Efficacy and safety of immune checkpoint inhibitors in recurrent or metastatic head and neck squamous cell carcinoma: A systematic review and meta‐analysis of randomized clinical trials

Abstract Background Immune checkpoint inhibitors (ICIs) showed antitumor activity for recurrent or metastatic head and neck squamous cell carcinoma (R/M HNSCC). However, the results from different studies were controversial. Methods Online databases were searched for randomized clinical trials (RCTs) evaluating ICIs for R/M HNSCC. The characteristics of the studies and the results of overall survival (OS), progression‐free survival (PFS), objective response rate (ORR), treatment‐related adverse events (TRAEs) were extracted. Results A total of 4936 patients from eight studies were included. Anti‐PD1/PDL1 monotherapy significantly improved OS in total population (hazard ratio, HR, 0.87, 95% CI, 0.79–0.95, p = 0.003) and PD‐L1 high expression patients (HR, 0.71, 95% CI, 0.55–0.90, p = 0.006) with significant lower incidence of any grade TRAEs (odds ratio, OR, 0.16, 95% CI, 0.07–0.37, p < 0.00001) and Grades 3–5 TRAEs (OR, 0.18, 95% CI, 0.10–0.33, p < 0.0001) compared with standard of care (SOC); however, the pooled results of PFS and ORR were not significant different. PD1/PDL1 inhibitors plus CTLA4 inhibitors did not improve OS, PFS, ORR compared with SOC or ICIs monotherapy; however, the incidence of Grades 3–5 TRAEs was significant higher compared with ICIs monotherapy (OR, 1.80, 95% CI, 1.34–2.41, p = 0.0001). Conclusions Anti‐PD1/PDL1 monotherapy could improve OS for R/M HNSCC with significant lower incidence of TRAEs compared with SOC. PD1/PDL1 inhibitors plus CTLA4 inhibitors showed no more benefit compared with both SOC and ICIs monotherapy, but the incidence of Grades 3–5 TRAEs was significant higher compared with ICIs monotherapy.


| INTRODUCTION
Head and neck squamous cell carcinoma (HNSCC) consist of tumors arising from oral cavity, pharynx, larynx, and paranasal sinuses.There were more than 743,000 new cases of HNSCC in the world annually, and it caused about 364,000 deaths during the same time period. 1 Recurrent or metastatic (R/M) HNSCC is one of the major causes of mortality.The prognosis of R/M HNSCC is poor, with a median survival time of less than 1 year. 2 The initial first line systemic treatment for R/M HNSCC was cisplatin based chemotherapy (cisplatin/5-FU). 3In 2008, a phase III randomized trial (EXTREME) found that the addition of cetuximab to the cisplatin/5-FU regimen improved response rate and median survival. 4Then the EXTREME regimen became the preferred first line regimen for the R/M HNSCC in the following decade.
Immune checkpoint inhibitors (ICIs) have been proved to be beneficial in varieties of tumors in the last decade. 5or R/M HNSCC, nivolumab and pembrolizumab were firstly evaluated in the platinum-refractory setting, and the overall survival (OS) were both prolonged. 6,7Subsequent study evaluating ICIs in the first line setting suggested that pembrolizumab plus chemotherapy or pembrolizumab monotherapy in the PD-L1 high expression patients also prolonged OS compared with EXTREME regimen, 8 while another study with nivolumab plus ipilimumab had negative result. 9Some other ICIs such as durvalumab, tremelimumab were also evaluated. 10,11There are still growing number of related researches, but the best strategy for ICIs in R/M HNSCC patients is still indeterminate.
We conducted a meta-analysis of all RCT studies, and comprehensively evaluated the efficacy and safety of ICIs in the treatment of R/M HNSCC.

| Data sources and search strategy
Relevant studies were searched from PubMed, Embase and Cochrane library up to June 31, 2023.We used the Medical Subject Headings (MeSH) terms and their entry terms as follows: "immune checkpoint inhibitors OR ipilimumab OR nivolumab OR PD-L1 Inhibitor OR programmed death-ligand 1 inhibitor OR CTLA-4 inhibitor OR cytotoxic T-lymphocyte-associated protein 4 inhibitor OR PD-1 inhibitor OR programmed cell death protein 1 inhibitor OR pembrolizumab OR cemiplimab OR camrelizumab OR atezolizumab OR durvalumab OR avelumab OR tremelimumab" AND "squamous cell carcinoma of head and neck OR mouth neoplasms OR laryngeal neoplasms OR oropharyngeal neoplasms OR hypopharyngeal neoplasms OR HNSCC OR SCCHN".

| Inclusion and exclusion criteria
This meta-analysis followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) statement. 12

| Inclusion criteria
The inclusion criteria were as follows: (a) study type was RCT; (b) patients with pathologically confirmed R/M squamous cell carcinoma of the oropharynx, oral cavity, hypopharynx, larynx; (c) there was at least one group that was treated with ICIs monotherapy or combination; (d) there was at least one outcome of OS, progression-free survival (PFS), objective response rate (ORR), treatmentrelated adverse events (TRAEs) was reported.

| Exclusion criteria
The exclusion criteria were as follows: (a) study type was non-RCT; (b) patients with squamous cell carcinoma of nasal cavity, paranasal sinuses, nasopharynx, cutaneous; (c) the intervention group was treated with ICIs plus radiotherapy; (d) the study data could not be extracted directly or indirectly; (e) the study was not published in English; (f) conference abstracts, case reports, comments, reviews, animal studies, and mechanistic studies.

| Quality assessment and data extraction
The risk of bias for each study was assessed according to the Cochrane Collaboration's tool by two investigators (Shoutao Dang and Xinyu Li).The data were extracted by the same two investigators.The following information were included: study name, first author, publication year, trial phase, study design, sample size, key inclusion criteria, interventions and control group, outcomes of OS, PFS, ORR, TRAEs.The disagreements were discussed and consensus.

| Statistical analysis
Statistical analysis was performed by Review Manager version 5.4 and Stata 17.The generic inverse variance was selected for the PFS and OS hazard ratio (HR) data, and dichotomous was used for ORR and TRAEs data.The heterogeneity of the studies was estimated using the I 2 statistics, and random-effect model was used for I 2 > 50%; otherwise, the fixed-effect model was used.Statistical significance was set at p < 0.05.

| Study and data selection
The study selection process is summarized in Figure 1.A total of 4556 studies were retrieved from the primary search strategy, of which 3260 were removed when the study type was restricted to clinical trials.And then 73 studies were excluded for duplicate records.A further 1145 studies were excluded due to conference abstracts, case reports, comments, reviews, animal studies, not published in English, and mechanistic studies.Additionally, 78 studies were assessed for eligibility, and 69 studies were excluded (non-RCT = 25, locoregionally advanced HNC = 23, concurrent with radiotherapy = 5, lack of study results = 8, identical published study = 6, others = 1).Ten studies were included in the review.However, two studies were excluded for metaanalysis because it could not be combined analysis with the others. 13,14Ultimately, eight studies with 4936 patients were enrolled in this meta-analysis.

| Study characteristics and quality assessment
The characteristics of the included studies are shown in Table 1.Five studies [6][7][8]10,11 compared anti-PD1/PDL1 monotherapy with standard of care (SOC) which comprised EXTREME regimen in the first line setting and single-agent methotrexate, docetaxel or cetuximab in the second line setting. Threestudies [9][10][11] compared anti-PD1/ PDL1 plus anti-CTLA4 with SOC.Two studies compared anti-PD1/PDL1 plus anti-CTLA4 therapy with anti-PD1/ PDL1 monotherapy, 15,16 and the CheckMate 714 study 15 included two different cohorts of participants as platinumrefractory and platinum-eligible patients.Two studies 10,11 had both monotherapy group and combination group, but these two groups were not compared directly.All the eight studies reported the incidence of TRAEs, ORR, OS, and PFS curve; however, some OS and PFS HR outcomes were not displayed.The Jayne F Tierney method was used to calculate the HR from the survival curve.17 All the eight studies were of high quality and mostly at low risk of bias, while six studies had unclear risk of allocation concealment (Figure S1).

| Sensitivity analysis and publication bias
We conducted sensitivity analysis for the OS results in PD-L1 high expression patients treated with PD1/PDL1 inhibitors alone.Stable result was observed by removing each trial in turn (Figure S4).Publication biases were evaluated for OS in the patients treated with ICIs monotherapy by Egger's test (total population, p = 0.414; PD-L1 high expression patients, p = 0.587) and Begg's test (total population, p = 0.806; PD-L1 high expression patients, p = 0.806).

| DISCUSSION
R/M HNSCC has poor prognosis, and the systemic treatment options are limited.ICIs had better efficacy and safety than conventional chemotherapy or targeted therapy for R/M HNSCC; however, the appropriate strategies for individual patients are still indefinite. 18Our meta-analysis indicated that ICIs monotherapy could improve the OS in total population and PD-L1 high expression patients with significant lower incidence of TRAEs compared with SOC.
To our knowledge, this is the first meta-analysis that has pooled results of ICIs monotherapy versus EXTREME in the first line setting, anti-PDL1 monotherapy versus SOC, PD1/PDL1 inhibitors plus CTLA4 inhibitors versus ICIs monotherapy.However, ICIs monotherapy did not significantly improve OS in the first line setting or for the anti-PDL1 drugs compared with SOC.Furthermore, ICIs combination therapy did not improve OS, PFS, ORR compared with ICIs monotherapy, but with significant higher incidence of Grades 3-5 TRAEs.
Immune checkpoint proteins such as CTLA4 and PD1 play an important role in suppressing T-cell activation and limiting the immune response which make tumor cells "escape" the immune surveillance. 19,20The ICIs can "release the brake" on the immune system and promoting anti-tumor response. 21This pooled analysis of five studies suggested that ICIs monotherapy could improve the OS for the total population of R/M HNSCC.This outcome consistent with a former meta-analysis without the recent KESTREL study. 22We suggested that for the total population in R/M HNSCC, ICIs alone might prolong the OS compared with SOC.However, the improvement of OS might be restricted to the second line setting or the anti-PD1 drugs used.Furthermore our meta-analysis also found no superior and heterogeneous PFS and even worse ORR in the first line setting.Biomarkers which could predict for the ICIs were demanded to select suitable patients.
The expression of PD-L1 was found to be associated with OS for R/M HNSCC in some studies; however, the results were still controversial. 23We pooled analysis for the PD-L1 high expression patients.We found that ICIs alone also reduced the risk of death compared with SOC in PD-L1 high expression patients.Subgroup analysis was further carried out due to the heterogeneity, and we found that the improvement of OS might also be restricted to the second line setting or the anti-PD1 drugs used.The different results from anti-PD1 and anti-PDL1 drugs might be one of the reasons for the heterogeneity.And further studies are needed to estimate PD-L1 in R/M HNSCC.Additionally, the evaluation methods for the expression of PD-L1 and the cut-off value varied in different studies.This might be another reason for the heterogeneity.Therefore although PD-L1 expression is a rational biomarker for R/M HNSCC, it still has some limitations.
ICIs were considered low toxicity; however, with the increasing ICIs applied, the number of immune related adverse events (irAEs) also raised. 24All of the included studies reported significant lower incidence of any grade and Grades 3-5 TRAEs compared with SOC, and also for our pooled results.Nonetheless, caution should always be made because the irAEs might be delayed-onset, permanent and even fatal.
There were also studies focused on the ICIs combine therapy.A former meta-analysis found this combination showed no more benefit compared with SOC. 25 Our pooled analysis found similar results that the combination regimen did not improve OS, PFS, ORR compared with SOC or ICIs monotherapy; however, the incidence of Grades 3-5 TRAEs was significantly higher in the ICIs combination group compared with ICIs monotherapy although it had better safety than SOC.Our pooled results suggested that ICIs combine therapy should be considered carefully for R/M HNSCC.
There are some limitations in this meta-analysis.Firstly, only eight RCT studies were included, and the studies for pooled analysis in each treatment strategies were even less.Secondly, these studies involved various types of drugs and different eligible populations which might cause heterogeneity in meta-analysis.Lastly, there was only one RCT evaluate ICIs plus chemotherapy, and none was done with ICIs plus targeted therapy although these treatment strategies were often used in the real world.High quality studies focus on this issue are urgently required.

| CONCLUSIONS
Anti-PD1/PDL1 monotherapy could improve the OS in total population and PD-L1 high expression patients with significant lower incidence of TRAEs compared with SOC.PD1/PDL1 inhibitors plus CTLA4 inhibitors did not improve OS, PFS, ORR compared with SOC or ICIs monotherapy; however, the incidence of Grades 3-5 TRAEs was significant higher compared with ICIs monotherapy.

1
Flow diagram of the screening and selection process.

F I G U R E 2
Forest plots of OS for total population in ICIs monotherapy VS SOC (A), ICIs combination VS SOC (B), ICIs combination VS ICIs monotherapy (C); for PD-L1 high expression patients treat with ICIs monotherapy (D).

F I G U R E 3
Subgroup analysis of OS for total population treated with ICIs monotherapy by therapy lines (A) and ICIs drugs used (B).F I G U R E 4 Forest plots of PFS for total population in ICIs monotherapy VS SOC (A); ICIs combination VS SOC (B) and ICIs combination VS ICIs monotherapy (C).F I G U R E 5 Forest plots of ORR for total population in ICIs monotherapy VS SOC (A); ICIs combination VS SOC (B) and ICIs combination VS ICIs monotherapy (C).
The characteristic of the eligible studies in the meta-analysis.