Head and neck cancers represent a clinically diverse group of tumours with distinctive molecular features. Understanding the importance of characteristic molecular changes has permitted the definition of a new clinicopathological entity – the Human Papillomavirus (HPV) related subset of oropharyngeal squamous cell carcinomas that are generally associated with a good prognosis. We briefly discuss the key clinical and pathological differences between HPV-related and HPV-unrelated oropharyngeal disease and the underlying molecular differences, and we also consider how these features can inform clinical management. For rarer head and neck tumours or those that lack effective systemic treatment options, such as salivary gland tumours, sinonasal carcinomas and NUT midline tumours, we discuss how an understanding of underlying molecular features can facilitate the exploration of novel treatment options. Thus we demonstrate in this brief review, that despite the rarity of most head and neck cancers, evolving insights into the key molecular drivers are impacting on clinical practice.
Though commonly grouped together on the basis of anatomical proximity, the generic terminology of ‘head and neck cancer’ (HNC) refers to a miscellany of clinically and molecularly diverse tumours that arise from more than 15 anatomical subsites and comprise several different histotypes. From a clinical perspective the term HNC is often used to refer to mucosal squamous cell carcinomas, but evidence of differences in patient outcomes according to anatomical subsite and aetiology, have highlighted the need for research in more uniform cohorts.
Despite the rarity, HNCs are generally successfully managed with combinations of surgery, radiotherapy and chemotherapy delivered by a multidisciplinary team. The complexities of the sensitive anatomical location, the toxicity of treatment, and the functional consequences of both the tumour and treatment mandates management of these tumours by an expert team, and there is evidence of better outcomes in larger more experienced centres.1
The identification of the molecular changes that characterise subsets in HNC has been useful to improve disease stratification and has impacted upon patient management.2-8 We will highlight selected key molecular insights in HNC.
The discovery of the causal role of the Human Papillomavirus (HPV) in the majority of oropharyngeal squamous cell carcinomas (OPSCCs) in many countries, including Australia, exemplifies how translational research has improved the clinical management of HNC.
The role of HPV has been clearly defined only in the oropharyngeal subsite, where overexpression of p16 is an established robust surrogate marker of the HPV-induced subset of OPSCC.3,9 Importantly, HPV-related OPSCC represents a distinct clinicopathological and molecular entity associated with a favourable prognosis regardless of treatment modality.3,10-14 A summary of clinicopathological differences between HPV-related and HPV-unrelated disease is provided in table 1.
From an epigenetic level to protein level, the clinical variances between HPV-related and HPV-unrelated disease are also reflected from a molecular perspective.15-21 Characteristic molecular differences include the lack of association with TP53 mutations or major chromosomal abnormalities in HPV-related disease,17,18,22 while the majority of non-HPV related head and neck squamous cell carcinomas (HNSCC) harbour TP53 mutations,19,20,23 and are known to demonstrate field cancerisation effect.19-21,23 A recent publication using the data from the Cancer Genome Atlas HNSCC Working Group identified the poor prognostic effect of 3p deletions in both HPV-related and HPV-unrelated disease.22 However, for HPV-unrelated disease with simultaneous TP53 mutation and 3p deletion, the additional presence of mir-548k miRNA and MUC5B gene mutations identified further subgroups with even worse survival.
Molecular dissimilarities between diseases may also serve as differential neo-antigenic stimuli for the host immune system,24-28 although HPV also employs mechanisms to directly facilitate immune evasion.28,29 Early clinical trials of antibodies targeting the programmed cell death-1 (PD-1) pathway have demonstrated efficacy in tumours characterised by genomic heterogeneity,30 of which HNSCC is one of the top most mutated tumours defined by whole exome sequencing.31 It is yet to be established whether the differences in immune responses observed for HPV-related and HPV-unrelated HNSCC impact on patient prognosis, or the effectiveness of immunotherapies.24,25,32
In the oropharynx, predominantly HPV subtype 16 contributes to carcinogenesis through the characteristic production of the E6 and E7 viral oncoproteins.33-35 E6 mediates the ubiquitinisation of p53, and E7 disrupts retinoblastoma protein (pRb) function. With the combined loss of both tumour suppressors, unrestricted cell cycling occurs and apoptosis is evaded. p16 overexpression detected by immunohistochemistry is an established surrogate marker for HPV-related OPSCC, and results secondary to loss of a negative feedback loop mediated through pRb.9 Thus, p16 overexpression is a surrogate marker of good prognosis that occurs only as a bystander effect of the HPV.
The definition of p16 ‘positivity’ or overexpression, is a key consideration to ensure its use as a robust surrogate marker for the presence of HPV. For OPSCC, p16 ‘positivity’ is commonly defined as strong intensity staining in more than 70% of cells, although less stringent criteria are still considered sufficient.3,9 For all other HNSCC, whether p16 immunohistochemistry is an adequate surrogate marker of HPV, and whether p16 overexpression by itself is prognostic, remains to be established.
In general, outside of the oropharyngeal subsite, HPV mediated disease is known to occur much less frequently than for OPSCC, although the exact incidence is hindered by diagnostic limitations of methodology chosen, the use of heterogeneous cohorts limiting insight into subsite specific differences, and the use of small cohorts with a small number of observed events.36-39 Caution should be used in interpreting the literature in light of the methodology used, as ultimately only methodology such as reverse-transcriptase PCR or in-situ hybridisation are able to detect transcriptionally active or integrated HPV to infer causality.40 Other techniques, including viral genotyping and serology may overestimate the prevalence of the virus through detection of viral DNA not contributing to carcinogenesis. p16 immunohistochemistry lacks specificity for HPV given that other somatic aberrations, such as gene amplification or mutation of Rb, can also alter p16 expression. Similarly, absence of p16 staining does not indicate the loss of the protein, but rather demonstrates a lack of overexpression, exemplified by the absence of staining of normal tissues.41
An additional noteworthy point is that p16 and HPV status should not be interpreted in isolation. The significance of p16 or HPV as a prognostic marker needs to be interpreted in the light of clinical information, given the abrogation of a favourable outcome for patients with a strong smoking history, T4 disease, or greater than N2a disease.3,42 Reports in cohorts of HPV-related OPSCC of late disseminating metastatic disease to locations unusual for HPV-unrelated disease, and of longer survival in patients with metastatic disease, further emphasises the different tumour biology observed between the two disease entities.43,44
While further research is required to understand the role of HPV in other HNSCC subsites, the ability to identify patients with OPSCC who have a favourable prognosis has led to interest in whether de-escalation of treatment can minimise toxicity without compromising efficacy. This is of particular relevance given the younger age of patients with HPV-related disease. Numerous trials are currently enrolling patients with HPV-related disease, including the Trans-Tasman Radiation Oncology Group (TROG) 12.01 study which investigates the benefit of weekly cetuximab and radiotherapy versus weekly cisplatin and radiotherapy, in patients with low risk HPV-related OPSCC.
Salivary gland tumours include a spectrum of rare but distinct cancers, including the more indolent adenoid cystic carcinomas and the highly aggressive salivary ductal carcinomas. Tumours can arise from the three major components of the organ – the ducts, the acini and the myoepithelial cells, with disease occurring most commonly in the parotid gland.5 The World Health Organisation classification of salivary gland tumours includes 24 malignant epithelial tumours, in addition to benign tumours associated with malignant counterparts.45 Treatment is generally limited to surgery and radiotherapy, with these tumours demonstrating a high propensity for metastasis and recurrence without proven effective chemotherapy options.
While adenoid cystic carcinomas can arise from any location, it most frequently occurs in the major salivary glands of the head and neck. A recurrent translocation, t(6:9)(q22-23;p23-24), has been commonly identified in adenoid cystic carcinomas regardless of the site of origin of disease, in both primary and recurrent disease.6,46-50 The most common fusion product occurs between MYB, an oncogene encoded for on chromosome 6, and NFIB, a nuclear factor encoded on chromosome 9. Disruption to the MYB locus been identified as a putative poor prognostic biomarker.50,51 MYB activation resulting from the fusion mediates carcinogenesis mainly through its role controlling transcriptional elongation, but is additionally involved in the maintenance of cellular proliferation, inhibition of cellular differentiation, apoptosis and cell adhesion.46,51 Due to variable breakpoints in MYB and NFIB, a large number of fusion transcript variants are expressed.46,50 Disruption of the MYB pathway has also been identified to occur through gene amplification, mutation and overexpression in up to 80% of all adenoid cystic carcinomas, indicating that it is likely to be a seminal event in carcinogenesis.6,48-50 A DNA fusion vaccination based immunotherapy targeting MYB has been developed and shows anti-tumour efficacy in mouse models.52 Other potentially targetable aberrations observed in adenoid cystic carcinomas include canonical activating PIK3CA mutations, FGFR activating mutations and alterations of the FGF-IGF-PI3K pathway in up to 30% tumours.6,48
Salivary ductal carcinomas are aggressive tumours that can arise from malignant transformation of pleomorphic adenomas or can occur de novo, and demonstrate both in-situ and invasive patterns.5 Of particular interest, salivary ductal carcinomas demonstrate features in common with other glandular carcinomas, like breast and prostate carcinomas, with these similarities lending insight into promising treatment options.53,54 A therapeutically exploitable feature of salivary ductal carcinomas, is its expression of hormone receptors (e.g. oestrogen, progesterone, androgen) and expression of transmembrane receptors (e.g. HER2, EGFR, c-kit).5,45,55,56 Compared to other salivary gland tumours, salivary ductal carcinomas may be defined by the presence of androgen receptor expression, which occurs in up to 40% of cases,55,57 with these tumours also known to overexpress HER2. Given the successful use of targeted therapies in prostate and breast carcinomas, similar treatment algorithms have been employed for salivary ductal carcinomas, with reports of the success of androgen deprivation therapy and of HER2 inhibition.56,58-60 A recent study also reports the identification of canonical PIK3CA mutations in salivary ductal carcinomas, which holds additional therapeutic promise.61 The rarity of salivary ductal carcinomas hinders the investigation of the efficacy of these therapies, although progressive molecular characterisation of the disease has assisted in the provision of promising therapeutic options, as demonstrated in the literature and in our experience (figure 1).
Although a relatively small anatomical region, the sinonasal cavities give rise to some of the most complex and histological diverse groups of tumours.7,45,62 Tumour types include intestinal-type adenocarcinomas, esthesioneuroblastomas (olfactory neuroblastoma) which only occur in the sinonasal subsite, sinonasal undifferentiated carcinomas, large and small cell neuroendocrine carcinomas and germ cell tumours. The clinical management of these diverse tumours is complex due to the proximity to the orbit and brain, with frequent neural involvement and resultant significant functional and aesthetic challenges for management with surgery and radiotherapy.
As the name implies, intestinal-type adenocarcinoma has a histological resemblance to the intestines, and has a predilection for the ethmoid sinus.7,45 Wood dust and leather dust exposure are recognised risk factors for the development of disease.63 Several sub-classifications of intestinal-type adenocarcinoma exist, but most commonly refer to the colonic, papillary, solid and mixed subtypes, all of which demonstrate patterns of CK7, CK20, CDX2 and MUC staining.64 Given the morphological similarities to intestinal tumours, it has been of interest to determine whether the molecular changes in intestinal-type adenocarcinoma are similar. Alterations of DCC are observed similar to intestinal malignancies, but in contrast, intestinal-type adenocarcinomas express an intact Wnt signalling pathway (APC and β-catenin),65 demonstrate lower frequency mutations in KRAS and BRAF and have intact mismatch repair gene function.66 Similar to other HNCs, TP53 and CDKN2A are frequently disrupted in intestinal-type adenocarcinomas, but EGFR overexpression is less common.20,65,67 In the advanced/metastatic setting, although not proven, these tumours are reported to demonstrate response to 5-fluorouracil and platinum-based chemotherapy regimens which are effective for both colonic and head and neck carcinomas.7
NUT midline carcinoma is recently described tumour characterised most commonly by a t(15:19) translocation. A fusion oncogene between NUT (nuclear protein in testis) and BRD4 (bromodomain-containing protein 4) results, although other fusion products between NUT and other bromodomain and extra-terminal domain associated genes have been detected.68-71 The fusion product is observed to inhibit squamous differentiation while maintaining cellular proliferation, and is also known to activate histone acetyl-transferase which indirectly, but paradoxically decreases overall acetylation levels.70 This genetically defined, very rare, poorly-differentiated variant of squamous cell carcinomas arises in midline anatomical regions, and occurs in the head and neck region second most frequently to the thorax, with a possible preference for the sinonasal subsite.72 These tumours occur in younger aged patients and confer a dismal median overall survival of 6.7 months.73 Though rare, NUT midline carcinomas are of clinical interest due to the therapeutic promise of bromodomain inhibitors and histone deacetylase inhibitors, partnered with the development of a robust immunohistochemical antibody to detect the NUT protein.70,72
Beyond HPV status, no validated biomarkers are known to direct therapeutic decisions even for the more common HNSCC. However, the emerging understanding of the molecular features of rare head and neck cancers, many of which do not respond well to current systemic therapies, is likely to lead to the development of more effective therapies.
We would like to thank Eddie Lau for the provision of the images.