Efficacy of immune checkpoint inhibitor treatment for head and neck mucosal melanoma recurrence in patients treated with carbon‐ion radiotherapy

Abstract Background Carbon‐ion radiotherapy (C‐ion RT) is effective for head and neck mucosal melanoma (HN‐MM), including radioresistant mucosal melanoma. Melanoma also responds effectively to immune checkpoint inhibitors (ICIs). Data on the efficacy and safety of ICIs for HN‐MM are insufficient. Aims To analyze the efficacy and safety of ICI salvage therapy in patients with HN‐MM recurrence after C‐ion RT. Methods and Results This retrospective study analyzed the medical records of 52 patients with HN‐MM treated with C‐ion RT between 2012 and 2020. A dose of 57.6 or 64.0 Gy (relative biological effectiveness) was provided in 16 fractions. The primary endpoint was 3‐year overall survival (OS) rate. The median follow‐up time was 26.8 months for all patients. A total of 29 patients had local recurrence or distant metastasis, and 16 patients who received ICI therapy. The 3‐year OS rate in the ICI group (n = 16) and best supportive care group (n = 13) were 53.8% and 0.0%, respectively (p = 0.837); the difference was not statistically significant. There were no deaths after 1 year among patients who underwent ICI therapy. No adverse events associated with C‐ion RT were related to or exacerbated by ICI. Conclusion ICI salvage therapy is effective and safe for patients with HN‐MM recurrence after C‐ion RT.


| INTRODUCTION
The United States National Cancer Database reported a cutaneous melanoma incidence of over 90%, with mucosal melanomas accounting for 1.3% of onsets at the head and neck. 1 Head and neck mucosal melanomas (HN-MM) have a very poor prognosis because of distant metastases, with a 5-year overall survival (OS) rate of approximately 20-30% after surgery, radiotherapy (RT), and chemotherapy. [1][2][3][4] Surgery and postoperative RT are the mainstays in treatment in most cases of HN-MM. 2,3 However, surgery is not feasible in all cases, owing to wide resection margins and difficulty in reconstruction, which could result in cosmetic and functional challenges that impact the patient's quality of life. 5 Mucosal melanoma has been considered radioresistant; however, there is evidence supporting the benefits of RT. A retrospective multicenter study by the Japan Carbon-Ion Radiation Oncology Study Group (J-CROS) reported a 5-year OS rate of 44.6% and a local control (LC) rate of 72.3% in patients with HN-MM after carbon-ion (Cion) RT, indicating that it is one of the most effective treatment modalities. 6 In a prospective study, HN-MM treated with C-ion RT and dacarbazine, nimustine, and vincristine (DAV) had a 3-year OS rate of 49.2% and an LC rate of 92.3%. C-ion RT had a reproducible effect on HN-MM; however, the most distant metastases were recurrent, with an approximately 1-year median duration of recurrence.
There were no significant cases with progression-free survival (PFS) and OS after the completion of 3 cycles of therapy with or without DAV. 7 The previous study showed that cancers, such as melanoma and lung, had the highest prevalence of somatic mutation and responded best to immune checkpoint inhibitors (ICIs). 8 Advanced melanoma was shown to have a 5-year OS of 52% in a treatment group of nivolumab plus ipilimumab, 44% in a treatment group of nivolumab, and 26% in a treatment group of ipilimumab. 9 Furthermore, ICIs and RT are thought to play a significant role in the activation of immune responses. 10 At present, recurrence of HN-MM after C-ion RT is treated with ICIs. However, both of their efficacy and safety are insufficient. Therefore, we arranged a retrospective study to evaluate the safety and efficacy in recurrent HN-MM after C-ion RT. F I G U R E 1 Flow chart of patients with mucosal malignant melanoma treated with carbon-ion radiotherapy.

| Carbon-ion RT procedure
The C-ion RT technique and treatment planning methods used in this study were previously reported. 7,11,12 Patients in personalized cradles (Moldcare; Alcare, Tokyo, Japan) were immobilized using thermoplastic shells (Shellfitter; Kuraray, Osaka, Japan) and positioned. A personalized mouthpiece immobilized the mandible. 13 CT simulation (thickness of 2 mm) was used for C-ion RT planning and MRI for baseline reference. The XiO-N system (Elekta, Stockholm, Sweden) was performed for treatment planning. Contouring for the gross tumor volume (GTV) was used as a reference contrast-enhanced MRI image.
The clinical target volume (CTV) criterion is a 5 mm margin around the GTV for all directions and used as reference of anatomical structure.
The planning target volume (PTV) had 2 mm margins around the CTV.
The target volumes of the clinical and C-ion RT planning margins were modified as necessary when the targets were close to the organs at risk (OAR), such as optic nerve, eye, brain, and brain stem. CTV and PTV were modified to protect OAR. The modification rate was 80.8% (42 of 52 cases). The nasal cavity and paranasal sinuses tended to be invaded to the proximity of the above OARs, which accounted for the majority of cases of correction. Calculations of the physical dose were based on the algorithm of the pencil-beam, and those of the clinical dose distribution were performed using the physical dose and the relative biological effectiveness (RBE), used as reference of human submandibular gland cells. 14 The C-ion RT dose was expressed as "Gy

| Follow-up
For the first 6 months, follow-ups for patients were every month, and every 3 months thereafter. Medical imaging, such as CT and MRI, were performed alternately every 2-3 months, and F-18 fluorodeoxyglucose positron emission tomography/CT was performed every year.
All patients, including those who developed lymph node or distant metastases, were assessed for LC during follow-up until death.

| ICI therapy
Treatment after C-ion RT was not prescribed. However, individuals who received ICIs at recurrence were identified through retrospective medical record review. did not. In Japan, ICI agents were approved for the treatment for melanoma patients in 2014.

| Endpoints
Three-year OS was the primary endpoint. LC rate, PFS, and adverse event occurrence were secondary endpoints.

| Statistical analysis
LC, OS, and PFS rates with C-ion RT were estimated by the Kaplan-Meier method and compared by log-rank test. Betweengroup differences were evaluated by t-test. Differences were considered statistically significant at p < .05. Statistical analyses were performed with IBM SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Table 1 shows T A B L E 2 Acute and late adverse events in patients treated with carbon-ion radiotherapy (n = 52).

| RESULTS
Adverse events after C-ion RT are summarized in Table 2. Acute grade 2 mucositis and grade 1 dermatitis were common adverse events, which improved by conservative therapy. Leukopenia, anemia, and thrombocytopenia were improved after DAV therapy. One patient required analgesia due to late effect grade 4 mucositis.
Another patient had grade 4 oral fistula due to tumor invasion of the maxilla and required a dental prosthesis. Two patients required hyperbaric oxygen therapy and analgesia due to grade 3 and skin, n = 1; and multiple sites, n = 2), and one had local recurrence and distant lymph node metastasis simultaneously ( Table 3). The ICI agents used were anti-PD-1 (nivolumab and pembrolizumab) and anti-CTLA4 (ipilimumab). Eight patients survived ( Table 3).
The 3-year OS rates of the ICI group (n = 16) and BSC group (n = 13) were 53.8% and 0.0%, respectively (95% CI, ICI group, 28.174%-71.543%; BSC group, 6.982%-19.802%) ( p = .482) ( Figure 3B). The OS ( Figure 3B) was calculated after the diagnosis of recurrence (any site) and treatment with C-ion RT. There was no significant difference in the 3-year OS rates between the ICI and BSC groups; however, there was no mortality after 1 year among patients who received ICIs ( Figure 3B).
The observed ICI-related adverse events are shown in Table 3.
None of the adverse events associated with C-ion RT were related to or exacerbated by ICI administration. Most adverse events were grade 1 gastrointestinal and hepatobiliary toxicities. Grade 3 adverse events included mucositis (n = 1) and dermatitis (n = 1) and Grade 5 adverse events included pneumonitis (n = 1) ( Table 3). The nivolumab-only F I G U R E 3 OS curve of patients with mucosal malignant melanoma treated with C-ion RT with or without ICIs. (A) ICI agents were approved for treating melanoma in 2014 in Japan. The patients (n = 52) were divided into two groups based on whether they had been diagnosed before 2014 (n = 8) or after 2014 (n = 44). The 3-year OS rates of those diagnosed before 2014 (dotted line) and those diagnosed after 2014 (straight line) were 0.0% and 76.4%, respectively ( p = .000) (B) The 3-year OS rates in the ICI group (n = 16) (straight line) and BSC group (n = 13) (dotted line) were 53.8% and 0.0%, respectively ( p = 0.482). This is the OS after the diagnosis of recurrence (any site) and treatment with C-ion RT. BSC, best supportive care; C-ion RT, carbon-ion radiotherapy; ICI, immune checkpoint inhibitor; OS, overall survival.
T A B L E 3 Patient characteristics treated with Immune checkpoint inhibitor therapy and therapeutic outcomes.

| DISCUSSION
In this study, the 3-year OS rate was 64.3% for all patients. Patients diagnosed with recurrence after 2014 were categorized based on whether they received ICIs or BSC. However, there was no significant difference between the groups. A retrospective multi-center study of C-ion RT by J-CROS showed similar outcomes, with a 2-year OS rate of 69.4% and 5-year OS rate of 44.6%. 6 However, in the J-CROS study, it was unclear whether ICIs were used in combination with Cion RT. Nevertheless, a good prognosis was showed in the ICI study, with a 5-year OS rate of 52% with the use of nivolumab plus ipilimumab for advanced melanoma. 9 In this study, the administration of nivolumab-only was used in 11 patients. A previous clinical trial showed good prognosis for the use of nivolumab-only, with a 5-year OS rate of 44%. 9 Therefore, the use of ICI agents for recurrence of HN-MM after C-ion RT may have contributed to survival.
The 3-year LC in the current study was 86.3%. Similar results were obtained in a multi-center study by J-CROS, in which the 2-year LC rates were 83.9% and 5-year LC rates were 72.3%. 6 Another prospective study reported a 3-year LC rate of 92.3%. 7 In our study, no cases of local recurrence due to modification of CTV or PTV were identified. Therefore, the C-ion RT had a good local effect while protecting the OAR, similar to the previous reports. 6,7 Good LC may indicate good OS. Furthermore, some studies reported a favorable response after a long duration of ICI treatment. 15,16 In the present study, good local efficacy for OS was maintained with ICI therapy, and no deaths were reported after 1 year in patients who received ICIs.
Therefore, a good OS can be attributed to both a good local response to C-ion RT and a durable response to ICIs.
In this study, adverse events were documented for both C-ion RT  (Table 3). At the minimum, C-ion RT or ICI agents could be safely administered. One study concluded that combining RT with ICI agents did not increase the risk of grade 3 or higher adverse events. 19 Furthermore, a few studies reported no increase in the rate of adverse events when ICIs were administered before or after C-ion RT. 20,21 A recent study on in-vivo melanoma models found that C-ion irradiation suppressed myeloid-derived suppressor cells while increasing the lymphoid cells to a greater extent than X-rays did with equivalent RBE. 22  been reported to be associated with the efficacy of ICI therapy. 24 In the current study, 13 of the 16 patients received nivolumab. PD-L1 expression is induced in tumors after C-ion RT, [25][26][27] which may contribute to the good prognosis against anti-PD-1/PD-L1 antibodies.
Therefore, PD-L1 evaluation of recurrent cases may also be helpful in prediction of prognosis of salvage treatment by ICIs. This is consistent with the results of Sato et al., 28 who showed that PD-L1 expression is enhanced by deoxyribonucleic acid damage signals. Summarily, radiation has been reported to stimulate the immune response. Therefore, C-ion RT may be a better immune stimulator than X-rays. Although there was a time lag between the C-ion RT and the administration of ICIs in this study, theoretically, a further improvement in efficacy may be expected using the combination of ICIs as early as possible.
This study had a few limitations. Firstly, several combinations and dosages of ICIs were used, the effects and adverse events of which should be further examined. Secondly, this study was retrospective and involved patients were enrolled from a single center. Future prospective studies with a larger sample size and including a greater number of institutions should be carried out. Moreover, it will be important to standardize ICI protocols in a prospective study to evaluate the effect of ICIs.
In conclusion, no significant differences were found in the survival rate between C-ion RT followed by ICI therapy and C-ion RT alone; however, patients who used ICIs for a longer period of time tended to survive longer. Therefore, ICI may be promising for lymph node metastases and distant metastases in patients with localized disease controlled by C-ion RT. In addition, the other novelty of this study was that patients who received ICIs and C-ion RT did not show a higher incidence of adverse events compared to patients treated with ICIs alone, and the possibility of combination therapy is clinically significant. As for the potential indications, the safety of the treatment has been confirmed in this report, and it has the potential to be fully writingreview and editing (supporting). Norichika Ota: Data curation (supporting); formal analysis (supporting); writingoriginal draft (supporting); writingreview and editing (supporting). Akiko Adachi: Data curation (supporting); formal analysis (supporting); writingoriginal draft (supporting); writingreview and editing (supporting). Masato Shino: Writingreview and editing (supporting). Osamu Nikkuni: Writingreview and editing (supporting). Shota Ida: Writingreview and editing (supporting). Katsuyuki Shirai: Methodology (supporting); project administration (supporting); writingoriginal draft (supporting); writingreview and editing (supporting). Jun-ichi Saitoh: Methodology (supporting); project administration (supporting); writingoriginal draft (supporting); writingreview and editing (supporting). Satoshi Yokoo: Writingreview and editing (supporting). Kazuaki Chikamatsu: Writingreview and editing (supporting). Tatsuya Ohno: Data curation (supporting); formal analysis (supporting); writingoriginal draft (supporting); writingreview and editing (supporting).

This work was supported by JSPS KAKENHI [grant Number 21K07693],
the Takeda Science Foundation, and the Uehara Memorial Foundation.
The funding sources had no involvement in the study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.