Real-World Outcomes of Fusion-Directed Targeted Therapy in Advanced Non-Small Cell Lung Cancer Harboring Actionable Gene Fusions

1. Introduction

Lung cancer remains the leading cause of cancer-related mortality worldwide, accounting for approximately 2.5 million new cases and 1.8 million deaths in 2022. Non-small cell lung cancer (NSCLC) comprises approximately 85% of lung cancers and remains a major global health burden. Over the past two decades, management of advanced NSCLC has shifted from a predominantly histology-based approach toward precision oncology, driven by identification of actionable molecular alterations and development of targeted therapies. Comprehensive molecular profiling by next-generation sequencing (NGS) is recommended for patients with advanced non-squamous NSCLC because identification of actionable alterations directly informs treatment selection.[1,2,3,4]

Among the most clinically relevant genomic alterations are oncogenic gene fusions, which arise through chromosomal rearrangements that juxtapose previously separate genes and may produce constitutively active fusion proteins. The principal actionable fusion groups in NSCLC include ALK, ROS1, RET, NTRK1/2/3, and NRG1. Although each is individually uncommon, together they identify a clinically important population eligible for fusion-directed therapy.[2,5,6]

The therapeutic landscape for fusion-positive NSCLC has evolved rapidly. Primary studies of next-generation ALK inhibitors, ROS1 inhibitors, selective RET inhibitors, tumor-agnostic TRK inhibitors, and zenocutuzumab for NRG1 fusion-positive cancer have established substantial and often durable antitumor activity across these molecular subsets.[7,8,9,10,11,12,13,14,15,16]

The widespread implementation of comprehensive genomic profiling has changed the diagnostic and therapeutic paradigm of advanced NSCLC. Contemporary NGS platforms can evaluate multiple cancer-associated genes simultaneously, improving the ability to identify both common and rare actionable alterations from limited specimens. Molecular-testing recommendations therefore support broad profiling before systemic treatment in advanced non-squamous NSCLC.[3,4]

Despite these advances, important knowledge gaps remain. Most efficacy evidence originates from small prospective trials with protocol-defined eligibility, whereas real-world investigations have generally focused on individual fusion subtypes or biomarker prevalence. Few cohorts have applied common outcome definitions across all currently actionable fusion groups within one health system.[17,18,19,20]

We therefore conducted a retrospective health-system cohort study to characterize the clinicopathologic features, treatment patterns, and survival outcomes of patients with advanced NSCLC harboring ALK, ROS1, RET, NTRK, or NRG1 gene fusions who received fusion-directed targeted therapy. We hypothesized that patients treated in routine clinical practice would experience durable disease control despite the greater clinical heterogeneity inherent to a real-world population.

2. Methods

2.1. Study Design and Setting

This retrospective observational cohort study was conducted at Memorial Cancer Institute, including Memorial Cancer Institute West and Memorial Cancer Institute East within Memorial Healthcare System in South Florida. Consecutive adults with advanced NSCLC harboring actionable oncogenic gene fusions who received care at either campus were identified through institutional electronic health records, molecular pathology reports, and oncology registries.

The study was approved by the Memorial Healthcare System Institutional Review Board (Protocol No. MHS.2025.175; January 13, 2026) and was conducted in accordance with the Declaration of Helsinki and applicable regulations. The requirement for informed consent was waived because of the retrospective design.

The study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.[21]

Patients diagnosed between May 14, 2014, and October 10, 2025, were eligible. Clinical follow-up continued through the prespecified cutoff of April 24, 2026.

2.2. Patient Selection

Eligible patients were identified through institutional molecular testing databases and thoracic oncology registries. Inclusion criteria were: age ≥18 years; histologically confirmed lung adenocarcinoma; stage IV or recurrent NSCLC; presence of an actionable ALK, ROS1, RET, NTRK, or NRG1 fusion confirmed by clinically validated molecular testing; receipt of at least one fusion-directed targeted therapy; and adequate follow-up for survival assessment. Patients without molecular confirmation or adequate follow-up were excluded.

2.3. Molecular Testing

Comprehensive molecular profiling was performed as part of routine clinical care using CLIA-certified next-generation sequencing platforms, including Tempus, Caris Life Sciences, and Guardant360. Fluorescence in situ hybridization (FISH) or other clinically validated molecular assays were used when appropriate depending on specimen availability and testing era.

Fusion nomenclature followed current IASLC recommendations. NTRK1, NTRK2, and NTRK3 rearrangements were analyzed collectively as NTRK fusion-positive NSCLC. One patient harboring an ALK rearrangement also demonstrated a concurrent EGFR alteration and received osimertinib during the disease course; this patient remained eligible for analysis.

2.4. Data Collection

Clinical data were abstracted from the electronic health record using a standardized data collection instrument. Variables included age, sex, race/ethnicity, smoking history, molecular subtype, targeted therapy, treatment line, dates of progression, death, and last clinical follow-up. Race and ethnicity were summarized descriptively. Treatment-naïve status was incompletely documented and therefore excluded from inferential analyses.

2.5. Study Outcomes

The primary endpoint was progression-free survival (PFS), defined as the interval from initiation of fusion-directed targeted therapy until objective disease progression or death from any cause, whichever occurred first. Overall survival (OS) was defined as the interval from initiation of fusion-directed targeted therapy until death from any cause. Secondary endpoints included the distribution of actionable fusion subtypes, treatment patterns, survival by fusion subtype, and exploratory analyses according to treatment line. Patients without a PFS event were censored at the date of last clinical follow-up or April 24, 2026, whichever occurred first; patients alive at that time were censored for OS.

2.6. Statistical Analysis

Continuous variables are summarized as medians with ranges or interquartile ranges, and categorical variables as frequencies and percentages. Median follow-up was estimated using the reverse Kaplan-Meier method.22 Progression-free and overall survival were estimated using the Kaplan-Meier method, and survival curves were compared using the log-rank test.23 Exploratory univariable Cox proportional hazards models estimated hazard ratios (HRs) and 95% confidence intervals (CIs) according to treatment line.24 Because only 30 PFS events and 16 deaths occurred, multivariable Cox regression was not performed to avoid model overfitting and unstable parameter estimates. Missing data were handled using complete-case analysis for inferential models. Given the exploratory nature of the subgroup analyses, no adjustment for multiple comparisons was performed. No formal sample-size calculation was undertaken because the study included all eligible patients treated during the prespecified accrual period. All statistical tests were two-sided, with P < .05 considered statistically significant. Analyses were performed using Python version 3.13.5 and stats models version 0.14.6. All subgroup and regression analyses were considered exploratory and hypothesis-generating.

3. Results

3.1. Patient Cohort

Between May 14, 2014, and October 10, 2025, 68 patients with advanced lung adenocarcinoma harboring an actionable gene fusion met the eligibility criteria and received fusion-directed targeted therapy at Memorial Cancer Institute. Clinical follow-up continued through April 24, 2026. The median follow-up, estimated using the reverse Kaplan-Meier method, was 43.1 months (95% CI, 24.4-50.0 months). The analytic cohort assembly is illustrated in Figure 1.

3.2. Clinicopathologic Characteristics

The median age at cancer diagnosis was 62 years (range, 27-90 years); 40 patients (58.8%) were female, and 37 (54.4%) were never-smokers. All tumors were adenocarcinomas. Race was recorded as White in 60 patients (88.2%), Black in 6 (8.8%), and Asian in 2 (2.9%). Hispanic ethnicity was recorded in 43 patients (63.2%); the remaining 25 patients had heterogeneous non-Hispanic, national-origin, or other descriptors in the source record and are presented descriptively in Table 1.

3.3. Molecular Landscape and Treatment Patterns

ALK rearrangements were the most common actionable alteration (34 patients, 50.0%), followed by ROS1 (15, 22.1%), RET (13, 19.1%), NTRK (4, 5.9%), and NRG1 (2, 2.9%). Fusion-directed therapy was administered in the first-line setting in 41 patients (60.3%), second-line setting in 19 (27.9%), and third-line setting in 8 (11.8%). Alectinib was the most frequently used ALK-directed therapy, entrectinib the most frequently used ROS1-directed therapy, selpercatinib the predominant RET-directed therapy, larotrectinib the predominant NTRK-directed therapy, and zenocutuzumab was used in both patients with NRG1 fusions. One patient with concurrent ALK and EGFR alterations received osimertinib (Table 2).

3.4. Survival Outcomes

At the data cutoff, 30 PFS events had occurred and 38 patients (55.9%) were censored for PFS. Sixteen deaths had occurred, and 52 patients (76.5%) were censored for OS. Median PFS for the overall cohort was 44.8 months (95% CI, 17.1 months to not estimable), whereas median OS was not reached. Estimated PFS rates at 12, 24, and 36 months were 67.5% (95% CI, 54.2%-77.7%), 57.9% (95% CI, 44.1%-69.4%), and 50.2% (95% CI, 36.0%-62.9%), respectively. Corresponding OS estimates were 90.4% (95% CI, 79.8%-95.6%), 80.9% (95% CI, 68.1%-89.0%), and 71.4% (95% CI, 56.5%-81.9%). Kaplan–Meier curves for the overall cohort are shown in Figure 2 and Figure 3.

3.5. Survival by Fusion Subtype

Among the three largest molecular subgroups, median PFS was not reached for ALK, 16.6 months for ROS1, and 29.8 months for RET. Median OS was not reached for ALK and ROS1 and was 61.8 months for RET. Survival distributions did not differ significantly among the ALK, ROS1, and RET subgroups for PFS (log-rank P = .342) or OS (log-rank P = .966). Estimates for NTRK and NRG1 were descriptive because of the small subgroup sizes (Table 3). Kaplan–Meier estimates of progression-free survival by fusion subtype are presented in Figure 4.

3.6. Exploratory Analysis by Treatment Line

Median PFS was 60.8 months among patients who received fusion-directed targeted therapy in the first-line setting and 16.6 months among patients treated in the second- or third-line setting (log-rank P = .031). Median OS was not reached in either group, although OS distributions differed by treatment line (log-rank P = .029). In exploratory univariable Cox models, later-line targeted therapy was associated with an increased hazard of progression or death (HR, 2.16; 95% CI, 1.05-4.44; P = .036) and death (HR, 2.87; 95% CI, 1.06-7.76; P = .037). Because treatment line was not randomized, the analyses were unadjusted, and no correction for multiple comparisons was applied, these findings should be interpreted as hypothesis-generating rather than causal (Table 4). Progression-free survival according to treatment line is illustrated in Figure 5.

Table 4. Exploratory Univariable Cox Regression by Treatment Line.

Outcome	Comparison	HR	95% CI	P value
PFS	Second/third line vs first line	2.16	1.05-4.44	.036
OS	Second/third line vs first line	2.87	1.06-7.76	.037

CI, confidence interval; HR, hazard ratio; OS, overall survival; PFS, progression-free survival. Models were univariable and exploratory. Treatment line was not randomized; results should not be interpreted causally.

Table 4. Exploratory Univariable Cox Regression by Treatment Line.

Outcome	Comparison	HR	95% CI	P value
PFS	Second/third line vs first line	2.16	1.05-4.44	.036
OS	Second/third line vs first line	2.87	1.06-7.76	.037

CI, confidence interval; HR, hazard ratio; OS, overall survival; PFS, progression-free survival. Models were univariable and exploratory. Treatment line was not randomized; results should not be interpreted causally.

Table 5. Contextual Comparison with Selected Landmark Trials.

Fusion	Study / therapy	Population	ORR	Median PFS	Median OS	Interpretive note
ALK	ALEX / alectinib7	Previously untreated	82.9%	34.8 mo	NR	Updated investigator-assessed PFS
ALK	CROWN / lorlatinib8	Previously untreated	76%	NR; 5-y PFS 60%	NR	5-year analysis
ALK	Current cohort	Mixed TKIs and lines	Not collected	NR	NR	Real-world heterogeneous cohort
ROS1	PROFILE 1001 / crizotinib9	Advanced ROS1-positive	72%	19.3 mo	51.4 mo	Single-arm phase 1 expansion
ROS1	Integrated entrectinib analysis	ROS1 TKI-naive	68%	15.7 mo	47.8 mo	Pooled prospective trials
ROS1	TRIDENT-1 / repotrectinib11	ROS1 TKI-naive	79%	35.7 mo	NR; 18-mo OS 88%	Primary efficacy population
ROS1	Current cohort	Mixed agents and lines	Not collected	16.6 mo	NR	Real-world heterogeneous cohort
RET	LIBRETTO-001 / selpercatinib12	Previously treated / untreated	61% / 84%	24.9 mo	NR	Updated registrational analysis
RET	LIBRETTO-431 / selpercatinib13	First-line randomized cohort	84%	24.8 mo	Immature	Versus platinum chemotherapy +/- pembrolizumab
RET	ARROW / pralsetinib14	Treatment-naive	72%	13.0 mo	NR / immature	Phase 1/2 update
RET	Current cohort	Selpercatinib or pralsetinib	Not collected	29.8 mo	61.8 mo	Small real-world subgroup
NTRK	Larotrectinib lung analysis15	TRK fusion-positive lung cancer	73%	35.4 mo	40.7 mo	15 efficacy-evaluable patients
NTRK	Current cohort	Larotrectinib or repotrectinib	Not collected	Descriptive	Descriptive	n=4
NRG1	eNRGy / zenocutuzumab16	NRG1 fusion-positive NSCLC	29%	6.8 mo*	Not mature	*PFS for overall efficacy population
NRG1	Current cohort	Zenocutuzumab	Not collected	Descriptive	Descriptive	n=2

Abbreviations: NR, not reached; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; TKI, tyrosine kinase inhibitor. Values are shown for context only. Cross-trial comparisons are not adjusted and should not be interpreted as evidence of equivalence or superiority.:.

Table 5. Contextual Comparison with Selected Landmark Trials.

Fusion	Study / therapy	Population	ORR	Median PFS	Median OS	Interpretive note
ALK	ALEX / alectinib7	Previously untreated	82.9%	34.8 mo	NR	Updated investigator-assessed PFS
ALK	CROWN / lorlatinib8	Previously untreated	76%	NR; 5-y PFS 60%	NR	5-year analysis
ALK	Current cohort	Mixed TKIs and lines	Not collected	NR	NR	Real-world heterogeneous cohort
ROS1	PROFILE 1001 / crizotinib9	Advanced ROS1-positive	72%	19.3 mo	51.4 mo	Single-arm phase 1 expansion
ROS1	Integrated entrectinib analysis	ROS1 TKI-naive	68%	15.7 mo	47.8 mo	Pooled prospective trials
ROS1	TRIDENT-1 / repotrectinib11	ROS1 TKI-naive	79%	35.7 mo	NR; 18-mo OS 88%	Primary efficacy population
ROS1	Current cohort	Mixed agents and lines	Not collected	16.6 mo	NR	Real-world heterogeneous cohort
RET	LIBRETTO-001 / selpercatinib12	Previously treated / untreated	61% / 84%	24.9 mo	NR	Updated registrational analysis
RET	LIBRETTO-431 / selpercatinib13	First-line randomized cohort	84%	24.8 mo	Immature	Versus platinum chemotherapy +/- pembrolizumab
RET	ARROW / pralsetinib14	Treatment-naive	72%	13.0 mo	NR / immature	Phase 1/2 update
RET	Current cohort	Selpercatinib or pralsetinib	Not collected	29.8 mo	61.8 mo	Small real-world subgroup
NTRK	Larotrectinib lung analysis15	TRK fusion-positive lung cancer	73%	35.4 mo	40.7 mo	15 efficacy-evaluable patients
NTRK	Current cohort	Larotrectinib or repotrectinib	Not collected	Descriptive	Descriptive	n=4
NRG1	eNRGy / zenocutuzumab16	NRG1 fusion-positive NSCLC	29%	6.8 mo*	Not mature	*PFS for overall efficacy population
NRG1	Current cohort	Zenocutuzumab	Not collected	Descriptive	Descriptive	n=2

Abbreviations: NR, not reached; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; TKI, tyrosine kinase inhibitor. Values are shown for context only. Cross-trial comparisons are not adjusted and should not be interpreted as evidence of equivalence or superiority.:.

4. Discussion

In this retrospective cohort of 68 patients with advanced lung adenocarcinoma harboring actionable gene fusions, fusion-directed targeted therapy was associated with durable disease control in routine clinical practice. Median progression-free survival (PFS) for the overall cohort was 44.8 months, and median overall survival (OS) was not reached after a reverse Kaplan–Meier median follow-up of 43.1 months. Survival outcomes did not differ significantly among the ALK, ROS1, and RET subgroups, although these comparisons were limited by small subgroup sizes and a low number of events. Patients who received targeted therapy in the first-line setting demonstrated longer PFS and OS than those treated in later lines. However, this exploratory association should not be interpreted as evidence of a causal treatment effect, as treatment line was not randomized and was likely influenced by factors such as treatment era, prior therapies, disease characteristics, and access to comprehensive molecular testing.

4.1. Contextual Comparison with Landmark Trials

The outcomes observed in the present cohort fall within the broad range reported across pivotal studies of contemporary fusion-directed therapies; however, direct cross-trial comparisons are not valid because of differences in eligibility criteria, treatment line, drug exposure, endpoint assessment, and follow-up. In ALK-positive disease, the updated ALEX analysis reported a median investigator-assessed PFS of 34.8 months with first-line alectinib, whereas the 7-year CROWN analysis found that median PFS with lorlatinib remained unreached and that 55 % of patients were alive without progression at 7 years.7,8 In our ALK subgroup, median PFS and OS were not reached despite inclusion of several ALK inhibitors and multiple treatment lines.

4.2. ROS1-Positive Disease

For ROS1 fusion-positive NSCLC, the updated PROFILE 1001 analysis reported median PFS of 19.3 months and median OS of 51.4 months with crizotinib.⁹ A long-term integrated analysis of entrectinib reported median PFS of 15.7 months and median OS of 47.8 months, whereas TRIDENT-1 reported median PFS of 35.7 months among ROS1 TKI-naive patients treated with repotrectinib.^10,11 The median PFS of 16.6 months in our ROS1 subgroup is therefore most similar to the crizotinib and entrectinib experience, which is clinically plausible given the mixture of agents, treatment eras, and treatment lines represented in this cohort.

4.3. RET Fusion-Positive Disease

In RET fusion-positive NSCLC, the updated LIBRETTO-001 analysis reported durable activity with selpercatinib, including median PFS of 24.9 months in the efficacy population.[12] In the randomized LIBRETTO-431 trial, first-line selpercatinib produced median PFS of 24.8 months and an objective response rate of 84%.[13] The updated ARROW analysis reported an objective response rate of 72% and median PFS of 13.0 months among treatment-naive patients receiving pralsetinib.[14] Median PFS in our RET subgroup was 29.8 months and median OS was 61.8 months. These results are directionally consistent with sustained benefit from selective RET inhibition, but the small number of RET patients and events precludes definitive comparison.

4.4. Rare NTRK and NRG1 Fusions

The NTRK and NRG1 cohorts were too small for meaningful survival inference, but their inclusion illustrates the clinical value of broad fusion testing. In a lung cancer-specific pooled analysis, larotrectinib produced an objective response rate of 73%, median PFS of 35.4 months, and median OS of 40.7 months in evaluable patients with TRK fusion-positive lung cancer.[15] In the eNRGy trial, zenocutuzumab demonstrated activity in NRG1 fusion-positive cancer, including NSCLC.[16] Our four NTRK-positive and two NRG1-positive patients provide descriptive real-world evidence only and should not be compared numerically with these prospective cohorts.

4.5. Real-World Evidence and Study Contribution

Most published real-world studies have focused on individual fusion subtypes or the prevalence of actionable gene fusions rather than comparing treatment outcomes across multiple fusion-defined populations. Hess and colleagues characterized patients with RET fusion–positive NSCLC using a large U.S. real-world database, while Lu and colleagues described outcomes in a contemporary Chinese RET fusion–positive cohort.17,18 Overbeck and colleagues screened 1,068 unselected NSCLC cases and identified NTRK fusions in only 0.2% of tumors, underscoring their rarity.19 The international eNRGy1 registry similarly characterized the clinical features of NRG1 fusion–positive lung cancers across multiple centers.20 In contrast, the present study evaluated patients with ALK, ROS1, RET, NTRK, and NRG1 fusion–positive lung adenocarcinoma within a single health system, using a unified clinical dataset and standardized survival definitions to enable consistent comparisons across fusion subtypes.

4.6. Precision Oncology in Routine Practice

These findings support comprehensive genomic profiling as a prerequisite for precision treatment in advanced nonsquamous NSCLC. Broad NGS can identify common drivers while also detecting rare, clinically actionable rearrangements that may be missed by sequential assays.[3,4] In our cohort, molecular testing identified all five fusion groups. Timely testing is particularly important because actionable results determine eligibility for fusion-directed therapy; however, the exploratory association between treatment line and outcomes may also reflect confounding by treatment era and prior therapy.

4.7. Implications for Immunotherapy

The study did not directly evaluate immunotherapy efficacy. Nevertheless, systematic reviews and meta-analyses have generally reported limited activity of immune checkpoint inhibitors in oncogene-driven NSCLC, including RET fusion-positive and other driver-defined populations.[25,26,27] These data support obtaining molecular results before systemic treatment whenever clinically feasible and prioritizing an approved targeted therapy when a relevant actionable fusion is identified.

5. Strengths and Limitations

This study has several strengths, including patient-level outcome data, a reverse Kaplan-Meier median follow-up exceeding 3.5 years, inclusion of all five clinically actionable fusion groups, and uniform outcome definitions within one health system. The cohort also includes a substantial Hispanic population. Important limitations include the retrospective design, single-system setting, modest sample size, treatment heterogeneity across an 11-year diagnostic period, and limited progression and death events. The dataset did not systematically capture performance status, brain metastases, response rate, adverse events, or molecular-testing platform at the individual-patient level; these variables therefore could not be analyzed. Ethnicity categories were recorded inconsistently in the electronic health record and were presented descriptively without inferential analysis. Finally, cross-trial comparisons are contextual only and cannot establish equivalence between this cohort and prospective trials.

6. Conclusions

Patients with actionable fusion-positive advanced NSCLC experienced durable disease control with fusion-directed targeted therapy in routine clinical practice. The survival outcomes observed in this cohort were broadly consistent with those reported in pivotal clinical trials and other real-world studies, while the inclusion of patients with rare NTRK and NRG1 fusions highlights the clinical value of comprehensive genomic profiling. Larger multicenter studies with standardized clinical data, treatment-specific cohorts, and longer follow-up are needed to better define comparative effectiveness, characterize outcomes across rare fusion subtypes, and inform optimal treatment sequencing.