Optimizing Treatment De‐Escalation in Head and Neck Cancer: Current and Future Perspectives

Abstract

Treatment of locoregionally advanced head and neck squamous cell carcinoma involves a multidisciplinary approach that combines surgery, radiotherapy, and systemic therapy. These curative strategies are associated with significant acute and long‐term toxicities. With the emergence of human papillomavirus (HPV) as an etiologic factor associated primarily with oropharyngeal squamous cell carcinoma, higher cure rates juxtaposed with substantial treatment‐related morbidity and mortality has led to interest in de‐escalated therapeutic strategies, with the goal of optimizing oncologic outcomes while reducing treatment‐related toxicity. Currently explored strategies include replacing, reducing, or omitting cytotoxic chemotherapy; reducing dose or volume of radiotherapy; and incorporation of less‐invasive surgical approaches. Potential biomarkers to select patients for treatment de‐escalation include clinical risk stratification, adjuvant de‐escalation based on pathologic features, response to induction therapy, and molecular markers. The optimal patient selection and de‐escalation strategy is critically important in the evolving treatment of locoregional head and neck cancer. Recently, two large phase III trials, RTOG 1016 and De‐ESCALaTE, failed to de‐escalate treatment in HPV‐associated head and neck cancer by demonstrating inferior outcomes by replacing cisplatin with cetuximab in combination with radiation. This serves as a cautionary tale in the future design of de‐escalation trials in this patient population, which will need to leverage toxicity and efficacy endpoints. Our review summarizes completed and ongoing de‐escalation trials in head and neck cancer, with particular emphasis on biomarkers for patient selection and clinical trial design.

Implications for Practice

The toxicity associated with standard multimodality treatment for head and neck cancer underscores the need to seek less‐intensive therapies with a reduced long‐term symptom burden through de‐escalated treatment paradigms that minimize toxicity while maintaining oncologic control in appropriately selected patients. Controversy regarding the optimal de‐escalation strategy and criteria for patient selection for de‐escalated therapy has led to multiple parallel strategies undergoing clinical investigation. Well‐designed trials that optimize multimodal strategies are needed. Given the absence of positive randomized trials testing de‐escalated therapy to date, practicing oncologists should exercise caution and administer established standard‐of‐care therapy outside the context of a clinical trial.

Short abstract

This review summarizes completed and ongoing clinical trials evaluating treatment de‐escalation in head and neck cancer, with an emphasis on biomarkers for patient selection and trial design.

Introduction

Historically, head and neck cancer (HNC) treatment has represented an epitome of a multidisciplinary approach that generally includes surgery, radiotherapy (RT), and systemic therapy. Approximately two‐thirds of patients present with locoregionally advanced disease treated with some combination of these three therapeutic modalities [1]. Over the past 2 decades, human papillomavirus (HPV) has emerged as an etiologic factor associated primarily with oropharyngeal squamous cell carcinoma (OPSCC) [2]. HPV‐associated disease results in higher rates of cure compared with HPV‐negative disease with current treatment paradigms [3]. Significant acute and long‐term toxicities are associated with curative multimodality approaches for locoregionally advanced HNC [4]. The notably improved clinical outcomes in HPV‐associated head and neck cancer juxtaposed with significant treatment‐related morbidity and mortality has led to interest in the development of de‐escalated therapeutic strategies with the goal of maintaining or further improving oncologic outcomes while reducing short and long‐term toxicity.

Multiple distinct strategies for treatment de‐escalation are currently being investigated with this goal in mind. Broadly, these strategies can be considered in the context of reducing, replacing, or omitting cytotoxic chemotherapy, reducing dose or volume of RT, and reincorporating less‐invasive surgical approaches. Identification of patients with head and neck squamous cell carcinoma (HNSCC) who are candidates for treatment de‐escalation varies across trials, and is a critical differentiator among de‐escalation strategies. The heterogeneity of approaches demands a concerted effort to review evidence from emerging de‐escalation strategies with a look toward future investigation.

The Changing Face of Head and Neck Cancer

Head and neck cancer remains a major cause of morbidity and mortality worldwide [5, 6]. Although HNSCCs attributable to tobacco and alcohol consumption have been decreasing in incidence since the 1980s, associated with reduced smoking rates in the U.S. [7], oropharynx cancer has initially remained constant and subsequently started to rise despite the reduction in tobacco use in the U.S. [8, 9] relating to an increasing fraction of oropharyngeal HNSCC associated with high‐risk HPV, in particular the HPV16 subtype [2, 8, 10]. Gillison et al. provided strong evidence of a causal relationship between HPV infection and oropharyngeal cancer [11].

HPV‐positive (HPV+) OPSCC has unique clinical and histopathologic characteristics such as early stage primary tumor and more advanced nodal disease [12, 13, 14]. Although historically associated with younger age, recently increased incidence in older adults has been observed [14]. Subset analyses of multiple large randomized controlled trials demonstrated that HPV+ OPSCC was associated with significantly higher cure rates compared with its HPV‐negative counterparts [3, 15, 16]. In the Radiation Therapy Oncology Group (RTOG) 0129 trial in which 433 patients were randomized to RT with either accelerated fractionation with concomitant boost or standard fractionation with concurrent cisplatin, it was noted that of the 74.6% of patients who underwent testing for HPV (assessed based on p16 status), clinical outcomes were significantly better in the HPV+ compared with the HPV‐negative patients, with an 8‐year overall survival of 71% versus 30% respectively (hazard ratio, 0.34; 95% confidence interval [CI], 0.22–0.52) [3].

These prognostic findings led to a reevaluation of staging HPV‐associated OPSCC [12], which has been adapted into the most recent Union for International Cancer Control and American Joint Committee on Cancer eighth edition staging manual to reflect the improved prognosis of this patient cohort [17]. To better prognosticate HPV+ disease, many patients previously categorized as seventh edition TNM stage III–IV are TNM stage I in the eighth edition staging [12]. Importantly, the eighth edition clinical staging paradigm for HPV+ disease was developed under historically intensive treatment regimens. As such, the modified clinical staging for HPV+ disease is not intended to guide clinical decision making. The trials discussed below are based on the seventh edition staging paradigm; however, as de‐escalation trials emerge for HPV+ disease, it will be critical to pay close attention to differences in staging systems used.

Treatment‐Related Toxicity Following Chemoradiotherapy in the Treatment of Locoregional HNSCC

The treatment for locoregional HNSCC has evolved regarding the role of radiation, surgery, and systemic therapy. Historically, in the first half of the 20th century, radiation therapy was the primary treatment, which shifted to include surgery in the middle 20th century with or without adjuvant radiation [1]. This strategy demonstrated high locoregional failure and distant failure rates of approximately 30% and 20%, respectively, with 5‐year survival rates around 40% [18, 19]. In the 1990s, a shift toward functional outcomes and introduction of chemotherapy led to the development of organ preservation strategies [1, 20]. Subsequent randomized clinical trials demonstrated a role for concomitant chemoradiotherapy (CRT) for the definitive treatment of locoregional HNSCC [21, 22]. This was further supported by the MACH‐NC meta‐analysis demonstrating a survival benefit with a concurrent rather than sequential strategy, which was greatest for platinum containing regimens [23]. The survival benefit noted with the addition of concomitant chemotherapy to radiation is predominantly through improved locoregional control [23]. Definitive therapy with concurrent CRT has thus become a standard treatment approach for locoregional HNSCC [24].

Despite these important advances, both acute and late complications of treatment with concurrent CRT remain a major concern. Late toxicities of CRT can have a significant impact on function and quality of life in long‐term survivors of HNSCC. Despite improvements in RT techniques, late toxicities such as xerostomia, long‐term poor dental and oral health, fibrosis, trismus, and esophageal stenosis greatly impact survivorship [25, 26, 27, 28, 29, 30]. Ototoxicity secondary to cisplatin concurrent with radiation results in long‐term ototoxicity rates of approximately 20% [31] and is a limiting factor in delivering optimal cumulative doses of cisplatin. A pooled analysis from three RTOG trials of patients with locally advanced HNSCC treated with concurrent CRT suggested that 43% of patients had a severe late toxicity [4]. Taken together, studies of long‐term complications of treatment have a significant impact on quality of life in long‐term head and neck cancer survivors [27, 28, 32, 33, 34].

Current Landscape of Treatment De‐Escalation Strategies

The emergence of HPV‐associated OPSCC as a highly curable malignancy, albeit at the cost of significant treatment‐related morbidity, has led to an interest in developing de‐escalated treatment strategies with the goal of maintaining oncologic outcomes while reducing short and long‐term toxicity. Broadly, strategies for deintensification of therapy include replacing, reducing, or omitting cytotoxic chemotherapy; reducing dose or volume of RT; and incorporation of less‐invasive surgical approaches (Fig. 1) [35, 36].

Figure 1.

Approaches to de‐escalation in head and neck squamous cell carcinoma.

Replacing, Reducing, or Omitting Cytotoxic Chemotherapy

One strategy of deintensification is combining RT with targeted therapy rather than cytotoxic chemotherapy [37, 40]. A large randomized phase III trial of concurrent cetuximab with RT versus RT alone demonstrated a statistically significant improvement in locoregional control of 24.4 months in the cetuximab arm compared with 14.9 months in the RT alone arm [37]. There was corresponding improvement in overall survival (OS) of 49 and 29.3 months in the cetuximab and RT alone arms, respectively (p = .03). This finding was also noted in subset analysis of the HPV‐positive subset of patients in this trial [38], which suggested the value of comparing the combination strategy of cetuximab with RT versus cisplatin with RT with the goal of maintaining high rates of disease control while reducing toxicity (Table 1) [39, 40].

Table 1.

Strategies to replace, reduce, or omit cytotoxic chemotherapy

Trial Name	Patient population	n	Design	Outcome measure	Reference
Replacing cytotoxic chemotherapy
DeESCALaTE	T3–4N0 or T1–4N1–3, <10 PY smoking, HPV+	334	CRT with cisplatin vs. cetuximab	2‐yr OS (97.5% vs. 89.4%; p = .001)	Mehanna et al. [39]
RTOG 1016	T3–4N0 or T1–4N1–3, HPV+	849	CRT with cisplatin vs. cetuximab	2‐yr OS 84.6% vs. 77.9%;	Gillison et al. [40]
Reducing cytotoxic chemotherapy
RTOG 0129	T3–4N0; T2‐4N2–3; ~64% HPV+	721	CRT with standard vs. accelerated fractionation with cisplatin 100 mg/m2 for 3 vs. 2 cycles	No significant difference in OS	Nguyen‐Tan et al. [3]
UNC/UF study	T0–3N0–2c, minimal smoking hx	44	CRT with cisplatin 30 mg/m2 weekly with RT to 60 Gy	3‐yr OS 95%	Chera et al. [43]
UNC/UF study LCCC1413	T0–3, N0–2, minimal smoking hx	114	CRT with weekly cisplatin 30 mg/m2, or RT alone T0–2N0–1	2‐yr PFS 88%	Chera et al. [45]
Omitting definitive chemotherapy
NRG‐HN002	T1–3N1–2b, T3N0, <10 PY	306	60 Gy of accelerated RT vs. 60 Gy with weekly cisplatin 40 mg/m2	2‐year PFS 90.5% (cisRT) vs. 87.6% (RT alone)	Yom et al. [41]

Abbreviations: cisRT, cisplatin radiotherapy; CRT, chemoradiotherapy; HPV, human papillomavirus; OS, overall survival; PFS, progression‐free survival; PY, pack‐years; RT, radiotherapy; UNC/UF, University of North Carolina and University of Florida.

Mehanna and colleagues reported results of the DeESCALaTE HPV trial in which 334 patients with low‐risk HPV‐positive oropharyngeal SCC were randomized to cisplatin or cetuximab with RT [39]. They found that there was no difference in overall severe toxicity and that there was worse 2‐year OS in the cetuximab arm of 89.4% compared with the cisplatin arm of 97.5% (p = .0012), attributed to worse disease control (16.1% vs. 6.0% respectively; p = .0007). A similar finding was seen by Gillison and colleagues in the NRG Oncology RTOG 1016 trial [40]. In this trial, 849 patients were randomized to either cisplatin or cetuximab with standard RT. Although this trial was designed as a noninferiority trial, OS was significantly worse in the cetuximab arm compared with the cisplatin arm, with estimated 5‐year OS of 77.9% and 84.6%, respectively. There was an increased risk of locoregional failure seen in the cetuximab arm compared with the cisplatin arm, with 5‐year LRF of 17.3% versus 9.9%, respectively. In combination, the results of the DeESCALaTE HPV trial and the RTOG 1016 trial underscore the importance of caution in deintensification strategies in this disease cohort [42].

The reduction in cytotoxic chemotherapy concurrent with RT has also been evaluated. In the RTOG 0129 study [3], patients were randomized to either high‐dose cisplatin 100 mg/m² every 3 weeks for 3 doses concurrent with standard fractionation of 70 Gy in 35 fractions over 7 weeks or cisplatin 100 mg/m² every 3 weeks for 2 doses concurrent with accelerated fractionation of 72 Gy in 42 fractions over 6 weeks. No difference in overall survival was noted; however, because of variability in both radiation and cisplatin, definitive conclusions are limited but suggest the safety and feasibility of reducing total cisplatin dose. Chera and colleagues [43], in collaboration between University of North Carolina and University of Florida, treated low‐risk patients (minimal smoking history, T0–3, N0–2c) with weekly cisplatin 30 mg/m² with concurrent radiation to 60 Gy, followed by biopsy of primary and lymph node dissection, demonstrating an 86% pathologic complete response rate. The 3‐year OS was excellent at 95%, and 39% of patients requiring a temporary feeding tube. A follow‐up study of an additional 114 patients demonstrated 2‐year progression‐free survival (PFS) of 88% [44].

Omission of concurrent cytotoxic chemotherapy was evaluated in the recently reported NRG‐HN‐002 study in which 306 patients with T1–3N1–2b disease or T3N0 disease were randomized to 60 Gy with accelerated radiation, compared with 60 Gy with standard fractionation concurrent with weekly cisplatin 40 mg/m² for 6 cycles [45]. This study demonstrated improved PFS in the concurrent cisplatin arm, with 2‐year PFS of 90.5% meeting the statistical noninferiority margin in the study, whereas PFS in the accelerated RT alone arm was 87.6% and did not meet the noninferiority margin. A subsequent NRG study, NRG‐HN005 study, randomizes patients to standard radiation to 70 Gy with concurrent cisplatin 100 mg/m² for 2 cycles versus 60 Gy of radiation with concurrent cisplatin 100 mg/m² for 2 cycles, and 60 Gy of radiation with concurrent nivolumab (NCT03952585) and is currently accruing.

These results further underscore the importance of caution in reducing, replacing, or omitting concurrent chemotherapy in the definitive management of locoregional HPV+ OPSCC. In particular, the RTOG 1016 and DeESCALaTE studies demonstrating inferior oncologic outcomes with concurrent cetuximab compared with cisplatin, as well as the results of the NRG‐HN‐002 study that demonstrates inferiority of radiation alone compared with concurrent cisplatin, further support that the favorable outcomes of HPV+ OPSCC in definitive chemoradiation trials is likely attributable at least in part to the increased chemotherapy and RT sensitivity of these diseases. Contribution to long‐term toxicity with definitive chemoradiation is in large part related to RT dose and volume [46, 47, 48], which is supported by comparable toxicity between arms in RTOG 1016 and DeESCALaTE.

Minimally Invasive Surgical Approaches with De‐Escalated Adjuvant (Chemo)radiotherapy

In recent years, minimally invasive surgical techniques have relaunched the discussion of risk/benefit considerations of upfront surgical versus chemoradiotherapeutic strategy. Transoral robotic surgery (TORS) is a minimally invasive surgical technique that permits resection of pharyngeal tumors through an open mouth, resulting in significantly improved surgical morbidity (Table 2) [49].

Table 2.

Minimally invasive surgical approaches with de‐escalated adjuvant (chemo)radiotherapy

Trial name	Patient population	Design	Clinical trial identifier
ECOG 3311 [51]	Stage III/IV HPV+ OPSCC	(a) Low risk (T1–2/N0–1) → TORS alone. (b) Intermediate risk → randomized to RT to 50 vs 60 Gy (c) High risk (+ECE, +extensive nodal disease) → CRT with weekly cisplatin 40mg/m2 to 60–66Gy	NCT01898494
PATHOS	HPV+ OPSCC	(a) Low risk → TORS alone. (b) Intermediate risk → randomized to RT to 50 or 60 Gy; (c) High risk (+margin or ECE) are randomized to RT to 60 Gy ± cisplatin (either 3 weekly or weekly according to local practice)	NCT02215265
MC1273 [53]	Locally advanced HPV+ OPSCC	TORS → 30–36 Gy in 1.5–1.8 Gy b.i.d. fractions over 2 weeks with docetaxel 15 mg/m2 × 2	NCT01932697
DART‐HPV Mayo	Locally advanced HPV+ OPSCC	Randomization to MC1273 regimen (described above) versus 60 Gy/2 Gy daily fractions ± concurrent weekly cisplatin 40 mg/m2	NCT02908477
Adjuvant de‐esc RT + nivo	HPV+ intermediate risk (>5+ LNs, N2c/3, +ECE > 1 mm, +margin)	TORS → 45–50Gy RT with concurrent and adjuvant nivolumab	NCT03715946

Abbreviations: +, positive; CRT, chemoradiotherapy; ECE, extracapsular extension; HPV, human papillomavirus; LN, lymph node; nivo, nivolumab; OPSCC, oropharyngeal squamous cell carcinoma; RT, radiotherapy; TORS, transoral robotic surgery.

Performing surgical resection upfront leads to precise pathologic information that can be used for risk stratification in deintensification [36]. This was evaluated in the ECOG 3311 trial in which patients with HPV+ OPSCC undergo TORS and are stratified based on pathologic results. Low‐risk patients (negative margins, N0–1, no extracapsular extension [ECE]) are treated with surgical therapy alone, whereas high‐risk patients (positive margins, >1 mm ECE, or ≥5 positive nodes) are treated with adjuvant concurrent CRT to 66 Gy with weekly cisplatin 40 mg/m² [51]. Intermediate‐risk patients (close margins <3 mm, 2–4 positive lymph nodes, ≤1 mm ECE, perineural invasion/lymphovascular invasion) are then randomized to 50 Gy or 60 Gy adjuvant RT. Results of this trial were recently presented demonstrating that among 206 intermediate‐risk patients randomized to 50 Gy versus 60 Gy adjuvant RT, 2‐year PFS was 95.0% and 95.9%, respectively [51]. This informs de‐escalated adjuvant therapy in patients who undergo surgery, but this upfront surgical strategy will need to be compared with a de‐escalated definitive chemoradiotherapeutic approach in terms of oncologic and functional outcomes to clarify whether surgery is necessary. Similarly, the PATHOS trial in the U.K. is stratifying patients by pathologic risk group. Low‐risk patients receive surgery alone, intermediate‐risk patients (T3, N2a–b with perineural or vascular invasion or close margins) are randomized to two arms either 60 Gy or 50 Gy, and high‐risk patients (positive margins or ECE) are randomized to RT to 60 Gy with or without concurrent cisplatin (Table 2) [36]. Other deintensification strategies in HPV+ disease are evaluating the question of treatment implications for tumors with ECE on surgical resection, given controversial retrospective data suggesting lack of prognostic significance in HPV+ disease [52]. The Adjuvant De‐escalation, Extracapsular Spread, P16+, Transoral trial attempted to randomize HPV+ patients to adjuvant CRT or RT alone (NCT01687413) however, this trial unfortunately closed because of poor accrual. Ultimately for patients requiring postoperative (C)RT, the benefit of upfront surgical resection versus a de‐escalated definitive CRT approach remains an open question. As such, an adjuvant paradigm will need to be considered in the context of evolving definitive de‐escalated CRT.

A novel adjuvant de‐escalation approach from the Mayo Clinic group is investigating a substantial reduction in radiation dose and duration with concurrent docetaxel. This trial included patients with p16+ tumors with minimal smoking history and negative margins who underwent surgical resection. A total of 37 patients without ECE were treated with 30 Gy delivered in 1.5‐Gy fractions over 2 weeks with weekly docetaxel 15 mg/m² for 2 weeks. An additional 43 patients with ECE were treated with a simultaneous integrated boost to nodal levels with ECE to 36 Gy in 1.8‐Gy twice‐daily fractions for 2 weeks with the same chemotherapy [53]. The 2‐year PFS was reported at 91.1% for this cohort overall, with a median follow‐up of 36 months. A follow‐up phase III trial is currently enrolling (NCT02908477) randomizing patients to this regimen versus standard adjuvant treatment of radiation with or without concurrent cisplatin. Results from this novel adjuvant approach are eagerly awaited as a strategy to optimize minimally invasive surgical approach while aggressively de‐escalating radiation dose. An ongoing trial is evaluating de‐escalated adjuvant therapy following minimally invasive surgery incorporating nivolumab, an anti–programmed death‐1 monoclonal antibody, both concurrent and following de‐escalated radiation (NCT03715946).

Reduction in Radiotherapy Dose Following Induction Chemotherapy

Response to induction therapy to select patients for subsequent de‐escalation is a strategy being investigated (Table 3). The ECOG 2399 study evaluated 96 patients with stage III or IV HNSCC treated with two cycles of induction paclitaxel and carboplatin followed by weekly concomitant paclitaxel with standard fractionation RT [54]. A total of 40% of patients had oncogenic HPV positivity detected. Patients with HPV‐positive HNSCC demonstrated higher response rates compared with HPV‐negative tumors, with response rates of 82% and 55%, respectively (p = .01). The 2‐year OS was also improved in the HPV‐positive cohort compared with the HPV‐negative cohort with 2‐year OS rates of 95% versus 62%, respectively (p = .005).

Table 3.

Strategies involving induction followed by risk‐adapted locoregional therapy

Trial name	Patient population	n	Design	Outcome measure	Reference
ECOG 2399	Stage III/IV oropharynx/larynx (40% HPV)	96	Carboplatin and paclitaxel ×2 cycles → CRT with weekly paclitaxel	2‐yr OS 95% (HPV+) vs. 62% (HPV−) RR: 82% (HPV+) vs. 55% (HPV−)	Fakhry et al. [54]
ECOG 1308	HPV+ OPSCC	90	Cisplatin, paclitaxel, and cetuximab → CRT with concurrent cetuximab to 54 Gy (CR) or 69.3 Gy (<CR)	2‐yr PFS 78%; 2‐yr OS 91%	Marur et al. [58]
RAVD	Locally advanced HNSCC (63% HPV+)	94	Cisplatin, paclitaxel, cetuximab ± everolimus → volume de‐escalation (>50% shrinkage)	G‐tube dependence at 6 mo (5.7% in de‐escalated vs. 32.6%; p = .005)	Villaflor et al. [59]
OPTIMA	Locally advanced HPV+ OPSCC	62	Carboplatin and nab‐paclitaxel ×3 cycles → (a) 50 Gy (low risk, >50% shrinkage) (b) CRT 45 Gy (low 30%–50%; high risk >50%) (c) CRT 75 Gy (high risk <30%)	2‐year PFS 94%	Seiwert et al. [60]
Chen et al.	Locally advanced HNSCC	45	Carboplatin and paclitaxel ×2 cycles → CRT with paclitaxel to 54 Gy (PR/CR) or 60 Gy (<PR)	2‐yr PFS 92%	Chen et al. [61]
OPTIMA 2	Locally advanced HPV+ HNSCC		Carboplatin/nab‐paclitaxel/nivoumab ×3 cycles → risk and response‐adapted locoregional therapy		NCT03107182
DEPEND	Locally advanced HPV‐ HNSCC		Carboplatin/nab‐paclitaxel/nivoumab ×3 cycles → risk and response‐adapted locoregional therapy		NCT03944915

Abbreviations: –, negative; +, positive; CR, complete response; CRT, chemoradiotherapy; HNSCC, head and neck squamous cell carcinoma; HPV, human papillomavirus; OPSCC, oropharyngeal squamous cell carcinoma; OS, overall survival; PFS, progression‐free survival; PR, partial response.

It has long been identified that response to induction chemotherapy is predictive of response to subsequent RT or CRT [55, 56, 57]. In the E1308 trial, 90 patients with HPV+ OPSCC were treated with 3 cycles of induction therapy with cisplatin, paclitaxel, and cetuximab followed by RT to 54 Gy with weekly cetuximab in patients with a primary‐site complete clinical response or 69.3 Gy with cetuximab in patients with less than complete clinical response. A total of 70% of patients achieved a clinical complete response (CR) and were treated to 54 Gy, whereas the remaining patients who did not achieve a CR were treated to 69.3 Gy, each with concomitant cetuximab. Patients treated to 54 Gy had 2‐year PFS and OS of 80% and 94%, respectively, and improved swallowing function and nutrition at 12 months [58, 62].

Our team at The University of Chicago has investigated beyond radiation dose de‐escalation, a role for radiation volume de‐escalation following induction therapy. A total of 94 patients with locally advanced HNSCC (63% HPV+) were treated with two cycles of induction chemotherapy followed by response‐adapted volume de‐escalation (elimination of elective nodal coverage) in patients with ≥50% tumor shrinkage following induction [59]. This strategy demonstrated reduced long‐term toxicity with significantly less G‐tube dependence at 6 months (5.7% versus 32.6%; p = .005), without compromising outcomes with >90% of locoregional failures developing within the radiation treatment volumes and occurring in the highest risk PTV1.

Subsequently, we built upon these published results with our institutional OPTIMA trial, a phase II dose and volume de‐escalation trial with risk stratification and response to induction therapy [60]. In this trial, 62 patients with HPV+ OPSCC were enrolled, of whom 28 were low risk (T1–3, N0–2b, and less than 10 pack‐years smoking history), and 34 were high risk (T4, N2c–N3, or greater than 10 pack‐years smoking history). All patients received induction with carboplatin and nab‐paclitaxel for 3 cycles. Patients with low‐risk disease with ≥50% response received de‐escalated radiation to 50 Gy; low‐risk patients with 30%–50% tumor shrinkage and high‐risk patients with ≥50% response were treated with de‐escalated concurrent CRT to 45 Gy with twice‐daily radiation and paclitaxel, 5‐fluorouracil, and hydroxyurea (TFHX) week‐on week‐off. Low‐risk patients with less than 30% shrinkage and high‐risk patients with less than 50% shrinkage were treated with standard‐dose concurrent CRT to 75 Gy with twice‐daily fractions concurrent with TFHX week‐on week‐off. Results of this trial demonstrated that in the low‐risk cohort, 2‐year PFS and OS were 95% and 100%, respectively. In the high‐risk cohort, 2‐year PFS and OS were 94% and 97%, respectively. Acute toxicities were significantly reduced in the de‐escalated cohorts, demonstrating significantly lower rates of grade 3+ mucositis, dermatitis, and need for enteral feeding.

A study evaluating a similar approach by Chen et al. evaluated 45 patients with locally advanced HNSCC who were treated with two cycles of induction with carboplatin plus paclitaxel, followed by reduced‐dose concurrent CRT with paclitaxel to 54 Gy in partial and complete responders, and 60 Gy in nonresponders [61]. The Mount Sinai group randomized patients to induction docetaxel, cisplatin, and 5‐FU with a favorable response to standard dose (70 Gy) versus dose reduced (56 Gy) with concurrent carboplatin (NCT01706939), which unfortunately closed because of poor accrual and lack of financial support. Other induction strategies have included neoadjuvant chemotherapy followed by transoral surgery [50] and neoadjuvant immunotherapy followed by transoral surgery [63]; however, these trials were small (17 and 28 patients, respectively) and therefore limit definitive conclusions.

The role of immunotherapy in the paradigm for treatment de‐escalation is evolving. The REACH study is randomizing patients to avelumab, an anti‐programmed death ligand 1 monoclonal antibody, with cetuximab and RT versus cisplatin or cetuximab and RT (NCT02999087). The JAVELIN head and neck study is randomizing patients to avelumab, with CRT versus CRT alone (NCT02952586). EMD‐Serono and Pfizer announced that the JAVELIN study will be terminated as the study is unlikely to show a statistically significant improvement in the primary endpoint of PFS based on a preplanned interim analysis [64]. The KEYNOTE 412 study is randomizing patients to pembrolizumab with CRT versus CRT alone (NCT03040999). At the University of Chicago, our team is incorporating immunotherapy both as a component of an induction chemoimmunotherapy regimen and in the adjuvant setting both for HPV+ disease (OPTIMA II; NCT03107182) and HPV‐negative disease (DEPEND; NCT03944915). Early results of our institutional OPTIMA II trial of patients treated with induction carboplatin, nab‐paclitaxel, and nivolumab followed by TORS or RT (50 Gy) alone demonstrated favorable results with excellent oncologic outcomes with reduced toxicity [65].

Novel Strategies for Selecting Patients for Treatment De‐Escalation

The heterogeneity of deintensification clinical trials in head and neck cancer include variability in strategies for patient selection for de‐escalation. HPV positivity has clearly been established as a favorable prognostic and predictive biomarker and, as a result, has been the major focus of deintensification efforts. Clinical factors that have been identified as high‐risk features include staging factors such as T4 tumor stage, bilateral nodal disease, or N3 nodal conglomerate. Significant smoking history has been identified as a high‐risk feature in HPV+ OPSCC and has been identified as an exclusion for multiple deintensification trials. Response to induction therapy as a biomarker for treatment de‐escalation is being applied in multiple trials as well. Improved predictive and prognostic biomarkers are needed to enhance the next generation of biomarker selected de‐escalation trials in head and neck cancer (See Fig. 2).

Figure 2.

Biomarker selection for treatment deintensification in head and neck squamous cell carcinoma. Abbreviations: ER, estrogen receptor; HPV, human papillomavirus; MATH, mutant allele tumor heterogeneity.

Recently, genomic biomarkers have become of interest to further characterize tumor biology. Mutations in PIK3CA, which is commonly mutated in HPV+ OPSCC and has been associated with worse disease‐free survival compared with PIK3CA wild‐type tumors in de‐escalation trials, suggests a potential role as a genomic biomarker [66]. Mutations in p53 have been associated with smoking‐related HPV‐negative cancer and likely characterize tumors driven by smoking oncogenesis and may identify a poor risk population [67]. Other molecular biomarkers have included mutant allele tumor heterogeneity (MATH), which is a quantitative measure of intratumor genetic heterogeneity and is associated with worse outcomes, and estrogen receptor (ER)‐α expression was found to be a favorable prognosis. Using composite biomarker of high MATH score, low ER‐α expression, and HPV negative, hazard ratio was 28.2 (p = .0001). TRAF3 and CYLD loss by inactivating mutations or deletions have been associated with a favorable prognosis among HPV+ OPSCC and is associated with episomal HPV and activation of NF‐kB [68].

Other biomarkers under active investigation include serum HPV circulating tumor DNA assays. Serum‐based biomarkers are of substantial interest in selecting patients not only for upfront prognostication but also as a dynamic biomarker of treatment response and risk of recurrence following treatment [69, 70]. A prospective clinical trial of 115 patients with HPV+ OPSCC treated with definitive concurrent CRT assessed the role of a plasma‐based circulating tumor HPV DNA biomarker [70]. Plasma‐circulating tumor HPV DNA demonstrated a negative predictive value of 100% (95% CI, 96%–100%), and two consecutively positive circulating tumor HPV DNA had a positive predictive value of 94% with a median lead‐time of HPV DNA positivity to biopsy proven recurrence of 3.9 months. These promising biomarkers will need to be validated in larger cohorts and in different patient populations and treatment paradigms for broader application, in particular with respect to HPV de‐escalation protocols.

Conclusion

The multitude of novel therapeutic approaches in patients with head and neck cancer continues to grow as we refine our therapeutic strategies with the goal of maximizing oncologic control while minimizing toxicity. The heterogeneity of de‐escalation approaches with patient selection and clinical trial design have limited the broader application of de‐escalated treatment for HPV+ OPSCC and thus remains investigational. Well‐designed, large, randomized, multicenter clinical trials are needed to refine, optimize, and establish a treatment paradigm for HPV+ OPSCC that optimizes oncologic outcomes while reducing acute and chronic toxicities.

Author Contributions

Conception/design: Ari J. Rosenberg, Everett E. Vokes

Collection and/or assembly of data: Ari J. Rosenberg

Data analysis and interpretation: Ari J. Rosenberg, Everett E. Vokes

Manuscript writing: Ari J. Rosenberg, Everett E. Vokes

Final approval of manuscript: Ari J. Rosenberg, Everett E. Vokes

Disclosures

Ari J. Rosenberg: EMD Serono (C/A); Everett E. Vokes: C/A: AbbVie, Amgen, AstraZeneca, Bristol‐Myers Sqibb, Celgene, Eli Lilly, EMD Serono, Genentech, GlaxoSmithKline, Merck, Novartis, Regeneron (C/A).

(C/A) Consulting/advisory relationship; (RF) Research funding; (E) Employment; (ET) Expert testimony; (H) Honoraria received; (OI) Ownership interests; (IP) Intellectual property rights/inventor/patent holder; (SAB) Scientific advisory board