Efficacy of ICIs vs. Chemotherapy in NSCLC

What is the effectiveness and safety of first-line immune checkpoint inhibitors (ICIs), as monotherapy or in combination, compared to platinum-based chemotherapy, with or without bevacizumab for people with advanced NSCLC, according to the level of PD-L1 expression

In advanced NSCLC patients with high PD-L1 expression, ICI monotherapy outperforms platinum-based chemotherapy for survival outcomes, though combination ICI regimens offer no additional advantage and safety profiles are similar.

Abstract

Immune checkpoint inhibitors yielded a survival benefit over platinum‐based chemotherapy primarily in patients with high PD-L1 expression. In one trial, pembrolizumab monotherapy in patients with a tumor proportion score of ≥50% achieved a median overall survival of 20.0 months compared with 12.2 months with chemotherapy (p = 0.0003). Two separate trials of durvalumab monotherapy in high PD-L1 populations reported median overall survival improvements of 1.8 to 3.4 months (p = 0.037 and p = 0.036, respectively). In contrast, combination regimens—such as durvalumab plus tremelimumab and sitravatinib plus nivolumab—did not consistently yield statistically significant overall survival benefits when compared with chemotherapy or monotherapy.

Reported safety outcomes were sparse. One study noted 0–1 dose‐limiting toxicities in a pembrolizumab plus chemotherapy regimen, and no study reported marked differences in grade 3–4 adverse events, treatment discontinuation, or treatment‐related deaths. Together, these findings indicate that first‐line immune checkpoint inhibitor monotherapy, especially with pembrolizumab or durvalumab, appears effective in extending overall survival among patients with high PD-L1 advanced non‐small cell lung cancer, whereas combination regimens have not demonstrated clear added benefit.

Methods

We analyzed 9 sources from an initial pool of 75, using 8 screening criteria. Each paper was reviewed for 6 key aspects that mattered most to the research question. More on methods

Papers identified with Elicit search

n = 75
Papers screened using: Population - Age and Stage, Treatment Line, Intervention, Comparator, PD-L1 Data, Study Design, Outcomes, Study Quality
n = 75
Papers screened out
n = 66
Papers included for extraction
n = 9

Paper search

Using your research question "What is the effectiveness and safety of first-line immune checkpoint inhibitors (ICIs), as monotherapy or in combination, compared to platinum-based chemotherapy, with or without bevacizumab for people with advanced NSCLC, according to the level of PD-L1 expression", we searched across all trials from the ClinicalTrials.gov corpus. We retrieved the 75 trials most relevant to the query.

Screening

We screened in sources based on their abstracts that met these criteria:

• Population - Age and Stage: Does the study focus exclusively on adult patients (≥18 years) with advanced/metastatic NSCLC?
• Treatment Line: Is the study conducted in a first-line treatment setting?
• Intervention: Does the study evaluate approved immune checkpoint inhibitors as monotherapy or in combination?
• Comparator: Does the study include a platinum-based chemotherapy (with or without bevacizumab) control arm?
• PD-L1 Data: Does the study report PD-L1 expression levels?
• Study Design: Is the study either a randomized controlled trial (RCT) or a systematic review/meta-analysis of RCTs?
• Outcomes: Does the study report at least one efficacy outcome (overall survival, progression-free survival, or response rates) OR safety outcome (adverse events or treatment discontinuation)?
• Study Quality: Is the study design something other than a case report, case series, non-comparative observational study, or Phase I trial?

We considered all screening questions together and made a holistic judgement about whether to screen in each paper.

Data extraction

We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.

• Study Design:

Identify and extract the specific study design details:

• Type of study (e.g., randomized controlled trial, phase III clinical trial)
• Randomization method (if applicable)
• Blinding status (if mentioned)
• Multi-center or single-center study
• Geographic locations of study sites

If information is incomplete or unclear, note "Not fully reported" and provide any available partial information. If multiple design elements are present, list all relevant details.

• Participant Characteristics:

Extract detailed participant information:

• Total sample size
• Age range/mean age
• Gender distribution
• Specific NSCLC characteristics (histology, stage)
• PD-L1 expression levels (specific percentage or category)
• Key inclusion/exclusion criteria

For numeric data, include both the number and percentage where possible. If ranges or means are provided, extract both. If data is incomplete, note which specific characteristics are missing.

• Intervention Specifics:

Detail the specific interventions:

• Name of immune checkpoint inhibitor(s)
• Dosage and administration schedule
• Combination therapy details (if applicable)
• Duration of intervention
• Whether used as monotherapy or in combination

Be precise about drug names, doses, and frequency. If multiple intervention arms exist, extract details for each. Use exact terminology from the source document.

• Comparison/Control Condition:

Describe the control or comparator treatment:

• Type of treatment (e.g., platinum-based chemotherapy)
• Specific chemotherapy regimen
• Dosage and administration details
• Whether bevacizumab was included

If multiple control groups exist, extract details for each. If control group specifics are not fully described, note "Insufficient details provided".

• Primary Outcome Measures:

Extract primary outcome measures:

• Specific outcomes measured (e.g., overall survival, progression-free survival)
• Measurement time points
• Reported results with statistical significance
• Confidence intervals or p-values (if provided)

Prioritize outcomes directly related to effectiveness and safety. If multiple primary outcomes are reported, extract details for all. Use exact numerical values and statistical indicators from the source.

• Clinical significance assessment of primary endpoint results::

Provide a brief assessment of whether the statistically significant results translate to clinically meaningful benefit:

• Compare effect size to established clinically important differences in the field
• Consider practical impact on patient outcomes relative to standard of care
• Note any limitations that affect real-world applicability

Format: "[Clinically meaningful/Marginal benefit/Statistically significant only] - [brief rationale]"

Remember that citations are formatted claim, and rules around special characters in the [] still apply.

Results

Characteristics of Included Studies

Effects

Efficacy by PD-L1 Expression Levels

Safety and Adverse Events

Comparative Analysis

Immune Checkpoint Inhibitor Monotherapy vs Chemotherapy

• Pembrolizumab monotherapy (Merck Sharp & Dohme LLC, 2023) was associated with a substantial and statistically significant overall survival benefit over platinum-based chemotherapy in advanced non-small cell lung cancer, particularly in patients with high PD-L1 expression (tumor proportion score ≥50%).
• Durvalumab monotherapy (AstraZeneca, 2025, "PEARL" and "Phase III Open Label Study") demonstrated a statistically significant but more modest overall survival benefit in high PD-L1 populations.
• In lower PD-L1 or unselected populations, the benefit of immune checkpoint inhibitor monotherapy was less pronounced or not statistically significant in these studies.

Immune Checkpoint Inhibitor Combinations vs Standard Treatment

• Combination regimens (durvalumab plus tremelimumab, sitravatinib plus nivolumab) were not found to consistently outperform chemotherapy or immune checkpoint inhibitor monotherapy in these studies.
• In the NEPTUNE study, durvalumab plus tremelimumab improved overall survival by 2.6 months in the blood tumor mutational burden ≥20 group, but this was not statistically significant.
• In the Phase III Open Label study, the combination arm did not show overall survival or progression-free survival benefit over chemotherapy.
• Sitravatinib plus nivolumab improved overall survival by 1.6 months compared to docetaxel, but this was not statistically significant.
• Adding canakinumab to pembrolizumab plus chemotherapy did not improve progression-free survival or overall survival.

Impact of PD-L1 Expression and Biomarkers

• The benefit of immune checkpoint inhibitor therapy was most pronounced in patients with high PD-L1 expression, as reported in studies of pembrolizumab and durvalumab monotherapy.
• Combination regimens have not demonstrated clear added value in these settings.
• Evidence for immune checkpoint inhibitor use in lower PD-L1 or biomarker-negative populations was less robust in these studies.

Limitations

• Early termination and sample size: Several studies were terminated early or had small sample sizes (Regeneron Pharmaceuticals and Sanofi, OSE Immunotherapeutics), which limits interpretability.
• Incomplete safety reporting: We didn’t find mention of safety and adverse event data in most studies, which limits assessment of tolerability.
• Open-label design: Most studies were open-label, which may introduce bias.
• Heterogeneity of populations: Some studies focused on post-immune checkpoint inhibitor or later-line settings, which may not be directly comparable to first-line studies.
• Biomarker selection: The generalizability of findings may be limited by biomarker selection and study populations.