Optimal drug regimens for squamous non-small cell lung cancer: a systematic review and a Bayesian network meta-analysis of randomized trials

Abstract

Background

Current treatment options for patients with squamous non-small cell lung cancer (sqNSCLC) include immune checkpoint inhibitors combined with chemotherapy, monotherapy with immune therapy, and platinum-based doublet chemotherapy, among others. However, there is still a lack of systematic comparisons regarding the relative efficacy and safety of these regimens. This study aims to comprehensively assess the clinical benefits [such as overall survival (OS), progression-free survival (PFS)] and adverse event rates across different treatment strategies via a network meta-analysis, providing evidence-based guidance for individualized treatment decisions in sqNSCLC patients.

Methods

We searched PubMed, Embase, Cochrane Library, and Web of Science from inception to May 3, 2025 for eligible randomized controlled trials (RCTs). The primary outcome measures were PFS and OS; secondary outcomes included objective response rate (ORR) and grade ≥3 treatment-related adverse events (≥3 TRAEs). A network meta-analysis was carried out using Stata 16.0 and R 4.4.0.

Results

A total of 32 RCTs, involving 9,652 participants, were included. This study analyzed 18 different treatment strategies. The network meta-analysis showed that for PFS, paclitaxel-based doublet chemotherapy combined with programmed cell death-1 and programmed death-ligand 1 [PD-(L)1] checkpoint inhibitors [surface under the cumulative ranking curve (SUCRA): 96.52%] and gemcitabine-based doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors (SUCRA: 85.16%) were the most effective. For OS, paclitaxel-based doublet chemotherapy combined with a Toll-like receptor-2 agonist (CADI-05) (SUCRA: 92.11%), paclitaxel-based doublet chemotherapy combined with PD-(L)1 and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) checkpoint inhibitors (SUCRA: 89.65%), and paclitaxel-based doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors (SUCRA: 83.58%) were the top three choices. The combination of paclitaxel-based doublet chemotherapy with PD-(L)1 checkpoint inhibitors (SUCRA: 87.00%) was associated with the best improvement in ORR.

Conclusions

For first-line treatment of sqNSCLC, the current evidence supports the use of paclitaxel-based doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors as one of the preferred treatment options. However, more high-quality trials are needed to confirm these findings.

Highlight box.

Key findings

• Paclitaxel-platinum chemotherapy plus programmed cell death-1 and programmed death-ligand 1 [PD-(L)1] blockade ranked best for progression-free survival (PFS) [surface under the cumulative ranking curve (SUCRA): 96.52%] and objective response rate (SUCRA: 87.00%).

• For overall survival, paclitaxel-platinum plus a Toll-like receptor 2 agonist (CADI-05) ranked highest (SUCRA: 92.11%), followed by dual immune checkpoint inhibition [PD-(L)1 + cytotoxic T-lymphocyte-associated protein 4 (CTLA-4)] plus paclitaxel-platinum (SUCRA: 89.65%).

What is known and what is new?

• Immune checkpoint inhibitors combined with chemotherapy or alone are standard first-line treatments for squamous non-small cell lung cancer (sqNSCLC). However, systematic comparisons across multiple regimens, including targeted therapies and antibody-drug conjugates (ADCs), have been limited.

• This is the first network meta-analysis simultaneously comparing 18 treatment strategies, including paclitaxel- and gemcitabine-based doublets, single-agent immunotherapy, dual checkpoint inhibition, ADCs, and other targeted agents. It provides novel rankings for efficacy and safety, highlighting the superior performance of paclitaxel-platinum plus PD-(L)1 blockade for PFS and response, and identifying promising regimens such as CADI-05 combinations.

What is the implication, and what should change now?

• Paclitaxel-platinum doublet chemotherapy combined with PD-(L)1 inhibitors should be considered a preferred first-line option for sqNSCLC based on current evidence.

• Regimens like CADI-05 warrant further high-quality trials before clinical adoption.

Introduction

Non-small cell lung cancer (NSCLC) accounts for approximately 85% of all primary lung cancers, with squamous cell carcinoma (SCC) comprising 20% to 30% of NSCLC cases. In contrast to adenocarcinoma, SCC exhibits rarely exhibits driver gene mutations, such as EGFR and ALK, but shows a higher frequency of programmed death-ligand 1 (PD-L1) overexpression, which is strongly correlated with tobacco exposure (1). These unique biological characteristics limit the application of targeted therapies in SCC, making traditional platinum-based doublet chemotherapy the long-standing standard for first-line treatment (2).

In recent years, the emergence of immune checkpoint inhibitors (ICIs) has revolutionized the treatment landscape. Whether as a monotherapy or in combination with chemotherapy, ICIs are now recommended as first-line standard treatment for squamous NSCLC (sqNSCLC) patients (3,4). Programmed cell death-1 and programmed death-ligand 1 [PD-(L)1] checkpoint inhibitors have greatly improved survival in sqNSCLC patients, particularly those with high PD-(L)1 expression (5). Additionally, the combination of immunotherapy and chemotherapy has been shown to be more effective than chemotherapy alone in first-line treatment for advanced or metastatic sqNSCLC patients (3,4,6-10). With the successful completion of the KEYNOTE-407 trial, pembrolizumab combined with chemotherapy was approved as the first anti-PD-1 regimen for treatment-naïve advanced sqNSCLC patients, regardless of PD-(L)1 expression levels (11). However, not all patients are suitable for ICIs, and histology-directed chemotherapy continues to play an important role in the treatment of sqNSCLC. Currently, first-line chemotherapy regimens for sqNSCLC include cisplatin or carboplatin combined with cytotoxic agents such as gemcitabine, paclitaxel, and docetaxel. However, the median overall survival (OS) is typically less than 12 months, and patients often face significant adverse events during treatment, such as bone marrow suppression and gastrointestinal symptoms (12-14).

Beyond the treatment options mentioned, there has been growing interest in exploring additional therapies to improve survival outcomes in sqNSCLC. Notably, combinations of chemotherapy with other agents are gaining attention, such as the second-generation recombinant human IgG1 EGFR antibody necitumumab, the human IgG1 monoclonal antibody targeting VEGFR-2 extracellular domain, ramucirumab, and the small molecule VEGFR inhibitor motesanib (15-18).

Despite the diversity of treatment strategies available for sqNSCLC, there are limited systematic comparisons of the relative efficacy and safety of different regimens. Existing network meta-analyses (NMAs) have mainly focused on comparing regimens with PD-(L)1 inhibitors combined with chemotherapy or specific patient populations (e.g., those with PD-L1 expression ≥50%) (19,20). In this study, we conducted the first NMA that simultaneously included multiple regimens, including immunotherapy combined with chemotherapy, single-agent immunotherapy, targeted therapy, and antibody-drug conjugates (ADCs), aiming to provide a more comprehensive evaluation of efficacy and offer evidence-based guidance for personalized treatment decisions in sqNSCLC patients, in order to inform clinical practice. We present this article in accordance with the PRISMA-NMA reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1-0152/rc).

Methods

The study protocol has been registered on the international prospective systematic review registry platform, Prospero (CRD420251103523). Since only publicly available published data were analyzed, patient consent and ethical approval from committees were not required.

Literature search

The search strategy combined subject headings and free-text terms and was conducted in four electronic databases: PubMed, Web of Science, Cochrane Library, and Embase. The search period spanned from the inception of each database to May 3, 2025. Additionally, to include comprehensive and updated results, reference lists of relevant articles were reviewed. To ensure the completeness and reliability of the evidence base for the systematic review, conference abstracts, clinical trial registrations, and similar documents were excluded. The detailed search strategy is available in Appendix 1.

Inclusion and exclusion criteria

Inclusion criteria were as follows: (I) participants: adults with histologically confirmed stage III–IV or recurrent sqNSCLC, irrespective of prior systemic therapy; (II) interventions and comparators: chemotherapy, immunotherapy, or targeted agents alone or in combination, with the control arm receiving chemotherapy alone or another active therapy; (III) study design: randomized controlled trial (RCT); (IV) outcomes: at least one of the following reported—progression‑free survival (PFS) (time from randomization to first progression or death), OS (time from randomization to death from any cause), objective response rate (ORR) (proportion of patients achieving tumor response), or incidence of grade ≥3 treatment-related adverse events (≥3 TRAEs).

Exclusion criteria were as follows: (I) the investigational drug has been discontinued and is not available in any country; (II) animal or cell experiments, case reports, scientific experiment plans, reviews, letters, editorials, conference papers, etc.; (II) literature with missing or seriously erroneous research data; (IV) duplicate publications; (V) studies with no full-text available.

Two reviewers (Y.Y. and L.S.) independently assessed the titles and abstracts according to the inclusion and exclusion criteria and retrieved the full-text articles to screen for eligibility. Disagreements were resolved through discussion or consultation with a third reviewer (Y.L.).

Literature screening and data extraction

Two reviewers (L.S. and Y.Y.) conducted the literature search strictly following the inclusion and exclusion criteria. All references were managed using Endnote X9 software. After excluding duplicate publications, studies that were preliminarily eligible based on titles or abstracts were retrieved for full-text screening. Full-text articles were then assessed for inclusion in the systematic review. Basic information from each study was extracted and cross-verified, and consistent units of measurement were applied. In cases of disagreement, decisions were made through discussion with a third reviewer (Y.L.). Data extracted from the studies included: (I) basic information: first author, publication year, country; (II) study characteristics: cancer stage, histological characteristics, sample size, treatment lines, interventions, and control measures; (III) outcome measures: reported clinical outcomes.

Quality assessment of included studies

Two reviewers independently assessed the literature quality. Disagreements were resolved by a third reviewer, and the final decision was reached through discussion. The Cochrane risk-of-bias tool (RoB2) (21) was used to evaluate the quality of the included studies. The evaluation focused on five aspects: bias arising from randomization, bias due to deviations from intended interventions, bias from missing outcome data, bias in outcome measurement, and bias in the selection of reported results. These aspects were classified into three categories: low risk, unclear, and high risk, with the results presented in a risk-of-bias graph (Figure S1).

Statistical analysis

Statistical analysis was carried out using R version 4.4.0. Hazard ratios (HRs) were used for survival endpoints, and relative risks (RRs) for binary endpoints, each reported with 95% confidence intervals (CIs). A difference was deemed statistically significant when the 95% CI excluded 1. Heterogeneity was interpreted using I² values of 25%, 50%, and 75% as thresholds for low, moderate, and high heterogeneity, respectively (22). A Bayesian NMA was undertaken via the Markov Chain Monte Carlo (MCMC) method (23,24). Four Markov chains were run, with a thinning interval set to 1. The first 20,000 iterations served as burn-in, followed by 50,000 iterations to achieve convergence. Model fit and global consistency were compared using the Deviance Information Criterion (DIC); if the absolute DIC difference between consistency and inconsistency models was less than 3, the consistency model was adopted. Local consistency was assessed with the nodesplitting method for closed loops within the network (25).

A random-effects model was applied to evaluate all treatment strategies. Outcome measures were presented as network plots, cumulative probability ranking plots, and league tables (26). The surface under the cumulative ranking curve (SUCRA) quantified the ranking probability, where larger SUCRA values indicated better intervention performance (27). Network graphs illustrating both direct and indirect comparisons were generated using Stata 16.0. All analyses were conducted with Stata 16.0 and R 4.4.0. Publication bias was evaluated via funnel plots, and the significance threshold was set at P<0.05.

Results

Literature search process

A total of 7,889 articles were initially retrieved. After removing 1,531 duplicates, 6,128 articles were excluded based on preliminary screening of titles and abstracts. The remaining 230 articles underwent full-text review, and strict inclusion and exclusion criteria were applied, resulting in the inclusion of 32 studies. The specific screening process is shown in Figure 1 (3,4,6-9,28-45). A total of 9,652 patients were included in the analysis, receiving 18 different treatment regimens, including: taxane-platinum doublet chemotherapy (Taxane-Pt); taxane-platinum doublet chemotherapy combined with veliparib (Taxane-Pt-Veliparib); taxane-platinum doublet chemotherapy combined with the vascular disrupting agent vadimezan (Taxane-Pt-Vadimezan); taxane-platinum doublet chemotherapy combined with the a Toll-like receptor-2 agonist (CADI-05) (Taxane-Pt-CADI-05); taxane-platinum doublet chemotherapy combined with motesanib (Taxane-Pt-Motesanib); taxane-platinum doublet chemotherapy combined with necitumumab (Taxane-Pt-Necitumumab); taxane-platinum doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors (Taxane-Pt-PD1orPDL1); taxane-platinum doublet chemotherapy combined with both PD-(L)1 and cytotoxic T-lymphocyte‑associated protein 4 (CTLA-4) checkpoint inhibitors (Taxane-Pt-PD1orPDL1-CTLA-4); single-agent taxane chemotherapy (only including docetaxel); taxane monotherapy combined with ramucirumab (Taxane-Ramucirumab); gemcitabine-platinum chemotherapy (Gemcitabine-Pt); gemcitabine-platinum combined with PD-(L)1 checkpoint inhibitors (Gemcitabine-Pt-PD1orPDL1); gemcitabine-platinum combined with necitumumab (Gemcitabine-Pt-Necitumumab); gemcitabine combined with taxanes (Gemcitabine-Taxane); single-agent PD-(L)1 checkpoint inhibitors; PD-(L)1 combined with CTLA-4 dual checkpoint inhibitors; vinorelbine-platinum (Vinorelbine-Pt); and ADCs. Table 1 provides detailed characteristics of the included studies.

Figure 1.

PRISMA flow diagram for literature search and selection.

Table 1. Characteristics of the included studies.

First author	Year	Registered number	Line of treatment	Country	Intervention modalities			Number of cases		Outcome
					Treatment 1	Treatment 2		Treatment 1	Treatment 2
Lara PN	2011	Unknown	1	Multi-countries	Paclitaxel 200 mg/m2 D1 plus carboplatin AUC 6.0 D1 plus vadimezan (ASA404) 1,800 mg/m2 D1, Q3W	Paclitaxel 200 mg/m2 D1 plus carboplatin AUC 6.0 D1 plus placebo 1,800 mg/m2 D1, Q3W		132	133	OS
Garon EB	2014	NCT01168973	2	Multi-countries	Docetaxel 75 mg/m2 D1 plus ramucirumab 10 mg/kg D1, Q3W	Docetaxel 75 mg/m2 D1 plus placebo 10 mg/kg D1, Q3W		157	171	PFS, OS, ORR
Novello S	2014	NCT00460317	1	Multi-countries	Motesanib 125 mg QD plus carboplatin AUC 6.0 D1 plus paclitaxel 200 mg/m2 D1 Q3W	Placebo 125 mg QD plus carboplatin AUC 6.0 D1 plus paclitaxel 200 mg/m2 D1 Q3W		182	178	OS, PFS, ORR, AEs
Thatcher N	2015	NCT00981058	1	Multi-countries	Gemcitabine 1,250 mg/m2 D1, 8 plus cisplatin 75 mg/m2 necitumumab 800 mg D1, 8, Q3W	Gemcitabine 1,250 mg/m2 D1, 8 plus cisplatin 75 mg/m2, Q3W		545	548	OS, PFS, ORR, TTF
Belani CP	2017	CTRI/2007/091/000037	1	India	TLR-2 agonist (CADI-05) 0.1 mL D8, 15 plus paclitaxel 175 mg/m2 D1 plus cisplatin 100 mg/m2 D1, Q3W	Paclitaxel 175 mg/m2 D1 plus cisplatin 100 mg/m2 D1, Q3W		42	34	OS
Lee SM	2017	NCT00801736	1	United Kingdom	Paclitaxel 175 mg/m2 D1 plus gemcitabine 1,250 mg/m2 D1, 8, Q3W	Cisplatin 75 mg/m2 D1 plus gemcitabine 1,250 mg/m2 D1, 8, Q3W		89	84	OS, PFS
Ramalingam SS	2017	NCT01560104	1	Multi-countries	Veliparib 120 mg BID D1–7 plus carboplatin AUC 6.0 D3 plus paclitaxel 200 mg/m2 D3, Q3W	Placebo 120 mg BID D1–7 plus carboplatin AUC 6.0 D3 plus paclitaxel 200 mg/m2 D3, Q3W		51	25	PFS, OS
Rittmeyer A	2017	NCT02008227	Subsequent-line (≥2 lines)	Multi-countries	Atezolizumab 1,200 mg, Q3W	Docetaxel 75 mg/m2, Q3W		112	110	OS
Spigel DR	2017	Unknown	1	Multi-countries	Necitumumab 800 mg D1, 8 plus paclitaxel 200 mg/m2 D1 plus carboplatin AUC 6.0 D1, Q3W	Paclitaxel 200 mg/m2 D1 plus carboplatin AUC 6.0 D1, Q3W		110	57	ORR, OS, PFS, DCR, safety
Wang Z	2019	NCT01236716	1	China	Nab-paclitaxel 135 mg/m2 D1, 8 plus carboplatin AUC 5.0 D1, Q3W	Gemcitabine 1,250 mg/m2 D1, 8 plus carboplatin AUC 5.0 D1, Q3W		60	64	ORR, OS, PFS, AEs
Watanabe S	2019	NCT01763788	1	Japan	Gemcitabine 1,250 mg/m2 D1, 8 plus cisplatin 75 mg/m2 D1 plus necitumumab 800 mg D1, Q3W	Gemcitabine 1,250 mg/m2 D1, 8 plus cisplatin 75 mg/m2 D1, Q3W		90	91	PFS, OS, ORR, TTF
Wu YL	2019	NCT02613507	2	Multi-countries	Nivolumab 3 mg/kg, Q2W	Docetaxel 75 mg/m2, Q3W		133	67	OS, PFS, ORR
Jotte R	2020	NCT02367794	1	Multi-countries	Atezolizumab 1,200 mg D1 plus carboplatin AUC 6.0 D1 plus nab-paclitaxel 100 mg/m2 D1, 8, 15, Q3W	Carboplatin AUC 6.0 D1 plus nab-paclitaxel 100 mg/m2 D1, 8, 15, Q3W		343	340	PFS, OS, ORR
Borghaei H	2021	NCT01642004	2	Multi-countries	Nivolumab 3 mg/kg, Q2W	Docetaxel 75 mg/m2, Q3W		135	137	OS, PFS, safety
Gettinger SN	2021	NCT02785952	1, 2, 3	United States	Nivolumab 3 mg/kg D1 Q2W plus ipilimumab 1 mg/kg D1 Q6W	Nivolumab 3 mg/kg D1 Q2W		125	127	OS, PFS
Grossi F	2021	EudraCT number 2012-003531-40	1	Multi-countries	Oral vinorelbine 60-80 mg/m2 D1, 8 plus cisplatin 80 mg/m2 D1, Q3W	Gemcitabine 1,250 mg/m2 D1, 8 plus cisplatin 80 mg/m2 D1, Q3W		57	56	DCR, PFS, OS, TTF, safety
Kawashima Y	2021	UMIN000010443	1	Japan	Cisplatin 80 mg/m2 D1 plus gemcitabine 1,000 mg/m2 D1, 8, Q3W	Carboplatin AUC 6.0 D1 plus nab-paclitaxel 75 mg/m2 D1, 8, Q3W		35	36	ORR, PFS, OS, DCR toxicities
Kogure Y	2021	UMIN000019843, jRCTs041180110	1	Japan	Carboplatin AUC 6.0 D1 plus nab-paclitaxel 100 mg/m2 D1, Q3W	Docetaxel 60 mg/m2, Q3W		95	95	OS, PFS, ORR, AEs
Ramalingam SS	2021	NCT02106546	1	Multi-countries	Veliparib 120 mg BID D(-2)-5 plus carboplatin AUC 6.0 D1 plus paclitaxel 200 mg/m2 D1, Q3W	Placebo 120 mg BID D−2 to D5 plus carboplatin AUC 6.0 D1 plus paclitaxel 200 mg/m2 D1, Q3W		486	484	OS, PFS, ORR
Zhou C	2021	NCT03629925	1	China	Sintilimab 200 mg D1 plus gemcitabine 1.0g/m2 D1, 8 plus platinum (cisplatin 5 mg/m2 D1 or carboplatin AUC 5.0 D1), Q3W	Placebo 200 mg D1 plus gemcitabine 1.0 g/m2 D1, 8 plus platinum (cisplatin 5 mg/m2 D1 or carboplatin AUC 5.0 D1), Q3W		179	178	PFS
Ren S	2022	NCT03668496	1	China	Camrelizumab 200 mg D1 plus carboplatin AUC 5.0 D1 plus paclitaxel 175 mg/m2 D1, Q3W	Placebo 200 mg D1 plus carboplatin AUC 5.0 D1 plus paclitaxel 175 mg/m2 D1, Q3W		193	196	PFS, OS, ORR
Shi YK	2022	NCT03150875	2	China	Sintilimab 200 mg, Q3W	Docetaxel 60 mg/m2, Q3W		145	135	OS, PFS, ORR, DCR
Zhang J	2022	NCT02996214	1	China	Paclitaxel liposome (Lipusu) 175 mg/m2 D1 plus cisplatin D1 75 mg/m2, Q3W	Gemcitabine 1,000 mg/m2 D1, 8 plus cisplatin 75 mg/m2 D1, Q3W		265	269	PFS, OS, ORR, DCR
Wang Z	2023	NCT03856411	1	China	Toripalimab 240 mg plus nab-paclitaxel 100 mg/m2 D1, 8, 15 plus carboplatin AUC 5.0, Q3W	Placebo 240 mg plus nab-paclitaxel 100 mg/m2 D1, 8, 15 plus carboplatin AUC 5.0, Q3W		147	73	PFS
Zhong J	2024									OS
Paz-Ares LG	2024	NCT05089734	Subsequent-line (≥2 lines)	Multi-countries	Sacituzumab govitecan 10 mg/kg D1, 8, Q3W	Docetaxel 75 mg/m2, Q3W		84	80	OS, PFS
Reck M	2024	NCT03215706	1	Multi-countries	Nivolumab 360 mg Q3W plus ipilimumab 1 mg/kg Q6W plus carboplatin AUC 6.0 Q3W plus paclitaxel 200 mg/m2 Q3W (two cycles of maintenance chemotherapy)	Carboplatin AUC 6.0 plus paclitaxel 200 mg/m2, Q3W (four cycles of maintenance chemotherapy)		115	112	OS, PFS
Wang J	2024	NCT03594747	1	China	Tislelizumab 200 mg D1 plus paclitaxel 175 mg/m2 D1 plus carboplatin AUC 5.0, D1, Q3W	Paclitaxel 175 mg/m2 D1 plus carboplatin AUC 5.0, D1, Q3W		120	121	PFS, OS, ORR
					Tislelizumab 200 mg D1 plus nab-paclitaxel 100 mg/m2 D1, 8, 15 plus carboplatin AUC 5.0, D1, Q3W	Paclitaxel 175 mg/m2 D1 plus carboplatin AUC 5.0, D1, Q3W		119	121
Zhong H	2024	NCT03866993	1	China	Penpulimab 200 mg D1 plus paclitaxel 175 mg/m2 D1 plus carboplatin AUC 5.0 D1, Q3W	Placebo 200 mg D1 plus paclitaxel 175 mg/m2 D1 plus carboplatin AUC 5.0 D1, Q3W		175	175	PFS, OS, ORR, DCR
Zhou C	2024	NCT04033354	1	Multi-countries	Serlulimab 4.5 mg/kg D1 plus nab-paclitaxel 100 mg/m2 D1, 8, 15 plus carboplatin AUC 5 or 6 D1, Q3W	Placebo 4.5 mg/kg D1 plus nab-paclitaxel 100 mg/m2 D1, 8, 15 plus carboplatin AUC 5 or 6 D1, Q3W		358	179	PFS, OS, ORR
Ahn MJ	2025	NCT04656652	Subsequent-line (≥2 lines)	Multi-countries	Datopotamab deruxtecan (Dato-DXd) 6 mg/kg, Q3W	Docetaxel 75 mg/m2, Q3W		65	71	PFS, OS, ORR
Zhou C	2025	NCT03789604	1	China	Sugemalimab 1,200 mg D1 carboplatin AUC 5.0 plus paclitaxel 175 mg/m2 D1, Q3W	Placebo 1,200 mg D1 carboplatin AUC 5.0 plus paclitaxel 175 mg/m2 D1, Q3W		129	63	PFS, OS

AEs, adverse events; AUC, area under the curve; BID, twice a day; D, day; DCR, disease control rate; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; Q2W, every 2 weeks; Q3W, every 3 weeks; Q6W, every 6 weeks; QS, quality score; TTF, time to treatment failure.

Quality assessment

The results of the risk of bias assessment for the 32 included studies are shown in Figure S1. All 32 studies were RCTs. Of these, 10 studies were of high quality, with low risk of bias in random sequence generation and blinding. Additionally, allocation concealment, completeness of outcome data, selective reporting, and other biases were controlled for at low risk. Four studies did not report allocation concealment, and eight studies were open-label, making blinding infeasible. The remaining 10 studies were both open-label and did not report allocation concealment, thus were considered to have a high risk of bias.

Pairwise meta-analysis of included trials

Through head-to-head meta-analysis, we evaluated the efficacy of various treatment regimens compared to taxane-platinum doublet chemotherapy in patients with sqNSCLC, with OS and PFS as outcome measures (Figure 2). Taxane-platinum doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors significantly improved PFS (HR =0.47, 95% CI: 0.39, 0.57) and OS (HR =0.72, 95% CI: 0.62, 0.84). Additionally, the combination of taxane-platinum doublet chemotherapy with PD-(L)1 and CTLA-4 checkpoint inhibitors significantly improved OS (HR =0.63, 95% CI: 0.42, 0.95).

Figure 2.

Forest plot illustrating the PFS (A) and OS (B) outcome measure. ADC, antibody-drug conjugate; CADI-05, Toll-like receptor-2 agonist; CI, confidence interval; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; OS, overall survival; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; Pt, platinum.

Network analysis results

Network evidence plots

Figure 3 presents the NMA plots for efficacy and safety outcomes in sqNSCLC patients, including PFS (Figure 3A), OS (Figure 3B), ORR (Figure 3C), and ≥3 grade TRAE (Figure 3D). The size of each node represents the number of participants for that intervention type, and the thickness of the lines between interventions indicates the number of studies comparing those interventions. Pt refers to platinum-based chemotherapy.

Figure 3.

Network diagram. (A) PFS; (B) OS; (C) ORR; (D) the ≥ grade 3 TRAE. ADC, antibody-drug conjugate; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; ORR, objective response rate; OS, overall survival; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; Pt, platinum; TRAE, treatment-related adverse event.

PFS

A total of 25 studies reported the PFS outcomes for 15 different interventions (3,4,6,8-10,12-18,30,31,33,35,39-45). The network structure for different interventions is shown in Figure 3A. The NMA results indicated that most treatment regimens performed better than ADCs in improving PFS. Specifically, gemcitabine combined with platinum and Necitumumab (HR =0.42, 95% CI: 0.21, 0.82) and gemcitabine combined with platinum and PD-(L)1 checkpoint inhibitors (HR =0.32, 95% CI: 0.15, 0.68) showed superior efficacy in improving OS when compared to gemcitabine combined with taxane. Furthermore, gemcitabine combined with platinum chemotherapy (HR =0.42, 95% CI: 0.21, 0.82), gemcitabine combined with platinum and PD-(L)1 checkpoint inhibitors (HR =0.25, 95% CI: 0.11, 0.59), gemcitabine combined with platinum and necitumumab (HR =0.33, 95% CI: 0.15, 0.73), PD-(L)1 checkpoint inhibitors (HR =0.56, 95% CI: 0.39, 0.83), and PD-(L)1 combined with CTLA-4 dual checkpoint inhibitors (HR =0.45, 95% CI: 0.23, 0.87) were all more effective than taxane monotherapy. The combination of taxane-platinum doublet chemotherapy with PD-(L)1 checkpoint inhibitors outperformed taxane-platinum doublet chemotherapy combined with veliparib (HR =0.58, 95% CI: 0.37, 0.94). Additionally, gemcitabine combined with platinum and PD-(L)1 checkpoint inhibitors (HR =0.33, 95% CI: 0.12, 0.90), taxane-platinum doublet chemotherapy with PD-(L)1 checkpoint inhibitors (HR =0.26, 95% CI: 0.12, 0.57), and taxane-platinum doublet chemotherapy with PD-(L)1 and CTLA-4 checkpoint inhibitors (HR =0.36, 95% CI: 0.14, 0.92) all performed better than taxane monotherapy combined with ramucirumab. No significant differences were found in other pairwise comparisons (Figures 4,5).

Figure 4.

NMA focusing on PFS as the primary outcome measure. Data are presented as hazard ratio (95% confidence interval). CTLA-4, cytotoxic T-lymphocyte-associated protein 4; NMA, network meta-analysis; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; Pt, platinum.

Figure 5.

Probability ranking results for optimal PFS among different interventions. The horizontal axis represents the various interventions and the vertical axis represents the ranking probability of each intervention. Each bar graph shows the probability of ranking different interventions from first to last. ADC, antibody-drug conjugate; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1; PFS, progression-free survival; Pt, platinum.

Based on cumulative probability results, taxane-platinum doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors had the highest SUCRA value of 96.52%, gemcitabine combined with platinum and PD-(L)1 checkpoint inhibitors (SUCRA: 85.16%), and taxane-platinum doublet chemotherapy combined with PD-(L)1 and CTLA-4 checkpoint inhibitors (SUCRA: 80.48%) are likely to be the top three interventions for improving PFS (Figure S2). The overall heterogeneity was low (I²=3%), indicating no variability between the included studies and supporting a high level of consistency in the NMA results.

OS

A total of 28 studies reported OS outcomes for 16 different interventions (3,4,6-10,12-18,28-36,39,41), with the remaining studies described in further references (43-45). The network structure for these interventions is shown in Figure 3B. The NMA results revealed that gemcitabine combined with platinum and Necitumumab significantly improved OS compared to gemcitabine combined with taxane (HR =0.54, 95% CI: 0.32, 0.88), gemcitabine combined with platinum (HR =0.59, 95% CI: 0.35, 0.99), and other regimens such as gemcitabine combined with platinum and necitumumab (HR =0.46, 95% CI: 0.25, 0.82), PD-(L)1 checkpoint inhibitors (HR =0.67, 95% CI: 0.55, 0.83), and PD-(L)1 combined with CTLA-4 dual checkpoint inhibitors (HR =0.59, 95% CI: 0.37, 0.93) when compared to taxane monotherapy. Additionally, taxane-platinum doublet chemotherapy combined with CADI-05 (HR =0.32, 95% CI: 0.14, 0.75), taxane-platinum doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors (HR =0.43, 95% CI: 0.24, 0.77), taxane-platinum doublet chemotherapy combined with PD-(L)1 and CTLA-4 checkpoint inhibitors (HR =0.37, 95% CI: 0.19, 0.75), and taxane-platinum doublet chemotherapy combined with veliparib (HR =0.49, 95% CI: 0.26, 0.92) were all more effective than taxane monotherapy combined with ramucirumab. No significant differences were found in other pairwise comparisons (Figures 6,7).

Figure 6.

NMA focusing on OS as the primary outcome measure. Data are presented as hazard ratio (95% confidence interval). ADC, antibody-drug conjugate; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; NMA, network meta-analysis; OS, overall survival; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1; Pt, platinum.

Figure 7.

Probabilistic ranking results for optimal OS in different interventions. ADC, antibody-drug conjugate; CADI-05, Toll-like receptor-2 agonist; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; OS, overall survival; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1; Pt, platinum.

Based on cumulative probability results, taxane-platinum doublet chemotherapy combined with CADI-05 (SUCRA: 92.11%), taxane-platinum doublet chemotherapy combined with PD-(L)1 and CTLA-4 checkpoint inhibitors (SUCRA: 89.65%), and taxane-platinum doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors (SUCRA: 83.58%) are likely to be the top three interventions for improving OS, as shown in Figure S3. The overall heterogeneity was low (I²=6%), indicating no variability between the included studies and supporting a high level of consistency in the NMA results.

ORR

A total of 22 studies reported ORR outcomes for 14 different interventions (4,6,8-10,12-18,33-41,45). The network structure for these interventions is shown in Figure 3C. The NMA results are presented in Figure 8 and Figure S4. Based on cumulative probability results, taxane-platinum doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors (SUCRA: 87.00%), taxane-platinum doublet chemotherapy combined with Necitumumab (SUCRA: 70.88%), and PD-(L)1 combined with CTLA-4 dual checkpoint inhibitors (SUCRA: 66.23%) are likely to be the top three interventions for improving ORR (Figure S5). The overall heterogeneity was low (I²=3%), indicating no variability between the included studies and supporting a high level of consistency in the NMA results.

Figure 8.

NMA focusing on ORR as the primary outcome measure. Data are presented as relative risk (95% confidence interval). ADC, antibody-drug conjugate; CTLA-4, cytotoxic T-lymphocyte-associated protein 4; NMA, network meta-analysis; ORR, objective response rate; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1; Pt, platinum.

Grade ≥3 TRAEs

A total of 11 studies reported ≥ grade 3 TRAEs for 9 interventions (4,6,8,9,12,13,15,16,30,36,41). The network structure of different interventions is shown in Figure 3D. The NMA results indicated that, except for PD-(L)1 checkpoint inhibitors significantly reducing the incidence of TRAEs compared to taxane monotherapy chemotherapy (RR =0.38, 95% CI: 0.15, 0.98), there were no statistically significant differences in the incidence of ≥ grade 3 TRAEs among the other treatments (see Figure 9). According to the cumulative probability results, PD-(L)1 checkpoint inhibitors (SUCRA: 92.50%), PD-(L)1 combined with CTLA-4 dual checkpoint inhibitors (SUCRA: 79.89%), and taxane monotherapy chemotherapy (SUCRA: 54.58%) may rank as the top three indicators in terms of safety. The overall heterogeneity was low (I²=4%), indicating no variability between the included studies and supporting a high level of consistency in the NMA results.

Figure 9.

NMA focusing on TRAE as the primary outcome measure. Data are presented as relative risk (95% confidence interval). CTLA-4, cytotoxic T-lymphocyte-associated protein 4; NMA, network meta-analysis; PD-1, programmed cell death-1; PD-L1, programmed death-ligand 1; Pt, platinum; TRAE, treatment-related adverse event.

Subgroup analysis

Subgroup NMA was conducted by treatment line and study design (Figures S6,S7). The results showed that taxane-platinum combined with PD-(L)1 checkpoint inhibitors demonstrated superior efficacy in both PFS and OS for first-line treatment, consistent with the main analysis. Furthermore, in both PFS and OS, the efficacy of taxane-platinum combined with PD-(L)1 checkpoint inhibitors in double-blind trials was superior to that in open-label trials (HR for PFS: 0.44 vs. 0.54; HR for OS: 0.68 vs. 0.81). This suggests that potential bias in open-label studies may have led to an underestimation of the efficacy of immunotherapy.

Publication bias

To assess publication bias for the included studies, funnel plots were used for visual evaluation (Figure S8). The results indicated that the funnel plot was approximately symmetrical, suggesting that publication bias is unlikely to have significantly affected the results of this study.

Discussion

To the best of our knowledge, this is the first NMA comparing the efficacy and safety of various treatment regimens for SqNSCLC patients. The NMA analyzed the most recent data from 32 eligible RCTs. The results of our study suggest that taxane-platinum doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors is the most effective regimen for improving PFS and ORR, while taxane-platinum doublet chemotherapy combined with CADI-05 is the most effective for improving OS. Furthermore, PD-(L)1 checkpoint inhibitors not only showed superior safety but also significantly reduced the incidence of grade ≥3 TRAEs when compared to taxane monotherapy. Overall, our findings demonstrate that combined chemotherapy and immunotherapy regimens exhibit superior efficacy compared to chemotherapy alone, which aligns with the results of a previous NMA (19). Among these regimens, the taxane-platinum doublet chemotherapy regimen combined with immunotherapy shows significant potential when compared to other chemotherapy-only regimens.

CADI-05 is an effective Toll-like receptor-2 (TLR-2) agonist that induces a cell-mediated immune response targeting the expression of desmocollin-3 (DSC3) on cancer cells. The innovative combination of taxane-platinum doublet chemotherapy with CADI-05 shows preliminary potential in improving OS, which may be related to the specific expression of DSC3 in SqNSCLC (29,46). However, only one RCT in this NMA included the CADI-05 regimen, involving 221 patients (109 in the chemo-immunotherapy group and 112 in the chemotherapy-only group), warranting further clinical trials to validate these findings.

Our NMA included several nodes supported by only a single or very few trials, particularly CADI-05, necitumumab, and certain veliparib combinations. Such sparse data can lead to network imbalance and imprecision, which may impair the validity of comparisons between nodes. For example, Necitumumab-based regimens were primarily investigated before the widespread adoption of ICIs. Consequently, these high SUCRA values should not be directly interpreted as superiority over contemporary immunotherapy-containing regimens in the absence of head-to-head comparisons. Clinicians and researchers should view these findings as hypothesis-generating rather than practice-changing at present.

Both PD-(L)1 and CTLA-4 inhibitors can improve survival benefits through distinct but complementary mechanisms, yet few studies have directly compared dual immune checkpoint inhibition combined with chemotherapy versus single immune checkpoint inhibition combined with chemotherapy. Our NMA findings show that single checkpoint inhibition plus chemotherapy was superior to dual checkpoint inhibition plus chemotherapy in improving PFS and ORR, while the opposite was true for OS. This deviates slightly from the results of the POSEIDON trial, which randomized 1,013 patients (1:1:1) to the CTLA-4 + PD-L1 + chemotherapy group, the PD-L1 + chemotherapy group, and the chemotherapy-only group. The study showed that the PD-L1 + chemotherapy group had a slight advantage over the CTLA-4 + PD-L1 + chemotherapy group in improving PFS (median PFS: 4.7 vs. 4.6 months) and OS (median OS: 11.5 vs. 10.4 months), with a better long-term OS benefit (5-year OS rate: 7.6% vs. 7.3%) (47,48). This discrepancy may be related to the small number of RCTs included in our NMA involving dual immunotherapy combined with chemotherapy (only one study).

We included two ADCs, datopotamab deruxtecan and sacituzumab govitecan, as treatments for later-line therapy in randomized RCTs. The results indicated that ADCs were among the least effective treatment options in all regimens. Notably, both trials had docetaxel as the control group. Datopotamab deruxtecan did not improve PFS (mPFS 2.8 vs. 3.9 months) or OS (mOS 7.6 vs. 9.4 months) compared to docetaxel, while sacituzumab govitecan showed improvements in PFS [HR 0.83 (95% CI: 0.56–1.22)] and OS [HR 0.83 (95% CI: 0.56–1.22)], but these differences were not statistically significant (43,45). Based on these two RCTs and the results of our NMA, we conclude that datopotamab deruxtecan is not recommended for later-line treatment of sqNSCLC, and sacituzumab govitecan should not be prioritized. This conclusion aligns with the ASCO Living Guideline (49).

Although our study suggests that taxane-platinum doublet chemotherapy combined with PD-(L)1 checkpoint inhibitors may carry a higher toxicity burden (SUCRA: 23.51%), this finding is limited by insufficient reporting in the original studies. In clinical practice, the core value of this regimen should not be dismissed based solely on this ranking. Notably, the overall safety profile of this combination has been well-established across multiple pivotal RCTs (4,6,8,10). The CameL-Sq trial (6) demonstrated that adding camrelizumab to chemotherapy did not significantly increase chemotherapy-related adverse events compared to chemotherapy alone, with no unexpected immune-related adverse events observed in either arm. Similarly, Wang et al. randomized 360 sqNSCLC patients to: Arm A (tislelizumab + paclitaxel/carboplatin, n=120), Arm B (tislelizumab + nab-paclitaxel/carboplatin, n=119), or Arm C (paclitaxel/carboplatin, n=121). Final analysis revealed comparable rates of grade ≥3 TEAEs across groups (89.2% vs. 87.3% vs. 84.6%, respectively) (8). However, the IMpower131 trial reported higher rates of serious TRAEs and treatment discontinuations in the atezolizumab-chemotherapy arm compared to chemotherapy alone. Investigators attributed this to known risks of atezolizumab, including febrile neutropenia, anemia, and pneumonitis. This observation underscores that the toxicity profile of the combination is well-characterized rather than representing unknown or unmanageable risks.

Collectively, these key trials indicate that while combination therapy introduces immune-related AEs and may potentiate some chemotherapy toxicities, its safety spectrum remains clearly defined. With proper management protocols, these risks remain overall controllable in clinical practice.

Limitations

However, there are several limitations in our study. Firstly, our NMA did not distinguish between different types of platinum agents, taxanes, or anti-PD-(L)1 drugs. There may be variations in efficacy among different platinum compounds (e.g., carboplatin, cisplatin), taxanes (e.g., paclitaxel, nab-paclitaxel, docetaxel), and anti-PD-(L)1 drugs (e.g., atezolizumab, durvalumab, cemiplimab). Secondly, since most of the trials included in our analysis were open-label, the lack of blinding may affect the credibility of the results. Thirdly, in our analysis, regimens such as ADCs and taxane plus ramucirumab were evaluated primarily in later-line settings. By contrast, most immunotherapy-containing regimens were studied in the first-line setting. Therefore, the relatively inferior performance of ADC and ramucirumab combinations observed in this NMA should not be extrapolated to first-line treatment decisions. Our subgroup analysis by line of therapy partly addresses this issue, but residual confounding by line of therapy remains a key limitation. Finally, the studies included in our analysis spanned from 2011 to 2025, and these temporal trends may influence outcomes independently of the treatment regimens. Therefore, an exploratory sensitivity analysis excluding studies published before 2018 (i.e., the pre-immunotherapy era) yielded consistent directions of efficacy for most modern regimens (see Figure S9). However, the early studies of CADI-05 and necitumumab remained susceptible to this bias. Future studies should perform detailed subgroup analyses based on different platinum agents, taxanes, and immune drugs, and adopt blinded designs.

Conclusions

In summary, our analysis found that the triple combination of taxane-platinum chemotherapy with PD-(L)1 inhibitors provides substantial survival benefits in the first-line treatment of sqNSCLC, solidifying its position as a preferred therapeutic option. While safety analyses indicate a discernible toxicity profile, the compelling efficacy advantage remains clinically decisive under existing evidentiary standards. The key to successful implementation lies in integrating standardized toxicity prevention, monitoring, and management strategies. Future investigation through rigorously designed, multicenter, large-sample, long-term follow-up RCTs will further validate the efficacy and safety of this combination approach.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.