Re-Irradiation for Recurrent Head and Neck Cancer: Freedom from Cancer Recurrence Rate

Salvage re-irradiation (rRT) for patients with locoregionally recurrent head and neck cancer (rHNC) remains challenging. A retrospective analysis was performed on 49 patients who received rRT between 2011 and 2018. The co-primary endpoint of the study was 2-year freedom from cancer recurrence rate (FCRR) and overall survival (OS), and secondary endpoints were 2-year disease-free survival (DFS), local failure (LF), regional failure (RF), distant metastases (DM), and RTOG grade 3 ≥ late toxicities. Adjuvant and definitive rRT were delivered to 22 and 27 patients, respectively. A total of 91% of patients were managed with conventional re-RT and 71% of patients received concurrent chemotherapy. The median follow-up after rRT was 30 months. The 2-year FCRR, OS, DFS, LF, RF, and DM were 64%, 51%, 28%, 32%, 9%, and 39% respectively. MVA showed that poor performance status (PS: 1–2 vs. 0) and age > 52 years were predictive of worse OS. In comparison, poor PS (1–2 vs. 0) and total dose of rRT < 60 Gy were predictive of worse DFS. Late RTOG toxicity of grade 3 ≥ was reported in nine (18.3%) patients. FCRR at 2 years after salvage rRT for rHNC was higher than other traditional endpoints and could be an important endpoint to be included in future rRT studies. rRT for rHNC at our cohort was relatively successful, with a manageable level of late severe toxicity. Replacing this approach in other developing countries is a viable option.

Management of LRR arising in a previously irradiated volume has always been a complex clinical and dosimetric situation in the context of old radiation techniques. The Radiation Therapy Oncology Group (RTOG) 9610 and 9911 had reported a higher rate of severe rRT-related morbidity, with approximately 8% of patients experiencing treatment-related deaths (for different reasons, e.g., fatal hemorrhage, febrile neutropenia, dehydration, shock, and pneumonitis). The use of historical rRT techniques for recurrent head and neck cancer (rHNC) was associated with low survival rates, with a 2-year overall survival rate (OS) of 15-26% [11,12]. However, a large multi-institutional cohort study using newer rRT such as intensity-modulated radiation therapy (IMRT) showed a relatively low rate of grade 4 acute toxicity (5.1%) and a higher 2-year OS (35.4%) compared to previously reported historical data [9,13].
Morbidity in rHNC is primarily caused by cancer recurrence and/or treatment-related toxicity. The advantages of rRT should outweigh the associated risks of such treatment. Patients require a multidisciplinary evaluation to determine the best management strategy [13,14]. Salvage surgery is the preferred choice for resectable cases, either with or without adjuvant rRT (+/− concurrent chemotherapy) as indicated by postoperative histopathological findings. Unfortunately, many recurrences present in the advanced T-and N-categories and are not surgically resectable. In unresectable or medically inoperable cases, definitive rRT with or without concurrent chemotherapy is the appropriate treatment option in carefully selected patients [9,11,13]. rRT can be administered via IMRT, stereotactic body radiation therapy (SBRT) whenever possible, or proton therapy whenever available [9,15,16].
The previous studies mainly focused on traditional outcomes such as OS, DFS, LF, RF, and DM. However, these endpoints do not adequately reflect the success of salvage rRT. Hence, in this study, we assessed the freedom from cancer recurrence rate (FCRR), which evaluates the success of salvage rRT. Additionally, we present the traditional oncologic outcomes of rRT for patients with rHNC at our institution.

Study Population
After institutional research board (IRB) approval (IRB No. 21 KHCC 189), we identified all patients with rHNC who had pathologically confirmed non-metastatic recurrent squamous cell carcinoma (SCC) of the head and neck (H&N) and who had previously received RT to the H&N region at a total dose of ≥44 Gy. These patients were staged according to the seventh edition of the TNM staging system jointly used by the American Join Committee on Cancer (AJCC) and Union of the International Cancer Control (UICC). Patients salvaged with curative-intent rRT at our institution between 2011 and 2018 were included in this retrospective analysis. Patients younger than 18 years, and those with histopathology other than SCC were excluded from this analysis. The patients' demographics and clinical information including outcomes were retrospectively collected from the patients' medical records.

Diagnostic Approach
Re-staging work up and pre-salvage treatment evaluation consisted of a comprehensive physical examination including assessment with fiberoptic endoscopy. Imaging evaluation included H&N MRI and PET/CT scans. After completion of re-staging work up, all patients were discussed and managed by a multidisciplinary H&N team, with evaluation by dedicated teams of dental oncologists, nutritionists, and speech/language pathologists prior to the initiation of rRT.

Treatment Approach
All patients who developed local and/or regional recurrence were evaluated by a HN surgeon for possible salvage surgery. Adjuvant rRT was considered for pT3/4, close resection margin(s), and/or multiple involved lymph nodes, while adjuvant concurrent chemotherapy (with rRT) was considered for involved resection margin(s) and/or pathologic extranodal extension (pENE). Patients with unresectable tumors were managed with definitive rRT with or without concurrent chemotherapy.
The dose and fractionation schedules of the rRT regimens were selected based on institutional guidelines. In general, conventional fractionation (1.8-2 Gy/fraction) rRT was used with a minimum total prescribed dose of 44 Gy and a maximum total prescribed dose of 70 Gy (whenever possible) according to: (1) intent of rRT (adjuvant vs. definitive), (2) the degree of overlap with previous radiation, (3) proximity to critical organs-at-risk (i.e., neuro-ocular structures), and (4) time interval between the two courses of radiation. At the discretion of the treating radiation oncologist (after a comprehensive review of the old radiation plan), a hyperfractionation rRT schedule (for more complex cases) or a hypofractionation rRT regimen (for less complex cases with small volume tumors far away from critical structures) could be used. rRT was delivered using IMRT. Concurrent chemotherapy (in the definitive or adjuvant setting) consisted of weekly cisplatin (40 mg/m 2 weekly) or carboplatin (area under the curve [AUC] of 1.5 weekly) during rRT.

Post-Treatment Evaluation and Follow-Up
In general, patients were reviewed in the radiation oncology clinic 2 weeks after the end of rRT, then every 3 months for the first 2 years, every 4 months in the third year, every 6 months in the fourth and fifth year, and annually thereafter until death. Post-treatment imaging to evaluate the response to therapy included H&N MRI and PET/CT scans, which were performed 10-12 weeks after the end of rRT, then as clinically indicated. Severe late rRT-related side effects were defined as late RTOG grade ≥ 3 toxicity starting > 6 months after the end of rRT.

Statistical Methods
The co-primary endpoint of the study was 2-year FCRR and OS, and secondary endpoints were 2-year local failure (LF), regional failure (RF), distant metastases (DM), disease-free survival (DFS), and late toxicities. OS and DFS were analyzed using the Kaplan-Meier method and compared using the log-rank test. LF, RF, and DM rates were estimated using the cumulative incidence method using Fine Gray's test, with death (without disease recurrence) as a competing risk. FCRR was estimated by the competing risk method (LF, RF, and DM are events, while death without LF, RF, and DM was considered a competing factor). Late toxicity rates were estimated by the cumulative incidence function. Multivariable analysis (MVA) using Cox proportional hazards regression was used to identify predictors of OS and DFS. All reported p-values were two-sided, with a statistical significance level of p ≤ 0.05. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA), and the figures were created using GraphPad PRISM 7.
Five patients developed RF (adjuvant rRT: n = 1 vs. definitive rRT: n = 4) at the median time of 7 (range: 2-72) months post-rRT. Of whom, one patient had LF and three patients had DM at the time of RF, and were treated with palliative chemotherapy.

Toxicity Outcomes
The grade 3 ≥ late RTOG toxicity was reported in nine (18.3%) patients. This included grade 3 dysphagia (n = 5), which necessitated the placement of a feeding tube 6 months after the end of rRT, grade 3 osteoradionecrosis (n = 3), grade 3 brain necrosis (n = 2), grade 3 neck fibrosis (n = 2), and grade 5 carotid blowout (n = 1). Out of 49 patients studied, 10 (20.4%) patients experienced grade 2 RTOG toxicity related to late dysphagia, while 20 (40.8%) patients experienced grade 2 RTOG toxicity related to late xerostomia. The cumulative prescribed BED 10 for patients who developed grade 3 ≥ late RTOG toxicity ranged between 91 Gy 10 and 121 Gy 10 as shown in Table 3. The patient, who developed carotid blowout, was a 57-year-old at the time of diagnosis of a recurrent unresectable regional recurrence on the background of previously irradiated T1N0M0 glottic cancer (originally received 63 Gy in 28 fractions using conventional technique with two opposing lateral fields). The unresectable neck recurrence involved the right neck nodal levels II-IV, the cumulative BED 10 was 98 Gy 10 , and the interval between original RT and rRT was 18 months.

Discussion
This paper introduces FCRR as an endpoint for rRT in rHNC. To our knowledge, this endpoint was not used previously in clinical trials. The significance of FCRR lies in the fact that part of the primary cause of morbidity in rHNC is cancer recurrence and treatmentrelated toxicity. Previously, death was used as an event in calculating DFS, potentially minimizing the impact of cancer recurrence on morbidity and quality of life (QoL). FCRR, on the other hand, focuses solely on the cancer recurrence rate. Our study found that 64% of patients had a FCRR at 2 years after rRT, which was higher than other outcomes such as OS, DFS, LF, RF, and DM. The relationship between FCRR and QoL in rHNC patients requires further exploration. This study indicates a substantial proportion of non-metastatic rHNC patients can achieve disease control and survival after rRT using the IMRT technique. These findings align with the results of previous studies [9,10,[17][18][19]. In a study by the University of Texas MD Anderson Cancer Center, they reported that after 2 years of receiving rRT with IMRT, 58% and 64% of 78 patients achieved OS and locoregional control, respectively [19]. Lee et al. reported 2-year PFS and OS rates of 30.9% and 54.6%, respectively, in 42 HNC patients who underwent salvage rRT for non-metastatic locoregional recurrence [18].
This study found that there was no significant difference in OS between patients who received adjuvant or definitive rRT for rHNC. This suggests that salvage surgery did not provide a survival benefit for this patient population. The study's outcome contradicts earlier research by the Multi-institution Radiotherapy Collaborative Group (MIRI), which showed a 2-year OS rate of 45% and 36% for postoperative and definitive rRT, respectively [13]. The Korean Radiation Oncology Group, in the study (KROG 1707), also reported that salvage surgery for rHNC was associated with superior OS (p = 0.002) [17]. The difference in results may be due to several factors including: optimal patient selection at a specialized cancer center, a larger number of patients in our study with NPC (known to be highly responsive to radiotherapy and chemotherapy, and high salvage success rates with rRT) [20,21], and more advanced T4-category among patients who underwent surgery compared to those who received definitive rRT (55% n = 12 vs. 33% n = 9, respectively). Furthermore, too many treatment regimens were used in our study that weaken the generalizability of our rRT endpoints. However, the data regarding rRT is sparse in the literature. Several studies show institutional experience of relatively very small numbers of patients [18,19]. There is no consensus recommendation regarding the actual regimen that can be used16. In addition, every case represents a challenging situation when constraints for critical organs at risk can be met. All these reasons make it different to generalize the conclusion for retrospective studies of rRT (including our study); however, it provides some practical guidance for managing rHNC cases.
In this study, the factors that were found to be associated with better OS included good PS (ECOG 0) and young age (below 52 years). Furthermore, the study results showed that improved DFS was observed in patients who had a PS of ECOG 0 and received a rRT dose > 60 Gys [14]. A higher radiotherapy dose was typically associated with improved local control of rHNC. Our results were similar to Roesch et al. who conducted a large multicenter analysis of dose-escalated rRT for rHNC in Germany. His study included 253 patients treated at 16 university hospitals. The results showed that patients with good ECOG PS and rRT doses above 50 Gy (EQD2) had median longer OS (17.8 months vs. 11.7 months, p < 0.01) and PFS (9.6 months vs. 6.8 months, p < 0.01) compared to those with poor ECOG PS and rRT below 50 Gy (EQD2) (p < 0.01) [22]. These findings provide important information for clinicians who are considering rRT treatment options for patients with rHNC [9,13,14].
Although severe late toxicity from salvage treatment of rHNC is a major concern, not providing treatment can also have severe consequences, as rHNC can cause significant morbidity and mortality [3]. In our study population, severe grade 3 ≥ late toxicity was seen in 18.3% of the cases, which is comparable to the 16.3% rate reported by the MIRI collaborative group [23]. These findings emphasize the importance of carefully weighing the potential benefits and risks of salvage rRT for rHNC.
The limitations of this paper include its retrospective design, limited sample size, and the diverse primary site locations of the rHNC tumors. Despite our findings, it is advisable to exercise caution when promoting the use of salvage rRT for rHNC patients. A multidisciplinary approach to managing rHNC remains the recommended course of action in these complex cases.

Conclusions
The 2-year FCRR after salvage rRT for rHNC was superior to other traditional endpoints. Given that the morbidity of rHNC is largely linked to locoregional recurrence regardless of survival, FCRR may serve as a new endpoint to consider in rRT studies for rHNC. For studies with a small sample size, FCRR may serve as a more suitable endpoint. The rate of late severe toxicity was acceptable.