Osimertinib in EGFR-mutated non-small cell lung cancer: a comprehensive narrative review of clinical evidence

INTRODUCTION

Lung cancer remains the leading cause of cancer-related mortality worldwide, and non-small cell lung cancer (NSCLC) accounts for approximately 85% of all lung cancer diagnoses^[1]. The therapeutic landscape of cancer has undergone a profound transformation over the past two decades, evolving from empirical cytotoxic chemotherapy to molecularly targeted and immunologically driven strategies that exploit tumor-specific vulnerabilities. Among the molecularly defined subsets of NSCLC, tumors harboring activating mutations in the epidermal growth factor receptor (EGFR) gene exemplify this paradigm shift, principally exon 19 deletions (ex19del) and the exon 21 L858R point mutation, which represent a clinically distinct entity characterized by sensitivity to EGFR tyrosine kinase inhibitors (TKIs)^[2-4]. The identification of these driver mutations and the subsequent development of genotype-directed therapies marked the advent of precision oncology in thoracic malignancies. First-generation (gefitinib, erlotinib) and second-generation (afatinib, dacomitinib) EGFR TKIs have demonstrated initial response rates of 60% to 70%, yet acquired resistance invariably develops within 9 to 14 months, with the EGFR T790M gatekeeper mutation accounting for approximately 50% to 60% of resistance cases^[5-10]. This unmet clinical need has led to the development of osimertinib (AZD9291, Tagrisso™), a third-generation EGFR TKI selectively designed to inhibit both EGFR-sensitizing mutations and the T790M resistance mutation while sparing wild-type EGFR^[4,11]. Since its initial accelerated approval for T790M-positive disease in 2015, the clinical development program of osimertinib has expanded markedly, from second-line salvage therapy to first-line metastatic treatment and subsequently to curative-intent perioperative and locally advanced treatment. This trajectory represents one of the most comprehensive and successful targeted therapy development programs in oncology history^[12-14]. Over the past decade, an extensive body of literature concerning osimertinib in EGFR-mutated NSCLC has accumulated; however, these contributions tend to be circumscribed by thematic focus or temporal scope^[2,12,15-25]. Broad clinical-positioning reviews have typically focused on one or two disease stages, most commonly advanced first-line and T790M-positive later-line settings, without systematically integrating the more recent adjuvant (ADAURA), locally advanced (LAURA), and neoadjuvant (NeoADAURA) evidence into a unified treatment continuum^[26-28]. Resistance-focused reviews, while offering valuable molecular depth, have generally cataloged mechanisms in isolation from clinical decision-making algorithms and have rarely provided a structured, line-dependent comparison of on-target vs. off-target resistance patterns across first-line, second-line, and adjuvant contexts^{[2,12,15,17-24,29]}. Meta-analyses and network meta-analyses have quantified systemic efficacy and, in selected cases, central nervous system (CNS)-specific outcomes but have not contextualized these findings within the rapidly shifting frontline landscape now shaped by combination intensification trials such as FLAURA2 and MARIPOSA^[30-41]. Similarly, guidelines and consensus documents, although clinically pragmatic, have largely predated the convergence of antibody-drug conjugates, EGFR-MET bispecific antibodies, and fourth-generation allosteric EGFR inhibitors into the postosimertinib therapeutic space^[24,42-46]. As a consequence, no single review to date has offered an integrated framework that simultaneously spans the full disease continuum from curative-intent adjuvant therapy through locally advanced consolidation to advanced first-line treatment and mechanism-guided postprogression management. The present review seeks to address this gap by providing, for the first time, a comprehensive and clinically oriented synthesis that unifies six treatment scenarios within a single narrative architecture. It systematically contrasts resistance biology as a function of treatment line and proposes a risk-adapted frontline selection framework informed by the latest biomarker and combination trial evidence. In addition, it maps each identified resistance mechanism to its corresponding therapeutic strategy, with explicit notation of evidence maturity. In doing so, this review also addresses key unresolved questions, including resistance after adjuvant osimertinib, optimal frontline intensity selection, sequencing of novel postosimertinib agents, and the role of minimal residual disease (MRD)-guided adaptive strategies that should inform the design of next-generation clinical trials. By consolidating the most current evidence through early 2026 into a cohesive decision-support framework, this work aims to serve not only as a reference for practicing clinicians navigating an increasingly complex treatment landscape but also as a roadmap for researchers seeking to prioritize the most impactful areas of investigation in EGFR-mutant NSCLC. Osimertinib (AZD9291) is a mono-anilino-pyrimidine compound that functions as an irreversible, covalent inhibitor of EGFR through selective binding to the C797 residue in the adenosine triphosphate (ATP)-binding pocket^[4]. Its molecular design strongly inhibits both EGFR-sensitizing mutations (exon 19 deletion, L858R) and the T790M resistance mutation, while it exhibits approximately 200-fold greater selectivity over wild-type EGFR, a property that underpins its favorable therapeutic index compared with earlier-generation TKIs^[4,27]. Osimertinib is orally administered at a standard dosage of 80 mg once daily, achieving steady-state plasma concentrations within approximately 15 days, with a terminal half-life of approximately 48 h that supports once-daily dosing. Metabolism occurs primarily via CYP3A4 and CYP3A5, generating two pharmacologically active metabolites (AZ5104 and AZ7550) that contribute to both systemic and intracranial activity^[25]. A distinguishing pharmacokinetic feature is the exceptional CNS penetration of osimertinib: preclinical studies have demonstrated a brain-to-plasma area under the concentration–time curve, ratio substantially higher than that of gefitinib, erlotinib, or afatinib, which is attributable to its physicochemical properties, including moderate lipophilicity and low susceptibility to P-glycoprotein-mediated efflux^[11]. This pharmacologic profile, mutation-selective potency, wild-type sparing, and CNS penetration provide the mechanistic foundation for the clinical efficacy data reviewed in subsequent sections.

EVIDENCE FROM CLINICAL TRIALS

Mutation-specific considerations

Common sensitizing mutations: exon 19 deletion vs. L858R. In first-line and ctDNA-selected trials, ex19del is consistently associated with superior outcomes compared with L858R. The Liquid Lung-O Cohort 1 data illustrate the following: an ex19del ORR of 91%, with a median PFS of 21.9 months, vs. an L858R/L861Q ORR of 43%, with a median PFS of 5.1 months^[64]. Similar differential patterns are observed across FLAURA and other datasets, reflecting intrinsic biological differences between these mutation subtypes.

Uncommon EGFR mutations (excluding exon 20 insertions). Three prospective studies reported osimertinib activity in patients with uncommon EGFR mutations. The KCSG-LU15-09 phase II trial (Korean Cancer Study Group; NCT03001505) demonstrated clinically meaningful activity, with an ORR of 50% and a median PFS of 8.2 months^[70], whereas the UNICORN trial extended this evidence to the first-line, untreated setting with meaningful responses^[71,72]. A pooled post hoc analysis by Eide et al. revealed an important mutation-specific gradient: compared with other uncommon variants, G719X-compound mutations markedly improved outcomes, suggesting that mutation subtypes should guide clinical expectations and treatment counseling^[73]. These data collectively support the use of osimertinib for several key uncommon genotypes, particularly G719X, L861Q, and S768I. Supplementary Table 8 summarizes osimertinib in patients with uncommon EGFR mutations.

Exon 20 insertions: the Korean LU17-19 phase II trial evaluated 80 mg osimertinib daily in patients with EGFR exon 20 insertion mutations after chemotherapy failure^[74]. Consistent with preclinical data suggesting structural resistance at standard dosing, the available evidence indicates limited activity. Standard-dose osimertinib is not an optimal therapy for most exon 20 insertions, reinforcing the need for exon 20-specific agents (e.g., amivantamab and mobocertinib) and demonstrating the boundaries of the mutation spectrum of osimertinib. The resistance mechanisms of osimertinib in lung cancer are summarized in Figure 1. Importantly, these resistance mechanisms differ substantially in their current therapeutic actionability and should be interpreted along a translational readiness spectrum rather than as a homogeneous catalog. Among the clinically targetable mechanisms with established or late-phase therapeutic options, MET amplification can be addressed with selective MET inhibitors (savolitinib, tepotinib, and capmatinib) or the EGFR-MET bispecific antibody amivantamab, all of which have generated confirmatory clinical efficacy data in osimertinib-resistant populations; Human epidermal growth factor receptor 2 (HER2) amplification is targeted by trastuzumab deruxtecan (T-DXd), which has demonstrated meaningful activity across HER2-altered NSCLC; oncogene fusions involving anaplastic lymphoma kinase (ALK), rearranged during transfection (RET), ROS proto-oncogene 1 (ROS1), or neurotrophic tyrosine receptor kinase (NTRK) can be managed with the corresponding approved selective TKI (e.g., alectinib/lorlatinib for ALK and selpercatinib for RET), provided that the fusion is identified through comprehensive genomic profiling at progression. Among the mechanisms currently under active clinical investigation but not yet supported by definitive practice-changing evidence, the tertiary C797S mutation in EGFR exon 20, which disrupts the covalent binding site of osimertinib, is being targeted by fourth-generation allosteric EGFR inhibitors (e.g., BLU-945 and BLU-525) in early-phase trials with preliminary signals of activity; PIK3CA (phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha) mutations and PTEN (phosphatase and tensin homolog) loss are being explored as targets for AKT (protein kinase B) inhibitors (e.g., capivasertib) in combination strategies, and RAS/MAPK (rat sarcoma viral oncogene homolog/mitogen-activated protein kinase) pathway activation is under investigation with MEK inhibitor combinations, although clinical data remain limited. Among the mechanisms for which no targeted therapeutic strategy is currently established, histological transformation to small-cell lung cancer (SCLC), The prevalence of SCLC transformation (2%–15%) and its management with platinum–etoposide chemotherapy remain accurately described. The parallel statement on squamous cell transformation correctly reflects the absence of a mechanism-specific therapeutic strategy. The discussion of rare or polyclonal resistance profiles preserves the original meaning regarding the limitations of current single-target approaches. This stratification of resistance mechanisms by therapeutic actionability has direct clinical implications: it underscores the critical importance of comprehensive molecular profiling at the time of osimertinib progression, ideally through both tissue rebiopsy and liquid biopsy, to distinguish patients who can be redirected to established targeted strategies from those for whom clinical trial enrollment or empirical chemotherapy-based approaches remain more appropriate.

Figure 1. Mechanisms of osimertinib resistance. Note: Resistance mechanisms are classified by current therapeutic actionability. Clinically targetable: MET amplification (savolitinib, tepotinib, and amivantamab), HER2 amplification (trastuzumab deruxtecan), and ALK/RET/ROS1/NTRK fusions (corresponding selective TKIs). Under clinical investigation: C797S mutation (fourth-generation allosteric EGFR inhibitors, e.g., BLU-945), PIK3CA/PTEN alterations (AKT inhibitors), and RAS/MAPK activation (MEK inhibitor combinations). No established targeted strategy exists for SCLC: histological transformation (managed with platinum-etoposide chemotherapy), squamous transformation, or polyclonal multipathway resistance.

EFFICACY OF THE CENTRAL NERVOUS SYSTEM

CNS metastases represent a dominant mode of treatment failure and a leading cause of morbidity in patients with EGFR-mutated NSCLC, with brain involvement occurring in approximately one-third of patients during the course of their disease and having a disproportionate effect on functional independence and quality of life^[75]. The limited CNS penetration of first- and second-generation EGFR-TKIs has been a longstanding therapeutic liability, as the blood-brain barrier restricts intracranial drug concentrations to subtherapeutic levels, permitting the CNS to serve as a pharmacologic sanctuary for residual disease even when systemic control is maintained^[11]. Osimertinib was rationally designed to overcome this limitation: preclinical studies demonstrated sustained brain exposure and antitumor activity in multiple EGFR-mutant NSCLC brain metastasis models, with substantially greater CNS penetration than gefitinib, erlotinib, and afatinib did, establishing the pharmacokinetic basis for its subsequent clinical CNS efficacy^[11]. This preclinical advantage has been validated across a remarkably broad spectrum of prospective clinical trials, evolving from early exploratory subanalyses to formal, protocol-specified end points that now constitute a central element of osimertinib’s therapeutic identity.

The first comparative clinical evidence for osimertinib’s intracranial activity emerged from the T790M-positive second-line setting. Preplanned CNS analysis of the AURA3 phase III trial evaluated osimertinib vs. platinum pemetrexed in patients with T790M-positive advanced NSCLC whose baseline brain metastases were confirmed by a blinded central neuroradiologic review^[76]. In this CNS-evaluable population, compared with chemotherapy, osimertinib resulted in significantly higher intracranial objective response rates and prolonged CNS progression-free survival. These findings were particularly impactful because they provided the first randomized evidence that a TKI could achieve not only systemic disease control but also measurable and durable intracranial tumor shrinkage in a head-to-head comparison with platinum-based chemotherapy, a treatment modality with inherently poor CNS penetration. The APOLLO study extended this evidence to a more clinically representative population: prospectively enrolling T790M-positive NSCLC patients with documented CNS metastases, the trial confirmed meaningful systemic and intracranial PFS with 80 mg of osimertinib daily^[77]. Uniquely among osimertinib trials, APOLLO integrated cerebrospinal fluid (CSF) pharmacokinetic measurements and next-generation sequencing analyses, providing direct evidence of osimertinib penetration into the CSF compartment and enabling molecular characterization of intracranial vs. extracranial disease evolution. This translational integration reinforced the mechanistic rationale for osimertinib’s CNS efficacy and offered insights into patterns of intracranial resistance that remain relevant to ongoing drug development efforts. Korean subgroup analysis of the AURA extension and AURA2 trials further confirmed this CNS activity in an East Asian population, with trial protocols explicitly permitting the enrollment of patients with stable or asymptomatic brain metastases, a design feature that enhanced the generalizability of the CNS efficacy signal beyond the narrowly selected populations typical of earlier targeted therapy studies^[69].

The transition of osimertinib to first-line metastatic therapy in the FLAURA trial brought CNS outcomes into even sharper focus. FLAURA enrolled patients with asymptomatic or stable CNS metastases and demonstrated that compared with standard-of-care first-generation EGFR-TKIs (gefitinib or erlotinib), osimertinib significantly reduced the risk of CNS progression^[3,56]. A dedicated CNS subgroup analysis from FLAURA reported superior CNS response rates and CNS PFS with osimertinib vs. the comparator arm in patients with identifiable brain lesions at baseline^[58]. Moreover, among patients without baseline CNS metastases, osimertinib reduced the incidence of new brain lesions during treatment, a preventive benefit that extends the CNS efficacy narrative beyond the treatment of established metastases to prospective CNS protection. This dual capacity, both therapeutic intracranial activity against existing lesions and prophylactic prevention of new CNS disease, distinguishes osimertinib from its predecessors and has become a decisive factor in clinical decision-making, particularly for patients with a high a priori risk of CNS relapse. The ctDNA-selected Liquid Lung-O Cohort 1 trial provides a striking illustration of this principle: Despite an extraordinarily high baseline incidence of CNS metastases (79%), patients with exon 19 deletion who were treated with first-line osimertinib achieved an ORR of 91% and a median PFS of 21.9 months^[64]. That such prolonged disease control was achievable in a cohort with near-universal intracranial involvement testifies to the magnitude of osimertinib’s CNS activity in the first-line setting and underscores its preferential value in patients presenting with brain disease at diagnosis.

The extension of osimertinib into curative-intent settings, adjuvant and postchemoradiation consolidation, has increased CNS disease control from a secondary consideration to a primary rationale for therapy. In the ADAURA trial, CNS DFS was incorporated as a protocol-specified secondary endpoint, reflecting the recognition that CNS relapse is a particularly devastating event in patients who have undergone curative resection^[48]. The results were unequivocal: the CNS DFS hazard ratio was 0.24 in the stage II-IIIA population, indicating a 76% reduction in the risk of CNS recurrence with adjuvant osimertinib compared with placebo. This magnitude of CNS protection is arguably the single most clinically persuasive argument for adjuvant osimertinib in resected EGFR-mutated NSCLC, given that isolated CNS relapse following complete surgical resection often signals the transition from curable to incurable disease and disproportionately compromises neurologic function and patient autonomy. Updated analyses with longer follow-up confirmed the sustained durability of this CNS DFS benefit, further reinforcing its relevance to long-term survival. In the locally advanced, unresectable setting, the LAURA trial similarly demonstrated a profound PFS benefit (39.1 vs. 5.6 months with placebo) following definitive CRT^[55]. While the trial’s abstract-level reporting does not provide a separate CNS PFS analysis, the extreme prolongation of overall PFS, which is more than sevenfold that of the placebo, necessarily incorporates a substantial component of CNS relapse prevention, given the high baseline risk of brain involvement in stage III EGFR-mutated disease and the known propensity for CNS failure following chemoradiation alone.

Taken together, the evidence of the efficacy of the CNS across the clinical development program of osimertinib reveals a coherent and compelling narrative that has itself undergone a significant methodologic evolution. In the earliest trials (AURA extension, AURA3), CNS outcomes were assessed through preplanned but exploratory subgroup analyses. As the clinical significance of intracranial disease control became evident, subsequent trials progressively formalized CNS endpoints: FLAURA incorporated CNS progression as a component of the primary PFS analysis and conducted dedicated CNS subgroup evaluations; ADAURA elevated CNS DFS to a protocol-specified secondary endpoint with structured neuroradiologic review; and the APOLLO study uniquely integrated CSF-based pharmacokinetic and genomic correlates. This methodologic maturation, from retrospective observation to prospective, protocol-mandated CNS assessment, reflects both the growing recognition of CNS relapse as a critical determinant of long-term outcomes in EGFR-mutated NSCLC and the field’s increasing confidence in osimertinib’s ability to address this challenge.

The clinical implications of this CNS evidence base are substantial and multidimensional. For patients who present with brain metastases at any stage of EGFR-mutated NSCLC, the advantages of osimertinib over both chemotherapy and first-generation TKIs in terms of intracranial ORR and CNS PFS suggest that osimertinib is the pharmacologic agent of first choice, with the potential to modify the need for, timing of, or extent of local CNS therapies such as stereotactic radiosurgery or whole-brain radiation. For patients without baseline CNS disease, the demonstrated reduction in new brain lesion development and the dramatic improvement in CNS DFS in the adjuvant setting (HR 0.24) provide a strong pharmacologic rationale for osimertinib as a CNS-protective strategy. With respect to the growing population of long-term osimertinib responders in adjuvant and consolidation settings, the prevention of CNS relapse translates directly into the preservation of neurocognitive function, functional independence, and quality of life outcomes, which are not captured in traditional PFS or OS endpoints but are of paramount importance to patients and are increasingly recognized by regulatory and guideline bodies. Taken together, the CNS activity of osimertinib is not merely an ancillary benefit appended to its systemic efficacy; it is a therapeutically decisive and pharmacologically distinctive property that pervades every clinical setting in which the drug has been evaluated and constitutes one of the strongest arguments for its position as the backbone EGFR-targeted agent across the full NSCLC disease continuum. Table 2 provides key evidence of the CNS efficacy of osimertinib.

COMPREHENSIVE SAFETY ASSESSMENT

Building on the efficacy data presented in Section "EVIDENCE FROM CLINICAL TRIALS", this section provides a consolidated, cross-sectional analysis of the safety profile of osimertinib. To avoid fragmented presentation of safety data across individual trial descriptions, all tolerability and adverse event information has been centralized here to facilitate meaningful cross-trial comparisons.

The safety evaluation of osimertinib benefits from an unusually consistent pharmacologic framework: across virtually all prospective interventional trials reviewed, spanning adjuvant, neoadjuvant, consolidation, first-line, and salvage settings, the drug has been administered at a uniform dosage of 80 mg once daily^{[52,53,66,69,70,73]}. This dosing consistency facilitates meaningful cross-trial safety comparisons and enables a cumulative, setting-stratified characterization of the tolerability profile that is rarely achievable for targeted oncologic agents. Table 3 summarizes the key adverse event profiles of osimertinib across all major clinical settings.

The most frequently reported adverse events are dermatologic and gastrointestinal in nature, reflecting on-target EGFR inhibition in epithelial tissues. In the AURA phase II extension cohort, which provides among the most granular safety data in the pretreated T790M-positive population, diarrhea occurred in 43% and rash in 40% of patients, predominantly at grades 1-2^[67]. The KCSG-LU15-09 trial in uncommon EGFR mutations reported a broadly similar but somewhat attenuated toxicity spectrum: rash 31%, pruritus 25%, decreased appetite 25%, diarrhea 22%, and dyspnea 22%, all at manageable grades^[70]. A defining feature of osimertinib tolerability, which is consistently observed across settings, is the lower incidence and reduced severity of these class-effect toxicities relative to first- and second-generation EGFR-TKIs. In FLAURA, compared with gefitinib/erlotinib, osimertinib resulted in notably fewer high-grade rash and diarrhea events^[3,56], while the AURA3 Japanese subgroup analysis strongly revealed this advantage: grade ≥ 3 treatment-related adverse events occurred in only 12.2% of osimertinib-treated patients vs. 54.5% of those receiving platinum pemetrexed^[66]. This favorable therapeutic index, lower toxicity burden coupled with superior efficacy, was instrumental in the rapid adoption of osimertinib as the preferred EGFR-TKI across treatment lines.

In light of this background of generally mild and predictable toxicity, interstitial lung disease (ILD) and pneumonitis have emerged as the most clinically consequential idiosyncratic risks, warranting particular vigilance. In the AURA phase II extension, ILD events were documented in approximately 4% of patients, including three fatal (grade 5) cases, indicating that even with a highly selective third-generation TKI, pulmonary toxicity remains a life-threatening concern^[67]. In the Liquid Lung-O program, both Cohort 1 (first-line ctDNA-selected) and Cohort 2 (second-line plasma T790M-positive) reported sporadic drug-related interstitial pneumonitis events leading to treatment discontinuation: one patient in Cohort 1^[64] and one grade 3 event among 19 treated patients in Cohort 2 (an overall adverse event rate of 89.5% and a grade 3 to 4 rate of 31.6%)^[68]. Large-scale real-world data provide a complementary perspective: a global ASTRIS study of > 3,000 T790M-positive patients reported ILD/pneumonitis in approximately 1% of the overall cohort^[78], whereas a Korean multicenter real-world study reported treatment discontinuation due to severe adverse events (including pneumonitis) in 1.8% of 558 patients^[79]. The apparent discrepancy between clinical trial ILD rates (approximately 4% in AURA extension^[67]) and real-world estimates (1%-2%) may reflect differences in reporting rigor, surveillance intensity, population selection, and the inherent difficulty of attributing pulmonary events in heavily pretreated patients with underlying lung disease.

The risk of ILD has particular clinical significance in the postchemoradiotherapy consolidation setting. In LAURA, the overlapping pathophysiology of radiation pneumonitis and TKI-related ILD creates a uniquely challenging diagnostic and management environment^[55]. Careful separation of radiotherapy-induced vs. drug-induced pulmonary toxicity is essential for appropriate clinical decision-making, including treatment continuation, dose modification, or discontinuation. Although LAURA’s abstract-level reporting does not provide granular ILD or pneumonitis incidence data, this competing risk is explicitly recognized as central to the benefit risk calculus of post-CRT osimertinib consolidation. Clinicians who prescribe osimertinib in this context must maintain heightened surveillance, particularly during the first months of therapy, when the temporal overlap between radiation-induced and TKI-induced pneumonitis is greatest. East Asian populations deserve additional consideration, as background ILD risk may be elevated in this demographic, a pattern observed across multiple EGFR-TKI classes^[66,67].

Cardiac toxicity represents a second domain of structured monitoring throughout the osimertinib program. Corrected QT interval (QTc) prolongation and a reduction in the left ventricular ejection fraction (LVEF) were prospectively assessed in the ADAURA adjuvant trial and its long-term extensions, with cardiac imaging and electrocardiographic surveillance incorporated into the study protocols^[47,48,50]. Reassuringly, no major new cardiac safety signals have emerged from these prolonged follow-up analyses, even with 3-year continuous dosing in the adjuvant context. Nevertheless, the cumulative cardiac exposure inherent in extended adjuvant therapy mandates ongoing monitoring, particularly in older patients and those with preexisting cardiovascular comorbidities. The absence of a dedicated cardiac-focused analysis across the combined prospective database remains a relative gap, and long-term postmarketing registries will be essential for definitively characterizing the incidence and clinical significance of osimertinib-associated cardiotoxicity over multiple years of exposure.

The safety profile of osimertinib is further modulated by the specific clinical context in which it is deployed. In perioperative settings, short-term neoadjuvant exposure (6 to 8 weeks) in the Blakely or Aredo trial and NEOS yielded high surgical feasibility, and resection was successfully performed in 89% of patients in NCT03433469, with no sign of impaired wound healing or elevated perioperative morbidity^[52-54]. This finding contrasts with the more prolonged perioperative toxicity profiles observed with some chemoimmunotherapy regimens and supports the safety of brief preoperative osimertinib courses. When osimertinib is combined with platinum-pemetrexed chemotherapy, as evaluated in FLAURA2 and OPAL, the predictable additive burden of chemotherapy-related toxicity, principally myelosuppression, nausea, vomiting, and fatigue, is layered onto the baseline profile of the TKI^[60]. The FLAURA2 and OPAL trial designs included safety as a coprimary or key secondary endpoint, and the combination was deemed manageable in both studies^[59-61], although the incremental toxicity cost relative to monotherapy must be weighed against the PFS benefit, particularly in patients for whom quality of life is a primary treatment goal. In special populations historically excluded from pivotal trials, the OPEN/TORG2040 experience in ECOG PS 2 to 4 patients (median age is 75 years) provides critical reassurance that osimertinib is tolerable even in elderly and frail cohorts, with evidence of performance status improvement alongside antitumor activity^[63,71]. This tolerability in functionally compromised patients further distinguishes osimertinib from cytotoxic chemotherapy and, to a lesser extent, from earlier-generation TKIs with higher rates of treatment-limiting dermatologic and gastrointestinal events.

By integrating the totality of safety evidence, several principles emerge for clinical practice. First, the overall risk and benefit profile of osimertinib is favorable and consistent across disease stages, treatment lines, and diverse patient populations, with dermatologic and gastrointestinal events being the most common but overwhelmingly manageable toxicities. Second, ILD or pneumonitis, although infrequent (1% to 4% across studies), is the principal life-threatening adverse event and demands proactive monitoring, prompt diagnostic workup upon symptom onset, and a low threshold for treatment interruption, especially in postradiation settings, East Asian populations, and heavily pretreated or ctDNA-selected programs where disease burden and prior thoracic injury may amplify risk. Third, cardiac monitoring should be incorporated into long-term treatment plans, particularly for adjuvant and consolidation therapies that extend over time. Fourth, the incremental toxicity introduced by chemotherapy combination regimens is real but manageable and must be contextualized against the magnitude of the additional efficacy benefit offered to each individual patient. Collectively, this safety evidence supports the continued expansion of the therapeutic role of osimertinib across the EGFR-mutated NSCLC continuum, with appropriate risk stratification monitoring as an essential component of the treatment paradigm.

EVIDENCE FROM REAL-WORLD STUDIES

While unusually comprehensive, the prospective clinical trial portfolio of osimertinib was necessarily conducted in selected populations meeting strict eligibility criteria, typically patients with good performance status, limited comorbidities, confirmed tissue-based molecular testing, and access to protocol-mandated follow-up. The extent to which these controlled efficacy and safety signals translate into heterogeneous clinical practice settings—where patients are older, frailer, more heavily pretreated, and managed under diverse healthcare systems—can only be addressed by large-scale real-world studies. Over the past several years, a substantial body of observational evidence has accumulated across multiple continents and clinical contexts, providing an essential complement to randomized trial data and progressively filling evidence gaps for populations underrepresented in registration programs^[78-83]. Tables 4 and 5 summarize evidence from real-world studies.

The largest and most geographically diverse real-world dataset comes from the ASTRIS study, a global, noninterventional program that included more than 3,000 patients with T790M-positive advanced NSCLC treated with osimertinib after progression on prior EGFR-TKIs^[78]. In this heterogeneous cohort, encompassing patients from multiple countries, practice settings, and clinical profiles that would not uniformly qualify for randomized trial enrollment, osimertinib achieved an investigator-assessed ORR of 57.1%, a median PFS of 11.1 months, and a median time to treatment discontinuation of 13.5 months. These effectiveness metrics are broadly concordant with the 12.3-month median PFS and high ORR observed in the AURA phase II extension^[67] and the PFS advantage demonstrated in AURA3^[65], providing critical reassurance that the controlled trial findings are reproducible across the diversity of real-world practice. Equally notable was the safety profile: ILD/pneumonitis was documented in approximately 1% of the ASTRIS population^[78], substantially lower than the 4% rate observed in the AURA extension^[67]. While this difference likely reflects, in part, the less intensive adverse event capture inherent in observational designs, the less stringent surveillance protocols, and the potential for underreporting in routine clinical practice, the low real-world ILD incidence nevertheless provides clinicians with a practical benchmark for counseling patients about risk magnitude outside the controlled trial setting.

Regional real-world cohorts from East Asia, where EGFR-mutated NSCLC has the highest prevalence and where osimertinib utilization is most extensive, offer further granularity. A Korean multicenter retrospective study analyzed outcomes in 558 patients treated with osimertinib for T790M-positive disease after prior EGFR-TKI therapy^[79]. This cohort, reflecting the spectrum of real-world clinical complexity, including varied numbers of prior treatment lines and diverse comorbidity profiles, had a median PFS of 14.2 months, an ORR of 55.2%, a median time to treatment discontinuation of 15.0 months, and a median OS of 36.7 months. The PFS of 14.2 months slightly exceeds the 12.3 months observed in the AURA extension and is notably longer than the 8.3 months reported in the plasma-only T790M-selected Liquid Lung-O Cohort 2^[68], likely reflecting the favorable prognostic profile of a tissue-confirmed T790M population in a healthcare system with a well-established molecular testing infrastructure. Discontinuation of severe adverse events, including pneumonitis, occurred in 1.8% of patients, confirming the manageable safety profile in routine Korean clinical practice. A Taiwanese early-access program provided a contrasting perspective: among 419 patients treated with osimertinib for T790M-positive locally advanced or metastatic NSCLC, the outcomes were more modest, with a median PFS of 10.5 months, a median OS of 19.0 months, an ORR of 32.5%, and a disease control rate (DCR) of 86.4%^[81]. Critically, this Taiwanese cohort was substantially more heavily pretreated than either the Korean cohort or the registration trial population, with 53% of patients having received three or more prior lines of therapy. The attenuated ORR (32.5% vs. 55% to 74% in less pretreated cohorts^[69,79]) and shorter OS (19.0 vs. 36.7 months^[79]) illustrate the expected inverse relationship between treatment-line burden and therapeutic efficacy and provide a realistic benchmark for the outcomes achievable when osimertinib is deployed in late-line, heavily pretreated settings that are common in clinical practice but poorly represented in prospective trials.

The transition to first-line real-world osimertinib data marked an important evolution in the evidence base. An Indian single-center study from a tertiary cancer center analyzed 129 consecutive patients with metastatic EGFR-mutated NSCLC who were treated with first-line osimertinib between March 2018 and May 2023^[83]. Despite the resource constraints and patient demographic characteristics of a developing country’s oncologic practice, the outcomes were a striking median PFS of 21.9 months, a median OS of 31 months, and an ORR of approximately 77.5%. These figures closely parallel the FLAURA trial results (PFS HR 0.46 vs. first-generation TKIs; median PFS approximately 18.9 months in the osimertinib arm) and demonstrate that the controlled trial efficacy is achievable in a geographically and socioeconomically distinct clinical setting. The Chinese CAPTRALung multicenter database analysis provided a relatively large-scale and methodologically more rigorous first-line comparison: among 1,556 patients with advanced EGFR-mutated NSCLC (202 receiving osimertinib and 1,354 receiving first-generation TKIs), propensity score matching at a 1:2 ratio was employed to adjust for baseline covariate imbalances^[80]. After matching, compared with first-generation TKIs, osimertinib significantly improved PFS, with a comparable safety profile effectively recapitulating the FLAURA findings in a real-world Chinese population and supporting the generalizability of osimertinib's first-line superiority across the region where EGFR-mutated NSCLC is most prevalent. A Taiwanese retrospective cohort study provided more equivocal results in a different comparative context: among 168 stage IIIb-IV patients treated between April 2020 and April 2023 (55, osimertinib; 113, afatinib/dacomitinib), no statistically significant difference in PFS was observed between osimertinib and second-generation EGFR-TKIs^[82]. However, this study was limited by its single-center design, modest sample size, and relatively short follow-up duration, factors that collectively limit its statistical power to detect the magnitude of PFS difference demonstrated in larger trials and may not adequately capture differences in CNS relapse prevention or OS that have been the more definitive discriminating endpoints in prospective randomized data.

When the totality of real-world evidence is appraised alongside the prospective clinical trial portfolio, several overarching patterns emerge with important implications for clinical practice and future research. First, the effectiveness of osimertinib in routine practice consistently approximates, and in some cohorts closely replicates, the efficacy observed in registration trials, both in the second-line T790M-positive setting (ASTRIS median PFS 11.1 months; Korean cohort 14.2 months; vs. AURA extension 12.3 months) and in the first-line context (Indian cohort PFS 21.9 months; CAPTRALung propensity-matched PFS advantage; vs. FLAURA). This consistency across geographies, across India, China, Korea, Taiwan, and global multicountry programs and healthcare systems provides robust external validation of the randomized evidence. Second, outcomes are meaningfully influenced by treatment-line burden and patient selection: heavily pretreated populations (Taiwanese EAP: 53% ≥ 3 prior lines; PFS 10.5 months, ORR 32.5%) show attenuated benefit compared with earlier-line cohorts (Korean: PFS 14.2 months, ORR 55.2%), underscoring the importance of optimizing the sequencing and timing of osimertinib deployment. Third, the safety profile in real-world practice is consistently reassuring, with ILD or pneumonitis rates of approximately 1% to 2% (ASTRIS 1%; Korean severe adeverse event (AE) discontinuation 1.8%), lower than the 4% observed in the more intensively monitored AURA extension, and no emergence of novel safety signals despite broader and less selected patient populations. Fourth, these real-world cohorts progressively fill evidence gaps for subgroups underrepresented or excluded from pivotal trials: patients with multiple comorbidities, advanced age, complex prior treatment histories, diverse ethnic and socioeconomic backgrounds, and variable access to molecular diagnostics and follow-up infrastructure. The growing contribution of data from low- and middle-income countries is particularly valuable, as it addresses the global equity dimension of targeted therapy access and effectiveness.

Certain limitations of the real-world evidence base must, however, be acknowledged. Observational studies are inherently subject to selection bias, immortal time bias, and confounding by indication even when propensity score methods are employed. Investigator-assessed endpoints in the absence of blinded independent central review may introduce measurement bias that favors either overestimation or underestimation of treatment effects depending on the context. Adverse events captured in routine practice are typically less systematic than those captured in clinical trials, potentially underestimating the true incidence of less common but clinically significant toxicities such as ILD and cardiac events. The retrospective, single-center design of several key studies limits generalizability and statistical power. Ongoing large-scale prospective registries and pragmatic clinical trials will be essential for further refining the real-world evidence base, particularly for long-term outcomes after multiyear adjuvant or consolidation therapy, comparative effectiveness against emerging combination regimens, and safety surveillance in populations with extended osimertinib exposure. Taken together, the real-world evidence for osimertinib provides a coherent and reassuring message: the transformative efficacy demonstrated in prospective randomized trials is translatable to the complexity and diversity of routine clinical practice across global healthcare settings. This body of evidence reinforces the position of osimertinib as a central EGFR-targeted agent for real-world patients with EGFR-mutated NSCLC while simultaneously highlighting the importance of treatment-line optimization, population-specific safety monitoring, and continued data generation from underrepresented geographies and clinical subgroups. The treatment continuum according to the Sankey diagram is shown in Figure 2. Notably, Figure 2 represents a conceptual synthesis of the treatment continuum informed by the aggregate evidence base rather than a direct patient-flow analysis derived from a single dataset.

Figure 2. Conceptual treatment continuum of osimertinib across disease stages (Sankey diagram). Note: This diagram represents a synthesized, evidence-informed conceptual framework illustrating the treatment flow across clinical scenarios rather than a strictly data-derived patient-level analysis. Treatment transitions are based on current guideline recommendations and pivotal trial evidence (ADAURA, LAURA, FLAURA, FLAURA2, AURA3). The bifurcation between osimertinib monotherapy and osimertinib plus chemotherapy in the first-line advanced setting reflects the risk-stratified approach informed by FLAURA2 data, where high-risk features (e.g., TP53 comutation, high tumor burden, and L858R mutation) may favor combination therapy. Dashed lines indicate emerging or investigational pathways not yet established as the standard of care.

CHRONOLOGIC EVOLUTION

The clinical development of osimertinib traces a remarkably coherent and strategically deliberate trajectory that, over approximately one decade, has expanded the drug's therapeutic role from a narrowly indicated salvage agent to a universal backbone therapy spanning the entire EGFR-mutated NSCLC disease continuum. This evolution can be understood through three distinct strategic phases, each driven by a unifying clinical question that naturally arose from the preceding evidence.

Phase I (2013-2017): Establishing proof-of-concept in T790M-mediated resistance. The earliest phase was defined by the imperative to address T790M-mediated acquired resistance, the most common mechanism of failure of first- and second-generation EGFR-TKIs^[84]. The AURA phase I/II expansion and subsequent AURA3 phase III trial established randomized superiority over platinum-pemetrexed chemotherapy, leading to global adoption as the standard second-line therapy for T790M-positive progression^[65,67]. CNS-focused analyses of AURA3 revealed a pharmacologically distinctive intracranial efficacy that is central to the therapeutic efficacy of osimertinib^[66,76]. During this same period, the LiquidLung-O Cohort 2 trial (Liquid biopsy in Lung cancer for Osimertinib and WJOG 8815L trials (West Japan Oncology Group study 8815L; NCT02771314) marked a parallel methodological evolution from tissue-based to plasma-based T790M detection^[68,85], whereas the APOLLO study uniquely integrated CSF pharmacokinetic measurements, reframing intracranial activity from a secondary observation to a primary therapeutic attribute^[77].

Phase II (2017-2020): Moving to first-line and optimizing deployment. With T790M-directed efficacy established, the logical question became whether earlier deployment could yield greater benefit. The FLAURA trial answered definitively, demonstrating both PFS and OS superiority over first-generation TKIs as first-line therapy and effecting a paradigm shift from salvage to universal frontline use^[3,56]. Subsequent studies refined this deployment: the LiquidLung-O Cohort 1 trial revealed striking mutation-specific heterogeneity of benefit that challenged the assumption of uniform efficacy across EGFR mutation subtypes^[57,64], and the OPEN/TORG2040 trial extended the evidence to the previously neglected poor-performance-status population^[62,63]. Real-world effectiveness data from multiple countries progressively confirmed the generalizability of these findings across diverse healthcare systems (Section “EVIDENCE FROM REAL-WORLD STUDIES”).

Phase III (2020-present): Extending to curative-intent settings and combination strategies. The most transformative phase extended osimertinib into curative-intent contexts, fundamentally redefining EGFR-mutated NSCLC from a metastatic-only treatment target to a molecular disease requiring targeted intervention across the entire disease trajectory. The ADAURA trial provided the first evidence that targeted adjuvant therapy is disease-modifying and capable of improving overall survival in patients with resected EGFR-mutated NSCLC^[47,50], whereas the LAURA trial established targeted consolidation after chemoradiotherapy in patients with unresectable stage III disease^[55]. Neoadjuvant programs (NCT03433469, NEOS) have demonstrated the feasibility of preoperative osimertinib, although the incremental benefit over adjuvant-only strategies remains an open question to be addressed by the ongoing NeoADAURA phase III trial^[52-54,86]. Paralleling this stage-based expansion, the mutation spectrum broadened through the KCSG-LU15-09 and UNICORN trials for uncommon mutations^[70], and combination intensification strategies matured, with FLAURA2 establishing the foundation for risk-stratified frontline treatment selection^[60]. This chronological evolution is illustrated in Figure 3, which maps the key milestone trials to their respective strategic phases.

Figure 3. Timeline of Milestone trials of osimertinib in lung cancer.

RESISTANCE MECHANISMS AND THE POST-OSIMERTINIB LANDSCAPE

Despite the transformative efficacy of osimertinib across clinical settings, acquired resistance remains an inevitable clinical reality that ultimately limits the duration of benefit for virtually all patients, and a growing body of narrative reviews has comprehensively mapped the complex, heterogeneous landscape of resistance biology that underpins this challenge. Multiple dedicated reviews, most notably those by Lazzari et al. (2020)^[20], He et al. (2021)^[45], and Nicoś et al. (2025)^[87], have classified resistance mechanisms into three broad categories: (1) on-target EGFR-dependent alterations, principally the tertiary C797S mutation in exon 20, which disrupts the covalent binding site of osimertinib and accounts for 0%-29% of resistance cases depending on the series; (2) off-target bypass pathway activation, including MET amplification (7%-24%), HER2 amplification, oncogenic fusions involving RET, ALK, FGFR, and other partners (1%-10%), and downstream signaling alterations in PIK3CA and RAS/MAPK pathways; and (3) histological transformation, most notably the phenotypic switch from NSCLC to SCLC, which occurs in 2%-15% of cases and results in a fundamentally altered treatment paradigm. A critically important insight emerging from these syntheses is that resistance patterns differ substantially depending on the line of therapy in which osimertinib is administered: first-line resistance tends to exhibit greater molecular heterogeneity with fewer on-target EGFR mutations and a greater proportion of bypass and transformation events, whereas second-line resistance after prior first/second-generation TKIs more frequently involves EGFR-dependent mechanisms, including C797S, a distinction with direct implications for resistance monitoring strategies and subsequent treatment selection. The heterogeneity and clinical unpredictability of bypass-mediated resistance are illustrated by the case reported by Yamaguchi et al. (2024), in which a patient with EGFR L858R-positive NSCLC developed acquired EML4-ALK variant 3a/b rearrangement after 15 months of first-line osimertinib, notably the first reported instance of ALK fusion arising under first-line osimertinib selective pressure independent of T790M, with the synchronous coexistence of two canonically mutually exclusive oncogenic drivers in the same tumor^[88]. The limited efficacy of subsequent brigatinib monotherapy in this dual-driver context suggests that single-agent targeting of only one pathway is insufficient for cross-driver resistance scenarios, underscoring the need for combination strategies and dedicated clinical trials in this molecularly defined subpopulation. This case serves as a reminder that resistance can involve unexpected oncogenic driver switching beyond the reach of population-level profiling, reinforcing the imperative for comprehensive genomic assessment, incorporating both tissue rebiopsy and liquid biopsy at the time of disease progression. Recognizing the central role of real-time molecular characterization in navigating this complexity, Ntzifa et al. (2024)^[89] provided a dedicated review of liquid biopsy strategies during osimertinib treatment, establishing a framework for longitudinal ctDNA and circulating tumor cell (CTC) monitoring that enables noninvasive detection of emergent resistance mechanisms, tracking of clonal evolution under therapeutic pressure, and timely adaptation of treatment, which is applicable across first-line, adjuvant, neoadjuvant, and unresectable stage III contexts where osimertinib is now employed. The therapeutic translation of resistance biology into actionable postosimertinib treatment pathways remains, however, at a relatively early stage of evidence maturity: Zhang et al. (2025)^[90] conducted the first network meta-analysis of subsequent-line regimens after EGFR-TKI resistance, comparing the efficacy and safety of distinct postprogression strategies, including chemotherapy, immunotherapy combinations, and targeted approaches. Notwithstanding this, the heterogeneity of the included populations, the diversity of resistance mechanisms within trial cohorts, and the absence of resistance-mechanism-stratified randomization in most contributing studies limit the precision with which these network meta-analysis (NMA) estimates can be applied to individual patients with molecularly defined resistance profiles. Nicoś et al. (2025)^[87] further surveyed the emerging therapeutic armamentarium designed to overcome osimertinib resistance, including fourth-generation EGFR-TKIs engineered to circumvent C797S (such as BLU-945 and BLU-525), bispecific antibodies (amivantamab), antibody‒drug conjugates (patritumab deruxtecan and datopotamab deruxtecan), MET-targeted combinations, and other mechanism-informed strategies currently under clinical investigation. However, none of these approaches has yet achieved the level of evidence supporting a definitive standard-of-care recommendation. Several of these strategies have nonetheless progressed beyond preclinical rationale to generate early-phase clinical evidence that warrants further discussion. Among fourth-generation EGFR TKIs, BLU-945, an allosteric inhibitor designed to overcome C797S while maintaining activity against the original sensitizing mutation, has demonstrated preliminary antitumor activity, including confirmed partial responses in C797S-positive patients in the phase I/II SYMPHONY trial^[87]. Among bispecific antibodies, the MARIPOSA-2 trial demonstrated that amivantamab plus lazertinib plus chemotherapy significantly improved PFS over chemotherapy alone in patients who progressed on osimertinib (median PFS 8.3 vs. 4.2 months; HR 0.44), providing the first randomized evidence for a postosimertinib combination regimen^[91]. Among antibody-drug conjugates, patritumab deruxtecan (HER3-DXd) achieved a confirmed ORR of 29.8% (95%CI: 23.9-36.2), a median duration of response of 6.4 months, and a median PFS of 5.5 months in heavily pretreated EGFR-mutant NSCLC patients in the HERTHENA-Lung01 phase II trial, with efficacy observed across diverse mechanisms of EGFR-TKI resistance and a confirmed CNS ORR of 33.3% in patients with nonirradiated brain metastases^[92], while datopotamab deruxtecan (Dato-DXd, targeting trophoblast cell-surface antigen 2 (TROP2)) is under evaluation in the TROPION-Lung05 and TROPION-Lung15 trials with preliminary signals of activity^[42,93]. Among MET-targeted combinations, savolitinib combined with osimertinib achieved a confirmed ORR of up to 62% in MET-amplified osimertinib-resistant patients in the TATTON trial^[94], and tepotinib plus osimertinib demonstrated similar activity in the INSIGHT 2 trial^[95]. These data, while predominantly from single-arm phase I/II studies, collectively support a mechanism-informed, biomarker-guided approach to postosimertinib treatment selection and underscore the critical importance of comprehensive genomic profiling at the time of resistance to match each patient’s molecular profile to the most appropriate emerging strategy. Collectively, these resistance-focused reviews provide the essential conceptual scaffolding for interpreting survival endpoints across trials and treatment eras: as the proportion of patients receiving first-line osimertinib increases and the postosimertinib treatment landscape evolves, the nature and frequency of resistance mechanisms encountered in clinical practice will shift, cross-trial comparability of overall survival will become increasingly contingent on postprogression therapies received, and the design of future evidence syntheses, whether meta-analyses or network meta-analyses, will need to explicitly account for resistance biology, liquid biopsy-guided treatment adaptation, and the sequential therapeutic context in which osimertinib is deployed^[20,45].

DISCUSSION

The evidence synthesized in this review establishes osimertinib as the most extensively validated targeted agent in the history of thoracic oncology, with a prospective clinical trial portfolio spanning the full disease continuum of EGFR-mutated NSCLC, from resectable early-stage and unresectable locally advanced to advanced metastatic disease. This portfolio also encompasses multiple molecular subsets, diverse patient populations, and both monotherapy and combination approaches.

CONCLUSIONS

Osimertinib has been established as a central EGFR-targeted agent across the full spectrum of early-stage, locally advanced, unresectable and metastatic EGFR-mutated NSCLC. Its clinical development program, encompassing multiple definitive phase III RCTs and complementary phase II studies across diverse settings, populations, and molecular subsets, represents one of the most comprehensive targeted therapy evidence bases in oncology. The consistent demonstration of efficacy (PFS/DFS and OS benefits), clinically significant CNS disease control, and a favorable and predictable safety profile collectively support the role of osimertinib as the backbone of EGFR-directed therapy. Ongoing and future studies addressing combination strategies, mutation-granular evidence, perioperative optimization, resistance management, and real-world implementation will further refine its use and maximally realize the promise of precision oncology for patients with EGFR-mutated lung cancer.