Afatinib Versus Osimertinib for Non‐Small Cell Lung Cancer With Uncommon EGFR Mutations: Real‐World Outcomes Authors: Yutaro Nagano, Hiroshi Yokouchi, Ryota Saito, Osamu Honjo, Toshiyuki Sumi, Takayuki Kaburagi, Kenichiro Ito, Yasuyuki Ikezawa, Keiki Yokoo, Hajime Kikuchi, Sayaka Kudo, Masami Yamazoe, Ryota Shigaki, Keita Tagami, Toshiyuki Tenma, Tatsuru Ishikawa, Mamoru Takahashi, Takaaki Sasaki, Hajime Asahina, Shiro Hinotsu, Hirofumi Chiba, Satoshi Oizumi Published: 2026 ABSTRACT ABSTRACTNon–small cell lung cancers (NSCLC) harboring uncommon epidermal growth factor receptor (EGFR) mutations (UMs) are clinically heterogeneous, and the optimal first‐line EGFR tyrosine kinase inhibitor (EGFR‐TKI), as well as the role of subsequent immune checkpoint inhibitor (ICI)–based regimens, remains uncertain. We conducted a multicenter retrospective cohort study including patients with advanced or recurrent non‐squamous NSCLC harboring UMs who received first‐line afatinib or osimertinib between January 2015 and January 2024 at 29 hospitals in Japan, excluding exon 20 insertions and de novo T790M mutations. Inverse probability of treatment weighting adjusted baseline imbalances, and post–EGFR‐TKI treatment patterns were evaluated. Among 162 patients (afatinib, n = 95; osimertinib, n = 67), weighted analyses demonstrated no significant differences in time to treatment failure (hazard ratio [HR], 1.04; 95% confidence interval [CI], 0.61–1.77) or overall survival (HR, 1.34; 95% CI, 0.73–2.44). Afatinib was associated with higher response rates and more frequent adverse events requiring dose reduction. Subgroup analyses suggested differential treatment effects according to mutation subtype, with osimertinib favoring L861X mutations and afatinib favoring compound mutations. Among 56 patients who received subsequent systemic therapy, clinical outcomes were comparable between ICI plus platinum doublet and platinum doublet alone. These findings indicate that afatinib and osimertinib provide comparable survival outcomes as first‐line therapies for NSCLC with UMs, with treatment effects varying by molecular subtype, while subsequent ICI‐based regimens may confer limited additional benefit. Abbreviations adverse event covariate balancing propensity score inverse probability of treatment weighting using covariate balancing propensity score confidence interval common mutation Common Terminology Criteria for Adverse Events disease control rate Eastern Cooperative Oncology Group performance status epidermal growth factor receptor epidermal growth factor receptor‐targeted tyrosine kinase inhibitor generalized linear model inverse probability of treatment weighting using generalized linear model Hokkaido Lung Cancer Clinical Study Group hazard ratio immune checkpoint inhibitor immune checkpoint inhibitor plus chemotherapy interstitial lung disease inverse probability of treatment weighting not available neuroendocrine carcinoma non–small cell lung cancer objective response rate overall survival programmed death‐ligand 1 progression‐free survival standardized mean difference time to treatment failure uncommon mutation vascular endothelial growth factor Introduction In patients with non‐squamous non–small cell lung cancer (NSCLC), mutations in the epidermal growth factor receptor (EGFR) gene are the most common genetic alterations observed, particularly in Asian populations [1]. The introduction of EGFR‐targeted tyrosine kinase inhibitors (EGFR‐TKIs) improved the prognosis of patients with EGFR‐mutant NSCLC [2, 3, 4, 5, 6, 7, 8, 9, 10]. The majority of EGFR mutations are common mutations (CMs), comprising exon 19 deletions and the exon 21 L858R point mutation. However, uncommon mutations (UMs), which account for approximately 10% of all EGFR mutations, are generally excluded from clinical trials [11]. Afatinib, a second‐generation EGFR‐TKI, and osimertinib, a third‐generation EGFR‐TKI, have demonstrated clinical activity against UMs. Afatinib has been evaluated in randomized trials compared to chemotherapy and in pooled post hoc analyses of prospective studies, showing effectiveness in patients with UMs [12, 13]. Similarly, osimertinib has been assessed in single‐arm prospective studies with promising results [14, 15, 16]. However, no direct head‐to‐head trials have been conducted to compare the efficacy of these EGFR‐TKIs in patients with UMs, rendering the relative effectiveness of these agents unclear. Data on subsequent treatments for patients with UMs are also limited. While recent studies suggest minimal benefit from adding immune checkpoint inhibitors (ICIs) to the chemotherapy of patients with CMs [17, 18, 19, 20, 21], evidence in patients with UMs is confined to a few small‐scale observational studies [22, 23]. Thus, the optimal strategy for first‐line and subsequent therapies in patients with UMs remains unclear. We therefore conducted a multicenter retrospective observational study to investigate the real‐world treatment patterns and outcomes of patients with non‐squamous NSCLC harboring UMs who received EGFR‐TKIs as first‐line therapy. Material and Methods Study Design and Patient Cohort This multicenter retrospective observational study (HOT‐Next‐001/HOT2501) was conducted in collaboration with the Hokkaido Lung Cancer Clinical Study Group (HOT) and involved 29 institutions across Japan. Patients with non‐squamous NSCLC harboring UMs who received first‐line EGFR‐TKI between January 2015 and January 2024 were included. To evaluate the optimal first‐line EGFR‐TKI for patients with UMs, the patients were categorized by treatment received: afatinib or osimertinib. Subsequent therapies after first‐line EGFR‐TKI discontinuation were classified as ICI plus chemotherapy (ICI‐Chemo), platinum doublet, ICI monotherapy, non‐platinum chemotherapy, and EGFR‐TKI rechallenge. We also compared ICI‐Chemo with platinum doublet among patients who were both chemotherapy‐ and ICI‐naïve at the time of EGFR‐TKI discontinuation (e.g., patients receiving second‐line therapy after EGFR‐TKI or third‐line therapy after sequential EGFR‐TKI treatments), to evaluate the potential added benefit of combining ICIs with platinum doublet chemotherapy in the post–EGFR‐TKI setting. If patients received both regimens sequentially, the first regimen administered was considered for analysis (e.g., patients treated with a platinum doublet followed by ICI‐Chemo were classified in the platinum doublet group). The eligible patients had histologically or cytologically confirmed non‐squamous NSCLC with UMs in the advanced or recurrent setting. UMs were defined as EGFR alterations other than exon 19 deletions or exon 21 L858R, excluding de novo T790M and exon 20 insertions. Patients were excluded if they had neuroendocrine histology, other active driver mutations, concurrent active malignancies, or prior EGFR‐TKI or ICI use, or were deemed ineligible due to substantial missing data or extremely short follow‐up. Patients with compound mutations involving both CMs and UMs were included, except for cases with de novo T790M or exon 20 insertions. Assessment and Endpoints Efficacy outcomes were evaluated by investigators at each participating institution according to the RECIST v1.1 criteria, without central radiologic review. The objective response rate (ORR) was defined as the proportion of patients achieving a confirmed complete or partial response, and the disease control rate (DCR) was defined as the proportion of patients achieving a complete response, partial response, or stable disease. Time to treatment failure (TTF) was measured from treatment initiation until treatment discontinuation, death, or censoring. Progression‐free survival (PFS) was measured from treatment initiation until documented progression, death, or censoring. Overall survival (OS) was measured from the initiation of first‐line EGFR‐TKI until death or censoring. Post–EGFR‐TKI OS was measured from the initiation of subsequent therapy until death or censoring. Patients without an event were censored at the last follow‐up before the cutoff (April 2025). Adverse events (AEs) were assessed only for first‐line EGFR‐TKIs, according to Common Terminology Criteria for Adverse Events (CTCAE) v5.0. Hematologic toxicities, interstitial lung disease (ILD), and hepatotoxicity were recorded regardless of grade, while other AEs were documented if grade ≥ 3. Statistical Analysis Propensity score methods with inverse probability of treatment weighting (IPTW) were used to adjust for baseline characteristics. Covariates for each comparison are listed in Table S1. Propensity scores were estimated using the covariate balancing propensity score (CBPS) method, which directly optimizes covariate balance based on the generalized method of moments [24, 25]. Efficacy comparisons were primarily conducted using IPTW with CBPS‐derived scores (CBPS‐IPTW). Sensitivity analysis was performed using IPTW estimated by a generalized linear model (GLM) with logistic regression (GLM‐IPTW). Covariate balance was evaluated before and after each adjustment method using standardized mean difference (SMD), with SMD < 0.1 indicative of adequate balance. Survival outcomes were analyzed using the Kaplan–Meier method, log‐rank test, and Cox proportional hazards models. ORR and DCR were compared using Fisher's exact test in unweighted analyses and logistic regression models in IPTW‐adjusted analyses. All statistical tests were two‐sided, and p‐values < 0.05 were considered statistically significant. Analyses were performed using the R software (version 4.4.2; R Foundation for Statistical Computing, Vienna, Austria). Results Patient Disposition and Baseline Characteristics of the First‐Line EGFR‐TKI Cohort During the study period, 199 patients with non‐squamous NSCLC harboring UMs received first‐line EGFR‐TKI therapy. After excluding 30 patients due to exon 20 insertions, de novo T790M, active concomitant malignancies, or mixed histology, 169 patients were included in the final analysis (Figure S1). Seven of these patients received first‐generation EGFR‐TKIs (gefitinib, n = 4; erlotinib, n = 3), whereas afatinib and osimertinib were administered to 95 and 67 patients, respectively. Given the limited number of patients treated with first‐generation agents, comparative analyses focused on the 162 patients who received afatinib or osimertinib. Before adjustment, patients in the afatinib group tended to be younger and more frequently exhibited an Eastern Cooperative Oncology Group performance status of zero compared to those in the osimertinib group (Table S2). Mutation subtype analysis revealed that the G719X mutation was more prevalent in the afatinib group. Moreover, compound mutations with CMs and UMs were more frequent in the osimertinib and afatinib groups, respectively. After IPTW adjustment, baseline imbalances were substantially reduced (Figure S2). With GLM‐IPTW, all covariates except age and stage IV disease achieved an SMD of < 0.1, and CBPS‐IPTW further improved covariate balance, with SMDs approaching zero across all variables. Effectiveness of First‐Line Afatinib vs. Osimertinib After CBPS‐IPTW The median follow‐up durations for censored cases in the afatinib and osimertinib groups were 26.5 months (range, 9.1–118.2 months) and 21.5 months (range, 0.7–73.1 months), respectively. After CBPS‐IPTW adjustment, the afatinib group demonstrated significantly higher ORR (70.2% vs. 47.2%; p = 0.029) and DCR (93.7% vs. 82.5%; p = 0.041) compared to the osimertinib group. The median TTF was 11.0 and 6.5 months in the afatinib and osimertinib groups, respectively (HR, 1.04; 95% CI, 0.61–1.77), and the median PFS was 11.3 and 5.9 months, respectively (HR, 1.04; 95% CI, 0.60–1.80), with no significant difference between the groups (Figure 1A,B). Moreover, the median OS of the afatinib and osimertinib groups was 28.0 and 25.5 months, respectively (HR, 1.34; 95% CI, 0.73–2.44), with no significant intergroup difference (Figure 1C). Time to treatment failure (A), progression‐free survival (B), and overall survival (C) after CBPS‐IPTW in patients treated with afatinib or osimertinib. CBPS, covariate balancing propensity score; CI, confidence interval; HR, hazard ratio; IPTW, inverse probability of treatment weighting; OS, overall survival; PFS, progression‐free survival; TTF, time to treatment failure. Overall, findings from subgroup analysis were consistent with those from the main analysis (Figure 2). EGFR variant and compound mutation status exhibited distinct patterns: patients harboring the L861X mutation exhibited more favorable outcomes with osimertinib, including a significant OS advantage, whereas those with compound mutations—particularly involving CMs—benefited more from afatinib. Stratification by PD‐L1 expression showed that among patients with PD‐L1 ≥ 50%, afatinib achieved longer TTF compared to osimertinib (HR, 0.34; 95% CI, 0.15–0.79), with a numerically longer OS (HR, 0.38; 95% CI, 0.07–1.93). In patients with brain metastases, TTF was significantly longer with afatinib than osimertinib (HR 0.46; 95% CI 0.23–0.91). To better understand the possible reasons for this finding, we re‐evaluated baseline characteristics within the brain metastasis subgroup after CBPS‐IPTW adjustment (Table S3). Several variables remained imbalanced between treatment groups. In particular, patients in the afatinib group tended to be younger, less likely to have an ECOG PS ≥ 2, and more likely to have received local treatment for brain metastases. Subgroup analyses of time to treatment failure (A) and overall survival (B) in patients treated with afatinib or osimertinib after CBPS‐IPTW. CBPS, covariate balancing propensity score; CI, confidence interval; CMs, common mutations; ECOG PS, Eastern Cooperative Oncology Group Performance Status; EGFR, epidermal growth factor receptor; HR, hazard ratio; IPTW, inverse probability of treatment weighting; PD‐L1, programmed death‐ligand 1; UMs, uncommon mutations. Sensitivity Analyses To assess the robustness of the results of CBPS‐IPTW, additional comparisons were performed using unadjusted data and GLM‐IPTW. Afatinib was consistently associated with higher ORR and DCR compared to osimertinib across all analytical approaches (Table S4), and results for TTF and OS were generally consistent with those obtained using CBPS‐IPTW, with no significant differences between the groups (Figure S3). Unadjusted subgroup analyses yielded similar overall trends observed with CBPS‐IPTW, including the disadvantage of osimertinib in tumors with PD‐L1 ≥ 50%, the favorable outcomes with osimertinib in tumors with L861X mutation, and the greater benefit of afatinib in patients with compound mutations (Figure S4). Treatment Exposure, Safety, and Treatment Sequencing of Afatinib and Osimertinib Treatment exposure and dose modifications are summarized in Table S5. The starting dose and subsequent dose reductions were determined at the discretion of the treating physician. Standard dose was defined as afatinib 40 mg/day and osimertinib 80 mg/day. Reduced starting doses and dose reductions during treatment were substantially more frequent in the afatinib group than in the osimertinib group (reduced starting dose: 41.1% vs. 6.0%; dose reduction during treatment: 63.2% vs. 26.9% for afatinib and osimertinib, respectively). Overall, treatment discontinuation occurred in 84 (88.4%) patients receiving afatinib and 59 (88.1%) receiving osimertinib. Disease progression was the most frequent reason for discontinuation (60.0% vs. 64.2%), followed by adverse events (22.1% vs. 19.4%). The overall incidence of grade ≥ 3 AEs was higher in the afatinib group compared to the osimertinib group (38.9% vs. 25.4%, Table 1). Treatment‐related deaths occurred in one patient in each group, both attributable to ILD. Dermatologic toxicities and diarrhea were more frequent and severe in the afatinib group, whereas hematologic toxicities such as thrombocytopenia and anemia were more common in the osimertinib group. TABLE 1Adverse events in patients treated with afatinib or osimertinib.Adverse events a , no. (%)Incidence of adverse eventsAfatinib (n = 95)Osimertinib (n = 67)Any eventsAny grade50 (52.6)36 (53.7)Grade ≥ 337 (38.9)17 (25.4)Treatment‐related deaths1 (1.1)1 (1.5)LeukopeniaAny grade2 (2.1)5 (7.5)Grade ≥ 30 (0)1 (1.5)AnemiaAny grade9 (9.5)7 (10.4)Grade ≥ 30 (0)1 (1.5)ThrombocytopeniaAny grade1 (1.1)8 (11.9)Grade ≥ 30 (0)1 (1.5)Interstitial lung diseaseAny grade12 (12.6)8 (11.9)Grade ≥ 37 (7.4)3 (4.5)HepatotoxicityAny grade12 (12.6)10 (14.9)Grade ≥ 32 (2.1)2 (3.0)Dermatologic toxicity b Grade ≥ 312 (12.6)1 (1.5)Enteritis/DiarrheaGrade ≥ 314 (14.7)2 (3.0)FatigueGrade ≥ 30 (0)2 (3.0)Anorexia/NauseaGrade ≥ 34 (4.2)3 (4.5)Others c Grade ≥ 32 (2.1)4 (6.0) a Leukopenia, anemia, thrombocytopenia, interstitial lung disease, and hepatotoxicity are reported as any grade; all other adverse events are reported as Grade ≥ 3. b Dermatologic toxicity includes paronychia and stomatitis. c In the afatinib group, Grades 3 and 4 events included fever and renal dysfunction, respectively. In the osimertinib group, Grade 3 events included pleuritis, pericarditis, and pulmonary embolism, and Grade 4 events included elevated pancreatic enzyme levels. Adverse events in patients treated with afatinib or osimertinib. Leukopenia, anemia, thrombocytopenia, interstitial lung disease, and hepatotoxicity are reported as any grade; all other adverse events are reported as Grade ≥ 3. Dermatologic toxicity includes paronychia and stomatitis. In the afatinib group, Grades 3 and 4 events included fever and renal dysfunction, respectively. In the osimertinib group, Grade 3 events included pleuritis, pericarditis, and pulmonary embolism, and Grade 4 events included elevated pancreatic enzyme levels. At data cutoff, the proportions of patients remaining on first‐line treatment were comparable between the two groups (11.6% vs. 11.9%). Transition to second‐line therapy occurred in 51.6% and 47.8% of patients in the afatinib and osimertinib groups, respectively (Figure S5). Subsequent treatment patterns differed: patients initially treated with afatinib more frequently received platinum doublet chemotherapy, whereas those treated with osimertinib were more often rechallenged with another EGFR‐TKI. Among the rechallenged patients, the second‐line EGFR‐TKIs included osimertinib (n = 4) or erlotinib plus an anti‐vascular endothelial growth factor (VEGF) agent (n = 1) after afatinib and afatinib (n = 10), gefitinib (n = 1), or erlotinib plus anti‐VEGF (n = 1) after osimertinib. Across all treatment sequences, T790M mutations were detected in only two patients (3.5%) among those who discontinued afatinib. Effectiveness of Subsequent Therapies Among 87 patients who received subsequent treatment after first‐line EGFR‐TKI, a total of 198 regimens were administered (Figure S6). The most common regimen was non‐platinum chemotherapy (n = 68), followed by EGFR‐TKI rechallenge (n = 54), platinum doublet (n = 43), ICI‐Chemo (n = 21), and ICI monotherapy (n = 12). Most ICI‐Chemo and platinum doublet regimens were administered in the third‐line setting, whereas rechallenge with the same treatment category became more frequent beyond the fourth line of treatment. Among the patients who received subsequent therapy, the overall median follow‐up duration for censored cases from the initiation of subsequent therapy was 15.4 months (range, 3.7–66.2). Median PFS was 5.8 months for patients who received platinum doublet, 7.3 months for ICI‐Chemo, 3.9 months for non‐platinum chemotherapy, 5.0 months for EGFR‐TKI, and 2.3 months for ICI monotherapy, across all regimens (Figure 3A). ICI monotherapy yielded the shortest median PFS and was significantly inferior to platinum doublet (HR 0.46; 95% CI, 0.23–0.91; p = 0.023) and ICI‐Chemo regimens (HR 0.47; 95% CI, 0.21–1.02; p = 0.049). Progression‐free survival after subsequent therapy for all treatment lines (A) and up to third‐line therapy (B). CI, confidence interval; Chemo, cytotoxic chemotherapy; EGFR‐TKI, epidermal growth factor receptor–tyrosine kinase inhibitor; HR, hazard ratio; ICI, immune checkpoint inhibitor; PFS, progression‐free survival. To minimize potential bias due to repeated regimens in later lines, PFS was reanalyzed by restricting to therapies administered up to the third line of treatment (Figure 3B). Again, ICI monotherapy demonstrated significantly shorter PFS compared to platinum doublet (HR 0.34; 95% CI, 0.13–0.90; p = 0.022) and ICI‐Chemo regimens (HR 0.27; 95% CI, 0.08–0.79; p = 0.010). Comparative Effectiveness of Subsequent ICI‐Chemo vs. Platinum Doublet Given that ICI monotherapy demonstrated the poorest outcomes, we focused on comparing ICI‐Chemo and platinum doublet regimens to evaluate the potential incremental benefit of adding ICIs to platinum‐based chemotherapy. Among the patients who were chemotherapy‐ and ICI‐naïve at the time of EGFR‐TKI discontinuation, 19 and 37 received ICI‐Chemo and platinum doublet, respectively (Table S6). We observed imbalances in baseline characteristics before adjustment, including age and PFS of first‐line EGFR‐TKI. After CBPS‐IPTW, all covariates achieved an SMD of < 0.1, indicating adequate balance (Table S6, Figure S7). The median follow‐up duration for censored cases was 9.6 months (range, 3.7–22.6) and 15.2 months (range, 7.1–66.2) in the ICI‐Chemo and platinum doublet groups, respectively. After CBPS‐IPTW, the ICI‐Chemo and platinum doublet groups exhibited ORRs of 23.9% and 47.3% and DCRs of 82.4% and 82.7%, respectively, with no intergroup differences in either parameter. Regarding time‐to‐event outcomes, median PFS was 6.7 and 7.3 months for ICI‐Chemo and platinum doublet, respectively (HR, 1.08; 95% CI, 0.53–2.19), with no significant intergroup difference (Figure 4). Post–EGFR‐TKI OS was also comparable between the groups (14.6 vs. 19.7 months; HR, 1.30; 95% CI, 0.63–2.68). Subgroup analyses stratified by PD‐L1 expression and other clinical factors showed no subgroup with a clear advantage of ICI‐Chemo over platinum doublet (Figure S8). Progression‐free survival (A) and overall survival (B) after CBPS‐IPTW in patients treated with subsequent ICI‐Chemo or platinum doublet therapy. CBPS, covariate balancing propensity score; CI, confidence interval; HR, hazard ratio; ICI, immune checkpoint inhibitor; IPTW, inverse probability of treatment weighting; OS, overall survival; PFS, progression‐free survival. To validate the robustness of the comparison of the efficacies of ICI‐Chemo and platinum doublet, sensitivity analyses were performed using unadjusted and GLM‐IPTW datasets. As summarized in Table S7, ORR and DCR did not differ significantly between groups across analytical approaches. Similarly, PFS and post–EGFR‐TKI OS were comparable between the ICI‐Chemo and platinum doublet groups, which was consistent with the findings from primary CBPS‐IPTW analysis (Figure S9). Discussion In this study, we evaluated the efficacy of first‐line EGFR‐TKIs and subsequent therapies in patients with non‐squamous NSCLC harboring UMs, excluding exon 20 insertions and de novo T790M. After adjusting for baseline characteristics using IPTW, we found that afatinib was associated with higher ORR and DCR compared to osimertinib, whereas OS outcomes were comparable. Subgroup analyses indicated heterogeneity of efficacy by PD‐L1 expression, EGFR subtype, and compound mutation status. In the post–EGFR‐TKI setting, ICI monotherapy showed limited benefit, and ICI‐Chemo did not yield better outcomes compared to platinum doublet. Our study represents the largest retrospective observational cohort to date, directly comparing first‐line EGFR‐TKIs in patients with UMs while systematically characterizing subsequent post–EGFR‐TKI treatments in this population. Direct comparative analyses of afatinib and osimertinib in UMs are scarce. Prior pooled and real‐world studies have suggested broadly similar efficacy with heterogeneity across subtypes [26, 27]. In our subgroup analyses, outcomes after osimertinib use were poorer in tumors with PD‐L1 ≥ 50%. Similar trends of reduced efficacy have been reported in patients with CMs [28, 29, 30]. However, given the limited number of cases and wide CIs, it remains unclear whether this difference is clinically meaningful. In this subgroup analysis, TTF was notably longer in patients with brain metastases treated with afatinib compared to those treated with osimertinib. However, osimertinib has been shown to have better central nervous system penetration and intracranial efficacy than first‐generation EGFR‐TKIs [10]. To better understand this unexpected result, we further explored baseline characteristics within the brain metastases subgroup. Several variables, including age, ECOG PS, and local treatments for brain metastases, remained imbalanced between the treatment groups despite CBPS‐IPTW adjustment, suggesting residual selection bias. Therefore, the subgroup findings, especially those related to PD‐L1 expression and brain metastases, should be interpreted with caution. However, these findings may also generate hypotheses for future studies exploring how mutation subtype and clinical context, including the presence of brain metastases, influence the activity of different EGFR‐TKIs. EGFR mutation subtypes are another key determinant of TKI sensitivity. The L861X mutation structurally resembles CMs [31], and in vitro functional studies using Ba/F3 cells have shown that L861Q exhibits half‐maximal inhibitory concentration values similar to CMs [32], supporting the favorable clinical activity of osimertinib observed in this subgroup [14, 15, 16]. Conversely, compound mutations have been associated with greater sensitivity to afatinib [33, 34], a pattern supported by our findings. These results highlight the potential importance of integrating mutation subtype into treatment selection for patients with UMs. Previous analyses have shown that clinical efficacy may be maintained with both a reduced starting dose and dose reductions during treatment, supporting the notion that afatinib can be effectively managed through dose adjustments [35, 36]. Indeed, treatment discontinuation rates in our cohort were comparable (22.1% with afatinib vs. 19.4% with osimertinib) despite the higher incidence of both initial dose reduction and dose reductions during treatment with afatinib. In contrast, osimertinib was associated with fewer high‐grade gastrointestinal or dermatologic AEs but more hematologic toxicities compared to afatinib, which was consistent with the published safety profiles of osimertinib in patients with UMs [14, 15, 16]. Thus, afatinib and osimertinib are manageable in clinical practice, albeit with distinct toxicity profiles. These differences in AE patterns and the need for dose adjustment should be considered when selecting first‐line EGFR‐TKIs for patients with UMs. Evidence regarding ICIs in UMs is limited. Although G719X has been linked to higher tumor mutation burden [37], there are few reports of a meaningful benefit from ICI in this setting [22, 23]. In our study, ICI monotherapy was associated with the shortest PFS, significantly inferior to that of platinum‐based regimens. Furthermore, the addition of ICI to platinum doublet therapy provided no incremental benefit, even after IPTW adjustment. These findings indicated that ICIs are unlikely to offer meaningful benefits in patients with UMs, and platinum doublet should remain the preferred option after EGFR‐TKI failure. Several limitations of this study must be acknowledged. First, as a retrospective observational study, selection bias cannot be entirely excluded; however, CBPS‐IPTW and sensitivity analyses with GLM‐IPTW were applied to minimize the imbalance. Second, the sample size was relatively small, particularly for subgroup and subsequent therapy analyses, which limits the statistical power of our results. Third, outcomes were investigator‐assessed using RECIST v1.1 without central review, and imaging intervals were not standardized. To address this limitation, we evaluated TTF, which may represent a more pragmatic endpoint in real‐world clinical practice. Finally, as this study was conducted exclusively in Japanese institutions, the generalizability of our findings to non‐Asian populations and different healthcare systems may be limited. Therefore, our findings should be interpreted as hypothesis‐generating real‐world evidence rather than definitive guidance for treatment selection. In conclusion, this multicenter analysis demonstrated comparable survival outcomes between afatinib and osimertinib in patients with UMs, with differential efficacy suggested in subgroups defined by EGFR subtype and compound mutations. In subsequent therapy, ICI‐based regimens showed limited benefits, and platinum doublet therapy might remain the standard of care. These findings underscore the importance of individualizing treatment selection in patients with UMs based on clinical and molecular contexts. Author Contributions Yutaro Nagano: conceptualization, methodology, software, data curation, investigation, formal analysis, funding acquisition, visualization, project administration, resources, writing – original draft, writing – review and editing. Hiroshi Yokouchi: methodology, project administration, investigation, resources, writing – review and editing. Ryota Saito: resources, investigation, writing – review and editing. Osamu Honjo: investigation, resources, writing – review and editing. Toshiyuki Sumi: investigation, resources, writing – review and editing. Takayuki Kaburagi: investigation, writing – review and editing, resources. Kenichiro Ito: investigation, writing – review and editing, resources. Yasuyuki Ikezawa: investigation, writing – review and editing, resources. Keiki Yokoo: investigation, writing – review and editing, resources. Hajime Kikuchi: investigation, writing – review and editing, resources. Sayaka Kudo: investigation, writing – review and editing, resources. Masami Yamazoe: investigation, writing – review and editing, resources. Ryota Shigaki: methodology, project administration, resources, investigation, writing – review and editing. Keita Tagami: methodology, project administration, writing – review and editing. Toshiyuki Tenma: methodology, project administration, writing – review and editing, investigation, resources. Tatsuru Ishikawa: writing – review and editing, project administration, resources, methodology, investigation. Mamoru Takahashi: supervision, writing – review and editing. Takaaki Sasaki: methodology, project administration, writing – review and editing. Hajime Asahina: methodology, writing – review and editing, project administration. Shiro Hinotsu: validation, supervision, writing – review and editing. Hirofumi Chiba: supervision, writing – review and editing. Satoshi Oizumi: supervision, writing – review and editing. Funding This work was supported by a grant from the Sapporo Medical University Grants for Programs promoting Academic advancements (grant number 246, June 30, 2025). The funder had no involvement in the study design, data collection, analysis, or decision to publish. Ethics Statement The study was approved by the Institutional Review Board of Sapporo Medical University Hospital (approval no. 362‐251, April 10, 2025) and conducted in accordance with the Declaration of Helsinki. Consent Informed consent for this study was obtained through an opt‐out approach under the Ethical Guidelines for Life Science and Medical Research Involving Human Subjects in Japan. Conflicts of Interest Yutaro Nagano received honoraria from MSD K.K., AstraZeneca, Chugai Pharmaceutical, Taiho Pharmaceutical, and Kyowa Kirin. Hiroshi Yokouchi received honoraria from AstraZeneca and Chugai Pharmaceutical, and has received research funding from AbbVie, Amgen, AstraZeneca, Bristol Myers Squibb, Chugai Pharmaceutical, Daiichi Sankyo, MSD K.K., Sanofi, and Takeda Pharmaceutical. Ryota Saito received honoraria from AstraZeneca, Chugai Pharmaceutical, MSD K.K., Taiho Pharmaceutical, Merck Biopharma, Ono Pharmaceutical, Daiichi Sankyo, Takeda Pharmaceutical, Kyorin Pharmaceutical, and Sanofi. Toshiyuki Sumi received honoraria from AstraZeneca, Boehringer Ingelheim, Ono Pharmaceutical, Chugai Pharmaceutical, MSD K.K., Takeda Pharmaceutical, Nippon Kayaku, Kyowa Kirin, Merck Biopharma, and Eli Lilly Japan. Keiki Yokoo received honoraria from Bristol Myers Squibb, AstraZeneca, Eli Lilly Japan, Chugai Pharmaceutical, Boehringer Ingelheim Japan, Pfizer Japan, Merck Serono, Takeda Pharmaceutical, Daiichi Sankyo, MSD K.K., Nippon Kayaku, Kyowa Kirin, and Taiho Pharmaceutical. Hajime Kikuchi received honoraria from AstraZeneca, Chugai Pharmaceutical, MSD K.K., Merck Biopharma, Nippon Kayaku, Bristol Myers Squibb, Ono Pharmaceutical, Daiichi Sankyo, and Takeda Pharmaceutical. Tatsuru Ishikawa received honoraria from AstraZeneca and Eli Lilly Japan. Takaaki Sasaki received honoraria from AstraZeneca, Chugai Pharmaceutical, and Amgen. Hajime Asahina received honoraria from AstraZeneca, Eli Lilly Japan, Chugai Pharmaceutical, MSD K.K., Bristol Myers Squibb, and Ono Pharmaceutical, and has received research funding from AstraZeneca. Satoshi Oizumi received honoraria from AstraZeneca, Chugai Pharmaceutical, and MSD K.K., and has received research funding from AbbVie, Amgen, AstraZeneca, Bristol Myers Squibb, Chugai Pharmaceutical, Daiichi Sankyo, Ono Pharmaceutical, Pfizer, MSD K.K., Sanofi, Taiho Pharmaceutical, and Takeda Pharmaceutical. The other authors declare no conflicts of interest. Supporting information Figure S1: Patient eligibility flow diagram. Figure S2:. Covariate balance between afatinib and osimertinib before propensity score analysis, after CBPS‐IPTW, and after GLM‐IPTW. Figure S3:. Time to treatment failure (A) before adjustment and (B) after GLM‐IPTW, and overall survival (C) before adjustment and (D) after GLM‐IPTW in patients treated with afatinib and osimertinib. Figure S4:. Subgroup analyses of time to treatment failure (A) and overall survival (B) in patients treated with afatinib or osimertinib before adjustment. Figure S5:. Subsequent treatment options after first‐line afatinib (A) and osimertinib (B). Figure S6:. Treatment pathways beyond first‐line EGFR‐TKI in patients with UMs. Figure S7:. Covariate balance between ICI‐Chemo and platinum doublet before propensity score analysis, after CBPS‐IPTW, and after GLM‐IPTW. Figure S8:. Subgroup analyses of progression‐free survival (A) and overall survival (B) in patients treated with ICI‐Chemo or platinum doublet after CBPS‐IPTW. Figure S9:. Progression‐free survival (A) before adjustment and (B) after GLM‐IPTW, and overall survival (C) before adjustment and (D) after GLM‐IPTW in patients treated with ICI‐Chemo and platinum doublet as subsequent therapy. Table S1:. Covariates used for propensity score estimation. Table S2:. Baseline characteristics of patients treated with afatinib or osimertinib before propensity score analysis and after IPTW. Table S3: Baseline characteristics of patients with brain metastases after CBPS‐IPTW. Table S4:. ORR and DCR of afatinib versus osimertinib across unadjusted and propensity score–weighted analyses. Table S5: Treatment exposure and dose modifications. Table S6:. Baseline characteristics of patients treated with ICI‐Chemo or platinum doublet as subsequent therapy before propensity score analysis and after IPTW. Table S7:. ORR and DCR of ICI‐Chemo versus platinum doublet across unadjusted and propensity score–weighted analyses.