The association between immune-related adverse events and survival outcomes in Asian patients with advanced melanoma receiving anti-PD-1 antibodies

Immune checkpoint inhibitors (ICIs), including ipilimumab, nivolumab, and pembrolizumab, have become standard therapies in advanced melanoma, regardless of genetic alteration [1‐3]. The combination of ipilimumab and nivolumab demonstrated greater effectiveness than monotherapy with either of these [4‐7]; however, combination therapy had the greater toxicity than monotherapy and insufficient power for overall survival (OS) over nivolumab limit the application of combination treatment in advanced melanoma. ICIs induce favorable response in some patients, so biomarkers are needed to determine the clinical course of ICIs in advanced melanoma.

One potentially effective clinical biomarker of ICI response in cancer patients is immune-related adverse event (irAE) [8]. Patients who experience irAEs during therapy with anti-PD-1 and anti-PD-L1 antibodies have been found to show favorable outcomes – overall response rate (ORR), progression-free survival (PFS), and OS – in cases of melanoma, lung cancer, and urothelial carcinoma [9‐15]. However, as compared with other cancer types, the association between irAE occurrence and anti-PD-1 antibody efficacy is not well established in metastatic melanoma patients. Several retrospective studies have reported improved outcomes in patients who experience irAEs; however, not all measured outcomes consistently improved in these patients [16‐19]. Besides, given most of the previous publications are studied on cutaneous melanoma, it raises the question that whether those results could be applied in Asian in which acral and mucosal melanoma are predominant.

Previously, we reported the clinical efficacy and safety of ICIs in our institution in 80 ICI-naïve melanoma patients [20]. As anti-CTLA-4, anti-PD-1 antibodies, and their combination showed distinct irAE patterns [2], we selected the patients undergoing anti-PD-l monotherapy, with either nivolumab or pembrolizumab, and analyzed the association between irAEs and survival outcomes. This study could provide additional evidence of irAEs as biomarkers of treatment outcomes.

In the current study, the association between irAEs and clinical outcomes was investigated using data from 49 patients treated with anti-PD-1 antibodies in single institute cancer center; it was found that the association was dependent on type and severity of the irAEs. The patients with mild-to-moderate irAEs (grade 1–2) had better PFS and OS. In addition, patients with skin irAEs/vitiligo/endocrine irAEs showed favorable PFS and OS.

Our findings were compatible with recently published meta-analysis investigating irAE and efficacy of ICIs and showing the occurrence of irAEs was significantly associated with a better ICI efficacy in cancer patients, particularly endocrine, skin, and low-grade irAEs [22]. In addition, this association was limited in PD-1 monotherapy but not anti-CTLA-4 nor combination therapy indicating different targets of ICIs might be associated with distinct irAE patterns which should be analyzed and discussed separately. Furthermore, ethnicity and melanoma subtype was found to be associated with distinct irAE profiling in the setting of PD-1 ICI by using 3 independent melanoma centers from the US. and China [23]. This finding indicates that not only melanoma subtype but also ethnicity influence the efficacy of anti-PD-1 ICI. Non- Caucasian had higher rates of skin and endocrine irAEs but low rates of pneumonitis than Caucasian. Following previous findings, current study conducted in Taiwan provides additionally important and valuable evidence showing high frequency of skin/endocrine irAE and low frequency of pneumonitis were associated with survivals in Asian melanoma undergoing PD-1 melanoma.

The results of previous studies regarding to irAEs and survival in melanoma patients undergoing therapy with anti-PD-1 ICI have been inconsistent. Okada et al. examined 15 melanoma patients undergoing nivolumab therapy and found that patients with irAEs were associated with better OS than patients without irAEs [9]. Indini et al. conducted a retrospective analysis of 173 patients with metastatic melanoma treated with anti-PD-1 antibodies and found that 59% of the patients experiencing irAEs showed improved PFS and OS, independent of other factors [16]. However, a large retrospective study analyzed the outcomes of 576 melanoma patients pooled from several studies treated with nivolumab [18] and no differences in PFS were found between patients with or without irAEs. Our cohort showed distinct results—that the prognostic value of irAE was severity dependent.

In terms of severity of irAEs, mechanistically as irAEs are considered to be the bystander effects of activation of T cells by ICIs, so patients who experience more severe irAEs should have higher T-cell activity and experience better outcomes than those who experience mild-to-moderate or no irAEs [24]. However, most of the previous studies on anti-PD-1 and anti-CTLA-4 antibodies rarely demonstrated the relationship between irAE severity and ICI efficacy. This variation in results could be attributed to the fact that patients experiencing severe irAEs tend to suffer from significant morbidity and sometimes mortality from the autoimmune reactions that compromise the benefit of ICIs [25]. In addition, severe toxicity is often associated with aggressive immunosuppression treatment, which may also influence the efficacy of ICIs [26]. Further, in the case of patients with severe irAEs, no ICIs were administered even after they completely recovered from the irAEs. Taken all together, these points could explain why the patients with severe irAEs had worse survival than patients without severe irAEs. One study in 858 older aged (≥65 years) melanoma patients treated with ipilimumab supported our findings as patients with non-severe irAE had improved OS compared to patients without irAE, and patients with severe irAE appeared to have the highest risk of death [27], although ipilimumab rather than anti-PD-1 was used in this study.

In current study, we found that different types of irAEs differentially predicted survival, as skin/vitiligo/endocrine irAEs were favorable irAEs. Sanlorenzo et al. found that patients with cutaneous irAEs had a significantly longer PFS than those without in a retrospective analysis of 83 metastatic cancer patients (including 66 melanoma patients) treated with pembrolizumab [28]. Yamazaki et al. followed 124 Japanese melanoma patients treated with nivolumab and reported that the occurrence of skin-related and endocrine-related irAEs had a significant impact on the PFS of the patients [29] although only the abstract is available currently. Moreover, Fujisawa et al. demonstrated that endocrine-related irAEs were associated with longer OS of melanoma patients treated with ipilimumab after nivolumab [30]. Other than melanoma, a significant correlation between endocrine irAEs and OS was observed (p = 0.019) in a pooling analysis of 12 randomized controlled trials of 3815 metastatic head and neck and lung cancer patients treated with ICIs [31]. All of above studies supported our finding that skin and endocrine irAEs represent favorable irAEs.

Vitiligo is a specific irAE in melanoma and is considered a predictive biomarker of the effectiveness of ICIs in advanced melanoma as T cells activated by ICIs may recognize the common antigens on tumor cells and host melanocytes. Therefore, the occurrence of vitiligo reflects that the T cells are activated and ready to kill the melanomas [32]. In a retrospective report, Indini analyzed various irAEs and found that vitiligo was associated with better OS than other irAEs although not a statistically significant level (p = 0.061) [16]. Only two patients experienced vitiligo in our cohort; therefore, the association between vitiligo and survival could not be undetermined.

Skin and endocrine irAEs in the current study were mild to moderate, so topical treatment and endocrine supplements were useful for managing the irAEs. Systemic steroids were not necessary for most of the patients who could continue ICI treatment. In contrast, pulmonary irAEs may be life threatening, so systemic steroids should be started as soon as pulmonary irAEs are suspected. The treatment course should be halted or terminated to improve the prognosis of patients with pulmonary irAEs. Fatigue is a nonspecific complaint resulting from ICI treatment, underlying malignancies or comorbidities. Therefore, patients with fatigue in our cohort may reflect disease progression rather than treatment-related AEs, which are difficult to differentiate in the initial presentation if patients have no other discomfort.

The association between irAEs and clinical outcomes was evident in not only metastatic melanoma but also resected melanoma with adjuvant therapy. In Keynote-054, pembrolizumab was found to improve recurrence-free survival (RFS) of stage III melanoma patients after complete resection [33]. The occurrence of an irAE was associated with longer RFS in the pembrolizumab arm, particularly endocrine AE [34]. Although patients appeared to have a low risk of recurrence or death after vitiligo onset in the pembrolizumab arm, statistical significance was not observed due to the limited number of cases (n = 24). In addition, the severe irAEs (grade 3–4) were not significantly associated with favorable RFS, but the patients who experienced severe irAEs seemed to be numerically worse than those had not experienced severe irAE in the pembrolizumab arm. These findings are compatible with our findings on advanced melanoma as the severity and types of irAE predict the survival outcomes.

As the concern of glucocorticoids/immunosuppressant may suppress the efficacy of ICIs [35], withholding ICIs rather than glucocorticoids unless great 3 irAE or pneumonitis [21]. The most frequent irAE in current study were skin and endocrine toxicities so topic glucocorticoids with or without oral antihistamine, and hormone replacement therapy were applied for those patients with skin and endocrine toxicities respectively. Only three patients experienced grade 3–5 irAE and two patients had pneumonitis, it is difficult to investigate the correlation and possible impact on survivals.

There are some limitations to the current retrospective study. Challenges of guarantee-time bias should be considered [36] because patients who experience irAEs are usually those who remain on ICI treatment for longer time periods and thus have a better prognosis than those who do not have irAEs. A retrospective study pooling melanoma patients from the randomized Checkmate 067 and Checkmate 069 trials suggested time in therapy is not the factor behind the relationship between irAE onset and ICI efficacy [37]. A Cox model with a time-varying covariate was used to avoid this bias for KEYNOTE-054, and the hazard of recurrence or death was lower in the pembrolizumab-treated patients after irAE onset (HR 0.37; 95% CI 0.24–0.57) than in those without or before irAE onset (HR 0.61; 95% CI 0.49–0.77) (p = 0.03) [34]. In current study, most skin irAE occurred in early period (< 30 days) of ICIs treatment indicating the early immune response of skin irAE may help clinician to predict the tumor response. In contrast, most endocrine irAE occurred after 3-month treatment so the prognostic value of endocrine irAE may be influenced by guarantee-time bias.

Small number of cases in current study limited the significance and some results should be interpreted cautiously. Only three patients experienced grade 3–5 irAE and two patients had pneumonitis so clinical significance of such irAE should not be confirmed using limited cases even statistical significance (p < 0.05, Figs. 2, 3). The current study did not analyze the impact of autoimmune diseases as preexisting autoimmunity may be associated with ICIs efficacy and irAE [38]. However, the previous studies showed ICIs lead to similar rates of irAEs in patients with coexisting autoimmune diseases compared with those without existing coexisting autoimmune diseases [39].

In conclusion, melanoma patients undergoing anti-PD-1 ICIs who experienced mild-to-moderate irAE (grade 1–2) had favorable survival outcomes than those without irAE or severe irAE (grade 3–5). Patients with skin/vitiligo/endocrine irAEs had favorable survival. Therefore, the clinical outcomes of melanoma patients undergoing anti-PD-1 ICIs may be severity- and type-dependent, as per the current cohort.