Programmed Cell Death Protein-1 Inhibitors Versus Programmed Death-Ligand 1 Inhibitors in Addition to Chemotherapy for the First-Line Treatment of Advanced NSCLC: A Systematic Review and Meta-Analysis.

INTRODUCTION: The addition of programmed cell death protein-1 (PD-1) or programmed death-ligand 1 (PD-L1) inhibitors to first-line chemotherapy (CT) improved the outcomes of advanced NSCLC. Nonetheless, no direct comparison exists between these combination treatments.
METHODS: We performed a meta-analysis of randomized clinical trials to evaluate and compare the efficacy and safety of PD-(L)1 inhibitors in combination with first-line CT for advanced NSCLC.
RESULTS: A total of eight randomized clinical trials were included. The addition of a PD-(L)1 inhibitor to CT improved progression-free survival, overall survival, and objective response rate compared with CT alone. The risk of grade greater than or equal to 3 treatment-related adverse events was slightly higher with the addition of a PD-(L)1 inhibitor to CT as compared with CT alone. A subgroup analysis according to the targeted receptor (PD-1 versus PD-L1) revealed that the addition of a PD-1 inhibitor to CT led to better objective response rate (p = 0.0001), progression-free survival (p = 0.006), and overall survival (p = 0.002) compared with that of a PD-L1 inhibitor. The risk of grade greater than or equal to 3 treatment-related adverse events was significantly increased with the addition of a PD-L1 inhibitor to CT, but not with the addition of a PD-1 inhibitor. A direct comparison using the meta-regression analysis confirmed the statistical significance of all previous findings.
CONCLUSIONS: On the basis of this meta-analysis, the addition of a PD-1 inhibitor to first-line CT revealed statistically significant better outcomes and less additional toxicity compared with that of a PD-L1 inhibitor, as compared with CT alone, in advanced NSCLC, regardless of PD-L1 status.

Introduction

Lung cancer represents the first cause of cancer-related death worldwide. NSCLC accounts for at least 85% of histologic diagnosis and mainly occurs at an advanced stage, leading to poor 5-year survival expectancy., With the advent of immune checkpoint inhibitors (ICIs) targeting the programmed cell death protein-1 (PD-1)-programmed death-ligand 1 (PD-L1) axis, the outcomes of patients with nononcogene-addicted advanced NSCLC have drastically changed. Pembrolizumab demonstrated its superiority to standard platinum-based chemotherapy (CT) as upfront treatment for patients with NSCLC with PD-L1 expression greater than or equal to 50%., In addition, cohort G of the phase 1/2 KEYNOTE 021 study opened the way to the combination of pembrolizumab and carboplatin-pemetrexed regimen as first-line treatment for nonsquamous NSCLC, revealing the improvement of both objective response rate (ORR) and progression-free survival (PFS) regardless of PD-L1 expression., Thereafter, several phase 3 trials confirmed these results for both squamous and nonsquamous NSCLCs treated with CT plus a PD-1 inhibitor, also highlighting a revolutionary benefit in terms of overall survival (OS).8, 9, 10, 11, 12 Similarly, the addition of atezolizumab, a PD-L1 inhibitor, to CT with or without the antiangiogenic drug bevacizumab outperformed standard CT doublets for nonsquamous NSCLCs.13, 14, 15, 16 Nevertheless, the absence of a head-to-head comparison between PD-(L)1 inhibitors as single-agent therapy and their combination with chemotherapeutic regimens, or between different ICI-CT combinations, does not allow evidence-based treatment decisions.

Herein, we performed a meta-analysis to weigh the efficacy and safety of randomized clinical trials (RCTs) comparing standard platinum-based regimens in combination with an anti–PD-(L)1 inhibitor or placebo and to explore whether the combination of a PD-1 inhibitor to CT outperforms that of a PD-L1 inhibitor to CT.

Materials and Methods

Search Strategies

RCTs reporting OS, PFS, ORR, and safety data published before February 1, 2021, were searched through the online databases MEDLINE (PubMed), EMBASE, and Cochrane Database of Systematic Reviews and Central Register of Controlled Trials (Wiley). Keywords used for the research were the following: “immunotherapy”, “pembrolizumab”, “atezolizumab”, “sintilimab”, “durvalumab”, “avelumab”, “nivolumab”, “chemotherapy”, “platinum based”, “cisplatin”, “carboplatin”, “pemetrexed”, “paclitaxel”, “nab-paclitaxel”, “non-small cell lung cancer”, “NSCLC”, “first-line”, “upfront”, “untreated”, “metastatic”, “stage IV”, and “advanced.” Inclusion criteria for study selection were as follows: (1) patients with previously untreated advanced NSCLC; (2) treatment with the combination of either a PD-1 or PD-L1 inhibitor and first-line CT-based treatment, with at least one control arm; and (3) availability of OS and PFS data regardless of PD-L1 expression. Exclusion criteria were as follows: (1) nonrandomized controlled trials; (2) absence of hazard ratio (HR) for efficacy outcomes (OS or PFS) independently from PD-L1 expression; and (3) evaluation of immunotherapy doublets (e.g., PD-1 or PD-L1 inhibitor plus anti–CTLA-4 agent).

Only articles published in peer-reviewed journals and written in the English language were considered. Data from additional resources, such as relevant abstract derived from proceedings of main international oncological meetings, were also included in the research. Studies were retrieved and reviewed by three different authors (A.D.F., A.D.G., and C.P.).

Records underwent a first screening for title and abstract. Relevant articles were subsequently screened for full text and analyzed to identify those meeting the inclusion criteria. The bibliography of each relevant article was finally searched.

The Preferred Reporting Items for Systematic Review and Meta-Analyses guidelines were adopted to conduct this meta-analysis.

Risk of Bias

The risk of bias of the included studies was independently evaluated by three authors (A.D.F., A.D.G., and C.P.), using the tools of the Cochrane Collaboration for assessing risk of bias (selection, performance, detection, attrition, and reporting bias). The results of this meta-analysis were interpreted according to the risk of bias, and any disagreement was resolved by discussion to reach consensus. A summary of the risk of bias evaluation is reported in Table 1. The presence of publication bias was evaluated through funnel plots (Fig. 1A–D).

Table 1

Summary of Authors’ Judgment on the Risk of Bias for Each Selected Randomized Controlled Clinical Trial According to the Cochrane Collaboration for Assessing Risk of Bias

Category	KEYNOTE021	KEYNOTE189	KEYNOTE407	IMpower130	IMpower131	IMpower132	IMpower150	ORIENT-11
Random sequence generation	Low	Low	Low	Low	Low	Low	Low	Low
Allocation concealment	Low	Low	Low	Low	Low	Low	Low	Low
Selective reporting	Low	Low	Low	Low	Low	Low	Low	Low
Blinding participants and personnel	High	Low	Low	High	High	High	High	High
Blinding outcome assessment	Low	Low	Low	Unclear	Unclear	Unclear	Unclear	Unclear
Incomplete outcome data	Low	Low	Low	Low	Low	Low	Low	Low
Other	Unclear	Unclear	Short follow-up duration	Unclear	Unclear	Low	Low	Unclear

Figure 1

Funnel plots of (A) PFS, (B) OS, (C) ORR, and (D) grade greater than or equal to 3 TRAEs revealing basic symmetry, suggesting the absence of publication biases. ORR, objective response rate; OS, overall survival; PD-1, programmed cell death protein-1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; RR, risk ratio; TRAEs, treatment-related adverse events.

Outcome Measures

Efficacy outcomes (OS and PFS, expressed as HR, and ORR, expressed as risk ratio [RR]) were extracted from the selected studies. OS was defined as the time from the date of treatment assignment to the date of death from any cause; PFS was defined as the time from treatment assignment to disease progression or death from any cause; ORR was defined as the proportion of patients who achieved a partial or complete response to therapy according to the Response Evaluation Criteria in Solid Tumors version 1.1. Grade greater than or equal to 3 treatment-related adverse events (TRAEs), classified according to WHO or National Cancer Institute Common Toxicity Criteria, were also collected for the analysis, and their incidence was expressed as RR.

Data Extraction and Analysis

This meta-analysis was performed using a dedicated software (Review Manager Version 5.4; The Cochrane Collaboration). Summary measures were HR with 95% confidence interval (CI) for OS and PFS and RR with 95% CI for ORR and grade greater than or equal to 3 TRAEs. For the time-to-event variables, including OS and PFS, HRs with 95% CIs were evaluated for each study. The inverse variance technique was adopted for the meta-analysis of the HR. RRs with 95% CIs were calculated for each study for the analysis of ORR and grade greater than or equal to 3 TRAEs. The Mantel–Haenszel method was used to combine the RRs of the analyzed trials. Finally, a meta-regression analysis was used to directly compare PD-1 inhibitors and PD-L1 inhibitors in terms of OS (HR), PFS (HR), grade greater than or equal to 3 TRAEs (RR), and ORR (RR). The meta-regression analysis was performed using a dedicated software (Comprehensive Meta-Analysis Version 3; Biostat Inc.). Statistical heterogeneity between studies was evaluated using the chi-square test and the I2 statistic. Considering the number of the included studies and their poor heterogeneity, the fixed-effect model was used for this meta-analysis. Nevertheless, all the analyses were repeated using the random-effect model to verify their consistency.

For KEYNOTE 021, KEYNOTE 189, and KEYNOTE 407, data from updated analyses were used for this meta-analysis.

Results

The initial database search yielded a total of 426 records eligible for inclusion (one identified through conference abstract research). On the basis of duplicate records and review of titles or abstracts, two articles were excluded, resulting in 424 potentially eligible studies. Through reviewing titles and abstracts of all articles, 402 articles were excluded in accordance with the exclusion criteria. Full texts of the 22 remaining articles were further reviewed in detail, and, finally, eight prospective RCTs (four RCTs testing atezolizumab plus CT versus CT alone, three RCTs testing pembrolizumab plus CT versus CT alone, and one RCT testing sintilimab plus CT versus CT alone), published from 2016 to 2021 and including a total of 4466 patients, matched the inclusion criteria and were selected for this meta-analysis. The included trials have almost superimposable inclusion and exclusion criteria. In particular, key eligibility criteria were the following: unresectable NSCLC; absence of previous treatment with CT or ICI; at least one measurable lesion per Response Evaluation Criteria in Solid Tumors 1.1; baseline Eastern Cooperative Oncology Group performance status score of 0 or 1; absence of untreated or symptomatic brain metastasis; absence of active immunosuppressive treatment. Three trials testing the PD-L1 inhibitor atezolizumab allowed the inclusion of patients with sensitizing EGFR mutations or ALK rearrangements if they have had disease progression (during or after treatment) or intolerance to treatment with at least one tyrosine kinase inhibitor, although their presence represented an exclusion criterion for the other trials. The search process is summarized in the flow diagram of the Preferred Reporting Items for Systematic Review and Meta-Analyses (Fig. 2). A brief summary of the included studies is reported in Table 2.

Figure 2

PRISMA study flow diagram. PRISMA, Preferred Reporting Items for Systematic Review and Meta-Analyses.

Table 2

Main Characteristics of the Randomized Clinical Trials Selected for the Meta-Analysis

Trial	Phase	Histology	No. of Intervention/Control	Arms of Treatment	Primary Outcome	TRAEs Reported
KEYNOTE 0216,7	II	Nonsquamous	60/63	Pembrolizumab + carboplatin + pemetrexedvs.carboplatin + pemetrexed	ORR	Yes
KEYNOTE 18910,11	III	Nonsquamous	410/206	Pembrolizumab + cisplatin or carboplatin + pemetrexedvs.cisplatin or carboplatin + pemetrexed	PFS, OS	No
KEYNOTE 4078,9	III	Squamous	278/281	Pembrolizumab + carboplatin + nab-paclitaxel or paclitaxelvs.carboplatin + nab-paclitaxel or paclitaxel	PFS, OS	Yes
ORIENT-1112	III	Nonsquamous	266/131	Sintilimab + pemetrexed + cisplatin or carboplatinvs.pemetrexed + cisplatin or carboplatin	PFS, OS	No
IMpower13015	III	Nonsquamous	483/240	Atezolizumab + carboplatin + nab-paclitaxelvs.carboplatin + nab-paclitaxel	PFS, OS	Yes
IMpower13113	III	Squamous	343/340	Atezolizumab + carboplatin + nab-paclitaxelvs.carboplatin + nab-paclitaxel	PFS, OS	Yes
IMpower13216	III	Nonsquamous	292/286	Atezolizumab + cisplatin or carboplatin + pemetrexedvs.cisplatin or carboplatin + pemetrexed	PFS, OS	Yes
IMpower15014	III	Nonsquamous	400/400	Atezolizumab + bevacizumab + carboplatin + paclitaxelvs.bevacizumab + carboplatin + paclitaxel	PFS, OS	Yes

ORR, objective response rate; OS, overall survival; PFS, progression-free survival; TRAE, treatment-related adverse event.

Funnel plots of PFS, OS, ORR, and grade greater than or equal to 3 TRAEs revealed basic symmetry, suggesting no publication bias (Fig. 1).

In the safety analysis, only trials reporting TRAEs were included (six of eight). KEYNOTE 189 and ORIENT-11, owing to a different method of safety recording, were excluded (Table 2).

PD-(L)1 Inhibitor Plus CT Versus CT Alone

Both PD-1 and PD-L1 inhibitors were found to improve the outcomes of patients with advanced NSCLC compared with CT alone, regardless of PD-L1 expression. The addition of a PD-(L)1 inhibitor to CT improved both PFS (HR = 0.59, 95% CI: 0.55–0.63) (Fig. 3A) and OS (HR = 0.76, 95% CI: 0.70–0.82) (Fig. 3B). ORR was also improved by the combination of an ICI to CT as compared with CT alone (RR = 1.52, 95% CI: 1.41–1.63) (Fig. 3C). The risk of grade greater than or equal to 3 TRAEs was slightly higher with the addition of a PD-(L)1 inhibitor to CT (RR = 1.19, 95% CI: 1.12–1.26) (Fig. 3D) than with CT alone.

Figure 3

Hazard ratios of (A) PFS and (B) OS and (C) risk ratios of ORR and (D) grade greater than or equal to 3 TRAEs of the addition of a PD-(L)1 inhibitor to first-line chemotherapy-based treatment. The subgroup analysis revealed favorable survival, improved response rate, and less toxicity with the addition of a PD-1 inhibitor to CT as compared with the addition of a PD-L1 inhibitor to CT. CI, confidence interval; CT, chemotherapy; ICI, immune checkpoint inhibitor; ORR, objective response rate; OS, overall survival; PD-1, programmed cell death protein-1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; TRAE, treatment-related adverse event.

PD-1 Inhibitor Plus CT Versus PD-L1 Inhibitor Plus CT

A subgroup analysis was performed according to the targeted receptor (PD-1 versus PD-L1) to evaluate if the addition of a PD-1 inhibitor or a PD-L1 inhibitor to histology-driven CT could lead to different outcomes in terms of efficacy and safety. We observed that the addition of a PD-1 inhibitor to CT had a greater impact on PFS (HR = 0.52, 95% CI: 0.46–0.58) than that of a PD-L1 inhibitor (HR = 0.63, 95% CI: 0.58–0.69) (Fig. 3A) as compared with CT alone (p for subgroup differences = 0.006, I2 = 86.6%). Similar results were observed in terms of OS ([HR for PD-1 inhibitors = 0.64, 95% CI: 0.56–0.73] versus [HR for PD-L1 inhibitors = 0.83, 95% CI: 0.75–0.92], p for subgroup differences = 0.002; I2 = 89.1%) (Fig. 3B) and ORR ([RR for PD-1 inhibitors = 1.86, 95% CI: 1.63–2.11] versus [RR for PD-L1 inhibitors = 1.36, 95% CI: 1.25–1.48], p for subgroup differences = 0.001, I2 = 93.5%) (Fig. 3C). Moreover, the risk of grade greater than or equal to 3 TRAEs was significantly increased (p for subgroup differences = 0.04, I2 = 75.6%) with the addition of a PD-L1 inhibitor to CT (RR = 1.22, 95% CI: 1.15–1.30), but not with the addition of a PD-1 inhibitor to CT (RR = 1.04, 95% CI: 0.90–1.20) (Fig. 3D).

A direct comparison of the two types of inhibitors using a meta-regression analysis confirmed that the addition of a PD-1 inhibitor to CT, as compared with that of a PD-L1 inhibitor to CT, was associated with improved PFS (HR = 0.82, 95% CI: 0.71–0.95, p = 0.007), OS (0.77, 95% CI: 0.65–0.91, p = 0.002), and ORR (RR = 1.33, 95% CI: 1.08–1.56, p = 0.0002).

A further comparison in the subgroup of patients with PD-L1 greater than or equal to 50% in terms of PFS and OS was performed using a meta-regression analysis. Trials excluded owing to lack of data for this population were KEYNOTE 021 in the PFS analysis and KEYNOTE 021, ORIENT-11, and IMpower150 in the OS analysis. No difference either in terms of PFS (HR = 0.79, 95% CI: 0.57–1.10, p = 0.166) or OS (HR = 1.04, 95% CI: 0.65–1.67; p = 0.865) was documented between the two groups.

Moreover, the addition of a PD-1 inhibitor to CT seemed to be associated with lower risk of grade greater than or equal to 3 TRAEs than that of a PD-L1 inhibitor (RR = 0.84, 95% CI: 0.73–0.98, p = 0.027).

Discussion

An increasing number of studies evaluated the addition of a PD-(L)1 inhibitor to CT as upfront treatment for nononcogene-addicted advanced NSCLC. As, to date, pembrolizumab, atezolizumab, and sintilimab have all been found to improve clinical outcomes obtained with CT alone, we believed that a meta-analysis to explore if there were differences depending on the targeted receptor would have been worthwhile. On the basis of our study, the addition of an ICI, either PD-1 or PD-L1 inhibitor, to CT significantly improved the outcomes of patients with NSCLC and only modestly increased the risk of developing severe adverse events. Despite the superiority of the overall effect of the combination strategy over CT, it should be highlighted that two studies in the PD-L1 inhibitor plus CT group did not meet one of their coprimary end points. In fact, both IMpower131 and IMpower132, which tested histology-driven CT plus atezolizumab versus CT alone in squamous and nonsquamous NSCLCs, respectively, failed to reveal an OS benefit with the combination strategy, despite reaching a significant PFS improvement., Therefore, the results of these two trials are those largely responsible for the poorer outcomes obtained by the PD-L1 inhibitor plus CT subgroup compared with PD-1 plus CT subgroup, at least in terms of OS. Nevertheless, definitive conclusions cannot be easily drawn. In fact, despite the difficulty in indirectly comparing different trials, it is worth of consideration that, excluding the only phase 2 trial,, control arms of phase 3 studies testing the PD-L1 inhibitor atezolizumab had numerically better outcomes in terms of median OS than those testing PD-1 inhibitors.8, 9, 10,13, 14, 15 Conversely, ORR and median PFS of the control arms were numerically similar in the two groups. Moreover, the IMpower132 experimental regimen has a survival benefit in the Asian population of the study, leading to its approval in Japan.

A recent cross-sectional meta-analysis conducted on 11,379 patients with cancer enrolled in 19 RCTs revealed favorable outcomes with PD-1 inhibitors compared with PD-L1 inhibitors in terms of both PFS (HR = 0.73, 95% CI: 0.56–0.96, p = 0.02) and OS (HR = 0.75, 95% CI: 0.65–0.86, p < 0.001), with no significant difference in their safety profile. In advanced NSCLC, a meta-analysis comparing PD-(L)1 inhibitors as monotherapy for previously treated patients revealed that PD-1 inhibitors were associated with a similar risk of any grade greater than or equal to 3 adverse events (except for the risk of pneumonitis, which was higher with PD-1 inhibitors) but better OS (HR = 0.67, 95% CI: 0.60–0.74) and PFS (HR = 0.81, 95% CI: 0.72–0.91) trends than PD-L1 inhibitors (HR for OS = 0.80, 95% CI: 0.71–0.90; HR for PFS = 1.02, 95% CI: 0.89–1.17). In agreement with preexistent data, our meta-analysis confirmed that both PD-1 and PD-L1 inhibitors improved the outcomes of patients with advanced NSCLC when added to first-line, histology driven, platinum-based CT., Nevertheless, the addition of a PD-1 inhibitor to CT seemed to be safer in terms of TRAEs of grade greater than or equal to 3 and to provide higher benefit in terms of PFS, OS, and ORR when compared with that of a PD-L1 inhibitor added to CT. We hypothesize three orders of reasons that could explain these findings, which are as follows: clinical, pharmacologic, and biological ones.

First, we analyzed and compared the characteristics of patients enrolled in studies evaluating the two ICIs. Inclusion and exclusion criteria of all studies were almost superimposable, and patients enrolled had similar baseline characteristics. Three of four trials testing the addition of a PD-1 inhibitor to CT (KEYNOTE 021, KEYNOTE 189, and ORIENT-11) and three of four trials testing atezolizumab (IMpower130, IMpower132, and IMpower150) enrolled patients with nonsquamous NSCLC, whereas the remaining two (KEYNOTE 407 and IMpower131) enrolled patients with squamous NSCLC.6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16 It is unlikely that the difference between the outcomes obtained by the addition of PD-1 or PD-L1 inhibitor could be attributed to dissimilar NSCLC histologies, as trials that tested each agent in both histologies (squamous and nonsquamous) have been included in this meta-analysis. Moreover, the subgroups (PD-1 versus PD-L1) were balanced as they included the same number of trials per histology (Table 2). It is worth considering that of the two included trials conducted on squamous NSCLC,,, only the one testing the addition of a PD-1 inhibitor to CT (KEYNOTE 407) demonstrated a survival benefit over CT alone. Nevertheless, as previously mentioned, also one trial testing the combination of the PD-L1 inhibitor atezolizumab plus CT versus CT alone in nonsquamous histology failed to reach a survival improvement (IMpower132), making difficult to attribute the better outcomes obtained by the anti–PD-1 subgroup to a poor performance of anti–PD-L1 in squamous histology. Moreover, atezolizumab recently was found to perform well as a single-agent therapy in either squamous or nonsquamous NSCLC with PD-L1 greater than or equal to 50%. The proportion of patients with PD-L1–positive NSCLC was slightly higher in studies testing PD-1 inhibitors (≃63.5%) than in those testing PD-L1 inhibitors (≃50%), but this little gap is unlikely to be responsible for statistically significant different outcomes. Consistently, the comparison performed in the subgroup of patients with PD-L1 greater than or equal to 50% revealed no PFS or OS differences. Differently from all the trials evaluating the addition of a PD-L1 inhibitor to CT, IMpower130, IMpower131, and IMpower150 trials also included a minority of patients with known EGFR gene mutation or ALK gene rearrangements, which are considered negative predictive factors of response to immunotherapy. Nevertheless, interestingly, the subgroup analysis of patients with EGFR-mutated NSCLC enrolled in the IMpower150 trial reported improved outcomes with the combination treatment as compared with CT alone.

The different pharmacokinetic and pharmacodynamic profiles of these agents may provide an additional reason for the observed results. Pembrolizumab and sintilimab retain higher binding affinity for its ligand (Kd = 0.029 nM and 0.074 nM, respectively) than atezolizumab (Kd = 1.75 nM)., Pembrolizumab and sintilimab are immunoglobulin G4, that represents the preferred class of immunoglobulin for immunotherapy. In fact, its low affinity for C1q and Fc receptors minimizes the risk to trigger host effector functions, such as activation of the complement system and antibody-dependent cytotoxicity. Conversely, the immunoglobulin G1 class, such as that of atezolizumab, harbors a higher potential of antibody-dependent cytotoxicity activation; to reduce this risk, atezolizumab was engineered with a modification in the Fc domain. Anyhow, the main diversity between these ICIs might stand in their different targeted receptors. In fact, although disrupting the interaction between PD-1 and its primary ligand, PD-L1, the inhibition of these targets prevents their bind with further side ligands, such as PD-L2 in the case of pembrolizumab and sintilimab or B7.1 in the case of atezolizumab. PD-L2 has lower level and range of expression compared with PD-L1, but its presence has been recently reported in a large number of tumor types., Similar to PD-L1, the interaction with PD-1 seems to dampen and regulate T-cell immune response during the induction and effector phases, thus having similar effects with PD-L1 and playing a role in cancer immune evasion.31, 32, 33 Nevertheless, although a wide number of studies documented the association between PD-L1 expression and impaired survival, the same correlation with PD-L2 is still debated.34, 35, 36, 37, 38, 39

Finally, the limitations of this meta-analysis should be acknowledged. Included studies had slightly different CT backbones. Nevertheless, all of them were platinum-based regimens that represent the standard chemotherapeutic options for patients with advanced NSCLC, with known superimposable outcomes. Moreover, HRs reported only indicate the incremental treatment effect added by the combination of an ICI and CT as compared with the same CT regimen without ICI. For these reasons, we do not see it as a relevant bias of the analysis. Consistently, for the one study (IMpower150) that included bevacizumab, the antiangiogenetic agent was comprised in the CT backbone; thus, the difference between the two treatment arms of the study was only represented by the addition of atezolizumab in one. Again, as we only evaluated the incremental effect of the ICI as compared with the same treatment without ICI, we do not see it as a relevant bias. A further limitation consisted in the exclusion of two of eight trials selected in the safety analysis. In fact, KEYNOTE 189 and ORIENT-11, which both belonged to the PD-1 inhibitor group, did not report TRAEs as safety recording., Finally, methodology of response assessment was different among the trials analyzed. In fact, some studies adopted a blinded independent assessment,6, 7, 8, 9, 10, 11 whereas response evaluation was investigator assessed in others.12, 13, 14, 15, 16 This aspect could represent a limit to data interpretation and cross-trial comparison.

In conclusion, our meta-analysis revealed that the addition of a PD-1 inhibitor to CT seems to be more effective and safer than that of a PD-L1 inhibitor. These findings need validation in prospective trials of direct comparison among different ICIs in combination with platinum-based CT.

CRediT Authorship Contribution Statement

Alessandro Di Federico: Conceptualization, Formal analysis, Writing—original draft.

Andrea De Giglio: Conceptualization, Data curation, Writing—original draft.

Claudia Parisi: Writing—review and editing.

Francesco Gelsomino: Supervision, Writing—review and editing.

Luca Boni: Formal analysis.

Andrea Ardizzoni: Methodology, Supervision, Project administration.