Salvage radiotherapy for recurrent hypopharyngeal and laryngeal squamous cell carcinoma (SCC) after first-line treatment with surgery alone: a 10-year single-centre experience

Purpose Salvage surgery of recurrent hypopharyngeal and laryngeal squamous cell carcinoma (SCC) results in limited local control and survival rates. As a result of recent technological progress, radiotherapy (RT) has become a valuable, potentially curative therapeutic option. Thus, we aimed to determine prognostic factors for survival outcome in order to optimize patient selection for salvage radiotherapy after failure of first-line treatment with surgery alone in this special patient cohort. Methods Seventy-five patients (85% male, median age of 64 years) underwent salvage RT in a secondary setting for recurrent hypopharyngeal or laryngeal SCC after prior surgery alone between 2007 and 2017. On average, patients were treated with one prior surgery (range 1–4 surgeries). Median time between surgery and salvage RT was 7 months (range 1–47 months) for initially advanced tumors (T3/4, N+, extracapsular spread) and 18 months (range 5–333 months) for initially early stage tumors. The majority of patients received concomitant chemotherapy (n = 48; 64%) or other kind of systemic treatment concurrent to radiotherapy (n = 10; 13%). Results Median follow-up was 41 months (range 3–120 months). Overall, fifteen patients were diagnosed with local failure (all were in-field) at last follow-up (20%). Median time to recurrence was 35 months (range 3–120 months) and 3-year local progression-free survival (LPFS) was 75%, respectively. Dose-escalated RT with 70.4 Gy applied in 2.1 Gy or 2.2 Gy fractions corresponding an EQD2 > 70 Gy (p = 0.032) and the use of concomitant cisplatin weekly chemotherapy (p = 0.006) had a significant positive impact on LPFS. 3-year OS and DPFS were 76 and 85%, respectively. No toxicity-related deaths occurred. Reported grade > 3 side effects were rare (n = 4/70, 6%). Conclusion Salvage radiotherapy resulted in excellent local control rates while radiation dose and the use of cisplatin weekly chemotherapy were identified as prognostic factors for LPFS. Nevertheless, patient selection for curative salvage treatment remains challenging.


(Continued from previous page)
Conclusion: Salvage radiotherapy resulted in excellent local control rates while radiation dose and the use of cisplatin weekly chemotherapy were identified as prognostic factors for LPFS. Nevertheless, patient selection for curative salvage treatment remains challenging.
Keywords: Squamous cell carcinoma, Recurrent hypopharynx and larynx carcinoma, Salvage radiotherapy, Function preservation Background Squamous cell carcinoma (SCC) of the larynx and hypopharynx is the most common tumor in the head and neck region, mainly observed in males over 50 years. Especially glottic larynx carcinomas are mostly diagnosed in early stages due to the disorder of essential laryngeal and hypopharyngeal functions presenting with initial symptoms, i.e. hoarseness, swallowing difficulties or dyspnea, resulting in significant reduced patient′s satisfaction and quality of life [1][2][3]. Therefore, besides achieving optimum local control, preservation of phonatory and swallowing function gain in importance. The role of chemoradiotherapy (CRT) in the organ-preserving treatment of hypopharyngeal and laryngeal malignancies was established by two important landmark trials, showing equal survival rates compared with surgery and a high rate of larynx-preservation in twothirds of the patients [4,5]. Over the last two decades, CRT is increasingly considered as a valuable alternative to total laryngectomy for advanced tumors. Nowadays, total laryngectomy is mostly used in highly selected patients with advanced diseases or reserved as salvage surgery in case of treatment failure after primary CRT. In the current literature, poor prognosis is reported for patients who were treated with salvage surgery after radiotherapy failure in several studies [6,7]. However, data concerning outcome of salvage CRT after failed first-line surgical treatment are still missing. The aim of this retrospective analysis is to assess clinical outcome in patients with recurrent hypopharyngeal and laryngeal SCC after first-line treatment with surgery alone, who received second-line RT in potentially curative intention and to determine prognostic factors for survival outcome to optimize patient selection for salvage radiotherapy.

Materials and methods
Evaluation Seventy-five patients with laryngeal or hypopharyngeal SCC treated for recurrence after prior surgery at the Department of Radiation Oncology of the University Hospital Heidelberg between 2007 and 2017 were identified retrospectively and patient′s records were analysed regarding local progression-free survival (LPFS), overall survival (OS) and distant progression-free survival (DPFS). Additionally, potential prognostic factors were assessed and calculated for local control (LC), OS and DPFS.
Follow-up was performed with magnetic resonance imaging (MRI) or computed tomography (CT) 6 weeks after completion of therapy, at three-month intervals during the first 2 years after treatment, every 6 months during the third year after treatment and then, once a year. Yearly CT scans of the chest and abdominal ultrasound were performed to identify distant relapse.
Acute and late toxicity were assessed according to the Common Terminology Criteria for Adverse Events version 4.03 (CTCAE v4.03) and tumor response to the current Response Evaluation Criteria in Solid Tumors (RECIST) [8,9]. TNM (tumor, nodal, metastasis) stage was assessed and adjusted to the eight edition of the TNM classification system [10].
Survival data were calculated from first diagnosis to the date of last follow-up, death or progression by using Kaplan-Meier estimates (IBM SPSS Statistics version 24). OS was calculated from the first diagnosis up to the last follow-up or death. LPFS and DPFS were considered as the time period between first diagnosis and occurrence of local progression, distant progression or death. LC was calculated from first day of treatment to last follow-up or local progression. Univariate analysis to identify potential prognostic factors for survival outcome were performed using the log-rank test. For multivariate analysis, the cox regression model was used. All tests were 2-tailed and the significance level was defined as ∝ < 0.05.

Patient characteristics
All patients received RT in a secondary setting for recurrence after first-line treatment with surgery alone (median number of operations 1, range 1-4). Patients who received a prior radiation treatment were excluded from the study. Initially, postoperative RT (n = 13, 17%) for UICC stage III/IV or CRT (n = 22, 29%) for incomplete resection margin (n = 11, 15%), lymph node metastases with extracapsular spread (ECS; n = 9, 12%) or both (n = 2, 3%) was obligatory in overall 47% of the patients but was declined from the patient. Therefore, two treatment groups were differed within the study population; early stage patients without an initial indication for postoperative RT (n = 40, 53%) and patients with advanced tumors who declined postoperative RT after first-line surgery (n = 35, 47%). Prior total laryngectomy was performed in 24% of the patients (n = 18), prior unilateral neck dissection (ND) in 9% of the patients (n = 7) and bilateral ND in 35% of the patients (n = 26). In 79%, lymph node metastases could be identified (n = 26), in 11/26 patients with ECS (for detailed treatment characteristics please see Table 2).

Treatment characteristics
For treatment planning, CT and MRI were performed and patients were immobilized with custom-made thermoplastic masks with shoulder fixation. Clinical target volume 1 (CTV1) including the macroscopic tumor and the potential microscopic spread and CTV2 including the CTV1 and the lymphatic drainage were outlined. Planning target volumes (PTVs) were generated with a margin of 5 mm around the CTVs and received at least 90% of the prescription dose. All patients received intensity modulated radiotherapy (IMRT) via tomotherapy with (n = 70, 93%) or without simultaneous integrated boost (SIB, n = 5, 7%) at a median time of 7 months after the first operation (range 1-333 months). Single doses and prescription doses differed, thus we calculated the equivalent dose in 2 Gy per fraction with the following formula for better dose comparibility: EQD2 = D x ((d + α/β)/(2 + α/β)) (D = total dose in Gy, d = single dose in Gy, α/β = 2).
The median prescribed total dose was 54 Gy (range 52.2-66 Gy) with a median single dose fraction of 1.8 Gy (range: 1.8-2.0 Gy) to the bilateral cervical lymph drainage and 66 Gy (range 58-72 Gy) with a median single dose fraction of 2.2 Gy (range 1.8-2.2 Gy) to the macroscopic tumor. The median equivalent dose to a 2 Gy single dose fraction (EQD2) prescribed on the macroscopic tumor was 67 Gy (range 60-73 Gy). In 5 cases, treatment could not be finished (7%). Thus, these patients received smaller cumulative doses. The median PTV1 was 101 cc (range 21-222 cc) and the median PTV2 was 740 cc (range 51-1301 cc). Overall, 77% of the patients were treated with concomitant systemic therapy (n = 58) of whom 64% received concomitant chemotherapy (n = 48) with carboplatin/5-fluoruracil in week 1 and 5 (n = 4) or cisplatin weekly (n = 44) and 13% concomitant immunotherapy with cetuximab weekly (n = 10). Treatment characteristics are depicted in Table 2.

Prognostic factors for survival
We analysed the impact of potential prognostic factors at time of first-line treatment with surgery alone (initial T, N, G stage, ECS, tumor site, Karnofsky performance status (KPS), prior laryngectomy vs. larynx-preserving surgery, prior ND yes vs. no, time period between first operation and first recurrence) and at time of second-line treatment with salvage RT (recurrent T, N, G stage, KPS, number of prior operations, EQD2 > 70 Gy vs. ≤70 Gy, PTV1 ≥ 101 cc vs. < 101 cc, cisplatin weekly chemotherapy vs. others, local recurrences yes vs. no after RT and metastases yes vs. no) on OS, LPFS and DPFS. Several potential prognostic factors were identified using the log-rank test for univariate analyses. Independent prognostic factors were assessed by the cox regression model for multivariate analysis. The results of univariate and multivariate analysis are shown in Table 3.
Regarding LPFS, we could identify the use of concomitant chemotherapy with cisplatin weekly (p = 0.006) and an EQD2 > 70Gy prescribed on the macroscopic tumor as positive prognostic factors (p = 0.032). Patients who received concomitant systemic therapy were included into analysis only. The use of concomitant chemotherapy with cisplatin weekly resulted in a 5-year LPFS of 86% vs. 44% compared with patients who received concomitant chemotherapy with carboplatin/5-FU in week 1 and 5 or cetuximab weekly (Fig. 1). With regard to the applied EQD2 on the macroscopic recurrent tumor, patients who received an EQD2 > 70 Gy had a 5-year LPFS of 90% vs. 68% compared with patients who received an EQD2 ≤ 70 Gy (Fig. 2).  Patients with local or locoregional recurrence after salvage RT had a significant worse OS with a 5-year OS of 25% vs. 74% estimated from Kaplan-Meier analysis compared with patients who were free from recurrence at last follow-up (p = 0.041), respectively. Furthermore, recurrent tumor site showed an estimated impact on patients′ OS. Thus, patients with glottic recurrence had a survival benefit with a 5-year OS rate of 89% compared with patients who had recurrences in the area of the supraglottic or subglottic larynx, hypopharynx or neopharynx with a 5-year OS rate of 25% (p = 0.05). Additionally, initial N stage seemed to have a further impact on OS with a survival benefit for patients without initial lymph node metastases, but in the multivariate analysis, the significance level could not be achieved (p = 0.06).
In terms of DPFS, recurrent T stage showed the most significant impact on DPFS with a 5-year DPFS of 90% for rT1/2 tumors and 70% for rT3/4 tumors (p = 0.028). Further, PTV1 (≥ median of 101 cc vs. < median of 101 cc) had a worse impact on DPFS with increasing volume (p = 0.038). At least, initial N stage could be diagnosed as further prognostic factor for DPFS with a significant

Toxicity
Overall, 20% of the patients reported acute grade 3 (n = 14) and 19% of the patients chronic grade 3 and grade 4 toxicity (n = 13). During and six weeks after therapy, no acute grade 4 toxicity could be identified. An overview of acute and late side effects is shown in Table 4. Acute grade 3 toxicity consisted of mucositis (n = 2, 3%), dysphagia (6%, n = 4), odynophagia (n = 5, 7%), dermatitis (n = 4, 6%), xerostomia (n = 1, 1%), hoarseness (n = 3, 4%) and lymphedema (n = 1, 1%). Three months after therapy, the majority of the acute grade 3 side effects disappeared. Nevertheless, 3 patients received tracheostomy for acute dyspnea due to grade 4 lymphedema of the larynx (4%) and in one patient a laryngoesophageal fistula could be diagnosed 6 months after therapy (1%). 3 patients claimed about chronic severe dysphagia with high-grade stenosis of the laryngoesophageal junction with the need of regular bougienage up to the last follow-up (4%). Further, one patient developed a laryngocutaneous fistula 12 months after RT (1%) and one patient developed a wound healing disorder in radiation field 24 months after therapy (1%). Under RT, 43% of the patients needed supportive nutrition for nutritional difficulties due to acute side effects (n = 30). Acute gastric tube dependence counted 34% (n = 24). 3 to 6 months after therapy, only 13% of the patients needed farther a gastric tube for nutrition (n = 9). Only one patient was dependent on a gastric tube for more than 2 years (1%).

Findings
The majority of the patients in the current analysis were treated with dose-escalated RT (93%) and concomitant cisplatin weekly chemotherapy (60%) for recurrent laryngeal or hypopharyngeal SCC after firstline treatment with surgery alone. Two thirds of the patients had tumors in advanced stages (UICC III and IV) before treatment. Nevertheless, we identified an excellent 3-year LPFS, OS and DPFS of 75, 76 and 85% and an estimated 5-year LPFS, OS and DPFS of 75, 64 and 82%, respectively. The 5-year local control rate was 79% for this unfavourable patient population. All recurrences occurred in-field and within the first two years after treatment. Larynx-preservation could be observed in 90% of the patients who received organ-preserving surgery as initial treatment. Local control was best achieved in patients receiving an EQD2 > 70 Gy on the macroscopic tumor and concomitant chemotherapy with cisplatin weekly. RT dose and the use of cisplatin chemotherapy did not correlate with patients' KPS (p = 0.139). OS depended negatively on the occurrence of local recurrences after salvage RT and on the recurrent tumor site before RT with survival benefit for recurrent tumors involving the glottis. DPFS differed significantly regarding recurrent tumor size and the initial nodal stage. Overall, compliance was well (93%) and RT was tolerated with moderate toxicity.

Survival results
Nowadays, CRT is seen as an equivalent therapy option to surgery regarding tumor control in laryngeal and hypoharyngeal tumors [11,12]. Thus, Mendenhall et al. reported 2001 RT alone for T1/2 N0 larynx tumors as an alternative curative therapy option to surgery with a comparable 5-year LPFS of 72 to 94% depending on T stage, a 5-year OS of 79% and a 5-year DPFS of 98% [13]. Nevertheless, surgery in form of organ-preserving endoscopic resection, laser surgery or open-neck partial laryngectomy is still considered as the gold standard in the treatment of early stage tumors with local control rates of 60 to 95%, declining with increasing T stage [12]. While local relapses are relatively rare after primary treatment of early stage tumors, local control in advanced hypopharyngeal and laryngeal tumors remains a challenge. As organ-preserving treatment strategies are increasingly used as first-line treatment, the use of organ-preserving RT in combination with chemotherapy for advanced stages gained in importance within the last decades [5,13,14]. However, 40 to 60% of patients with advanced tumors relapse after primary CRT [5,15]. Salvage treatment options for these patients are limited, as re-irradiation is mostly limited by the tumor site and the necessary prescription dose to the recurrent tumor. In these cases, salvage surgery is mostly required. The effectiveness of salvage surgery after failure of primary CRT (2-year OS between 27 and 71%) is reported by several studies [16][17][18]. Taguchi et al. reported a 5-year OS and disease-specific survival of 61 and 66% for salvage surgery after primary CRT vs. 10 and 10% for patients who failed primary CRT but did not undergo salvage surgery [18]. While poor prognosis with a 5-year OS of 16% is described for recurrent hypopharyngeal tumors after salvage  surgery, the same treatment method offers 5-year OS rates ranging from 57 to 70% for recurrent laryngeal tumors [19][20][21]. Recurrent hypopharyngeal tumors are mostly considered inoperable and should, therefore, be treated with alternative salvage methods [22]. Salvage CRT is generally reserved for recurrent tumors after primary total laryngectomy, for patients with inoperable recurrent tumors or in cases, where the patient rejects total laryngectomy as the only remaining treatment option in order to further preserve the organ function. Patients who receive salvage CRT mostly appear with initially early stage tumors and represent a non-comparable patient collective to patients who receive salvage surgery after failure of primary CRT. Nevertheless, studies describing a homogeneous patient population after salvage CRT including recurrent hypopharyngeal and laryngeal tumors only are lacking. Lee

Prognostic factors
Several meta-analyses and randomized studies have proven the beneficial role of concomitant chemotherapy in combination with RT, especially regarding cisplatin chemotherapy, showing superior local control and OS rates in laryngeal/hypopharyngeal tumors as well as in other HNC either in the primary or postoperative setting [5,15,[25][26][27]. Besides an EQD2 > 70 Gy achieved by using fractionation doses >2Gy, we could identify a significant impact of cisplatin chemotherapy on LPFS only [28,29]. Nevertheless, Pignon et al. could show in the MACH-NC meta-analysis that chemotherapy with carboplatin and 5-fluoruracil is considered to be equivalent to cisplatin chemotherapy. The results of the current study could possibly be explained by patient selection bias between both groups [30]. In the current literature, further factors influencing LPFS, i.e. T stage, N stage, G stage, sex, age, vocal cord invasion, overall treatment time and RT field size are discussed in the first-line treatment [13,[31][32][33][34]. Glottic tumors seem to have a survival benefit compared with other tumor sites, thus in several studies superior OS rates are described [17][18][19][20]. The occurrence of local recurrence after treatment, tumor size and N stage are accepted as further prognostic factors [31][32][33][34][35]. For second-line treatment, initial tumor in the hypopharynx vs. the larynx, recurrent tumor in the hypopharynx vs. the larynx, advanced primary tumor, advanced recurrent tumor, advanced primary N stage and advanced initial and recurrent G stage were most frequently associated with decreased progression-free survival resulting in a poor prognosis [23,24,36]. In multivariate analysis, we could not identify an impact of tumor size and N stage on OS, possibly due to the low patient number but on DPFS. Thus, tumors with initial N+ stage and increased recurrent tumor size (PTV1, rT stage) influenced distant control negatively [34]. Additionally, tumor differentiation (G stage) can be considered as further prognostic factor regarding DPFS [31].

Toxicity
Superior dose conformity and decreased toxicity due to improved preservation of organs at risk compared with 3D-RT is described for IMRT [37][38][39][40]. Nevertheless, toxicity after IMRT remains high [41][42][43][44][45]. Especially in the hypopharynx and larynx region, significant higher rates of late side effects occur due to the proximity of several organs at risk compared with other regions of the head and neck [45]. For patients treated with concurrent CRT, Forastiere et al. could show in the RTOG trial 91-11 a high rate of severe late grade 3 and 4 toxicity, especially regarding mucositis (43%) [44]. In a RTOG analysis of three RTOG trials (RTOG 91-11, RTOG 97-03, RTOG 99-14), Machtay et al. identified 43% late side effects after CRT for locally advanced squamous cell carcinoma of the head and neck, and described grade 3 and 4 pharyngeal and laryngeal dysfunction in 39%, gastric tube dependence longer than 2 years in 13% and treatment-related death within 3 years in 10% of the patients. Swallowing limitations, aspiration, laryngoesophageal stricture and dysphagia dominated regarding laryngoesophageal dysfunction [45]. Caudell et al. reported a 3-year laryngoesophageal dysfunction-free survival (LEDFS) of 32% for patients who were treated with CRT for advanced SCC of the larynx and hypopharynx [42]. For salvage treatment methods as well, high toxicity rates are reported. Several authors described complication rates between 44 and 59% for salvage surgery after first-line CRT [16,17,46]. A systematic review and meta-analysis of the complications of salvage total laryngectomy including 3293 patients by Hasan et al. showed a complication rate of 68% with 29% pharyngocutaneous fistula [47].In the current analysis, in contrast, only 19% of the patients claimed about severe chronic toxicity. Especially laryngoesophageal dysfunction consisting of severe dysphagia (4%), fistula (3%), dyspnea (4%) or wound healing disorder (1%) were observed. Only one patient showed feeding tube dependence over 2 years after RT (1%). We could not identify any treatment-related deaths within the follow-up time. Despite first-line surgery and second-line dose-escalated RT, we identified moderate toxicity rates making salvage CRT after failed surgery a safe therapy alternative.

Conclusion
Salvage radiotherapy is an effective curative therapy option for recurrent hypopharyngeal and laryngeal SCC after prior surgery with excellent local control rates and moderate toxicity comparable to prior results concerning primary radiotherapy. We recommend dose-escalated IMRT with an EQD2 > 70 Gy as well as the use of concomitant cisplatin weekly chemotherapy for superior LPFS. Nevertheless, patient selection for curative salvage treatment remains challenging.