Purpose:

No standard treatment exists for platinum-refractory, recurrent/metastatic nasopharyngeal cancer (NPC). This phase II study (NCT02605967) evaluated progression-free survival (PFS) of spartalizumab, an antiprogrammed cell death protein-1 (PD-1) monoclonal antibody, versus chemotherapy, in NPC.

Patients and Methods:

Patients with nonkeratinizing recurrent/metastatic NPC who progressed on/after platinum-based chemotherapy were enrolled. Spartalizumab was dosed 400 mg once every 4 weeks, and chemotherapy was received per investigator's choice.

Results:

Patients were randomized to receive either spartalizumab (82 patients) or chemotherapy (40 patients). The most common spartalizumab treatment-related adverse events were fatigue (10.3%) and pruritus (9.3%). Median PFS in the spartalizumab arm was 1.9 months versus 6.6 months in the chemotherapy arm (P = 0.915). The overall response rate in the spartalizumab arm was 17.1% versus 35.0% in the chemotherapy arm. Median duration of response was 10.2 versus 5.7 months in the spartalizumab versus chemotherapy arms, respectively. Median overall survival was 25.2 and 15.5 months in the spartalizumab and chemotherapy arms, respectively. Tumor RNA sequencing showed a correlation between response to spartalizumab and IFNγ, LAG-3, and TIM-3 gene expression.

Conclusions:

Spartalizumab demonstrated a safety profile consistent with other anti–PD-1 antibodies. The primary endpoint of median PFS was not met; however, median overall survival and median duration of response were longer with spartalizumab compared with chemotherapy.

Translational Relevance

There is an unmet therapeutic need for patients with platinum-refractory, recurrent/metastatic nasopharyngeal cancer (NPC). This phase II study investigated the efficacy of the anti–PD-1 monoclonal antibody spartalizumab (PDR001) versus investigator's choice of chemotherapy in patients with recurrent/metastatic NPC who had progressed on/after platinum-based chemotherapy. The primary endpoint of prolonged progression-free survival with spartalizumab versus chemotherapy was not met. Median overall survival and duration of response were longer in the spartalizumab arm versus chemotherapy arm. Spartalizumab demonstrated a safety profile consistent with other anti–PD-1 antibodies. Further studies are required to identify the appropriate patient population, for example by PD-L1 and/or IFNγ signature gene expression, for combination therapies with anti–PD-1 agents in NPC.

Nasopharyngeal carcinoma (NPC) is a form of head and neck cancer that usually develops around the ostium of the eustachian tube in the lateral wall of the nasopharynx (1). The World Health Organization classifies NPC histologically into several subtypes. Type 1, keratinizing squamous cell carcinoma, consists of well-differentiated cells that produce keratin, whereas type 2 is nonkeratinizing carcinoma, differentiated or undifferentiated and does not produce keratin. Type 3, basaloid squamous cell carcinoma, is also nonkeratinizing and characterized by less differentiated cell types (2, 3). The association with Epstein–Barr virus (EBV) makes NPC distinct from other head and neck cancers; type 2 and type 3 NPCs are EBV-associated, whereas EBV infection is generally absent in type 1 (1). Although rare on a global scale, NPC is a leading form of cancer in Southeast Asia and Southern China, where the prevalence rate is 25 to 50 cases per 100,000 people (4).

Approximately 70% of patients with newly diagnosed NPCs present with locally advanced disease; for these patients, platinum-based chemoradiotherapy (CRT) has become the standard of care (5). However, although high initial response rates and survival benefits have been reported with CRT compared with radiotherapy alone (6–10), 15% to 58% of patients with NPC will experience recurrent disease and must undergo re-treatment (11). A pivotal phase III trial has shown that gemcitabine plus cisplatin significantly improves progression-free survival (PFS) with a 45% lower risk of disease progression (or death) in recurrent or metastatic NPC, compared with fluorouracil plus cisplatin (12). This established the gemcitabine and cisplatin combination as the standard first-line treatment option for recurrent or metastatic NPC. For patients with locally advanced recurrent or metastatic NPC who have progressed on or after the platinum-based first-line therapy, multiple single-arm phase II studies with chemotherapy combinations have been conducted (13, 14). Despite this, there is currently no standard-of-care treatment for this patient population.

Spartalizumab (PDR001) is a humanized IgG4 monoclonal antibody that binds to programmed cell death protein-1 (PD-1) with subnanomolar affinity, blocking interaction with ligands, programmed death-ligand 1 (PD-L1) and PD-L2 (15). Recently published, single-arm phase I/II studies of monotherapy PD-1 inhibitors have shown preliminary activity in recurrent/metastatic NPC (16–18).

Here, we report the first randomized controlled phase II study, designed to assess the efficacy of spartalizumab versus chemotherapy, in patients with recurrent or metastatic nonkeratinizing NPC, who progressed on/after platinum-based chemotherapy.

Clinical study design

This is a phase II, open-label, multicenter, randomized controlled study of spartalizumab in patients with recurrent or metastatic nonkeratinizing NPC. The data cutoff date was October 11, 2018; enrollment to this study has been completed.

This Novartis-sponsored study was performed according to the principles of the Declaration of Helsinki and performed in compliance with Good Clinical Practice. The study protocol was approved by an Independent Ethics Committee or Institutional Review Board before the start of the study. Written informed consent was obtained from each patient.

Study objectives

The primary objective was to determine the efficacy of spartalizumab versus chemotherapy in patients with locally advanced recurrent or metastatic NPC by assessing PFS per Response Evaluation Criteria In Solid Tumors (RECIST) v1.1 (central assessment). Secondary objectives included further assessing the antitumor activity of spartalizumab versus chemotherapy by measuring overall survival (OS), overall response rate [ORR: complete response (CR) + partial response (PR)] and duration of response (DOR) as per RECIST v1.1, using central assessment. Additional secondary objectives included investigating the safety and tolerability of spartalizumab, determining the pharmacokinetic (PK) profile of spartalizumab, assessing the emergence of spartalizumab antibodies (data will be reported in a separate publication), observing the pharmacodynamic effect of spartalizumab in peripheral blood, and assessing potential predictive biomarkers of the efficacy of spartalizumab.

Treatment plan and drug administration

Patients were randomized in a 2:1 ratio to the spartalizumab or chemotherapy arm. Spartalizumab was administered intravenously at 400 mg every 4 weeks in 28-day cycles until unacceptable toxicity, progressive disease per immune-related response criteria (irRC), and/or treatment discontinuation due to patient/physician's decision. Chemotherapy per investigator's choice was dosed according to the drug label(s).

Patients in the chemotherapy arm were permitted to cross over to the spartalizumab arm if there were evidence of radiologic progression (as per RECIST v1.1), assessed by independent central review, and the investigator believed this was the best treatment option for the patient.

Safety and efficacy assessments

Tumor response was evaluated locally and centrally per RECIST v1.1 and irRC, at baseline and every 8 weeks until week 40. Subsequent assessments were taken every 12 weeks, until progression or patient withdrawal. Regular safety assessments were based on physical examination, Eastern Cooperative Oncology Group (ECOG) performance status, laboratory parameters, and cardiac assessments. Adverse events (AEs) were assessed and graded according to the Common Terminology Criteria for Adverse Events (CTCAE) v4.03.

Statistical analysis

The primary analysis was planned after approximately 70 PFS events had occurred across both treatment arms, giving 85% power to detect a hazard ratio of 0.55 (corresponding to a median PFS of 10 months in the spartalizumab arm and 5.5 months in the comparator arm), using a log-rank test at a one-sided significance level of 0.1. This assumption was based on historical PFS/time-to-progression (TTP) data from studies of second-line treatment of metastatic or recurrent NPC with single-agent and/or doublet chemotherapy, which reported a median TTP of 5 to 6 months (19–24). Considering a uniform recruitment time of approximately 14 months and a 15% dropout rate at 12 months, approximately 114 patients needed to be randomized (2:1) across the two arms in order to observe 70 PFS events approximately 20 months after First Patient First Visit. The sample size was computed using EAST trial design software v5.4.

Time to event endpoints were summarized using the Kaplan–Meier method and categorical endpoints using frequency and percentages. The primary endpoint of PFS was analyzed by study arm using a stratified log-rank test with an overall one-sided 10% level of significance and the hazard ratio was estimated using a stratified Cox regression. The secondary objectives of OS, DOR, best overall response (BOR), and ORR were summarized by treatment arm. For OS, a stratified Cox regression model was used to estimate the hazard ratio, along with 95% confidence intervals (CIs). For ORR, the corresponding 95% CI based on the exact binomial distribution was used. Note that for patients who crossed over to the spartalizumab arm, data collected post crossover were excluded from the analysis.

Patients

Adults aged ≥ 18 years with nonkeratinizing locally advanced recurrent or metastatic NPC who progressed on/after platinum-based chemotherapy were enrolled. The key inclusion criteria included patients with at least one measurable lesion (per RECIST v1.1) progressing or new since the last antitumor therapy. All patients had an ECOG performance status ≤ 2 and received at least one prior therapy for recurrent or metastatic disease (and up to two prior systemic therapies in the recurrent or metastatic setting). Patients were not allowed to participate in additional parallel investigational drug or device studies, or have a history of receiving prior PD-1– or PD-L1–directed therapy. Other exclusion criteria are described in the study protocol.

Pharmacokinetic and biomarker analyses

Blood samples were collected at specified time points from patients in the spartalizumab arm for PK and cytokine analyses. PK parameters included total systemic exposure (AUCtau), peak serum concentration (Cmax), time of maximum concentration observed (Tmax), and elimination half-life (T1/2). A 30-plex Meso Scale Discovery (MSD) analysis was performed on plasma samples taken pretreatment and at several on-treatment timepoints, to quantify soluble cytokine levels.

Archival or newly obtained tumor samples were collected at screening. Protein expression of immunologic markers was assessed by immunohistochemistry (IHC); PD-L1 expression in tumor samples was measured according to percent positive tumor cells and CD8 according to percent marker area. IHC staining for PD-L1 was performed on the Dako Autostainer Link 48 system with the 22C3 mouse monoclonal primary antibody and EnVision FLEX visualization system, as described in the PD-L1 IHC 22C3 pharmDx package insert. The percentage of tumor cells with partial or complete membranous staining of PD-L1 was assessed. IHC staining for CD8 was performed on the Ventana Benchmark XT system with the Dako CD8 mouse monoclonal primary antibody (clone C8/144B). Images of whole tumor sections were captured using a Mirax scanner (Zeiss) and MSD with Definiens (Definiens AG). DAB (3,3′-diaminobenzidine) intensity was quantified as percent positive pixels.

RNA sequencing (RNA-seq) was used to assess gene expression and EBV transcripts in the tumor. For RNA-seq, single gene analysis was performed on TIM-3 and LAG-3, as preselected immunologic markers of interest. IFNγ signaling was assessed using a composite gene signature (CCL5, CD2, CD3D, CD3E, CIITA, CXCL10, CXCL13, CXCR6, GZMB, GZMK, HLA-DRA, HLA-E, IDO1, IL2RG, LAG-3, NKG7, STAT1, and TAGAP; ref. 25). RNA was extracted from formalin-fixed paraffin-embedded tissue biopsies. As previously described (26), mRNA enrichment was performed using RNase H digestion to deplete rRNA. This rRNA-depleted RNA was then fragmented and converted to cDNA. Sequencing libraries were constructed from the cDNA with the TruSeq RNA Library Preparation Kit v2 (Illumina No. RS-122-2001 and No. RS-122-2002). An Illumina HiSeq sequencing system was used to sequence the resulting libraries with 100-bp paired-end reads to a target depth of 50 million total reads per sample. Next-generation sequencing Data Processing Sequencing reads were aligned to the human reference genome (hg19) using STAR (27). HTSeq was used to quantify the number of reads aligned to each gene in the RefSeq transcriptome (28). Sequencing data were evaluated for quality, and low complexity libraries with < 2 million estimated unique read pairs were excluded from downstream analysis. Gene count data were normalized using the trimmed mean of M values method as implemented in edgeR (29). All downstream gene signature analyses were performed on the log2 of the normalized gene count data. EBV transcript number in tumor was assessed by aligning RNA-seq reads against NC_007605 (ref. 30; Release 95 of NCBI's Viral genomes database) in both chemotherapy- and spartalizumab-treated patients.

In this study, plasma EBV DNA for all patients were centrally analyzed at a third-party laboratory. An issue during stability check which had led to the use of unstable laboratory samples for the measurement of plasma EBV DNA was identified by the laboratory and was communicated to us on June 15, 2021. The baseline values of plasma EBV DNA for two patients were affected by this issue and cannot be removed from the current analysis without a new data transfer from the laboratory, the latter would require the signing of a new contract. We have performed an impact assessment and concluded that the overall impact is minimal as only a limited number of samples were affected and plasma EBV DNA assessment is not part of this study eligibility criteria. We also concluded that the issue does not have any impact on our study primary and secondary endpoints, and therefore does not change the overall conclusions of this study.

Patient population and treatment

At data cutoff (October 11, 2018), 122 patients with nonkeratinizing NPC were randomized. Patients randomized to spartalizumab treatment are defined as the “spartalizumab arm”; patients randomized to chemotherapy treatment as the “chemotherapy arm”; patients who crossed over from chemotherapy treatment to spartalizumab treatment as the “crossover group”; and all patients in the study who received spartalizumab treatment as the “all-spartalizumab group” (spartalizumab arm + crossover group).

Specifically, 82 patients were randomized to the spartalizumab arm and 40 patients to the chemotherapy arm; in the chemotherapy arm, one patient was not treated due to the physician's decision (Supplementary Fig. S1).

At data cutoff (October 11, 2018), 66 patients (80.5%), 37 patients (92.5%), and 22 patients (88.0%) in the spartalizumab arm, chemotherapy arm, and crossover group, respectively, had discontinued treatment, mostly due to disease progression (Supplementary Table S1).

Patient demographics and other baseline characteristics were well balanced between the spartalizumab and chemotherapy arms (Table 1). The median age was 51 years (range, 21–74 years) in the spartalizumab arm and 50 years (range, 26–78 years) in the chemotherapy arm. More white patients were enrolled in the spartalizumab arm than in the chemotherapy arm (14.6% vs. 5.0%). In addition, the spartalizumab arm had more patients with an ECOG performance status of 0 than the chemotherapy arm (42.7% vs. 27.5%).

Table 1.

Patient demographics and characteristics.

CharacteristicSpartalizumab 400 mg every 4 weeks (N = 82)Chemotherapya (N = 40)Crossover to spartalizumab 400 mg every 4 weeks (N = 25)
Median age, y (range) 51 (21–74) 50 (26–78) 49 (26–78) 
Sex, n (%) 
 Male 68 (82.9) 33 (82.5) 20 (80.0) 
 Female 14 (17.1) 7 (17.5) 5 (20.0) 
Race, n (%) 
 Asian 70 (85.4) 37 (92.5) 23 (92.0) 
 White 12 (14.6) 2 (5.0) 1 (4.0) 
 Black 1 (2.5) 1 (4.0) 
ECOG PSb, n (%) 
 0 35 (42.7) 11 (27.5) 6 (24.0) 
 1 45 (54.9) 28 (70.0) 19 (76.0) 
 2 2 (2.4) 
Prior lines of systemic therapy, n (%) 
 1 16 (19.5) 9 (22.5) 5 (20.0) 
 2 47 (57.3) 23 (57.5) 14 (56.0) 
 3 18 (22.0) 7 (17.5) 6 (24.0) 
 ≥ 4 1 (1.2) 1 (2.5) 
Best response at last line of therapy, n (%) 
 CR 5 (6.1) 3 (7.5) 2 (8.0) 
 PR 18 (22.0) 10 (25.0) 5 (20.0) 
 SD 8 (9.8) 7 (17.5) 5 (20.0) 
 PD 28 (34.1) 8 (20.0) 6 (24.0) 
 UNK 5 (6.1) 2 (5.0) 2 (8.0) 
 NA 18 (22.0) 10 (25.0) 5 (20.0) 
Plasma EBV DNA titer at baseline, n (%) 
 Negative 7 (8.5) 
 < 108 (copies/mL) 5 (6.1) 1 (2.5) 
 108–1,000,000 (copies/mL) 62 (75.6) 33 (82.5) 21 (84.0) 
 > 1,000,000 (copies/mL) 7 (8.5) 4 (10.0) 4 (16.0) 
 Missing 1 (1.2) 2 (5.0) 
PD-L1 expression in archival or new baseline sample, percent positive tumor cells 
n, (%) 78 (95.1) 38 (95.0) 24 (96.0) 
 Median (range) 20.0 (0–100) 60.0 (0–100) 60.0 (0–100) 
CD8 expression, percent marker area 
n (%) 78 (95.1) 37 (92.5) 24 (96.0) 
 Median (range) 4.23 (0.1–31.8) 5.74 (0.3–29.8) 5.58 (0.3–20.4) 
CharacteristicSpartalizumab 400 mg every 4 weeks (N = 82)Chemotherapya (N = 40)Crossover to spartalizumab 400 mg every 4 weeks (N = 25)
Median age, y (range) 51 (21–74) 50 (26–78) 49 (26–78) 
Sex, n (%) 
 Male 68 (82.9) 33 (82.5) 20 (80.0) 
 Female 14 (17.1) 7 (17.5) 5 (20.0) 
Race, n (%) 
 Asian 70 (85.4) 37 (92.5) 23 (92.0) 
 White 12 (14.6) 2 (5.0) 1 (4.0) 
 Black 1 (2.5) 1 (4.0) 
ECOG PSb, n (%) 
 0 35 (42.7) 11 (27.5) 6 (24.0) 
 1 45 (54.9) 28 (70.0) 19 (76.0) 
 2 2 (2.4) 
Prior lines of systemic therapy, n (%) 
 1 16 (19.5) 9 (22.5) 5 (20.0) 
 2 47 (57.3) 23 (57.5) 14 (56.0) 
 3 18 (22.0) 7 (17.5) 6 (24.0) 
 ≥ 4 1 (1.2) 1 (2.5) 
Best response at last line of therapy, n (%) 
 CR 5 (6.1) 3 (7.5) 2 (8.0) 
 PR 18 (22.0) 10 (25.0) 5 (20.0) 
 SD 8 (9.8) 7 (17.5) 5 (20.0) 
 PD 28 (34.1) 8 (20.0) 6 (24.0) 
 UNK 5 (6.1) 2 (5.0) 2 (8.0) 
 NA 18 (22.0) 10 (25.0) 5 (20.0) 
Plasma EBV DNA titer at baseline, n (%) 
 Negative 7 (8.5) 
 < 108 (copies/mL) 5 (6.1) 1 (2.5) 
 108–1,000,000 (copies/mL) 62 (75.6) 33 (82.5) 21 (84.0) 
 > 1,000,000 (copies/mL) 7 (8.5) 4 (10.0) 4 (16.0) 
 Missing 1 (1.2) 2 (5.0) 
PD-L1 expression in archival or new baseline sample, percent positive tumor cells 
n, (%) 78 (95.1) 38 (95.0) 24 (96.0) 
 Median (range) 20.0 (0–100) 60.0 (0–100) 60.0 (0–100) 
CD8 expression, percent marker area 
n (%) 78 (95.1) 37 (92.5) 24 (96.0) 
 Median (range) 4.23 (0.1–31.8) 5.74 (0.3–29.8) 5.58 (0.3–20.4) 

Abbreviations: CD8, cluster of differentiation 8; CR, complete response; EBV, Epstein–Barr virus; ECOG PS, Eastern Cooperative Oncology Group performance status; NA, not available; PD, progressive disease; PD-L1, programmed death-ligand 1; PR, partial response; SD, stable disease; UNK, unknown.

aPer investigator's choice.

bECOG PS is missing in one patient in the chemotherapy arm.

Data cutoff: October 11, 2018.

Of the 39 patients treated in the chemotherapy arm, 27 patients received monochemotherapy and 12 patients received chemotherapy combinations with two or more drugs. Specific monochemotherapy and chemotherapy combinations are detailed in Supplementary Table S2.

In the chemotherapy arm, 25/40 patients crossed over to receive spartalizumab after centrally confirmed radiologic progression. Patients who crossed over from the chemotherapy arm had similar demographics and baseline characteristics compared with all patients in the chemotherapy arm (Table 1).

All patients in the study had received prior antineoplastic therapies, and radiotherapy was the most common type (84.1% and 92.5% in the spartalizumab and chemotherapy arms, respectively). The majority of patients received ≥ 2 prior systemic therapies in any setting (80.5% in the spartalizumab arm and 77.5% in the chemotherapy arm). The most common prior chemotherapy given to patients was cisplatin in both the spartalizumab arm (84.0%) and the chemotherapy arm (82.5%). Around 48% to 58% were previously treated with carboplatin, fluorouracil, or gemcitabine.

At the last line of prior therapy, the BOR was recorded; complete response was experienced by five patients (6.1%) in the spartalizumab arm compared with three patients (7.5%) in the chemotherapy arm, and partial response was experienced by 18 patients (22.0%) in the spartalizumab arm and 10 patients (25.0%) in the chemotherapy arm. Progressive disease was experienced by 28 patients (34.1%) and 8 (20.0%) in the spartalizumab and the chemotherapy arms, respectively.

Preliminary analysis showed that the median interval between the end date of last prior chemotherapy and the first dose of study treatment was comparable among the spartalizumab, chemotherapy, crossover, and the all-spartalizumab groups (Supplementary Table S3).

Overall, most patients received chemotherapy after the end-of-study treatment across the spartalizumab, chemotherapy, and crossover groups (Supplementary Table S4). Similar results were seen among patients who experienced a BOR of progressive disease.

Exposure

The median duration of exposure to the study treatment was 14.43 weeks (range, 3.1–120.1) in the spartalizumab arm and 19.29 weeks (range, 3.0–77.4) in the chemotherapy arm. Patients in the crossover group had a median duration of 8.86 weeks (range, 0.7–78.3) exposure to spartalizumab. The overall median duration of exposure in the all-spartalizumab group was 12.43 weeks (range, 0.7–120.1). In total, 27 patients (32.9%) and 18 patients (46.2%) were exposed for > 24 weeks in the spartalizumab and chemotherapy arms, respectively. In the all-spartalizumab group, 33 patients (30.8%) were exposed to spartalizumab treatment for > 24 weeks.

Safety

As of the data cutoff, the safety profiles of the spartalizumab arm and the crossover group were similar (Table 2). Therefore, the “all-spartalizumab group” was compared with the chemotherapy arm for the safety data analyses in this section.

Table 2.

Adverse events suspected to be treatment related (≥ 10% incidence of all grades events in any of the patient treatment arms).

Spartalizumab 400 mg every 4 weeks (N = 82)Chemotherapya (N = 39b)Crossover to spartalizumab 400 mg every 4 weeks (N = 25)All-spartalizumab (N = 107)
Preferred term, n (%)All gradesGrade ≥ 3All gradesGrade ≥ 3All gradesGrade ≥ 3All gradesGrade ≥ 3
Total 59 (72.0) 14 (17.1) 34 (87.2) 16 (41.0) 11 (44.0) 4 (16.0) 70 (65.4) 18 (16.8) 
Fatigue 8 (9.8) 13 (33.3) 1 (2.6) 3 (12.0) 1 (4.0) 11 (10.3) 1 (0.9) 
Pruritus 7 (8.5) 2 (5.1) 3 (12.0) 10 (9.3) 
Decreased appetite 6 (7.3) 1 (1.2) 7 (17.9) 3 (12.0) 9 (8.4) 1 (0.9) 
Rash 9 (11.0) 5 (12.8) 9 (8.4) 
AST increased 8 (9.8) 5 (12.8) 8 (7.5) 
Anemia 6 (7.3) 1 (1.2) 16 (41.0) 4 (10.3) 6 (5.6) 1 (0.9) 
ALT increased 5 (6.1) 1 (1.2) 4 (10.3) 5 (4.7) 1 (0.9) 
Nausea 4 (4.9) 8 (20.5) 1 (2.6) 1 (4.0) 5 (4.7) 
Arthralgia 1 (1.2) 2 (5.1) 3 (12.0) 4 (3.7) 
Hypoesthesia 3 (3.7) 4 (10.3) 1 (4.0) 4 (3.7) 
Stomatitis 3 (3.7) 7 (17.9) 1 (2.6) 3 (2.8) 
Diarrhea 1 (1.2) 5 (12.8) 2 (5.1) 1 (4.0) 2 (1.9) 
Weight decreased 2 (2.4) 4 (10.3) 2 (1.9) 
Neuropathy peripheral 1 (1.2) 5 (12.8) 1 (0.9) 
PPE syndrome 6 (15.4) 1 (4.0) 1 (0.9) 
Alopecia 9 (23.1) 
Febrile neutropenia 4 (10.3) 4 (10.3) 
Neutropenia 12 (30.8) 10 (25.6) 
Peripheral sensory neuropathy 4 (10.3) 
Platelet count decreased 7 (17.9) 3 (7.7) 
White blood cell count decreased 8 (20.5) 4 (10.3) 
Spartalizumab 400 mg every 4 weeks (N = 82)Chemotherapya (N = 39b)Crossover to spartalizumab 400 mg every 4 weeks (N = 25)All-spartalizumab (N = 107)
Preferred term, n (%)All gradesGrade ≥ 3All gradesGrade ≥ 3All gradesGrade ≥ 3All gradesGrade ≥ 3
Total 59 (72.0) 14 (17.1) 34 (87.2) 16 (41.0) 11 (44.0) 4 (16.0) 70 (65.4) 18 (16.8) 
Fatigue 8 (9.8) 13 (33.3) 1 (2.6) 3 (12.0) 1 (4.0) 11 (10.3) 1 (0.9) 
Pruritus 7 (8.5) 2 (5.1) 3 (12.0) 10 (9.3) 
Decreased appetite 6 (7.3) 1 (1.2) 7 (17.9) 3 (12.0) 9 (8.4) 1 (0.9) 
Rash 9 (11.0) 5 (12.8) 9 (8.4) 
AST increased 8 (9.8) 5 (12.8) 8 (7.5) 
Anemia 6 (7.3) 1 (1.2) 16 (41.0) 4 (10.3) 6 (5.6) 1 (0.9) 
ALT increased 5 (6.1) 1 (1.2) 4 (10.3) 5 (4.7) 1 (0.9) 
Nausea 4 (4.9) 8 (20.5) 1 (2.6) 1 (4.0) 5 (4.7) 
Arthralgia 1 (1.2) 2 (5.1) 3 (12.0) 4 (3.7) 
Hypoesthesia 3 (3.7) 4 (10.3) 1 (4.0) 4 (3.7) 
Stomatitis 3 (3.7) 7 (17.9) 1 (2.6) 3 (2.8) 
Diarrhea 1 (1.2) 5 (12.8) 2 (5.1) 1 (4.0) 2 (1.9) 
Weight decreased 2 (2.4) 4 (10.3) 2 (1.9) 
Neuropathy peripheral 1 (1.2) 5 (12.8) 1 (0.9) 
PPE syndrome 6 (15.4) 1 (4.0) 1 (0.9) 
Alopecia 9 (23.1) 
Febrile neutropenia 4 (10.3) 4 (10.3) 
Neutropenia 12 (30.8) 10 (25.6) 
Peripheral sensory neuropathy 4 (10.3) 
Platelet count decreased 7 (17.9) 3 (7.7) 
White blood cell count decreased 8 (20.5) 4 (10.3) 

Note: A patient with multiple severity grades for an AE was only counted under the maximum grade.

Abbreviations: AE, adverse event; ALT, alanine aminotransferase; AST, aspartate aminotransferase; PPE, palmar-plantar erythrodysesthesia.

aPer investigator's choice.

bOne patient in the chemotherapy arm did not receive the treatment and was therefore excluded from the safety set.

Overall, 96.3% of patients in the all-spartalizumab group and 94.9% in the chemotherapy arm experienced at least one AE during the study, regardless of relationship to study treatment (Supplementary Table S5). Suspected treatment-related AEs were experienced in 70 patients (65.4%) in the all-spartalizumab group compared with 34 patients (87.2%) in the chemotherapy arm (Table 2). Grade 3/4 AEs suspected to be related to treatment were less common in the all-spartalizumab group compared with the chemotherapy arm (16.8% vs. 41.0%). Serious AEs (SEAs) of all grades, suspected to be treatment related, were reported in 12 patients (11.2%) in the all-spartalizumab group compared with seven patients (17.9%) in the chemotherapy arm. The classification of AEs in the crossover group is shown in Supplementary Fig. S2. AEs of special interest (AESI), including immune-related AEs, are shown in Supplementary Table S6 for patients treated with spartalizumab. Liver enzyme increase was the only grade 3/4 AESI observed in more than one patient (five patients; 4.7%). Grade 3 pneumonitis, observed in one patient in the crossover arm, was the only AESI leading to treatment discontinuation, the patient later died of disease progression.

AEs leading to discontinuation were less frequent in the all-spartalizumab group (1.9% vs. 10.3%) compared with the chemotherapy arm. On-treatment deaths were as follows: five deaths (6.1%) in the spartalizumab arm, three deaths (7.7%) in the chemotherapy arm, and three deaths (12.0%) in the crossover group. No treatment-related deaths occurred.

Of the five deaths reported in the spartalizumab arm, four deaths were due to study indication (NPC; one patient also had a fatal SAE of aspiration pneumonia as a contributing reason for death), and one patient died due to an SAE of sepsis. Of the three deaths in the chemotherapy arm, one death was due to study indication (NPC) and two deaths were due to SAEs of acute respiratory failure and myocardial infarction, respectively. The three deaths reported in patients crossing over to spartalizumab were all due to study indication (NPC).

Efficacy

Treatment with spartalizumab did not improve PFS compared with chemotherapy and the primary endpoint was not met (Fig. 1A).

Figure 1.

Kaplan–Meier plot of PFS (A) and OS (B) in patients treated with spartalizumab versus chemotherapy (RECIST v1.1).

Figure 1.

Kaplan–Meier plot of PFS (A) and OS (B) in patients treated with spartalizumab versus chemotherapy (RECIST v1.1).

Close modal

Based on central review, the median PFS in the spartalizumab arm was 1.9 months (95% CI, 1.8–3.6) versus 6.6 months (95% CI, 3.7–9.3) in the chemotherapy arm. The hazard ratio was 1.36 (95% CI, 0.87–2.12) with a one-sided stratified log-rank test P = 0.915. For the 25 patients who crossed over from chemotherapy to spartalizumab, the median PFS was 1.7 months (95% CI, 1.6–1.9).

Immune-related PFS (irPFS) was also assessed by irRC (central assessment). Median irPFS in the spartalizumab and crossover arms were consistent with median PFS assessed by RECIST v1.1 (1.9 months and 1.7 months, respectively).

The OS was assessed by an intention-to-treat analysis. Consequently, patients who crossed over from chemotherapy to spartalizumab were considered under the chemotherapy arm. The median OS in the spartalizumab arm was 25.2 months [95% CI, 13.1–not estimable (NE)] versus 15.5 months (95% CI, 8.3–21.3) in the chemotherapy arm (P = 0.138; Fig. 1B). In the chemotherapy-treated patients who did not cross over to spartalizumab, the median OS was 17.6 months (95% CI, 4.2–NE).

The ORR in the spartalizumab arm was 17.1% (95% CI, 9.7–27.0; Table 3). In the chemotherapy arm, the ORR was 35.0% (95% CI, 20.6–51.7). In addition, the ORR was 25.9% for patients treated with monochemotherapy and 58.3% for patients who received a combination of two or more chemotherapies.

Table 3.

Best overall response (RECIST v1.1) based on central review.

ResponseSpartalizumab 400 mg every 4 weeks (N = 82)Chemotherapya (N = 40)Crossover to spartalizumab 400 mg every 4 weeks (N = 25)
Best overall response, n (%) 
 CR 1 (1.2) 1 (2.5) 
 PR 13 (15.9) 13 (32.5) 2 (8.0) 
 SD 15 (18.3) 14 (35.0) 4 (16.0) 
 PD 41 (50.0) 9 (22.5) 11 (44.0) 
 Non-CR/non-PD 6 (7.3) 
 UNK 6 (7.3) 3 (7.5) 8 (32.0) 
ORRb, n (%) 14 (17.1) 14 (35.0) 2 (8.0) 
[95% CI] [9.7–27.0] [20.6–51.7] [1.0–26.0] 
DCRc, n (%) 35 (42.7) 28 (70.0) 6 (24.0) 
[95% CI] [31.8–54.1] [53.5–83.4] [9.4–45.1] 
ResponseSpartalizumab 400 mg every 4 weeks (N = 82)Chemotherapya (N = 40)Crossover to spartalizumab 400 mg every 4 weeks (N = 25)
Best overall response, n (%) 
 CR 1 (1.2) 1 (2.5) 
 PR 13 (15.9) 13 (32.5) 2 (8.0) 
 SD 15 (18.3) 14 (35.0) 4 (16.0) 
 PD 41 (50.0) 9 (22.5) 11 (44.0) 
 Non-CR/non-PD 6 (7.3) 
 UNK 6 (7.3) 3 (7.5) 8 (32.0) 
ORRb, n (%) 14 (17.1) 14 (35.0) 2 (8.0) 
[95% CI] [9.7–27.0] [20.6–51.7] [1.0–26.0] 
DCRc, n (%) 35 (42.7) 28 (70.0) 6 (24.0) 
[95% CI] [31.8–54.1] [53.5–83.4] [9.4–45.1] 

Abbreviations: CI, confidence interval; CR, complete response; DCR, disease control rate; ORR, overall response rate; PD, progressive disease; PR, partial response; RECIST, Response Evaluation Criteria In Solid Tumors; SD, stable disease, UNK, unknown.

aPer investigator's choice.

bORR = CR + PR.

cDCR = CR + PR + SD.

The disease control rate (DCR) was 42.7% (95% CI, 31.8–54.1) and 70.0% (95% CI, 53.5–83.4) in the spartalizumab and chemotherapy arms, respectively.

Interestingly, in the spartalizumab arm, the median DOR was 10.2 months (95% CI, 7.4–NE), and in the chemotherapy arm, it was 5.7 months (95% CI, 3.7–7.4). The estimated probability of maintaining the response at 12 months was 45.5% (95% CI, 18.1–69.5) and 9.3% (95% CI, 0.6–33.7) in the spartalizumab and chemotherapy arms, respectively, based on the Kaplan–Meier distribution. Supplementary Fig. S3 shows the best percentage change in the sum of diameters of target lesions from baseline per central review, for the spartalizumab and chemotherapy arms.

Pharmacokinetic data

Following administration of spartalizumab at 400 mg every 4 weeks, the geometric mean [coefficient variation (CV%)] AUCtau and Cmax in cycle 1 were 1,340 (25.8) day•μg/mL and 116 (21.9) μg/mL, respectively (Supplementary Table S7). Predose concentrations were all 0 μg/mL at cycle 1 day 1 (C1D1). During treatment, trough (predose) concentrations (geometric means; geo CV%) were 26.3 (43.1) μg/mL on C2D1, 35.1 (71.7) μg/mL on C3D1, 50.4 (49.4) μg/mL on C4D1, 61.5 (42.6) μg/mL on C5D1, and 57.3 (52.3) μg/mL on C6D1. Overall, the PK data are similar to spartalizumab studies in other tumor types.

Biomarker analysis

At baseline in the spartalizumab arm, seven patients (8.5%) were negative in plasma EBV DNA titer versus no patients in the chemotherapy arm; EBV-negative patients consisted of three (3.7%) Asian patients and four (5.0%) white patients. The median EBV DNA level in plasma was 31,000 copies/mL for patients in the spartalizumab arm compared with 14,000 copies/mL for patients in the chemotherapy arm. The overall weighted median EBV DNA level was 19,000 copies/mL across both arms. A total of 45 patients (55.0%) in the spartalizumab arm and 17 patients (43.0%) in the chemotherapy arm had EBV DNA levels higher than the weighted median.

Exploratory analysis of response rates in patients with EBV DNA levels < the weighted median compared with levels ≥ the weighted median suggested that patients with higher levels of EBV DNA had poorer outcomes. In the spartalizumab arm, an ORR of 22.2% (95% CI, 10.1–39.2) was observed in patients with EBV DNA levels < 19,000 copies/mL and 11.1% (95% CI, 3.7–24.1) in patients with EBV DNA levels ≥ 19,000 copies/mL. In the chemotherapy arm, the ORRs were 42.9% (95% CI, 21.8–66.0) in patients with EBV DNA levels < 19,000 copies/mL and 29.4% (95% CI, 10.3–56.0) in patients with EBV DNA levels ≥ 19,000 copies/mL.

Patient tumor biopsies were analyzed by IHC for baseline expression of PD-L1 and CD8+ lymphocytes. The baseline median percentage of tumor cells positive for PD-L1 expression in the spartalizumab arm was lower (20%; range, 0–100%) compared with the chemotherapy arm (60%; range, 0–100%). The baseline median CD8 expression (percent marker area) was similar between the spartalizumab (4.23%; range, 0.1%–31.8%) and chemotherapy (5.74%; range, 0.3%–29.8%) arms (Table 1). It should be noted that most baseline samples were archival and varied in both age—with some being relatively old—and prior intervening treatments between collection and analysis.

RNA-seq data from baseline tumor samples showed a correlation between the response to spartalizumab treatment (best percentage change from baseline in sum of diameters of target lesions) and baseline expression of preselected markers, TIM-3, LAG-3, and IFNγ signatures; this was not observed in the chemotherapy-treated patients (Fig. 2). The observed correlations were stronger in patients who had biopsies taken within a year of study treatment.

Figure 2.

Gene expression of TIM-3 (A), LAG-3 (B), and IFNγ (C) and response to spartalizumab or chemotherapy treatment

Figure 2.

Gene expression of TIM-3 (A), LAG-3 (B), and IFNγ (C) and response to spartalizumab or chemotherapy treatment

Close modal

Baseline tumor biopsies were also analyzed by RNA-seq for EBV transcripts. Higher levels of EBV transcripts in tumor samples were associated with greater tumor shrinkage in spartalizumab-treated patients with available samples (n = 15), but not in chemotherapy-treated patients with available samples (n = 13; Fig. 3A). It should be noted that the sample sizes for these analyses were small.

Figure 3.

Correlation of (A) EBV transcript number in tumor and treatment responses, (B) circulating baseline levels of MCP-1, and CXCL10 to response.

Figure 3.

Correlation of (A) EBV transcript number in tumor and treatment responses, (B) circulating baseline levels of MCP-1, and CXCL10 to response.

Close modal

Finally, baseline plasma samples were analyzed for circulating cytokine levels. Higher levels of monocyte chemoattractant protein (MCP-1) and C-X-C motif chemokine 10 (CXCL10), also known as interferon gamma-induced protein 10 (IP-10), correlated with poorer responses in spartalizumab-treated patients (Fig. 3B).

Treatment with spartalizumab was generally well tolerated in this phase II study with no unexpected safety concerns. The primary efficacy endpoint of median PFS was not met. However, the median OS and median DOR, as per RECIST v1.1, were both longer in patients who received spartalizumab compared with those who had chemotherapy. The long DOR with spartalizumab treatment compared with chemotherapy suggests a preliminary long-lasting antitumor effect of spartalizumab in NPC. Overall, however, these results are not conclusive as the study was not sufficiently powered to assess significance for these endpoints. It should be noted that there was no evidence of pseudoprogression in this study. Additionally, several immuno-oncology trials have shown a benefit in OS but no difference in PFS, highlighting the fact that assessment of efficacy may differ depending on the selected outcome measure for the primary endpoint (31).

In addition, almost a third of patients in the control arm received combination regimens, many of which included platinum-based agents, which may have contributed to the observed results. In contrast, the phase III KEYNOTE-122 trial, which evaluated pembrolizumab versus chemotherapy in patients with recurrent/metastatic NPC, was limited to single-agent capecitabine, gemcitabine, or docetaxel (32).

The results from this study are consistent with other studies of different anti–PD-1 agents in patients with recurrent/metastatic NPC. The observed ORR of 17.1% (n = 14; 95% CI, 9.7–27.0) and DOR of 10.2 months in the spartalizumab arm is comparable to the ORR of 25.9% (n = 7; 95% CI, 11.1–46.3) and DOR of 17.1 months (per investigator review, 10.7 months per central review) reported in a phase Ib study of the treatment of PD-L1–positive advanced NPC with pembrolizumab (17). Treatment with nivolumab in PD-L1–unselected, recurrent/metastatic NPC reported an ORR of 20.8% (n = 5; 95% CI, 7–42) and DOR not reached in a phase I/II study (33) and an ORR of 20.5% (n = 9; 95% CI, 9.8–35.3) and DOR of 9.3 months in a phase II study (16). In a phase II POLARIS-02 study of patients with recurrent/metastatic NPC, treatment with toripalimab led to an ORR of 23.9% in patients who had failed two prior lines of chemotherapy; in the overall population, the ORR was 20.5% (n = 39; 95% CI, 15.0–27.0), with a DOR of 12.8 months (34).

The OS values were also comparable between these studies; the median OS in the phase II study of nivolumab in NPC was 17.1 months (95% CI, 10.9–NE; ref. 16) and in the phase Ib study of pembrolizumab in NPC was 16.5 months (95% CI, 10.1–NE; ref. 17) compared with 25.2 months (95% CI, 13.1–NE) in the current study. Although the data have not yet been published, the phase III KEYNOTE-122 trial, evaluating pembrolizumab versus chemotherapy in patients with recurrent/metastatic NPC (35), did not meet its primary endpoint of OS. In POLARIS-02, median OS was 17.4 months (95% CI, 11.7–22.9; ref. 34); toripalimab has been granted conditional approval for the treatment of recurrent/metastatic NPC, after failure of at least two lines of prior systemic therapy, by the National Medical Products Administration (NMPA) of China.

The overall PK exposure (Cmax, AUCtau) with spartalizumab, in both cycles 1 and 3, was consistent with the spartalizumab PK observed in a previous study (15). The total systemic exposure (AUCtau) was higher in cycle 3 compared with cycle 1 and the accumulation factor according to the geometric mean was 1.61 (CV% 19.6). Serum concentrations of spartalizumab decreased with a median T1/2 of 20.1 days in cycle 1 and 22.7 days in cycle 3, respectively.

Plasma EBV DNA levels were higher in patients in the spartalizumab arm compared with the chemotherapy arm. Higher concentrations of plasma EBV DNA have previously been linked to poorer outcomes (36), and so it might be expected that these patients would exhibit poorer responses overall than those in the chemotherapy arm. Intriguingly, a preliminary report of the ipilimumab/nivolumab combination in a similar NPC population suggested higher responses in patients with low plasma EBV levels compared with those with high EBV levels (37). We observed that response rates were favorable in patients with plasma EBV levels lower than the weighted median; this trend was seen in both treatment arms. It has been hypothesized that high levels of EBV might favor upregulation of PD-L1 through EBV-driven IFNγ (38); however, we did not see an association between elevated EBV and upregulation of PD-L1.

The median percentage of tumor cells positive for PD-L1 expression was lower in the spartalizumab arm compared with the chemotherapy arm, which could also have contributed to the low response rates seen with spartalizumab. Although PD-L1 has been shown to be a poor prognostic marker in NPC in the context of conventional therapy, tumor biopsies positive for PD-L1 expression have been associated with greater response rates in the context of PD-1–targeted therapies (38, 39). In an NPC study with nivolumab, no statistical correlation between ORR and the biomarkers was reported; however, a descriptive analysis showed that the proportion of patients who responded was higher in patients with PD-L1–positive tumors (>1% expression) than those with PD-L1–negative tumors (16). A similar trend was seen in the POLARIS-02 study, with ORRs of 27.1% and 19.4% in patients with PD-L1–positive and –negative disease, respectively (34).

RNA-seq data from archival tumor samples showed a correlation between the response to spartalizumab and TIM-3, LAG-3, and IFNγ signature gene expression at baseline, in all analyzed tumor samples and more robustly in samples obtained within 12 months before starting treatment (Fig. 2). Although further study is needed, these preliminary data may help to support the rationale of combining anti–TIM-3, anti–LAG-3, and anti–PD-1 agents in NPC. Overall, the identification and validation of potential predictors of response to treatment is limited by the number of available biomarker samples and cohort differences, including variable PD-L1 expression and EBV levels between the two treatment arms; further investigations will be required to identify biomarkers associated with treatment response.

As NPC is a highly chemosensitive disease, the combination of chemotherapy and an anti–PD-1 inhibitor may prove to be a promising therapeutic strategy (18).

In the first-line setting, the use of PD-1/PD-L1 inhibitors in combination with gemcitabine plus cisplatin (GP) in recurrent/metastatic NPC has shown improvements in median PFS. Recent results from the phase III CAPTAIN-1st trial reported a median PFS, as assessed by an independent review committee of 10.8 months in the camrelizumab (a PD-L1 inhibitor) plus GP arm compared with a median PFS of 6.9 months in the placebo plus GP arm (HR = 0.51; 95% CI, 0.37–0.69; one-sided P < 0.0001; ref. 40). Similarly, another phase III study (JUPITER-02) investigating toripalimab in combination with GP (a PD-1 inhibitor) resulted in a median PFS, as assessed by an independent review committee of 11.7 months in the toripalimab plus GP arm versus 8.0 months in the placebo plus GP arm (HR = 0.52; 95% CI, 0.36–0.74); two-sided P = 0.0003; ref. 41). These data support the investigation of immunotherapy as a promising strategy for NPC and highlight the effectiveness of PD-1/PD-L1 inhibition in combination with standard-of-care chemotherapy regimen, in first-line recurrent/metastatic NPC.

Based on our data, PD-L1 expression alone may not be sufficient to determine which patients are most likely to benefit from PD-1–targeted treatment. Further studies using an appropriate selection of the patient population, for example, by PD-L1 and/or IFNγ signature gene expression, should be considered when investigating potential combination therapies with anti–PD-1 agents in NPC.

C. Even reports personal fees from Bristol Myers Squibb, Innate Pharma, Merck Sharp & Dohme, and Merck Serono outside the submitted work. R. K-C. Ngan reports personal fees from Novartis, AstraZeneca, Sanofi, Zai Lab, Pfizer, Eisai, Merck Sharp & Dohme, Eli Lilly, and Roche outside the submitted work. L. Zhang reports grants from Beigene and Hengrui Pharmaceuticals during the conduct of the study. B.B.Y. Ma reports other support from Bristol Myers Squibb and Merck Sharp & Dohme, personal fees and other support from Viracta; personal fees from Y-Biologics, and grants from Novartis during the conduct of the study. V.H.F. Lee reports personal fees from Pfizer and AstraZeneca, grants from Pfizer, AstraZeneca, Merck Sharp & Dohme, Eli Lilly, Amgen, and Boston Scientific outside the submitted work. Z. Li reports other support from NYU Langone Health during the conduct of the study. A.I. Spira reports grants from Novartis during the conduct of the study and grants and personal fees from American Society of Clinical Oncology outside the submitted work. J. Guigay reports personal fees from AstraZeneca, Innate Pharma, Merck Sharp & Dohme, and Roche and grants and personal fees from Bristol Myers Squibb and Merck outside the submitted work. Y. Yao reports currently working for EOC Pharma in China. R. Séchaud reports other support from Novartis outside the submitted work. L. Manenti reports other support from Novartis during the conduct of the study. D. W-T. Lim reports grants from Bristol Myers Squibb, personal fees and nonfinancial support from Merck and Boehringer-Ingelheim, and personal fees from Roche, Pfizer, Taiho, and AstraZeneca outside the submitted work. No disclosures were reported by the other authors.

C. Even: Resources, supervision, investigation, writing–original draft, writing–review and editing. H-M. Wang: Resources, Investigation, writing–review and editing. S-H. Li: Resources, writing–review and editing. R.K-C. Ngan: Resources, investigation, writing–review and editing. A. Dechaphunkul: Resources, writing–review and editing. L. Zhang: Conceptualization, resources, investigation, writing–review and editing. C-J. Yen: Resources, writing–review and editing. P.C. Chan: Resources, validation, investigation, writing–original draft, writing–review and editing. S. Chakrabandhu: Resources, writing–review and editing. B.B.Y. Ma: Resources, validation, investigation, visualization, writing–original draft, writing–review and editing. S. Tanasanvimon: Resources, data curation, investigation, writing–review and editing. V.H.F. Lee: Resources, formal analysis, validation, investigation, writing–original draft, writing–review and editing. P-J. Lou: Resources, writing–review and editing. Z. Li: Investigation, writing–review and editing. A.I. Spira: Resources, data curation, supervision, investigation, project administration, writing–review and editing. A. Sukari: Resources, data curation, investigation, writing–original draft, project administration, writing–review and editing. J. Guigay: Resources, writing–review and editing. S. McCune: Conceptualization, supervision, validation, investigation, methodology. J. Gonzalez-Maffe: Data curation, formal analysis, visualization, methodology, writing–original draft, writing–review and editing. S. Szpakowski: Data curation, software, formal analysis, supervision, validation, visualization, methodology, writing–original draft, writing–review and editing. Y. Yao: Conceptualization, investigation, writing–review and editing. H. Liang: Resources, supervision, project administration, writing–review and editing. J. Mataraza: Conceptualization, resources, data curation, supervision, validation, visualization, methodology, writing–original draft, writing–review and editing. R. Séchaud: Resources, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. L. Manenti: Conceptualization, supervision, writing–original draft, writing–review and editing. D.W-T. Lim: Resources, investigation, visualization, writing–review and editing.

This work was supported by Novartis Pharmaceuticals Corporation. The authors would like to thank the patients who participated in the trial and their families. The authors would also like to thank the physicians, nurses, research coordinators, and other staff at each site who assisted with the study. Hongqian Wu is thanked for her role in data analysis. Editorial assistance was provided by Manoshi Nath, MSc, and was funded by Novartis Pharmaceuticals Corporation.

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.

1.
Chou
J
,
Lin
YC
,
Kim
J
,
You
L
,
Xu
Z
,
He
B
, et al
Nasopharyngeal carcinoma–review of the molecular mechanisms of tumorigenesis
.
Head Neck
2008
;
30
:
946
63
.
2.
Amin
MB
,
Greene
FL
,
Edge
SB
,
Compton
CC
,
Gershenwald
JE
,
Brookland
RK
, et al
The eighth edition AJCC cancer staging manual: continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging
.
CA Cancer J Clin
2017
;
67
:
93
9
.
3.
Cobanoglu
BAS
. 
Nasopharyngeal cancer: an update on diagnosis and treatment
.
Mucosa
2019
;
2
:
1
5
.
4.
Ren
YQ
,
Yuan
Y
,
Ye
J
,
Tian
Y
,
Wen
B
,
Zhang
W
, et al
Evaluation of 7th edition of AJCC staging system for nasopharyngeal carcinoma
.
J Cancer
2017
;
9
:
1665
72
.
5.
Agulnik
M
,
Siu
LL
. 
State-of-the-art management of nasopharyngeal carcinoma: current and future directions
.
Br J Cancer
2005
;
92
:
799
806
.
6.
Al-Sarraf
M
,
LeBlanc
M
,
Giri
PG
,
Fu
KK
,
Cooper
J
,
Vuong
T
, et al
Chemoradiotherapy versus radiotherapy in patients with advanced nasopharyngeal cancer: phase III randomized Intergroup study 0099
.
J Clin Oncol
1998
;
16
:
1310
7
.
7.
Wee
J
,
Tan
EH
,
Tai
BC
,
Wong
HB
,
Leong
SS
,
Tan
T
, et al
Randomized trial of radiotherapy versus concurrent chemoradiotherapy followed by adjuvant chemotherapy in patients with American Joint Committee on Cancer/International Union against cancer stage III and IV nasopharyngeal cancer of the endemic variety
.
J Clin Oncol
2005
;
23
:
6730
8
.
8.
Chan
AT
,
Leung
SF
,
Ngan
RK
,
Teo
PM
,
Lau
WH
,
Kwan
WH
, et al
Overall survival after concurrent cisplatin-radiotherapy compared with radiotherapy alone in locoregionally advanced nasopharyngeal carcinoma
.
J Natl Cancer Inst
2005
;
97
:
536
9
.
9.
Zhang
L
,
Zhao
C
,
Peng
PJ
,
Lu
LX
,
Huang
PY
,
Han
F
, et al
Phase III study comparing standard radiotherapy with or without weekly oxaliplatin in treatment of locoregionally advanced nasopharyngeal carcinoma: preliminary results
.
J Clin Oncol
2005
;
23
:
8461
8
.
10.
Lee
AW
,
Tung
SY
,
Chan
AT
,
Chappell
R
,
Fu
YT
,
Lu
TX
, et al
Preliminary results of a randomized study (NPC-9902 Trial) on therapeutic gain by concurrent chemotherapy and/or accelerated fractionation for locally advanced nasopharyngeal carcinoma
.
Int J Radiat Oncol Biol Phys
2006
;
66
:
142
51
.
11.
Xu
T
,
Tang
J
,
Gu
M
,
Liu
L
,
Wei
W
,
Yang
H
. 
Recurrent nasopharyngeal carcinoma: a clinical dilemma and challenge
.
Curr Oncol
2013
;
20
:
e406
19
.
12.
Zhang
L
,
Huang
Y
,
Hong
S
,
Yang
Y
,
Yu
G
,
Jia
J
, et al
Gemcitabine plus cisplatin versus fluorouracil plus cisplatin in recurrent or metastatic nasopharyngeal carcinoma: a multicentre, randomised, open-label, phase 3 trial
.
Lancet (London, England)
2016
;
388
:
1883
92
.
13.
Beldjilali
Y
,
Yamouni
M
,
Benhadji
KA
,
Khellafi
H
,
Abdelaoui
A
,
Betkaoui
F
, et al
Combination of gemcitabine and carboplatin chemotherapy for recurrent nasopharyngeal carcinoma
.
J Clin Oncol
2010
;
28
:
e16006
e
.
14.
Chen
C
,
Wang
FH
,
Wang
ZQ
,
An
X
,
Luo
HY
,
Zhang
L
, et al
Salvage gemcitabine-vinorelbine chemotherapy in patients with metastatic nasopharyngeal carcinoma pretreated with platinum-based chemotherapy
.
Oral Oncol
2012
;
48
:
1146
51
.
15.
Naing
A
,
Gelderblom
H
,
Gainor
JF
,
Forde
PM
,
Butler
M
,
Lin
CC
, et al
A first-in-human phase I study of the anti-PD-1 antibody PDR001 in patients with advanced solid tumors
.
J Clin Oncol
2016
;
34
:
3060
.
16.
Ma
BBY
,
Lim
WT
,
Goh
BC
,
Hui
EP
,
Lo
KW
,
Pettinger
A
, et al
Antitumor activity of nivolumab in recurrent and metastatic nasopharyngeal carcinoma: an international, multicenter study of the mayo clinic phase 2 consortium (NCI-9742)
.
J Clin Oncol
2018
;
36
:
1412
8
.
17.
Hsu
C
,
Lee
SH
,
Ejadi
S
,
Even
C
,
Cohen
RB
,
Tourneau
LC
, et al
Safety and antitumor activity of pembrolizumab in patients with programmed death-ligand 1-positive nasopharyngeal carcinoma: results of the KEYNOTE-028 study
.
J Clin Oncol
2017
;
35
:
4050
6
.
18.
Fang
W
,
Yang
Y
,
Ma
Y
,
Hong
S
,
Lin
L
,
He
X
, et al
Camrelizumab (SHR-1210) alone or in combination with gemcitabine plus cisplatin for nasopharyngeal carcinoma: results from two single-arm, phase 1 trials
.
Lancet Oncol
2018
;
19
:
1338
50
.
19.
Foo
KF
,
Tan
EH
,
Leong
SS
,
Wee
JT
,
Tan
T
,
Fong
KW
, et al
Gemcitabine in metastatic nasopharyngeal carcinoma of the undifferentiated type
.
Ann Oncol
2002
;
13
:
150
6
.
20.
Zhang
L
,
Zhang
Y
,
Huang
PY
,
Xu
F
,
Peng
PJ
,
Guan
ZZ
. 
Phase II clinical study of gemcitabine in the treatment of patients with advanced nasopharyngeal carcinoma after the failure of platinum-based chemotherapy
.
Cancer Chemother Pharmacol
2008
;
61
:
33
8
.
21.
Ngeow
J
,
Lim
WT
,
Leong
SS
,
Ang
MK
,
Toh
CK
,
Gao
F
, et al
Docetaxel is effective in heavily pretreated patients with disseminated nasopharyngeal carcinoma
.
Ann Oncol
2011
;
22
:
718
22
.
22.
Peng
PJ
,
Ou
XQ
,
Chen
ZB
,
Liao
H
,
Peng
YL
,
Wang
SY
, et al
Multicenter phase II study of capecitabine combined with nedaplatin for recurrent and metastatic nasopharyngeal carcinoma patients after failure of cisplatin-based chemotherapy
.
Cancer Chemother Pharmacol
2013
;
72
:
323
8
.
23.
Yau
TK
,
Shum
T
,
Lee
AW
,
Yeung
MW
,
Ng
WT
,
Chan
L
. 
A phase II study of pemetrexed combined with cisplatin in patients with recurrent or metastatic nanopharyngeal carcinoma
.
Oral Oncol
2012
;
48
:
441
4
.
24.
Chua
DT
,
Kwong
DL
,
Sham
JS
,
Au
GK
,
Choy
D
. 
A phase II study of ifosfamide, 5-fluorouracil and leucovorin in patients with recurrent nasopharyngeal carcinoma previously treated with platinum chemotherapy
.
Eur J Cancer
2000
;
36
:
736
41
.
25.
Ayers
M
,
Lunceford
J
,
Nebozhyn
M
,
Murphy
E
,
Loboda
A
,
Kaufman
DR
, et al
IFN-gamma-related mRNA profile predicts clinical response to PD-1 blockade
.
J Clin Invest
2017
;
127
:
2930
40
.
26.
Middleton
G
,
Yang
Y
,
Campbell
CD
,
André
T
,
Atreya
CE
,
Schellens
JHM
, et al
BRAF-mutant transcriptional subtypes predict outcome of combined BRAF, MEK, and EGFR Blockade with dabrafenib, trametinib, and panitumumab in patients with colorectal cancer
.
Clin Cancer Res
2020
;
26
:
2466
76
.
27.
Dobin
A
,
Davis
CA
,
Schlesinger
F
,
Drenkow
J
,
Zaleski
C
,
Jha
S
, et al
STAR: ultrafast universal RNA-seq aligner
.
Bioinformatics
2013
;
29
:
15
21
.
28.
Anders
S
,
Pyl
PT
,
Huber
W
. 
HTSeq–a Python framework to work with high-throughput sequencing data
.
Bioinformatics
2015
;
31
:
166
9
.
29.
Robinson
MD
,
McCarthy
DJ
,
Smyth
GK
. 
edgeR: a Bioconductor package for differential expression analysis of digital gene expression data
.
Bioinformatics
2010
;
26
:
139
40
.
30.
Brister
JR
,
Ako-Adjei
D
,
Bao
Y
,
Blinkova
O
. 
NCBI viral genomes resource
.
Nucleic Acids Res
2015
;
43
:
D571
7
.
31.
Anagnostou
V
,
Yarchoan
M
,
Hansen
AR
,
Wang
H
,
Verde
F
,
Sharon
E
, et al
Immuno-oncology trial endpoints: capturing clinically meaningful activity
.
Clin Cancer Res
2017
;
23
:
4959
69
.
32.
Ng
QS
,
Spreafico
A
,
Lee
V
,
Ngan
RKC
,
To
KF
,
Ahn
M-J
, et al
KEYNOTE-122: phase 2 study of pembrolizumab versus standard-of-care chemotherapy in platinum-pretreated, recurrent or metastatic nasopharyngeal carcinoma
.
Ann Oncol
2016
;
27
:
48TiP
.
33.
Delord
J-P
,
Hollebecque
A
,
Boer
JPD
,
Greve
JD
,
Machiels
J-PH
,
Leidner
RS
, et al
An open-label, multicohort, phase I/II study to evaluate nivolumab in patients with virus-associated tumors (CheckMate 358): efficacy and safety in recurrent or metastatic (R/M) nasopharyngeal carcinoma (NPC)
.
J Clin Oncol
2017
;
35
:
6025
.
34.
Wang
FH
,
Wei
XL
,
Feng
J
,
Li
Q
,
Xu
N
,
Hu
XC
, et al
Efficacy, safety, and correlative biomarkers of toripalimab in previously treated recurrent or metastatic nasopharyngeal carcinoma: a phase ii clinical trial (POLARIS-02)
.
J Clin Oncol
2021
;
39
:
704
12
.
35.
BioSpace
. 
Merck announces fourth-quarter and full-year 2020 financial results
. <https://www.biospace.com/article/releases/merck-announces-fourth-quarter-and-full-year-2020-financial-results>.
Accessed 2021 May 24
.
36.
Lin
JC
,
Wang
WY
,
Chen
KY
,
Wei
YH
,
Liang
WM
,
Jan
JS
, et al
Quantification of plasma Epstein-Barr virus DNA in patients with advanced nasopharyngeal carcinoma
.
N Engl J Med
2004
;
350
:
2461
70
.
37.
Darren
W-TL
,
Quan
S Ng
,
Ruey-Long
H
,
Daniel
TS
,
Eng-Huat
T
,
Goh
B-C
, et al
AACR 2020
. 
Poster presentation
.
Abstract CT203
.
38.
Fang
W
,
Zhang
J
,
Hong
S
,
Zhan
J
,
Chen
N
,
Qin
T
, et al
EBV-driven LMP1 and IFN-γ up-regulate PD-L1 in nasopharyngeal carcinoma: Implications for oncotargeted therapy
.
Oncotarget
2014
;
5
:
12189
202
.
39.
Khunger
M
,
Hernandez
AV
,
Pasupuleti
V
,
Rakshit
S
,
Pennell
NA
,
Stevenson
J
, et al
Programmed cell death 1 (PD-1) ligand (PD-L1) expression in solid tumors as a predictive biomarker of benefit from PD-1/PD-L1 axis inhibitors: a systematic review and meta-analysis
.
JCO Precis Oncol
2017
;
1
:
1
15
.
40.
Li
Z
,
Yunpeng
Y
,
Song
Q
,
Jin-Gao
L
,
Chaosu
H
,
Mingjun
X
, et al
ASCO 2021
. 
Oral presentation
.
Abstract 6000
.
41.
Rui-hua
X
,
Hai-Qiang
M
,
Qiu-Yan
C
,
Dongping
C
,
Chaosu
H
,
Kunyu
Y
, et al
ASCO 2021
. 
Oral presentation
.
Abstract LBA2
.