The landscape of immune checkpoint inhibitor plus chemotherapy versus immunotherapy for advanced non‐small‐cell lung cancer: A systematic review and meta‐analysis

Abstract Background Lung cancer is the leading cause of cancer‐related deaths worldwide and the prognosis remains poor. The recent introduction of the immune checkpoint inhibitor (ICI), or plus chemotherapy, both resulted in the survival benefit for patients with advanced non‐small‐cell lung cancer (NSCLC), but it remains unanswered which is superior. The current study aimed to estimate the comparative efficacy and safety of ICI‐chemotherapy versus ICI‐monotherapy in advanced NSCLC. Methods Studies were identified by searching PubMed, Embase, and Cochrane library. The randomized controlled trials (RCTs) that ICI monotherapy or ICI plus chemotherapy compared with chemotherapy in NSCLC were included with available primary endpoints of progression‐free survival (PFS), overall survival (OS), objective response rate, or treatment‐related adverse events. A fixed‐effect or random‐effects model was adopted depending on between‐study heterogeneity. Results A total of 20 RCTs involving 12,025 patients with NSCLC were included. Both ICI‐monotherapy and ICI‐chemotherapy resulted in significantly prolonged survival compared to chemotherapy and the former led to significantly longer PFS. The magnitude of survival benefits appeared to be greatest among those treated with pembrolizumab plus platinum‐based chemotherapy (OS, 0.56; PFS, 0.54). Additionally, OS and PFS advantages of ICI therapies were observed in patients with NSCLC with low or high programmed cell death 1 ligand 1 (PD‐L1) expression level, but not in intermediate PD‐L1 TPS. Conclusions Pembrolizumab plus platinum‐based chemotherapy was recommended as the optimal first‐line therapy for advanced patients with NSCLC. Additionally, PD‐L1 alone is not recommended as an adequate molecular biomarker to identify eligible patients for routine clinical practice in immunotherapy.

PD-L1 alone is not recommended as an adequate molecular biomarker to identify eligible patients for routine clinical practice in immunotherapy.

K E Y W O R D S
chemotherapy, efficacy, immune checkpoint inhibitor, non-small-cell lung cancer, safety 1 | INTRODUCTION Lung cancer is the leading cause of cancer death worldwide (Davies, Cheng, Field, Liu, & Li, 2019;Siegel, Miller, & Jemal, 2018). Most patients with lung cancer are diagnosed at an advanced stage with metastasis . At the metastatic stage, the 5-year survival rate is no more than 5% since no curative treatment options.
However, only a small population of patients (16%) are diagnosed at an early stage, for which the 5-year survival rate amounts to 56% (Siegel et al., 2018). Non-small-cell lung cancer (NSCLC) presents as the most prevalent histological subtype (>85%) of lung cancer (Herbst, Morgensztern, & Boshoff, 2018). Squamous-cell NSCLC, accounting for approximately 30% of all cases of NSCLC and lacking the targetable genetic aberrations, is related with worse prognosis than is nonsquamous-cell NSCLC .
The standard-of-care therapies as first-line treatments for patients with advanced NSCLC include platinum-doublet chemotherapy for those with nonsquamous cancer and targeted treatments for those with targetable genetic aberrations (Planchard et al., 2019;Sandler et al., 2006). However, the clinical outcomes of patients with NSCLC remain poor. Some reasons may explain this. On the one hand, platinum-based chemotherapy (PBC) only provides advanced NSCLC patients with a 15-30% response rate. On the other hand, new agents, such as docetaxel (DOC), can significantly improve survival benefits comparing with supportive care, but at the expense of a higher risk of adverse events (Fillon, 2018). Furthermore, only a small proportion of patients with NSCLC could benefit from targeted therapy due to the lack of targetable mutations (Camidge, Doebele, & Kerr, 2019).
Recently, growing evidence indicates that immune checkpoint inhibitor (ICI) therapies are promising therapeutic options for patients with NSCLC. Several ICI drugs have been approved by Food and Drug Administration and they could inhibit downregulation of antitumor responses through blocking programmed death 1 (PD-1), programmed death ligand 1 (PD-L1), and cytotoxic T-cell lymphocyte antigen 4 (CTLA-4) pathway in mechanism (Ribas & Wolchok, 2018). NSCLC tumor cells escape immune attack and induce tumor tolerance through developing immune checkpoints. For example, the tumor express ligand PD-L1, which is prevalent in NSCLC, is engaged by the activated T expressed receptor PD-1 to downregulate the antitumor function of T cells and promote immune escape (Pardoll, 2012). By blocking such immune checkpoints, the activation of T cells could be continued and the immune system could eliminate NSCLC cancer cells (Pardoll, 2012). The anti-PD-1 monotherapy is the first-line treatment for patients with NSCLC However, the relative efficacy and safety of different ICI strategies for advanced patients with NSCLC remains controversy.
In clinical practice, current ICI strategies contain two or three of the following treatment or different doses of the same ICI drug, including nivolumab (NIV), pembrolizumab (PEM), avelumab, atezolizumab (ATE), ipilimumab (IPI), and conventional therapy (chemotherapy or and targeted therapy). Furthermore, it is hypothesized that ICI plus chemotherapy (ICI-chemotherapy) might exhibit synergistical effects on survival benefits because the antitumor activity could be mediated by the cytotoxic effects of chemotherapy and the immunological effects of ICI therapies (e.g., chemotherapy could induce PD-L1 expression on tumor cells and infiltrating immune cells therefore enhancing the therapeutic effects of ICI therapies; Havel, Chowell, & Chan, 2019). ICI therapies, harnessing the immune system, are demonstrating promising results in combination with chemotherapy . Nevertheless, in a Phase III randomized controlled trial (RCT), ATE plus PBC failed to show any benefits over chemotherapy with respect to OS  The keywords included ICIs (anti PD-1 or anti PD-L1 or anti CTLA-4), specific ICI drug names (avelumab, ATE, durvalumab, IPI, NIV, PEM, tremelimumab), and lung cancer (The search strategy is detailed in Table   S1). We retrieved additional studies from major conference proceedings of the American Society of Clinical Oncology, the European Society of Medical Oncology, the American Association for Cancer Research, and the World Conference on Lung Cancer (WCLC). In terms of duplicated studies, the most complete data of the study was enrolled.
Study selection corresponded with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (Knobloch, Yoon, & Vogt, 2011;Liberati et al., 2009). Both exclusion and inclusion criteria were prespecified. The eligible RCTs met the following inclusion criteria as follow: (a) population: pathologically confirmed advanced patients with NSCLC; (b) intervention: treated with PD-1/PD-L1/CTLA-4 inhibitors (avelumab, ATE, durvalumab, IPI, NIV, PEM, or tremelimumab) with or without chemotherapy irrespective of dosage and duration; (c) comparison: treated with chemotherapy; (d) outcomes: PFS or OS measured as hazard ratios (HRs), objective response rate (ORR), and treatment-related adverse events (TRAEs) of any grade or grade ≥3 measured as risk ratios (RR). Studies were excluded based on the following criteria as follows: (a) designed as retrospective or prospective observational cohort studies; (b) lack of related data; (c) published as reviews, case reports, letters, commentaries, editorials, or meta-analysis; and (d) duplicated articles.
Manual search was performed through reviewing the reference lists of all trials fulfilling the eligibility criteria for additional relevant studies.

| Data extraction and quality assessment
The following items were extracted for each trial: first author, year of publication, acronym of the trial, trial phase, histology type, number of patients, OS, PFS, ORR, and TRAEs of any grade and grade ≥3. We carried out the methodological quality assessment of the enrolled trials with the Cochrane Risk of Bias Tool (Higgins et al., 2011), which consists of six items: random sequence generation; allocation concealment; blinding of participants and personnel to the study protocol; blinding of outcome assessment; incomplete outcome data; and selective reporting. An item identified as "low risk" is considered as applicable. Two authors (W. Q. and Y. J.) independently extracted data and performed quality assessment in this process and discrepancies were resolved by consensus (C. W.).

| Statistical analysis
The χ 2 test and I 2 statistic were applied to evaluate heterogeneity. The random effect models were chosen if I 2 was more than 50%, implying obvious heterogeneity, otherwise, the fixed-effect models were applied (Liberati et al., 2009). The primary outcomes were OS and PFS, presented with HRs, 95% CIs, and p values, which were calculated using the inversevariance-weighted method. The integrated analysis for ORR, Grade 1-5 TRAEs, and Grade 3-5 TRAEs were conducted based on the Mantel-Haenszel method. The Bucherʼs method was employed to make each of the pairwise indirect comparisons separately (Sultan, 2009).

| Characteristics of the studies and quality assessment
All RCTs were international multicenter studies published between 2012 and 2019, which were funded by the pharmaceutical industry.
All trails were completed in advanced or metastatic settings including Stage IIIB or IV or recurrent patients with NSCLC, who had the Eastern Cooperative Oncology Group performance-status (ECOG PS) score of 0 or 1. 17 of 20 eligible trials belonged to Phase III studies, and POPLAR and CA184-041 were Phase II trials. There were 6,490 patients enrolled in the intervention group (ICI-chemotherapy or ICImonotherapy) and 5,535 patients allocated to the chemotherapy control group. Among patients in the intervention arm, 53.5% were treated with ICI monotherapy compared with chemotherapy and 46.5% were treated with ICI-chemotherapy compared with chemotherapy. Four studies were conducted with squamous lung cancer, six with nonsquamous lung cancer, and 12 with mixed types of squamous and nonsquamous cancer. Overall, all but seven studies WANG ET AL.

| 4915
(35%) demonstrated an OS advantage for patients receiving ICI therapies compared with patients receiving chemotherapy. In subgroup analyses, eight studies (40%) showed an OS advantage from ICI monotherapy and five studies (25%) showed the same advantage from ICI-chemotherapy compared with chemotherapy.
Two RCTs had unique designs and were needed to warrant further explanation. KEYNOTE-010 was a Phase II/III trial and evaluated two different doses of PEM (2 mg/kg and 10 mg/kg) every 3 weeks, which was considered as two studies-(a) KEYNOTE-010 and (b) KEYNOTE-010.
CheckMate 227 trial explored the efficacy of two NIV-based treatments (NIV plus IPI or NIV) with an additional PBC compared with chemotherapy, which was descried as (a) CheckMate 227 and (b) CheckMate 227 in this study. The main characteristics of the eligible RCTs are shown in Table 1 and the quality of these trails was satisfactory (Table S2). Sensitivity analysis illustrated our results remained stable by omitting trails sequentially ( Figure S2 and S3) and the funnel plot revealed no evidence of publication bias ( Figure S4). nonsquamous: HR, 0.73; p < .001) compared to those treated with chemotherapy. The ORR benefit of squamous lung cancer patients using ICI therapies was marginal (p = .073), while there was a significant improvement of ORR in the subgroup of nonsquamous NSCLC compared with chemotherapy (p < .001; Figure S6).

| Subgroup analysis
3.4.1 | ICI-chemotherapy vs ICI monotherapy Table 2 and Table S3  In nonsmoker, ICI-chemotherapy revealed an OS advantage over ICI

| PEM plus PBC versus PEM monotherapy
The greatest improvement for OS (HR, 0.56) and PFS (HR, 0.59) was achieved in patients treated with anti-PD-1 blocker plus chemotherapy. Moreover, the subgroup analysis revealed that PEM plus PBC maximized the survival benefits in the ICI-chemotherapy subgroup (OS: HR, 0.56; PFS, HR, 0.54; Figure S6). And in the ICI monotherapy subgroup, the minimum risk of death or disease progression was detected in PEM (OS: HR, 0.70; PFS, HR, 0.72; Figure S6) Figure S7).

| DISCUSSION
The pooled analysis, including 20 RCTs of high quality involving 12,025 patients, revealed that ICI therapies were associated with significantly better therapeutic effect across all the efficacy and safety end points than chemotherapy alone in advanced NSCLC populations. Compared with chemotherapy, ICI-chemotherapy resulted in significantly prolonged PFS than ICI monotherapy with significant difference between the two subgroups. Furthermore, PEM plus PBC led to the greatest improvement for OS and PFS than the other treatments, and consequently it is recommended as the optimal first-line option for advanced patients with NSCLC.  (Rizvi et al., 2015). This is in accordance with the observation that nonsynonymous mutation burden is elevated in squamous-cell NSCLC patients (CheckMate 227, a Phase III RCT, focused on patients with NSCLC with a high tumor mutational burden). Then certain somatic mutations increase the burden of neoantigens, which is crucial for the clinical response of PD-1 inhibitors against tumor (Łuksza et al., 2017;Rizvi et al., 2015). Consequently, nonsynonymous mutation burden strongly induces reactivity of T-cell and results in tumor regression in the context of anti-PD-1 therapy (Le et al., 2017;Riaz et al., 2017;Rizvi et al., 2015).
We found that the anti-PD-1 therapy appeared to illicit greater treatment benefits compared with the anti-PD-L1 therapy, which is consistent with the previous hypothesis that NSCLC patients with anti-PD-1 therapy are more likely to experience prolonged survival and a more tolerable safety prolife than anti-PD-L1 therapy Wei et al., 2017;Xu et al., 2018). Theoretically, the interaction between PD-1 and PD-L1 as the dominant ligand, can be inhibited by both PD-L1 blockers and PD-1 blockers. In addition, PD-1 rather than PD-L1 inhibitors can also block the binding of PD-1 to PD-L2, which is 2-6 folds stronger than the affinity of PD-1 binding to PD-L1 (Ribas & Wolchok, 2018). PD-L1 immunotherapy spares the mutual effects between PD-L1 and PD-L2. However, few RCTs are reported to have specifically investigated the role of PD-L2 in the immunotherapy compared with conventional therapy. Further research are needed to understand the role of activation of PD-1 pathway in the antitumor immunity on the whole landscape.
ICI-chemotherapy and ICI monotherapy have never been directly compared in RCTs, partially because sponsors of RCTs are competitive pharmaceutical enterprises. In the presence of statistically significant interaction between ICI-chemotherapy and ICI monotherapy, the current study revealed that there were greater F I G U R E 4 Forest plots of HRs comparing overall survival between ICI therapies and chemotherapy according to PD-L1 status. CI, confidence interval; HR hazard ratio; ICI, immune checkpoint inhibitor; PD-L1, programmed cell death 1 ligand 1 PFS benefits from ICI-chemotherapy among subgroups involving woman, young (<65 years old) patients, never smokers and nonsquamous-lung cancer patients compared with those from ICImonotherapy. Our finding further confirmed that anti-PD-1 blocker plus chemotherapy provided OS and PFS advantages over anti-PD-1 therapy alone with significant difference between the two intervention groups. Previous molecular interactions aside, the observation of synergetic effects in patients with NSCLC treated with anti-PD-1 blocker plus chemotherapy seemed to accord with the hypothesis that chemotherapy may upregulate PD-L1 expression level as well as PD-L1 negative also experienced longer survival from ICI therapy (Borghaei et al., 2015). In the IMpower131 trial, compared with chemotherapy alone, although ATE plus PBC led to statically significant improvements in OS and PFS on the whole, the OS and PFS in the subgroups of PD-L1 <1% or 1-49% did not have significance . As previously discussed, the biological function of PD-1 pathway appears to be more important than PD-L1 status alone in forecasting the prognosis of patients with NSCLC treated with immunotherapy. Another possible explanation is that positive survival benefits for patients with PD-L1 negative were observed from studies of CheckMate 017 (squamous-cell NSCLC), , and KEYNOTE-407 (squamous-cell NSCLC) studies Gandhi et al., 2018;Paz-Ares et al., 2018 Abbreviations: CI, confidence interval; ECOG PS, Eastern Cooperative Oncology Group Performance Status; HR, hazard ratio; ICI, immune checkpoint inhibitor; PD-1, programmed cell death 1; PD-L1, programmed cell death 1 ligand 1; PFS, progression-free survival.
cross study analysis with data from Phase III RCTs of high quality.
However, these results should be interpreted with caution. Third, the indirect methods of comparison require that the enrolled RCTs should be comparable with respect to potential factors of therapeutic effects and the weak heterogeneity across the indirect comparisons indicated that our results were true. Fourth, the toxicity profile is as crucial as survival benefits to determine the optimal treatment choice for patients with NSCLC. Although overall, we took Grade 1-5 and Grade 3-5 TRAEs into account, we could not deal with the issue in the subgroups because data concerning TRAEs of involved populations stratified by PD-L1 status were not available.

| CONCLUSION
In conclusion, for advanced patients with NSCLC, ICI therapies with or without PBC are promising therapeutic options with advantageous survival, clinical response rate and a manageable safety profile than chemotherapy. Furthermore, PEM plus PBC is recommended as the optimal first-line therapy option for patients with NSCLC without targetable genomic mutations. In addition, PD-L1 alone is not recommended as an adequate molecular biomarker to identify eligible patients for routine clinical practice in immunotherapy.

ACKNOWLEDGMENTS
The author thanks the data provided by the authors of included RCTs. This work was supported by grants 91859203, 81871890 from National Natural Science Foundation of China.