Best regimens for treating chemo‐naïve incurable squamous non‐small cell lung cancer with a programmed death‐ligand 1 tumor proportion score of 1%–49%: A network meta‐analysis

Abstract Background Non‐small cell lung cancer (NSCLC) is the leading cause of cancer‐related mortality worldwide. It is advisable to select the appropriate treatment based on characteristics of the cancer such as pathology, mutations, and programmed death‐ligand 1 (PD‐L1) levels. In this study, by remarking squamous NSCLC with low PD‐L1 expression without mutations, we investigated the efficacy and safety of regimens that included molecularly targeted drugs such as immune checkpoint inhibitors (ICIs) through a network meta‐analysis. Methods Databases were searched systematically to identify appropriate articles, in which randomized trials with incurable squamous NSCLC were described. Suitable studies were manually checked by two reviewers. A random model network meta‐analysis was conducted, in which the primary outcome was the overall survival rate. Results We identified 48 studies, which included 16 391 patients. When a platinum + third‐generation cytotoxic agent regimen (platinum regimen) was a reference, the platinum regimen + pembrolizumab (Pemb) yielded the best results in regard to the overall survival rate when compared with chemotherapy (hazard ratio [HR] = 0.57, 95% confidence interval [CI] = 0.36–0.90, p = 0.016) followed by the platinum regimen + nivolumab (Niv) + ipilimumab (Ipi) (HR = 0.61, 95% CI = 0.44–0.84, p = 0.003). However, the efficacy of ICI monotherapy was not statistically different from that of the platinum regimen. Conclusions The combination therapies, which were the platinum regimen + Pemb and the platinum regimen + Niv + Ipi, rather than ICI monotherapy were effective first‐line agents for treating squamous NSCLC with low PD‐L1 levels.

INTRODUCTION Non-small cell lung cancer (NSCLC) is the leading cause of mortality in the world. 1 Platinum doublet chemotherapy is the standard of care for patients diagnosed with inoperable NSCLC lacking sensitizing mutational drivers. Immune checkpoint inhibitors (ICIs) have broadened the options of cancer treatment with a new mechanism that suppresses immune evasion by tumor co-suppressors and activates antitumor immunity. 2 For example, programmed cell death protein and cytotoxic T-lymphocyte-associated antigen 4 are immune checkpoint proteins in lung cancer. 3 According to recent guidelines, drugs that are used for combination therapy or monotherapy are considered for each programmed deathligand 1 (PD-L1) expression and histological type. [4][5][6][7] The expression of PD-L1 is classified as a tumor proportion score (TPS) of 50% or higher, 1%-49%, and <1%. NSCLCs are classified predominantly as adenocarcinoma and squamous cell carcinoma. These two types differ in smoking history, tumor location, and clinical outcomes. 8 Appropriate treatment must be selected based on the characteristics of cancer. Incidentally, kinase inhibitors such as target epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase, B-Raf proto-oncogene, and ROS proto-oncogene 1 are specifically effective in NSCLC with mutations, but these are not recommended for patients who do not have these mutations. 9 The superiority of regimens is often determined in randomized clinical trials (RCTs), however, all regimens, especially ICI regimens, are not compared. Understanding the differences between ICI regimens is important during the process of choosing a treatment. Network meta-analysis is an optimal analytical method that allows for indirect comparisons among multiple regimens. In 2017, we conducted a network meta-analysis of cytotoxic regimens for first-line incurable NSCLC. 10 At this time, remarking squamous NSCLC with low PD-L1 expression (TPS 1%-49%) and no mutations, we investigated the efficacy and safety of regimens including molecularly targeted drugs, such as ICIs, by conducting a network meta-analysis.

Protocol registration
The protocol of our analysis conformed to the Preferred Reporting Items for Systematic Reviews reporting checklist 11 and was registered at the University Hospital Medical Information Network Center, Japan. 12 The contents of the protocol have been partially revised (Text S1). In the systematic review, informed consent of the patient and approval of the institutional review board was not required.

Study search
To identify eligible articles, the MEDLINE, Web of Science Core Collection, Embase, and Cochrane Central Register of Controlled Trials databases were searched systematically on October 15, 2020. The equations used in MEDLINE are listed in Text S2. The titles and abstracts of the shortlisted papers were screened, and those that met the inclusion criteria were reviewed in the full text. The suitable studies were manually checked by two review authors (N.F. and K. S.). Any dispute in the selection process was discussed and resolved by the two reviewers (NF and N.H.).

Selection criteria: publication type and trial design
The published sources had to be written in English. RCTs that enrolled patients with advanced NSCLC were included in this study. Trials that focused on NSCLC with oncogenic mutations were excluded. Conference abstracts were allowed. Studies that included treatment only in the form of first-line chemotherapy were included. Studies that included maintenance therapy, second-line therapy, and later-line therapy were excluded.

Selection criteria: chemotherapy
Appropriate treatments included platinum doublet chemotherapy, ICIs, and molecularly targeted therapies. Clinical studies on platinum plus an angiogenesis inhibitor have also been conducted. ICI can be administered alone or in combination with platinum-based treatments.
There is a controversy regarding the difference between cisplatin (CDDP) and carboplatin (CBDCA), 13,14 but in recent clinical trials, physicians can choose either CDDP or CBDCA. 15,16 Keynote 407, which is one of the material RCTs related to pembrolizumab (Pemb), 17 paclitaxel (Ptx), and nanoparticle albumin-bound Ptx (nabPtx), was considered similarly. For this reason, CDDP/CBDCA and Ptx/nabPtx were considered identical during analysis. Nedaplatin expression was also differentiated.
A network meta-analysis based on individual regimens cannot evaluate the physicians' choice regimens. Two models were established to assay these items. The "main model" did not make a distinction between CDDP or CBDCA plus third-generation cytotoxic drug regimen (platinum regimen). The "separate model" evaluated each platinum regimen differently.
Because the use of kinase inhibitors in patients without driver-gene mutations is not standardized, they were excluded from the study. [4][5][6][7] Trials that examined pre-or post-operative chemotherapy and concomitant chemoradiation were also excluded from this analysis.

Selection criteria: patients
RCTs enrolled patients with advanced or recurrent squamous NSCLC. Regardless of whether classification of the tumor-lesion-metastasis was revised in the past, we adopted the stage described in the article. Patients with a history of surgery or radiotherapy were included in the study. Studies that focused on those with a low performance status or the elderly were excluded. NSCLCs with a low or high PD-L1 level (<1% or >50%) were excluded because recent guidelines have used the cutoff of PD-L1 levels at 1% or 50%. [4][5][6][7] If a portion of the surveyed population met our criteria, we analyzed that data subset. If a study provided data for three separate populations with TPSs of 0%, 1%-49%, and ≥50%, the data for those with a TPS of 1%-49% were used.
Trials focusing on non-squamous carcinoma, oncogenic mutations, and patients with a TPS ≥50% were excluded. However, we accepted studies that did not clearly describe the pathological classification, oncogenic mutations, and PD-L1 expression. This was necessary, because the majority of studies related to NSCLC would have been excluded otherwise. It has been reported that there are differences in PD-L1 assay positivity. 18 We thought that these differences were slight and they could be taken into consideration.

Quality assessment
The quality of each incorporated study was checked using the Cochrane risk of bias tool. This judgment tool evaluated selection, performance, detection, attrition, reporting, and other biases. 19

Outcomes
The primary outcome was overall survival (OS); the secondary endpoints were progression-free survival (PFS), adverse events based on the Common Terminology Criteria for Adverse Events grade III or higher, 20 and treatment-related death. Hazard ratios (HRs) were used to evaluate the OS and the PFS. Odds ratios (ORs) were used to evaluate adverse events (ORae) (>grade III) and treatment-related death. Disease progression was determined according to the Response Evaluation Criteria in Solid Tumors guidelines issued in 2000, which were revised in 2009. 21 If possible, imaging evaluations performed by an independent reviewer were preferred.

Data extraction
The data of studies including the first author name, publication year, phase, primary outcome, pathology, stage, performance status, regimens, sample size (if possible, intention-to-treat analysis), and trial outcomes were independently extracted by two reviewers (N.F. andK.S.). Discrepancies were resolved through mediation by a third reviewer (N.H.). Parmar's method was used to analyze survival data. 22 The regimens were defined according to the drug compounding independently of the dose and schedule. The subset data were collected by subtraction using a fixedmodel meta-analysis. The results of the "PD-L1 1% to 49%" subgroup were calculated by subtracting data of the "PD-L1 <1%" subgroup from data of the "PD-L1 <50%" subgroup. Data on adverse events and treatment-related deaths could be obtained from NSCLC with pathology or any PD-L1 TPS because stratified data were seldom noted.

Statistical analyses
Our study applied the frequentist weighted least-squares approach random-model network meta-analysis. 23 The OR and HR were calculated using log-conversion. Platinum + Ptx was used as a common reference in the separate model because this regimen was recently evaluated in trials for the treatment of squamous carcinoma. 17 The results were F I G U R E 1 Preferred reporting items for systematic reviews and meta-analyses flow chart for study search. RCT, randomized clinical trials

Study search
There were 1386 articles retrieved by electronic and hand searches. There were 897 articles that met the inclusion criteria. There were 489 articles removed because they were duplicate studies; 719 articles were removed during the process of screening title/abstract, and 130 articles were removed after reviewing the full article. Finally, 48 appropriate studies were identified (Figure 1; Supplementary References).  (Table S1).

Safety analysis
Safety was considered in regard to grade III adverse events and chemotherapy-related deaths. Pemb had a significantly lower risk of grade III adverse events than did the platinum regimen (OR = 0.20, 95% CI = 0.09-0.42, p < 0.001), followed by Niv and durvalumab (Figure 3(c),(d)). The addition of ICI to the platinum regimen did not result in a statistically higher risk compared to standard chemotherapy. The regimens were not significantly different for chemotherapy-related deaths.

DISCUSSION
We conducted a network meta-analysis to compare regimens for chemo-naïve incurable squamous lung cancer with a PD-L1 TPS of 1%-49%. In addition to cytotoxic regimens and ICIs, molecularly targeted therapies were also integrated. The analysis showed that the combination of platinum regimen and ICIs was the best regimen in terms of OS and PFS, without considerably increasing the risk of adverse events. The high statistical power acquired by aggregating RCTs increases the certainty and accuracy of the results.
The platinum regimen is reported to enhance the activity of immunotherapy even though the tumor microenvironment is non-immunogenic. 25 The death of cancer cells triggers phagocytosis and promotes the maturation of dendritic cells, leading to tumor eradication. 26 Therefore, combination therapy is reasonable. The main models in our study indicated that HRos of the platinum regimen + Pemb significantly decreased when compared with those of the platinum regimen (HR = 0.57, 95% CI = 0.36-0.90); this trend was followed by HRpfs of the platinum regimen + Pemb (HR = 0.56, 95% CI = 0.32-0.97) (Figure 3(a),(b)). ICI monotherapy has shown great efficacy in some patients, leading to long-term responses. 27 Because the OS curve of Niv + Ipi crossed that of chemotherapy in the early stage of CheckMate 227, 16 patients receiving ICI monotherapy do not receive the benefits that come with early disease control. When chemotherapy is combined with ICIs, the possibility of rapidly controlling the initial disease increases. There is concern about the adverse effects of combination therapy with ICIs. In our analysis, the ORae of the platinum regimen + Pemb (OR = 1.08, 95% CI = 0.38-3.07) and that of the platinum regimen + Niv + Ipi (OR = 1.24, 95% CI = 0.42-3.66) were not significantly higher than that of the platinum regimen alone (Figure 3(c)). ICI monotherapy, for example, Pemb (ORae = 0.20, 95% CI = 0.09-0.42), has also shown to be safer than the platinum regimens. However, the HRos of ICI monotherapy were not statistically different from those of the platinum regimen. Therefore, we recommend combination therapy rather than ICI monotherapy for patients with a PD-L1 TPS of 1%-49%. ICI monotherapy can be used when patients are intolerant to cytotoxic regimens.
In our study, combination therapies with drug exclusive ICIs were analyzed. The platinum regimen + Nctm was better than the platinum regimen alone in terms of OS (HR = 0.82, 95% CI = 0.73-0.92) (Figure 3(a)). In the separate model, the platinum regimen + gemcitabine + Nctm indicated the same result (HR = 0.79, 95% CI = 0.69-0.92) ( Figure S3). Nctm is a monoclonal antibody against EGFR and has been reported to be effective in patients with high EGFR or squamous cancer scores. Adverse events such as skin rash, venous thromboembolism, and eye disorders should be noted. 28 For squamous lung cancer with PD-L1 TPS of 1%-49%, platinum + Atz is recommended by the guidelines of the American Society of Clinical Oncology, the National Comprehensive Cancer Network, and the European Society for Medical Oncology. However, the results of Impower 130, 29 which compared platinum + Ptx + bevacizumab + Atz and platinum + Ptx + bevacizumab against squamous cancer, did not show any effect on OS. In our study, Figure 3(a),(b) indicated that HRos (HR = 1.08, 95% CI = 0.81-1.44) and HRpfs (HR = 0.70, 95% CI = 0.43-1.14) of the platinum regimen + Atz were not significantly different from those of the platinum regimen alone.
Our analysis had some limitations. First, although there were differences in the drug, amount, and period, the platinum regimen in the main model was identified to be the same. Complementarily, we conducted a separate model, and the result was similar to that of the main model. Second, according to the Cochrane tool criteria, all incorporated papers had a high risk of bias. Unfortunately, it is difficult to conduct a double-blind study without a sponsor. We believe that these factors did not significantly reduce the reliability of this study.
In summary, we performed a systematic review and network meta-analysis of patients with squamous NSCLC with a PD-L1 TPS of 1%-49%. For the 16 391 patients diagnosed with NSCLC and part of 48 RCTs, the platinum regimen + Pemb and the platinum regimen + Niv + Ipi were considered appropriate first-line agents for treating squamous NSCLC with low PD-L1.