Treatments for metastatic non-small cell lung cancer: a pathways pilot with systematic review, evidence synthesis and economic model.

Background: National Institute for Health and Care Excellence technology appraisals assess the effectiveness and cost-effectiveness of medicines at a single point in the treatment pathway. However, for some disease areas, such as non-small cell lung cancer, there are many recommendations, making it difficult to use National Institute for Health and Care Excellence guidance. The treatment pathway for metastatic stage 4 non-small cell lung cancer can be divided into decision points (nodes) based on histology (squamous or non-squamous), programmed death-ligand 1 expression, presence of tumour mutations and line of therapy. The National Institute for Health and Care Excellence commissioned this pilot to assess the potential of taking a 'pathways approach' to technology appraisals. The aim was to build a single disease-specific cost-effectiveness model for metastatic stage 4 non-small cell lung cancer patients not eligible for targeted therapies at first line, that can be updated with economic and clinical data as required. Methods: We conducted a systematic review (searches last updated 11 July 2025) and network meta-analysis of treatment efficacy and safety at each decision node in the pathway. We used flexible fractional polynomial models for primary outcome progression-free survival, required for a model of treatment sequences. We built a novel cost-effectiveness model that compared sequences of treatments, and was populated using network meta-analyses for progression-free survival, data on overall survival after last-line therapy, evidence on treatment sequences from an analysis of systemic anticancer therapy data, and quality of life, cost and resource use estimates from previous technology appraisals. Drug list prices were used, but confidential discounts are available. Results: We included 15 randomised controlled trials and 1 single-arm study in the review, judged as some concerns or low risk of bias. Immunotherapies, in combination with doublet platinum chemotherapy, were most effective first-line treatments, although with higher adverse event rates. Immunotherapy monotherapies were most effective at second line, unless patients were suitable for targeted therapies. Sequences starting with atezolizumab + bevacizumab + doublet platinum chemotherapy had similar costs and quality-adjusted life-years to sequences starting with pembrolizumab + doublet platinum chemotherapy. Sequences starting with pemetrexed + platinum chemotherapy had the lowest costs but also the lowest total quality-adjusted life-years. Sequences for non-squamous non-small cell lung cancer, programmed death-ligand 1 ≥ 50%: Sequences starting with pembrolizumab + doublet platinum chemotherapy had highest quality-adjusted life-years, but higher costs compared to other sequences. Sequences starting with pemetrexed + platinum chemotherapy had the lowest cost and the lowest number of quality-adjusted life-years. Sequences for squamous non-small cell lung cancer, programmed death-ligand 1 < 50%: Sequences starting with pembrolizumab + doublet platinum chemotherapy had higher quality-adjusted life-years and higher costs than sequences starting with platinum chemotherapy. Sequences for squamous non-small cell lung cancer, programmed death-ligand 1 ≥ 50%: Sequences starting with atezolizumab had the highest predicted quality-adjusted life-year gains, slightly higher than for sequences starting with pembrolizumab. Sequences starting with platinum chemotherapy had the lowest quality-adjusted life-years, but the lowest total costs. Patients in the systemic anticancer therapy analysis had a shorter time on treatment than those in trials, resulting in lower treatment costs and causing the immunotherapy sequences to appear more cost effective. Conclusions: Our model can be used to estimate either the most cost-effective sequence of treatments or the most cost-effective treatment at a given point in the pathway, although our results are based on drug list prices and these would need to be updated to draw conclusions about the relative cost-effectiveness of different treatment sequences. We were able to use real-world data from systemic anticancer therapy to estimate model parameters and sequences that reflect clinical practice. Our model can readily be updated and used as a reference model for metastatic non-small cell lung cancer. Study registration: The study is registered as PROSPERO CRD42023470119. Funding: This award was funded by the National Institute for Health and Care Research (NIHR) Evidence Synthesis programme (NIHR award ref: NIHR136097) and is published in full in