Document resource
WHAT IS ALREADY KNOWN ON THIS TOPIC Although highly effective, many novel cancer therapeutics for early non-small cell lung cancer (NSCLC) prove to be prohibitively expensive. Our comprehensive cost–utility analysis, adopting a health system perspective in Sri Lanka and using an Low and middle income countries (LMIC) appropriate willingness-to-pay threshold (US$6700/quality-adjusted life-year), sought to identify economically viable options.WHAT THIS STUDY ADDS Our findings offer clear, policy-actionable evidence: only adjuvant osimertinib, alectinib and neoadjuvant nivolumab were determined to be cost-effective. Crucially, all other immune checkpoint inhibitor regimens assessed were highly cost-ineffective, requiring substantial price reductions to be considered for public funding.HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY This study identifies the most economically viable novel neoadjuvant and adjuvant therapies in stage II–III NSCLC which could be currently integrated into Sri Lanka’s state health sector. The findings further quantify the price reductions required for some regimens to be considered for sustainable funding.Introduction Lung cancer represents a significant public health challenge in Sri Lanka, with approximately 2000 new cases identified annually. 1 This makes it the second most common cancer among males in the country, although it is only the ninth most prevalent malignancy among females.1 Significantly, the incidence of lung cancer is not only rising steadily across both genders but is also outpacing the general increase in the incidence of all other cancers1. Lung cancer survival rates in Sri Lanka are disappointingly low.2 This is primarily due to the disease often being diagnosed at an advanced stage.2 Contributing factors also include the poor performance status of patients on presentation and limited access to innovative diagnostic tools and therapeutic treatments.3Early-stage non-small cell lung cancer (NSCLC) is potentially curable.4 Surgery is the standard of care for stages I–II, with stereotactic body radiotherapy increasingly being considered as a comparable alternative.5 Most stage IIIA NSCLC patients would receive chemoradiotherapy, with few undergoing curative surgery.4 Despite treatment, nearly 50% of early-stage NSCLC (stages I–II) and over 70% of stage III patients develop progressive disease, often distant metastases.4 Consequently, adjuvant systemic treatments have been investigated with a view to reducing cancer recurrence and improving survival. A large meta-analysis confirmed that adjuvant chemotherapy after surgery offers a 5% improvement in overall survival.6 Following publication of this landmark meta-analysis, adjuvant platinum-based chemotherapy has become standard of care in stage II–III NSCLC.Newer cancer treatments such as small molecule tyrosine kinase inhibitors and monoclonal antibodies that restore anti-tumour immunity have improved survival in the metastatic setting.4 Clinical trials now show their benefit as adjuvant or neoadjuvant treatments with surgery or chemoradiotherapy for early and locally advanced disease.4 The ADAURA and LAURA trials confirmed that adjuvant osimertinib helps lung adenocarcinoma patients with epithelial growth factor receptor (EGFR) mutations who received curative surgery and chemoradiotherapy, respectively.7 8 Adjuvant alectinib also improved survival for adenocarcinoma patients with anaplastic lymphoma kinase (ALK) mutations.9 For patients without these mutations, neoadjuvant nivolumab before surgery improved disease-free survival and overall survival in stage II-III NSCLC.10 Adjuvant pembrolizumab and atezolizumab after surgery also improved disease-free survival and overall survival.11 12 Nivolumab, pembrolizumab, and durvalumab have improved survival as perioperative therapy (neoadjuvant followed by adjuvant treatment) in stage II and III tumours undergoing curative surgery.13–15 For stage III disease treated with chemoradiotherapy, consolidation with durvalumab improved both disease-free survival and overall survival.16Regulatory approvals for these treatments have slight variations. Neoadjuvant and perioperative nivolumab is approved by the European Medicines Authority only for patients with over 1% PD-1 expression, while the Federal Drug Agency (FDA) of the USA approved it for all patients.4 Adjuvant atezolizumab is FDA-approved for patients with over 1% PD-1 expression, but the European Medicines Authority limited its approval to those with over 50% PD-1 expression.4Despite improved survival, these novel drugs are costly, posing a funding challenge for most health systems. Consequently, they are not included in the WHO essential drugs list.17 A Markov modelling-based cost–utility analysis and a budget impact analysis (BIA) were conducted to determine their cost-effectiveness in Sri Lanka from a health system perspective.Methods We performed a three-state Markov model ( figure 1) for the following treatments in stage II–III NSCLC using TreeAge Pro 2025:Adjuvant osimertinib following curative surgery for EGFR mutant adenocarcinoma.Adjuvant osimertinib following chemoradiotherapy for EGFR mutant adenocarcinoma.Adjuvant alectinib following curative surgery for ALK mutant adenocarcinoma.Neoadjuvant nivolumab for resectable NSCLC with expression of PD-L1 >1%.Adjuvant pembrolizumab after curative surgery for NSCLC.Adjuvant atezolizumab after curative surgery for NSCLC with expression of PD-L1 >50%.Perioperative treatment with pembrolizumab in resectable NSCLC.Perioperative treatment with durvalumab in resectable NSCLC.Consolidation treatment with durvalumab following chemoradiotherapy for stage III NSCLC.Figure 1Markov model with health states and possible transitions.The target population comprised adults diagnosed with stage II–III NSCLC, with a median age of 60 years. Sex-specific analyses were conducted for males and females to reflect differences in overall survival and recurrence outcomes, with results for females reported in the online supplemental appendix.SP110.1136/bmjonc-2025-001046.supp1Supplementary dataAlthough perioperative nivolumab is approved by the European Medicines Agency for NSCLC patients with PD-L1 expression >1%, we excluded it from this analysis since its HR was very similar to neoadjuvant nivolumab alone.10 15 Perioperative treatment requires significantly more doses of nivolumab with substantially higher costs, and it is unclear whether this treatment is superior to neoadjuvant alone, particularly in patients with PD-L1 expression >1%.10 15We adhered to a pre-established economic evaluation plan for this study and presented our results in accordance with the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) guidelines.18Model structure The cost-effectiveness model ( figure 1) features three mutually exclusive health states: disease-free, recurrence and death. Patients begin in the ‘disease-free’ state after definitive treatment. In each cycle, patients can stay ‘disease-free’ or progress to ‘recurrence’ or ‘death’. The ‘recurrence’ state signifies advanced disease, from which patients can only transition to ‘death’, with no possibility of returning to earlier states.The Markov model simulated a lifetime horizon using 1-month cycles, moving through three predefined health states. Quality-adjusted life-years (QALYs) were used to assess effectiveness, with both QALYs and costs discounted at an annual rate of 3%. The analysis was conducted from the perspective of the Sri Lankan public healthcare system.Data sources Transition probabilities Transition probabilities for moving from a disease-free state to death were determined using age-specific and sex-specific Sri Lankan population life tables. For patients receiving standard treatment, probabilities of transitioning from a disease-free state to recurrence were derived from published clinical trial outcomes. These inputs are summarised in table 1.19 20 For novel treatments, disease-free to recurrence transition probabilities were adjusted using HRs reported in the corresponding clinical trials.7–14 16 In the absence of sex-stratified recurrence estimates in the trial literature, post-recurrence mortality was adjusted using a female-to-male HR of 0.8 derived from external observational evidence, reflecting consistently better survival outcomes among women with lung cancer.21 In the absence of sex-specific estimates for disease-free to recurrence transitions, the same HR was applied for females.Table 1Parameters used in the Markov model (male population)Probabilistic sensitivity analysisParameter estimateSEDistributionReferenceAdjuvant treatment with tyrosine kinase inhibitorsTransition probabilities (monthly) Osimertinib after surgery Disease-free to recurrence (SC)0.0370.0025Beta7 HR0.170.03Normal7 Distant metastasis to death0.0230.0015Beta19 Osimertinib after chemoradiotherapy Disease-free to recurrence (SC)0.1180.0034Beta8 HR0.160.04Normal8 Distant metastasis to death0.0230.0015Beta19 Intervention: alectinib after surgery Disease-free to recurrence (SC)0.0190.0025Beta9 HR0.240.07Normal9 Distant metastasis to death0.0580.0008Beta20 Disease-free to death (All)Life tableBetaMonthly cost estimates (US$) Osimertinib after surgery First 3 years: intervention400102GammaLocal First 3 years: standard care0 Distance recurrence576GammaLocal Osimertinib after chemoradiotherapy First 2 years: intervention40068GammaLocal First 2 years: standard care0 Distance recurrence576GammaLocal Alectinib after surgery First 2 years: intervention2167221GammaLocal First 2 years: standard care0 Distance recurrence758GammaLocal Subsequent years (only 5 years)152GammaLocalAdjuvant and neoadjuvant treatment with immunotherapy Transition probabilities (monthly) Neoadjuvant nivolumab Disease-free to recurrence (SC)0.0580.0008Beta10 HR0.410.07Normal10 Perioperative pembrolizumab Disease-free to recurrence (SC)0.0370.0020Beta13 HR0.580.07Normal13 Perioperative durvalumab Disease-free to recurrence (SC)0.0270.0020Beta14 HR0.680.09Normal14 Adjuvant atezolizumab after surgery Disease-free to recurrence (SC)0.0190.0017Beta11 HR0.430.08Normal11 Adjuvant pembrolizumab after surgery Disease-free to recurrence (SC)0.0160.0017Beta12 HR0.760.07Normal12 Durvalumab after chemoradiotherapy Disease-free to recurrence (SC)0.1150.0037Beta16 HR0.510.06Normal11 Distant metastasis to death0.0580.0008Beta20 Disease-free to death (All)Life tableBetaCost estimates (US$) Neoadjuvant nivolumab First year: intervention91794GammaLocal First year: standard care0 Perioperative pembrolizumab First year: intervention4292438GammaLocal First year: standard care0 Perioperative durvalumab First year: intervention4042412GammaLocal First year: standard care0 Adjuvant atezolizumab after surgery First 3 years: intervention3167323GammaLocal First 3 years: standard care0 Adjuvant pembrolizumab after surgery First year: intervention4292438GammaLocal First year: standard care0 Durvalumab after chemoradiotherapy First year: intervention2208225GammaLocal First year: standard care0 Subsequent years (only 5 years)152GammaLocal Distance recurrence758GammaLocalUtility estimates Disease free0.780.04Beta24 Recurrence0.690.02Beta24SC, Standard Care.In metastatic EGFR or ALK mutant adenocarcinoma of lung, the median overall survival is 30 months, while for oncogene non-addicted NSCLC, it is 10 months.19 20 These estimates were used to calculate the transition probabilities from ‘distant recurrence’ to ‘death’.Transition probabilities reported on an annual basis were converted to monthly probabilities to match the model cycle length, assuming constant hazards over time and using standard rate-to-probability transformations.22Cost estimates A comprehensive breakdown of treatment expenses is detailed in the online supplemental appendix: Cost calculations and is summarised in table 1. The analysis adopted a health system perspective, encompassing the costs of treatments, investigations, administration and the human resource costs of healthcare professionals.The private sector market prices of novel therapeutic drugs were used for this analysis as these drugs are currently unavailable through the state health system. For osimertinib and alectinib, the average price of generic products was considered due to their significantly lower cost compared with the original formulations. A similar approach was taken for pembrolizumab, for which a biosimilar product is registered in Sri Lanka. However, for atezolizumab, durvalumab and nivolumab, only the original products are available, and their private market prices were therefore used in the analysis.For all other drugs and consumables, cost data were sourced from the Medical Supplies Division of the Ministry of Health. Other costs were determined using published local data. The costs of managing additional toxicities were also accounted for, as further elaborated in the online supplemental appendix.In our practice, patients are followed up for 5 years and the annual cost of the disease-free state comprises clinic visits and routine imaging in keeping current management protocols during this period. Because high-cost targeted therapies such as osimertinib, alectinib and immunotherapy agents are not currently funded for patients with advanced lung cancer within the state health sector, these treatments were excluded from the cost calculations for the recurrence state. Instead, the model only incorporated the costs of therapeutic interventions and supportive care services that are presently available and accessible to patients within the existing national health system.In the absence of published variance estimates for local cost inputs, uncertainty around cost parameters was modelled by applying a symmetric ±20% range around each point estimate. This range was selected as a pragmatic and conservative assumption to reflect plausible variation in local prices, procurement conditions and resource-use estimates, while avoiding unsupported precision in the absence of empirical variance data. The range was informed by expert opinion and reviewed by the Sri Lankan oncology clinical expert and the health economists on the study team. The ±20% range was treated as representing an approximate 95% uncertainty interval, from which the corresponding SE was derived to parameterise gamma distributions for the probabilistic analysis. This approach is consistent with established decision-modelling guidance recommending the use of plausible ranges and appropriate distributional assumptions when empirical variance data are unavailable.23All costs were reported in 2024 US Dollars.Utility estimates Since local utility data were unavailable, we used regional and global estimates to assign health-state utility values of 0.78 for the disease-free state and 0.69 for the recurrence state. 24 To address the potential quality-of-life impact of adverse events and treatment burden during prolonged active therapy, we applied a utility decrement of −0.02 during the on-treatment period.25 26 This decrement was applied for the planned duration of active treatment: 3 years for adjuvant osimertinib, 2 years for adjuvant alectinib, and the corresponding active treatment duration for each immunotherapy regimen. The decrement was applied to the disease-free utility value during active treatment only.Model evaluation Our decision-analytic model simulated patient cohorts through various health states. Patients accumulated costs and QALYs based on clinical outcomes as they moved between these states. The cost-effectiveness analysis then compared the total accumulated costs and QALYs of each new adjuvant therapy with its standard care counterpart.To characterise parameter uncertainty, a probabilistic analysis was conducted and used as the base-case analysis. This involved assigning probability distributions to model inputs to reflect their variability: beta distributions were used for transition probabilities and utility values (constrained between 0 and 1), and gamma distributions were applied to cost parameters, accounting for the typical right-skewed distribution of healthcare costs. A total of 1000 Monte Carlo simulations were performed. In each simulation, input values were randomly sampled from their assigned distributions (refer table 1), and costs and QALYs were estimated. Mean costs and mean QALYs across simulations are reported as the main results. The incremental cost-effectiveness ratio (ICER) was calculated as the difference in mean costs divided by the difference in mean QALYs between the intervention and standard care arms.A willingness-to-pay threshold of US$6700 per QALY gained was applied in the primary analysis. This threshold corresponds to approximately twice Sri Lanka’s gross domestic product (GDP) per capita and was used as a GDP-based reference threshold commonly applied in Low and Middle Income Countries (LMIC) economic evaluations.27 Given increasing concerns that GDP-multiple thresholds may not reflect health opportunity costs in resource-constrained settings, we conducted scenario analyses using lower opportunity-cost-informed thresholds of US$1500 and US$3000 per QALY gained. These thresholds were selected to represent conservative and intermediate values below one times GDP per capita, informed by methodological work by Woods et al and Ochalek et al.28 29 Cost-effectiveness conclusions were interpreted across all three thresholds. To ensure robustness under parameter uncertainty, the total cost, effectiveness and ICER were estimated from the mean outcomes across all probabilistic sensitivity analysis simulations.An incremental cost-effectiveness scatterplot was created to visualise the combined distribution of cost and effectiveness outcomes. Each point on the plot signifies the incremental cost and QALY outcome from a single simulation. The willingness-to-pay threshold of US$6700 per QALY gained is represented by the dashed line.To represent the uncertainty surrounding the cost-effectiveness of each intervention across various willingness-to-pay thresholds, a cost-effectiveness acceptability curve was constructed. This curve illustrates the probability of each strategy being cost-effective compared with standard care, given different willingness-to-pay values per QALY gained. The cost-effectiveness acceptability curve was derived from the proportion of probabilistic sensitivity analysis simulations where the intervention yielded the highest net monetary benefit at each willingness-to-pay level. Ultimately, the curve offers decision-makers insight into the likelihood that a particular intervention provides value for money under different threshold assumptions.A BIA was conducted to estimate the expected financial consequences of adopting the cost-effective novel therapies within the Sri Lankan public health system. The analysis was performed from the same health system perspective and followed recommended good-practice principles for budget impact modelling.30 The eligible population was estimated using national incidence data for stage II–III NSCLC and epidemiological estimates of biomarker prevalence (EGFR mutation, ALK rearrangement and PD-L1 expression). The budget impact compared current standard care with four implementation scenarios assuming 25%, 50%, 75% and 100% uptake of each eligible novel therapy among clinically eligible patients. The analysis included treatment and follow-up costs but did not include additional diagnostic infrastructure, biomarker testing or implementation costs that may be required to identify eligible patients and support adoption. The analysis incorporated both male and female patients. The budget impact was estimated over a 5-year time horizon without discounting, consistent with standard BIA methodology. Cumulative incremental budget impact estimates were calculated as the difference in total projected expenditure between the two scenarios for therapies identified as cost-effective in the main probabilistic analysis.Results Among the tyrosine kinase inhibitors evaluated, osimertinib demonstrated clear cost-effectiveness in both treatment settings in the male population ( table 2 and figure 2). When administered after surgery, osimertinib produced an ICER of US$2134 per QALY gained with a 99% probability of being cost-effective at the willingness-to-pay threshold of US$6700 per QALY. Following curative chemoradiotherapy, osimertinib remained highly favourable, yielding an ICER of US$1933 per QALY and the same 99% probability of cost-effectiveness. Alectinib, given as adjuvant therapy after surgery for ALK-mutated disease, showed borderline cost-effectiveness (ICER US$6335 per QALY; 61% probability of cost-effectiveness). Gender-specific analyses in females (online supplemental table 2) demonstrated consistent conclusions. Adjuvant osimertinib after surgery and after chemoradiotherapy remained highly cost-effective, with ICERs of US$2355 and US$1283 per QALY, respectively, each with a 99% probability of cost-effectiveness. Alectinib showed similar economic performance in females (ICER US$6255 per QALY; 61% probability of cost-effectiveness).SP210.1136/bmjonc-2025-001046.supp2Supplementary dataFigure 2Incremental cost-effectiveness (ICE) scatterplots: results from probabilistic sensitivity analysis. QALY, quality-adjusted life-year; WTP, willingness-to-pay.Table 2Cost-effectiveness (CE) of adjuvant, neoadjuvant and perioperative therapies for stage II–III non-small-cell lung cancer: results based on 1000 Monte-Carlo simulations (male population)Cost (US$)Incremental cost (US$)QALYIncremental QALYICER(US$/QALY)Probability of being CEOsimertinib after surgery Intervention11 00987838.124.12213499% Standard care22274.00Osimertinib after chemoradiotherapy Intervention770754045.522.80193399% Standard care23022.73Alectinib after surgery Intervention23 90822 6058.343.57633561% Standard care13034.77Neoadjuvant nivolumab Intervention10 41990654.921.67542978% Standard care13543.25Perioperative pembrolizumab Intervention44 02542 6644.521.5527 563<0.1% Standard care13612.97Perioperative durvalumab Intervention42 03540 7114.921.1535 537<0.1% Standard care13243.77Adjuvant atezolizumab after surgery Intervention31 27830 0187.172.4112 451<0.1% Standard care12604.76Adjuvant pembrolizumab after surgery Intervention46 20444 9806.140.8255 016<0.1% Standard care12245.33Consolidation durvalumab after chemoradiotherapy Intervention19 04317 7172.030.5930 258<0.1% Standard care13261.44ALK, anaplastic lymphoma kinase; EGFR, epithelial growth factor receptor; ICER, incremental cost-effectiveness ratio; QALY, quality-adjusted life-year.Among immunotherapy-based regimens, neoadjuvant nivolumab demonstrated the most favourable economic profile in both sexes. In males, the ICER was US$5429 per QALY with a 78% probability of cost-effectiveness, while in females it was US$4640 per QALY with a 87% probability of cost-effectiveness. In contrast, adjuvant atezolizumab and pembrolizumab, perioperative pembrolizumab and durvalumab, and consolidation durvalumab after chemoradiotherapy were not cost-effective in either sex, with ICERs substantially exceeding the willingness-to-pay threshold and probabilities of cost-effectiveness below 1%. Overall, conclusions were consistent across sexes, with slightly greater QALY gains observed in females.A scenario analysis using lower willingness-to-pay thresholds showed that the cost-effectiveness conclusions were sensitive to the threshold selected. At a threshold of US$3000 per QALY, both osimertinib strategies remained highly likely to be cost-effective, with 99% probability of cost-effectiveness for osimertinib after surgery and after chemoradiotherapy. However, at lower thresholds, the probability of cost-effectiveness decreased substantially. Osimertinib after surgery had only a 1% probability of being cost-effective at US$1500 per QALY, while osimertinib after chemoradiotherapy had a 10% probability at US$1500 per QALY. Alectinib after surgery and neoadjuvant nivolumab were not cost-effective under the lower opportunity-cost-informed thresholds, with probabilities of cost-effectiveness below 1% at both US$1500 and US$3000 per QALY. Thus, while all four therapies were cost-effective at the original US$6700 per QALY threshold, only osimertinib-based strategies remained robustly cost-effective under the US$3000 per QALY scenario.For immunotherapy-based regimens, substantial price reductions would be necessary to achieve a 50% probability of cost-effectiveness (online supplemental table 1). Pembrolizumab, when used either perioperatively or as adjuvant therapy, would need a price reduction of 77%–88%, from US$51 500 to US$12 000 and US$6000, respectively, to reach cost-effectiveness. Durvalumab would require a decrease from US$48 500 to US$9000 (an 81% reduction) in the perioperative setting, and from US$26 500 to US$5500 (a 79% reduction) when used as consolidation therapy after chemoradiotherapy. Atezolizumab would also need a 46% price reduction, to reach similar cost-effectiveness probabilities.5-year cumulative budget impact estimates for adopting cost-effective therapies for stage II–III NSCLC in Sri Lanka under 25%, 50%, 75% and 100% uptake scenarios are presented in table 3. Under a 50% uptake scenario, the cumulative 5-year incremental budget impact was US$1.48 million for osimertinib after surgery, US$1.9 million for osimertinib after chemoradiotherapy, US$0.77 million for alectinib after surgery and US$2.29 million for neoadjuvant nivolumab. At full implementation, the corresponding 5-year budget impact increased to US$2.78 million, US$3.3 million, US$1.49 million and US$4.30 million, respectively. Across all cost-effective therapies, total 5-year budget impact was approximately US$6.5 million under the 50% uptake scenario and US$11.8 million under the 100% uptake scenario. Neoadjuvant nivolumab was associated with the largest cumulative financial impact across all uptake scenarios.Table 35-year cumulative budget impact of adopting cost-effective therapies for stage II–III NSCLC in Sri Lanka (25%, 50%, 75% and 100% uptake scenario)TherapyEstimated eligible patients per year (male; female)25% uptake scenario50% uptake scenario75% uptake scenario100% uptake scenarioOsimertinib after surgeryMales: 50; females: 25828 0651 477 0522 126 0392 775 026Osimertinib after chemoradiotherapyMales: 100; females: 351 234 6951 923 4872 612 2783 301 0709Alectinib after surgeryMales: 10; females: 5404 573765 9441 127 3151 488 686Neoadjuvant nivolumabMales: 75; females: 251 291 2092 293 3993 295 5894 297 779NSCLC, non-small cell lung cancer.The cost-effectiveness acceptability curves illustrate the probability of each treatment being cost-effective across a range of willingness-to-pay thresholds (online supplemental figure 1). Among the novel agents assessed, osimertinib, both after surgery and following chemoradiotherapy, and alectinib after surgery demonstrated the highest probability of cost-effectiveness, exceeding 50% at willingness-to-pay values well below the Sri Lankan threshold of US$6700 per QALY. Neoadjuvant nivolumab reached a 50% probability at a willingness-to-pay of approximately US$5500 per QALY, also within the national threshold. In contrast, atezolizumab, pembrolizumab and durvalumab required substantially higher willingness-to-pay values often exceeding US$13 000–US$60 000 per QALY to achieve similar probabilities, indicating that these therapies remain economically unviable under current pricing.Discussion Our analyses show that in Sri Lanka’s state health system, four approved indications of novel cancer therapeutics for stage II-III NSCLC emerged as cost-effective at the primary willingness-to-pay threshold of US$6700 per QALY: adjuvant osimertinib administered after both surgery and chemoradiotherapy, adjuvant alectinib after surgery and neoadjuvant nivolumab prior to surgery. However, these conclusions were sensitive to the threshold applied. Under lower opportunity-cost-informed thresholds, only the osimertinib-based strategies remained robustly cost-effective at US$3000 per QALY, while alectinib and neoadjuvant nivolumab were no longer cost-effective.Several characteristics that indicate the cost-effectiveness of a treatment can be discerned from our results: the availability of generic or biosimilar products, reduced overall treatment exposure (which lowers costs), and the existence of predictive biomarkers to identify patients who will most likely benefit from the treatment. Osimertinib demonstrates a highly favourable HR of 0.16–0.17, and its overall drug cost is substantially reduced due to the availability of generic products.7 8 Neoadjuvant nivolumab’s cost-effectiveness at the primary threshold could also be attributed to a favourable HR of 0.41 and lower total dose.10 Costs are expected to drop significantly in the coming years as nivolumab biosimilars enter the market.While approximately 2000 new lung cancer cases are diagnosed annually, most patients in Sri Lanka present at a stage where curative treatment is not feasible due to advanced disease (stage IIIB and IV) or poor performance status.1 2 Adjuvant osimertinib is indicated only for patients with EGFR mutant tumours, neoadjuvant nivolumab is recommended for patients with tumours expressing PD-L1 >1%. Therefore, we estimate that only about 150–200 patients per year would be eligible for these treatments. Given the low incidence of ALK mutations in adenocarcinoma of lung, total funding for adjuvant alectinib would likely not have a significant budgetary impact.31 As shown in table 3, an additional annual budgetary footprint of around US$2.36 million would be required to fund these drugs, an amount which appears feasible.17A group of patients for which a novel treatment is not found to be cost-effective are NSCLC patients without EGFR, ALK mutations and PD-L1 expression <1%. While robust data on PD-L1 expression in NSCLC in Sri Lanka is lacking, international data suggests that 30%–40% of patients may have negative PD-L1 expression.32 Consequently, a substantial price reduction in durvalumab and pembrolizumab would be necessary to meet an acceptable cost-effectiveness threshold and to avoid a significant budgetary impact, given the sizeable number of potentially eligible patients. For NSCLC patients without EGFR mutations treated with curative intent chemoradiotherapy consolidation durvalumab is the only approved indication. Once again, a significant price reduction of the drug would be required for it to be competitive at the defined willingness to pay threshold.Our cost analysis relied on private sector retail drug prices because these agents are unavailable in the government-funded state health sector. State sector procurements, typically involving bulk purchases, would likely result in substantially lower drug costs compared with the private sector retail market. This could lead to significant cost savings and a more favourable QALY gain for these drugs. Prices may fall further when the patency for these agents expire in the region and more generic and biosimilar products enter the market.While our model captures the primary drivers of cost and clinical outcomes, it is subject to certain limitations. When considering the costs of the ‘recurrence’ health state, we based our calculations on the current clinical and financial realities reflecting the standard clinical pathway in the Sri Lankan state health sector in which several novel cancer drugs such as osimertinib, alectinib and immunotherapy are not approved for use in advanced NSCLC. While this cost asymmetry undoubtedly impacted on the derived ICERs, by focusing on currently available treatments in the recurrence state, the analysis ensures that the estimated costs of recurrence accurately reflect the actual budgetary burden and standard of care of the state health sector. These simplifications were necessary to ensure model stability and relevance to the local resource-constrained setting.Furthermore, empirical variance estimates were unavailable for several local cost inputs. We therefore used a uniform ±20% expert-informed uncertainty range for cost parameters in the probabilistic analysis. Although this approach provides a pragmatic method for reflecting cost uncertainty in the absence of local variance data, it may not fully capture differences in uncertainty across cost categories.The female analysis should also be interpreted with caution. In the absence of sex-specific transition probabilities from the pivotal trials, we applied a female-to-male HR of 0.8 from a single Japanese lung cancer registry study to adjust recurrence and post-recurrence mortality estimates. Although this reflects evidence that women with NSCLC often have better survival than men, applying one HR uniformly across treatment indications, biomarker subgroups, disease stages and transition types is a simplifying assumption. The Japanese registry population may also differ from the Sri Lankan population in ethnicity, smoking patterns, treatment access and background mortality. Therefore, the female results should be considered supportive scenario analyses rather than definitive sex-specific estimates.Apart from these, we relied on regional and international estimates for transition probabilities and utility weights due to the absence of local data, potentially not fully reflecting Sri Lankan patients’ quality-of-life valuations. Indirect costs, such as patient out-of-pocket expenditures and productivity losses, were excluded as the analysis is focused on the public health system perspective, and their inclusion could alter intervention value estimates.Finally, we applied a willingness-to-pay threshold of US$6700 per QALY (approximately two times Sri Lanka’s GDP per capita), a benchmark commonly used in LMIC economic evaluations and frequently adopted where locally derived thresholds are unavailable. However, GDP-multiple thresholds are increasingly regarded as imperfect because they may not reflect the health opportunity costs faced by constrained public budgets, ie, the health gains that could be produced if the same resources were allocated to alternative services. Recent LMIC methodological literature has therefore emphasised opportunity cost-based thresholds, which aim to approximate the marginal productivity of health spending and are often lower than GDP-based benchmarks.28 In view of this, we conducted scenario analyses using lower opportunity-cost-informed thresholds. These analyses showed that conclusions were threshold-sensitive: only osimertinib-based strategies remained robustly cost-effective at US$3000 per QALY, while alectinib and neoadjuvant nivolumab were no longer cost-effective at lower thresholds. Therefore, our findings should be interpreted alongside affordability, budget impact and price negotiation considerations.Nevertheless, this study provides a consistent set of comparative cost-effectiveness estimates for multiple recently approved treatment strategies for stage II–III NSCLC, analysed from the Sri Lankan public health system perspective. By presenting results side-by-side and quantifying the price reductions required for selected agents to meet plausible thresholds, our findings can help decision-makers prioritise candidates for initial adoption and guide price negotiations, alongside feasibility and budget impact considerations.