, thus the need to wait for final pathology report to assess adequate margins.

Anyhow, surgery should be discussed even though we can only provide a close but negative margin and not a wide free negative margin of safety as desired.

According to Ha et al., a positive margin will mean a drop on overall survival from 75–35% (P = .008) [21]. In a meta-analysis by Smeele et al. [26], in 1998, it was already state clearly that patients benefit on overall and disease-free survival when complete resection was achieved versus incomplete resection (P < .001). The latter group was still better off compared with those who did not undergo resection at all (P < .01). Survival curves show a dramatic drop on overall survival from around 50% at 5 years for complete resection to less that 25% at 3 years for incomplete resection.

When discussing the management of the neck, it is widely agreed that regional spread of osteogenic sarcomas is rare, thus prophylactic dissection of N0 patients is not indicated, regardless of histologic grade or tumor size. Therefore, selective neck dissection is only indicated in patients with clinical/radiologic nodal metastases [5].

Although there is no general consensus, nodal localization should be treated surgically and should be considered an adverse feature when evaluating adjuvant treatments.

Unlike in the management of most other head and neck cancers, prophylactic neck dissection is not advised for high-grade or large osteosarcomas of the head and neck region.

Further research in this regard would be advisable, though, as the only data available on this matter are now old and suggest that prophylactic lymph node dissection has a detrimental effect on patients overall survival [27, 28].

Figure 7 shows the surgical resection, osteo-integrated implants, and reconstruction of patient presented in Figure 1, a 35-year-old female with a maxillary chondroblastic osteosarcoma, and Figure 8 shows the surgery for patient shown in Figure 3, a 27-year-old male patient with osteosarcoma of the left mandible.

Figure 7.

Surgical resection, osteo-integrated implants and reconstruction of patient presented in Figure 1, a 35-year-old female with a maxillary chondroblastic osteosarcoma.

Figure 8.

Surgery for patient shown in Figure 3, a 27-year-old male patient with osteosarcoma of the left mandible. (a) 3D model for preoperative surgical planning. (b) Surgical tumor excised (left mandibulectromy). (c) Mandibule with the cutting guide for medial osteotomy and surgical defect after mandibulectomy. (d) Fibula-free flap harvested for reconstruction according to surgical preoperative planning and flap in-situ on left mandibule defect.

7.2 Chemotherapy

Surgery still is the main therapeutic modality for cure on HNOS. However, many trials indicate the benefit of adjuvant chemotherapy in improving survival of patients with extremity OS. Treatment approach for this disease had a major shift when several studies evidenced that chemotherapy improved significantly overall and disease-free survival [29]. Link et al. evidenced an improvement on 2 years of disease-free survival from 17 to 66% with the addition of chemotherapy to the treatment of long bones OS [30]. After that, implementation of standardized treatment protocols involving both neoadjuvant and adjuvant chemotherapy has resulted in significantly improved overall survival up to 60–80% for extremity osteosarcomas, compared with 10–20% with surgery alone [31, 32]. Multimodal treatment has also shown to improve disease-free survival, and some trials on the role of neoadjuvant chemotherapy were even successful on facilitating limb preservation in selected patients [33].

However, these trials repeatedly avoided enrolment of head and neck cases because of significant difference in clinical presentation, course of disease, prognosis, and the need for multidisciplinary treatment.

There are few retrospective studies, meta-analyses, or reviews that assess the role of chemotherapy specifically in HNOS; however, they have shown conflicting results [26, 32, 34, 35, 36, 37, 38]. In addition, these studies have small samples and use different chemotherapeutic agents or their combinations, which limits the evaluation of chemotherapy as an independent factor impacting treatment outcomes and prognosis. As a result, the benefit of chemotherapy remains unclear.

Guadagnolo et al. studied 119 patients with HNOS and failed to find a survival benefit in patients who received chemotherapy plus surgery versus surgery alone [10]. Chen et al. reported comparable findings in their study of 160 patients with HNOS [39].

On the contrary, Smeele et al. published on 201 patients with HNOS treated between 1974 and 1994 and did found a statistically significant survival benefit in patients who underwent chemotherapy and surgery versus surgery alone, on overall survival and disease-free survival. Moreover, chemotherapy was found to increase survival even in those cases of incomplete surgical resection [26].

In 2017, Boon et al. on a retrospective single-institution study of 77 patients with HNOS, where 30 patients received chemotherapy, reported an improved disease-free survival of 33% vs. 67% with addition of neoadjuvant/adjuvant chemotherapy in HNOS vs. non-chemotherapy treated patients, while the overall survival and disease-free survival were non-significant when all other cofactors were analyzed [34], a similar observation was made by Thariat et al. [35] in mandibular osteosarcomas. Nonetheless, the study by Boon et al. [34] did demonstrate a significant improvement in local recurrence rates among patients with intermediate or high-grade tumors, aged younger than 75 years, who received chemotherapy, in both univariate and multivariable analyses, postulating that the benefit of chemotherapy is thereby likely to depend on individual tumor characteristics, including grade and the presence of positive surgical margins.

While the Cooperative Osteosarcoma Study Group (COSS) protocols (neoadjuvant chemotherapy plus surgery plus adjuvant chemotherapy) have demonstrated significantly better disease-free survival in patients with extremity osteosarcoma, there is no consensus as to whether this treatment approach provides a survival benefit in patients with HNOS and timing of chemotherapy in HNOS continues to be heavily debated [7, 8, 26].

In 2021, a study by Shim et al. [32], using data from the NCDB with HNOS from 2004 to 2016, demonstrated a shift in treatment trends since the last HNOS-specific retrospective NCDB analysis was completed in 2003 [8], which mirrors treatment approach of extremity OS, showing a steady increase in the utilization of neoadjuvant and adjuvant chemotherapy in addition to surgery, with fewer patients being treated with surgery alone. However, interestingly, with no corresponding changes in estimated 2-year and 5-year overall survival, as they did not demonstrate a long-term survival benefit for HNOS patients treated with perioperative chemotherapy or radiation therapy in addition to surgery. Nonetheless, this study, which included 694 HNOS patients for the treatment analysis, found that patients treated with neoadjuvant chemotherapy and surgery plus adjuvant chemotherapy demonstrated significantly improved survival in the first 18 months after treatment compared with patients treated with surgery alone, although there was no difference in OS [32]. This observed trend in early survival could be due, in part, to benefits of neoadjuvant chemotherapy in decreasing the confines of the tumor, allowing for a more complete surgical resection. Anyhow, early increases in survival dissipate beyond 5 years after treatment. This phenomenon is perhaps a result of the tumor’s propensity for local recurrence and progression, to which patients eventually succumb.

As explained before, complete resection with negative margins is essential to adequately treating osteosarcoma. While this is relatively straightforward in extremity osteosarcoma, HNOS present unique anatomic challenges to R0 resections. Thus, neoadjuvant chemotherapy could help by shrinking the primary tumor burden, allowing higher rates of negative surgical margins, thereby reducing rates of local recurrence.

On the other hand, for OS, the rationale for adjuvant chemotherapy is treating occult disease and preventing distant metastases, which are common in extremity osteosarcomas (up to 44–49%), 9,15, with pulmonary micro metastases known to be present up to 80%. Unlike long bones OS, HNOS metastasize much less frequently (7–17%), with disease progression or failure more likely due to local recurrence [13]. The study by Shim et al. [32] found only 39 out of 1035 (3.8%) HNOS patients in their cohort with metastatic disease. Given the lower metastases rates in HNOS in comparison to extremity osteosarcoma, caution should be used before extrapolating treatment protocols aimed at preventing distant metastases as adjuvant therapy is associated with several adverse effects, including increased risk of secondary malignancy.

Finally, the moment in which to administer chemotherapy in HNOS remains under discussion. Neoadjuvant chemotherapy reported a poor response (a good response being <10% viable tumor) in the COSS study group in 66% of patients in a subgroup of maxillofacial OS patients (n = 16) [40]. However, it becomes important to assess the results from studies specifically focusing on the use of neoadjuvant chemotherapy on HNOS. Thariat et al. [35] evidenced improvement on disease-free and metastatic-free survival and an increased in clear margins rates from 50% to 68% with the use of neoadjuvant chemotherapy for HNOS. Mücke et al. also evidenced, in 2014, that neoadyuvant chemotherapy improved survival for HNOS versus surgery alone and proved to be an independent factor impacting survival on multivariate analysis [37]. Thus, Neoadyuvante Chemotherapy allows for the determination of percent tumor kill at the time of surgical resection and guides requisite changes in chemotherapeutic regimens after surgery; furthermore, it allows the evaluation of response to chemotherapy, and this may be useful as a prognostic marker or to determine adjuvant treatment [16, 36].

As for the chemotherapeutic agents studied and validated for use, Cisplatin, Doxorubicin, Adriamycin, Ifosfamide, Methotrexate, cyclophosphamide, and leucovorin are described in different combinations and schemes.

In summary, chemotherapy has shown to improve survival when added to surgery as multimodal treatment for selected tumors; however, the moment of administration (adjuvant vs. neoadjuvant) is still debated.

Figure 9 shows the final appearance, 6 months after surgery and chemotherapy, of patient presented in Figure 1.

Figure 9.

Appearance 6 months after treatment (surgery and chemotherapy) of patient presented in Figure 1, a 35-year-old female with a maxillary chondroblastic osteosarcoma.

7.3 Radiotherapy

It has been stated that conventional osteosarcoma is relatively resistant to RT; however, RT may have the positive effect of reducing the rate of local recurrence [41, 42].

Generally, radiotherapy (RT) is indicated only in HNOS patients who have close or positive resection margins [6], as the combined treatment of surgery and radiotherapy has shown to have impact on local control and on disease-free survival on HNOS patients with unknown or close margins [10]. However, its impact on overall survival has evidenced conflicting results [43].

Guadagnolo et al. studied on 119 patients, of which 92 underwent surgery alone and in the other 27 cases surgery was followed by RT [10]. They revealed on a multivariate analysis that only the margin status predicted overall survival. Analysis by resection margin status demonstrated that the combined use of surgery and 55–60 Gy dose radiotherapy was superior to surgery alone and could improve overall survival (80 vs. 31%) and disease-free survival (80 vs. 35%) in patients with positive or uncertain margins. Moreover, the addition of adjuvant RT did not improve local control for those with negative margins but did improve local control for those with positive or uncertain margins, concluding that this high-risk group is inclined to get the best results, while no advantage is expected for patients with negative margins. However, the rates of RT-associated complications were 40% and 47% at 5 years and 10 years, respectively, and severe RT complications were observed in five (19%) of 27 patients.

In addition, while the evidence supports the use of RT in patients with positive or uncertain surgical margins, the role of combined adjuvant chemoradiotherapy is not established [10]. Some experts alternatively offer chemoradiation, typically with concurrent cisplatin as a radiosensitizer [41], extrapolating from the treatment approach used for squamous cell carcinoma of the head and neck. However, since there are limited data to support the use of chemoradiation in HNOS, the decision to use it should be made in a multidisciplinary setting. If both adjuvant chemotherapy and RT are being used, some groups chose to delay RT until the end of adjuvant chemotherapy.

The optimal dose for RT on HNOS is probably similar to that used for carcinomas and is the one commonly reported to be used in different series.

The use of heavy-particle radiation such as proton beam or carbon ion therapy is promising, particularly in patients with unresectable HNOS [44]. Proton therapy may offer some benefit to those with skull base lesions, allowing to reduce the dose to the eye and central nervous system, decreasing the risk of long-term complications [6].

There is also concern for increased risk of long-term complications of adjuvant chemotherapy and radiotherapy, including development of secondary malignancies [5, 10].

In summary, key points in the treatment of HNOS:

• The treatment for HNOS cannot be extrapolated directly from that of OS in extremities, due to substantial differences in its biological behavior.

The rationale of implementing neo/adjuvant chemotherapy is always questioned as it has a notoriously lower rate of distant metastasis, and the need for a complete resection becomes increasingly important as it has greater failure due to local recurrence.

• Surgery remains the cornerstone of treatment.

• Low-grade tumors could be candidates for exclusive surgical treatment.

• Regardless of whether the planned approach is uni or multimodality treatment, the effort must be placed on achieving a complete resection, since positive or close margins have an important impact on survival.

• Controversy persists as to what we define as adequate margin.

• In high-grade and advanced-stage tumors, multimodality treatment using surgery and chemotherapy is accepted as standard and has shown to improve survival outcomes vs. surgery alone. However, the order in which treatments provide the greatest benefit is not yet clearly established. This controversy is especially centered on the management of resectable tumors, since unresectable tumors will probably be approached primarily with chemotherapy, which may also serve as a prognostic marker.

• Controversy persists as to when to indicate chemoradiotherapy, but the indication for adjuvant radiotherapy is fairly well accepted in HNOS patients with close or positive margins.

8. Prognostic factors

Unfortunately, regardless of the treatment regimen, survival of patients afflicted with osteosarcomas of the head and neck remains poor.

While there are several single and multicenter studies reported, population data are scarce and controversial. On a literature review by Mendehall et al. [6] in 2011, overall survival shows a very wide range among the different series, going from 24 to 86% at 5 years. In Smith et al. [8] NCDB study with 496 patients, the 5-year disease-specific survival was reported to be 59.7%, for patients who were diagnosed with HNOS between 1985 and 1991. Similarly, in Lee et al. [9] cohort of the SEER tumor registry, with 541 patients, the 5-year overall and disease-specific survival were 52% and 62%, respectively, and the 10-year overall and disease-specific survival were 35% and 54%, respectively, with a median overall survival of 8 years.

Both of this large series report multiple patient, tumor, and treatment factors that were associated independently with a significantly worse survival. These risk factors included: skull or facial bone sites (vs. mandible); age older than 60 years; tumor size >6 cm; osteoblastic/NOS histology; high histological grade; advanced stage at presentation; nonsurgical initial therapy; and the presence of residual disease after surgical resection [8, 9].

Different series confirm the finding that mandibular tumors appear to have a better prognosis than maxillary tumors, showing significant difference in the metastatic rate and local recurrence [8, 9, 16]. The median overall survival for osteosarcoma of the mandible has been reported to be 10.4 years vs. 6.3 years for osteosarcoma of the skull/facial bones, including the maxilla [9]. This may be due to a greater metastatic potential in more vascularized sites such as the maxilla and skull base and/or to a more economical resection in these areas, compared with the mandible, with smaller oncologic safety margins, given the greater anatomic complexity and proximity to vital structures. Maxillary tumors show greater rate of positive margins [16]. In any case, this observation emphasizes the need for greater therapeutic consideration.

Age is another factor named repeatedly by different authors to impact survival, generally 60 years and older. Advanced age could affect by an age-dependent T-lymphocyte depletion, an intolerance to the cytotoxic effects of chemotherapy, a higher propensity to development of metastatic disease, or differences in management based on age [9].

In addition, a tumor size, histology, and grade have also shown to impact survival. Of note, osteoblastic OS has evidenced worse overall survival [8], and fibroblastic and chondroblastic osteosarcoma had the best prognosis of histological subtypes, while osteosarcoma in Paget disease had a particularly poor outcome with an overall survival not longer than 6 months [9].

Histological Grade has been proved to impact severely on survival, thus is now taken into account for staging [8, 9, 16, 39]. Within the NCDB [8], patients with low-grade HNOS tumors showed a 74% 5-year survival rate, compared with 42% for patients with high-grade tumors. Ha et al. [21] also noted a marked discrepancy in survival for patients with high-grade versus low-grade HNOS tumors, with a difference at 5 years of nearly 60%. However, on the SEER tumor registry [9], the multivariate analysis found that grade was not considered to be and independent significant determinant of survival, which could perhaps be explained by the high percentage of cases defined to be unknown.

On another hand, treatment has shown to impact dramatically patients’ outcome, and surgical resection has been found to improve 5-year overall and disease-specific survival [8, 9, 16]. Complete surgical resection with wide margins has been reported as the most significant prognostic factor in HNOS [6, 13, 15]. It is reasonable to state that patients with HNOS who are not candidates for surgical resection because of advanced disease or notable comorbidities at presentation may bias the survival advantage seen in the different cohorts. However, according to Smith et al. [8], patients who did not undergo surgical therapy had a markedly worse survival with a 5-year overall survival of 21.7% compared with 74.7% for patients treated with surgery alone and 71.3% surgery plus chemotherapy.

Finally, prognosis follows a dynamic course that has been and will continue to change as the best treatment approach for these patients becomes clearer. In a study by Granowski-LeCornu et al. [16], where patients with OS of the jaw were treated from 1967 to 1991 where compared with patients treated from 1992 to 2009, this second group had better prognosis than patients in the earlier treatment group (overall 5-yeral survival rate of 77% vs. 52%). They discuss that this could be explained by several factors favoring the latter group, such as better imaging, both CT and MRI, allowing for earlier diagnosis, thus, smaller tumors at diagnosis and better treatment planning. Also, more sophisticated reconstructive techniques allow wider ablative procedures and better chance of achieving clear margins.

For more recent HNOS patients, we now offer improved diagnostic tools, more aggressive treatment, and better surveillance. The role of neoadjuvant chemotherapy remains to be elucidated and may perhaps, added to all this other factors, continue to improve prognosis.

9. Conclusions

HNOS is a rare and complex disease, for which its treatment approach is still under debate. It shows different clinical and oncologic behavior from OS of the extremities, thus requiring specific studies, which are scarce due to its infrequency. Further population-based studies are required to determine the therapeutic approach that will prove most successful. Surgical resection with large, clear margins remains the mainstay of optimal treatment, and adjuvant treatments should be discussed on a case-by-case basis. For the time being, there is consensus in that it should be managed in tertiary centers that concentrate the cases and that can offer the tools to confirm the correct diagnosis and perform the correct staging, evaluate and offer the possibility of adequate ablative and reconstructive surgery, adjuvant treatment if required, and the correct follow-up.

Acknowledgments

Dr. Cristobal Araya, maxillofacial surgeon and pathologist at Instituto Nacional del Cáncer Chile.

Head and Neck Surgery Department at Instituto Nacional del Cáncer Chile, Drs. Rodrigo Montes, Fabio Valdés, Marcelo Veloso, Luis Pizarro, Felipe Bustos, and Katherine Lozano.

Conflict of interest

The author declares no conflict of interest.