Actionable driver gene alterations in early-stage non-small cell lung cancer: a review.

Introduction
Early detection and treatment of non-small cell lung cancer (NSCLC) at stage I–III is crucial, as surgical resection offers the only potential cure for these patients.
1
However, even after complete resection, the risk of disease recurrence remains distressingly high.2,3 Historically, the addition of platinum-based chemotherapy in the adjuvant setting yielded only a modest survival benefit—equating to an absolute improvement of ~5% in 5-year overall survival—for resected stage II–III NSCLC.4,5 Thus, there has been an urgent need for more effective systemic therapies to improve outcomes in early-stage disease. In the last few years, the therapeutic landscape of resectable NSCLC has begun to shift toward a precision oncology approach, mirroring the successes achieved in advanced NSCLC. Targeted therapies directed against oncogenic driver mutations and immunotherapies have dramatically improved outcomes in the metastatic setting. These advances are now entering earlier stages. The landmark ADAURA trial demonstrated that adjuvant osimertinib (a third-generation EGFR tyrosine kinase inhibitor (TKI)) can dramatically reduce the risk of postoperative recurrence in EGFR-mutant NSCLC (by ~80% vs placebo), leading to regulatory approval of this targeted therapy in the adjuvant setting.
6
Similarly, the phase III ALINA trial showed an unprecedented improvement in disease-free survival (DFS) with adjuvant alectinib (an ALK inhibitor) compared to chemotherapy in resected ALK-positive NSCLC (3-year DFS ~88%–89% with alectinib vs ~53%–54% with chemotherapy; hazard ratio ~0.24).
7
In parallel, immune checkpoint blockade has proven beneficial in the perioperative setting: adjuvant or neoadjuvant immunotherapy (e.g., anti-programmed cell death protein 1 (PD-1)/programmed death-ligand 1 (PD-L1) antibodies) has yielded significant gains in disease-free and event-free survival—and in some cases overall survival—for resectable NSCLC without EGFR/ALK alterations.
8
As a result, biomarker-driven treatment has become a reality in early-stage NSCLC, prompting routine testing for certain molecular markers at diagnosis. Currently, EGFR mutations and ALK rearrangements (to identify candidates for adjuvant TKIs), along with PD-L1 expression levels (to guide immunotherapy use), are the only biomarkers with approved targeted therapies in stage I–III NSCLC.9,10 In advanced NSCLC, multiplex next-generation sequencing (NGS) panels are standard, enabling parallel analysis of numerous actionable genes; this ensures that oncogenic drivers such as ROS1 fusions, RET fusions, BRAF V600E, MET exon 14 skipping (METex14), NTRK fusions, ERBB2 (HER2) mutations, KRAS mutations, and others are identified whenever present.11,12 These alterations occur in early-stage tumors as well, often at similar frequencies as in advanced disease, although targeted therapies against many of them have yet to be evaluated in the curative-intent setting.
13
By contrast, comprehensive molecular profiling has not been uniformly adopted in early-stage disease, since—prior to the advent of these new therapies—there was no clear utility for broad testing when surgery and chemotherapy were standard care.
Despite this progress, significant knowledge gaps remain regarding actionable driver gene alterations in early-stage NSCLC. Moreover, the impact of specific driver mutations on the risk of post-surgery recurrence has not been clearly defined. For example, it is unclear whether patients whose tumors harbor certain alterations (such as KRAS G12C, METex14, or others for which new therapies exist) face a higher likelihood of relapse if not given additional targeted therapy—knowledge that could refine adjuvant treatment decisions and follow-up strategies. Understanding these nuances is increasingly important: as ADAURA and ALINA have shown, identifying a targetable mutation can radically alter a patient’s postoperative management and outcomes. Conversely, for emerging biomarkers that lack approved adjuvant treatments, recognizing their presence might motivate clinical trial enrollment (e.g., trials of novel targeted agents in the adjuvant setting) or intensified surveillance for early relapse detection. In addition, the advent of sensitive liquid biopsy assays to detect minimal residual disease (MRD) via circulating tumor DNA (ctDNA) offers a new dimension to risk stratification in early-stage NSCLC, potentially integrating with genomic data to guide therapy timing and escalation.14–17
The following review will delve into the current state of molecular testing in early-stage NSCLC, the prevalence of key driver alterations in resected tumors, the potential clinical implications of these genomic findings, and the evolving role of liquid biopsy in the perioperative management of NSCLC. Through this overview, we aim to synthesize existing knowledge and identify future directions at the intersection of tumor genomics and early-stage lung cancer care.

Current molecular testing in the early-stage setting
Early detection of NSCLC, including the assessment of clinically relevant biomarkers, is crucial for optimizing patient outcomes. Although patients with stage I–III NSCLC are potentially curable through surgical resection,
18
the risk of disease recurrence remains significant and must be appropriately addressed.
1
In this context, adjuvant and/or neoadjuvant therapies play a pivotal role in reducing relapse rates and improving long-term survival. Notably, while chemotherapy has demonstrated only modest benefits in this setting,19,20 the advent of targeted therapies has marked a major therapeutic advancement.
10
To date, the only approved biomarkers for stage I–III NSCLC are EGFR mutations and ALK rearrangements, which guide the use of TKIs, and PD-L1 expression, which informs immunotherapy administration.21–23 In advanced stages, a critical issue lies in the selection of the most appropriate molecular testing strategy to determine the status of clinically actionable biomarkers. Among single-gene testing approaches, immunohistochemistry (IHC), often followed by fluorescence in situ hybridization (FISH), remains a standard method for detecting ALK rearrangements.24–26 IHC is currently the only approved technique for evaluating PD-L1 expression, mostly as a tumor proportion score.27,28 Conversely, IHC is not suitable for the detection of EGFR mutations.
12
Techniques such as quantitative PCR (qPCR) or reverse transcription PCR (RT-PCR) are commonly employed to detect known EGFR mutations and ALK rearrangements, but these methods are limited by their inability to identify novel or uncommon alterations.29,30 In this regard, NGS is strongly recommended, where available, as it provides a comprehensive molecular profile. NGS enables the simultaneous analysis of multiple biomarkers from a minimal nucleic acid input and offers high multiplexing capacity. This is particularly advantageous for the molecular characterization of emerging biomarkers relevant to stage I–III NSCLC, including RET, ROS1, BRAF, MET, NTRK, and KRAS,
31
as well as for identifying co-occurring genomic alterations.
29
The choice between single-gene assays and broad NGS for early-stage patients hinges on practical factors—such as tissue availability, turnaround time, cost, and local access to testing—but there is a growing consensus that comprehensive genomic profiling is advantageous when feasible, as it can capture the full spectrum of clinically relevant biomarkers in one analysis. Indeed, NGS allows efficient detection of both common drivers and rare or “uncommon” alterations from limited tissue, and it can reveal co-occurring mutations that might influence prognosis or treatment (e.g., comutations in TP53 or STK11). In addition, although the higher upfront cost of NGS remains a common concern, multiple studies have demonstrated that a comprehensive NGS-based strategy can reduce personnel workload and lower the overall cost per patient when compared to sequential single-gene testing.9,11,32,33 Another important consideration is the timing of when testing should be performed. In particular, in this setting, reflex molecular testing following histological diagnosis—using a comprehensive NGS approach—may be both feasible and clinically relevant. This strategy can increase the number of patients eligible for targeted therapies while also reducing turnaround times.34,35

The role of liquid biopsy
Liquid biopsy has significantly revolutionized the management of NSCLC patients. Generally, the term “liquid biopsy” refers to the analysis of ctDNA extracted from plasma. However, in a broader context, it can be extended to various biofluids (e.g., urine, saliva, effusions, and cerebrospinal fluid) as well as different analytes (e.g., ctRNA, circulating tumor cells, and extracellular vesicles).36–41 Beyond advanced disease stages, liquid biopsy has gained an important role in monitoring MRD after surgical resection. Several studies have explored the prognostic value of ctDNA-based MRD detection in early-stage NSCLC patients.
39
Among these, the TRACERx trial employed a tumor-informed approach based on Anchored Multiplex PCR chemistry, which enhances amplification of targeted regions; subsequently, in EGFR-mutant NSCLC, specific EGFR mutations in ctDNA were tracked using droplet digital PCR.
42
Notably, the use of broad gene panels capable of detecting passenger or non-canonical variants, in addition to known driver mutations, significantly increased MRD detection rates, with approximately 50% of landmark MRD-positive samples harboring non-canonical variants. This approach may improve risk stratification for recurrence in resected NSCLC patients.43,44 Another critical consideration is disease stage. For example, Xia et al.
45
reported hazard ratios for recurrence-free survival (RFS) comparing ctDNA-positive and ctDNA-negative patients of 18.0 (95% confidence interval (CI), 7.0–46.0; p < 0.001) in stage I and 5.5 (95% CI, 2.9–10.7; p < 0.001) in stage II–III disease groups.
45
The timing of liquid biopsy testing after surgery is also crucial; studies have described blood sampling performed from 3 days up to 120 days post-surgery. In the LUNGCA-1 study, RFS hazard ratios for ctDNA-positive versus ctDNA-negative patients were 8.6 (95% CI, 4.7–15.6; p < 0.001) and 14.3 (95% CI, 7.9–25.9; p < 0.001) for samples collected at 3 days and 1 month post-surgery, respectively.
45
Similarly, the PROPHET study reported hazard ratios for DFS in ctDNA-positive patients of 5.31 (p < 0.001) and 16.4 (p < 0.001) at 3–7 and 30 days after surgery, respectively.
43
An emerging area of investigation involves the use of liquid biopsy in the perioperative management of early-stage NSCLC. Preoperative ctDNA positivity may represent a novel poor prognostic factor in these patients.
46
Another crucial point is that the sensitivity of ctDNA assays can be improved when known mutations from initial tumor tissue testing are available. In such cases, specific point mutations can be actively searched for during analysis—an approach known as “tumor-informed” testing.

Prevalence of driver gene alterations in resected or resectable patients
In the context of early-stage NSCLC, assessing the molecular status of clinically relevant biomarkers—alongside evaluating PD-L1 expression levels—has become central to guiding treatment decisions. As previously mentioned, the only biomarkers currently approved for stage I–III NSCLC include EGFR mutations, ALK rearrangements (for TKIs), and PD-L1 expression levels (to guide immunotherapy).21–23
The Cancer Genome Atlas (TCGA) reported EGFR mutations in approximately 11% of patients with early-stage lung adenocarcinoma.
47
In a large German study on resected NSCLC using an amplicon-based NGS approach, EGFR mutations were detected in 12.7% of early-stage adenocarcinoma cases. Notably, the majority (72.2%) were “common” mutations—namely exon 19 deletions and exon 21 p.L858R point mutations. These were followed by “uncommon” mutations (9.7% consisting of point mutations and indels within exons 18–21, and 9.7% exon 20 insertions), and “compound” mutations (8.3%). Nearly half of the EGFR-mutated cases (52.8%) harbored co-occurring genomic alterations.
31
Another large NGS-based cohort found EGFR mutations in 12.6% of early-stage adenocarcinoma patients, with common exon 19 deletions and exon 21 p.L858R mutations accounting for 32.5% and 31.1% of cases, respectively. “Uncommon” EGFR mutations represented 30.2% (within exons 18–21), while exon 20 insertions accounted for 6.2%.
48
Similarly, Varela et al.
49
reported EGFR mutation rates of 14.5% (via RT-qPCR, Idylla™; Biocartis, Mechelen, Belgium) and 19.5% (via NGS, Oncomine™; Thermo Fisher Scientific Inc., Waltham, MA, USA), emphasizing the capacity of NGS to detect concurrent alterations, as remarked by most recent retrospective cohorts reporting higher (>20%) prevalence of EGFR mutations.13,50 As expected, the prevalence of EGFR mutations was higher among Asian patients (53.0%) compared to other ethnicities.
51

TCGA reported KRAS mutations in about 32% of early-stage non-squamous NSCLC patients.
47
Stephan-Falkenau et al. found KRAS mutations in 40.4% of NSCLC patients (78% adenocarcinoma)—17.3% had the exon 2 p.G12C mutation and 23.1% had non-p.G12C mutations. Most KRAS mutations were located in exon 2 (95.2%), followed by exon 3 (4.4%) and exon 4 (0.4%). Only one patient carried two distinct KRAS mutations. Among patients with the p.G12C variant, comutations in TP53 and STK11 were found in 34.7% and 9.6% of cases, respectively.
31
Bossè et al. similarly identified KRAS mutations in 45.7% of early-stage patients, predominantly in exon 2 (92.3%) and exon 3 (7.7%). The most common mutation was p.G12C, representing 43.2% of KRAS mutations and 20.0% of the total cohort. They also reported comutations involving BRAF, ERBB2, MET, TP53, and PIK3CA.
48

TCGA reported BRAF mutations in 7% of early-stage NSCLC cases.
47
Stephan-Falkenau et al. found BRAF mutations in 5.7% of cases, including 1.8% with exon 15 p.V600E and 3.9% with non-V600E variants. Some BRAF-mutated cases had co-alterations in KRAS.
31
Bossè et al. identified BRAF mutations in 4.6% of patients, with 21.6% of those occurring at codon 600—most commonly exon 15 p.V600E (20.3%). One patient had a concurrent EGFR exon 20 insertion.
48

ERBB2 mutations were reported in 2.2% of patients by TCGA,
47
and in 0.88% by Stephan-Falkenau et al.,
31
with two patients also harboring TP53 mutations. Bossè et al.
48
reported ERBB2 mutations in 1.6% of cases, including one patient with a co-occurring EGFR exon 20 insertion.
ALK rearrangements were identified in 1.3% of cases by TCGA,
47
and in 1.2% by Stephan-Falkenau et al.
31
—almost all involving EML4, with one involving KIF5B. Bossè et al.
48
found ALK rearrangements in 0.9% of patients, mostly with EML4, and one each with KIF5B and HIP1.
ROS1 rearrangements were reported in 1.7% of patients by TCGA.
47
Stephan-Falkenau et al.
31
identified a single case (0.2%) with an SDC4::ROS1 fusion, while Bossè et al.
48
reported a 0.3% frequency, with fusion partners including CD74, EZR, and SDC4.
RET rearrangements were found in 0.9% of cases by TCGA,
47
and 0.7% by both Stephan-Falkenau et al.
31
and Bossè et al.,
48
with KIF5B and CCDC6 as fusion partners.
METex14 alterations were present in 4.3% of TCGA cases,
47
2.3% in Stephan-Falkenau et al.
31
’s cohort—with TP53 comutations in 38.5%—and 2.9% in Bossè et al.
48
’s study.
No NTRK fusions were identified in either Stephan-Falkenau et al.
31
or Bossè et al. cohorts.
48

Regarding PD-L1 expression (using Dako 28-8 validated assay), Kerr et al. found that 43.4% of resected NSCLC patients were positive (⩾1%). Using cut-offs of 5%, 25%, and 50%, the proportion of positive cases was 34.0%, 22.6%, and 16.6%, respectively.
22

Potential implications in clinical practice
The identification of actionable driver gene alterations in early-stage NSCLC has relevant clinical implications that extend beyond the current standard of care, which includes targeted adjuvant therapies for EGFR- and ALK-positive tumors.52–54 In this context, recent data suggest that a considerable proportion of resected stage I–III NSCLC tumors harbor genomic alterations for which targeted therapies are approved in the metastatic setting but not yet implemented perioperatively.13,50 This includes alterations in KRAS, MET, HER2, BRAF, and RET, among others.
Data from the Italian AGA-R cohort showed that driver alterations were present in 71% of resected non-squamous tumors, with a comparable prevalence (73%) even in stage I disease.13,55
KRAS mutations were the most frequent, followed by EGFR alterations—whose prevalence (29% in stage I) was notably higher than historically observed in advanced-stage Western populations. METex14 mutations also appeared enriched in early stages (8% in stage I).
13
These findings highlight the potential role of NGS in identifying patients who may benefit from future adjuvant or surveillance strategies tailored to their molecular profile.
Currently, osimertinib and alectinib represent the only targeted agents approved in the adjuvant setting6,7 (Table 1). However, more than half of the molecular alterations detected in early-stage NSCLC are associated with targeted therapies already approved in metastatic disease, including MET inhibitors, BRAF/MEK inhibitor combinations, RET inhibitors, HER2-targeting agents, and KRAS G12C inhibitors, not yet explored in randomized adjuvant trials.
56
To date, few clinical trials are planned in the adjuvant setting according to mutational status (NCT04302025), and in the earlier stage IA-B stage setting (NCT06955325). In addition, the TARGET clinical trial is exploring the need to prolong osimertinib adjuvant treatment after 3 years,
57
and further investigation should be set on the role of rechallenging the adjuvant TKI at disease progression after completion of adjuvant treatment.
A key observation in the AGA-R study was the association between driver mutations and recurrence risk, with 39.6% recurrence among patients harboring oncogenic drivers, compared to 29.6% in wild-type tumors.13,55 Notably, 70% of recurrences in stage I occurred in patients with a driver alteration, supporting the hypothesis that molecular features may contribute to early relapse risk—even in tumors traditionally considered low risk. Specific alterations, including gene fusions, EGFR exon 20 insertions, and BRAF non-V600 mutations, were associated with higher recurrence rates (up to 75%). By contrast, METex14 mutations and uncommon EGFR mutations showed lower recurrence rates (15%–17%), suggesting variability in biological behavior.
13

These results highlight the potential for oncogene-specific recurrence risk stratification, which could inform postoperative surveillance intensity, duration of follow-up, and patient selection for future adjuvant studies, including earlier intervention in high-risk subsets.
Overall, the incorporation of broad molecular profiling at the time of diagnosis in early-stage NSCLC can provide critical insights beyond EGFR and ALK status. These findings underscore the need for dedicated clinical trials evaluating the role of targeted therapies in the perioperative setting for other actionable alterations, as well as the integration of molecular risk factors into clinical decision-making and surveillance strategies, particularly in stage I disease. In addition, in this complex scenario, although only a limited number of actionable biomarkers are currently approved for clinical use, a wide range of potential prognostic and predictive biomarkers should also be considered. Therefore, it is preferable to adopt methodologies that allow for the simultaneous analysis of multiple biomarkers across different patients—such as NGS—while reserving single-gene approaches (such as PCR-based assays, IHC, or FISH) for specific situations. These include cases not covered by NGS or when confirmatory testing is needed using orthogonal methods.

Conclusion
The landscape of early-stage NSCLC is evolving rapidly, driven by the integration of molecular profiling and the expansion of targeted treatment options beyond advanced disease. Recent evidence demonstrates that a high proportion of resected NSCLC tumors harbor actionable driver gene alterations. To date, EGFR and ALK remain the only biomarkers guiding adjuvant targeted therapy, but this narrow focus may limit clinical benefit for a broader population. Recurrence rates vary significantly across molecular subgroups, suggesting that driver mutations may influence post-surgical outcomes and should be considered when designing adjuvant strategies and follow-up protocols.
In summary, recent advances have set the stage for a more personalized approach to early-stage lung cancer, highlighting the rationale for comprehensive molecular assessment in resectable NSCLC, as the field moves beyond a “one size fits all” paradigm to incorporate molecularly actionable driver gene alterations into risk assessment and adjuvant treatment planning (Figure 1).