Association between immune-related adverse events and efficacy of PD-1 inhibitors in Chinese patients with advanced melanoma

Programmed cell death 1 (PD-1) checkpoint inhibitor therapy leads to immune-related adverse events (irAEs). We sought to evaluate whether the development of irAEs correlates with the treatment response in Chinese patients with advanced melanoma. In this study, we conducted a retrospective study of advanced melanoma patients who received PD-1 inhibitor therapy in China between August 2014 and March 2018. A total of 93 patients treated with PD-1 inhibitors including pembrolizumab and nivolumab were enrolled. The most frequent irAEs were pruritus, rash, vitiligo, and fatigue. The median time to onset of irAEs was 6.1 weeks. The overall response rate (ORR) and disease control rate (DCR) were higher in patients with irAEs than those without irAEs (P = 0.004 and P = 0.003, respectively), and better in patients who experienced three or more irAEs than those with none (P <0.001 and P <0.001, respectively). The ORR and DCR were significantly better in patients with grade 1 to 2 irAEs when compared with those with none (P = 0.002 and P = 0.003, respectively). In addition, the median progression-free survival and overall survival were longer in patients with irAEs than in those without irAEs (P = 0.007 and P = 0.002, respectively). In conclusion, our data demonstrated that irAEs were associated with a better clinical outcome after treatment with PD-1 inhibitor therapy in Chinese patients with advanced melanoma.


INTRODUCTION
Malignant melanoma is the leading cause of cancerrelated deaths worldwide, and its incidence in China has been increasing in the recent years [1,2]. The treatment of advanced melanoma has remarkably improved due to the successful clinical development of immune checkpoint inhibitors (ICIs) that reactivate the anticancer immune response [3,4]. The antiprogrammed cell death 1 (PD-1) receptor antibody represents the second breakthrough in immune checkpoint blockade therapy of melanoma after the approval of ipilimumab [5][6][7][8]. Two anti-PD-1 antibodies, pembrolizumab, and nivolumab were approved for the treatment of melanoma in 2014 [6,9].
AGING pembrolizumab and nivolumab can reverse this T-cell suppression and induce long-lasting antitumor responses in patients with advanced solid tumors, including advanced melanoma [15][16][17]. By activating the immune system, the anti-PD-1 antibodies lead to autoimmune-like toxicities known as immune-related adverse events (irAEs) through immune cell infiltration into normal, noncancerous tissues [6]. Such irAEs were not frequently observed with cytotoxic chemotherapy or other classes of targeted agents [5,18]. The irAEs in response to ICIs have varying times to onset and include organ-specific toxicities in skin, endocrine, gastrointestinal, hepatobiliary, pulmonary, and renal, as well as non-organ-specific toxicities such as fatigue, pyrexia, appetite loss, arthralgia, and myalgia [19][20][21][22]. Their presentation can range from mild and manageable, to severe and life threatening if not recognized early and treated with appropriate measures such as corticosteroids [5,23].
Recent studies have demonstrated that the development of irAEs in melanoma [11,[24][25][26][27] and non-small cell lung cancer (NSCLC) [12,18,20,28,29] patients treated with ICIs could correlate with the clinical response. Chinese melanoma patients have a higher proportion of the acral and mucosal types, which have distinct genetic and clinical characteristics, lower somatic mutational burden, and poorer prognoses [30][31][32][33]. However, the relationship between the irAEs and clinical outcomes remains unclear. In this study, we performed a retrospective analysis of the clinical data obtained from 93 advanced melanoma patients treated with anti-PD-1 antibodies at the Sun Yat-sen University Cancer Center and evaluated the association between irAEs and clinical outcomes. The results of this study will help identify patients with advanced melanoma who are most likely to benefit from PD-1 checkpoint inhibitor therapy.

Association between irAEs and response rates
In this study, the objective response rate (ORR) and disease control rate (DCR) in patients who received PD-1 inhibitors were 22.6% and 40.9%, respectively ( Table  3). The ORR and DCR were significantly better in patients who experienced irAEs than those who did not (33.3% versus 7.7%; P = 0.004 and 53.7% versus 23.1%; P = 0.003, respectively) ( Table 3). The ORR and DCR were a little higher in patients who experienced one to two irAEs than those with no irAEs (19.4% versus 7.7% and 35.5% versus 23.1%, respectively), although the results were not statistically significant (P = 0.148 and P = 0.254, respectively) ( Table 3). Moreover, ORR and DCR were significantly better in patients who experienced three or more irAEs than those who experienced no irAEs (ORR: 42.2% versus 7.7%; P < 0.001 and DCR: 78.3% versus 23.1%; P < 0.001) and one to two irAEs (ORR: 42.2% versus AGING 19.4%; P < 0.001 and DCR: 78.3% versus 35.5% P < 0.001) ( Table 3). In addition, patients with grade 1 to 2 irAEs had significantly higher ORR and DCR than those with no irAEs (40.0% versus 7.7%; P = 0.002 and 54.3% versus 23.1%; P = 0.003, respectively) ( Table 3). In contrast, no significant difference was found in the ORR and DCR in patients with grade 3 to 4 irAEs when compared with those with no irAEs (12.5% versus AGING  (Table 3). In addition, the clinical outcomes in patients with grade 3 to 4 irAEs were poorer when compared with those in patients with grade 1 to 2 irAEs (ORR: 12.5% versus 40%; P = 0.176 and DCR: 50.0% versus 54.3%; P = 0.820, respectively) ( Table 3).
The analysis of the association between clinical outcomes and common irAEs revealed that increased PFS was significantly associated with skin irAEs (median 11.0 months; 95% CI, 6.  (Figure 2A, 2B, and 2E). In contrast, no differences in PFS and OS were observed between patients with and without hepatobiliary and gastrointestinal irAEs ( Figure 2C and 2D).  Figure 3A). However, there were no statistically significant differences in the PFS and OS in patients with one to two irAEs when compared with those with no irAEs ( Figure 3A). In addition, patients with grade 1 to 2 irAEs when compared with those with no irAEs showed longer PFS (median 8.7 months; 95% CI, 3.1-14.3 versus 2.8 months; 95% CI, 2.7-2.9, P = 0.005) and OS (22.3 months; 95% CI, 10.6-34.0 versus 8.0 months; 95% CI, 6.7-9.3 P = 0.001) ( Figure 3B). There were no significant differences in the PFS and OS of patients with grade 3 to 4 irAEs when compared with those with grade 1 to 2 or no irAEs ( Figure 3B). Moreover, the effects of irAEs on the prognosis of patients treated  AGING with PD-1 inhibitors were further evaluated by the Cox proportional hazards regression analyses. Multivariable analysis revealed that any irAE and the number of irAEs were significantly associated with increased PFS. Increased OS was also significantly associated with any irAE and the number of irAEs (Table 4).

DISCUSSION
In this study, we retrospectively analyzed the irAE profiles of PD-1 inhibitors including pembrolizumab and nivolumab. Using a relatively large sample size, we demonstrate that irAEs occurring before disease progression are associated with clinical outcomes in Chinese patients with advanced melanoma. Among the 93 patients, we found that 58.1% of them experienced irAEs with a potential immunological etiology. Most of the irAEs were typically mild to moderate in intensity, and only 8.6% of the patients experienced severe irAEs (grade 3 to 4). Consistent with previous reports [14,28], the most commonly observed irAEs were rash, pruritus, vitiligo, thyroiditis, hepatitis, fatigue, and pyrexia. However, the incidence rates of certain types of irAEs were slightly different from those previously reported. In our study, after the commencement of treatment with PD-1 inhibitors, irAEs such as pruritus, rash, vitiligo, and hepatitis occurred in 30.1%, 24.7%, 16.1% and 8.6% of patients, respectively. These proportions are marginally higher than those reported in previous clinical trials (14.1-21%, 12-15%, 6-10.7%, and 0-1.1%, respectively) [4,[34][35][36]. However, the incidence rates of diarrhea and fatigue (5.4% and 15.1 %, respectively) in our study were lower than those previously reported (8)(9)(10)(11)(12)(13)(14)(15)(16).9% and 19.9-25%, respectively) [4,[34][35][36]. One reason for this discrepancy is the different ethnicity, and thereby leads to differences in the baseline characteristics of patients in our study compared with those in other studies. Compared with our study, there were relatively fewer Chinese patients enrolled in the KEYNOTE-001, KEYNOTE-002, KEYNOTE-006, CheckMate-037, and CheckMate-066 trials, which could account for the differences in the irAE profiles.
It has been proposed that the development of irAEs is associated with response to ICIs. This phenomenon was first described in melanoma patients who received ipilimumab therapy [24]. More recently, a retrospective analysis of 576 patients with advanced melanoma received nivolumab monotherapy in United States found a significantly higher ORR in patients who experienced irAEs of any grade than those who did not, with greater benefit in patients with three or more or one to two treatment-related AEs than those with none  Abbreviations: irAEs, immune-related adverse events; PFS, progression-free survival; OS, overall survival; HR, hazard ratio, CI, confidence interval a P value < 0.05 b Covariables included metastasis stage (yes versus no), LDH levels (≤UNL versus>UNL), and liver metastases (yes versus no).

AGING
[27]. Consistent with these findings, our study also demonstrated that the ORR and DCR were significantly better in patients who experienced irAEs of any grade than those who did not. However, we found stronger associations between clinical outcomes and those with three or more irAEs than those with less than three irAEs. We also found that while the ORR and DCR were significantly higher in patients with grade 1 to 2 irAEs than those with no irAEs, but no significantly better outcomes were observed in patients with grade 3 to 4 irAEs than those with no irAEs. Furthermore, we found that patients with grade 3 to 4 irAEs showed poorer clinical outcomes when compared with those with grade 1 to 2 irAEs. A recent retrospective study also found that while ORR was significantly associated with low-grade irAEs in non-melanoma patients treated with PD-1 inhibitors, no significant relationship was found with high-grade irAEs [37]. The poorer clinical response in patients who developed grade 3 to 4 irAEs is potentially due to the termination of immunotherapy when faced with irAEs and the use of steroids.
Prior studies have demonstrated an association between the development of irAEs and clinical survival in tumor patients treated with PD-1 inhibitors. Analysis of 148 melanoma patients treated with nivolumab in the United States, the OS was greater in patients with irAEs especially three or more than those who had no irAEs [25]. A previous prospective study also reported that early irAEs are associated with a better PFS in NSCLC patients receiving nivolumab treatment [18]. Another retrospective study analyzed NSCLC patients treated with pembrolizumab from a single center in the KEYNOTE-001 trial and found that the patients who experienced irAEs had increased PFS and OS, compared to those who did not [20]. Consistent with these previous studies, we also found that the development of irAEs was significantly associated with increased PFS and OS, particularly in patients with three or more irAEs. In addition, we found that prolonged survival of advanced melanoma patients treated with PD-1 inhibitors was associated with skin and endocrine irAEs, as well as fatigue, but not with hepatobiliary and gastrointestinal irAEs. Our findings are consistent with some other studies that have showed an association between skin irAEs and prolonged survival in patients with NSCLC and melanoma treated with nivolumab [25,28]. Moreover, a previous study on malignant melanoma revealed clear survival benefits in patients who received immunotherapy and developed vitiligo-like depigmentation [26]. In contrast to our results, it has been reported that hypothyroidism or hyperthyroidism was not associated with prolonged survival in malignant melanoma [25]. However, thyroid AGING dysfunction was associated with improved OS of NSCLC patients treated with pembrolizumab [12].
The mechanism underlying the association of irAEs with the efficacy and clinical outcomes of treatment with PD-1 inhibitors remains unclear. However, previous studies have demonstrated that melanocytes and melanomas shared common antigens (e.g., MART-1, gp100, and tyrosinase related proteins 1 and 2) and lymphocytes directed against the tumor could crossreact with normal melanocytes and cause irAEs in the skin [11,13,28,38]. In view of these findings, antigen sharing was considered the most likely cause of this association. Therefore, we speculate that irAEs can predict a strong activation of the immune system following the inhibition of PD-1, which can explain the better prognosis in melanoma patients, give the sensitivity of these tumors to immunotherapy.
In conclusion, this retrospective study demonstrated that irAEs were associated with the efficacy of PD-1 inhibitors in Chinese patients with advanced melanoma. Moreover, we showed that patients with at least three irAEs of grade 1 or 2, which include dermal, endocrine and fatigue were most likely to benefit from the inhibition of PD-1. However, our findings should be validated prospectively in subsequent analyses in larger cohorts of patients with advanced melanoma.

Patient population
We performed a retrospective study of Chinese patients with advanced melanoma who received PD-1 inhibitor monotherapy at the Sun Yat-sen University Cancer Center between August 2014 and March 2018. Patients with histologically confirmed advanced melanoma derived from skin and non-skin sections were included in this study. The included patients were treated with intravenous pembrolizumab (2 mg/kg every 3 weeks) or nivolumab (3mg/kg every 2 weeks). After excluding 11 patients who received only one dose of the drug, and 2 patients who died before the first imaging evaluation, a total of 93 patients were finally enrolled in our study. The electronic medical records were reviewed to obtain information including patient demographic, ECOG status, primary site, metastasis stage, BRAF V600E status, prior therapy, number of anti-PD-1 drug doses received, any irAEs, use of corticosteroids, findings of the imaging evaluation, date of progression and start of new treatment or death. The follow-up was ended on June 1, 2018. The Institutional Review Board of the Sun Yat-sen University Cancer Center approved the study, and all the patients provided written informed consent.

Assessments
IrAEs are defined as AEs with a potential immunologic cause that require frequent monitoring and intervention with immunosuppressive and/or endocrine therapy [5,[39][40][41][42][43][44]. All irAEs were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (version 4.0). Radiological evaluations (CT or MRI) were performed at baseline and subsequently at every 8 to 12 weeks to assess tumor responses. Tumor assessments were made based on the Response Evaluation Criteria in Solid Tumors (RECIST version 1.1), and included complete remission (CR), partial remission (PR), stable disease (SD), and progressive disease (PD). The ORR (CR + PR) and DCR (CR + PR + SD) were also calculated. PFS was defined as the interval between the start of the treatment and disease progression or death due to any cause. OS was defined as the interval between the start of the treatment and death due to any cause. Patients who did not progress or were still alive at the last follow-up date were censored.

Statistical analysis
Clinical and demographic characteristics of the patient with and without irAEs were compared using the Fisher exact test for categorical variables. ORRs and DCRs with 95% CIs were estimated using the Clopper-Pearson method. The Fisher exact tests were used to determine the associations between the number/grade of irAEs and the tumor response rates including ORRs and DCRs. The differences in the PFS and OS curves (estimated by the Kaplan-Meier method) based on the absence or presence of any irAEs which were observed before disease progression was evaluated using the logrank test. We used a Cox proportional hazards model to calculate HRs and 95% CIs. A multivariable analysis was performed with adjustment for metastasis stage, LDH levels, and liver metastases. A two-tailed p value < 0.05 was considered to be statistically significant. All the statistical analyses were performed using the IBM SPSS 19.0 software.