Salvage surgery in head and neck cancer

Abstract Salvage surgery after failed organ preservation treatment offers challenges for both the patient and the surgeon. The outcome is often uncertain and even today, 5‐year overall survival does not exceed 50 per cent. The chemoradiotherapy induced toxicity asks for meticulous discussion and planning in a multidisciplinary manner in a changing environment of increasing incidence of human papillomavirus induced oropharyngeal tumours, evolving surgical techniques and patient participation. Herein, we discuss the latest literature on salvage surgery and the need for identifying the proper prognosticators to ensure for an optimal treatment plan in potentially salvageable patients.


| INTRODUC TI ON
Salvage surgery in Head and Neck Cancer (HNC) after failed (chemo) radiotherapy is a complex and increasingly important issue with high stakes for the patients. Patients eligible for SS have previously been through the process of HNC treatment with accompanying anxiety and uncertainties. Therefore, they should be guided accordingly and informed in a truthful and concise manner meaning that salvage surgery is a last resort treatment with an often uncertain outcome both considering cure and function impacting quality of life. Salvage surgery should never be considered a fall back option in case patients elect organ preservation treatment over an advised primary surgical treatment. Performing surgery in previously irradiated tissue, especially when combined with systemic treatment enhancing toxicity, is a very difficult and comes with many both short-and long-term complications.
Although surgical and radiation techniques have improved, salvage surgery remains a journey not easily embarked on with current success rates often not exceeding 30%. This paper sets out to give an overview of current literature with regard to prognosticators in salvage surgery in light of developments such as increasing incidence of human papillomavirus in oropharyngeal squamous cell carcinoma, the use of transoral robotic surgery, necessity of multidisciplinary management and the increasing awareness for value-based health care.

| E VOLVEMENT OF SALVAG E SURG ERY
Although salvage surgery has always played a role in HNC, its prospects have changed over the last decades with the introduction of combined modality treatment for advanced stage disease. The use of mainly cisplatin and later cetuximab in platinum unfit patients has added toxicity causing bigger challenges for uneventful outcome in salvage surgery (Bonner et al., 2006;Pignon, Bourhis, Domenge, & Designé, 2000;Rovira et al., 2017). Every head and neck surgeon has experienced the setbacks of poor healing tendency and disappointing functional outcome caused by severe fibrosis and inferior perfusion, despite reports of improved outcome over the last two decades (Jayaram et al., 2016). The use of transoral robotic surgery (TORS) should currently be added to the surgeon's armamentarium.
TORS can be employed in salvage surgery for mainly early-stage recurrent disease and can be utilized for salvage in well-selected cases (White et al., 2013).

| Site
Success rates in salvage surgery differ per site. Historically, reported outcome in recurrent laryngeal cancer is relatively good, specifically in early-stage recurrences (Bonner et al., 2006;Chung et al., 2019;Chung, Park, Kwon, & Rho, 2020;Elbers et al., 2019;Goodwin, 2000;van der Putten et al., 2015). Laryngeal recurrences are relatively salvageable due to the possibilities of achieving adequate surgical margins and low nodal spread in early-stage recurrences. Goodwin (2020) reported a 2-year overall survival (OS) of well above 60% (83.4% for recurrent stage I/II). Since salvage pharyngo-laryngectomy is the mainstay of treatment in recurrent hypopharyngeal squamous cell carcinoma (SCC), these series are often analysed together with laryngeal recurrences. Outcome in recurrent hypopharyngeal SCC is inferior to recurrent laryngeal SCC.
Complication rate is higher in salvage PL probably due to high percentage of prior chemotherapy and the notoriously poor outcome of hypopharynx cancer (Chung et al., 2019;Elbers et al., 2019;van der Putten et al., 2015). Van der Putten et al. (2015) found a 5-year OS of 27% for salvage (pharyngo)laryngectomy after primary chemoradiation ( Figure 1).
On the other side of the spectrum is neck recurrence. For an isolated neck recurrence, OS drops to below 20% at 18 months (Chung et al., 2020). Radicality in salvage neck dissection is often difficult to achieve, specifically in case of extracapsular spread amidst of fibrosis with limited or no options for re-irradiation. Adjuvant chemoradiation in the salvage setting is seldom possible because of additional induction of non-acceptable toxicity.
Oral cavity squamous cell carcinoma (OCSCC) is different in that respect since surgery (with or without radiotherapy) is the primary treatment option. Reported recurrence rates are 25%-45% and even 50% for advanced stage disease. Locoregional recurrence after salvage surgery is around 60% (Lim, Lim, Kim, Byeon, & Choi, 2010;Tam et al., 2017). Several reports have shown that presence of lymph node metastasis at time of SS and positive surgical margins are negative prognosticators (Ho et al., 2014;Matsuura et al., 2018). In case of positive surgical margins, (chemo)re-irradiation does not seem to improve OS as reported by Zenga et al. (2019) in a multi-institutional study of both OCSCC and oropharyngeal squamous cell carcinoma (OPSCC) after initial surgery with or without radiotherapy.
Oropharyngeal squamous cell carcinomas can both be treated surgically (early stage) and with primary (chemo)radiation. Due to the increasing incidence of HPV-positive OPSCC, this particular subsite has been highlighted since HPV-positive OPSCC is a biologically different disease (Ang et al., 2010). Although HPV-positive tumours generally have a more favourable outcome, still more than 10% of patients experience (loco)regional failure. HPV-positive OPSCCs do have a superior outcome in salvage surgery (Fakhry et al., 2014).

| Stage
In his report in 2000, Goodwin (Bonner et al., 2006) stated that tumour stage is a stronger prognosticator for salvage outcome than site.
Indeed, advanced stage HNC has a higher recurrence rate and warrants primary chemoradiation or extensive primary surgery with or without chemoradiation. In his prospective study, Goodwin (2000) found a 2-year disease-free survival postsalvage surgery of 73%, 67%, 33% and 22% for stage I, II, III and IV, respectively (p = .0005). Therefore, he concluded that recurrent stage was a highly significant predictor of recurrence-free survival where he could not confirm this for specific sites. The fact that stage means more than site has been supported by the majority of authors, with the important note that that data on non-laryngeal advanced stage disease are sparse (Elbers et al., 2019;Hamoir et al., 2018;Pivot et al., 2001;van der Putten et al., 2015).

| Organ preservation strategies: Influence of chemotherapy
Since the emergence of combined modality treatment, advanced stage HNC aimed at organ preservation is treated with radiotherapy combined with cisplatin in a concurrent fashion. Induction chemotherapy, for example, docetaxel, fluorouracil plus cisplatin (TPF), may be used to assess chemosensitivity and/or to reduce the radiation field. It is known that the addition of platinum-based therapy to radiation gives a survival benefit of 4%-8%. Besides this positive effect, chemotherapy also increases toxicity making SS more challenging. It has been reported that previous chemotherapy in salvage candidates for HNC is negative prognosticator. The primary choice for chemoradiation portends an aggressive course of the diseaseadvanced stage disease or high grade features-which could be predictive of a recurrence (Gillison et al., 2019). As for cisplatin used in the re-irradiation setting after SS, improved disease-free survival is reported without improvement of OS (Janot et al., 2008).

| Human papillomavirus
The incidence of Human Papillomavirus (HPV)-positive OPSCCs is increasing. A new staging system for p16-positive OPSCC has been introduced in the eight edition of the UICC/AJCC (Brierley, Gospodarowicz, & Wittekind, 2017). Since its behaviour is distinct, efforts are made to tailor primary treatment. This mostly concerns de-escalation of treatment for which the first trials have reported results regarding platinum-based superiority over cetuximab in combined modality treatment (Mehanna et al., 2019;Gillison et al., 2019). All these effort may influence outcome in the salvage surgery setting. De-escalation trials regarding adjuvant treatment are expected to report results over the next few years (Howard et al., 2018;Owadally et al., 2015). Ma et al. (2020)  Early-stage HPV + OPSCCs are preferably primarily treated surgically to attempt to avoid radiotherapy-induced toxicity. In case of clear margins of the index tumour and limited nodal involvement (single node without extracapsular spread), surveillance is sufficient.
In this group of relatively young patients, toxicity reduction is key to prevent xerostomia, dysphagia, carotid artery atherosclerosis and risk reduction for a radiotherapy-induced tumour. The introduction of TORS has improved the accessibility for oropharyngeal resection and has also taken its role in the unknown primary setting by means of tongue base mucosectomy increasing the identification rate from 40% to 80% (van Weert et al., 2020). Because of possible primary avoidance of toxicity, HPV + OPSCC patients can benefit from (adjuvant)(chemo)radiation in case of locoregional failure.

| Margins and N-status
In recurrent laryngeal cancer, surgical margins are relatively eas- Neck recurrence is correlated with poor outcome in salvage surgery, both isolated and in combination with a local recurrence. The most favourable outcome of salvage neck dissection is reported in patients with an initial treatment with surgery alone and a N1 recurrence, preferably in the undissected neck (Lim et al., 2010).

| Impact of disease-free interval
The time interval between initial treatment and recurrence (diseasefree interval) is impacting outcome of salvage surgery. The majority of recurrent HNCs are diagnosed within 18 months after initial treatment (Hamoir et al., 2018;Stell, 1991). A short DFI is a poor prognosticator. Some authors use a cut-off point of 6 months (because of definition of persistent versus recurrent disease) and found significantly different OS rates where others use 12 months (Hamoir et al., 2018;Liao et al., 2008;Lim et al., 2010;Stell, 1991). Stell (1991) reported a 20% drop in OS in case of a DFI <9 months. A short DFI may reflect aggressive disease with limited response to treatment and problematic salvage scenarios (Ho et al., 2014). Results show that MDT discussed cases, mainly stage IV patients, have superior outcome with regard to 5-year OS. MDT discussed patients were more likely to receive multi-modality treatment than non-MDT discussed patients (Friedland et al., 2011;Liao et al., 2016;Philouze et al., 2017;Pignon et al., 2000). Patient referral to a tertiary centre with MDT has reportedly led to changes in staging and treatment in up to 60% of cases (Bergamini et al., 2016). In the recurrent/ salvage setting, there is still room for improvement. Guy  Table 2 summarizes the potential positive prognosticators in SS.

| PATIENT S ELEC TI ON AND OP TIMIZ ATI ON
In addition to tumour characteristics, adequate patient selection and optimization in the pre-, per and postoperative period is paramount. to be an important factor to consider (Kim et al., 2015). Specifically

| RECON S TRUC TI ON IN SALVAG E SU RG ERY
To optimize the chances of uneventful recovery, introduction of well-vascularized tissue in a toxicity exposed area is vital. Initially, pedicled flaps were used for this purpose such as the pectoralis major flap and the latissimus dorsi flap (Ariyan, 1979). These ver-

| COMPLI C ATI ON S AF TER SALVAG E SU RG ERY
Since salvage is performed in previously treated tissue, complication rates are relatively high. Previously induced fibrosis and toxicity deteriorate healing tendency. Goodwin (Bonner et al., 2006) reported complication rates of 6% and 30% for early and advanced stage recurrences, respectively. Complication rates as high as 67% have been reported with additional risk in patients previously treated with chemoradiation or needing neck dissection. Besides perioperative complications, long-term complications as progressive fibrosis, prolonged feeding tube or tracheotomy dependence are not rare after salvage surgery (Kostrzewa et al., 2010;Nichols et al., 2011;White et al., 2013;Zafereo et al., 2009). The Clavien-Dindo classification has been adopted for head and neck surgery to ensure uniformity and reproducibility of complication registration (Dindo, Demartines, & Clavien, 2004  and shared decision-making has contributed to patient participation (Roman, Awad, & Patel, 2015). Patients are consulted accordingly in case of presumed salvage surgery. Patients not eligible for salvage surgery should be considered for participation in immunotherapy clinical trials (Gavrielatou, Doumas, Economopoulou, Foukas, & Psyrri, 2020). Although major improvements have been made in managing patients with recurrent and residual HNC, a 5-year postsalvage OS ranging from 30% to 50% nowadays is still modest. Time should be taken to extensively discuss a case both in a MDT as with the patient. Every possible prognosticator should be carefully weighed (Leemans, 2017). In case of poor general condition, short DFI, advanced stage disease and significant nodal tumour burden, a plan for salvage surgery should at least be reconsidered. The patient should be informed thoroughly so to decide whether these high stakes of salvage surgery are worth the effort and uncertainty. SS is indeed a last resort treatment with an often unpredictable outcome and will remain so in the near future.

| CON CLUS ION
Further development of predictive modelling may aid in decisionmaking in salvage surgery.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/odi.13582.