OBJECTIVES: Immune checkpoint inhibitor (ICI)-containing treatment is currently prescribed as first-line treatment for all patients with advanced non-small cell lung cancer (NSCLC) without targetable driver mutations. However, only 30-45% of patients show no progression within 12 months after treatment start. Various biomarkers are being studied to save costly and potentially harmful treatment in non-responders. We evaluated the cost-effectiveness of implementing a hypothetical predictive biomarker for ICI-containing treatment response compared with standard of care (e.g., no implemented biomarker) for pembrolizumab-containing treatment in patients with advanced NSCLC in the Netherlands. MATERIALS AND METHODS: Standard-of-care-based and predictive-biomarker-based strategies were compared using Markov models for three first-line pembrolizumab-containing treatments depending on a patient's tumor programmed cell death ligand-1 (PD-L1) expression and histology. A Dutch healthcare system perspective was adopted. Assuming a receiver operating characteristic-area under the curve of 1.0 in identifying responders, alternative treatments were offered for non-responders in the predictive-biomarker-based strategy. Parameters and assumptions were based on real-world data from surveys, literature using a targeted search, expert opinion, and registries. Outcomes included differences in costs, survival (life years (LYs)), and survival corrected for health-related quality of life (QoL) quality-adjusted life-years (QALYs) between the predictive-biomarker- and standard-of-care-based strategy. RESULTS: Implementing a predictive biomarker in pembrolizumab-carboplatin-paclitaxel treatment led to a mean survival reduction of 24 days (- 0.067 LYs) (18 days corrected for QoL (- 0.049 QALYs)), with cost savings of €22,606 compared with standard of care. Pembrolizumab monotherapy and pembrolizumab-pemetrexed-platinum treatments showed survival reductions of 4.5 and 3.9 months, respectively (3.6 and 2.8 months corrected for QoL), with cost savings of €24,345 and €28,456. Sensitivity analyses confirmed consistent cost savings and survival reductions. Survival losses were mainly observed due to the lower survival rates associated with the alternative first-line treatment options available for non-responders in the predictive-biomarker-based strategy within each pembrolizumab-containing treatment regimen. Pembrolizumab-carboplatin-paclitaxel treatment also showed survival gains under certain conditions related to QoL and survival estimates. CONCLUSIONS: Our study highlights the importance of careful de-implementation of ICI-treatments in advanced NSCLC, balancing costs reductions and side effects without comprising survival. In the pembrolizumab-carboplatin-paclitaxel treatment regimen, the survival loss could be considered negligible. Future research should define acceptable tradeoffs and thresholds for de-implementation, considering factors such as survival of alternative treatments and responder classification to guide predictive biomarker implementation and optimize health resource allocation.