Squamous Cell Carcinoma of Head and Neck

Abstract

Squamous cell carcinoma of head and neck (SCCHN) is the most common cancer arising in the head and neck region. Smoking and heavy alcohol drinking are still the well-established causes of most cases worldwide; however, human papillomavirus (HPV) infection is the concerning cause in the Western world. The different pathogenesis, pathophysiology, and prognosis between HPV-driven and non-HPV SCCHN would lead to the different treatment approaches. Breakthroughs in radiation techniques, better organ-preserving surgical strategies, and multidisciplinary management modalities are the major reasons for the curability rate among patients with early and locally advanced SCCHN. Unfortunately, among patients with advanced, recurrent, or metastatic diseases, the treatment remains an area of need. Such patients usually die within a few years. The immune checkpoint inhibitors have been shown to provide astonishingly better survival, but only among a small and not definitely known proportion of patients. Investigating the more specific biomarkers predicting the treatment response and novel therapeutic options is warranted. In this review, we highlight the latest advances in pathophysiology, treatment, and the future direction of researches.

1. Introduction

Squamous cell carcinoma of head and neck (SCCHN) is cancer arising from the squamous epithelium in the oral cavity, pharynx, and larynx. It is the most common cancer that develops in the head and neck. According to the Global Cancer Statistics 2020, approximately 750,000 new cases were estimated and nearly 360,000 cases died annually. The lip and oral cavity are the most common sites, while, in respective order, larynx, oropharynx, and hypopharynx are less common sites [1]. Men are significantly more likely to develop SCCHN than women with an incidence ratio around 2:1 to 4:1 [2]. The average age of diagnosis is 50–70 years [3]. Globally, the incidence of HNSCC has increased by 36.5% over the past decade [4]. The prevalence of SCCHN varies across regions of the world and has been presumably correlated with tobacco use, excessive alcohol consumption, or both. There has been a significant decline in tobacco uses in Western countries during the last few decades, which has led to a sharp decline in smoking-related SCCHN [5]. On the contrary, there has been a significant increase in global epidemics of human papillomavirus (HPV)-associated SCCHN [6, 7, 8]. Around one-eighth of the incident cases of SCCHN comprise oropharyngeal squamous cell carcinoma (OPSCC), with HPV being an important associated risk factor for its development. HPV infection would be implicated with a small number of other SCCHN subsites. Data in this regard are quite inconclusive, presumably as a result of insufficient details on anatomical tumor localization and different HPV detection methods [9]. SCCHN of the oral cavity, hypopharynx, and larynx are associated with smoking and are categorized into HPV-negative SCCHN. No screening strategy has proved to be effective, unfortunately. Vigilant and careful physical examination of the population at risk remains the effective approach for early detection [3, 4]. With the exception of early-stage carcinomas of lip and oral cavity which surgery is the main curative treatment (with radiotherapy or chemo-radiotherapy as the adjuvant treatment depending on disease stage and clinical risk or laryngeal cancers that are amenable to either surgery or radiotherapy/chemo-radiotherapy), the majority of patients with SCCHN need multi-modality approaches. In this review, the latest advances in the pathophysiology of both HPV and non-HPV-driven SCCHN and their impact on the management will be elucidated. Perspectives on future directions will be provided as well.

2. Epidemiology and pathophysiology of HNSCC

2.1 HPV-associated SCCHN

2.1.1 Pathophysiology

Human papillomavirus (HPV) has been recognized as the major cause of oropharyngeal squamous cell carcinoma (OPSCC). HPV is sexually-transmitted. Early sexual experience as well as a high number of sexual partners, especially oral sex partners, and previous genital warts are the risks for HPV-associated (HPV(+)) OPSCC. There is a higher prevalence of HPV(+) OPSCC in men compared with women, and white populations compared with Asians and black populations [10, 11, 12]. The prevalence varies from less than 10% of all OPSCC cases in developing countries to 60–70% in the United States [13, 14, 15]. These variabilities are supposed to be, at least in part, due to the difference in sensitivity and specificity of the HPV detection assays, characteristics of the study cohorts, and their confounded risk factors especially tobacco uses and alcohol consumptions in the study population [16]. HPV infects the stratified squamous epithelia, both cutaneous and mucosal including the skin of hands and feet, as well as the anogenital tract, mouth, throat, and respiratory tract. Tonsillar crypt cells, similar to uterine cervical squamocolumnar junction cells, are arranged in a discontinuous single-layered epithelium that is more susceptible to cellular transformation than cells within other parts of the head and neck region [17, 18]. These perplexed invaginated crypts are naturally designed to entrap bacteria and foreign materials, driving the expression of programmed cell death-1 ligand-1 (PD-L1) [13]. The PD- L1 is responsible for immune evasion by binding programmed death-1 (PD-1) receptors expressed by the immune cells; therefore, the PD-L1 overexpression in tonsils promotes persistent HPV infection allowing carcinogenesis [14]. Moreover, in tonsillar crypts, the establishment of a biofilm composed of bacterial microcolonies encased in a glycocalyx matrix contributes to the HPV capability to escape the immune system [15]. The HPV family composes of circular, double-stranded DNA viruses of 8000 base pairs encoding proteins involved in viral replication (E1 and E2/E4) and assembly (L1 and L2), as well as accessory proteins (E5, E6, and E7). High-risk HPV types, including HPV16, HPV18, HPV31, HPV33, HPV35, HPV39, HPV45, HPV51, HPV52, HPV56, HPV58, HPV59, and HPV68, can induce malignant transformation of the infected oropharyngeal epithelium [19] by disrupting cell-cycle check points through E6- and E7-mediated degradation of p53 and Rb proteins, respectively [20]. Furthermore, the integration of the viral genome leads to E6/E7 expression, followed by disruption of the E2 coding region and dysregulation of E6/E7 themselves. This enables HPV to create the condition of persistent infections and replications. While the infected epithelial cells are gradually differentiated, the viral proteins are synthesized; however, no viral particle is actually produced. This non-productive infection by HPV is a key for cancer formation [21]. During carcinogenesis, the HPV E6 and E7 oncoproteins decrease the levels of p53 and functional Rb by post-translational regulation, resulting in aberrant overexpression of the cell-cycle protein p16 [22], which can be detected by immunohistochemistry (IHC) [23], rationalizing its use as a surrogate marker for the emergence of high-risk HPV-inducing cancer transformation rather than the infection as detected by an HPV DNA assay. Nevertheless, p16 expression is frequently lost in most cancers due to gene mutations, deletions, or promoter methylation [24], and p16 silencing has been reported to be as a result of tobacco exposure [25], indicating that this marker should be used carefully in the co-presence of risk factors. Interestingly, alcohol consumption is also correlated with p16 loss [26]. Thus, in non-OPSCC patients (mostly HPV negative) other risk factors can affect p16 status.

2.1.2 Prognosis

HPV status classifies OPSCC in particular into two distinct diseases (HPV(+) and HPV(−)) and it is still the only clinically biomarker that has been extensively validated from both retrospective and prospective studies. Several studies have shown that patients with HPV(+) OPSCC, as identified through PCR, in situ hybridization, or p16 immunohistochemistry (IHC) on tumor tissues, have shown that HPV(+) tumors are more likely to present with earlier T stage (T1-T2) and well-defined borders [27] or even occult primary tumors but have higher N stage (usually cystic and involving multi-level of the cervical lymph nodes) [28], and usually have poorer tumor differentiation, either non-keratinizing or basaloid histopathology [27]. Furthermore, the studies on correlation with clinical and imaging characteristics revealed that it usually had well-defined contours with exophytic growth (less necrotic and less ulcerated) and trended to uninvolved muscle tissue [28, 29]. Smoking also influenced clinical and imaging manifestations of HPV(+) OPSCC that were different from HPV(+) OPSCC in non-smoker [29]. The incidence of distant metastases seemed to be lower and if occurred, metastases usually developed later and with a very different pattern from patients with HPV(−) one. HPV(+) OPSCC had a 28% reduction in the risk of death and a 49% reduction in the risk of disease recurrence [30]. Second, primary tumor in patients with HPV(+) OPSCC is uncommon, and more likely to have a better overall survival rate compared to patients with HPV(−) ones [31]. The significantly better overall and disease-free survival compared with patients with HPV(−) have been confirmed by many prospective phase 3 clinical trials [30, 32, 33, 34, 35, 36, 37, 38, 39]. Even after adjustment for differences in favorable prognostic factors that were usually associated with HPV-positive patients (younger age, better performance status, fewer co-morbidities, less-or never-smoking) [32]. Consequently, numerous studies agreed on the fact that HPV(+) OPSCC led to a survival difference of up to 26.9% over HPV(−) one [40]. Smoking, on the other hand, has a negative impact on treatment efficiency and survival outcome. Smoking blunted the positive impact of HPV on survival as well. HPV(+) non-smoker had better prognosis compared with HPV(+) smoker, regardless of HPV detection methods (HPV DNA PCR, ISH, p16 IHC) used [41]. Since HPV(+) differs from and has a more favorable prognosis than HPV(−) OPSCC, the recently released Eighth Edition of the American Joint Committee on Cancer proposed the staging algorithm specifically for HPV(+) OPSCC, distinguishing it from HPV negative OPSCC. According to the new staging system, 92% of patients with HPV-positive OPSCC could be downstage, and up to 64% of patients were now staged as stage I disease [42]. Van Gysen et al. conducted a retrospective cohort study to validate the new AJCC staging system for HPV(+) OPSCC and demonstrated that this system better discriminated between stage I and stage II HPV(+) OPSCC with respect to OS compared with the seventh edition staging system. However, further investigation was required for stage III or IV patients [43].

2.1.3 HPV detection strategies

HPV testing in OSCC varies in detection targets including HPV DNA polymerase chain reaction (PCR) for E6/E7 viral oncogenes, HPV E6/E7 mRNA detection quantitative reverse transcription-PCR (qRT-PCR), DNA in situ hybridization (ISH), RNA ISH, and p16 immunohistochemical staining (IHC) as a surrogate marker for HPV status [44]. There is still no consensus regarding which method should be the gold standard. The PCR techniques are complex and have low specificity because they cannot distinguish between HPV acting as an oncogenic driver and transcriptionally silent virus from non-pathogenic or contaminated one [45]. Even though the detection of viral E6/E7 mRNA by RT-PCR is widely accepted as the gold standard due to its higher sensitivity, more specificity in detection of the oncogenic viral mRNA/DNA target, and its feasibility on formalin-fixed, paraffin-embedded tissue (FFPE) block [46], there are serious limitations including its time-consuming and its decreasing sensitivity depending on the quality of samples. The DNA ISH is another molecular test with high specificity, which enables direct detection of the presence of oncogenic HPV in matched topographical relationship with pathological samples, ascertaining that the pathogenic viral DNA emerges from tumor cells and not surrounding normal tissues. Furthermore, ISH has the advantage of being feasible in both FFPE tissues; however, this technique still lacks sufficient clinical validation. The DNA ISH is currently not used in routine practice [47]. Lydiatt et al. suggested that the test should be simple, inexpensive, and reproducible [42]. Not only the p16 IHC is the well-established marker as a surrogate for the presence of the oncogenic HPV, but also an independent positive prognostic factor among patients with OPSCC [48]. In particular, the cutoff point for p16 overexpression is diffuse (more than 75%) tumor expression localized to tumor cell nuclei and cytoplasm, with at least moderate (2+ or 3+) staining intensity. Either cytoplasmic-only staining or staining on other non-oropharyngeal sites is considered non-specific [49].

2.2 HPV-negative (−) SCCHN

2.2.1 Pathophysiology

Tobacco smoking is classified as being carcinogenic to humans by the International Agency for Research on Cancer (IARC) [50] and there are strong supporting evidences of association with head and neck cancer [51]. The HPV(−) SCCHN is characteristically seen in patients with a history of heavy tobacco and alcohol use [52]. Interestingly, HPV(−) SCCHN can occur in relatively young patients without a history of tobacco use, and the incidence has been rising with unclarified etiology [53]. Different brands of cigarettes and cigars use varying formulations of blended tobaccos affecting the various amount of nicotine and carcinogen content, thus impacting the toxicity of the smoke [54]. Smokeless tobacco products such as chewing tobacco, snuff, gutkha, and betel quid have been associated with oral cavity cancer for several decades. Some of the added components to these products can promote or modulate the absorption rate of nicotine. Areca nut commonly used in chewing tobacco products in India and Southeast Asia contains the alkaloid drugs arecoline, muscarine, and pilocarpine, which cause cholinergic mood-relaxing effects promoting addictive effect [18]. The pH of a tobacco product affects the rate of nicotine absorption. Buffering substances such as slacked lime or calcium carbonate that usually consist in gutkha and betel quid result in elevating the pH; therefore, they enhance the rate of nicotine absorption [55]. Both tobacco and especially tobacco fume are loaded with polycyclic aromatic hydrocarbons (PAH) and nitrosamines, which are the established human carcinogens and are strongly related to the risk of SCCHN. Their reactive metabolites, if not appropriately detoxified and excreted, lead to DNA damage, characteristically by promoting the generation of bulky DNA adducts. If such damages are not accurately and promptly repaired, the permanent damages as demonstrated by mutations, deletions, and amplifications can emerge. The TP53 and CDKN2A that encode p53 and p16^INK4A, respectively, are the most common genomic abnormalities in HPV(−) SCCHN. Furthermore, signaling pathways such as phosphoinositide 3-kinase (PI3K)–AKT–mTOR and MAPK pathway genes are mutagenically over-activated. Such genetic abnormalities are associated with the onset, progression, and adverse prognosis of HPV-negative SCCHN [56]. The fact that TP53 mutations in SCCHN have distinct signatures from TP53 mutations with aging and ultraviolet light exposure indicates the truly different mechanisms of genetic damage [57]. In contrast to HPV(+) OPSCC that the TP53 gene is rarely altered, since p53 in HPV(+) OPSCC is eliminated by the action of E6 [56]. Excessive alcohol drinking is known to synergize with tobacco use to promote the carcinogenesis of HPV(−) SCCHN, in particular. It is postulated that alcohol would function as a solvent for carcinogens to enhance the exposure of epithelial cells to these substances [58]. Alcohol is metabolized to acetaldehyde inducing the DNA adducts [59].

2.2.2 Prognosis

One of the unique characteristics of tobacco smoking-induced squamous cell carcinoma is the development of synchronous and/or metachronous second primary tumors arising at an exceptionally high rate after the diagnosis of an initial primary tumor and can be occurred along the aero-digestive tract including the head and neck, esophagus, and lungs [60]. The concept of “field cancerization” suggests that the carcinogens trigger the genomic damages along with the large anatomical fields as far as the tobacco fume can approach [61]. There are evidences supporting that the size of the damaged anatomical field may increase with patient age, as well [62]. A pooled analysis within the International Head and Neck Epidemiology Consortium demonstrated that besides older ages at diagnosis and advanced tumor staging that were the consistent adverse prognostic factors of survival, cigarette smoking was an independent unfavorable prognostic factor for overall survival (OS) among patients with OPSCC. Intense smoking (as defined as >20 cigarettes/day) was also an independent prognostic factor for OS among patients with oral cavity cancer. On the contrary, among patients with laryngeal cancer, low educational level was rather a deleterious prognostic factor for OS; moreover, the intensity of alcohol drinking was the prognostic factor for both of the OS and head-and-neck-cancer-specific survival [63].

Apart from the carcinogenic effects of tobacco, abundant evidence have demonstrated adverse effects of tobacco on various treatment-related outcomes in patients with SCCHN, including radiotherapy efficacy, surgical outcomes, and wound complications. Chen et al. reported that continued smoking during receiving radiotherapy was associated with inferior 5-year overall survival, locoregional control, and disease-free survival [64]. In addition, patients undergoing surgical treatment for SCCHN who were current or former smokers were more likely to have various postoperative complications including wounds and infections resulting in a longer length of stay than never smokers [65]. Marin et al. conducted a prospective study on the effect of smoking on wound healing in smokers undergoing free tissue transfer and concluded that those who were current smokers as indicated by high serum cotinine concentration, a metabolite of nicotine, would predict an increased risk of wound complication [66].

3. Management of newly diagnosed, early, and locally advanced SCCHN

3.1 Primary surgical approaches in SCCHN

Primary surgical treatment is usually considered as the standard of care for oral-cavity cancers, carcinoma of the true glottis, and sino-nasal cancers. The oral cavity cancers (OCC) are easily accessible trans-orally. Both the true glottis larynx (vocal folds) and paranasal sinus are void of lymphatics. Therefore, the primary surgical approach is the treatment of choice for such cancers and is associated with high cure rates with acceptable morbidity. In the case of oral cavity cancers, it has been recognized that the prognosis of OCC worsens when the tumor is thicker. The revised AJCC 8th edition cancer staging manual emphasizes the significance of the depth of invasion (DOI). For every 5-mm increase in DOI (microscopically measured from the level of the basement membrane of the closest adjacent normal mucosa), both cT and pT categories are right now categorized to one level increasing according to the following: ≤ 5 mm, >5 mm but ≤10 mm, and > 10 mm [42]; therefore, the DOI should be taken into account prior to consideration of surgery with curative intent. Besides oral, laryngeal, and sino-nasal cancers, the primary surgical approach is more suitable for stage I-II diseases. Oropharyngeal cancers (OPC) may be managed by either primary surgery or radiotherapy with the comparable outcomes [67, 68].

Among patients with laryngeal cancer, definitive chemoradiotherapy (CRT) is the time-honored preferred treatment, especially for the reason of organ preservation. Surgery is usually reserved as salvage treatment. According to the landmark RTOG 91–11, local control and larynx-preservation rates were significantly higher with concomitant cisplatin and RT (definitive CCRT); however, the OS curves separated after 4 years, favoring induction chemotherapy (IC) followed by RT alone. Although CCRT had the lowest rates of cancer-related deaths, it also had the highest rate of non-cancer deaths compared with IC, presumably attributed to late swallowing dysfunction and aspiration [69]. Patel et al. conducted a real-world retrospective analysis of treatment outcomes between surgical (partial laryngectomy (PL) or total laryngectomy (TL) with or without adjuvant radiotherapy with or without concomitant chemotherapy) and non-surgical (definitive CCRT)) management among 8703 patients with non-T1, stage III/IV, glottic, and supraglottic cancer from the National Cancer Data Base during 2003–2011. They demonstrated that non-surgical definitive CRT resulted in equivalent survival outcomes among patients with non-T4, low nodal burden (N0-N1) disease, and even better than surgical management (with or without adjuvant treatment) among patients with non-T4, high nodal burden (N2-N3) disease. However, patients with T4 disease, TL followed by adjuvant RT, or CCRT had superior survival outcomes compared with non-surgical definitive CCRT [70]. This study did not analyze the outcomes of IC, which was commonly used in the current practice of organ preservation. Also noted was the fact that not all T4 diseases were amenable to curative surgery especially those with T4b and those with supraglottic primary.

Since head and neck cancers have a propensity to metastasize to cervical lymph nodes, lymphatic drainage of the primary site and the risk of occult metastatic spread guide decisions regarding additional therapy. Radical neck dissection (RND) is to remove neck nodes of levels I–V, accessory nerve, internal jugular vein, and sternomastoid muscle, and this procedure is compulsory for most of the diseases with obviously clinical lymph node metastasis. If preservation of one or more of the accessory nerve, internal jugular vein, or sternomastoid muscle is possible, the procedure is called modified radical neck dissection (MRND) (types I, II, III, respectively). If removal of one or more additional lymphatic and/or non-lymphatic structures(s) relative to a RND, for example, level VII, retropharyngeal lymph nodes, hypoglossal nerve is necessary to eradicate all of the suspicious sites, it is called extended radical neck dissection (ERND). If the reservation of one or more levels of lymph nodes is possible, the procedure is called selective node dissection (SND). SND is usually indicated for clinical node-negative disease with a higher chance of occult lymphatic metastasis, that is, when a primary tumor deeply invades (ipsilateral removal of the most possible lymph node groups with tumor spread) or when primary tumor at a site with rich lymphatics crosses the midline (contralateral removal) [71].

More recently, transoral robotic surgery (TORS) and transoral laser microsurgery (TLM) have emerged as the primary surgical modalities in the management of oropharyngeal carcinoma (OPSCC), in particular. As compared with standard open surgery, these techniques have been shown to reduce hospital stays and feeding tube requirements at 1 year [72]. In OPSCC, TORS is suitable for the management of small-sized tumors (cT1–T2) [73]. Even though, about 17–31% of patients who underwent TORS eventually needed to receive salvage radiotherapy and/or chemotherapy. Thus, about a quarter of these patients eventually needed multi-modality therapy.

3.2 Multi-modality approach in locally advanced SCCHN

More than 60% of patients present with stage III or IV disease, which is characterized by large tumors with extensive local invasion (clinical T3–4) and metastases to regional nodes, or both. Locally advanced disease is associated with a high risk of local recurrence (15 to 40%) and distant metastasis, with a grave prognosis (5-year OS, <50%) [74]. A dilemma usually exists between whether it would take to completely remove a bulky or extensively infiltrated tumor versus the impact such a resection would have on the patient’s quality of life and self-image. When surgical resection is not feasible due to in-operable diseases, pre-existing serious co-morbidities, or on the condition that curative surgery would lead to unacceptable long-term functional and or cosmetic outcomes, the definitive concurrent chemo-radiotherapy (CCRT) is the most suitable choice.

The recent American Joint Commission on Cancer (AJCC) revised the T-staging classifications of head and neck cancers [42]. The T4 disease is subcategorized into T4a and T4b designations. Most patients with T4b tumors are generally defined as definitely unresectable, whereas the T4a tumors are potentially resectable. However, they need more devoted surgical techniques, if curative intent is primarily considered. The T4b is characterized by one of the following features: 1) vascular encasement and invasion, 2) prevertebral space invasion, and 3) invasion of mediastinal structures. The locally advanced disease generally harbors high-risk adverse features. The independent pathologic risk factors for recurrence include T3 or T4 tumors, multiple involved nodes, lymphovascular invasion, anatomic location, low neck location of lymph node involvement, extra-nodal extension (ENE), perineural invasion, and close/positive margins [75, 76, 77]. The more number of risk factors, the more the chance such a patient has increased risk of recurrence [78]. In addition, poorly differentiated tumors and, especially for oral cavity cancers, the depth of invasion (DOI) would be at risk for recurrence [79]. Consequently, all patients with stage III-IV diseases need postoperative radiotherapy (PORT) as a part of treatment with curative intent. Two large randomized controlled trials, the EORTC 22931 [80] and the RTOG 9501 [81], demonstrated a statistically significant loco-regional control (LRC) benefit with the addition of cisplatin of 100 mg/m² to PORT, given every 3 weeks for three cycles for patients with “high-risk” features. Both studies reported that serious (grade ≥ 3) toxicity occurred more than doubled with the addition of chemotherapy. However, the rate of distant metastasis did not differ between arms. When the results of the EORTC and RTOG trials were analyzed in a meta-analysis, a statistically significant OS improvement was determined in favor of postoperative adjuvant chemo-radiotherapy, especially for patients with ENE and close/positive surgical margins [82]. An OS benefit of an alternative schedule of cisplatin administration of weekly 50 mg/m² added to PORT was also demonstrated for stage III/IV SCCHN with ENE; however, the LRC was not proven [83]. The Japanese group (JCOG 1008) trial also showed the non-inferiority of the alternative schedule weekly cisplatin 40 mg/m² to three-weekly cisplatin in high-risk patients with microscopically positive margin and/or ENE [84]. In general, PORT should be commenced within 6 weeks after surgery [85]. A meta-analysis by Matuschek et al. showed no improvement in OS, PFS, or LRC with postoperative accelerated fractionation compared with conventional fractionation [86]; however, among patients with stage III/IV and a prolonged interval from surgery to RT, the accelerated fractionation seemed to result in better DFS or LRC benefit [87].

The recent MACH-NC meta-analysis revealed that chemotherapy when used concomitant with radiotherapy (RT) either as definitive treatment or following surgery in case of pathological adverse features led to an absolute benefit of 6.5% at 5 years and 3.6% at 10 years. Concomitant chemotherapy showed a significant effect on loco-regional failure (LRF), but not on distant failure (DF). For event-free survival (EFS), the meta-analysis demonstrated the highest effect for poly-chemotherapy with platin salt (HR = 0.74 [0.67; 0.82]) and the lowest for mono-chemotherapy without platin salt (0.86 [0.80, 0.93]). However, for OS, the interaction was borderline significant. Notable, the effect of chemotherapy on survival was decreasing with increasing age, probably due to the fact that more non-cancer deaths were reported more among patients 70 or over. The induction chemotherapy (IC), on the other hand, did not improve the LRF rate, OS, and EFS, even though it could decrease DF. There was no significant variation of the effect on OS according to the type of induction chemotherapy. The docetaxel/cisplatin/5-FU (TPF) regimen seemed to perform best in terms of EFS, but not OS compared with other regimens. Neither OS nor EFS was improved with adjuvant chemotherapy (AC). Moreover, AC led to a significant decrease in LRF and DF, as well as the deleterious effect on 120-day mortality [88]. Gau et al. suggested that IC (TPF regimen, in particular) might have less interest in OPSCC, especially among those associated with HPV. The IC would be most preferential if functional laryngeal preservation is the priority [89]. Petit et al. performed an individual patient data network meta-analysis to compare various kinds of multi-modality management with loco-regional therapy alone (surgery, RT, or both) in patients with locally advanced SCCHN. Hyper-fractionated RT with CCRT (HFCRT) was ranked as the best treatment for OS, whereas IC with TPF regimen followed by locoregional therapy (IC_Tax-PF-LRT), accelerated RT with CCRT (ACRT), IC with TPF regimen followed by concomitant chemo-radiotherapy (IC_TaxPF-CLRT) and CCRT with platinum-based chemotherapy (CLRT_P) were ranked less in consecutive order. Unfortunately, the treatment-related toxicities were not included in the analysis, even though HFCRT would be too toxic to be assigned in routine practice [90].

Among patient’s ineligible to cisplatin, either carboplatin/5-FU or carboplatin/paclitaxel are both the acceptable alternatives used in concomitant with RT [67].

3.3 De-escalation strategy for HPV-associated OPSCC

Since HPV(+) OPSCC is characteristically more radio-sensitive and chemo-sensitive than cancers caused by smoking and alcohol, the conventional paradigms like postoperative RT involving both tumor bed and neck with or without concomitant chemotherapy or definitive high-dose of RT in concomitant with high-dose cisplatin seem to be too toxic without incremental survival benefits. De-escalation of therapy has been proposed for this particular sub-group based on data demonstrating high OS and PFS [91]. De-escalation strategy includes minimally-invasive surgery, reduced dose and target volumes of adjuvant RT, and potential omission of chemotherapy.

Transoral robotic surgery (TORS) is a minimally invasive approach that reduces morbidity compared with traditional, open surgery most suitable for patients with resectable tonsil or base of tongue tumors when the adequate functional outcome can be preserved. The ORATOR2 study is an ongoing randomized trial investigating de-escalated definitive RT-based treatment in comparison with surgery with de-escalated adjuvant therapy. Both survival and swallowing quality of life will be evaluated [92].

ECOG-ACRIN 3311 was a randomized trial investigating reduced dose adjuvant RT for patients with intermediate postoperative risk factors. Patients with “low-risk” (AJCC 7th ed. pT1-T2, N0–1) disease with negative margins were observed, while patients with “intermediate risk” (close margins, < 1 mm of ENE, 2 to 4 involved lymph nodes, perineural invasion, or lymphovascular invasion) were randomized to postoperative RT of either 50 or 60 Gy. “High-risk” (positive margins, > 1 mm of ENE, or ≥ 5 involved nodes) patients received standard concurrent RT with cisplatin. At a median follow-up of 35.1 months, the outcomes were comparable or even better than historical results, and the 3-year PFS rates were 96.9%, 94.9%, 93.5%, and 90.7% for the low-risk, 50 Gy, 60 Gy, and high-risk arms, respectively [93]. The PATHOS trial is an ongoing randomized trial investigating a reduction in adjuvant RT and chemotherapy. Patients with low-risk disease are observed postoperatively, patients with intermediate-risk factors are randomized to 50 Gy vs. 60 Gy, and patients with high-risk features are randomized to 60 Gy alone or 60 Gy with cisplatin [94].

Omission of chemotherapy was studied in the randomized phase 2 NRG-HN002 trial. Patients with p16-positive, AJCC 7th ed. T1-T2 N1-N2b or T3 N0-N2b OPSCC with 10 or fewer pack-years of smoking were randomized to the “standard-of-care,” 60 Gy of RT in 6 weeks with cisplatin vs. 60 Gy of RT in 5 weeks without any systemic therapy. The 2-year PFS rate was 90.5% for cisplatin/RT vs. 87.6% for RT alone. Unfortunately, RT alone arm did not meet the prespecified 95% confidence interval threshold for PFS superiority to 85%; moreover, there was no difference in swallowing quality of life between the 2 arms [95].

Cetuximab, an epidermal growth factor receptor (EGFR) chimeric IgG1 monoclonal antibody, was shown to improve survival when used concurrently with RT in patients with locally advanced SCCHN as compared with RT alone leading to its popularity for use concurrently with RT as an alternative to cisplatin, especially among patients ineligible to cisplatin due to impaired renal function. Subgroup analysis showed superior survival benefit among patients with p16+. Cetuximab also had less nephrogenic (usually mild hypomagnesemia and hypokalemia) than cisplatin. Therefore, cetuximab had been investigated as a de-escalation strategy; however, three randomized phases 3 trials RTOG 1016 [96], De-ESCALaTE HPV [97], and TROG 12.01 [98] demonstrated the statistically significant detriment in both PFS and OS for patients treated with cetuximab/RT. In addition, toxicities were not overall reduced with cetuximab/RT compared with cisplatin/RT.

In conclusion, while results from well-designed clinical trials are gathering, the current data are still insufficient to recommend any de-intensified treatment strategy for HPV-associated OPSCC to be implemented in routine clinical practice.

3.4 Post-treatment response assessment and the role of salvage surgery

Since more than two-thirds of SCCHN patients present with locally advanced disease, usually not amenable to curative surgery, definitive chemo-radiotherapy is, therefore, the treatment of choice. Recurrence rates as high as 60% within 2 years of treatment have been reported with 20–30% of patients developing the distant metastatic disease [99]. Such patients mandate accurate staging and response assessment to guide appropriate management. Combined fluorine-18 fluorodeoxyglucose (F¹⁸-DG) positron emission tomography-computed tomography (PET-CT) is capable of both determining extent of locoregional and distant disease at initial staging and assessment post-treatment response [100]. Data from randomized controlled trials have shown that PET-CT is a precise and cost-effective technique for assessing response and would lead to spare 80% of patients from unnecessary salvage neck dissection [101]. Nevertheless, post-treatment-related changes in the neck can make assessment problematic in some cases, due to evidence suggesting that HPV(+) OPSCC may behave differently to HPV(−) SCCHN [102]. The semi-quantitative methods of treatment response assessment using standardized uptake value (SUV) alone have not been shown to be accurate at assisting a physician’s decision. The more reproducible qualitative interpretative criteria have been developed and validated in clinical trials to determine their value in predicting residual loco-regional disease and would help limit the number of problematically equivocal scan results [103, 104, 105]. Zhong et al. conducted a clinical trial to compare the accuracy of four different qualitative interpretative criteria (NI-RADS, Porceddu, Hopkins, Deauville) for predicting loco-regional control and progression-free survival (PFS) in patients with SCCHN treated with curative-intent non-surgical treatment who underwent baseline and response assessment FDG PET-CT and demonstrated that all four criteria had similar diagnostic performance characteristics; however, Porceddu and Deauville seemed to provide the best trade-off limiting indeterminate scores while maintaining a high negative predictive value (NPV) [106].

From the practical point of view, the role of PET-CT in determining the necessity of elective neck dissection (END) after definitive chemoradiotherapy (CCRT) was evaluated in phase 3 non-inferiority trial where 564 SCCHN patients with advanced nodal stages were randomly assigned to receive either elective neck dissection (END) within 4–8 weeks after CCRT or PET-CT scans at 12 weeks. The 2-year OS rate in PET-CT surveillance group was not inferior to the planned-surgery group (84.9 and 81.5%, respectively). PET-CT-guided surveillance also led to obviously fewer neck dissections than planned neck dissection surgery (54 and 221 cases, respectively). The PET-CT-guided surveillance would also be cost-effective [107]. Therefore, PET-CT-guided surveillance after non-surgical definitive treatment should be considered. Even though there is still a lack of the widely accepted consensus on the extent of neck surgery for residual neck disease after definitive non-surgical treatment, most of the experts suggest repeating PET-CT (preferentially), CT or MRI, or ultrasonography of the neck, 10–12 weeks after the conclusion of definitive CCRT for patients with clinical N2 and N3 diseases and consider observation if no evidence of viable residual disease exists; either observation or performing selective node dissection (SND), if post-treatment imaging studies interpret inconclusive, and performing SND or comprehensive neck dissection if residual neck disease is definitely revealed. In case of isolated neck node recurrence (without residual disease in the primary site and distant metastasis), comprehensive neck dissection is advocated, if feasible [67, 68].

4. Management of recurrent and/or metastatic HNSCC

As previously mentioned, most patients with recurrent disease are rarely suitable for salvage local therapy since such patients usually harbor extensive and intrinsically chemo- and radio-resistant diseases not amenable to salvage surgery. Furthermore, innovative radiation techniques are beneficial only in highly selective cases with very localized and low-tumor burden diseases. Among patients with metastatic disease, the prognosis is poor with a median OS of less than 1 year [108].

4.1 Cytotoxic chemotherapy

The platinum doublet therapy has been demonstrated to improve overall response rates over single-agent therapy, especially platinum in combination with either 5-fluorouracil (5-FU) or paclitaxel. Both of these combinations were shown equivalent in terms of both response rate (RR) (27% vs. 26%) and OS (8.7 vs. 8.1 months) [109]. The triplet of paclitaxel, ifosfamide, and either cisplatin or carboplatin was shown to have a much higher response rate (58 and 59%, respectively) [110, 111]. The TPF (docetaxel, cisplatin, and 5-FU) regimen was also a proven potent triplet with an RR of 44% [112]. Unfortunately, the triplets were associated with an unacceptably too high incidence of febrile neutropenia, despite the use of primary G-CSF prophylaxis. Therefore, the triplet is not recommended in the palliative setting. Since none of these platinum-based combination regimens has been demonstrated an OS benefit over single-agent methotrexate, the platinum doublet is recommended exclusively for symptomatic and fit patients whose immediate symptomatic relief is the primary aim of management.

A substantial number of conventional single agents have been investigated. The most commonly used agents are methotrexate, cisplatin, 5-FU, and bleomyin. These agents produced a modest response rate of 15 to 30% and a very short duration of response of around 3–5 months. Pemetrexed, vinorelbine, irinotecan, a fluoropyrimidine analog (capecitabine and S-1), and a taxane (paclitaxel and docetaxel) were among the newer agents. However, the taxanes are among the most potent agents to be proven in various kinds of tumor characteristics, with RR around 20 to 43% [113].

4.2 Anti-EGFR antibodies

Overexpression of epidermal growth factor receptor (EGFR) is commonly observed in SCCHN. Since the survival benefit of cetuximab, a chimeric mouse/human monoclonal anti-EGFR antibody with concurrent RT had been investigated in locally advanced setting [114], and cetuximab was then investigated in recurrent or metastatic setting in the EXTREME trial [115]. OS was improved with the addition of cetuximab to cisplatin/5-FU (also known as the EXTREME regimen) (10.1 vs. 7.4 months; HR 0.80; p = 0.04). In addition, the EXTREME regimen also prolonged both PFS from 3.3 to 5.6 months and RR from 20–36% as compared with the cisplatin/5-FU combination. Another platinum doublet in combination with cetuximab, the TPEx (docetaxel, cisplatin, and cetuximab) regimen failed to demonstrate OS benefit over the EXTREME regimen in the phase 2 study, even though the overall toxicities were less than the EXTREME regimen. However, their OS was astonishingly longest among all randomized trials ever reported of 14.5 vs. 13.4 months in TPEx and EXTREME, respectively [116]. Notably, this study allowed an immune checkpoint inhibitor as a subsequent treatment upon progression. Panitumumab, a humanized monoclonal anti-EGFR antibody was also investigated in combination with cisplatin/5-FU in the SPECTRUM trial. Although panitumumab with cisplatin/5-FU led to significantly longer PFS compared with the platinum doublet alone, but it failed to show OS improvement as the primary endpoint (11.1 vs. 9 months; HR = 0.0873, p = 0.14) [117]. The fact that the SPECTRUM trial allowed crossover upon disease progression, while the EXTREME trial did not would be a reason for failure to demonstrate OS benefit.

4.3 Immune checkpoint inhibitors

The efficacies of various immunotherapy checkpoint inhibitors (ICI) were first assessed in the later-line setting of recurrent or metastatic SCCHN. In phase 1 KEYNOTE-012 trial, pembrolizumab resulted in a response rate of 18% and median OS of 8 months for all patients. Notably, 65% of those who responded were continuing to respond at the time of final analysis [118]. Pembrolizumab was then investigated in the phase 3 trial KEYNOTE-040 comparing with the standard of care (methotrexate, docetaxel, or cetuximab) and shown to have OS benefit among the intention-to-treat population (8.4 vs. 6.9 months; HR 0.80, p = 0.0161). The RR was 14.6% overall, and 17.3 and 26.6% in those with PDL-1+ (combined positive score, CPS >1%) and PDL-1+ (tumor positive score, TPS >50%), respectively. The OS was also remarkably longer among those with PDL-1+ (TPS > 50%), compared with the intent-to-treat population and those with PDL-1+ (CPS > 1%) (with a significant P-value for the interaction of 0.015) [119]. Nivolumab was assessed in the phase 3 CheckMate-141 trial [120], comparing nivolumab vs. a single-agent (methotrexate, docetaxel, or cetuximab) of an investigator’s choice. As reported in the 2-year update, nivolumab improved OS significantly vs. treatment of an investigator’s choice (7.7 vs. 5.1 months; HR 0.68); however, PFS was similar between treatment arms. Estimated OS rates with nivolumab were consistent regardless of the PD-L1 expression (<1% vs. ≥1%). HPV status was not the predictive marker.

Pembrolizumab was then further investigated in the first-line setting. The KEYNOTE-048 trial randomly assigned platinum-sensitive patients to receive either the EXTREME regimen, or a single-agent pembrolizumab, or a combination of the platinum (cisplatin or carboplatin)/5-fluorouracil doublet with pembrolizumab [121]. Owing to its sophisticated hierarchical data analysis using superiority or non-inferiority design according to the selected hypothesis, this trial must be interpreted cautiously. Although the PD-1 inhibitor did not provide an improvement in both PFS and RR in the overall population, monotherapy with pembrolizumab did improve OS in patients with a PD-L1+ (CPS ≥1) (12.3 vs. 10.3 months; HR 0.78, p = 0.0086). Furthermore, in the overall population, OS of single-agent pembrolizumab was determined to be non-inferior to the EXTREME regimen (11.6 vs. 10.7 months), while in patients with a PD-L1+ (CPS ≥1), pembrolizumab with platinum/chemotherapy was even superior to the EXTREME regimen (13.6 vs. 10.4 months; HR 0.77). Interestingly, the seemingly decreasing efficacy in analyses of populations with more patients having lower CPS was a concerning issue, apparently when shifting the cutoff threshold from CPS ≥20 to CPS ≥1 and then to the total populations. In the CPS <1 subgroup, there was neither OS advantage of pembrolizumab alone over EXTREME (7.9 vs. 11.3 months, HR 1.51) nor pembrolizumab/chemotherapy over EXTREME (11.3 vs. 10.7 months; HR 1.21). In the CPS 1–20 population, the benefit was confined only to the pembrolizumab/chemotherapy (12.7 vs. 9.9 months; HR 0.71) but not pembrolizumab monotherapy (10.8 vs. 10.1 months; HR 0.86). In the CPS ≥ 20 groups, pembrolizumab monotherapy yielded the longest survival, skeptically even better than pembrolizumab/chemotherapy. With debatably insufficient statistical power, the results from such post hoc analyses are subjected to be biased and should be speculated cautiously. The treatment-related side effects of the experimental agent were not significantly higher than EXTREME regimen. The immune-related adverse events (ir-AEs) were also as expected and manageable. Also noted, the KEYNOTE-048 excluded patients with progressive disease (PD) within six months of completion of curatively intended systemic treatment for loco-regionally advanced disease. The subgroup analysis revealed the lack of survival benefit attributable to pembrolizumab in patients presenting with local and/or regional recurrence only. Another issue of concern is the fact that the proportion of patients in the KEYNOTE-048 study who had progressive disease as the best response was indeed greater in the pembrolizumab alone group than in the EXTREME arm (41% versus 12% in the total population), irrespective of CPS explaining why PFS was not improved overall in every subgroup of CPS. Whether these results support the hyper-progression phenomenon in the pembrolizumab single-agent arm remains to be realized. This rate remains relatively consistent around 40% when anti-PD-1 agents are used alone in both the first- and second-line settings. Unexpected early disease progression seems to be unavoidable even when immunotherapy is used in combination with chemotherapy. Pembrolizumab and chemotherapy would likely act independently [122]. The provocative results from the updated data analysis on survival according to the second-line treatment from TPExtreme trial showed that the taxane-based TPEx regimen followed by an ICI in the second line provided longer OS (21.9 and 19.4 months in TPEx and EXTREME, respectively) [123]. Whether this sequential cetuximab-based and then an immune checkpoint inhibitor would be the best treatment option warrants further investigations. Those with symptomatic extensive loco-regional recurrent disease or with low PD-L1 expression would be more suitable for the platinum doublet (cisplatin/5-FU or cisplatin/docetaxel) with cetuximab as the first line. An ICI should be reserved as a second-line option in this scenario.

Other ICIs, including durvalumab, a PD-L1 inhibitor, and tremelimumab, a CTLA-4 inhibitor, has been also investigated in the second-line setting. Although the phase II HAWK trial showed promising activity of durvalumab, with a modest RR of 16.2% [124], the survival benefit of durvalumab with or without tremelimumab over chemotherapy in the overall population was not proven in the phase III EAGLE trial [125].

As revealed across the reported trials, the toxicity profile of the ICIs is significantly more favorable than conventional cytotoxic chemotherapy and is similar to what is seen with their use in other solid malignancies. They are generally well-tolerated with fatigue (20%) and nausea (9%) being the most common adverse events. Specifically, unique for these agents, the ir-AEs include thyroiditis, pneumonitis, hepatitis, colitis, nephritis, hypophysitis, myocarditis, myositis, neuritis, adrenal insufficiency, rash, and neurological toxicities. Immune-mediated thyroiditis is the most commonly seen (15%), whereas the others are less common (<5%). Most of these reactions are usually transient and minor (grades 1 and 2). Nonetheless, grade 3/4 can still occur in ~1–2% of the cases and may be fatal in some of them. Management depends on the type and severity of the ir-AEs. Thyroiditis is usually managed with thyroid hormone supplement alone, while the rests, if serious, generally involve either withholding or discontinuing treatment and administering high-dose steroids with a slow taper over weeks to months until the toxicity alleviates to grade 1 or completely resolves. Unusual cases with fatal irAEs require more potent immunosuppressive agents such as mycophenolate mofetil and infliximab (contraindicated in immune-mediated hepatitis) [126].

Despite the positive results from randomized studies showing superiority in objective response of immune checkpoint inhibitors over chemotherapy, responses are shown in 30% overall at best and half of the responders would live dramatically longer for years. On the other hand, around 40% of patients did not respond at all and some of them had unexpectedly rapid progression during treatment. Identifying the ideal biomarkers of response is therefore essential and still remains a subject under investigation. There was significant heterogeneity among the available tests for PD-L1. The definitive cutoff for PD-L1 positivity was variable across the studies and for a specific kind of tumor. Moreover, studies also differed as to whether they evaluated the PD-L1 expression on tumor cells only (tumor proportion score or TPS) or tumor cells and tumor-infiltrating immune cells (combined positive score or CPS) [127]. The tumor mutation burden (TMB), tumor immune microenvironment, HPV status [128], oral and gut microbiome [129] are among the promising alternative biomarkers that have been extensively studied; however, no definite conclusion has been proposed.

5. Conclusion

The squamous cell carcinoma of the head and neck is one of the most fatal solid malignancies. The promotion of smoking cessation, restricted alcohol consumption, and possibly HPV vaccination would be the most economically effective in cancer control. Breakthroughs in surgical and radiotherapy techniques lead to improvement in oncological, functional, and cosmetic outcomes, especially among those with earlier diseases. Association with HPV is the proven prognostic factor of survival, although it is still not a sole predictive marker for treatment guidance. Concurrent chemo-radiotherapy has been consistently demonstrated to be the best paradigm of management as the definitive treatment for patients with locally advanced disease and as the postoperative adjuvant therapy for resectable disease with high-risk features. The multi-modality treatment is very effective in cancer management but in exchange for significant toxicities. Flail and extreme aging patients are vulnerable to serious treatment-related adverse events. Carefully-tailored management is more suitable for such patients. The recurrent or metastatic SCCHN is almost fatal. Although the immune-checkpoint inhibitors have been able to show promising survival outcomes, the potentially durable responses are observed in an unknown particular subgroup of patients. The researches on biomarkers of treatment response and proper sequence between an ICI and other proven systemic therapies are still ongoing. The incorporation of novel therapies into clinical practices is also an interesting area to be followed.

Conflicts of interest

The author received honoraria from Astra-Zeneca, Merck Sharp & Dohme, Bristol-Myers Squibb, Roche, Eisai, and Lilly for ad hoc Scientific Advisory Board participation.

Funding

This article received no external funding.