Over the past decades, the epidemiology of oropharyngeal squamous cell carcinoma (OSCC) has been characterised by an increased incidence attributed to human papillomavirus (HPV) infection [1, 2]. HPV-related (HPV+) OSCCs differ from other head and neck squamous cell carcinomas commonly associated with chronic smoking and drinking intoxication. They occur in healthier individuals with little or no tobacco consumption, and despite a high rate of nodal extension at diagnosis, locoregional recurrence rates decrease and survival improves irrespective of disease stage or treatment option when compared to HPV-unrelated (HPV−) OSCC patients [3]. One of the hallmarks of HPV+ OSCC is overexpression of p16, a cell cycle regulator protein whose positivity in immunochemistry is considered to be a reliable surrogate marker of an HPV-driven oncogenic process [4]. This has resulted in the development and validation of a specific staging system for HPV-related OSCC based on p16 status [5]. In OSCC, treatment options consist of either surgery followed by adjuvant radio(chemo)therapy (surgical treatment) or definitive radio(chemo)therapy (RCT) [6]. Although HPV-related OSCCs are a distinct entity and p16 overexpression represents a major prognostic factor, limited data are available on its predictive significance. No strategy has been identified as more effective and no predictive factors guide the treatment decision; nevertheless, further evidence on this issue is required. To assess the effect and the late toxicity profile of each strategy based on p16 expression, we retrospectively reviewed all consecutive patients treated for OSCC with curative intent between January 2009 and December 2013 at our tertiary cancer centre.

Patients and methods

As part of an institutional board-approved study, all consecutive patients treated with curative intent for OSCC between January 2009 and December 2013 were identified from a prospective departmental database (N = 278). Clinical records were retrospectively reviewed to verify patient and tumour characteristics, treatment details and clinical outcomes. Ninety-five patients were excluded from analyses for the following reasons: metastatic disease at diagnosis (N = 4), multiple synchronous tumour sites (N = 22), history of prior head and neck carcinoma (N = 20), three-dimensional conformal radiotherapy (N = 26) or material not available for p16 review (N = 23).

Pretherapeutic evaluation included:

  • physical examination of the head and neck by a surgeon and radiation oncologist,

  • triple endoscopy under general anaesthesia,

  • biopsies and biological tests,

  • cervical and thoracic CT scan (fluorodeoxyglucose positron-emission tomography [FDG PET] was optional),

  • orthopantomogram and a dedicated dental consultation,

  • nutritional assessment by a dietitian,

  • vocal and swallowing evaluation by a dedicated speech and swallowing therapist.

Tumours were staged according to the American Joint Committee on Cancer (AJCC; 7th edition) and each case was discussed by an institutional multidisciplinary head and neck cancer board before treatment was initiated.

Treatment decisions were made at the discretion of the institutional multidisciplinary head and neck tumour board. Definitive radio(chemo)therapy was indicated in cases of unresectable tumours, when general anaesthesia was contraindicated due to medical conditions or in some cases of small primary cancer of the tonsil or base of the tongue. Surgery consisted of en bloc resection of the lateral wall of the oropharynx, the close portion of the soft palate, partial glossectomy and mandibulectomy in case of bone extension. When the primary tumour did not extend over the midline, a unilateral I to V dissection was performed. In cases of significant resection leading to expected functional impairment, free flap reconstruction was provided for. Detailed operative approaches are depicted in Table 2. In some cases of small tumours with bulky nodal extension (≥3 cm), an initial neck dissection was performed before definitive radio(chemo)therapy as previously reported [7]. During radiotherapy (RT), all patients were immobilised in a supine position using a five-point thermoplastic mask. A contrast-enhanced CT scan was obtained for treatment planning. All available diagnostic MRI and/or PET scans were fused to the treatment planning CTs. RT was delivered using intensity modulated RT (IMRT): step and shoot, volumetric-modulated arc therapy (VMAT) or Tomotherapy® (Accuray Incorporated, Sunnyvale, CA, US) using the integrated boost technique as previously reported [8]. In the definitive setting, patients were prescribed a dose of 66–70 Gy in 30–35 fractions to a high-risk planning target volume (PTV) and 54–56 Gy in 30–35 fractions to low-risk PTV. Adjuvant RT was indicated in cases of T3–T4 tumours or large nodal extension (>2 involved nodes or diameter >15 mm). In the postoperative setting, the prescription dose was 54–63 Gy in 27–30 fractions to the high-risk PTV and 54 Gy in 30 fractions to the low-risk PTV. Concurrent platin-based chemotherapy was added to adjuvant radiotherapy in cases of extra-capsular spreading or involved final resection margin. Follow-up consisted of three-monthly physical examinations, which included direct fibreoptic nasopharyngeal laryngoscopy by a radiation oncologist or a head and neck surgeon. Contrast-enhanced CT evaluation was performed 3 months after treatment completion and annually thereafter, or if failure was suspected. All patients were seen by a dedicated dentist and a speech and swallowing therapist twice a year for the first year after RT completion and annually thereafter. Salivary and swallowing functions were closely monitored during follow-up and were classified as “normal” function, “moderate” or “severe impairment or feeding tube dependency”, and were retrospectively scored according to the Common Terminology Criteria for Adverse Events (CTCAE) scale (version 4.3) at last follow-up or before any locoregional failure within a minimum of 6 months after RT and/or surgery completion. Delayed severe toxicities were defined as grade 3 or higher dysphagia, xerostomia or osteoradionecrosis.

Immunostaining of p16INK4A was performed on 3 µm thick formaldehyde-fixed paraffin-embedded tissue sections, which were deparaffinised using high-pH solution. As a primary p16INK4A antibody, clone E6H4 (mouse monoclonal, 1:30 dilution; Roche, Almere, the Netherlands) was used and detected using Powervision (DAKO A/S, Glostrup, Denmark) and peroxidase-DAB visualisation [4]. Two independent dedicated head and neck pathologists (TG or EUC) performed evaluations of the immunostained samples and a consensus was reached on the scores. Positive p16 status was defined by continuous strong nuclear p16INK4A positivity with or without cytoplasmic staining observed in all tumour cells. In each analysis, negative and positive controls were included. The data were summarised as frequency and percentage for categorical variables and by median, range (min.–max.) and interquartile range (Q1–Q3) for continuous variables. The chi-squared or Fisher’s exact test was used to compare categorical variables, and the Kruskal–Wallis test was used for continuous variables. All survival times were calculated from the date of diagnosis and were estimated by the Kaplan–Meier method with 95% confidence intervals (CI) using the following first-event definitions: locoregional relapse for locoregional control (LRC), metastatic relapse for freedom from metastases (FFM), locoregional relapse, metastatic relapse, other cancer or death from any cause for disease-free survival (DFS) and death from any cause for overall survival (OS). Univariate and multivariate analyses were performed using the logrank test and Cox proportional hazards model, respectively. All reported p-values were two sided. For all statistical tests, a statistical difference was considered significant at the 5% level. Statistical analyses were conducted using Stata®, version 13 (StataCorp LLC, College Station, TX, USA).

Results

A total of 183 patients were included in this study. Among them, 89 (49%) presented with p16 overexpression (p16+). When compared to p16-negative (p16−) patients, p16 overexpression correlated with better medical condition (International Prognostic Score [IPS] score 0: 84 vs. 36%; p < 0.001), smaller primary tumour (T1–2: 70 vs. 48%; p = 0.003) and a high rate of nodal extension (N2–N3: 64 vs. 54%; NS). There were no significant differences regarding final resection (margins involved or ≤2 mm: 37 vs. 34%) or extracapsular spreading (58 vs. 47%). The overall cohort repartition according to Ang’s risk profile was equal to 38, 12 and 50% in low-, intermediate- and high-risk subgroups, respectively [9]. Patient and tumour characteristics according to p16 status are listed in Table 1.

Table 1 Initial demographic and disease characteristics according to p16 status
Full size table

Overall, 106 patients (58%) underwent definitive radio(chemo)therapy. Among them, 25 (24%) underwent initial neck dissection prior to definitive radio(chemo)therapy. In 77 patients (42%) surgical resection was performed, followed by adjuvant radio(chemo)therapy in 68 cases (88%). When compared to the definitive RCT group, surgical patients presented no significant difference in terms of median age (58 vs. 59 years; p = 0.74), performance status ≥1 (42 vs. 38%; p = 0.51), AJCC tumour stage ≥III (76 vs. 82%; p = 0.37) or p16 overexpression (48 vs. 49%; p = 0.86).

Treatment modalities and pathological findings according to p16 status are listed in Table 2.

Table 2 Treatment characteristics and pathological findings according to p16 status
Full size table

Overall, 142 patients were assessable for delayed toxicities (follow-up ≥6 months without locoregional disease). Delayed severe toxicities, i. e. dysphagia, xerostomia and osteoradionecrosis according to treatment modality and p16 status are detailed in Table 3 and 4. In the p16+ group (N = 84), surgical patients (N = 37) presented with a significantly higher rate of delayed severe dysphagia as compared to definitive RCT patients (N = 47): 35 vs. 4% respectively; p = 0.0002.

Table 3 Delayed severe toxicity (≥grade 3) according to treatment modality among p16-negative patients (n = 58)
Full size table
Table 4 Delayed severe toxicity (≥grade 3) according to treatment modality among p16-positive patients (n = 84)
Full size table

After a median follow-up of 4.2 years (95%CI: 3.8–4.4 years), 3‑year OS, DFS, FFM and locoregional control (LRC) were 76% (95%CI: 68–81), 64% (95%CI: 57–71), 86% (95%CI: 79–90) and 78% (95%CI: 71–83), respectively. The factors significantly associated with DFS in univariate analysis were p16 status (p16+ versus p16−; 82 vs. 47%; p < 0.0001), tumour stage (T1–2 vs. T3–4; 74 vs. 50%; 0.0007), treatment modality (initial surgery vs. definitive RCT; 74 vs. 57%; p = 0.007), IPS status (0 vs. ≥1; 78 vs. 44%; p < 0.0001) and tobacco consumption (<10 vs. ≥10 pack-years; 84 vs. 57%; p = 0.0002). In multivariate analyses, p16 status, T stage and treatment modalities remained associated with DFS. Univariate and multivariate analyses of prognostic factors for DFS are outlined in Table 5. When considering p16 status, p16+ patients did not display any survival benefit from an initial surgical approach (N = 37) compared to definitive RCT (N = 47; 3‑year DFS: 83 vs. 82%; p = 0.485; Fig. 1).

Table 5 Univariate and multivariate analyses of risk factors associated with DFS from the overall cohort
Full size table
Fig. 1
figure 1

Disease-free survival according to treatment strategies: overall cohort (a) and p16+ patients (b)

Full size image

Discussion

This is a large series reporting on 183 OSCC patients treated in recent years (2009–2013) in the era of IMRT and modern surgical techniques (transoral robotic surgery, free flap reconstruction), including a review of p16 expression and a delayed toxicity evaluation. The optimal treatment for OSCC remains controversial; as such, our study focused on treatment-related outcomes according to p16 status, which represents a major prognostic factor in OSCC and is easily available in routine practice. Almost half of our cohort (49%) presented with p16 overexpression, which was associated with a significantly better medical condition (score IPS 0: 84 vs. 36%; p < 0.0001), smaller primary tumours (T1–2: 70 vs. 48%; p = 0.003) and a high rate of nodal extension at presentation (N2–N3: 64 vs. 54%; NS), and which correlated with a better prognosis (3-year DFS: 82 vs. 47%; p < 0.0001) as compared to p16-negative patients. Our data are consistent with previously reported findings [10, 11]. After a median follow-up of 4.2 years, the 3‑year DFS of initial surgery and definitive RCT patients was 74 vs. 57%, respectively (p = 0.007). When considering p16 expression, p16+ patients did not display any survival benefit from an initial surgical approach as compared to definitive radio(chemo)therapy (3-year DFS: 83 vs. 82%; p = 0.485). Although p16+ OSCC patients are likely to be eligible for initial surgery due to small primary tumours and better medical condition, pathological findings often require adjuvant treatment. In our series, among surgical p16+ patients (N = 37), 37% presented a clear final margin ≤2 mm and 58% presented extracapsular spreading. The recent observation that p16+ OSCC patients have positive outcomes regardless of the considered treatment option has led many authors to consider that these patients are being overtreated and that a single-modality treatment should be preferred [12]. Lacau St Guily et al. recently suggested that the absence of upfront surgery worsened PFS in HPV-unrelated OSCC patients to a greater extent than in HPV-related patients [13]. Bossi et al. demonstrated distinct outcomes after definitive radio(chemo)therapy or initial surgical treatment followed by RT for HPV+ OSCC in favour of RCT for these patients [14]. In our study, initial surgery was associated with an overall higher rate of delayed severe dysphagia when compared to medical treatment: 35 vs. 4%, respectively (p = 0.0002). Open radical surgery, even with state-of-the-art free flap reconstruction, yielded poor functional outcomes [15]. Nevertheless, alternative minimally invasive surgical techniques could be offered to p16+ OSCC patients, such as transoral robotic surgery, which results in improved functional outcomes; however, the final result is impaired in the case of adjuvant radio(chemo)therapy [16]. In a matched-pair study, Jackson et al. recently observed that adjuvant radio(chemo)therapy associated with transoral surgery in HPV-related OSCC was correlated with improved DFS when compared to surgery alone (hazard ratio: 0.067; 95% CI 0.01–0.62) [17]. The rate of late dysphagia after surgery ± adjuvant RT is similar irrespective of p16 status, i.e. 35% for each group. When focusing on the definitive RCT group, the rate of severe delayed dysphagia was 4 vs. 30% for p16+ and p16− patients, respectively. Interestingly, this is not explained by the initial tumour stage, nodal extension or adjuvant treatment, which were relatively similar between the two groups (p16+ vs. p16−) except for cisplatin use (Table 2). The most important characteristics distinguishing p16+ patients from p16− patients in our cohort are the proportion of tobacco consumption >10 pack-years: 40 vs 98% (p < 0.01) and alcohol abuse 22.5 vs 86% (p < 0.01), respectively. Tobacco consumption is a well-known predictive factor for late normal tissue complications following radiotherapy [18, 19]. In our study, given the important rate of high tobacco exposure in p16− patients, one might hypothesize that this may have contributed to enhancing radio-induced toxicity even without initial surgery. That HPV-related OSCC might respond more to RT is supported by a growing body of preclinical evidence underlining a higher intrinsic radiosensitivity of HPV+ OSCC cell lines as compared to HPV-unrelated cell lines, which could explain the lack of benefit of initial surgery among p16+ OSCC patients [20]. Indeed, these tumours have their own oncogenic process: cellular infection by viral DNA promotes overexpression of oncoproteins E6 and E7 that inhibit cell cycle regulators pRb and p53, favouring transcription of the tumour suppressor gene p16 that would in turn inhibit radiation-induced DNA damage repair [21]. Interestingly, HPV-related carcinomas from other sites (i. e. cervix and anal squamous cell carcinoma) are treated with definitive radio(chemo)therapy at first intent [22, 23]. Furthermore, tumour inflammation and PD-L1 expression are present at a higher degree in HPV+ OSCC when compared to HPV− OSCC, which may have a synergistic interaction with RT to elicit immune tumour recognition [24]. Optimum reduced-dose treatment regimens for these patients are being investigated by various groups [25]. Chera et al. reported a pathological response in 86% of patients after chemoradiotherapy with dose of 60 Gy and weekly cisplatin [26].

There are several limitations to our study in addition to its retrospective nature. Firstly, direct retrospective comparisons of surgical or medical approaches are hazardous given the numerous biases associated with selection of operated patients that favour surgery: resectable tumours and no contraindication for general anaesthesia. When considering definitive RCT or initial surgical treatment, even if the two groups appear well balanced with regard to tumour AJCC stage, performance status, median age and p16 overexpression, we cannot preclude that some patients treated with definitive radio(chemo)therapy were not eligible for initial surgery due to tumoural extension or medical condition, which are commonly recognised as poor prognostic factors contributing to impair the results of patients treated with definitive RCT. However, when focusing on p16+ patients, the 3‑year DFS did not vary depending on whether patients underwent definitive RCT or an initial surgical strategy. Secondly, although dysphagia and xerostomia were closely monitored by our dedicated speech and swallowing therapist and our dentist, some other delayed toxicities that contribute to quality of life impairment following neck dissection may have been under-evaluated during follow-up, such as cervical fibrosis or painful shoulder. There are still a number of unanswered issues regarding the role and the optimal timing of neck dissection. In our series, 22% of our p16+ patients (N = 20) underwent initial neck dissection before definitive radio(chemo)therapy. Considering their high rate of nodal extension at diagnosis, we may not exclude the need for nodal resection in the definitive radio(chemo)therapy setting. Given their higher intrinsic radiosensitivity, we may hypothesise that not all p16+ locally advanced OSCC patients would require neck dissection after definitive RT. FDG PET performed within 3–4 months after RT completion is a reliable tool for shifting from initial planned dissection to surgery in the case of residual fixation [27, 28]. Likewise, Garden et al. reported equivalent neck recurrence rates among 401 locally advanced HPV+ OSCC patients treated with definitive RT, irrespective of whether they presented complete nodal response or underwent neck dissection because of residual FDG PET fixation (i. e. 8%; p = 0.4) [29].

Conclusion

In our cohort, p16+ patients did not benefit from an initial surgical approach as compared to definitive RCT (3-year DFS: 83 vs. 82%; p = 0.485), whereas initial surgery did result in a significantly higher rate of delayed severe dysphagia, 35 vs. 4%; p = 0.0002. Our findings—combined with those from previous studies—emphasise the distinct outcomes after initial surgery or definitive RCT according to p16 status, which should be taken into consideration during the decision process.