This study investigated alternative dosing regimens for ICIs, including weight-based, fixed, and varied dosing frequencies and vial sharing, to reduce economic and environmental impact while maintaining efficacy and safety. Although the analysis did not adopt a country-specific healthcare payer perspective, the cost and emission estimates were primarily derived from Dutch sources and reflected a Western European hospital context. A user-friendly application was developed to allow tailoring of dosing strategies based on virtual populations and location-specific variables. Our results demonstrate that alternative dosing based on FDA criteria can reduce costs by up to 28% and carbon emissions by up to 30%. In comparison, dosing strategies guided by the therapeutic window yielded cost reductions of up to 69% and emission reductions of up to 63%. Furthermore, vial sharing provided additional cost savings of up to 31% when used in conjunction with the lowest-cost dosing regimens, as summarized in Tables 2 and 3. These findings suggest that implementing alternative dosing strategies can reduce costs and environmental impact without compromising therapeutic efficacy.

Numerous studies have investigated alternative dosing regimens for atezolizumab, nivolumab, and pembrolizumab [1, 13,14,15, 18, 19, 40]. Our findings indicate that alternative dosing regimens for atezolizumab and nivolumab can substantially reduce costs, which may explain the extensive research focus on these agents. For instance, a study of nivolumab dosing reported a 66% cost reduction with Cmin,ss levels above the threshold in over 90% of patients, which aligns with our results showing a 47% cost reduction and over 90% of patients achieving target concentrations [15]. Similarly, a recent study in real-world patients suggested that annual drug costs with atezolizumab could be reduced from €68,153 to €16,200 or €14,007 per patient, depending on the de-escalation strategy (i.e. shifting from 1200 mg Q3W to 1200 mg every 12 weeks, or alternatively 92 mg Q3W), while maintaining trough levels above the target of 6 mg/L [41, 42]. In contrast, avelumab and durvalumab have been less thoroughly examined. The high target Cmin of durvalumab of 50 mg/L, potentially due to differences in binding affinity [43] or exposure target determination, and shorter half-life (21 vs 27 days for atezolizumab) limit opportunities for cost savings [12, 16]. Similarly, the shorter half-life of 6 days for avelumab restricts the extension of dosing intervals, making dose reductions necessary for cost savings (Tables 2 and 3). Additionally, avelumab exposure–response studies have suggested that dose adjustments could affect efficacy [12], although this effect may be confounded by disease status, as a matched analysis found no significant difference between exposure groups [44].

4.1 Weight-Based Dosing Versus Fixed Dosing

The results of our study reinforce previous findings that highlight cost reduction as the primary driver of the transition from weight-based to flat dosing for ICIs [17]. Flat dosing regimens are particularly effective in minimizing drug waste when vial sizes align with the prescribed doses (Tables 1, 2, and 3). Notably, some studies report significant savings with weight-based dosing, with reductions of 16–24% for pembrolizumab and 14% for nivolumab [27, 45, 46]. However, these studies frequently overlooked the cost implications of unused drug inherent in weight-based dosing. Our vial sharing strategies, which minimize waste, demonstrate that weight-based dosing can be more cost efficient under such conditions. Given that hospitals bear the cost of entire vials rather than merely that of the used drug, our findings offer a more realistic cost representation without vial sharing. Consequently, fixed doses result in lower overall costs when accounting for vial-based drug wastage [27].

4.2 Decreased Dose Versus Increased Interval

For atezolizumab, the lowest-cost strategies involved extending dosing intervals and reducing doses, with maximum cost reduction achieved through a combination of these approaches, owing to its wide therapeutic window (therapeutic index 201 vs 26–137 for other drugs). Pembrolizumab also shows benefits from extended dosing intervals and dose reductions, though not in combination, likely due to its fixed 100 mg vial size, which restricts dose adjustments at higher intervals. For avelumab, 2-week dosing intervals were recommended because of its short half-life, with cost savings primarily achieved through dose reductions. In the case of durvalumab and nivolumab, recommended regimens aligned with approved intervals, with cost reductions driven mainly by substantial dose reductions, probably driven by the availability of multiple vial sizes and low Cmax thresholds, which limit the potential for high doses and extended intervals. In general, extended dosing intervals can reduce the logistical burden on healthcare centers and contribute to improved patient workforce participation by reducing the frequency of hospital visits, while also potentially improving efficacy by optimizing administration timing [47]. However, schedules of concurrent medications should be taken into consideration.

4.3 Vial Sharing

Vial sharing yielded modest to moderate cost savings (0–31%) across the most economical dosing regimens in Tables 2 and 3 (calculated as ((cost after vial sharing − cost before vial sharing) / cost before vial sharing) × 100%), with the most significant reduction for atezolizumab, where vial sharing made a previously more expensive regimen comparatively less costly (e.g., €58,435 for reference regimen vs €25,848 for 120 mg Q2W; €65,880 without vial sharing, Table S3 in ESM-2). Fixed-dosing regimens generally produced minimal benefit, as vial sizes often matched the required dose, whereas all weight-based regimens resulted in cost reductions. Consistent with previous findings [18, 27], savings from vial sharing were influenced by vial size differences and drug cost per milligram. Atezolizumab (available in 840 mg and 1200 mg vials) showed notable savings, whereas reductions were smaller for nivolumab (40 and 100 mg vials) and durvalumab (120 mg vials), as also clearly visible in Fig. 3. Pembrolizumab, despite its smaller 100 mg vial size, also had savings due to its high cost per milligram. Although most savings were <5%, they still led to significant absolute reductions [19]. Automated systems could further optimize vial sharing for patients with doses scheduled within a 7-day window of each other, reducing waste [20]. In practice, organizing clinics to implement vial sharing requires additional planning, infrastructure, and staff coordination. This could incur additional operational costs depending on the local setup. Simultaneously, the rise of fixed dosing may limit the effectiveness of vial sharing. Increasing the availability of vial sizes or introducing smaller vials, as suggested by prior studies [18, 19], could enhance cost efficiency, as shown by the impact of the now-discontinued pembrolizumab 50 mg vials [19].

4.4 Reduction in Carbon Emissions

The carbon emissions of alternative dosing strategies closely mirror cost patterns, with estimated reductions ranging from 0 to 63% (as can be seen in Tables 2 and 3 and in Table S3 in ESM-2). Although the potential environmental benefits are notable, they are only considered because they align with the overarching goal of maintaining effective treatment outcomes. Our predicted 29% reduction in CO2e emissions for pembrolizumab aligns with a previously reported 24% reduction [7]. The maximum total CO₂e savings across all five drugs range from approximately 39 to 501 kg CO₂e per patient cohort, which is roughly equivalent to the annual carbon footprint of 0.004–0.047 for average EU citizens (based on 10,700 kg CO₂e per capita in 2022) [48]. The importance of accounting for drug packaging is emphasized by the 1560 mg Q4W dosing regimen for atezolizumab, which, despite a small cost reduction, showed higher predicted emissions. Avelumab showed higher emissions than durvalumab and atezolizumab, whereas nivolumab resulted in higher emissions than pembrolizumab (Table 2). This is largely due to differences in drug volume and solution density: durvalumab (50 mg/ml) and atezolizumab (60 mg/ml) are denser than avelumab (20 mg/ml), and pembrolizumab (25 mg/ml) is denser than nivolumab (10 mg/ml).

Although we assumed production emissions to be constant, they may also influence outcomes. Additionally, our analysis did not account for emissions associated with infrastructure—such as facility construction, heating, or fill-and-finish processes—which may contribute modestly to total emissions. Furthermore, our carbon footprint estimates did not include upstream corporate emissions such as sales, general, and administrative costs, which may contribute substantially to pharmaceutical emissions according to recent studies [55]. Lastly, our analysis did not account for the indirect carbon rebound effect—that is, the emissions associated with reinvesting saved healthcare funds elsewhere in the system. In European countries, healthcare spending has been estimated to generate approximately 0.5 kg CO₂e per $US spent (∼€0.87) [49], although this figure is higher in countries with more carbon-intensive healthcare systems. As such, reinvested cost savings could partially offset the environmental benefits of dosing optimizations, depending on how and where the savings are utilized.

4.5 FDA Exposure Criteria

The results of this study reveal that some approved alternative dosing regimens do not meet FDA criteria (Table 2), a discrepancy that can be explored using the developed Shiny application. This may be due to differences in virtual populations and covariate distributions compared with previous simulations. More likely, pre-guideline simulations did not adhere strictly to FDA criteria, instead prioritizing efficacy with high Cmin or AUC values. Strategies that extend dosing intervals by doubling the dose and interval length maintain AUC values but result in higher Cmax values. These findings suggest that FDA guidelines may be excessively stringent, particularly concerning Cmax thresholds, which restricts the adoption of alternative dosing regimens, although the guidelines do specify that exceptions to the 125% rule for bioequivalence may be justified if the safety profile is adequately explained [21]. Additionally, finding alternative doses that adhere to FDA exposure guidelines offers the significant advantage of requiring no further clinical studies, allowing for direct clinical implementation.

4.6 Therapeutic Window Exposure Criteria

Conversely, using therapeutic windows as the exposure target allows for more flexible dosing regimens, offering greater potential for reducing both financial costs and environmental impact compared with FDA-based exposure criteria, as demonstrated in Fig. 2. However, the use of therapeutic windows is the subject of ongoing debate regarding the adequacy of Cmin targets. Several studies have used this target for alternative dosing strategies [12, 22,23,24,25], whereas others have argued that they are based on inadequate clinical endpoints and may be set too low [26]. Additionally, concerns arise about increased clearance due to target-mediated drug disposition and potential non-linear clearance at lower exposures [50]. Indeed, although two unrandomized studies demonstrated that nivolumab doses substantially lower than the labeled dose were equally as effective as the approved dose or more effective than no nivolumab [51, 52], a randomized study evaluating low-dose nivolumab at 0.3 mg/kg Q3W (approximately 20 mg for a patient weighing 70 kg) found lower survival than with more standard dosing [53]. Our simulations suggest that intermediate doses (e.g., nivolumab 180 mg Q2W or 340 mg Q3W) may maintain therapeutic exposure; however, given the uncertainty surrounding Cmin thresholds, widespread implementation of such regimens would require additional trials to demonstrate therapeutic equivalence. Notably, our cost-minimization analysis assumed equivalent efficacy and safety across compared regimens. However, our findings indicate that fewer patients meet Cmin thresholds after the first dose compared with steady state, as concentrations often fall below the Cmin threshold, even for approved doses. For drugs such as durvalumab and nivolumab, initiating treatment with a higher loading dose could enhance efficacy early on. Therapeutic drug monitoring and model-informed precision dosing are recommended to ensure exposures remain above Cmin or to individualize dosing intervals [12, 13, 54].

4.7 Study Limitations

The study's limitations arise largely from assumptions in the simulations. Treatment duration differences [35] and regional, financial, and logistical disparities could also influence outcomes. The analysis was based on currently approved branded products and did not account for the future entry of biosimilars, which could further reduce costs. Conversely, manufacturers may respond to reduced sales volumes from more efficient dosing by raising unit prices, potentially offsetting projected savings. Variations in virtual populations, including patient demographics such as weight, may affect the differences observed between flat and weight-based dosing strategies, potentially influencing the generalizability of results across regions. Simulations were based on current vial sizes and a group size of two to eight for vial sharing. To address all these issues and enhance adaptability to diverse settings, we created a Shiny app that enables users to customize simulations for specific regional, financial, and logistical contexts and change the underlying assumptions. Additionally, we did not account for change in clearance over time (reported average decrease of 22–30% [12, 28, 56,57,58]), potentially allowing for further dose reductions or increased intervals over treatment duration. This was because the magnitude and direction of time-varying clearance vary significantly between individual patients, as it is influenced by factors such as disease status, treatment response, and cancer type. Consequently, optimized dosing over time would likely require a more individualized approach. Moreover, concerns about dose reductions or interval extensions may arise due to faster clearance at lower doses, likely linked to target-mediated drug disposition. In such cases, the linear population pharmacokinetic model may overestimate low exposures. However, FDA-based dosing strategies are designed to match exposures, reducing this risk. For therapeutic window regimens, proposed dose adjustments should still maintain median steady-state trough concentrations above critical levels [12, 50]. Finally, our analysis assumed an 8-month treatment period. Although this aligns with current clinical practice, there is an opportunity to significantly reduce costs by shortening treatment durations. Such reductions are currently being evaluated in several clinical trials [59]. These studies will need to confirm whether a significant reduction in treatment duration can be achieved without compromising treatment efficacy.