Tyrosine kinase inhibitors (TKIs) have transformed outcomes for patients with oncogene-addicted advanced non-small cell lung cancer (NSCLC), establishing first-line targeted therapy as the standard of care for EGFR-positive disease.1 Nevertheless, uninterrupted TKI administration carries cumulative burdens: Financial toxicity, impaired quality of life, and the selection of resistant subclones. As the population of long-term survivors in this setting continues to grow, the clinical question of whether, and under what molecular conditions, targeted therapy can be safely and temporarily discontinued has gained increasing relevance.

A study by Dong and colleagues, published in JAMA Oncology in 2024, takes a meaningful step toward answering this question.2 They enrolled 60 patients with advanced oncogene-addicted NSCLC (93% EGFR-positive) who had achieved complete radiological clearance after TKI therapy combined with local consolidative therapy (LCT), predominantly surgery. Crucially, eligibility for treatment discontinuation required not only the absence of measurable disease per RECIST 1.1, but also undetectable circulating tumor DNA (ctDNA) by a 338-gene NGS panel and a normal carcinoembryonic antigen, a recognition that imaging alone is insufficient to detect minimal residual disease (MRD) in solid tumors.3

Study Results

Outcomes were stratified by the trigger for retreatment. Group A (23% of patients) remained treatment-free throughout the follow-up period, with a median treatment break exceeding 20 months. Group B (52%) received retreatment based on ctDNA or carcinoembryonic antigen positivity before imaging-confirmed progression, achieving a median progression-free survival (PFS) of 20.2 months and a 96% retreatment response rate.

Group C (25%) reached imaging-defined progression and fared substantially worse, with a median PFS of only 5.5 months. Notably, nine of the 15 Group C patients had undetectable ctDNA at progression, six of whom had isolated brain metastases, highlighting the limitations of blood-based ctDNA in detecting sanctuary-site disease.4 The 96% retreatment response rate across groups addresses a longstanding concern—that a treatment break might compromise subsequent TKI efficacy or accelerate resistance. These data suggest it does not, at least in this carefully selected population.

Study Considerations and Limitations

The study by Dong et al. represents a novel proof of concept for ctDNA-guided adaptive de-escalation in advanced solid tumors. It also raises a terminological point worth noting: In patients who retain radiological clearance in the metastatic setting, the term Molecular Residual Disease (MoRD) may more precisely characterize their status than MRD, which carries established meaning in hematological oncology and curative-intent solid tumor settings.5

Several limitations temper immediate clinical enthusiasm. As the authors themselves acknowledge, the single-center, non-randomized design with no concurrent control arm and only 60 patients makes any definitive claims about efficacy premature. The 19-month median follow-up is insufficient to draw conclusions on overall survival, and without a comparator arm receiving uninterrupted TKI, it is impossible to determine whether the observed outcomes reflect non-inferiority to the standard treatment plan. Moreover, the data from the FLAURA2 and MARIPOSA-2 trials raise concerns about the new standard treatment in this population.

Three additional methodological gaps deserve attention. First, the broad-panel NGS assay (oncoMRD-B, 338 genes) is susceptible to clonal hematopoiesis of indeterminate potential (CHIP), which can generate spurious ctDNA-positive signals. Second, ctDNA was employed as a clinical decision tool rather than as a pre-specified regulatory endpoint, a distinction with direct implications for drug development, as the November 2024 FDA Guidance on ctDNA requires prospective validation correlating ctDNA dynamics with long-term survival outcomes before ctDNA clearance can support regulatory approval. Lastly, the absence of pre-TKI baseline ctDNA data precludes the retrospective identification of non-shedding tumors. Published data show preoperative ctDNA detection rates as low as 13% to 24% with standard assays in early-stage NSCLC,6,7 and while the metastatic setting yields higher rates, non-shedder biology remains a relevant source of false-negative MRD results. A confirmatory randomized trial stratified by baseline ctDNA tumor fraction8 and LCT modality, incorporating patient-reported outcomes, is the necessary and urgent next step.

Conclusions

The Dong et al. study is innovative and provides early but meaningful evidence that ctDNA-guided adaptive de-escalation of TKI therapy is feasible in a carefully selected subset of patients with advanced NSCLC who have achieved both molecular and radiological clearance. The 96% retreatment response rate and the prolonged treatment-free intervals in Groups A and B are signals worth pursuing. Clinicians considering this approach must clearly communicate to patients that ctDNA undetectability does not equate to disease eradication, that this strategy remains strictly investigational outside a clinical trial setting, and that the psychosocial consequences of treatment interruption in the metastatic context require dedicated oncology-informed psychological support.

In a discipline that has historically prioritized treatment intensification, the hypothesis that molecularly guided de-escalation may benefit a select subset of patients with oncogene-addicted advanced NSCLC represents a scientifically compelling and clinically meaningful paradigm shift. What the Dong et al. study provides is not yet practice-changing evidence but a rigorously conceived proof of concept that justifies and demands the randomized trials necessary to validate it.

References

• 1. Soria J-C, Ohe Y, Vansteenkiste J, et al: Osimertinib in Untreated EGFR-Mutated Advanced Non–Small-Cell Lung Cancer. N Engl J Med 378:113–125, 2018
• 2. Dong S, Wang Z, Zhang J-T, et al: Circulating Tumor DNA-Guided De-Escalation Targeted Therapy for Advanced Non−Small Cell Lung Cancer: A Nonrandomized Controlled Trial. JAMA Oncol 10:932, 2024
• 3. Chaudhuri AA, Chabon JJ, Lovejoy AF, et al: Early Detection of Molecular Residual Disease in Localized Lung Cancer by Circulating Tumor DNA Profiling. Cancer Discovery 7:1394–1403, 2017
• 4. Boire A, Brandsma D, Brastianos PK, et al: Liquid biopsy in central nervous system metastases: a RANO review and proposals for clinical applications. Neuro-Oncology 21:571–584, 2019
• 5. Jantus-Lewintre E, Calabuig-Fariñas S, Gandara D, et al: Harmonizing definitions in liquid biopsy: A terminology framework by the International Society of Liquid Biopsy. J Liq Biopsy 12:100462, 2026
• 6. Black JRM, Bartha G, Abbott CW, et al: Ultrasensitive ctDNA detection for preoperative disease stratification in early-stage lung adenocarcinoma. Nat Med 31:70–76, 2025
• 7. Jung H-A, Ku BM, Kim YJ, et al: Longitudinal Monitoring of Circulating Tumor DNA From Plasma in Patients With Curative Resected Stages I to IIIA EGFR-Mutant Non–Small Cell Lung Cancer. J Thorac Oncol 18:1199–1208, 2023
• 8. Rolfo CD, Khushman MM, Russo A, Borea R, et al: Chemotherapy response monitoring with DNA methylation-based ctDNA tumor fraction: Evidence from a real-world cohort of patients with advanced common solid malignancies. J Liq Biopsy 10:100442, 2025