Identifying optimal first-line immune checkpoint inhibitors based regiments for advanced non-small cell lung cancer without oncogenic driver mutations: A systematic review and network meta-analysis

Background Immune checkpoint inhibitors (ICIs) have changed the treatment pattern of advanced and metastatic NSCLC. A series of ICI based therapies have emerged in the first-line treatment field, but the comparative efficacy was unclear. Method We searched multiple databases and abstracts of major conference proceedings up to Apri1, 2022 for phase III randomised trials of advanced driver-gene wild type NSCLC patients receiving first-line therapy. Outcomes analyzed included progression free survival (PFS), overall survival (OS), and et al. Results Thirty-two double-blind RCTs were included, involving 18,656 patients assigned to 22 ICI-based first-line regimens. A series of ICI regiments (including ICI plus chemotherapy), ICI monotherapy, doublet ICIs, doublet ICIs plus chemotherapy) emerged, and showed significant PFS and OS benefit than chemotherapy and chemotherapy + bevacizumab (BEV) for advanced wild-type NSCLC. In comprehensive terms of PFS, chemoimmunotherapy (CIT) were significantly more effective than ICI monotherapy and doublet ICIs. In terms of OS for patients with non-squamous NSCLC, pembrolizumab containing CIT was associated with a median rank of the best regimens, and followed by Atezolizumab+BEV based CIT; while for OS in patients with squamous NSCLC, Cemiplimab and sintilimab based CIT were the most effective regimens. For more than 2 years follow-up, the atezolizumab, pembrolizumab, nivolumab and durvalumab containing ICI therapy all provide a durable long-term OS benefit over chemotherapy and BEV + chemotherapy. Conclusions The findings of the present NMA represent the most comprehensive evidence, which might suggest or provide basis for first-line ICI therapy decision for advanced NSCLC patients without oncogenic driver mutations.


Contribution to the field
Immune checkpoint inhibitors (ICIs) including anti-programmed death-1 (PD-1) monoclonal antibody and PD-L1 have greatly changed the treatment pattern of advanced and metastatic NSCLC.Recently, a series of phase 3 trials evaluated the efficacy of new anti-PD-1 antibodies have been published, and new combined ICIs approved indications for advanced NSCLC.At ASCO 2020, impressive long-term PFS and OS data were published with the updated final analysis of KEYNOTE-189 [4] and CheckMate 227.Moreover, at WCLC &ESMO 2021, data of several new first-line phase 3 clinical trials assessed ICI therapy have been published as abstracts.Increasing treatment options may reshape the first-line ICI interventions for advanced NSCLC patients.However, there remains considerable debate about the best therapeutic model (i.e., immunotherapy as single agent or in combination, and how to choose optimal combination therapy), and the potential differences in tolerability between individual treatment strategies.With the marketing of new ICI regiments and increasing numbers of trials published and long-term outcomes updated recently, we conducted an updated network meta-analysis to compare the treatment efficacy of different first-lines regimens for advanced NSCLC patients without oncogenic driver mutations.
Lung cancer remains the leading cause of cancer mortality worldwide, which caused 1.8 million deaths in 2020.Non-small cell lung cancer (NSCLC) accounts for about 80-90% of lung cancer cases [1].Immune checkpoint inhibitors (ICIs) including anti-programmed death-1 (PD-1) monoclonal antibody and PD-L1 have greatly changed the treatment pattern of advanced and metastatic NSCLC [2][3][4][5][6].ICI based therapies have been recommended as standard first-line regimens in metastatic or advanced NSCLC patients without oncogenic drivers, as monotherapy or combined with chemotherapy according to PD-L1 expression [7].
Currently, evidence suggests that the OS benefit from monotherapy was mainly observed in patients with high PD-L1 expression (PD-L1 TPS�50%) in first-line therapies for advanced NSCLC, such as pembrolizumab in KEYNOTE 024, atezolizumab in IMpower 110 or cemiplimab in EMPOWER-Lung [5,6,8,9].Whereas for patients with PD-L1-intermediate expression (1%� PD-L1 < 50%), ICIs + chemotherapy is generally considered the best option.Based on IMpower 150, atezolizumab+ cevacizumab (BEV) + chemotherapy (frequently referred to as ABCP) are also recommended by the US FDA and the European Medicines Agency (EMA) as the first-line treatment of patients without oncogenic drivers [10,11].The innovative CheckMate 9LA study demonstrated rapid disease control with limited-course chemotherapy plus IO doublet, and provided a new first-line treatment option [12].Additionally, a series of phase 3 trials evaluated the efficacy of new anti-PD-1 antibodies developed by Chinese pharmaceutical companies have been published, and new combined ICIs approved indications for advanced NSCLC by the National Medical Products Administration (NMPA) in china [13][14][15][16][17][18].At ASCO 2020, impressive long-term PFS and OS data were published with the updated final analysis of KEYNOTE-189 [4] and CheckMate 227 [19,20], both regardless PD-L1 expression.Moreover, at WCLC & ESMO 2021, data of several new first-line phase 3 trials analyzed the efficacy of ICI therapies were published as abstracts [21,22].Increasing treatment options may reshape the first-line ICI therapeutic strategy for advanced NSCLC.
There remains a lot of controversy about the best therapeutic model (i.e., immunotherapy as single agent or in combination, and how to choose optimal combination therapy), and the differences in treatment-related adverse events between individual regiments.With the marketing of new ICI regiments, the most recent clinical trials findings, and long-term outcomes updated recently [4,[19][20][21][22], we conducted an updated network meta-analysis to compare the different first-lines regimens in term of short and long term PFS and OS for advanced NSCLC patients without oncogenic driver mutations.

Ethical approval
Ethics approval was obtained from the Ethics Committee of Shandong Provincial Hospital affiliated with Shandong First Medical University.

Search strategy and selection criteria
This study is registered with PROSPERO, number CRD42021291015, and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) extension statement for network meta-analysis.
We searched PubMed, the Cochrane Library, Embase, the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP), including conference proceedings of the World Conference on Lung Cancer (WCLC), the American Society of Clinical Oncology (ASCO), the European Society of Medical Oncology (EMSO), and the Chinese Society of Clinical Oncology (CSCO) Academic Annual Conference, from inception until Apri1, 2022, with no language restrictions.The following Medical Subject Headings were used to search: non-small cell lung cancer (including non-squamous lung cancer, squamous lung cancer), and immunotherapy (including all currently known ICIs: pembrolizumab, atezolizumab, ipilimumab, nivolumab, avelumab, sintilimab, durvalumab, tremelimumab, tislelizumab, toripalimab, sugemalimab, camrelizumab, and et al.).Reference lists of relevant studies were also screened.
The inclusion criteria were as follows: studies that analyzing the efficacy and safety of ICIs (ICI monotherapy or double ICIs) alone or combined with chemotherapy and or antiangiogenetic drugs, as first-line treatments for advanced NSCLC; and reports contained patients' data, complete protocol, and at least one of key clinical outcomes, such as progression-free survival (PFS), overall survival (OS), and objective response rate (ORR), as well as the incidence of treatment-related adverse events (TRAEs).Phase III trials that analyzing the efficacy of chemotherapy in combination with BEV or placebo as first-line therapy for nonsquamous NSCLC were also included for comparison and to form a connected network.
The following studies were excluded from this review: those involving previously treated patients with advanced NSCLC; studies involving operable NSCLC who were treated with ICI therapy before surgery or after surgery; studies not reported key clinical outcomes (PFS, OS, ORR, the incidence of TRAEs); and studies lacking valid data for evaluating the safety and efficacy of ICI first-line treatment.

Data extraction
Two investigators (YL and WF) independently examined the titles and abstracts of retrieved articles to assess the eligibility.The full articles were evaluated if a decision could not be made based on the titles and abstracts.Data were extracted by the same two reviewers (YL and WF) using a predefined spread sheet.The extracted data included: trial name, year of publication, study design, number of participants, median/mean age, percent males, histology type, disease stage, PD-L1 expression level, median duration of follow-up, and outcome data.Outcome data including PFS, OS, and ORR were extracted; TRAEs associated with each intervention were also extracted in this network meta-analysis (S1 Table ).
The full texts screen was performed, when a decision could not determine based on the titles and abstracts.Two pairs of investigators (YL, WF, SWM ans SLL) extracted data from the studies or from supplementary materials, and evaluated the risk of bias.Disagreements were resolved through discussion.
We use the data of PFS, and ORR assessed by blinded independent central review (BICR) or independent radiographic review committee (IRRC), if available.Otherwise, investigatorassessed PFS, and ORR were used.Hazard ratio (HRs) and 95% confidence intervals (CIs) for PFS and OS were extracted, percentage for long-term PFS and OS rate were also extracted.Likewise, dichotomous ORR data, grade �3 AEs and immune-mediated were clustered.

Evidence networks
We performed fixed-effects network meta-analyses (NMA) in our study.
According to trial characteristics, heterogeneity analysis was performed on all eligible studies.The NMA meets the technical requirement that each treatment should be represented by at least one clinical study to create a viable comparison network.Under the assumption of consistency, the NMA model associated data from the individual studies with basic parameters that reflect the (pooled) relative treatment effect of each intervention compared with the reference therapy.Based on these parameters, the relative treatment effects of each contrast treatment in the network were obtained.Under the Bayesian framework, we could calculate the probability of being the best treatment out of all treatments included in the connected network.It is also possible to calculate which of all the interventions included in the connected network are the best and which are the next best.The 'rank probability' function was performed to calculate rank-probabilities, for each MCMC iteration, the treatments are ranked according to their effect relative to baseline.A frequency table was constructed according to these rankings and normalized by the number of iterations to give the rank probabilities.
Three histology-based evidence networks were constructed in our analysis: a mixed-histology network, a squamous NSCLC network and a non-squamous NSCLC network.The results of the NMA for PFS and OS were presented with estimates of the treatment effects of each intervention with respect to the reference therapy, chemotherapy.The posterior distributions of relative treatment effects were performed by the median and 95% CIs, which were made up of the 2.5th and 97.5th percentiles of the posterior distributions.If the 95% CIs were completely below or above 1, the respective estimated ratios were considered significant.Additionally, we used SUCRA to perform the rank of all treatments, the closer the SUCRA value is to 1, the better the treatment effect is.

Search results
A total of 2,375 citations were identified by the comprehensive search strategy (Fig 1).After excluding duplicate citations, 1765 citations underwent title/abstract screening and 65 studies were retrieved for full text screen.Of the 69 full-text articles screened, 37 studies not meeting the inclusion criteria were excluded.Overall, 32 first-line phase III RCTs (comprising 18,656 participants) done between 2016 and 2022 were included in the analysis.Study characteristics of included trials are listed in Table 1.The mean sample size was range from 1274 to 305 participants.

Outcomes of interest
PFS and OS were co-primary endpoints in most of the trials included.Similar definitions of OS were used across trials, defined as the time from randomization to death from any reason.In majority trials, PFS was defined as the time from randomization to disease progression or death from any cause, as assessed by the IRRC according to Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1.Whereas, the series of IMpower trials only reported investigator assessed PFS, ORR and DOR [8,10,11,[31][32][33][34][35][36].In the present analysis, investigator and IRRC assessed outcomes were assumed to be comparable.

PFS-NMA according to PD-L1 expression
PD-L1 TPS > 50% cohort.For patients with high PD-L1 TPS expression, all ICIs therapy except nivolumab monotherapy (HR, 0.62 [0.38-1.0])were significantly more effective than chemotherapy.Camrelizumab based CIT has a 90.4% probability of being the best treatment for PFS in patients with high PD-L1 TPS expression, followed by sintilimab (82.3%) based CIT and ABCP (79.1%).All CIT regiments appear superior to ICI monotherapy for PFS in patients with high PD-L1 TPS expression; camrelizumab and sintilimab containing CIT were both significantly more effective than pembrolizumab or atezolizumab alone, and pembrolizumab + ipilimumab (P<0.05) (

Objective response rate
In the overall NMA analysis of ORR, all CIT regiments except ipilimumab containing CIT significantly improved ORR compared with chemotherapy.The CIT regiments were superior to single-agent ICI (atezolizumab, pembrolizumab and nivolumab) for ORR.Pembrolizumab containing CIT showed numerical superiority over other ICI regimens for ORR, with a 92.2% probability of being the best treatment, followed by penpulimab (89.1%) and sugemalimab (78.0%) based CIT (S4 Fig).

Discussion
This NMA is based on 32 first-line RCTs for advanced NSCLC, which involving 18,656 patients randomized to 22 ICI-based regimens with 13 different ICI agents, and including recent published data of long-term follow-up of immunotherapy trials.The present analysis is substantially more comprehensive than previous meta-analysis for first-line ICI therapy for advanced NSCLC [2,3,22,23].The much larger evidence base, obtained through exhaustive search for published article and abstracts from recent major conference proceedings of the ASCO, EMSO, and WCLC.The main findings of this study including: (1) A series of ICI regiments (including ICI plus chemotherapy, doublet ICIs, doublet ICIs plus chemotherapy) emerged, and showed significant PFS and OS benefit than chemotherapy and chemotherapy +BEV for advanced wild-type NSCLC.(2) In comprehensive terms of PFS, CIT were significantly more effective than ICI monotherapy and doublet ICIs, penpulimab based CIT and ABCP showed numerical superiority over other ICI regimens for PFS.(3) In terms of OS for patients with non-squamous NSCLC, pembrolizumab containing CIT ranked to be the best treatment regimen, and followed by sintilimab based CIT; while for OS in patients with squamous NSCLC, camrelizumab and sintilimab based CIT were the most effective regimens.Cemiplimab monotherapy showed numerical superiority over other CIT regimens for OS in patients with high PD-L1 TPS expression.( 4) No statistically significant benefit for OS or PFS was observed for doublet ICIs based therapies comparing with single ICI based therapies.(5) For >24 months follow-up, the atezolizumab, pembrolizumab, nivolumab and durvalumab containing ICI therapy (including monotherapy, CIT, doublet ICIs, doublet ICIs plus chemotherapy) all provide a durable, clinically meaningful long-term OS benefit over chemotherapy and BEV+ chemotherapy.
For overall study cohort, we observed that CIT regiments were overall superior to ICI monotherapy and doublet ICIs in terms of PFS and ORR.The PFS benefit of a series of CIT regiments including penpulimab, camrelizumab, pembrolizumab, sugemalimab, tislelizumab, toripalimab, sintilimab based CIT and ABCP is comparable for overall population.In the OS NMA for overall study population, pembrolizumab containing CIT ranks being the best treatment, followed by sintilimab containing CIT.These two CIT regimens were also statistically superior to ICI monotherapy (atelizumab or pembrolizumab) and atelizumab based CIT, but not statistically superior to camrelizumab, cemiplimab, sugemalimab, atelizumab+bevacizumab, or nivolumab+ipilimumab based CIT.The similar PFS and OS benefits achieved for the above CIT regiments suggest new better practice-changing options for advanced NSCLC without sensitizing EGFR or ALK genetic alterations.Additional data including costeffectiveness and toxicity of ICIs regiments need to be considered to formulate national and health policy decisions.
We found that the majority of the ICIs therapies were more efficacious compared with chemotherapy (Table 2).However, some ICIs monotherapy and CIT regiments don't provide statistically significant PFS or OS benefit over chemotherapy.Durvalumab plus tremelimumab or durvalumab monotherapy did not significantly improved OS or PFS vs chemotherapy in patients with NSCLC and PD-L1 TC � 25% [24].Similarly, nivolumab monotherapy wasn't significantly improved PFS compared with chemotherapy for patients with advanced or recurrent NSCLC whose PD-L1 expression > 5%, and OS benefit was also negative [25].Although PFS and OS was statistically significantly improved with first-line nivolumab plus ipilimumab than chemotherapy, the PFS and OS benefit from this doublet ICIs regiment is relative inferior to some other ICI treatment examined.ACP regiment also provide less benefit in PFS or OS compared with ABCP, camrelizumab, pembrolizumab and sintilimab containing CIT, with lower SUCRA value in PFS and OS NMA.The inferior efficacy, high cost of doublet ICIs, and toxicity making these strategies less favourable options.
For patients with squamous NSCLC, the PFS NMA analysis suggest sugemalimab and camrelizumab based CIT was ranked as the first-and second-best treatment option, which were statistically superior to pembrolizumab and atezolizumab based CIT, and with similar efficacy with penpulimab, cemiplimab, sintilimab, or tislelizumab based CIT.In the OS NMA of squamous subgroup, camrelizumab and cemiplimab based CIT ranked as the first-and secondbest therapy, followed by sintilimab based CIT.The OS benefit of these three CIT regimens were numerical superior to pembrolizumab containing CIT.The OS data is byfar immature for sugemalimab, penpulimab, and tislelizumab based CIT [25].As OS should be considered as the standard measure of clinical benefit for cancer patients, the present NMA results suggest that camrelizumab, cemiplimab and sintilimab based CIT may be recommended as an initial first-line intervention for those with advanced, squamous NSCLC.
For nonsquamous subgroup, ABCP with the highest probability to be the best treatment in PFS NMA, and followed by sintilimab, pembrolizumab based CIT.However, the PFS benefit of ABCP did not translate into the best OS benefit.Pembrolizumab containing CIT has the highest probability (94.6%) to be the best ICI regimen for OS in patients with advanced, nonsquamous NSCLC.Atezolizumab+bevacizumab-based CIT regimen ranked second for OS benefit, whereas the mean rank (67.0%) is substantially lower than pembrolizumab based CIT.ABCP is numercialy inferior to pembrolizumab and sintilimab based CIT in term of survival benefit.The results were generally consistent with with previous network meta-analysis by Frederickson et al., which study showed pembrolizumab plus platinum-based chemotherapy had statistically significant OS benefit compared with other interventions in advanced nonsquamous NSCLC patients without sensitive oncogenic drivers [23].
For PD-L1-high patients with NSCLC, CIT regiments exhibit superior PFS benefit compared with single agent ICI or doublet ICIs.Camrelizumab based CIT ranks first for PFS benefit, followed by sintilimab based CIT and ABCP.Whereas, the best survival benefit was observed for cemiplimab monotherapy, which performed marginally better than pembrolizumab containing CIT.Both of cemiplimab monotherapy and pembrolizumab containing CIT are numerical superior to pembrolizumab alone, cemiplimab based CIT and ABCP.Atezolizumab, nivolumab, and durvalumab alone didn't significantly improved OS for PD-L1-high patients with NSCLC.Findings from a previous network meta-analysis by Liu et al. showed for patients with PD-L1-high expression, CIT regiments should be superior to pembrolizumab alone.However, based on the recent additions to the treatment space of advanced NSCLC, the present NMA suggests cemiplimab monotherapy should also be recommended as the initial treatment choice for PD-L1-high patients.
In the PD-L1 negative and intermediate cohort, CIT regiments exhibit superior PFS and OS benefit to ICI alone.Cemiplimab based CIT has a mean rank of being the best treatment for OS in patients with intermediate PD-L1 TPS expression, followed by camrelizumab and pembrolizumab based CIT, without an apparent disparity between the three combinations.
Nivolumab+ipilimumab, camrelizumab, and pembrolizumab containing CIT show similar superior effectivity for patients with negative PD-L1 expression.This evidence suggests that camrelizumab, pembrolizumab based CIT generally be preferred first-line treatments for PD L1-intermediate or negative advanced NSCLC patients; cemiplimab based CIT may also be recommended as a preferred first-line treatment for PD L1-intermediate cohort and Nivolu-mab+ipilimumab plus chemotherapy for PD-L1 negative patients.

Long-term
After more than 2 years follow-up, all ICI regiments were consistently resulted in substantially increased long-term OS and PFS rate compared with chemotherapy alone and BEV plus chemotherapy.Substantial long-term survival benefit was observed across single ICI, CIT, and doublet ICIs regiments.Specially, single ICI and doublet ICIs regiments showed numerical superiority over CIT regimens.Although durvalumab monotherapy didn't provide statistically significant median OS or PFS benefit over chemotherapy, durvalumab has a mean rank of being the best treatment for long-term OS benefit, followed by Nivolumab+ipilimumab and pembrolizumab monotherapy.There is an ongoing debate about which is the most adequate primary endpoint in clinical trial for NSCLC.The current NMA results suggested that median OS or PFS is generally incomplete in settings for evaluating the treatment effects of ICI regiments for NSCLC.Previous studies on NSCLC showed that an increase in PFS does not necessarily result in OS benefit; however, post-progression survival is strongly associated with OS after early-line treatment.Sustained survival benefit from ICI therapy suggests that both median OS and post-progression survival have increased along the years, which illustrated the delayed treatment effect of ICI therapy.Moreover, first-line pembrolizumab monotherapy and pembrolizumab based CIT were shown to improve PFS-2, and the PFS-2 was nearly doubled for patients in the pembrolizumab based CIT group.The outcome of PFS-2 suggests that despite the high crossover rate, treatment benefit of the first line therapy was maintained into the next line.As a result, post-progression survival and PFS-2 should also be considered as endpoint in clinical trials of ICI therapy for NSLCL.Substantial long-term survival benefit suggests target and discover specific patients who respond best to ICI treatment has become the major issue.However, despite the PD-L1 expression was approved as a standard biomarker in the setting of ICI treatment for advanced NSCLC, the response prediction was imperfect.There are other potential predictive biomarkers which could be factored into identify populations who respond best to specific combinations.Further research is required to identify additional biomarkers, which will help to align specific ICI regiments to specific patient groups.
The literature search for the present NMA was comprehensive, with the largest number of phases III RCTs for first-line immunotherapy treatments for advanced NSCLC, including 15748 patients enrolled in 32 studies.The 32 studies included recent published trials, longterm update of previous trials and abstracts from 2021 major conference proceedings to date.A previous NMA by Liu et al., based on one phase II and nine phase III studies, involving 6,124 patients with metastatic NSCLC, shows CIT intervention is superior to pembrolizumab monotherapy for patients with PD-L1 level � 1% and especially for those with PD-L1 � 50%.For non-squamous NSCLC, Bev +chemotherapy should also be recommended as an initial treatment for patients with PDL1 � 1%, since Bev +chemotherapy was not inferior to Pembrolizumab alone.However, with incorporated a considerable amount of recent published data, the present NMA shows the best survival benefit for PD-L1 � 50% cohort was observed for cemiplimab monotherapy, which performed marginally better than pembrolizumab containing CIT.For non-squamous NSCLC, the present NMA suggest all ICI regiments show longterm OS benefit compared to Bev +chemotherapy.A recent NMA was conducted in 2021 by Herbst et al., which including 17 clinical trials, limited on patients with stage IV NSCLC and high PD-L1 expression subgroup.Whereas, our NMA explored the relative benefit of ICI regimens on OS/PFS by PD-L1 levels and irrespective of PD-L1-positivity.
Our review has some limitations.First, most of the recent published data with short duration of follow-up, the number of OS events was not mature and the long-term PFS and OS rates were not available.The short follow-up prevented the full assessment of specific ICI regiment.Second, the majority studies included implement the 22C3 pharmDx assay to test the PD-L1 TPS, which is defined based on the proportion of tumor cells in membranous PD-L1 staining.38PD-L1 positivity refer to a tumour proportion score of 1% or higher.Whereas, IMpower studies utilize SP142 assaying to test PD-L1 level [15,17,40].Different assay method and reference value might lead to misclassification, affecting outcome assessment.Third, the different disease stage of study cohort may also be a confounding factor.Keynote, IMpower, and CheckMate studies entrolled stage IV NSCLC, whereas others included patients with stage IIIB or IIIC disease, meaning that the comparison need to be interpreted with great caution.Fouth, we lacked data from head-to-head comparisons of these ICIs regiments, and many conclusions are reliant on indirect comparisons.The limited value of indirect comparisons of results obtained from multiple different regimens, with varying follow-up time, suggests the results need to be taken with caution.Last, non-Asian patients were not recruited in the ORI-ENT, RATIONALE, Camel and CHOICE studies, and the results limited to Chinese subpopulation and therefore should be interpreted with caution.
Notwithstanding these limitations, the findings from this network meta-analysis represent the most comprehensive currently available evidence base for initial choice about first-line immunotherapies for advanced NSCLC patients without known EGFR mutations or ALK translocations.However, the findings from comparisons among ICI regiments should be interpreted by the potential limitations of the study design, and the heterogenous patient populations.The results of the present NMA might suggest or provide basis for first-line ICI therapy decision for advanced NSCLC patients without oncogenic driver mutations.