Safety, Efficacy, and Biomarker Analysis of Toripalimab in Previously Treated Advanced Melanoma: Results of the POLARIS-01 Multicenter Phase II Trial

In recent years, immune checkpoint inhibitors (ICI) have greatly improved the clinical outcomes of advanced melanoma. As the first indication approved for both nivolumab and pembrolizumab in 2014 (1, 2), melanoma has been considered as an immunogenic cancer type with favorable outcomes for PD-1 blockade therapy. Chronic sun exposure associated ultraviolet (UV)-induced DNA damage is the underlying disease mechanism for melanoma in the white population. Melanoma from chronic sun damage skin (arms, neck, and face) and intermittent sun damage skin (trucks and legs) each comprise approximately half of the cutaneous melanoma in the United States and other western countries (3). In contrast, acral lentiginous melanoma (ALM) and mucosal melanoma are the two most common melanoma subtypes in Asia, accounting for about 50% and 20% of all melanomas respectively in Asian regions (4).

In comparison with the UV-related cutaneous subtype, acral and mucosal melanomas are more aggressive malignancies with lower response rates to standard treatments, including chemotherapy and PD-1 blockade therapy (5–7). Genome-wide mutational study has revealed that acral and mucosal melanomas harbor limited and unique DNA mutations with unknown etiology (8), which provides a molecular basis for discordant clinical response in Asian versus Caucasian melanoma patient population for standard treatments. Retrospective studies showed immunotherapies were less effective treating acral and mucosal subtypes when compared with the UV-related cutaneous melanoma (9). Consistently, in a recent published study of KEYNOTE-151, pembrolizumab showed 13.3% and 15.8% objective response rates (ORR), respectively, in mucosal and acral melanoma subtypes in Chinese patients with advanced melanoma at second-line setting (10). In comparison, nivolumab and pembrolizumab had ORR of 27% and 21%, respectively, in patients with melanoma in the United States at similar settings (11, 12).

Toripalimab, also known as JS001 or TAB001, a humanized IgG₄ mAb against PD-1, has shown preliminary clinical activities in a phase I study in previously treated melanomas (13). Here we report the result from a single-arm open-label phase II registration study (POLARIS-01) evaluating the safety and efficacy of toripalimab in Chinese patients with advanced melanoma refractory to prior systemic treatments.

This study is a phase II, multicenter, single-arm, open-label, clinical trial (ClinicalTrial.gov Identifier: NCT03013101) evaluating the safety and clinical activity of toripalimab in previously treated patients with advanced melanoma. The study protocol and all amendments were approved by the institutional ethics committees of all participating centers. This study was conducted in accordance with the Declaration of Helsinki and the international standards of good clinical practice.

Eligible patients were at least 18 years old with pathologically confirmed local advanced or metastatic melanoma who were previously treated with systemic therapy. Patients must have at least one measurable lesion per Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 at baseline, with Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, adequate organ and bone marrow function, and willingness to provide consent for biopsy samples. Exclusion criteria included history of autoimmune diseases, ongoing infections, or prior anti-PD-1/PD-L1/PD-L2–based immunotherapies.

Melanoma with unknown primary in the study was defined according to Utter and colleagues (14). In this study, 26 (20.5%) patients had no history of primary cutaneous, ocular, or mucosal melanoma and no pathology reports documented a primary lesion. They were classified to melanoma with unknown primary after a thorough physical examination without identifying lesions in areas including skin, eye, anal, and genital at the time of enrollment.

Patients received toripalimab 3 mg/kg once every two weeks via intravenous infusion until disease progression, intolerable toxicity, or voluntary withdrawal of informed consent. Adverse events were monitored continuously and graded according to the National Cancer Institute Common Terminology Criteria (CTCAE) version 4.0. In this study, “definitely related,” “probably related,” and “possibly related” were classified as “treatment-related” AE (TRAE). “Possibly unrelated” and “Definitely unrelated” were classified as “treatment unrelated.” Radiographic imaging was performed before treatment, then once every 8 weeks in the first year and once every 12 weeks from the second year until disease progression and evaluated by investigators using both RECIST v1.1 and Immune-related Response Evaluation Criteria in Solid Tumors (irRECIST). Patients who initially developed progressive disease per RECIST v1.1 were allowed to continue therapy if the investigator considered patients might benefit from further treatment per irRECIST.

The primary endpoint of this study was safety and clinical efficacy by objective response rate (ORR) determined by independent radiologic review committee per RECIST v1.1. The secondary endpoints included pharmacokinetics and immunogenicity of toripalimab (anti-drug antibody, ADA), disease control rate (DCR), duration of response (DOR), progression-free survival (PFS), and overall survival (OS).

Archival or fresh tumor biopsy samples were obtained from patients prior to treatment. PD-L1 expression was determined by IHC staining with SP142 antibody in a central lab (15). PD-L1 expression was evaluated by certified pathologists. PD-L1–positive status was defined as the presence of membrane staining of any intensity in ≥ 1% of tumor cells.

Whole-exome sequencing (WES) was performed with SureSelect Human All Exon V6 kit (Agilent) on tumor biopsies and matched peripheral blood mononuclear cells (PBMC) samples. Genomic alterations including microsatellite stability status, nucleotide variants (SNV), short and long insertions/deletions (INDEL), copy number variants (CNV), and gene rearrangement and fusions were assessed. The tumor mutational burden (TMB) was determined by analyzing somatic mutations including coding base substitution and INDELs per mega-base (Mb).

RNA was extracted from unstained FFPE sections and cDNA synthesis was performed followed by sequencing on NovaSeq 5000/6000 platform. The relative abundance of each annotated transcript was expressed as transcripts-per-million (TPM) and log₂-transformed before analysis. Gene panels covering inflammation or angiogenesis markers were used to obtain a single signature score for each sample. The mean expression of the genes composing the signature was calculated.

At a one-sided significance level of 0.05, a total of 120 patients could provide 90% power to show the efficacy of toripalimab at targeted ORR of 20% versus 10% for alternative second-line therapy using Clopper–Pearson method. A 120-patient sample size was thus planned for this study and 128 patients were enrolled.

Safety analysis included all patients who received at least 1 dose of the study drug (n = 128). One patient with no recurrent cancer was excluded from efficacy analysis (n = 127). ORR and its 95% exact confidence interval (CI) were determined by Clopper and Pearson methodology. Fisher's exact test was used to compute two-tailed P values from contingency tables. PFS and OS were plotted using the Kaplan–Meier method, with median and corresponding two-sided 95% CI. Statistics analyses were performed with SAS version 9.4 or GraphPad Prism software.

From December 26, 2016 to September 15, 2017, a total of 161 patients with melanoma refractory to systemic treatments were screened from six participating centers. 128 patients were enrolled into the study (Supplementary Fig. S1). Baseline characteristics are summarized in Table 1. The average age was 52.5 years old with 57 (45.5%) male and 71 (55.5%) female patients. One patient was later diagnosed with no recurrent cancer and was excluded from efficacy analysis. Among the melanoma subtypes, 50 (39.4%) were acral, 22 (17.3%) were mucosal, 29 (22.9%) were nonacral cutaneous, and 26 (20.5%) were of unknown primary. BRAF mutations were documented in 34 (26.6%) patients at the enrollment. Majority of the patients were heavily pretreated, as 88 (68.7%) patients had at least two prior lines of systemic treatments. Ninety-nine (78.0%) patients had received prior systemic chemotherapy, 68 (53.5%) received interferon, 12 (9.4%) received IL2, and 9 (7.0%) patients were previously treated with ipilimumab. Among chemo-treated patients, 83(65.4%) had received platinum-based therapy.

By the cut-off date of August 15, 2019, 23 months after the last patient was enrolled, patients had received a median of 10 doses of toripalimab (range: 1–73 doses). 116 of 128 (90.6%) of patients experienced treatment-related adverse events (TRAE). The majority of TRAEs were grade 1 or 2. Common (>5%) TRAEs were listed in Table 2. Treatment-related serious adverse events (SAE) occurred in 10 (7.8%) patients. Twenty-five (19.5%) patients experienced grade 3 and above TRAE, including 13 (10.2%) grade 3 and 12 (9.4%) grade 4 (Supplementary Table S1). No patient experienced grade 5 TRAE. Treatment termination due to TRAE occurred in 15 (11.7%) patients; dose delay due to TRAE occurred in 8 (6.3%) patients. TRAEs of special interest that occurred less than 5% included 5 (3.9%) vitiligo, 3 (2.3%) liver injury (2 Grade 4), 2 (1.6%) acute pancreatitis, 2 (1.6%) interstitial lung disease (1 Grade 3), 2 (1.6%) adrenal insufficiency, 2 (1.6%) hypopituitarism, and 1 (0.8%) uveitis (Grade 3).

As of August 15, 2019, among intention-to-treat (ITT) population (n = 127), 61 (48.0%) had died, 46 (36.2%) discontinued treatment, and 20 (15.7%) patients remained on study. The median treatment duration was 4.6 months (range 0.2–33.6 months). Among 127 patients assessed by IRC per RECIST v1.1, 22 patients achieved confirmed objective responses [1 complete response (CR) and 21 partial responses (PR)]. The ORR was 17.3% (95% CI, 11.2–25.0) per RECIST v1.1 and 18.1% (95% CI, 11.8–25.9) per irRECIST (Table 3). Decrease of target lesions of any size from baseline was observed in 49 (38.6%) patients (Fig. 1A). Three stable disease (SD) patients had PRs initially, but the responses were unable to confirm. The median time to response was 3.5 months (95% CI, 1.7–3.6). The median duration of response (DOR) was not reached, as only 9 of 22 patients experienced disease progression after the initial responses (Fig. 1B and C). The DCR was 57.5% (95% CI 48.4 to 66.2) per RECIST v1.1 and 59.8% (95% CI, 50.8–68.4) per irRECIST (Table 3). The median PFS was 3.6 months (95% CI, 2.7–5.3) per RECISTv1.1 and 3.7 months (95% CI, 3.3–9.1) per irRECIST. The estimated overall median OS for the ITT population (n = 127) was 22.2 months (95% CI, 15.3–not reached).

The median OS for patients who experienced objective response (n = 22) or SD (n = 51) was not reached, as only 2 of 22 PR/CR patients and 17 of 51 SD patients had died. The median OS for patients with progressive disease (n = 54) was 9.7 months.

Among four melanoma subtypes of acral, mucosal, nonacral cutaneous and unknown primary, independent review committee (IRC) assessed ORRs were 14.0%, 0%, 31.0%, and 23.1%, respectively, and the median PFS were 3.2, 1.9, 5.5, and 7.3 months per RECIST v1.1 (Supplementary Table S2-1; Supplementary Fig. S2A). The median OS for acral and mucosal and subtypes were 16.9 months and 10.3 months, while the median OS for nonacral cutaneous and melanomas with unknown primary were not reached by the cut-off date (Supplementary Fig. S2B).

The steady-state median trough plasma concentration of toripalimab was 39.8 μg/mL (range 4.9–92.4 μg/mL), well above the full PD-1 blocking concentration of 1.5 μg/mL (10 nmol/L; ref. 16). Detection of ADA was performed in all 128 patients. Samples from 16 (12.5%) patients were ADA positive. Among them, 3 were positive before treatment reflecting reexisting conditions and 13 (10.2%) were ADA positive after treatment. However, none of ADA-positive individual had consecutive positive samples. Only 1 of 16 patients had decreased toripalimab trough concentration coincided with the detection of ADA, indicating a neutralizing activity. There were no significant differences in the incidence of AE, SAE, grade 3 and above AE, termination, or dose delay, or clinical efficacy between ADA-positive and ADA-negative patients.

Tumor biopsy samples were obtained from all 127 patients and PD-L1 expression was evaluated by IHC staining using SP142 antibody. Twenty-six (20.5%) PD-L1 positive and 84 (66.1%) PD-L1⁻ patients were identified (Fig. 2A). PD-L1 expression status from 17 (13.4%) patients was unknown. Among the four melanoma subtypes, PD-L1⁺ percentage were significantly lower in acral (6.8%) and mucosal (10.5%) subtypes than nonacral cutaneous (37.5%) and melanomas with unknown primary (52.2%; Fig. 2B). PD-L1⁺ patients responded favorably to toripalimab treatment than PD-L1⁻ patients in both ORR (38.5% vs. 11.9%, P = 0.0065) and DCR (80.8% vs. 48.8%, P = 0.006). Furthermore, PD-L1⁺ patients also showed significant survival advantage than PD-L1⁻ patients in PFS, 7.7 months versus 2.7 months[ HR = 0.53 (95% CI, 0.32–0.88); P = 0.013] as well as in OS, not reached versus 14.4 months [HR = 0.35 (95% CI, 0.19–0.63); P = 0.0005; Fig. 2C and D].

As previously described, whole exome sequencing (WES) was performed on both tumor biopsies and paired PBMCs (17). Valid results were obtained from 98 patients (Fig. 2A). TMB was determined by analyzing somatic mutations within the coding region of the human genome. TMB was generally low in this study with median TMB at 1.5 mutations per million base pairs (Muts/Mb). No MSI-high patients were identified. Only 6 patients had TMB over 10 Mutations/Mb, including three patients over 20 Muts/Mb. Among the four melanoma subtypes, mucosal melanoma had the lowest TMB with median TMB at 1.6 Muts/Mb and only 1 patient (6.7%) had TMB over 3.6 Muts/Mb (Fig. 2B). A cutoff of the top 20% of TMB value (3.6 muts/Mb) was selected to define TMB-high population in this study as suggested by Samstein and colleagues (18). Using 3.6 mutations per Mb as the cutoff, patients with TMB≥3.6 Mutations/Mb (n = 20) had 30% ORR while patients with TMB <3.6 Mutations/Mb (n = 78) had 12.8% ORR. The difference was not statistically significant (P = 0.088; Fig. 2A; Supplementary Table S2-2). TMB ≥3.6 Mutations/Mb group also showed numerically better PFS and OS, but the differences were also not statistically significant (Fig. 2E and F). Notably, TMB ≥3.6 Mutations/Mb (n = 20) and PD-L1⁺ (n = 26) groups were largely two independent population in this study, as only 7 of 26 PD-L1⁺ patients had TMB ≥3.6 Mutations/Mb (Fig. 2A).

WES identified 19,278 genetic alternations from 98 available patients using paired tumor biopsy and PBMCs, including 7,964 missense mutations, 509 gene deletions, 482 rearrangements, 288 alternative splicing sites, 129 frameshift truncations, and 8,157 gene amplifications. After excluding genes frequently mutated in public exomes, the most frequently altered genes (≥10%) were BRAF (33%), TERT (32%), CDKN2A (18%), NRAS (16%), CDK4 (12%), APOB (11%), CCND1 (11%), AGAP2 (11%), NF1 (10%), LRP1B (10%), MDM2 (10%), and KIT (10%; Fig. 3).

The sequencing results from 98 available patients showed distinctive patterns of genomic alterations among melanoma subgroups. BRAF mutations occurred at higher frequencies in nonacral cutaneous subtype (11/23, 48%) and melanomas with unknown primary (14/21, 67%), while occurred less often in acral (7/39, 18%) and mucosal (0/15, 0%) subtypes from this study. In contrast, NF1 mutations were more enriched in mucosal melanoma (4/15, 27%) than the other three subtypes (8%, 4%, and 10%; Supplementary Table S3). In addition, CDK4 or CCND1 (Cyclin D1) amplifications were observed in 33% (13/39) acral and 20% (3/15) mucosal subtypes, while only found in 9% (2/23) nonacral cutaneous and 10% (2/21) melanomas with unknown primary. Notably, 70% (7/10) CCND1 copy number variants as part of the 11q13 genomic amplification were observed in the acral subtype.

Genomic alternations were analyzed for correlation with clinical efficacy. Patients with either NRAS mutations (n = 16) or CCDN1 amplifications (n = 10) from this study were associated with poor response to toripalimab treatment (ORR 6.3% and 0%, respectively).

As described previously, RNA sequencing and expression profiling was performed on mRNA extracted from tumor biopsies (17). Valid results were obtained from 46 patients. The gene expression signatures of IFNγ-related gene panel (IDO1, CXCL10, CXCL9, HLA-DRA, STAT1, IFNG), inflammation-related gene panel (IL6, CXCL1, CXCL2, CXCL3, CXCL8, PTGS2), and angiogenesis-related gene panel (VEGFA, KDR, ESM1, PECAM1, ANGPTL4, CD34) were evaluated (Supplementary Fig. S3). There were no significant differences of angiogenesis panel, IFNγ-related gene panel and inflammation-related gene panel signature scores from patients with clinical benefit (CR+PR+SD) and patients with progressive disease (Supplementary Fig. S3).

Additional biomarkers or subgroups analyzed for correlation with clinical efficacy included age, gender, baseline ECOG score, prior lines of treatment, prior systemic chemotherapy, prior treatment with ipilimumab, lactate dehydrogenase (LDH) levels, and ADA status (Supplementary Table S4). Among the subgroups, patients with 1 or 2 lines of prior systemic treatments (n = 71) had significantly better ORR than patients with three lines or more of prior treatments (n = 56; 23.9% vs. 8.9%; P = 0.03). Patient with baseline LDH level in normal range also had numerically better ORR than those above upper limit of normal (19.6% vs. 10.0%), but the difference was not statistically different. Nine patients from this study have been previously treated with ipilimumab in a prior trial. No significant difference was found in ORR between ipilimumab-treated and nontreated patients (22.2% vs. 16.9%, P = 0.65).

The paradigms of disease management for metastatic melanoma have been greatly reshaped within the past decade due to the breakthrough of checkpoint blockade therapy (19). However, most clinical studies have focused on nonacral cutaneous melanoma, the major subtype in the white population, while limited data are available for acral and mucosal melanomas, the dominant subtypes in Asia and Africa. It has been well documented that despite a common tissue origin, nonacral cutaneous, acral, and mucosal melanomas are unique in terms of risk factors, epidemiology, genomic alternations, biological behaviors, clinical features, and response to standard treatments (4, 8). Mucosal melanoma is a known aggressive subtype with poor prognosis among melanomas. In a 2016 report, among 3,454 patients with melanoma, the median OS for mucosal, uveal, acral, nonacral cutaneous, and unknown primary melanoma was 9.1, 13.4, 11.4, 11.7, and 10.4 months, respectively (20).

In retrospective analysis, acral and mucosal subtypes also appeared less responsive to PD-1 blockade than nonacral cutaneous melanoma. As reported from the CHECKMATE-172 study, 723 (71.7%) nonacral cutaneous, 103 (10.2%) ocular, 63 (6.3%) mucosal, and 55 patients with (5.5%) acral melanoma were enrolled and treated with nivolumab. Among the subtypes, acral melanoma had similar overall survival with nonacral cutaneous melanoma (25.8 vs. 25.3 months), while ocular and mucosal melanoma had lower median overall survival at 12.6 and 11.5 months, respectively. In a recent published result of KEYNOTE-151 study, pembrolizumab showed 19.5%, 15.8%, 13.3%, and 12.5% ORRs respectively in nonacral cutaneous, acral, mucosal, and melanoma with unknown primary in Chinese patients with advanced melanoma at second-line setting (10).

Here we report, the results of safety and efficacy of PD-1 blockade in patients with advanced melanoma in China with predominantly acral and mucosal subtypes. Toripalimab demonstrated a manageable safety profile and durable antitumor activity. No novel safety signal was identified when compared with the same class of drugs. After treatment for over 2 years, median duration of response was not reached as majority of responses were still ongoing. Toripalimab appeared more efficacious for nonacral cutaneous than acral and mucosal subtypes (ORR: 31.0%, 14.0%, and 0%; mPFS: 5.5, 3.2 and 1.9 months, mOS: not reached, 16.9 and 10.3 months, respectively). Notably, patients with positive PD-L1 staining in tumor biopsies responded significantly better than PD-L1⁻ patients in terms of ORR (38.5% vs. 11.9%, P = 0.0065), PFS (7.7 months vs. 2.7 months, HR = 0.53; 95% CI, 0.32–0.88, P = 0.013), and OS (not reached vs. 14.4 months, HR = 0.35; 95% CI, 0.19–0.63; P = 0.0005).

In recent years, tumor mutational burden (TMB) was found to be associated with clinical responses to immune checkpoint inhibitors in patients with NSCLC (21) and gastric cancer (22). TMB value was surprisingly low in this study with median TMB at 1.5 Muts/Mb, presumably due to high percentage of acral and mucosal subtypes in the cohort. Neither acral nor mucosal melanoma is related to chronic UV exposure and thus both subtypes have low DNA mutational burden in comparison with cutaneous melanoma. A cutoff of the top 20% of TMB value (3.6 muts/Mb) was selected to define TMB high population in this study as suggested by Samstein and colleagues (18). Patients with TMB≥3.6 Mutations/Mb had 30% ORR while patients with TMB <3.6 Mutations/Mb had 12.8% ORR. The difference was not statistically significant (P = 0.088; Fig. 2A). Top 20% TMB value in each subtype was also used as cutoff for efficacy analysis. As in shown Supplementary Table S2-2, none of the differences in subgroups were statistically significant. The use of TMB to predict clinical response in patients with melanoma warrants further evaluation.

No significant difference of IFNγ-related gene signature scores was observed for available responders and nonresponders from this study. It is possible that IFNγ signatures might have different prognostic value in different melanoma subtypes. In contrast to the nonacral cutaneous melanoma, the application of IFNγ signatures scores in acral and mucosal subtypes need to be further investigated.

Both WES and RNA-seq data were uploaded to GSA public database (https://bigd.big.ac.cn/gsa-human/s/IlFfvCn6) and clinical characteristics were shown in Supplementary Table S5.

Among melanoma subgroups, PD-L1⁺ percentage were high among nonacral cutaneous (37.5%) and melanomas with unknown primary (52.2%), but low in acral (6.8%) and mucosal (10.5%) subtypes. In addition, mucosal melanoma also had the lowest TMB value with only one patient (6.7%) with TMB over 3.6 Muts/Mb. Combined lack of PD-L1 positive expression and low TMB value indicated a poorly immunogenic melanoma subtype of mucosal origin, which was least responsive to toripalimab monotherapy in this study. As shown in Supplementary Table S2-1, investigator assessed ORR per RECIST v1.1 for mucosal melanoma subgroup in this study was 9.1% (2PR). However, both responses were reviewed as SD by IRC. In addition, mucosal melanoma patients from this study (n = 22) were heavily pretreated with a median of two prior lines of treatment and 45% (10/22) patients had at least three prior lines of treatment.

We performed WES on tumor biopsies and paired PBMCs to explore the mutational landscape and correlation with clinical efficacy. Consistent with previous reports (23, 24), BRAF mutations occurred at significant lower frequencies in acral (7/35, 20%) and mucosal (1/13, 7.7%) melanomas than nonacral cutaneous melanomas (10/17, 58.8%) from this study. As BRAF and MEK inhibitors were not commercially available in China during the enrollment of this study, only 1 of 34 patients with documented BRAF mutations was treated with vemurafenib in a previous clinical trial.

Notably, patients with NRAS mutations (n = 16) in this study, including 7 acral, 3 mucosal, 3 cutaneous, and 3 unknown primary, had 6.3% (1/16) ORR to toripalimab treatment, indicating NRAS mutation as a potential resistance mechanism. This result is in sharp contrast to previous reports that NRAS mutants were associated with better objective response to anti-PD-1 therapies in patients with melanoma from Europe (25) and United States (26). Limited sample size and NRAS mutations from different melanoma subtypes might underlie the observed difference.

In this study, we observed differential clinical responses and unique mutation signatures among melanoma subtypes. For acral and mucosal subtypes in particular, combination treatment strategies are likely needed to further improve the clinical outcome of PD-1 blockade. We recently reported high frequency of CDK4 pathway aberrations from a large acral melanoma cohort (n = 514) and showed efficacy of CDK4/6 targeting therapy in preclinical xenograft tumor models (27). Consistently, in this study, we found CDK4 or CCND1(Cyclin D1) amplifications in 33% (13/39) acral and 20% (3/15) mucosal subtypes, but only in 9% (2/23) nonacral cutaneous melanomas and 10% (2/21) melanomas with unknown primary. Intriguingly, CCND1 copy number variations (n = 10) as part of 11q13 genomic amplification correlated with poor response to toripalimab (0% ORR), suggesting a potential CDK4/6 targeting therapy or combination strategy of CDK4/6 inhibitor with anti-PD-1 for CCND1-amplified melanomas. For advanced mucosal melanoma, we recently showed evidence of enhanced efficacy of the combination of toripalimab with a VEGFR inhibitor axitinib in a phase Ib study (17). Patients may benefit from the combination therapy regardless of tumor PD-L1 expression and TMB value, suggesting that toripalimab combined with axitinib is a promising treatment option for advanced mucosal melanoma, who otherwise might have limited response to PD-1 blockade alone.

In summary, this phase II study provides evidence for the safety and efficacy of toripalimab for patients with pretreated advanced melanoma in China. The safety profile of toripalimab was manageable. Patients with PD-L1⁺ tumor biopsy preferentially benefited from the treatment. Toripalimab monotherapy appeared to be more effective treating nonacral cutaneous and melanoma with unknown primary than acral and mucosal subtypes. On the basis of the results of this trial, toripalimab was approved by the Chinese National Medical Product Administration (NMPA) for second-line treatment of metastatic melanoma on December 17, 2018. A phase III randomized trial comparing toripalimab monotherapy versus chemotherapy for first-line treatment of advanced melanoma is currently being conducted to confirm the safety and clinical efficacy of toripalimab (NCT03430297).

W. Wang is an employee of OrigiMed. H. Feng is an employee of and holds ownership interest (including patents) in Shanghai Junshi Bioscience Co. Ltd. S. Yao is an employee of Shanghai Junshi Biosciences Co. Ltd. J. Guo is an unpaid consultant/advisory board member for MSD, Roche, Pfizer, Bayer, Novartis, Simcere, Shanghai Junshi Bioscience, and Oriengene. No potential conflicts of interest were disclosed by the other authors.

Conception and design: Z. Chi, H. Song, H. Wu, J. Guo

Development of methodology: B. Tang, Z. Chi, L. Dong, H. Song, J. Guo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B. Tang, Z. Chi, Y.-B. Chen, X. Liu, D. Wu, J. Chen, X. Song, L. Dong, S.-K. Qin, X. Zhang, J. Guo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B. Tang, Z. Chi, W. Wang, H. Song, S. Yao, S.-K. Qin, X. Zhang, J. Guo

Writing, review, and/or revision of the manuscript: B. Tang, Z. Chi, Y.-B. Chen, X. Liu, D. Wu, J. Chen, X. Song, W. Wang, S. Yao, S.-K. Qin, X. Zhang, J. Guo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B. Tang, Z. Chi, Y.-B. Chen, X. Liu, D. Wu, J. Chen, X. Song, H. Feng, S.-K. Qin, X. Zhang, J. Guo

Study supervision: J. Guo

This study is supported by National Major Science & Technology Major Projects (2015ZX09102017, 2017ZX09302009), National Natural Science Foundation of China (81672696, 81772912), Beijing Municipal Administration of Hospitals Clinical medicine Development of special funding support (ZYLX201603), and Beijing Municipal Science & Technology Commission (Z161100000516062).

This study is sponsored by Shanghai Junshi Biosciences. The authors thank the patients who participated in this study and their families.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.