First-line osimertinib in advanced EGFR-mutated NSCLC: real-world outcomes, clinicogenomic correlates, and oligoprogression management in a multicenter Spanish cohort Authors: G. Suay, S. Cases-Esteban, P. Capdevila, A. García-Márquez, J. Garde-Noguera, P. Espinosa-Olarte, M. Arnal, S. Sandiego-Contreras, F. Aparisi-Aparisi, C. Escoin-Pérez, P. Martín-Martorell, J. García, A. Blasco-Cordellat, S. Blasco-Mollá, J.D. Benítez-Fuentes, C.D. Ávila-Andrade, M. Borregón, O. Juan-Vidal Published: 2026 ABSTRACT BackgroundFirst-line osimertinib is standard for advanced EGFR-mutated non-small-cell lung cancer (NSCLC), but upfront combinations (FLAURA2, MARIPOSA) are reshaping this paradigm. Trial practice gaps and the optimal management of oligoprogressive disease (OPD) remain unclear. We evaluated real-world effectiveness, progression patterns, and the role of local ablative therapy (LAT) in a Western European cohort.Patients and methodsMulticenter retrospective cohort from eastern Spain including patients with advanced EGFR-mutated NSCLC treated with first-line osimertinib. Outcomes were objective response rate (ORR), real-world progression-free survival (rwPFS), real-world overall survival (rwOS) and access to second-line therapy. Prespecified analyses examined exon 19 deletion versus p.Leu858Arg, baseline central nervous system metastases, OPD treated with LAT while continuing osimertinib, re-biopsy at progression, and time toxicity.ResultsAmong 268 patients, median age was 72 years, 27.6% had Eastern Cooperative Oncology Group performance status ≥2, and 42.5% had brain metastases. Osimertinib achieved an ORR of 73.5%, rwPFS of 17.5 months, and rwOS of 28.1 months, but only 50.1% of patients who progressed initiated second-line therapy. Among 180 patients with documented progression, 51 (28.3%) had OPD and 82.4% received LAT. Exploratory analyses suggested clinically meaningful improvements in median PFS and median OS versus the overall cohort. Re-biopsy identified resistance mechanisms (e.g. MET amplification, small-cell transformation), and time toxicity measures were modest.ConclusionsIn routine practice, first-line osimertinib shows robust effectiveness but lower rwOS than expected and substantial post-progression attrition. These findings underscore the need for risk-adapted treatment strategies and support prospective evaluation of OPD management approaches, including integration of LAT alongside contemporary systemic options. Introduction The identification of oncogenic driver mutations has transformed the therapeutic landscape of non-small-cell lung cancer (NSCLC), enabling the development of molecularly targeted agents that have markedly improved patient outcomes compared with conventional chemotherapy.1,2 Among these, activating mutations in the epidermal growth factor receptor (EGFR) gene represent the most common actionable alteration in lung adenocarcinoma, occurring in ∼10%-15% of patients of European ancestry and up to 40%-50% in Asian populations.3 The therapeutic landscape for treatment-naïve, EGFR-mutated metastatic NSCLC was significantly transformed by osimertinib. In the pivotal FLAURA phase III trial, osimertinib demonstrated superior progression-free survival (PFS) and overall survival (OS) compared with first-generation EGFR tyrosine kinase inhibitors (TKIs) [median PFS (mPFS) 18.9 versus 10.2 months, hazard ratio (HR) for progression or death 0.46, P < 0.01; median OS (mOS) ∼38.6 versus 31.8 months, HR for death 0.80, P = 0.04] in patients with exon 19 deletions or p.Leu858Arg (L858R) mutations, and exhibited enhanced central nervous system (CNS) activity, reducing the risk of intracranial progression by >50% versus comparator agents.1,2,4 Nevertheless, subgroups with L858R (exon 21) mutations or baseline brain metastases continue to have poorer outcomes in both clinical trials and real-world practice.5 More recently, first-line combination strategies have expanded the therapeutic arsenal, including osimertinib plus platinum–pemetrexed (FLAURA2) and amivantamab plus lazertinib (MARIPOSA), both of which improved PFS compared with osimertinib monotherapy in randomized phase III trials.6, 7, 8, 9 However, the absence of head-to-head comparisons between these approaches creates uncertainty regarding optimal first-line selection, and guideline recommendations currently endorse these options without preference.10, 11, 12 In this setting, real-world data (RWD) from heterogeneous patient cohorts, often including older and frailer populations underrepresented in trials, can help contextualize outcomes with osimertinib monotherapy and inform which patients might be considered for upfront intensification. In parallel, key knowledge gaps persist between trial populations and routine practice, particularly regarding progression patterns and the optimal management of oligoprogression during first-line osimertinib. Oligoprogressive disease (OPD) is frequently encountered in EGFR-mutated NSCLC, yet evidence guiding patient selection and timing of local ablative therapy (LAT) while continuing EGFR TKI treatment remains limited. Therefore, in this study, we evaluate the real-world effectiveness of first-line osimertinib, characterize progression patterns with a focus on OPD and LAT use, and describe post-progression trajectories including subsequent lines and biomarker-guided management at progression in a multicenter Spanish cohort. Patients and Methods Study design and patients The primary objective of this multicenter, retrospective, observational study was to assess the clinical features and management of patients diagnosed with NSCLC and EGFR mutations in eastern Spanish hospitals of the GIDO (Grupo de Investigación y Divulgación Oncológica) group who received osimertinib as first-line treatment for advanced disease (RWD). Patients were recruited if they received osimertinib treatment between June 2021 (date when osimertinib was financed in eastern Spain) and December 2024 to ensure data maturity. The study was observational, and participation did not change therapeutic or follow-up management of patients. The secondary objectives were (i) to analyze parameters not included in the registrational phase III clinical trial FLAURA, such as median time to progression after OPD treatment and the presence of baseline active CNS disease; (ii) to compare mPFS and mOS in the FLAURA trial with RWD according to clinical practice; (iii) to evaluate mechanisms of resistance to osimertinib; and (iv) to evaluate how EGFR status was determined [PCR versus next-generation sequencing (NGS)]. Other inclusion criteria included age ≥18 years, diagnosis of advanced NSCLC bearing classical EGFR mutations (EGFR exon 19 deletion, EGFR L858R exon 21 mutation) or uncommon mutations (excluding EGFR exon 20 insertions). No first or second generation EGFR-directed TKIs were allowed. The study was approved by the Research Ethics Committee of La Fe University and Polytechnic Hospital on 22 March 2024 (reference number GIDO/2024/01) and was executed in accordance with the Declaration of Helsinki and local ethical and legal requirements. Data collection and analysis Data were retrospectively collected using routine electronic medical records from 11 public hospitals in eastern Spain, ensuring continuous longitudinal capture of systemic therapies, imaging, and survival status across medical oncology, radiotherapy, and inpatient settings. No formal sample size calculation was carried out; patients were included if eligible within the study window. Patients were uniquely identified within each site and cross-checked against regional identification numbers; potential duplicates across sites were queried and resolved by manual review. Raw data underwent range checks, date consistency checks, and logic rules. Outliers were verified against source records. A prespecified data dictionary was used across sites. The following data were collected: age; tumor–node–metastasis (TNM) staging (8th edition)13; clinical history; comorbidities; treatment-related time toxicity; clinical Eastern Cooperative Oncology Group performance status (ECOG-PS); histological and molecular data at diagnosis; treatment details, toxicity, and response; treatment and follow-up beyond progression (special focus on OPD); and successive treatments after osimertinib. The number of comorbidities was obtained by taking into account the disorders included in the Charlson comorbidity index.14 Treatment-related time toxicity is an emergent parameter that provides insight into the number of days that are lost in visits to health care facilities and may help analyze if drug-related survival improvements are not achieved through a quality-of-life loss; it was defined as the total number of days the patient had to come to the clinic (ordinary visits or toxicity-associated consultations) during treatment.15 EGFR mutations were detected by PCR or NGS, depending on local availability and guidelines when patients were diagnosed in each hospital. In cases for which NGS was available, co-mutations were evaluated. Tumor assessment was carried out according to clinical practice by the treating oncologists using RECIST v1.1. PFS was calculated from the first day of treatment until any relapse, local or distant, or death (whichever occurred first); OS was calculated from the first day of treatment until the last date of follow-up or death. For OPD, defined as progression in no more than five lesions,16,17 cohort-specific OS, PFS2, and PFS beyond first osimertinib progression (PFSBOP) were calculated. PFS2 was calculated from the first day of treatment until any second relapse, local or distant, or death (whichever occurred first) after a first progression on osimertinib, whereas PFSBOP was analyzed from the first day of treatment with osimertinib beyond progression until any second relapse, local or distant, or death (whichever occurred first) after a first progression on osimertinib. To mitigate bias, we prespecified covariates for multivariable models, verified proportional hazards, and treated LAT analyses as exploratory due to potential selection and immortal time biases. Statistical data were analyzed using the RStudio software v.2025.09.1+401. Kaplan–Meier and Cox regressions (univariate and multivariate) were carried out in R with the ‘survival,’ ‘survminer,’ and ‘ggplot2’ packages.18,19 Cox models were fitted using complete case analysis for the covariates included in each model. TP53 status was available only in the NGS subset; therefore, TP53 was evaluated in prespecified sensitivity models restricted to patients with available NGS data. To calculate follow-up, the inverse Kaplan–Meier method was used. The Sankey diagram was generated with SankeyMATIC.20 Results Baseline characteristics A total of 268 patients with EGFR-mutated NSCLC from 11 hospitals were included in the analysis (Table 1 and Supplementary Figure S1, available at https://doi.org/10.1016/j.esmorw.2026.100725). Median follow-up was 35.12 months. Baseline demographic and clinical characteristics are summarized in Table 1. The median age was 72 years (range 31-96), and most patients were female (67.9%). The majority of patients were never smokers (60.8%). Comorbidities were present in 75.7% of the cohort, with a median of two comorbid conditions (range 0-8). The median number of concomitant medications was three (range 0-15).Table 1Baseline demographic and clinical characteristicsVariableTotalAge, years72 (31-96)Sex Male86 (32.1%) Female182 (67.9%)Smoking Never163 (60.8%) Active or former smoker105 (39.2%)Patients with comorbidities203 (75.7%) Median number of comorbidities2 (0-8) Concomitant treatment, median number of drugs3 (0-15)ECOG-PS 033 (12.3%) 1161 (60.1%) 261 (22.8%) 3/413 (4.8%)Stage at NSCLC diagnosis I11 (4.1%) II12 (4.5%) III15 (5.6%) IV230 (85.8%)Stage at osimertinib prescription II-III8 (3%) IV260 (97%)Brain metastases Yes114 (42.5%) No154 (57.5%)Histology Adenocarcinoma263 (98.1%) Squamous1 (0.4%) Other4 (1.5%)Values are n (%) or median (interquartile range).ECOG-PS, Eastern Cooperative Oncology Group performance status; NSCLC, non-small-cell lung cancer. Table 1Baseline demographic and clinical characteristicsVariableTotalAge, years72 (31-96)Sex Male86 (32.1%) Female182 (67.9%)Smoking Never163 (60.8%) Active or former smoker105 (39.2%)Patients with comorbidities203 (75.7%) Median number of comorbidities2 (0-8) Concomitant treatment, median number of drugs3 (0-15)ECOG-PS 033 (12.3%) 1161 (60.1%) 261 (22.8%) 3/413 (4.8%)Stage at NSCLC diagnosis I11 (4.1%) II12 (4.5%) III15 (5.6%) IV230 (85.8%)Stage at osimertinib prescription II-III8 (3%) IV260 (97%)Brain metastases Yes114 (42.5%) No154 (57.5%)Histology Adenocarcinoma263 (98.1%) Squamous1 (0.4%) Other4 (1.5%)Values are n (%) or median (interquartile range).ECOG-PS, Eastern Cooperative Oncology Group performance status; NSCLC, non-small-cell lung cancer. Baseline demographic and clinical characteristics Values are n (%) or median (interquartile range). ECOG-PS, Eastern Cooperative Oncology Group performance status; NSCLC, non-small-cell lung cancer. At the time of NSCLC diagnosis, 85.8% of patients had stage IV disease, whereas only 15.2% were diagnosed at earlier stages (I-III). Similarly, at osimertinib initiation, nearly all patients (97.0%) were at stage IV. Brain metastases were documented in 42.5% of cases. Adenocarcinoma represented the predominant histological subtype (98.1%), with only rare cases of squamous (0.4%) or other histologies (1.5%). Molecular characteristics are detailed in Table 2. EGFR mutations were identified using NGS in 62.3% of patients and PCR in 37.7%. The most frequent alteration was exon 19 deletion (61.6%), followed by exon 21 L858R (35.8%), while uncommon mutations accounted for 2.6% of cases. Among exon 19 deletions, p.Glu746_Ala750del was the most prevalent variant (24.6% of the overall cohort), followed by p.Leu747_Pro753delinsSer (4.5%) and p.Glu746_Ser752delinsVal (2.2%). Among patients with available NGS data, the most frequent co-mutation was TP53 (14.97%), followed by EGFR amplification (7.78%) and PIK3CA mutations (5.38%). Other concomitant alterations were present in 14.97% of patients, while 17.37% showed no detectable co-mutations.Table 2Molecular characteristicsCharacteristicPatients, n (%)EGFR detection method NGS167 (62.3%) PCR101 (37.7%)EGFR mutation Exon 19 deletion165 (61.6%) Exon 21 mutation96 (35.8%) Uncommona7 (2.6%)Type of EGFR mutation p.Leu858Arg96 (35.8%) p.Glu746_Ala750del66 (24.6%) p.Leu747_Pro753delinsSer12 (4.5%) p.Glu746_Ser752delinsVal6 (2.2%) Other exon 19 deletions22 (8.21%) Uncommon mutations7 (2.6%) Unknown (only PCR data)66 (24.62%)Co-mutationsb TP5325 (14.97%) EGFR amplification13 (7.78%) PIK3CA9 (5.38%) Other mutations25 (14.97%) No co-mutation29 (17.37%)EGFR, epidermal growth factor receptor; NGS, next-generation sequencing; PCR, polymerase chain reaction.aUncommon EGFR variants included p.Gly719Ala (n = 2), p.Leu861Gln (n = 2), p.Leu861Arg (n = 1), p.Thr790Met (n = 1), and p.E709_T710delinsD (n = 1).bCo-mutations are reported for patients with NGS data. Table 2Molecular characteristicsCharacteristicPatients, n (%)EGFR detection method NGS167 (62.3%) PCR101 (37.7%)EGFR mutation Exon 19 deletion165 (61.6%) Exon 21 mutation96 (35.8%) Uncommona7 (2.6%)Type of EGFR mutation p.Leu858Arg96 (35.8%) p.Glu746_Ala750del66 (24.6%) p.Leu747_Pro753delinsSer12 (4.5%) p.Glu746_Ser752delinsVal6 (2.2%) Other exon 19 deletions22 (8.21%) Uncommon mutations7 (2.6%) Unknown (only PCR data)66 (24.62%)Co-mutationsb TP5325 (14.97%) EGFR amplification13 (7.78%) PIK3CA9 (5.38%) Other mutations25 (14.97%) No co-mutation29 (17.37%)EGFR, epidermal growth factor receptor; NGS, next-generation sequencing; PCR, polymerase chain reaction.aUncommon EGFR variants included p.Gly719Ala (n = 2), p.Leu861Gln (n = 2), p.Leu861Arg (n = 1), p.Thr790Met (n = 1), and p.E709_T710delinsD (n = 1).bCo-mutations are reported for patients with NGS data. Molecular characteristics EGFR, epidermal growth factor receptor; NGS, next-generation sequencing; PCR, polymerase chain reaction. Uncommon EGFR variants included p.Gly719Ala (n = 2), p.Leu861Gln (n = 2), p.Leu861Arg (n = 1), p.Thr790Met (n = 1), and p.E709_T710delinsD (n = 1). Co-mutations are reported for patients with NGS data. Clinical response and survival Real-world PFS (rwPFS) of the cohort was 17.5 months (95% CI 14.8-19.8), while real-world overall survival (rwOS) of the cohort was 28.1 months (95% CI 23.9-35.4) (Supplementary Figure S2, available at https://doi.org/10.1016/j.esmorw.2026.100725). The objective response rate (ORR) was 73.5% (95% CI 67.9% to 78.7%), whereas the disease control rate was 82.5% (95% CI 78% to 86.9%). In the prespecified Cox model for rwPFS, including age (≥72 versus <72 years), sex, ECOG-PS (≥2 versus 0-1), baseline CNS metastases, and EGFR subtype, we found that ECOG-PS ≥2 (HR 1.72, 95% CI 1.22-2.43, P = 0.02) and L858R (versus exon 19 deletion; HR 1.38, 95% CI 1.01-1.88, P = 0.046) were independently associated with shorter rwPFS. Baseline CNS metastases, sex, and age were not statistically significant. Full adjusted estimates are shown in Supplementary Figure S3, available at https://doi.org/10.1016/j.esmorw.2026.100725. In the prespecified Cox model for rwOS, including age (≥72 versus <72 years), sex, ECOG-PS (≥2 versus 0-1), baseline CNS metastases, and EGFR subtype, we found that ECOG-PS ≥2 (HR 2.72, 95% CI 1.9-3.9, P < 0.01) and L858R (versus exon 19 deletion; HR 1.41, 95% CI 1.01-1.98, P = 0.047) were independently associated with shorter rwOS. Baseline CNS metastases, sex, and age were not statistically significant. Full adjusted estimates are shown in Supplementary Figure S4, available at https://doi.org/10.1016/j.esmorw.2026.100725. Because TP53 status was available only in the NGS subset (62.3% of patients, n = 167), we carried out prespecified sensitivity multivariable Cox models restricted to patients with available NGS data, additionally including TP53. In these sensitivity analyses, TP53 was not significantly associated with rwPFS (P = 0.39) or rwOS (P = 0.26), and estimates for some covariates were less precise, consistent with the smaller sample size. Full adjusted estimates for the NGS subset are shown in Supplementary Figures S5 and S6, available at https://doi.org/10.1016/j.esmorw.2026.100725. Among the 261 patients with classical EGFR mutations (Figure 1A and B), rwPFS was better (although statistical significance was not achieved) in patients with exon 19 deletion than in patients with exon 21 L858R mutation (19.3 months, 95% CI 16.7-26.5, versus 16.1 months, 95% CI 12.3-20.5; HR 0.74, 95% CI 0.55-1.00, P = 0.05). rwOS was significantly better in patients with exon 19 deletion than in patients with exon 21 L858R mutation (35.4 months, 95% CI 30-46.5, versus 24.1 months, 95% CI 20.6-29.5; HR 0.67, 95% CI 0.48-0.93, P = 0.02). In patients with CNS baseline disease (Figure 1C and D), although rwPFS was significantly worse (14.3 months, 95% CI 10.8-18.9, versus 19.8 months, 95% CI 17.3-25.3; HR 1.40, 95% CI 1.04-1.88, P = 0.02), there were no statistically significant differences in rwOS (24.6 months, 95% CI 17.6-37.7, versus 30 months, 95% CI 25-39.8; HR 1.18, 95% CI 0.85-1.63, P = 0.32). ECOG-PS was analyzed (Figure 1E and F) by comparing patients that would comply with the FLAURA trial inclusion criteria (ECOG-PS 0-1) with those that would have been excluded (ECOG-PS 2-4). rwPFS and rwOS were significantly worse in patients with ECOG-PS 2-4 (rwPFS: 9 months, 95% CI 5.45-17.4, versus 19.3 months, 95% CI 17.35-24.1; HR 1.73, 95% CI 1.26-2.39, P < 0.01; rwOS: 35.4 months, 95% CI 29.5-45.5, versus 13.8 months, 95% CI 7.66-22.5; HR 2.74, P < 0.01).Figure 1Kaplan–Meier curves estimating PFS and OS. PFS according to classical EGFR mutation type (A), baseline CNS disease (C), and performance status (E). OS according to classical EGFR mutation type (B), baseline CNS disease (D), and performance status (F). CNS, central nervous system; EGFR, epidermal growth factor receptor; OS, overall survival; PFS, progression-free survival. Kaplan–Meier curves estimating PFS and OS. PFS according to classical EGFR mutation type (A), baseline CNS disease (C), and performance status (E). OS according to classical EGFR mutation type (B), baseline CNS disease (D), and performance status (F). CNS, central nervous system; EGFR, epidermal growth factor receptor; OS, overall survival; PFS, progression-free survival. Time toxicity was calculated in 184 patients, showing a median number of 22 days (95% CI 20-25). Treatment-related toxicity, which required dose reduction from 80 mg to 40 mg, was observed in 48 patients (17.9%), while treatment discontinuation due to toxicity was required in 9 patients (3.4%). Treatment beyond progression and subsequent lines Among 180 patients (67.2%) who experienced disease progression to osimertinib, 51 patients (28.33%) experienced OPD (Figure 2). Baseline characteristics are summarized in Supplementary Table S1, available at https://doi.org/10.1016/j.esmorw.2026.100725. Forty-six of 51 patients (90.2%) continued osimertinib beyond progression, and 42 of 51 patients (82.35%) received LAT for the OPD (40 patients received radical radiotherapy, while 2 patients had surgery and suspended osimertinib). Twenty-two of 42 patients treated with LAT received second-line treatment.Figure 2Kaplan–Meier curves estimating the real-world PFS and OSbeyond osimertinib progressionstratified by receipt of LAT. (A, B) PFS2 and real-world OS of the oligoprogression cohort. (C, D) Real-world PFS beyond osimertinib progression (C) and OS (D) stratified by receipt of LAT. Descriptive comparison only. Kaplan–Meier curves for real-world PFS beyond osimertinib progression (LAT versus no-LAT) are shown for descriptive purposes, but the comparison is not intended for causal inference due to selection and immortal time biases and the small no-LAT subgroup. Log-rank P values are shown on the curves for descriptive purposes only and should not be interpreted causally.LAT, local ablative therapy; OS, overall survival; PFS, progression-free survival. Kaplan–Meier curves estimating the real-world PFS and OSbeyond osimertinib progressionstratified by receipt of LAT. (A, B) PFS2 and real-world OS of the oligoprogression cohort. (C, D) Real-world PFS beyond osimertinib progression (C) and OS (D) stratified by receipt of LAT. Descriptive comparison only. Kaplan–Meier curves for real-world PFS beyond osimertinib progression (LAT versus no-LAT) are shown for descriptive purposes, but the comparison is not intended for causal inference due to selection and immortal time biases and the small no-LAT subgroup. Log-rank P values are shown on the curves for descriptive purposes only and should not be interpreted causally. LAT, local ablative therapy; OS, overall survival; PFS, progression-free survival. In the OPD cohort, rwOS was 46.5 months (95% CI 34.4-nor estimable; Figure 2B), while rwPFS2 was 25 months (95% CI 21.5-45.1; Figure 2A), and rwPFSBOP was 7.62 months (95% CI 5.19-11.8). Most patients with OPD received LAT while continuing osimertinib. In descriptive subgroup analyses, rwPFSBOP appeared longer among patients treated with LAT (8.97 months, 95% CI 6.8-16.4) than among those who continued osimertinib without LAT (2.07 months, 95% CI 1.02-not estimable; Figure 2C), and rwOS showed a similar numerical trend (46.5 months, 95% CI 43.7-not estimable, versus 39.7 months, 95% CI 14.6-not estimable; Figure 2D). However, given the retrospective design, the very small no-LAT subgroup, and the potential for selection and immortal time biases, comparisons between LAT and no-LAT groups are reported descriptively and should not be interpreted as evidence of causal benefit. Of 180 patients who had a definitive progression to osimertinib, only 91 (50.5%) began second-line treatment (Figure 3), which mainly comprised a platinum-based doublet (e.g. carboplatin–pemetrexed) in 44 patients (48.35%), a clinical trial in 22 patients (24.18%), and use of immunotherapy monotherapy/combinations (with platinum-doublet and/or antiangiogenic drugs) in 15 patients (16.48%).Figure 3Sankey diagram showing the flow of the patients after receiving first-line osimertinib for advanced disease. Among the 91 patients who reached second-line treatment (50.5%), the most common second-line treatment was carboplatin–pemetrexed (48.35%), followed by access to a clinical trial (24.18%).CBDCA, carboplatin. Sankey diagram showing the flow of the patients after receiving first-line osimertinib for advanced disease. Among the 91 patients who reached second-line treatment (50.5%), the most common second-line treatment was carboplatin–pemetrexed (48.35%), followed by access to a clinical trial (24.18%). CBDCA, carboplatin. Re-biopsy at progression was carried out in 57 patients (31.67% of the 180 patients with documented progression) to assess histological transformation and resistance mechanisms by NGS when feasible. MET amplification was identified in 10 cases (17.54% of re-biopsies), and small-cell lung cancer transformation in 3 cases (5.26% of re-biopsies). Four of the 10 patients with MET amplification were enrolled in clinical trials evaluating osimertinib combined with a MET inhibitor (savolitinib or tepotinib). Amivantamab was administered to 11 patients in the second-line setting or beyond; however, exposure occurred across heterogeneous lines and regimens (including different clinical trial protocols), and only four patients received amivantamab outside a clinical trial (expanded access/compassionate use). Accordingly, amivantamab use is reported descriptively, and no comparative outcome analyses were carried out for amivantamab-treated patients. Discussion In this multicenter real-world cohort from eastern Spain, first-line osimertinib showed robust effectiveness in advanced EGFR-mutated NSCLC, with rwPFS close to that reported in FLAURA but lower rwOS in the overall population. Our study adds detailed Western European RWD2,21, 22, 23, 24 in a predominantly older and less selected cohort and also provides detailed information on OPD management with LAT while continuing osimertinib.2,21, 22, 23, 24 Compared with pivotal phase III trials, our population was older (median age 72 versus 64 in FLAURA, 64 in MARIPOSA, and 61 in FLAURA21,6,8), included 27.6% of patients with ECOG-PS ≥2, and had a high prevalence of brain metastases (42.5%). Although ORR and rwPFS were broadly consistent with FLAURA, rwOS was clearly shorter. This likely reflects the broader case mix of routine practice, including frailer patients and those with symptomatic CNS disease or other characteristics often underrepresented in trials. Supporting this interpretation, patients with ECOG-PS 0-1 in our cohort showed outcomes closer to those observed in FLAURA (Figure 1E and F). Western real-world series also report lower OS, and in some cases shorter PFS, than pivotal trials,2,21,23 underscoring an ongoing unmet need in routine practice. Our findings also highlight clinically relevant heterogeneity within EGFR-mutated disease. Patients with L858R had worse rwOS than those with exon 19 deletion, and baseline CNS metastases were associated with shorter rwPFS. Together with the substantial attrition observed after progression, these results support a risk-adapted first-line strategy. In our cohort, only half of patients with progression received second-line therapy, consistent with other real-world reports.2,21,23 This loss of post-progression treatment opportunity strengthens the rationale for considering upfront intensification in selected higher-risk patients, such as those with L858R mutations or baseline CNS disease. Progression patterns were also informative. OPD occurred in 28.33% of patients with documented progression, and most of these patients continued osimertinib and received LAT. Outcomes in the OPD cohort were encouraging, with rwOS of 46.5 months and rwPFSBOP of 7.62 months. Because most OPD patients received LAT, and the no-LAT subgroup was small, comparisons between LAT and no-LAT must remain descriptive. Even so, our results support the pragmatic clinical value of local treatment while maintaining EGFR inhibition in selected patients, in line with current guideline recommendations and retrospective studies in the literature.11,12,25,26 They should not, however, be interpreted as proof of causal benefit. One previous phase II trial reported no-LAT benefit for OPD, although it combined patients who received first-line osimertinib with patients who were pretreated (e.g. with first/second generation TKI or chemotherapy) and may explain the worse outcomes; pure first-line studies such as the ongoing LAT-FLOSI may offer further insight in this population.27 Although randomized evidence specifically evaluating LAT at acquired OPD remains limited, the recently reported randomized phase II NORTHSTAR trial evaluated local consolidative therapy after response to first-line osimertinib (oligopersistent/residual disease), rather than LAT delivered at the time of acquired OPD. In that context, NORTHSTAR reported improved PFS with consolidative local therapy (17.0 to 25.4 months; HR 0.60, P = 0.02), thereby supporting the broader principle that appropriately timed local therapy can deepen disease control, but it should not be interpreted as direct evidence for LAT in OPD.28,29 These results in the consolidation scenario are also supported by the study of Sampath et al.30 Randomized evidence more directly addressing LAT at acquired oligoprogression remains scarce. The phase II CURB trial evaluated stereotactic body radiotherapy (SBRT) for oligoprogressive cancers and reported a PFS benefit in the NSCLC cohort; however, most NSCLC cases lacked an actionable driver alteration, and only a small minority had a ‘driver mutation’ (three in standard of care and five in the SBRT arm, without further molecular specification), limiting the applicability of CURB to EGFR-mutated NSCLC treated with first-line osimertinib.31 It will be important to determine whether LAT for OPD can further enhance long-term outcomes beyond those reported with FLAURA2 and MARIPOSA. Re-biopsy at progression should be considered whenever feasible to identify actionable resistance mechanisms (e.g. MET amplification) or histologic transformation32,33; however, we acknowledge that feasibility constraints in routine practice limit uptake, as reflected by the re-biopsy rate in our cohort. The MARIPOSA-2 trial has shown improved PFS, including intracranial control, with chemotherapy plus amivantamab after osimertinib, adding a targeted option beyond progression. In our cohort, amivantamab exposure was limited and largely trial-driven, with only four patients treated outside clinical trials via expanded access/compassionate use; therefore, we did not undertake comparative analyses of outcomes with amivantamab-based regimens. Despite European regulatory approvals, real-world uptake may be constrained by country-level reimbursement and local access pathways, which can lag behind approvals and influence observed treatment patterns.34 In MET-driven resistance, osimertinib plus the MET inhibitor savolitinib has demonstrated clinically meaningful activity in a phase II trial (SAVANNAH),35 with confirmatory evaluation ongoing in the phase III SAFFRON trial36; key practical questions include defining an appropriate MET cut-off and ensuring assay standardization. Regarding immunotherapy-based strategies, randomized phase III evidence does not support routine addition of pembrolizumab to platinum–pemetrexed chemotherapy after EGFR TKI resistance: KEYNOTE-789 did not demonstrate significant improvements in PFS or OS with pembrolizumab plus platinum–pemetrexed versus chemotherapy alone.37 The IMpower150 trial (atezolizumab plus bevacizumab and carboplatin–paclitaxel) has been considered in selected cases after EGFR TKI failure38; however, the role of immunotherapy remains controversial and was used sparingly in our cohort. Another relevant observation is the relatively modest burden of time toxicity with osimertinib monotherapy in our cohort.15 This is clinically meaningful because newer combination regimens may improve efficacy at the cost of greater treatment burden and toxicity.39 Real-world studies comparing survival gains with the time cost of treatment will therefore be important, particularly in older or frailer populations. Methodologically, this retrospective RWD study adhered to recognized conventions (rwPFS, rwOS, real-world time to discontinuation, and real-world time to new therapy), reports RWD from an older Western population, and provides granular insight into OPD and LAT. Alignment with ESMO-GROW recommendations40 and Friends of Cancer Research endpoint definitions41 should enhance interpretability and reproducibility. Limitations include the retrospective design, the limited size of some subgroups, the use of PCR rather than NGS in a subset of patients, and incomplete re-biopsy at progression. In particular, the OPD/LAT analyses are exploratory and susceptible to selection and immortal time biases. Overall, our data support several clinical messages: firstly, osimertinib remains highly effective in routine practice, especially in fitter patients; secondly, real-world survival in an unselected Western cohort is lower than that in pivotal trials; thirdly, substantial second-line attrition supports careful consideration of upfront intensification in selected high-risk patients; and fourthly, continuation of osimertinib with LAT is a pragmatic strategy for selected patients with oligoprogression, while re-biopsy at progression should be prioritized whenever feasible. Conclusions In a large Spanish real-world cohort, first-line osimertinib demonstrated robust effectiveness, yet mOS was substantially lower than in FLAURA, particularly in specific subgroups (e.g. EGFR L858R, exon 21). Moreover, approximately half of patients progressing on osimertinib did not receive second-line therapy. Considering the survival gains reported in MARIPOSA and FLAURA2, adoption of combination strategies as standard options is justified to reduce loss of opportunity while maintaining a risk-adapted approach for frail patients. At progression, re-biopsy should be prioritized to guide histology- and mechanism-directed treatments (e.g. MET amplification), and LAT for OPD offers meaningful additional disease control; its integration alongside contemporary first-line combinations warrants further study.