Deferring or delivering stereotactic radiosurgery in EGFR-mutated non-small cell lung cancer: lessons from the STARLET randomized analysis

The management of brain metastases in patients with epidermal growth factor receptor (EGFR)-mutated non-small cell lung cancer (NSCLC) has undergone a fundamental shift. Historically, intracranial disease prompted immediate local therapy with stereotactic radiosurgery (SRS), whole-brain radiation therapy, or surgery. The development of central nervous system (CNS)-penetrant tyrosine kinase inhibitors (TKIs), particularly osimertinib, introduced a competing strategy in which systemic therapy is initiated first, and radiation, typically in the form of SRS, is reserved for intracranial progression. Although clinical guidelines permit osimertinib with deferred local therapy for selected patients with asymptomatic brain metastases, this recommendation has rested largely on subgroup analyses, retrospective studies, and expert consensus rather than randomized evidence.

The STARLET collaboration therefore addresses a clinically important and persistent question (1). Does upfront SRS improve outcomes when added to osimertinib in patients with untreated brain metastases from EGFR-mutated NSCLC? Lee and colleagues prospectively pooled individual participant data from the randomized phase II OUTRUN and LUOSICNS trials, comparing SRS followed by osimertinib with osimertinib alone. The investigators should be commended for completing two multinational randomized studies in a highly selected molecular and clinical population for whom equipoise, referral patterns, and rapidly changing standards of care make prospective enrollment particularly difficult.

A negative primary endpoint that meaningfully informs practice

STARLET included 79 patients with up to 10 brain metastases amenable to SRS. Most participants had an Eastern Cooperative Oncology Group (ECOG) performance status of 0 or 1, minimally symptomatic intracranial disease, and relatively limited tumor burden. At a median intracranial follow-up of 39 months, the 12-month intracranial progression-free survival rate was 68% with SRS plus osimertinib and 57% with osimertinib alone. Median intracranial progression-free survival was 21.9 and 17.2 months, respectively, with a hazard ratio of 0.80 and a wide 95% confidence interval (0.46–1.37). Although these differences did not reach statistical significance, the numerical improvements are directionally supportive of a potential benefit from upfront SRS and should not be interpreted as evidence that SRS lacks value.

These results should not be interpreted as establishing equivalence between the two strategies. STARLET was initially designed as two larger trials but was modified because of slow accrual and the coronavirus disease 2019 pandemic. The resulting pooled sample of 79 patients remained insufficiently powered to detect more modest but clinically meaningful differences in treatment efficacy. Consequently, a clinically meaningful benefit from upfront SRS cannot be excluded. Even so, the absence of a clear improvement in the prespecified primary endpoint provides the first randomized evidence supporting a strategy already commonly used in practice: initiating osimertinib without SRS and closely monitoring low-volume, asymptomatic or minimally symptomatic brain metastases.

This conclusion is strengthened by several features of the study. Intracranial surveillance was performed using serial magnetic resonance imaging (MRI) rather than symptom-triggered imaging. Progression was assessed using the Response Assessment in Neuro-Oncology Brain Metastases (RANO-BM) criteria. Patients had not received prior cranial radiotherapy, and the randomized design reduced the treatment-selection bias that has complicated retrospective comparisons. Another strength is the mature follow-up (median, 39 months), which provides comprehensive long-term outcome data and increases confidence in the durability of intracranial control and survival benefits. The consistency of the treatment effect across OUTRUN and LUOSICNS also supports the validity of the pooled analysis despite differences in eligibility criteria, randomization procedures, imaging schedules, and SRS dose-fractionation approaches.

At the same time, the study population defines the boundaries of its conclusion. STARLET does not establish that SRS may be safely deferred in patients with substantial neurologic symptoms, lesions causing clinically important mass effect, leptomeningeal disease, large metastases threatening eloquent structures, or unreliable access to frequent MRI surveillance. Patients with such features were absent or underrepresented. Rather than eliminating SRS from the treatment pathway, STARLET helps clarify when immediate SRS may not be routinely necessary.

One limitation of the study was the lack of stratification by EGFR mutation status. Emerging data from the IGnITE study suggest that patients with L858R and atypical EGFR mutations have inferior intracranial outcomes with osimertinib monotherapy compared with those harboring exon 19 deletions and may preferentially benefit from the addition of upfront SRS. Without mutation-specific analyses, the applicability of the findings across biologically distinct EGFR subgroups remains uncertain.

Intracranial progression-free survival may not capture the entire value of SRS

The choice of intracranial progression-free survival as the primary endpoint was reasonable, but its interpretation requires consideration of what SRS is intended to accomplish. SRS is a lesion-directed treatment that provides durable control of treated metastases but does not prevent the development of new metastases elsewhere in the brain. Osimertinib, in contrast, acts throughout the CNS and extracranial compartments. A composite endpoint that includes local progression, distant intracranial progression, and death may therefore dilute a treatment effect that is primarily confined to local control.

Interestingly, STARLET did not demonstrate a clear reduction in local progression with upfront SRS. The 12-month cumulative incidence of local progression was 13% with SRS plus osimertinib and 10% with osimertinib alone. Several explanations are possible. Osimertinib may provide sufficient early control of small EGFR-mutated metastases to minimize the incremental local benefit of SRS within the first year. Classification of enlarging treated lesions may also be challenging because treatment-related change and, rarely, radionecrosis can resemble tumor progression. An effort is underway by RANO to account for transient post-SRS enlargement of brain metastasis, which may not be tumor progression. Furthermore, the relatively small number of events produces unstable estimates and limits meaningful comparisons of specific progression patterns.

Longer-term endpoints may be particularly relevant in EGFR-mutant NSCLC because patients can survive for several years after a brain metastasis diagnosis (2,3). SRS may prevent late regrowth of lesions that initially respond to systemic therapy but are not eradicated. Conversely, treating every visible lesion upfront may expose some patients to radiation who would otherwise experience prolonged or complete intracranial responses to osimertinib. Future studies should therefore assess not only conventional intracranial progression-free survival, but also time to neurologic deterioration, freedom from salvage treatment, cumulative intracranial tumor burden, preservation of neurocognition, corticosteroid exposure, intracranial symptom-free survival, and patient-reported quality of life.

The pending patient-reported and neurocognitive analyses from STARLET will be essential. A strategy that produces similar survival but reduces treatment burden, anxiety, toxicity, and disruption of systemic therapy may be preferred. Alternatively, early SRS could be valuable if it lowers the risk of symptomatic progression or urgent hospitalization even without a statistically significant difference in conventional progression-free survival.

The overall survival signal should generate questions, not conclusions

Median overall survival was numerically longer with SRS plus osimertinib than with osimertinib alone, at 46.1 vs. 29.1 months. The hazard ratio was 0.66, but its 95% confidence interval crossed unity, and the difference was not statistically significant. As a secondary endpoint in a small, open-label phase II dataset, this finding should not be considered evidence of a survival advantage. However, the magnitude and direction of the observed effect are hypothesis-generating and support further investigation into whether improved early local control may translate into longer overall survival for selected patients.

Among patients who developed intracranial progression, 57% in the osimertinib-alone group received salvage SRS, whereas almost one-third died without receiving SRS or whole-brain radiation therapy. This observation illustrates an important limitation of a deferred-SRS strategy. Deferral is not equivalent to omission, and its safety depends on surveillance, timely recognition of progression, continued performance status, and access to salvage treatment. Some patients may deteriorate between surveillance examinations, develop multifocal progression no longer readily amenable to SRS, or experience simultaneous extracranial progression that complicates subsequent treatment.

These competing possibilities cannot be resolved from the present dataset. The survival difference may reflect chance, residual imbalances, subsequent systemic therapy, or differences in salvage treatment. It may also indicate that preventing the progression of selected intracranial lesions has consequences that are not fully captured by the primary endpoint. Reporting causes of death, neurologic death, symptomatic intracranial progression, salvage-treatment eligibility, post-progression therapy, and longitudinal performance status would help clarify this issue in future trials.

Tumor burden may be more informative than lesion count alone

Perhaps the most clinically provocative STARLET finding is the post hoc association between baseline intracranial tumor burden and treatment effect. Among patients with measurable disease, SRS appeared more favorable as the sum of lesion diameters increased. Using a data-derived threshold of 19 mm, the hazard ratio for intracranial progression or death was 0.27 in the higher-burden subgroup, compared with 0.97 in patients below that threshold.

This analysis is exploratory and vulnerable to small sample size, missing non-measurable lesions, multiple testing, and overfitting from selection of an optimal cut point. It should not be used as a definitive treatment threshold. However, the biological and clinical premise is plausible. Larger metastases may contain more heterogeneous cellular populations, have less uniform drug exposure, require longer to regress, and carry a greater risk of causing symptoms if progression occurs. Retrospective TURBO-NSCLC data similarly suggested that the benefit of adding upfront SRS was concentrated among patients with lesions at least 1 cm in size, although that study included both EGFR- and anaplastic lymphoma kinase (ALK)-driven tumors and was subject to treatment selection bias.

These findings suggest that total intracranial disease burden, lesion diameter, anatomic location, edema, growth kinetics, and anticipated neurologic consequences may be more useful than a simple count of metastases. A patient with eight punctate lesions may be an excellent candidate for osimertinib-first management, whereas a patient with one 1.2-cm lesion near the motor cortex may benefit from immediate local therapy despite having fewer metastases. Cumulative intracranial tumor volume has been shown to be an important predictor of outcome in patients with brain metastases treated with SRS (4). Treatment selection should therefore remain risk-adapted rather than based solely on whether brain metastases are technically asymptomatic.

Toxicity remains part of the therapeutic equation

Upfront SRS was generally well tolerated, but radionecrosis was not negligible. Any-grade radionecrosis occurred in 21% of patients in the combination group, and grade 3 or 4 radionecrosis occurred in 5%. Even when uncommon, clinically significant radionecrosis may lead to corticosteroid dependence, seizures, neurologic deficits, surgery, bevacizumab or laser interstitial thermal therapy treatment, repeated imaging, and difficulty distinguishing treatment effect from tumor progression (5). SRS dose reduction using concurrent precision medicine approaches may reduce the risk of adverse radiation effects while still achieving a high degree of local tumor control (6).

These risks must be weighed against the toxicity of intracranial progression and the possibility that delayed SRS will be required in a substantial proportion of patients. The relevant comparison is therefore not “radiation versus no radiation”, but immediate treatment of all visible lesions versus selective treatment of lesions that persist or progress. In STARLET, only a subset of patients assigned to osimertinib alone ultimately underwent SRS, suggesting that initial systemic therapy spared many patients from cranial irradiation during the study period.

An adaptive strategy may ultimately provide a better balance (Figure 1). Osimertinib could be initiated with an early MRI at approximately 6–8 weeks, followed by SRS for enlarging lesions, anatomically threatening lesions, or residual metastases that fail to demonstrate a meaningful response. Such an approach would use early treatment response as a biomarker while avoiding unnecessary SRS in patients with rapid and complete intracranial regression. Prospective evaluation is needed because delayed treatment of a nonresponding lesion could still result in preventable neurologic morbidity.

Figure 1 Proposed risk-adapted framework for integrating SRS with osimertinib in patients with untreated brain metastases from EGFR-mutated NSCLC. Management should be individualized based on neurologic symptoms, intracranial disease burden, lesion characteristics, and feasibility of close MRI surveillance. *, total intracranial tumor volume or sum of measurable lesion diameters may be more informative than lesion count alone. EGFR, epidermal growth factor receptor; MRI, magnetic resonance imaging; NSCLC, non-small cell lung cancer; PFS, progression-free survival; SRS, stereotactic radiosurgery.

Applying STARLET in an evolving systemic-treatment landscape

The systemic treatment of EGFR-mutated NSCLC is already moving beyond osimertinib monotherapy. Osimertinib plus platinum-pemetrexed chemotherapy and amivantamab plus lazertinib have entered the first-line landscape, raising new questions about their intracranial efficacy and interactions with local treatment. The results of STARLET cannot be automatically extrapolated to these combinations, which may alter the depth and durability of CNS control, as well as toxicity. The treatment decision may also be influenced by EGFR mutation subtype, concurrent genomic alterations, extracranial disease burden, prior TKI exposure, and mechanisms of resistance (7).

Future randomized trials will require broader international collaboration and pragmatic designs to overcome the accrual difficulties experienced by OUTRUN and LUOSICNS. Rather than repeating a simple comparison of SRS for all lesions versus no initial SRS, studies should enrich for patients in whom the decision is most uncertain. Candidate populations include those with larger lesions, higher volumetric burden, incomplete early response to systemic therapy, lesions in high-risk locations, or limited access to surveillance. Central imaging review, volumetric response assessment, standardized SRS techniques, neurologic-death endpoints, and longitudinal patient-reported outcomes should be incorporated.

Conclusions

STARLET does not show that SRS lacks value in EGFR-mutated NSCLC. Instead, it demonstrates that routine upfront SRS cannot be assumed to improve 12-month intracranial progression-free survival when added to osimertinib in patients with asymptomatic or minimally symptomatic, relatively low-burden brain metastases. For such patients, osimertinib with rigorous MRI surveillance is now supported not only by systemic therapy subgroup analyses and guidelines but also by prospective randomized evidence.

The central clinical lesson is conditional. Deferring SRS appears reasonable when intracranial disease is small, neurologically low risk, responsive to osimertinib, and amenable to close follow-up. Immediate local therapy remains appropriate for symptomatic disease, substantial mass effect, high-risk anatomy, greater intracranial burden, uncertain adherence to surveillance, or circumstances in which progression could rapidly compromise neurologic function. STARLET therefore moves the field away from a uniform treatment sequence and toward a more individualized model in which the timing of SRS is determined by intracranial risk, treatment response, and patient priorities.

Acknowledgments

None.

Footnote

Provenance and Peer Review: This article was commissioned by the editorial office, Translational Lung Cancer Research. The article has undergone external peer review.

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0940/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2026-0940/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Lee CK, Lefresne S, Soon YY, et al. Osimertinib and Stereotactic Radiosurgery for Brain Metastases in EGFR-Mutated Lung Cancer: The STARLET Joint Analysis of OUTRUN and LUOSICNS Randomized Trials. J Thorac Oncol 2026;21:103549. [Crossref] [PubMed]
2. Wandrey NE, Gao D, Robin TP, et al. Multicenter analysis of stereotactic radiosurgery for multiple brain metastases from EGFR and ALK driven non-small cell lung cancer. Lung Cancer 2023;176:144-8. [Crossref] [PubMed]
3. Dutta SW, Mack ML, Aliotta E, et al. Intracranial disease control for EGFR-mutant and ALK-rearranged lung cancer with large volume or symptomatic brain metastases. J Neurooncol 2020;149:357-66. [Crossref] [PubMed]
4. Hirshman BR, Wilson B, Ali MA, et al. Superior Prognostic Value of Cumulative Intracranial Tumor Volume Relative to Largest Intracranial Tumor Volume for Stereotactic Radiosurgery-Treated Brain Metastasis Patients. Neurosurgery 2018;82:473-80. [Crossref] [PubMed]
5. Vellayappan B, Lim-Fat MJ, Kotecha R, et al. A Systematic Review Informing the Management of Symptomatic Brain Radiation Necrosis After Stereotactic Radiosurgery and International Stereotactic Radiosurgery Society Recommendations. Int J Radiat Oncol Biol Phys 2024;118:14-28. [Crossref] [PubMed]
6. Chung JH, Tos SM, Mantziaris G, et al. Stereotactic Radiosurgery Dose Reduction for Patients With Brain Metastases From Non-Small Cell Lung Primary on Immunotherapy or Targeted Therapy. Neurosurgery 2026; [Epub ahead of print]. [Crossref] [PubMed]
7. Miao E, Wo E, Boe L, et al. Impact of Genomic Subtype on Intracranial Outcomes in Treatment-naive EGFR mutant NSCLC with Osimertinib (IGnITE). J Thorac Oncol 2026;104114. [Epub ahead of print]. [Crossref] [PubMed]