Molecular Characterisations and the Association with Clinical Factors in RET Fusion-Positive NSCLC: A Retrospective Study of the Single Center Cohort

Introduction

Gene fusion is a key molecular variation in non-small cell lung cancer (NSCLC), affecting 8%–12% of NSCLC patients. Approved tyrosine kinase inhibitors (TKIs) targeting distinct gene fusions have brought substantial benefits to affected patients.

Rearranged during transfection (RET) fusion, identified post EGFR/ALK, is an emerging driver in NSCLC, with an incidence of 1%–2%.^1–3 RET fusion-positive patients often have distinct clinical features: younger age, non-smoking status, adenocarcinoma histology, better performance status, and similar male/female prevalence.^4,5 Brain metastasis is common, and RET fusions are generally considered mutually exclusive with other oncogenic drivers such as EGFR mutations and ALK rearrangements, although rare co-occurrences have been reported. However, Tsuta K et al⁶ and Michels S et al⁷ found no significant gender/smoking differences in RET fusion cases. Cheng Ying et al’s Phase II study⁸ identified KIF5B, CCDC6, and NCOA4 as RET fusion partners in brain metastasis cases, but other partners remain unclear current conclusions on RET fusion-related clinical factors are inconsistent. According to the ESMO recommendations,⁹ for patients with RET fusion-positive NSCLC, it is recommended to use next-generation sequencing technology (NGS) to sequence samples from different tissue types. Although significant progress has been made in the study of RET fusions in NSCLC, most current studies have focused primarily on common fusion partners such as KIF5B and CCDC6, and systematic investigations into the molecular characteristics and clinical significance of rare fusion partners remain limited. Recent studies have suggested that different RET fusion partners may influence the clinical characteristics and therapeutic responses of patients. A study by Sun et al¹⁰ found that patients with non-KIF5B fusion partners had a significantly longer median progression-free survival (mPFS) when treated with pralsetinib than those with KIF5B fusions (17.0 months vs. 5.5 months, P = 0.0473), indicating that the type of fusion partner may affect sensitivity to targeted therapy. In addition, Wang et al¹¹ reported a rare case of RET:FOXJ3 fusion in which the patient showed no response to pralsetinib, further demonstrating that rare fusion partners may exhibit distinct biological behaviors and clinical implications.

At present, descriptive data based on NGS testing regarding the spectrum of rare RET fusion partners and their associations with clinical factors in RET fusion-positive NSCLC remain scarce, particularly reports involving systematic identification using large multigene panels.

The present study aimed to systematically analyze the molecular features and clinicopathological factors of RET fusion-positive NSCLC patients in a single-center cohort from Jilin Province using multigene NGS, with a focus on the identification of rare fusion partners and their potential clinical significance. This study was designed to supplement the existing data on rare fusion partners and provide new evidence for the precise classification of RET fusion-positive NSCLC.

Materials and Methods

We performed a retrospective study involving 6,204 consecutive lung cancer patients who underwent next-generation sequencing for treatment plan formulation in the Translational Oncology Research Lab of Jilin Cancer Hospital between June 2018 and September 2024. We collected tissue samples via surgery or puncture for patients from whom tissue could be obtained, and used blood, malignant effusion, or cerebrospinal fluid specimens for those unable to provide tissue samples for NGS detection. Patients with RET fusion identified by NGS sequencing were included in the positive rate statistics of this study. We extracted basic clinical and pathological characteristics of the patients from the electronic medical record system. These characteristics included age, gender, smoking status, clinical stage at diagnosis, pathological type, metastatic status, as well as other variables. All included cases are consecutive NGS-tested cases without any purposeful artificial screening. In the statistics of clinical characteristics, if there is missing data for patients, the missing data are handled by listwise deletion, and only cases with complete data for all variables are retained.

Table 1 The RET Fusion Partners Discovered in Lung Cancer Patients in This Study

The NGS panels used for RET fusion detection in DNA level included 8‑gene, 59‑gene, 68‑gene, 168‑gene, 520‑gene, and 1021‑gene. The 8‑gene panel does not cover FAM107B‑RET (F1:R8), but covers 10 out of the 11 rare fusion variants listed in Table 1, accounting for 90.9% (10/11). Sequencing was performed on the Illumina NovaSeq 6000 and Gene+ Seq‑2000 platforms. Gene fusions were identified using FusionCaller v1.2, STAR‑Fusion v1.10, an in‑house pipeline markSV v4.2.4, and NCSV2 v1.2.0 software. A threshold of ≥4 supporting split reads was applied for samples sequenced on the Illumina platform, and ≥10 supporting split reads for those on the Gene+ Seq‑2000 platform. All final fusions were manually reviewed for breakpoint accuracy and sequencing background noise. Limit of detection in our panels was 2% and the sensitivity was >98%.

Sequence data were mapped to the reference human genome (hg19) using Burrows-Wheeler Aligner version 0.7.10. Local alignment optimization, duplication marking and variant calling were performed using Genome Analysis Tool Kit version 3.2, and VarScan version 2.4.3. Tissue and plasma samples were compared against their own white blood cell control to identify somatic variants. Variants were filtered using the VarScan fpfilter pipeline, loci with depth less than 100 were filtered out. Analysis of gene fusion was performed using Factera version 1.4.3.

Statistical Analysis

Patient clinical characteristics were described using percentages. Clinical characteristics between different groups, including sex, age, smoking history, histology, and metastatic status, were compared using the χ²-test or Fisher’s exact test. The strength of association was quantified using odds ratios (OR) and 95% confidence intervals (CI). Statistical significance was defined as a two‑sided P value < 0.05, and Benjamini–Hochberg false discovery rate (FDR) correction was applied for multiple comparisons.

Results

Patients’ Characteristics

This study included a total of 102 lung cancer patients with RET fusion. There were more female patients, accounting for 54.90% (56/102); 53.92% (55/102) of the patients were under the age of 60, with a median age of 60.5 years (ranging from 19 to 79 years); 27.45% (28/102) of the patients had a smoking history; the majority were in the advanced stage, accounting for 68.63% (70/102); surgical patients accounted for 24.51% (25/102); and distant metastasis was observed in 69.61% (71/102) of patients. The patients with the lowest physical condition score (Performance Status, PS score) were the most, accounting for 90.20% (92/102); the most common histological type was adenocarcinoma, accounting for 98.04% (100/102), and other histological types included 2 patients with sarcomatoid carcinoma.(Table 2).

Table 2 Clinical Characteristics of RET Fusion-Positive NSCLC Patients

Table 3 The Common and Rare Fusion Partners Characteristics of RET Fusion-Positive NSCLC Patients

Table 4 KIF5B-RET (N=51) and CCDC6-RET (N=23) of Clinical Characteristic

Metastatic sites were defined based on radiologic and pathological data documented in the patient records from the case system. Among the RET fusion-positive patients, 71 developed metastasis. As shown in Figure 1, lymph nodes were the most frequent metastatic site (n=53, 74.65%), followed by the pleura (n=33, 46.48%), bone (n=30, 42.25%), lung (n=28, 39.44%), brain (n=15, 21.13%), liver (n=9, 12.68%), and kidney (n=5, 7.04%).

Figure 1 Distribution of metastatic sites in RET fusion-positive NSCLC patients. This bar chart depicts the prevalence of distant metastatic sites in the cohort of RET fusion-positive NSCLC patients. Among 71 patients with confirmed distant metastasis, lymph node involvement was the most common (n=53, 74.6%), followed by pleural (n=33, 46.5%), bone (n=30, 42.3%), lung (n=28, 40.5%), brain (n=15, 21.2%), liver (n=9, 12.7%), and kidney (n=5, 7.0%) metastases.

Molecular Characteristic

To investigate RET fusion patterns, we performed ultra-deep targeted sequencing with six panels covering exons/introns of different genes. Among 102 RET fusion‑positive patients (1.6% fusion rate, 102/6204), a total of 134 RET fusion events were detected (Supplementary Table 1), 31 patients with multiple fusion events detected in a single sample (Supplementary Table 2). The predominant fusion partners were KIF5B (50.00%, 51/102) and CCDC6 (22.55%, 23/102). Other partners included NCOA4 and additional genes (Figure 2). Most RET breakpoints were in exon 12; others were exons 8, 9, 11, 13, 19. KIF5B’s main breakpoint was exon 15 (84.31%, 43/51), CCDC6’s exon 1 (91.30%, 21/23). Novel partners including C16orf95, CARNMT1-AS1, FAM107B,MYRFL and MTUS1 were identified (Table 1).

Figure 2 Distribution of RET fusion partners in RET fusion-positive NSCLC patients. The pie chart shows the proportional distribution of RET fusion partners identified in this study. KIF5B (38.06%) and CCDC6 (17.16%) were the most prevalent common fusion partners. Intergenic fusions constituted 20.15% of all detected events. A total of 24 rare and novel fusion partners were identified, with each rare/novel partner accounting for 0.75% to 1.49% of the cohort, including 5 previously unreported novel partners: C16orf95, CARNMT1-AS1, FAM107B, MTUS1, and MYRFL. Percentages were calculated based on the total number of RET fusion events (n=134) as multiple fusion events were detected in a single sample.

The RET Fusion Positivity in NSCLC is Often Associated with Age and PS Score

According to the description in the RETING study regarding the common fusion partners of RET fusion in NSCLC, they are KIF5B and CCDC6. We define KIF5B and CCDC6 as the common fusion partners. The other fusion breakpoints are considered as rare partners. We compared the occurrence of common fusion partners with that of rare fusion partners and investigated their association with those clinical factors. Through statistical analysis, it was found that the occurrence of KIF5B and CCDC6 was related to age (P = 0.023) and PS score (P = 0.040), but not related to gender, smoking history, clinical stage, metastasis history, and tissue subtype in Table 3. We compared the association between the two common fusion partners KIF5B and CCDC6 and those clinical factors, and found that their occurrence was related to smoking history (P = 0.037), but not related to gender, age, PS score, clinical stage, metastasis history, and tissue subtype (Table 4).

Accompanied by Genetic Mutation Characteristics

Among the 102 patients with RET fusion detected, 54 patients had other mutation forms coexisting with RET mutations (54/102, 52.94%). Among them, TP53 was the most common change (15/102, 14.71%), followed by EGFR (11/102, 10.78%), CDKN2A (7/102, 6.86%), KRAS (5/102, 4.90%), and SETD2 (3/102, 2.94%). Other genomic alterations were also present, including MYC, CDK4, MET, FGFR3, and PIK3CA, etc. Among the 93 patients with RET fusion, 19 (18.63%, 19/102) harbored concurrent driver gene alterations, including EGFR L858R (n=3), EGFR exon 19 deletion (n=2), KRAS G12X (n=2), and EML4-ALK (n=1), among others (Figure 3).

Figure 3 Genomic alteration landscape in RET fusion-positive NSCLC patients. The percentages on the right side indicate the overall alteration prevalence in the cohort: RET (100%), EGFR (11%), ALK (3%), KRAS (3%), MET (2%), ERBB2 (1%), ROS1 (1%), and BRAF (1%). Different alteration types are color-coded according to the legend, including missense mutations, small insertions/deletions (indel), copy number amplification (CN_amp), gene fusions (Fusion), and other variants (Others). This visualization comprehensively illustrates the co-occurrence patterns of driver genes, such as concurrent EGFR mutations, ALK rearrangements, and KRAS mutations, alongside RET fusions in this cohort.

Discussion

In this study, we retrospectively analyzed genetic data from 6,204 lung cancer patients and identified 102 with RET fusions, yielding a detection rate of 1.6%, consistent with the reported prevalence of 1%–2% in NSCLC.¹⁷ In terms of the discovery of RET fusion partners, at least 35 partner genes have been reported to fuse with RET, included ACBD5,AFAP1L2,AKAP13,BCR, CCDC6,CLIP1,CUX1,EML4,EPHA5,ERC1,FGFR1OP,FKBP15,FRMD4A,GOLGA5,HOOK3,KIAA1217,KIAA1468,KIF5B,KTN1,MYH13,NCOA4,PARD3,PCM1,PICALM,PPFIBP2,PRKAR1A,PRKG1,RFG9,RUFY2,SNRNP70,SPECC1L,SQSTTM1,TBL1XR1,TNIP1,TRIM24, TRIM27, TRIM33.¹⁸ Another study reported 11 partners, such as NCOA4, TSSK4, SORBS1, SIRT1, PTPRK, ADD3-AS1, PRKG1, IL2RA, CCNYL2, CCDC186, and ANKS1B.¹⁹ And in our cohort, we identified 28 distinct fusion partner types, especially reported multiple RET fusions that were not reported before, mainly including C16rf95, CARNMT1-AS1, FAM107B, MTUS1 and MYRFL.

The distribution of RET fusion partners varies considerably across published studies. The most frequently cited proportions of KIF5B-RET and CCDC6-RET are 83.6% and 15.1%, respectively, based on early comprehensive profiling studies.¹⁸ However, in our cohort KIF5B-RET and CCDC6-RET accounted for 50.0% and 22.5%, respectively. More recent large-scale studies have reported a wider range of KIF5B-RET frequencies. In a Chinese cohort of 380 RET fusion-positive patients, Wang et al reported KIF5B-RET in 51.1% of cases and CCDC6-RET in 23.4%^20. In the global LIBRETTO-001 trial, KIF5B-RET accounted for 61.9% of pretreated and 69.6% of treatment-naïve patients.²¹ Similarly, Sun et al reported a KIF5B-RET proportion of 65% in a Chinese cohort of 268 patients.¹⁰

The observed variation in fusion partner proportions across studies likely reflects a combination of selection effects and biological heterogeneity. Regarding selection effects, differences in detection methodologies may contribute to this variation. Studies employing targeted RNA sequencing or comprehensive DNA/RNA panels may capture a broader spectrum of fusion partners, potentially resulting in a relatively lower proportion of KIF5B-RET compared with studies using FISH or RT-PCR that focus on known common partners. Additionally, differences in case mix and referral patterns are important considerations. Single-center studies may reflect regional referral biases, whereas multicenter trials such as LIBRETTO-001 enroll patients from diverse geographic and clinical settings, potentially influencing the distribution of fusion subtypes.

Regarding biological heterogeneity, geographic and ethnic differences may also play a role. Notably, the KIF5B-RET proportion reported in Asian cohorts^10,20 and this study ranges from 51% to 65%, which is somewhat lower than the 70%–90% range cited in some Western-focused reviews.²² This raises the possibility of ethnic differences in the prevalence of specific RET fusion partners, although larger multi-ethnic comparative studies are needed to confirm this observation. Importantly, regardless of the variation in reported proportions, KIF5B-RET and CCDC6-RET consistently emerge as the two most common fusion partners across all studies, underscoring their central role in RET-driven NSCLC. Furthermore, previous studies¹⁴ have reported the fusion partners such as NCOA4, TSSK4, SORBS1, SIRT1, PTPRK, ADD3-AS1, PRKG1, IL2RA, CCNYL2, CCDC186, and ANKS1B. This study discovered multiple RET fusions that were not reported before, mainly including, C16rf95, CARNMT1-AS1, FAM107B, MTUS1 and MYRFL.

Multiple studies have explored the correlation between RET fusions and clinical demographics of lung cancer. Most studies have shown that RET fusions are more likely to occur in lung adenocarcinoma. In our study, we observed similar results, with 98.0% of RET fusion patients having a pathological type of lung adenocarcinoma. However, previous studies had differences in other factors such as gender and age. Michels’s report⁷ study showed that the proportion of RET fusion in the European cohort was higher in males (59% vs 41%). Tsuta K.’s study⁶ on Japanese patients indicated that RET fusion was not related to gender (p = 0.524). In our cohort, we observed that RET fusion tended to occur in female patients who were non-smokers (72.6% vs 26.4%) (54.9% vs 45.1%). Rui Wang.’s study⁴ showed that RET fusion occurred more frequently in younger populations, with a median age of 60 years. This study showed the same result, with a higher proportion occurring in people younger than 60 years (52.9% vs 47.1%), and a median age of 60.5 years. This may be due to differences in population race, lifestyle, environmental factors, or molecular heterogeneity. Therefore, further studies are needed to explore the potential relationships between RET fusion patients and these factors.

Brain metastasis is more common in patients with NSCLC with RET fusion positive. Multiple studies from different countries have shown that the frequency of brain metastasis ranges from 25% to 47.5%.^23–25 In this study, the metastatic sites more prone to occur in RET fusion positive patients are lymph nodes, pleura, bone, lung, brain, liver, and kidney, in descending order. The frequency of brain metastasis is 21.2%, slightly lower than that reported in various literatures. In the II phase clinical study of LIBRETTO-321,⁸ it was shown that the fusion partners of RET fusion positive patients that cause brain metastasis include KIF5B, CCDC6 and NCOA4. In this study, in addition to these, one novel types of fusion partners, MYRFL was also found to cause brain metastasis.

This study also analyzed the accompanying mutations of RET fusion patients, and found that TP53 is the most common accompanying mutation. At the same time, there are also accompanying mutations of EGFR and other driver genes. Whether RET fusion is mutually exclusive with other carcinogenic driving factors remains controversial. Recently, Qingsong Gao.²⁶ analyzed the fusion situations of 33 types of cancers and emphasized the exclusivity between fusion and mutations. However, Wang²⁷ reported that a unique mutation feature in Chinese NSCLC patients is an increase in EGFR mutation rate associated with RET and ALK gene fusion. In our study, concurrent EGFR mutations were identified in 11 RET fusion-positive patients. However, due to the retrospective nature of the study and the absence of longitudinal treatment history, we cannot determine whether these co-occurrences represent primary driver events, acquired resistance following EGFR TKI therapy, or sequencing artifacts. Co-occurrence of RET fusions and EGFR mutations has been reported in patients with acquired resistance to EGFR TKIs,²⁸ but the functional significance of such co-occurrences in treatment‑naïve patients requires further investigation.

Our study has limitations. Firstly, this study is a retrospective study and was completed in a single center., the data can not fully represent all the region. Larger sample sizes andand multi-center RET fusion data need to be collected and included to make our study more persuasive. In addition, since this study is a retrospective analysis based solely on DNA-based NGS testing, orthogonal validation at the RNA or FISH level was not performed and could not be conducted retrospectively. Another limitation is that Secondly, the treatment information of RET fusion positive NSCLC patients was not statistically analyzed. Lastly, further investigations into the molecular mechanisms that might explain these novel fusions were not performed. The prospective studies, large sample sizes and functional analysis are required in further study.In the next step, we will continue to improve this part.

Our study discovered 5 novel RET fusion aberrations using DNA-based NGS assay based upon the clinicopathologic and genomic features of NSCLC patients in a top tier cancer center located in northeast of China. In addition, we observed that TP53 was predominant co-mutation in both tissue and blood samples. The findings provide valuable insights into the genetic landscape of NSCLC and more evidence for regional differences or the inclusion of rare partners detected by NGS, which might be useful for clinical precise medicine.