Lung cancer (LC) is one of the most common cancers worldwide, with an estimated 2.5 million new cases and 1.8 million deaths reported globally in 2022.1 In Europe, the incidence per 100,000 people is 33.3–49.0 among men and 11.6–26.8 among women.1 The most recent mortality figure available in Spain is from 2023, with a total of 22,827 deaths, including 16,582 deaths among men and 6,245 among women.2 Although the incidence of LC among men is decreasing in Spain, it remains the leading cause of cancer-related death. However, in women, both incidence and mortality doubled between 2002 and 2020.3

Lung cancer is typically diagnosed between the ages of 55 and 75 years.3 Smoking remains the primary cause in most patients (71%). Approximately 80% of cases occur in men and 20% in women.4

Lung cancer is broadly classified into 2 major groups: non-small-cell lung cancer (NSCLC), accounting for approximately 85% of cases, and small-cell lung cancer (SCLC).4 Within NSCLC, a distinction is made between squamous and non-squamous histological subtypes, the latter corresponding mainly to adenocarcinoma.

One of the most important prognostic factors is the stage. 60–70% of NSCLC cases are diagnosed at stage IV. The 5-year survival rate is 61.2% for localized NSCLC, 33.5% for regionally advanced NSCLC, and 7% for metastatic NSCLC.

Traditionally, NSCLC has been treated with surgery, radiotherapy and chemotherapy; however, in recent years, immunotherapy and targeted therapies directed against specific mutations have been incorporated into clinical practice guidelines.

Currently, the standard of care for advanced and metastatic NSCLC without actionable mutations is PD-L1 inhibitors.5 The National Comprehensive Cancer Network (NCCN) guidelines recommend durvalumab as first-line treatment for patients with unresectable stage III disease whose disease has not progressed following chemoradiotherapy. For stage IV disease, pembrolizumab, atezolizumab, cemiplimab or tislelizumab are recommended for tumors with PD-L1 ≥ 50% in non-squamous NSCLC; pembrolizumab or cemiplimab combined with chemotherapy for tumors with PD-L1 > 1%; and finally, tislelizumab is recommended for squamous NSCLC regardless of PD-L1 expression.5

The cost of NSCLC treatment continues to rise year after year, partly driven by the widespread use of immune checkpoint inhibitors (ICIs). However, there remains considerable uncertainty as to which agent offers the greatest benefit in relation to its real-world cost.6 Cost-effectiveness studies help to assess the incremental costs and benefits associated with a given therapy.

Previous studies comparing the effectiveness and cost-effectiveness of ICIs have generally relied on model-based analyses incorporating structural assumptions and extrapolations from clinical trial data, which limits their external validity.7

The incremental cost-effectiveness ratio (ICER) reflects the additional cost of one treatment over another for a given improvement in effectiveness. Variations in ICER depend on costs per patient-month, which may vary according to factors such as dose, administration frequency and vial utilization, as well as on effectiveness, assessed by outcomes such as overall survival (OS) and progression-free survival (PFS).7

The objective of this study was to evaluate the effectiveness and efficiency of 4 ICIs (atezolizumab, durvalumab, nivolumab, and pembrolizumab) in patients with NSCLC in a real-world clinical setting.

A retrospective, single-center observational study was conducted. Patients diagnosed with NSCLC at La Princesa University Hospital (Madrid, Spain) were included. Inclusion criteria were being aged 18 years or older, having received at least one dose of treatment with an ICI (atezolizumab, durvalumab, nivolumab, or pembrolizumab) between January 2016 and December 2021, and having sufficient information in the electronic patient records (EPRs) to meet the study objectives. Patients who were participating in any clinical trial during the recruitment period were excluded. The follow-up of enrolled patients lasted until December 2023.

Demographic (age, sex and body mass index) and clinical characteristics (tumor histology and stage, genetic mutations, ECOG performance status, date of diagnosis, PD-L1 expression, LDH and C-reactive protein levels, neutrophil-to-lymphocyte ratio at the start of treatment, and adverse reactions) were collected from the EPR. Treatment-related variables included the drug administered, line of therapy, dose, number of cycles, concomitant treatment, time on treatment, delays in administration, reason for treatment discontinuation, disease progression and death.

The economic evaluation was conducted in accordance with methodological recommendations and the CHEERS 2022 reporting guidelines8 for health economic evaluations (Appendix 1). The economic evaluation was performed as a cost-effectiveness analysis from the perspective of the Spanish National Health System as the healthcare provider. Only drug costs were included as direct costs, as all other costs (monitoring, adverse reactions, etc.) were assumed to be similar among the drugs included in the study. For this calculation, the fixed doses recommended in the Summary of Product Characteristics were used as the reference, taking vial sharing into account, as this is considered standard practice in hospital pharmacy departments. The cost of the medications was obtained from Nomenclátor Web using the official ex-factory price (PVL, precio de venta de laboratorio) and expressed as cost per patient per 4 weeks based on fixed-dose regimens (Table 1). Indirect costs were excluded from the analysis.

The ICER was calculated as the ratio of the difference in costs to the difference in effectiveness between 2 treatments: ICER = (cost/patient for Option A – cost/patient for Option B) / (effectiveness of A – effectiveness of B). The ICER was calculated using OS as the measure of effectiveness, as it provides greater consistency compared with PFS.

Given the uncertainty associated with some variables, and the assumptions inherent in any economic analysis, a one-way sensitivity analysis was performed on those variables. The variables analyzed were the average cost per patient-year and the effectiveness achieved.

Statistical analysis was performed using the R Core Team software (2024). A 95% confidence level was used for hypothesis testing and confidence intervals. Analyses were performed on the available cases. To describe quantitative variables, the mean and standard deviation were used for normally distributed variables, and the median and interquartile range for those that were not. To describe qualitative variables, the number and percentage of patients were used. To compare groups of variables, the t-test or ANOVA were used if the quantitative variable followed a normal distribution. For qualitative variables, the chi-square test or Fisher's exact test was used, depending on the type of data. Analysis of both OS and PFS was performed using Kaplan–Meier curves and Cox regression. A p-value of <0.05 was considered statistically significant.

This economic evaluation was based exclusively on the observed survival data available from the study cohort, without extrapolating outcomes beyond the follow-up period. Parametric extrapolation was not performed because the follow-up duration was limited and the survival data were not sufficiently mature to support a reliable fit of standard parametric models.

A total of 237 patients with NSCLC were included: 25 (10.6%) were treated with atezolizumab, 21 (8.9%) with durvalumab, 74 (31.2%) with nivolumab, and 117 (49.4%) with pembrolizumab. A total of 65% of all patients were men, with a median age at baseline of 68.8 years (SD 9.7) (Table 2). Among patients with stage IV disease, 16 were treated with atezolizumab, 58 with nivolumab, and 94 with pembrolizumab. Among patients with stage III disease receiving palliative treatment, 5 were treated with atezolizumab, 16 with nivolumab, and 20 with pembrolizumab. A total of 16 patients receiving durvalumab with curative intent were also included.

Among patients with stage IV disease, 16 received atezolizumab, with a median PFS of 3.29 months (95% CI: 0.90–4.89), 58 received nivolumab, with a median PFS of 4.85 months (95% CI: 3.12–6.48), and 94 received pembrolizumab, with a median PFS of 6.86 months (95% CI: 4.80–8.55) (p = 0.397). The median OS was 9.24 months (95% CI: 5.72–11.90) for atezolizumab, 10.26 months (95% CI: 4.08–16.43) for nivolumab, and 15.49 months (95% CI: 8.99–21.98) for pembrolizumab (p = 0.60) (Table 3).

Among patients with stage III disease receiving palliative treatment, 5 received atezolizumab, with a median PFS of 7.41 months (95% CI: 0–16.54), 16 received nivolumab, with a median PFS of 6.49 months (95% CI: 4.80–8.02), and 20 received pembrolizumab, with a median PFS of 30.77 months (95% CI: 0–70.35) (p = 0.024). The median OS was 10.62 months (95% CI: 0–26.48) for atezolizumab, 17.03 months (95% CI: 4.79–29.27) for nivolumab, and 34.64 months (95% CI: 21.19–40.09) for pembrolizumab p = 0.017) (Table 4).

In addition, 16 patients receiving durvalumab with curative intent were included, with a median PFS of 26.22 months (95% CI: 15.52–32.47) and median OS of 30.18 months (95% CI: 22.34–37.54).

The ICER for OS in patients with stage IV disease was €517.92 per month of life gained for pembrolizumab versus atezolizumab and €512.66 per month of life gained for pembrolizumab versus nivolumab. Furthermore, the ICER for OS in patients with stage III disease was €135.76 per month of life gained for pembrolizumab versus atezolizumab and €152.26 per month of life gained for pembrolizumab versus nivolumab.

The sensitivity analysis, using the lower and upper bounds of the CI for OS, yielded results similar to those obtained using the median values. The ICER for OS in patients with stage IV disease, using the lower bound of the CI, was €989.90 per month of life gained for pembrolizumab versus atezolizumab and €546.07 per month of life gained for pembrolizumab versus nivolumab. The ICER for OS in patients with stage IV disease, using the upper bound of the CI, was €321.13 per month of life gained for pembrolizumab versus atezolizumab and €483.10 per month of life gained for pembrolizumab versus nivolumab.

The ICER for OS in patients with stage III disease, using the lower bound of the CI, was €152.76 per month of life gained for pembrolizumab versus atezolizumab and €163.49 per month of life gained for pembrolizumab versus nivolumab. The ICER for OS in patients with stage III disease, using the upper bound of the CI, was €237.84 per month of life gained for pembrolizumab versus atezolizumab and €250.09 per month of life gained for pembrolizumab versus nivolumab.

As nivolumab is expected to be the first ICI for which a biosimilar would become available, a sensitivity analysis assuming a 50% price reduction for this drug was performed. The ICER for OS in patients with stage IV disease for pembrolizumab versus nivolumab would be €1166.58 per month of life gained, while that in stage III disease would be €346.46 per month of life gained.

To date, no direct comparisons between the different ICIs for NSCLC are available, and there is insufficient evidence to determine which agent should be preferred as first-line treatment.

In our study, pembrolizumab showed the longest OS in stage IV disease, followed by nivolumab and atezolizumab. In stage III disease, pembrolizumab was also associated with longer OS, with a marked difference in PFS compared with the other agents. This finding is likely influenced by the fact that 15 of the 20 patients treated with pembrolizumab had PD-L1 expression ≥ 50%.

In Spain, as in most European countries, no official cost-effectiveness threshold has been established to reflect society's willingness to pay per additional unit of effectiveness. The introduction of new therapies can significantly increase direct costs, potentially leading to inequalities in access among European countries, even after regulatory approval.9

Despite the higher costs of pembrolizumab, its lower ICER suggests that it could be cost-effective depending on the willingness-to-pay threshold applied. On the other hand, the introduction of nivolumab biosimilars could alter the economic landscape in the future. Nevertheless, cost reductions could also be achieved through the implementation of efficiency measures, such as weight-based dosing up to a maximum threshold above which a fixed dose would be administered, as recommended by the Madrid Regional Health Authority in 2024. Several factors may influence ICER estimates, including patient heterogeneity, variability in the notified prices reported by individual hospitals, and differences in individual treatment responses.

In a multicentre study conducted in Spain to evaluate the economic impact of nivolumab in NSCLC, the cost of nivolumab per patient was estimated at €19,910, and the cost per life-year gained at €110,026 (€77,557–€231,171), values exceeding the thresholds recommended by the WHO.10 Our OS results for nivolumab (10.26 months) are consistent with those reported in that study, which reported a median OS of 9.7 months.

In Italy, a real-world study was conducted to compare the effectiveness and cost-effectiveness of ICIs in second-line treatment of advanced NSCLC.11 The study included 1,607 patients treated with nivolumab, pembrolizumab and atezolizumab. The researchers reported that nivolumab and pembrolizumab were associated with poorer cost-effectiveness, as patients treated with these drugs achieved OS similar to that of atezolizumab, but at a higher cost. In our study, atezolizumab is also associated with a lower cost than the other drugs, but pembrolizumab showed better efficacy, with higher OS rates.

Regarding the study limitations, the limited sample size is noteworthy, as it prevents stratification of patients by line of treatment for the calculation of treatment effectiveness, which may introduce significant bias into both the effectiveness results and the economic analysis. Specifically, in stage III, all patients treated with atezolizumab, 81% of those treated with nivolumab, and 61% of those treated with pembrolizumab received treatment in the second line or beyond, whereas all patients treated with durvalumab received it as consolidation therapy with curative intent. Efficacy would be expected to be lower in later lines of treatment than in the first-line setting.

The analysis did not include parametric survival extrapolation because the immaturity of the data did not permit a reliable model fit. As a result, the area under the survival curve may be underestimated. This represents a major limitation, particularly in the context of immunotherapy, where a subgroup of patients may achieve prolonged survival.

Another limitation is that factors such as patients' quality of life and associated indirect costs—such as hospitalization or serious adverse events—were not taken into account. Had these data been available, they could have provided a more comprehensive assessment of the economic impact of treatment. Several studies evaluating the costs associated with NSCLC management have shown that drug acquisition costs account for 75–80% of total healthcare expenditure.12 The exclusion of costs related to adverse events may therefore have led to an underestimation of the true economic burden, as the management of treatment-related toxicities has been shown to have a substantial economic impact.13 Furthermore, as noted by the ISPOR Drug Cost Task Force, the estimation of actual drug costs may be affected by the lack of transparency surrounding discounts and pricing agreements, introducing additional uncertainty into cost estimates.14

In our study, efficiency was assessed using direct drug acquisition costs alone; therefore, the overall budget impact may differ when other healthcare resource costs reported in the literature are taken into account. Taken together, these findings support the need to individualize treatment decisions based on tumor and patient characteristics, PD-L1 expression, and associated costs. This study provides a basis for informing therapeutic decision-making in routine clinical practice, while also highlighting the need for multicentre studies with larger patient populations to validate these findings.

This study evaluates the cost-effectiveness of 4 immune checkpoint inhibitors for the treatment of NSCLC in routine clinical practice at a tertiary care hospital. The findings suggest that the drugs evaluated are cost-effective in Spain according to the thresholds proposed by the WHO. They also highlight the need to implement efficiency measures.

The authors of this article accept the responsibilities defined by the ICMJE, available at http://www.icmje.org/. This manuscript has not been submitted to other journals or conferences.

The study was approved by the Drug Research Ethics Committee of La Princesa University Hospital in July 2023 (protocol code MOR-ICI-001).

Ana Álvarez-Yuste: Writing – review & editing, Writing – original draft, Visualization, Validation, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. María Pérez-Abánades: Writing – review & editing, Supervision, Methodology, Conceptualization. Ramón Colomer-Bosch: Writing – review & editing, Supervision, Methodology, Formal analysis, Data curation, Conceptualization. Irene Cuadrado-Berrocal: Writing – review & editing, Supervision, Methodology, Data curation, Conceptualization. Jacobo Rogado-Revuelta: Writing – review & editing, Validation. Tomás Gallego-Aranda: Writing – original draft, Validation, Methodology, Conceptualization. Ainhoa Aranguren-Oyarzábal: Writing – review & editing, Validation, Formal analysis.

None declared.