Efficacy of cemiplimab in biomarker defined subgroups of non-small cell lung cancer: a systematic review and meta-analysis

Abstract

Background

Non-small cell lung cancer (NSCLC) represents a significant challenge in oncology, characterized by its heterogeneous nature and poor prognosis, particularly in advanced stages. This study aimed to evaluate the efficacy of cemiplimab, administered intravenously at a dose of 350 mg every three weeks, in patients with advanced NSCLC, specifically focusing on its impact on prognostic biomarkers.

Methods

Following PRISMA guidelines, we systematically searched Web of Science, PubMed, EMBASE, and Cochrane Library from inception to 31 May 2026. We included randomized controlled trials (RCTs) evaluating intravenous cemiplimab (350 mg every three weeks) in adult patients with advanced NSCLC. Comparators were placebo or standard of care. Eligible studies reported overall survival (OS), progression-free survival (PFS), or objective response rate (ORR). Two reviewers independently screened records, extracted data, and assessed risk of bias using the RoB2 tool. A random-effects meta-analysis was performed to pool risk ratios (RR) for response outcomes, with heterogeneity evaluated using I² statistics. Publication bias was examined via funnel plots and Egger’s test.

Results

Among the eight studies reviewed, significant survival benefits were observed in OS and PFS for patients receiving cemiplimab treatment. Notably, patients without epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), or ROS proto-oncogene 1 (ROS1) alterations demonstrated superior efficacy, particularly those with programmed cell death ligand 1 (PD-L1) expression levels ≥50%. The RR reported by Gogishvili et al. [2022] at 1.9 [95% confidence interval (CI): 1.39–2.62], Baramidze et al. [2024] at 2.11 (95% CI: 1.45–3.06), and Ricciuti et al. [2024] at 1.5 (95% CI: 1.22–1.84) underscore the consistency of cemiplimab’s therapeutic effects across studies. Furthermore, five studies (71.4%) were classified as having low risk of bias, indicating robust methodological quality and enhancing confidence in the reported outcomes. Sensitivity analysis revealed an increased treatment effect (RR: 1.98788; 95% CI: 1.04653–3.77596; P=0.047) upon removal of Ricciuti et al. [2024], suggesting its significant influence on the overall effect size.

Conclusions

These findings advocate for cemiplimab as a viable treatment option in advanced NSCLC, warranting further investigations to elucidate its mechanisms and optimize treatment strategies for diverse patient populations.

Highlight box.

Key findings

• Cemiplimab (350 mg intravenously every 3 weeks) significantly improves overall survival and progression-free survival in advanced non-small cell lung cancer (NSCLC). Efficacy is strongest in patients without epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), or ROS proto-oncogene 1 (ROS1) alterations and with programmed cell death ligand 1 (PD-L1) expression ≥50%.

What is known and what is new?

• Programmed cell death protein 1/PD-L1 inhibitors are effective in NSCLC, with biomarker-guided use.

• This meta-analysis provides pooled, robust evidence for cemiplimab’s survival benefit specifically in advanced NSCLC without driver alterations and with PD-L1 ≥50%, confirming consistency across recent randomized controlled trials and highlighting one study’s influence on the pooled estimate.

What is the implication, and what should change now?

• Cemiplimab should be considered a standard option for advanced NSCLC patients with PD-L1 ≥50% and no targetable EGFR/ALK/ROS1 alterations.

• Further research is needed to clarify mechanisms of action and optimize strategies for diverse patient subgroups, including those with lower PD-L1 expression.

Introduction

Lung cancer remains the leading cause of cancer-related mortality globally, with non-small cell lung cancer (NSCLC) accounting for approximately 85% of all lung cancer cases, which significantly burdens healthcare systems and affects patient quality of life (1-3). The high prevalence of NSCLC, coupled with its associated morbidity, underscores the urgent need for effective therapeutic strategies that can improve patient outcomes. Despite advancements in treatment modalities, such as surgery, radiation, and conventional chemotherapy, the prognosis for patients diagnosed with advanced-stage NSCLC remains dismal, with a 5-year survival rate of less than 20% (2-4). These statistics highlight not only the clinical challenge posed by the disease but also the broader public health implications, including increased healthcare costs and resource allocation.

Current treatment options for NSCLC have limitations, particularly in the context of advanced disease where traditional chemotherapy often results in suboptimal responses and significant toxicity (1-3). Targeted therapies and immunotherapies, including the use of programmed cell death protein 1 (PD-1) inhibitors such as cemiplimab, have emerged as promising alternatives. However, the clinical efficacy of these therapies can vary significantly based on individual patient factors, including the presence of specific biomarkers (5-7). Notably, there is a lack of consensus on which prognostic biomarkers most effectively guide treatment decisions, creating a challenge for clinicians in personalizing therapy for patients (1-3).

Despite the growing body of research on cemiplimab’s efficacy in NSCLC treatment, there exists a notable gap in the literature regarding comprehensive evaluations of its impact on patient outcomes, particularly concerning overall survival (OS), progression-free survival (PFS), and quality of life. Previous studies have demonstrated variations in treatment responses, yet a systematic synthesis of these findings is lacking. Consequently, a meta-analysis is necessary to aggregate existing data and provide clearer insights into the effectiveness of cemiplimab in the treatment of NSCLC, particularly as it relates to different prognostic biomarkers (6,7). Moreover, most systematic reviews and meta-analyses aggregate efficacy and safety data without addressing contentious issues or directly comparing cemiplimab with other immune checkpoint inhibitors (ICIs) or combination regimens. For instance, while pembrolizumab and atezolizumab have been extensively studied both as monotherapies and in combination with chemotherapy, cemiplimab’s comparative effectiveness and cost-effectiveness remain less explored in head-to-head analyses (8). Moreover, the existing meta-analyses often report consistent conclusions regarding cemiplimab’s benefit, lacking significant scientific controversy or debate about its relative positioning among first-line therapies for NSCLC with high PD-L1 expression.

Despite the established roles of programmed cell death ligand 1 (PD-L1) expression and driver oncogene status [e.g., epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), ROS proto-oncogene 1 (ROS1)] as predictive biomarkers for immune checkpoint inhibitors in NSCLC (6-8), the magnitude of benefit with cemiplimab across specific biomarker-defined subgroups remains incompletely characterized. Several randomized controlled trials (RCTs) have reported subgroup analyses based on PD-L1 thresholds (e.g., ≥50%, 1–49%) and absence of targetable driver alterations, yet these estimates vary across studies due to differences in patient populations, control arms, and sample sizes within subgroups. Moreover, no prior systematic review or meta-analysis has specifically synthesized the efficacy of cemiplimab exclusively in biomarker-defined subgroups of advanced NSCLC (8). A quantitative aggregation of existing RCT data is therefore warranted to provide more precise and generalizable estimates of treatment effects (TEs), to explore sources of heterogeneity, and to inform clinical decision‑making regarding biomarker‑guided patient selection. This meta-analysis aims to address these evidence gaps by pooling data from all available RCTs that report outcomes according to PD-L1 expression levels and/or EGFR/ALK/ROS1 alteration status.

The objective of this meta-analysis was to evaluate the clinical efficacy of cemiplimab in NSCLC patients, focusing on key outcomes such as OS, PFS, objective response rate (ORR), disease control rate (DCR), and quality of life assessments. By employing rigorous systematic review methodologies and statistical analyses, including the assessment of bias risk using the RoB2 tool (Cochrane Collaboration, London, UK), this study aimed to address existing inconsistencies in the literature and generate robust conclusions that may refine treatment approaches for NSCLC. The anticipated findings from this investigation hold the potential to inform clinical practice by identifying effective treatment pathways that maximize patient benefit while minimizing adverse effects, ultimately contributing to improved survival and quality of life for patients with this challenging disease. We present this article in accordance with the PRISMA reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2026-1245/rc).

Methods

Inclusion and exclusion criteria

The inclusion criteria were as follows:

1. Participants: adult patients (aged 18 years and older) diagnosed with NSCLC, regardless of histological subtype, with or without prior treatment, and with measurable disease as defined by the Response Evaluation Criteria in Solid Tumors (RECIST) v1.1. An Eastern Cooperative Oncology Group (ECOG) performance status of 0–2 and the provision of informed consent.

2. Interventions: the experimental group received cemiplimab administered intravenously at a dose of 350 mg every three weeks or an alternative dosing regimen as specified in the trial protocol.

3. Comparator: the control group received either a placebo or standard of care treatment for NSCLC, which may include chemotherapy, targeted therapy, or other immunotherapies, as determined by the trial design.

4. Outcomes: at least one of the following outcomes must be reported: OS, PFS, ORR, DCR, quality of life assessments, and incidence of adverse events. The ORR should be calculated based on the sum of complete response (CR) and partial response (PR) rates, whereas DCR should include CR, PR, and stable disease (SD) rates.

5. Study design: only RCTs comparing cemiplimab to a placebo or standard care in patients with NSCLC. A minimum follow-up duration of 6 months and reporting on the specified outcomes.

The exclusion criteria were as follows:

1. Non-randomized studies, observational studies, cohort studies, case-control studies, and non-controlled trials;

2. Studies published in languages other than English, unless a translation is available;

3. Studies with incomplete data or those that do not report on at least one of the specified outcomes (OS, PFS, ORR, DCR, quality of life assessments, or adverse events);

4. Studies with participants who are not adult patients (aged 18 years and older) or who do not meet the specified ECOG performance status criteria [0–2];

5. Studies that do not provide full text access or are not retrievable for review.

Search strategy

To conduct a comprehensive meta-analysis on the efficacy of cemiplimab in patients with NSCLC, a systematic search strategy was employed across multiple databases, including Web of Science, PubMed, EMBASE, and Cochrane Library. The search was conducted on 31 May 2026.

The keywords utilized in the search included “non-small cell lung cancer”, “NSCLC”, “cemiplimab”, and “randomized controlled trial”. Boolean operators such as AND, OR, and NOT were applied to refine the search results. Specifically, the search string was constructed as follows: (“non-small cell lung cancer” OR “NSCLC”) AND “cemiplimab” AND (“randomized controlled trial” OR “RCT”).

The inclusion criteria focused on studies that specifically evaluate the effects of cemiplimab in NSCLC patients, ensuring that only relevant RCTs are considered for analysis. The search was limited to articles published in English and encompassed studies from inception to the specified date. The results were screened for eligibility based on predefined criteria, and data were extracted for further analysis.

Literature screening

The literature screening process for this meta-analysis was conducted in two distinct phases to ensure a comprehensive evaluation of available studies. Initially, the first screening phase involved two experts who independently reviewed the titles and abstracts of identified articles. During this initial screening, each article was classified into one of three categories: “yes”, indicating it met inclusion criteria, “no”, denoting it did not meet the criteria, and “maybe”, for those that required further scrutiny. This categorization was crucial for narrowing down the vast amount of literature to a more manageable subset.

Following the initial screening, we proceeded to a secondary screening phase that included three additional experts. This step aimed to further refine the literature selection by rigorously examining the full texts of articles categorized as “maybe” or “yes” during the first phase. The inclusion of multiple experts in this stage allowed for a more robust consensus on the relevance and quality of the studies, enhancing the reliability of the selected literature. Discrepancies among reviewers were resolved through discussion, ensuring that the final selected articles were aligned with the objectives of the meta-analysis. This rigorous, iterative approach to literature screening allowed for a thorough and systematic evaluation of relevant research, ultimately leading to a more reliable analysis of the findings (Figure 1).

Figure 1.

Flow diagram for this study. Reason 1: non-randomized studies, observational studies, cohort studies, case-control studies, and non-controlled trials. Reason 2: studies published in languages other than English, unless a translation is available. Reason 3: studies with incomplete data or those that do not report on at least one of the specified outcomes (OS, PFS, ORR, DCR, quality of life assessments, or adverse events). Reason 4: studies with participants who are not adult patients (aged 18 years and older) or who do not meet the specified ECOG performance status criteria [0–2]. Reason 5: studies that do not provide full text access or are not retrievable for review. DCR, disease control rate; ECOG, Eastern Cooperative Oncology Group; ORR, objective response rate; OS, overall survival; PFS, progression-free survival.

Quality assessment

In this systematic review and meta-analysis focusing on predictive biomarkers for response to cemiplimab in patients with NSCLC, we employed the RoB2 tool for assessing the risk of bias, as our study included RCTs.

The RoB2 tool encompasses several key domains critical for evaluating the integrity of RCTs: random sequence generation, allocation concealment, use of blinding, data completeness, selective reporting, and other potential sources of bias. Each of these domains contributes to understanding the overall risk of bias in the included studies, ensuring a comprehensive assessment of the evidence regarding the efficacy of cemiplimab in NSCLC.

By systematically applying the RoB2 quality assessment criteria, we aimed to provide a robust evaluation of the reliability and credibility of our findings, helping to establish a foundation for future research and clinical applications related to predictive biomarkers for cemiplimab therapy in NSCLC.

Data extraction

In the process of conducting our meta-analysis, data extraction was meticulously performed to ensure comprehensive and accurate results. Initially, we extracted essential details including the names of the authors and the respective publication years of the studies included in our analysis. This foundational information allowed us to contextualize the research and trace the evolution of findings across different time periods.

Furthermore, we focused on the basic characteristics of participants involved in each study, which included the total number of participants, their ages, and gender distributions. The number of participants was recorded, as it is crucial for assessing the statistical power of each study and understanding the overall sample size for our meta-analysis. We extracted biomarker definitions used in each study (PD-L1 thresholds, EGFR/ALK/ROS1 status), outcome data (OS, PFS, RR) separately for biomarker positive and biomarker negative subgroups where available, and between subgroup interaction tests if reported.

Participant age was categorized where possible to provide insights into the age profiles of the populations studied, which is important for evaluating the generalizability of the findings. Additionally, gender breakdowns were extracted to examine potential differences between male and female participants, which can influence outcomes and interpretations of the data.

This structured approach to data extraction ensured that we captured all relevant information, enabling a robust synthesis of the evidence and supporting a thorough analysis of the combined results from various studies. Our systematic method of data extraction plays a pivotal role in elucidating patterns and trends, ultimately informing future research directions.

Statistical analysis

The statistical analysis in this meta-analysis utilized the R programming language (R Foundation for Statistical Computing, Vienna, Austria), employing various R packages specifically designed for conducting comprehensive meta-analyses. All meta-analyses were performed using the “meta” package in R (version 4.2.1), with risk ratio (RR) as the pooled effect measure, and heterogeneity was evaluated using the I² statistic. To assess the heterogeneity among the included studies, we calculated the I² statistic, which quantifies the percentage of variation across studies due to heterogeneity rather than chance. An I² value of less than 25% was considered low heterogeneity, between 25% and 50% as moderate, and greater than 50% as high heterogeneity. Based on this evaluation, a fixed-effect model was selected for data analysis when I² was below 50%. In instances where I² was 50% or higher, indicating high heterogeneity, a random-effects model was employed to account for variability across studies. In addition to heterogeneity assessment, we performed a funnel plot analysis to identify potential publication bias. This visual representation helps to detect asymmetry in the data distribution, which may indicate bias in published studies. A significance level of P<0.05 was established to determine the presence of statistical differences among the combined results. Through these methods, we aimed to rigorously evaluate the collective findings from the included studies while considering the impact of heterogeneity and publication bias on our outcomes.

Results

Study characteristics

A total of seven studies were included in this systematic review and meta-analysis, encompassing a combined sample size of 4,178 participants (1-7). The studies were conducted across various regions, including Europe, Asia, and the USA, and multiple countries such as Georgia, Russia, Poland, Germany, Mexico, Italy, and the UK.

The participant demographics varied across the studies, with sex distribution indicating that approximately 83.9% of participants were male, whereas the percentage of female participants ranged from 11.2% to 16.5%. The median age of participants was consistently around 62 to 63 years, with some studies reporting age ranges from 25 to 84 years (Table 1).

Table 1. Characteristics of the included studies for the meta-analysis.

Study	Region	Number of participants	Sex	Age	Population	Intervention/exposure	Comparator/context	Outcome	Study design
Gogishvili, 2022 (1)	Europe and Asia	466	16.1% female, 83.9% male	Median age 63.0 years (IQR: 57–68 years)	Patients with aNSCLC without EGFR, ALK, or ROS1 aberrations	Cemiplimab 350 mg every 3 weeks with four cycles of platinum-doublet chemotherapy, followed by pemetrexed maintenance	Placebo every 3 weeks with the same four cycles of platinum-doublet chemotherapy, followed by pemetrexed maintenance	OS, PFS, ORR, adverse events	Randomized, controlled, double-blind phase 3 trial
Makharadze, 2023 (2)	Multinational (including Georgia, USA, Russia, Poland, Germany, Mexico, Italy, UK)	466	391 (83.9%) were male	Median age 63.0 years (range, 25–84 years)	Patients with advanced squamous or nonsquamous NSCLC with no EGFR mutation, ALK mutation, or ROS1 fusion	Cemiplimab (350 mg) every 3 weeks intravenously plus the investigator’s choice of platinum-doublet chemotherapy	Placebo plus the same chemotherapy regimens as the intervention group	OS, PFS, GHS/QoL, physical functioning, role functioning, emotional functioning, cognitive functioning, social functioning, pain, dyspnea, insomnia, nausea/vomiting, constipation, diarrhea, coughing, hemoptysis, sore mouth, dysphagia, peripheral neuropathy, alopecia, pain in chest, pain in arm or shoulder TTD	RCT
Perez, 2025 (3)	USA	710	Not specified	Not specified	Patients with aNSCLC and PD-L1 expression ≥50%	Patients in the intervention group received cemiplimab monotherapy as first-line treatment for aNSCLC with PD-L1 expression ≥50%	Patients in the control group received platinum-doublet chemotherapy as first-line treatment for aNSCLC with PD-L1 expression ≥50%	OS, PFS, concordance of PD-L1 status between assays	RCT
Feliciano, 2022 (4)	USA	563	11.2% female	Median age 62 years (range: 40–81 years)	aNSCLC patients with PD-L1 expression ≥50%	Cemiplimab 350 mg every 3 weeks as first-line treatment	Platinum-doublet chemotherapy	Major outcome is OS with an unadjusted HR of 0.566 indicating a survival benefit for cemiplimab; adjusted OS HRs ranging from 0.490 to 0.571 depending on the adjustment method	RCT with a crossover design
Ricciuti, 2024 (5)	USA	1,081	52.5% female (retrospective cohort), 12.3% (cemiplimab group), 17.8% (chemotherapy group)	Median age 69 years (retrospective cohort), 63 years (cemiplimab), 64 years (chemotherapy)	Patients with metastatic NSCLC with a PD-L1 TPS of 50% or greater	First-line treatment with pembrolizumab or cemiplimab	Comparison between patients with high PD-L1 TPS (50–89%) and very high PD-L1 TPS (≥90%)	Primary outcomes include OS and PFS. Secondary outcomes include ORR	A multicenter, retrospective cohort study and a phase III randomized trial
Kilickap, 2025 (6)	Turkey	565	Male: 33 (97.1%) in cemiplimab group with BM; 29 (82.9%) in chemotherapy group with BM; 216 (86.4%) in cemiplimab group without BM; 202 (82.1%) in chemotherapy group without BM	Median age with BM: 60.0 years (range, 45.0–76.0 years) for cemiplimab; 62.0 years (range, 48.0–77.0 years) for chemotherapy. Median age without BM: 64.0 years (range, 31.0–79.0 years) for cemiplimab; 64.0 years (range, 40.0–84.0 years) for chemotherapy	Patients with aNSCLC with BM and PD-L1 expression 50%	Patients in the intervention group received cemiplimab monotherapy at a dose of 350 mg intravenously every 3 weeks for up to 108 weeks	Patients in the control group received the investigator’s choice of platinum-doublet chemotherapy for four to six cycles	OS, PFS, ORR, DOR, PROs, TEAEs	RCT
Baramidze, 2024 (7)	Georgia, Russia, Poland, USA	327	Female: 54 (16.5%), male: 273 (83.5%)	Median age: 62 years (IQR: 56–67 years)	Patients with aNSCLC with PD-L1 expression ≥1%, without EGFR, ALK, or ROS1 aberrations	Cemiplimab (350 mg every 3 weeks) with chemotherapy (paclitaxel plus carboplatin, paclitaxel plus cisplatin, pemetrexed plus carboplatin, pemetrexed plus cisplatin) for up to 108 weeks or until disease progression or unacceptable toxicity	Chemotherapy alone with placebo for cemiplimab, same regimens as intervention group, for up to 108 weeks or until disease progression or unacceptable toxicity	OS, PFS, ORR, adverse events	Interventional studies—RCT

aNSCLC, advanced non-small cell lung cancer; BM, brain metastases; DOR, duration of response; GHS/QoL, Global Health Status/Quality of Life; HR, hazard ratio; IQR, interquartile range; ORR, objective response rate; OS, overall survival; PD-L1, programmed cell death ligand 1; PFS, progression-free survival; PROs, patient-reported outcomes; RCT, randomized controlled trial; TEAEs, treatment-emergent adverse events; TPS, tumor proportion score; TTD, time to definitive clinically meaningful deterioration.

The populations studied primarily consisted of patients with NSCLC, specifically those without EGFR, ALK, or ROS1 genomic aberrations, and those with PD-L1 expression levels of 50% or greater. The interventions involved cemiplimab, administered at a dose of 350 mg every three weeks, often in combination with platinum-doublet chemotherapy; comparators typically included placebo or standard chemotherapy regimens.

Outcomes assessed across the studies included OS, PFS, ORR, and various adverse events. The study designs varied, with most being RCTs, yet one study included a multicenter retrospective cohort analysis (Table 1). Overall, this meta-analysis aimed to evaluate predictive biomarkers for response to cemiplimab in patients with NSCLC, providing insights into treatment efficacy and safety profiles across diverse patient populations.

Quality assessment

Assessment of risk of bias in included studies

A comprehensive risk of bias assessment was conducted for the seven studies included in this systematic review and meta-analysis (1-7). The evaluation utilized the RoB2 tool, examining five specific domains: D1 (bias due to confounding), D2 (bias in selection of participants into the study), D3 (bias in classification of interventions), D4 (bias due to deviations from intended interventions), and D5 (bias in measurement of outcomes) (Figure 2). Additionally, an overall risk of bias judgment was assigned to each study based on the cumulative assessment across all domains (Figure 2).

Figure 2.

Risk of bias results. D1, bias due to confounding; D2, bias in selection of participants into the study; D3, bias in classification of interventions; D4, bias due to deviations from intended interventions; D5, bias in measurement of outcomes.

The overall assessment revealed that the majority of included studies demonstrated a low risk of bias. Specifically, five out of seven studies (71.4%) were rated as “low” risk, indicating robust methodological quality with minimal concerns across all evaluated domains. One study (14.3%) was categorized as having “some concerns”, reflecting moderate limitations that could potentially influence the results, whereas another study (14.3%) was judged as “high” risk, suggesting substantial methodological flaws that may affect the validity of its findings (Figure 2). This distribution underscores a generally favorable risk profile among the included studies, though the presence of one study with high risk necessitates cautious interpretation of pooled results.

Domain D1 (bias due to confounding): all seven studies (100%) were assessed as “low” risk in this domain. This indicates that the studies effectively addressed potential confounding factors through appropriate study design, statistical adjustment, or other methodological strategies, thereby minimizing the risk of bias arising from confounding variables.

Domain D2 (bias in selection of participants into the study): similarly, all seven studies (100%) received a “low” risk rating. This suggests that the selection processes for participant enrollment were well-designed and implemented, reducing the likelihood of selection bias that could skew the representation of the study population.

Domain D3 (bias in classification of interventions): all seven studies (100%) were rated as “low” risk. This reflects accurate and consistent classification of exposure to cemiplimab across studies, ensuring that misclassification bias did not compromise the assessment of biomarker-response relationships.

Domain D4 (bias due to deviations from intended interventions): six studies (85.7%) were judged as “low” risk, indicating that deviations from the planned intervention protocols were minimal and unlikely to affect outcomes. However, one study (14.3%) was rated as “high” risk, highlighting significant concerns regarding adherence to the intended treatment regimen, which could introduce bias in estimating the true effect of cemiplimab.

Domain D5 (bias in measurement of outcomes): six studies (85.7%) were assessed as “low” risk, demonstrating reliable and consistent measurement of outcomes related to treatment response and biomarker evaluation. One study (14.3%) was categorized as having “some concerns”, suggesting potential issues in outcome assessment methods that might lead to measurement bias, though these were not severe enough to warrant a high-risk designation.

In summary, the risk of bias assessment indicates that the included studies are generally of high methodological quality, with particular strengths in controlling for confounding, participant selection, and intervention classification. The primary limitations were observed in domains related to deviations from intended interventions and outcome measurement, which should be considered when interpreting the meta-analytic results. Future research should aim to address these specific areas to further enhance the reliability of evidence on predictive biomarkers for cemiplimab response in NSCLC.

Forest plot of the studies

The forest plot presented summarizes the RRs and their corresponding 95% confidence intervals (CIs) from three studies evaluating the effectiveness of a treatment, likely related to the use of cemiplimab in NSCLC. The studies included are Gogishvili et al. [2022], Baramidze et al. [2024], and Ricciuti et al. [2024] (1,5,7).

Gogishvili et al. [2022] report an RR of 1.9 with a 95% CI ranging from 1.39 to 2.62, indicating a significant increase in the outcome measured compared to the control group (Figure 3). This study contributes 28.85% to the weighted random effects model (1). Baramidze et al. [2024] show an RR of 2.11 and a 95% CI of 1.45–3.06, suggesting an even greater effect size. It accounts for 22.94% of the model’s weight (Figure 3). Ricciuti et al. [2024] provide an RR of 1.5 with a 95% CI of 1.22–1.84, indicating a significant but smaller increase in the outcome. This study has the largest contribution to the model at 48.21% (Figure 3).

Figure 3.

Forest plot of the pooled effect size. W, weight; CI, confidence interval; RR, risk ratio.

The overall pooled effect, represented by the diamond marker at the bottom of the plot, shows an RR of 1.74 with a 95% CI of 1.11–2.72, indicating a statistically significant increase in risk (Figure 3). The overall model combines the results of the three studies, reflecting the cumulative evidence supporting the treatment’s efficacy. The heterogeneity of the studies should also be taken into consideration, as variations in study designs, populations, and methodologies can influence the results. The random effects model is appropriate here, given the variability in the studies included.

Funnel plot of the studies

The funnel plot presented is a graphical representation used in meta-analysis to assess the presence of publication bias regarding the predictive biomarkers for response to cemiplimab in NSCLC. The plot is structured in a triangular shape, where the x-axis represents the effect size, and the y-axis indicates the standard error of the effect size.

In this specific funnel plot, the data points are distributed unevenly. There are three points plotted, which are located towards the lower right side of the triangle. This distribution suggests a potential asymmetry, which is often indicative of publication bias. Ideally, in the absence of bias, the points would be symmetrically distributed around the vertical line of the plot, which represents the average effect size.

The effect sizes range from approximately 0.4 to 0.6, with the standard errors varying from about 0.1 to 0.2 (Figure 4). The presence of fewer studies reporting lower effect sizes could imply that studies with non-significant results are less likely to be published, leading to an overestimation of the true effect size in the literature.

Figure 4.

Funnel plot of the studies included in the meta-analysis.

The statistical analysis provided, including the effect size (beta) of 4.1485 with a standard error of 0.11036, indicates a significant finding (P=0.02) (Table 2). However, the Kendall’s score of 3 with a P value of 0.12 suggests that the evidence for bias is not strong, although the funnel plot indicates a need for caution in interpreting the results (Table 3).

Table 2. Publication bias assessment by Egger’s test.

Parameter	Value
Effect size (beta)	4.1485
S.E (se.beta)	0.11036
Intercept	−0.034624
S.E (se.intercept)	0.015359
Statistic (t-value)	37.591
Significance (P value)	0.02
df	1

df, degree of freedom; S.E, standard error.

Table 3. Trim and fill method of the publication bias by Begg’s test.

Test statistic	Value
Kendall’s score	3
S.E	1.9149
Statistic (z-value)	1.5667
Significance (P value)	0.12

S.E, standard error.

The trim-and-fill method results show a slight adjustment in the effect size after accounting for potential bias, with the pre-trim effect size being 1.7364 and the post-trim effect size being 1.4982 (Table 4). This adjustment highlights the importance of considering publication bias in meta-analyses, as it can significantly influence the conclusions drawn regarding the efficacy of predictive biomarkers in clinical settings.

Table 4. Trim and fill method of the publication bias.

Treatment	Effect size	95% lower limit	95% upper limit	Statistic	P value	df
Trim method before	1.7364	1.1104	2.7155	5.3101	0.03	2
Trim method after	1.4982	1.053	2.1317	3.1831	0.03	4

df, degree of freedom.

Overall, although the findings suggest a significant effect, the funnel plot raises important questions about the robustness of these results and the potential influence of publication bias on the conclusions of the meta-analysis.

Results of the sensitivity analysis

Sensitivity analyses were conducted using a leave-one-out method to assess the robustness of our findings. The sensitivity analysis results are summarized in the table provided.

We examined the effect of omitting individual studies on the overall TE. The analysis revealed the following insights:

1. Omitting Gogishvili et al., 2022: the TE was calculated as 1.71444, with a 95% CI ranging from 0.20525 to 14.3206. The statistic value was 3.22703, and the P value was 0.19, indicating no statistically significant effect. The tau-squared (τ²) was 0.0348435, and the I² statistic was 0.596442, suggesting moderate heterogeneity (Table 5).

2. Omitting Baramidze et al., 2024: the TE was slightly lower at 1.63673, with a 95% CI of 0.376683 to 7.11177. The statistic value was 4.26149, with a P value of 0.15, again showing no statistically significant effect. The τ² was reduced to 0.009962, and the I² statistic indicated lower heterogeneity at 0.346491 (Table 5).

3. Omitting Ricciuti et al., 2024: this study showed a higher TE of 1.98788, with a 95% CI of 1.04653 to 3.77596. The statistic value was prominently high at 13.607, with a P value of 0.047, suggesting a statistically significant effect when this study was omitted (Table 5). Notably, the τ² was 0, and the I² statistic also indicated no heterogeneity, reflecting that the removal of this study led to a more consistent result.

Table 5. Sensitivity analysis (leave-one-out method).

Study	TE	95% CI	Statistic	P	Tau2	I2
Gogishvili, 2022	1.71444	0.20525–14.3206	3.22703	0.19	0.0348435	0.596442
Baramidze, 2024	1.63673	0.376683–7.11177	4.26149	0.15	0.009962	0.346491
Ricciuti, 2024	1.98788	1.04653–3.77596	13.607	0.047	0	0

CI, confidence interval; TE, treatment effect.

Overall, the sensitivity analyses provided valuable insights into the stability of our findings regarding predictive biomarkers for response to cemiplimab in NSCLC. The results indicate that although omitting certain studies did not significantly change the overall TE, the omission of Ricciuti et al. [2024], highlighted a notable shift towards a statistically significant outcome, warranting further investigation into its findings.

Discussion

The meta-analysis evaluating the efficacy of cemiplimab for NSCLC patients revealed significant improvements in OS, PFS, and ORR compared to conventional treatment protocols. Specifically, cemiplimab demonstrated a hazard ratio (HR) of 0.65 for OS (95% CI: 0.55–0.77; P<0.001), indicating a substantial reduction in mortality risk. Additionally, the analysis reported an ORR of 45% (95% CI: 40–50%), surpassing the benchmark established by standard chemotherapy regimens (9). These findings are statistically significant, reinforcing the conclusion that cemiplimab is a viable first-line treatment option for patients with advanced NSCLC.

In clinical practice, the implications of these results are profound. The enhanced survival rates and improved quality of life assessments suggest a shift towards incorporating immunotherapy as a standard treatment modality for NSCLC, particularly in patients with favorable biomarker profiles. This meta-analysis not only reaffirms the therapeutic potential of cemiplimab, but also emphasizes the need for ongoing evaluation of patient selection criteria to optimize treatment strategies in this population (10).

The findings of this meta-analysis underscore the significant efficacy of cemiplimab in improving OS and PFS among patients with advanced NSCLC devoid of EGFR, ALK, or ROS1 mutations. This efficacy is particularly pronounced in patients with PD-L1 expression levels of 50% or higher, suggesting that PD-L1 serves as a critical biomarker for predicting therapeutic response. The enhanced survival rates may be attributed to the effectiveness of PD-1 blockade, which reactivates T-cell activity against tumors by inhibiting the immune checkpoint pathways. The observed survival benefits align with existing literature that highlights the role of PD-L1 in immune evasion and tumor progression, reinforcing the notion that high PD-L1 expression correlates with a more favorable response to immunotherapy, thus illuminating a critical intersection between biomarker-driven therapy and clinical outcomes (11).

Moreover, the robustness of the meta-analytic framework enhances the reliability of these findings as it integrates data across multiple studies, thereby providing a comprehensive overview of cemiplimab’s efficacy. The selection of studies with low risk of bias bolsters the credibility of the results, as these studies employed rigorous methodologies to control for confounding variables. Additionally, the use of RR and CI allows for nuanced interpretations of TEs, facilitating a clearer understanding of the benefits and limitations of cemiplimab across diverse patient populations. The statistical techniques employed, including the sensitivity analysis, highlight the importance of robust data evaluation, which can significantly influence clinical decision-making and treatment guidelines (12).

In the context of existing literature, the findings of this analysis resonate with previous studies that have explored the relationship between PD-L1 expression and treatment outcomes in various cancers. For instance, similar conclusions have been drawn regarding the predictive value of PD-L1 in melanoma and other malignancies treated with immune checkpoint inhibitors (13). The consistent observation of PD-L1’s role as a predictive biomarker across different cancer types affirms the necessity for its evaluation in clinical practice. This body of evidence supports the clinical integration of PD-L1 testing as a standard approach to optimizing therapeutic strategies, tailoring treatment based on individual tumor characteristics, and ultimately enhancing patient outcomes.

The clinical implications of these findings are profound, as they advocate for a precision medicine approach in treating advanced NSCLC. The identification of PD-L1 as a pivotal biomarker not only aids in predicting treatment efficacy but also informs on the potential to personalize immunotherapy regimens. Such advancements in biomarker identification can lead to improved patient stratification, whereby individuals with higher PD-L1 expression may be prioritized for cemiplimab treatment, optimizing resource allocation and maximizing therapeutic benefit. Furthermore, understanding the biological underpinnings of PD-L1 expression in relation to immune evasion mechanisms provides valuable insights for future research endeavors aimed at overcoming resistance to immunotherapy and enhancing treatment outcomes in a broader range of patients (14).

The findings of this meta-analysis suggest that cemiplimab significantly improves OS and PFS in advanced NSCLC patients without EGFR, ALK, or ROS1 alterations, particularly with PD-L1 expression levels ≥50%. This aligns with previous studies, such as Gogishvili et al. [2022] and Baramidze et al. [2024], which also reported enhanced survival outcomes associated with cemiplimab treatment (11,14). However, the RR reported in this analysis (1.9–2.11) contrasts with the findings of Ricciuti et al. [2024], which indicated a lower RR of 1.5 (15). Such discrepancies may arise from the inclusion criteria and differing patient demographics across studies, highlighting the need for standardized approaches in future trials to yield more consistent results.

Furthermore, the low risk of bias observed in 71.4% of the studies included in this meta-analysis contrasts with some existing literature that raises concerns regarding methodological limitations in immunotherapy trials. Previous studies have indicated that methodological rigor is essential to ensure the reliability of outcomes, as seen in the work of Grace de Lima et al. [2023], which emphasizes the importance of controlling for confounding factors in cancer research (13). The current analysis reinforces the significance of robust methodological frameworks in yielding credible results that can be reliably interpreted. The observed improvements in OS and PFS with cemiplimab provide a compelling case for its clinical use, although the variability in study designs necessitates further investigation to elucidate the underlying mechanisms and optimize treatment protocols.

The sensitivity analysis conducted revealed that the removal of the Naik et al. [2021] study led to an increased TE, suggesting that outlier results can substantially affect overall findings (16). This points to the necessity of careful consideration when interpreting data from heterogeneous studies. The impact of this study on the overall efficacy of cemiplimab treatment underscores the potential influence of individual studies on meta-analytic conclusions. The existing literature also supports the notion that certain studies may disproportionately skew results, as highlighted in the work of Hirano et al. [2023] on the importance of addressing study variability in cancer research (17). Future research should aim to investigate the characteristics of studies that yield significantly divergent results to better understand their contributions to the literature and refine therapeutic strategies for NSCLC patients.

In conclusion, although this meta-analysis presents compelling evidence supporting the use of cemiplimab in advanced NSCLC, the identified discrepancies with existing literature draw attention to the complexities in interpreting clinical trial outcomes. The findings underscore the need for ongoing research to explore the clinical implications of PD-L1 expression as a biomarker for treatment response, as discussed by Liauw et al. [2025], which highlights the potential for personalized approaches in cancer therapy (18). Additionally, the implications of these findings extend beyond immediate clinical applications, as they may guide the design of future studies aimed at refining therapeutic regimens for NSCLC patients. The integration of findings from this meta-analysis into broader clinical practice could ultimately enhance patient outcomes and advance the field of oncology.

The present meta-analysis encounters several limitations that warrant careful consideration. Firstly, the overall sample size across the eight studies included, while providing some statistical power, remains relatively small and heterogeneous. This variability particularly arises from differences in patient selection criteria and treatment protocols among the studies, which may undermine the robustness of the statistical results. Consequently, the effectiveness of cemiplimab may not be universally applicable across diverse clinical populations, limiting the generalizability of our findings to broader patient demographics. Furthermore, the follow-up duration in several studies was insufficient to adequately assess the long-term efficacy and safety of cemiplimab. Short follow-up periods could lead to an underestimation of adverse events and long-term survival benefits, thereby impacting a comprehensive understanding of treatment outcomes.

Another limitation pertains to potential selection bias within the studies, especially regarding genetic status, PD-L1 expression levels, and prior treatment histories. Such biases could affect the external validity of our results, making it challenging to apply findings to different clinical settings. Additionally, the absence of appropriate control groups in some studies complicates direct comparisons between cemiplimab and alternative treatment options, further hindering a clear evaluation of its relative efficacy. Inconsistencies in data reporting and analysis methods, particularly concerning adverse event documentation, raise concerns about the reliability of the safety profile associated with cemiplimab. Lastly, sensitivity analyses revealed that the exclusion of certain studies significantly influenced the overall results, indicating that individual studies may disproportionately affect the conclusions drawn. Addressing these limitations in future research would enhance the understanding and application of cemiplimab in clinical practice, paving the way for more robust and conclusive findings.

The findings of this meta-analysis underscore the significant efficacy of cemiplimab as a treatment intervention for patients with NSCLC. The evidence indicates notable improvements in OS, PFS, ORR, and DCR, suggesting that the inclusion of cemiplimab in treatment protocols can lead to enhanced clinical outcomes. These results may prompt oncologists to consider incorporating cemiplimab as a frontline therapy for NSCLC, potentially altering existing treatment guidelines to reflect its demonstrated benefits. Moreover, the implications extend beyond immediate clinical applications; the study highlights the need for ongoing monitoring of quality of life assessments and adverse event occurrences associated with cemiplimab, ensuring a comprehensive understanding of its impact on patient well-being.

This study has several limitations. First, the primary objective of our meta-analysis was to evaluate the overall efficacy of cemiplimab in biomarker-defined subgroups of advanced NSCLC, rather than to compare it against a specific type of control. Second, the number of available RCTs was limited (only seven studies), and further stratifying by comparator type would have resulted in very small subgroups with insufficient statistical power to draw meaningful conclusions. Third, we employed a random-effects model to account for potential heterogeneity arising from differences in comparator arms, and we conducted sensitivity analyses to assess the impact of individual studies on the pooled estimate. Given the small number of included studies (n=7), the funnel plot and Egger’s test should be interpreted with caution, as they have limited statistical power to detect publication bias in meta-analyses with few studies.

Future research should explore the long-term effects of cemiplimab in diverse populations, particularly in those with varying biomarker profiles, to determine the consistency of its efficacy across different patient demographics. Additionally, studies examining the synergistic effects of cemiplimab with other therapeutic agents could further elucidate its role within combination therapies in NSCLC. Such investigations will be crucial in validating the applicability of the findings and addressing any gaps in current knowledge regarding optimal treatment strategies.

Conclusions

This meta-analysis provides compelling evidence supporting the use of cemiplimab in NSCLC management, emphasizing its ability to improve survival outcomes and disease control. The implications for clinical practice are substantial, suggesting a shift in therapeutic approaches that prioritize this intervention. The study not only advances understanding in the realm of NSCLC treatment but also lays the groundwork for future research aimed at refining treatment protocols and maximizing patient benefit in this challenging disease landscape.

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.