SALVO: Single-Arm Trial of Ipilimumab and Nivolumab as Adjuvant Therapy for Resected Mucosal Melanoma

This article is featured in Highlights of This Issue, p. 2171

Mucosal melanoma is an uncommon, but highly aggressive malignancy. This rare type of melanoma is biologically distinct from cutaneous melanoma and accounts for about 1%–4% of melanoma diagnoses (1). Sinonasal, anorectal, and vulvar melanoma represent the most common sites of origin, all with comparable outcomes. Many patients present with symptoms that are often mistakenly attributed to other etiologies (e.g., sinus congestion, hemorrhoids). Because of this, many patients present with locally advanced and/or metastatic disease. Unfortunately, even for patients who present with resectable disease, recurrence rates approach 100% within a short time. In a recent adjuvant study by Lian and colleagues (2) all patients randomized to the observation arm recurred within 2 years of follow-up. Lymph node sampling is not routinely performed, and there is no formal AJCC staging system for mucosal sites other than head/neck (3), so prior approvals for stage III melanoma often do not apply. As the disease is relatively rare, there is near absence of randomized clinical trials.

In the largest adjuvant study to date, Lian and colleagues randomized 189 patients with resected mucosal melanoma to one of three arms: Observation alone, standard IV IFN α2b x 1 year, or temozolomide plus cisplatin every 3 weeks for 6 cycles (2). Median recurrence-free survival (RFS) was 5.4, 9.4, and 20.8 months, respectively. Estimated median overall survival (OS) was 21.2, 40.4, and 48.7 months at the time of publication. As cytotoxic therapy has not generally been associated with improved outcomes for metastatic melanoma and given the improvements in outcome with novel therapies for cutaneous melanoma in recent years, utilization of the cisplatin/temozolomide regimen has been tempered. More recently Lian and colleagues (4) looked at 145 resected patients with mucosal melanoma randomized to either high-dose IFN (HDI) versus toripalimab (anti–PD-1). Median RFS was similar between the two groups (13.9 months in the HDI and 13.6 months in the toripalimab groups, respectively; ref. 4). However, subgroup analysis showed improved median RFS in patients who had PD-1–positive tumors and received toripalimab (17.4 months) versus those who received HDI (11.1 months; ref. 4). The National Comprehensive Cancer Network (NCCN) v2.2022 melanoma guidelines do not specifically address management of mucosal melanoma, Adjuvant radiotherapy for both sinus/nasal cavity and oral cavity/oropharynx/larynx/hypopharynx head and neck melanoma is recommended to be strongly considered for stage T3N0, and is recommended for more advanced stages as it has been associated with improved locoregional control but not OS in a meta-analysis (5, 6).

Immunotherapy has been demonstrated to have antitumor activity in mucosal melanoma in small studies (2–4), with only one prospective Phase II clinical trial by Nathan and colleagues (7). Although responses were seen, these were lower than those with cutaneous origin as well as OS. The CheckMate 067 study established ipilimumab (3 mg/kg q3w × 4 cycles) plus nivolumab (1 mg/kg q3w × 4 cycles, then 3 mg/kg q2w monotherapy) as a standard of care for advanced melanoma over nivolumab or ipilimumab alone, albeit at the cost of higher toxicity. Although there does appear to be a dose-dependent relationship for OS outcomes with regards to dosing of ipilimumab, nevertheless these come at the price of higher toxicity (8). Lower doses of ipilimumab were studied in the CheckMate 511 study where patients were randomized to ipilimumab/nivolumab at standard dosing (3 mg/kg and 1 mg/kg q3w × 4 then maintenance nivolumab at 480 mg q4w) or lower dose ipilimumab (1 mg/kg q3w × 4) and nivolumab (3 mg/kg q3w × 4 then 480-mg maintenance). The primary endpoint of a decrease in grade 3–5 immune-related adverse events was met, with a lower incidence in the lower dose ipilimumab arm (33.9% vs. 48.3%). Although the study was not powered for efficacy, a descriptive secondary analysis showed an approximately 5% difference in efficacy as measured by 36 months. Progression-free survival between the two cohorts (favorable to the IPI3 arm), but this was not statistically significant.

Immunotherapy for mucosal melanoma has been associated with some durable responses, but response rates appear to be lower than for cutaneous melanoma (9), although a recent combination of anti-PD1 antibody toripalimab and axitinib, a tyrosine kinase inhibitor (TKI), showed a response rate of 48% with a median duration of response of 13.3 months (10). In the CheckMate-238 study, a double-blinded randomized phase III study for adjuvant therapy of resected stages IIIB/C and IV melanoma comparing nivolumab 3 mg/kg q2w versus ipilimumab 10 mg/kg q3w × 4, then q3 m × 3, 29 patients with mucosal melanoma were included in the study population (16 nivo, 13 ipi). In this small subset, the ipilimumab appeared to outperform the nivolumab arm in RFS, although it did not reach statistical significance [HR, 1.71; 95% confidence interval (CI), 0.68–4.29]. The CheckMate 915 study recently evaluated the addition of low-dose ipilimumab (1 mg/kg q6w) to nivolumab 240 mg q2w. Although it has been reported in abstract form as a negative trial for the primary outcome, it has not been published in article form. Patients with mucosal melanoma were eligible, but no detail is available for any patients with mucosal melanoma, and it is unclear if the lower dose of ipilimumab may have affected the outcome (11).

Given the high recurrence rate for resected mucosal melanoma, along with the promise of greater tolerability for the “flip” dosing of the ipilimumab/nivolumab combination, it was felt that this combination represented a rational approach for adjuvant therapy for the mucosal melanoma population resected with curative intent. We tested this hypothesis in a single arm, phase II study of ipilimumab and nivolumab as adjuvant therapy for resected mucosal melanoma (SALVO). Here, we report on the primary and secondary endpoints of the study and outcomes across mucosal melanoma subgroups treated with adjuvant ipilimumab and nivolumab.

A prospective, single-arm phase II trial was conducted at 6 sites through the Midwest Melanoma Partnership in cooperation with the Hoosier Cancer Research Network. The study was approved by the Mayo Clinic Institutional Review Board and all patients signed informed, written consent. The study was conducted in accordance with the Declaration of Helsinki. Appropriate IRB review and approval also was obtained at all institutions. Patients without evidence of distant metastatic disease were eligible for enrollment following R0 or R1 resection of mucosal melanoma from any site except the eye. Patients were required to be registered within 90 days of surgical resection. Adjuvant radiotherapy was permitted, provided it was completed within the 90 day window and at least 2 weeks before study enrollment. Eligible patients were ≥18 years of age with ECOG performance status of 0 or 1, as well as adequate organ function and blood counts, not pregnant or breastfeeding, who had not recieved prior neo-adjuvant or other systemic therapy for mucosal melanoma. Exclusion criteria included contraindications to immunotherapy, including active autoimmune disease requiring systemic immune suppression.

Once enrolled, patients were treated with 4 cycles of ipilimumab (1 mg/kg) and nivolumab (3 mg/kg) intravenously every three weeks for four cycles. Patients subsequently underwent restaging and if no recurrence was found, then completed 11 additional cycles of adjuvant nivolumab 480 mg every four weeks. Patients were actively assessed for recurrence with CT scans every 3 months while on active adjuvant therapy and after completion of therapy, with planned passive follow-up (event monitoring) for recurrence for a total of five years After 2 years on study, imaging follow-up was left to the discretion of the treating provider.

The primary endpoint was RFS, defined as the time from date the patient receives first dose of study medication to the date of documentation of first disease recurrence or death due to any cause. Patients without a documented date of disease recurrence or death were censored on the last tumor assessment date in which they are confirmed to be alive and recurrence-free. Secondary endpoints included estimating the OS as well as assessment of the safety profile of the combination of ipilimumab/nivolumab in this patient population. Correlative endpoints included performing in depth genetic analyses of the primary tumor to identify commonly found somatic mutations including NRAS, BRAF, and KIT.

Descriptive statistics were used to summarize the patient population, disease characteristics, and outcomes. Time-to-event outcomes were estimated using the Kaplan–Meier method. Exact confidence intervals were constructed for outcomes of interest. Exploratory analyses were conducted using a two-sided log-rank test to investigate whether differences in RFS were observed between primary sites of disease, and the Fisher's exact test to investigate whether adjuvant radiotherapy was associated with local recurrence. Given the exploratory nature of these analyses, no P value assessments were performed.

The study sample size of 35 patients was selected to ensure 85% power to detect an improvement in RFS between 5.5 and 9.5 months using a one-sided, α = 0.05, log rank test.

Data presented in this study can be accessed by contacting the corresponding author.

Between September 2017 and August 2021, 44 patients were approached to enter this study. Nine patients were excluded (eight did not meet eligibility criteria, and one declined participation), leaving 35 eligible patients (Supplementary Fig. S1). As of this time of this article, 33 patients have completed the treatment phase, 12 patients are off-study and 8 have died. Details are included in the CONSORT diagram shown in Supplementary Fig. S1.

Descriptive statistics for the entire study population are presented in Table 1. The mean patient age was 65.8 years and 21 (60.0%) were female. The primary site of disease was vulvovaginal in 12 (34.3%) patients, sinonasal in 10 (28.6), anorectal in 9 (25.7) and other site (including penile urethra, hard palate of mouth, anus, urethral) in 4 (11.4%). Twenty-nine patients (82.9%) had an R0 resection before enrollment. Somatic driver mutations were rare, with only 3 tumors (8.6%) having a KIT mutation, and one (2.9%) each having a NRAS or BRAF mutation. Seven patients (20%) received adjuvant radiotherapy before starting study therapy.

Treatment information is provided in Table 2. Twenty-eight (80.0%) patients received all 4 doses of ipilimumab and nivolumab induction, whereas 9 (25.7%) had a dose delayed or omitted. Ten (28.6%) patients did not go on to maintenance nivolumab, and 12/25 patients (48%) required a dose omission or delay of maintenance nivolumab. Additional treatment details can be found in Table 2.

The most common adverse events of any grade were fatigue (82.9%), hypertension (54.3%), pruritus (54.3%), and diarrhea (51.4%). A total of 23 (65.7%) patients experienced a grade 3 adverse event, but no patient experienced a grade 4 or above adverse event. All grade 3 treatment-related adverse events are described in Table 3, with all adverse events, all grade 3 adverse events, and all related events, respectively, in Supplementary Tables S1–S3.

Outcomes are summarized in Table 4. Median (95% CI) recurrence-free survival was 14.3 months (95% CI, 6.2–40.9 m) and is shown in Fig. 1. Notably, the lower bound of the 95% CI is greater than the a priori protocol defined hypothesized median RFS of 5.5 months. One-year RFS was 53.7% (95% CI, 34.9%–69.2%), whereas OS was 90.6% (95% CI, 73.6%–96.9%). Results are shown in Fig. 2. Median OS was not reached at the time of data lock. The following are the treatment discontinuations: Completed protocol 8 (22.9%) patients, Disease progression 14 (40%) patients, Adverse events 10 (28.6%) patients, withdrawal 1 (2.9%) patients.

Descriptive statistics assessing diagnostic site, site of recurrence, R0/R1 resection, receipt of radiotherapy are included in the Supplementary Tables. None achieved statistically significant associations with outcome.

Mucosal melanoma is a rare subtype of melanoma that behaves quite aggressively, where there has been little research specifically conducted in the adjuvant space for patients treated with curative intent. After complete surgical resection of a primary tumor is performed, patients are known to be at high risk for recurrence, but prospective clinical trials of adjuvant therapies are lacking. The best evidence to date at the time of study development favored utilization of a cytotoxic chemotherapy combination, including temozolomide and cisplatin. However, cytotoxic chemotherapy has little place in the treatment of other types of melanoma given its inability to produce durable responses, and often fleeting efficacy. Given the advancements in contemporary immunotherapy for melanoma dating back to 2011, in addition to emerging evidence of the benefit of checkpoint inhibitors in mucosal melanoma (4, 7, 9), this study provides the first prospective evidence for dual checkpoint inhibitor immunotherapy benefit in the adjuvant setting.

The single treatment arm met its primary endpoint of recurrence-free survival over that expected for observation alone. Median RFS was 14.3 months overall, meeting our predetermined criteria for success. Lian and colleagues (2) previously noted median RFS of 5.4 months for observation, 9.4 months for high-dose interferon, and 20.8 months for adjuvant temozolomide and cisplatin and median RFS of 13.9 months in adjuvant HDI and 13.6 months in adjuvant toripalimab (4). In our study, there appears to be some leveling of the RFS curve, suggesting that there may be a subpopulation of treated patients who will achieve a durable survival benefit (Fig. 1) Although Lian and colleagues (2) observed 100% recurrence by 2 years in their observation arm, in our study, 37.8% are estimated to be recurrence free at 2 years. Given prior recurrence rates approaching 100% for resected mucosal melanoma, this represents a significant advance in clinical care. Further follow-up will be necessary to determine long-term outcomes for these patients and whether there is a subgroup with durable disease control. Given the small sample size, there is insufficient power to determine whether significant differences exist among anatomic sites of mucosal melanoma (Supplementary Table S4). However, and in keeping with prior reports (2), there were no unexpected patterns or obvious outliers that would impact the fundamental interpretation of the trial based on primary tumor location.

The combination of 1 mg/kg of ipilimumab and 3 mg/kg of nivolumab has been referred to as “flip dosing” of these agents, in contradistinction to the original higher dosing for ipilimumab and lower dose of nivolumab studied in patients with metastatic disease (12). In the OpACIN-neo study, this combination appears to preserve efficacy of the dual antibody combination, with decreased side effects related to immunotherapy. As we had to choose one combination for this single-arm study in a rare tumor, we have not been able to compare with either standard dose ipilimumab plus nivolumab or to single-agent anti-PD1 therapy. We chose a combination trial targeting CTLA-4 and PD-1 due to the highly aggressive nature of mucosal melanoma. Although low-dose ipilimumab has since failed to show benefit in adjuvant therapy of melanoma, any patients with mucosal melanoma included have not been reported (13). It is likely that ipilimumab is an important component as in the Checkmate-238 study, mucosal primary was the only subset that appeared to favor the use of high-dose ipilimumab over nivolumab in the adjuvant setting (HR 1.71; 95% CI, 0.68–4.29), although statistical significance was not reached in this subset analysis (14). A national trial to answer these fundamental questions is needed. In addition, targeted therapy toward mutations in KIT or BRAF may be beneficial for patients whose tumors carry these mutations, but these represent a small subset of an already rare malignancy. An ongoing study through the Alliance is examining the role of adjuvant nivolumab with or without cabozantinib, a TKI (NCT05111574), with estimated study completion in late 2023.

Limitations of the study include the single-arm design without randomization, and a lack of staging conventions for mucosal melanoma across primary sites other than head and neck. Conclusions from the study are also limited by its small size (N = 35, although given the rarity of the disease, is still a notable accomplishment. We enrolled at 6 (5 academic/1 community) centers, limiting generalizability, although the multiinstitutional structure of the study does represent a strength, in addition to broad representation of primary sites, mean age and sex (Supplementary Table S5). In addition, for vulvovaginal and anorectal primary tumors, it is sometimes difficult to determine whether or not mucosal origination is present versus origination in nearby skin, and there can be gray areas in this interpretation. The authors made this determination to the best of our ability with input from pathology and surgeons. There are also emerging data that the mutational profiles of melanoma arising in these “gray” areas align with that of mucosal primary tumors (15).

Also, given potential morbidity of primary resection, including that for all sites of origin, a neoadjuvant approach may be beneficial. Patients with rapid appearance of metastatic disease can be spared the morbidity of a surgery, there is an ability to assess response to systemic therapy, and potential to downsize the tumor to optimize the extent of operation and improve likelihood of negative margins. Clearly, there are times where this approach makes sense, versus other times where the diagnosis is not necessarily known until the surgery is performed and pathology is obtained. Future trials should assess the neoadjuvant approach for mucosal melanoma, and given the results of our current study, checkpoint inhibition with ipilimumab and nivolumab merit further research in that setting as well.

In our single-arm, multi-center SALVO trial, the combination of ipilimumab and nivolumab met its primary endpoint of relapse-free survival for patients with resected mucosal melanoma. Future randomized trials should be performed to confirm these findings, determine the optimal dosing of adjuvant immunotherapy, address neoadjuvant approaches, and determine long-term outcomes for this rare, extremely high-risk patient cohort. The combination of adjuvant ipilimumab and nivolumab is a potentially valuable addition to the standard of care for patients with resected mucosal melanoma.

L.A. Kottschade reports personal fees from Immunocore outside the submitted work. G.R. Pond reports grants from AstraZeneca, Merck, Takeda, and Profound Medical outside the submitted work; in addition G.R. Pond has a close family member who is an employee of Roche Canada and owns stock in Roche Ltd. A.J. Olszanski reports personal fees from Merck, BMS, Pfizer, Takeda, Novartis, Sanofi, Eisai, Nektar, and IntilBio outside the submitted work. Y. Zakharia reports advisory board participation with Bristol-Myers Squibb, Amgen, Roche Diagnostics, Novartis, Janssen, Eisai, Exelixis, Castle Bioscience, Genzyme Corporation, AstraZeneca, Array, Bayer, Pfizer, Clovis, EMD Serono, and Myovant; grant/research support to institution from NewLink Genetics, Pfizer, Exelixis, and Eisai; data and safety monitoring committee membership at Janssen Research; and development consultant honoraria from Pfizer and Novartis. E. Domingo-Musibay reports grants from Clinigen Group, Iovance Biotherapeutics, and Instil Bio outside the submitted work. B.D. Curti reports nonfinancial support from Hoosier Cancer Research Network during the conduct of the study as well as grants from BMS outside the submitted work. M.S. Block reports grants from Bristol-Myers Squibb during the conduct of the study as well as grants from Alkermes, Genentech, Merck, nference, Pharmacyclics, Regeneron, Sorrento, TILT Biotherapeutics, Transgene, and Viewpoint Molecular Therapeutics outside the submitted work. T. Hieken reports grants from Genentech and SkylineDX BV outside the submitted work. R.R. McWilliams reports grants from BMS during the conduct of the study as well as personal fees from Zeno/Zentalis Pharmaceuticals outside the submitted work. No disclosures were reported by the other authors.

L.A. Kottschade: Conceptualization, writing–review and editing. G.R. Pond: Conceptualization, data curation, formal analysis, writing–original draft, writing–review and editing. A.J. Olszanski: Data curation, writing–review and editing. Y. Zakharia: Data curation, writing–review and editing. E. Domingo-Musibay: Data curation, writing–review and editing. R.J. Hauke: Data curation, writing–review and editing. B.D. Curti: Data curation, writing–review and editing. S. Schober: Data curation, writing–review and editing. M.M. Milhem: Data curation, writing–review and editing. M.S. Block: Data curation, writing–review and editing. T. Hieken: Data curation, writing–review and editing. R.R. McWilliams: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, writing–review and editing.

The funding for the study was provided by Bristol-Myers Squib. Identifiers are NCT03241186 and MMP-16–252.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.