Clinical Characteristics and Molecular Patterns of RET-Rearranged Lung Cancer in Chinese Patients

Abstract

RET rearrangement has been proven as an oncogenic driver in patients with lung cancer. However, the prevalence, clinical characteristics, molecular features, and therapeutic options in RET-rearranged patients remain unclear, especially in Chinese lung cancer patients. We retrospectively collected 6,125 Chinese lung cancer patients who have been profiled using next-generation sequencing (NGS). The clinical demographics and molecular features of RET rearrangement-positive patients were analyzed. RET rearrangements were identified in 84 patients with a proportion of 1.4% in our cohort. The median age at diagnosis was 58 years, and it mainly occurred in females with adenocarcinoma histology. KIF5B-RET was the most frequent fusion type and accounted for 53.8% (57/106) of all RET fusions identified, with K15-R12 as the most frequent variant (71.9%). Among 47 RET ⁺ patients profiled with larger panels, 72.3% (34/47) harbored concurrent alterations. TP53 ranked as the most common concurrent alteration, and concomitant EGFR oncogenic alterations were identified in seven patients. Moreover, an adenocarcinoma patient harboring concurrent RET fusion and EGFR L858R responded to combinatorial treatment of cabozantinib and osimertinib, with a progression-free survival of 5 months. Our study improved knowledge of clinical characteristics and molecular features of RET-rearranged lung cancers in China. It might be helpful to guide clinicians for more effective personalized diagnostic and therapeutic approaches.

INTRODUCTION

The proto-oncogene gene RET encodes a receptor tyrosine kinase that can activate downstream pathways such as MAPK/ERK, PI3K/AKT, and JAK/STAT¹. RET plays a critical role in cell proliferation, migration, and differentiation^2–5. RET rearrangement was first identified in NIH-3T3 cells transfected with lymphoma DNA in 1985⁶, and chromosomal rearrangement could lead to constitutive activation of RET kinase and downstream signaling events, which cause tumorigenesis.

Fusion of RET more frequently occurs in radiation-induced papillary thyroid cancer^7,8. The prevalence of RET rearrangement is 0.7%–2% in lung cancer and 1%–2% in non-small cell lung cancer (NSCLC)⁹. To date, several fusion partners have been identified, such as KIF5B, CCDC6, NCOA4, and TRIM33 in lung cancer, with KIF5B-RET fusion accounting for the major proportion¹⁰. The coiled-coil domain of RET partner gene KIF5B can activate RET tyrosine kinase domain by ligand-independent homodimerization and autophosphorylation, leading to the constitutive activation of downstream pathways and tumorigenesis¹¹.

Investigation of the prevalence, clinical demographics, and molecular pattern of oncogenic rearrangements may provide comprehensive genomic profiling as well as aiding in the selection of patients for optimal therapies. However, results from studies of RET rearrangements in lung cancer are still inconclusive, especially in Chinese patients. Moreover, previous reports about the prevalence and clinical characteristics are conflicting. Lin et al. reported that RET rearrangements were more prone to occur in younger age, never-smokers, females with adenocarcinoma in lung cancer¹². Some studies revealed that there was no statistically significant difference in gender and smoking status, or even drew an opposite conclusion in terms of gender^13,14. Therefore, we retrospectively analyzed 6,125 Chinese lung cancer patients for RET rearrangement using next-generation sequencing and identified 84 RET fusion-positive patients. This study demonstrated clinical demographics and molecular features of RET rearrangement in Chinese lung cancer patients.

MATERIALS AND METHODS

Patient Information

A total of 6,125 samples (either tissue or plasma) from lung cancer patients were consecutively collected from September 2015 to July 2017 in this retrospective RET rearrangement study. There was no preselection with smoking, gender, clinical stage, or age of patients. Eligible patients were histologically diagnosed as lung cancer according to the latest World Health Organization Criteria. Cancer stage was evaluated based on the 7th TNM classification. RET rearrangements were identified using next-generation sequencing. Our profiling panels (Burning Rock Biotech, Guangzhou, P.R. China), consisting of 8, 56, 168, or 295 cancer-related genes, were designed and validated for identification of base substitutions, insertions, deletions, copy number variations, and gene fusion. This entire study was approved by the institutional review board of Union Hospital, Tongji Medical College.

Preparation of Tissue DNA and Plasma Cell-Free DNA

Tissue DNA and plasma cell-free DNA were extracted using QIAamp DNA formalin-fixed paraffin-embedded (FFPE) tissue kit (Qiagen, Valencia, CA, USA) and QIAamp Circulating Nucleic Acid kit (Qiagen), respectively, following the manufacturer’s instructions.

NGS Library Preparation and Sequencing

DNA shearing was performed using Covaris M220 (Covaris Inc., Woburn, MA, USA) and followed by end repair, phosphorylation, and adaptor ligation. Then 200- to 400-bp fragments were selected by bead (Agencourt AMPure XP Kit; Beckman Coulter, Brea, CA, USA) and hybridized with capture probes baits (SureSelectXT Custom 1kb-499kb; Agilent, Santa Clara, CA, USA). Hybrid selection was performed using magnetic beads (Dynabeads™ MyOne™ Streptavidin T1; Thermo Fisher Scientific, Waltham, MA, USA) and followed by PCR amplification. Bioanalyzer (LabChip GX Touch 24 Nucleic Acid Analyzer; Perkin-Elmer, Waltham, MA, USA) was used to evaluate DNA quality and size by high-sensitivity DNA assay. Indexed samples were sequenced on NextSeq 500 (Illumina, Inc., San Diego, CA, USA).

NGS Data Analysis

All the reads were mapped to the human genome (hg19) with Burrows–Wheeler Aligner (BWA)¹⁵. Local alignment optimization, mark duplication, and variant calling were performed using Genome Analysis ToolKit (GATK) 3.2¹⁶, picards, and VarScan¹⁷. Gene rearrangements were called with FACTERA¹⁸, and CNVs were analyzed based on sequencing depth. Variants were filtered using the VarScan fpfilter pipeline, and loci with depth less than 100 were filtered out. At least two and five supporting reads were needed for INDELs, while eight supporting reads were needed to call SNVs, in both plasma and tissue samples. According to the ExAC, 1000 Genomes, dbSNP, ESP6500SI-V2 database, variants with population frequency over 0.1% were grouped as SNP and excluded from further analysis. Remaining variants were annotated with ANNOVAR¹⁹ and SnpEff v3.6²⁰.

RESULTS

Patient Characteristics

To interrogate RET fusion patterns, we retrospectively screened 6,125 NSCLC patients from September 2015 to July 2017 and identified 84 patients harboring RET fusion, giving an overall frequency of 1.4%. Thirty-six (42.9%) of them were males and 47 (56.0%) were females. Gender information of the remaining patient (1.2%) was not recorded. Compared to patients in the original cohort (n = 6,125; 3,411 males, 2,578 females, and 136 unknown), the prevalence of females was significantly higher in the RET ⁺ cohort (n = 84, p = 0.0228, Pearson’s chi-squared test). Median age of patients harboring RET fusion was 58 years, ranging from 35 to 81 years. Compared to patient age in the 6,125 original cohort (median age = 61 years), no preference pattern in terms of age was found in the RET ⁺ cohort. As to histological subtype, a majority of them (62/84, 73.8%) were diagnosed as lung adenocarcinoma. One patient was diagnosed with lung squamous cell carcinoma, and three patients had a mixture of adenocarcinoma and squamous cell carcinoma. Histological information of the other 18 lung cancer patients was not recorded. The detailed patient characteristics are summarized in Table 1.

Table 1.

Summary of Baseline Characteristics of Patients Harboring RET Rearrangement (N = 84)

Patient Characteristics	n (%)
Gender
Male	36 (42.9%)
Female	47 (56.0%)
Unknown	1 (1.2%)
Age (years)
Median	58
Range	35–81
Histological types
LUAD	62 (72.8%)
LUSC	1 (1.2%)
Mixed LUAD and LUSC	3 (3.6%)
Other lung cancers	18 (21.4%)

LUAD, lung adenocarcinoma; LUSC, lung squamous cell carcinoma.

Overview of RET Fusion Patterns

To interrogate RET fusion patterns, we performed ultra-deep targeted sequencing on plasma using panels covering critical exons and introns of lung cancer-related genes. Among them, 32 patients used a panel consisting of 168 lung cancer-related genes, spanning 170 kb of human genome; 10 patients used a 56-gene panel, and the 37 patients used a panel consisting of 7 well-known lung cancer driver genes plus KRAS, a well-established prognostic factor. Five patients used a panel consisting of 295 cancer-related genes.

In this cohort, a total of 106 RET rearrangements in 84 patients were identified with RET rearrangement. Structures of all the RET rearrangements with detailed demonstration of breakpoints are presented in Figure 1. The most commonly seen partner was KIF5B, and 57 KIF5B-RET fusion events were identified with a frequency of 53.8% (57/106) in all RET rearrangements and in 67.9% (57/84) patients (Fig. 2). The most frequent variant of KIF5B-RET was K15-R12, occurring in 71.9% (41/57) of KIF5B-RET ⁺ patients (Fig. 1). This result was in agreement with previous literature, which reported the occurring frequency of about 75%^11,21–23.

Figure 1.

Structure and breakpoints of 106 RET fusions detected in 84 lung cancer patients by next-generation sequencing.

Figure 2.

Distribution of various RET rearrangement partners identified in the 84 lung cancer patients. Different colors and sizes indicate the occurrence frequency of each RET fusion partner in the overall RET fusion identified (n = 106).

The second- and third-ranked fusion partners were CCDC6 and NCOA4, occurring in 17.0% (18/106) and 2.8% (3/106) of all RET fusions, and in 21.4% (18/84) and 3.6% (3/84) of patients, respectively. Several rare and novel RET fusion partners were identified in our study, including TSSK4, SORBS1, SIRT1, PTPRK, ADD3-AS1, PRKG1, IL2RA, CCNYL2, CCDC186, and ANKS1B. Table 2 lists detailed breakpoints and patient histology of novel RET fusions identified in this study, which have not been reported before to the best of our knowledge.

Table 2.

Novel RET Fusion Partners Identified in Chinese Lung Cancer Patients

Fusion	Breakpoints	Histology
TSSK4–RET	Intron1_Intron11	Adenocarcinoma
SORBS1–RET	Intron8_Intron11	Adenocarcinoma
SIRT1–RET	Intron8_Intron11	Adenocarcinoma
PTPRK–RET	Intron3_Intron11	Adenocarcinoma
ADD3–AS1–RET	Intron1_Intron11	Adenocarcinoma
PRKG1–RET	Intron7_Intron11	Adenocarcinoma
IL2RA–RET	Intergenic_Intron11	Adenocarcinoma
CCNYL2–RET	Intron6_Intron15	Not available
CCDC186–RET	Intron10_Intron11	Adenocarcinoma
ANKS1B–RET	Intron1_Intron11	Adenocarcinoma

In addition, we observed that multiple RET fusions can be detected in individual patients, and 21 patients harbored more than one RET fusion. Most studies revealed that for an individual patient harboring multiple fusions, the hotspot partner commonly serves as driver mutation²⁴.

Concurrent Genomic Alterations in RET Fusion-Positive Patients

Since the discovery of RET fusion, there has been no definitive conclusion about the mutual exclusivity between RET fusion and other genomic alterations^14,25,26. We interrogated the mutation spectrum of lung cancer patients with positive RET fusions. Patients tested with the eight-gene panel were excluded from this analysis due to the rare chance of harboring dual drivers. Among 47 patients tested with larger panels, 34 of them (72.3%) harbored concurrent mutations, with TP53 being the most commonly seen concurrent mutation, occurring in 42.5% (20/47) of patients. It was followed by EGFR and MYC, occurring in 14.9% (7/47) and 10.6% (5/47) of patients, respectively (Fig. 3). The underlying mechanisms of RET co-occurrence with other mutations and influence on clinical outcomes are needed to be addressed in further studies.

Figure 3.

Concurrent genetic alternation analysis demonstrated by oncoPrint. The top bar indicates the number of mutation in each patient. The right-side bar demonstrates the number of patient harboring a specific mutation. Different colors mean different mutation type categories.

It has been regarded that actionable driver mutations were commonly mutually exclusive^27–29. However, the coalteration of EGFR and other driver mutations such as ALK in a subset of lung cancers has been reported and challenged previous dogma^30–32. In this cohort, the co-occurrence of RET fusion with EGFR oncogenic genetic alterations was observed in seven patients, consisting of five exon 19 deletions, two L858R mutations, and two T790M mutations. All seven patients had received previous treatment before the positive detection of RET and EGFR alterations. One case received previous chemotherapy, and the other six cases received previous EGFR-TKI treatment (including the two T790M⁺ patient), indicating that RET fusion maybe one of the mechanisms that contributes to resistance of EGFR TKI.

Interestingly, we found that no KIF5B-RET was identified in the seven patients who harbored EGFR mutations. The mutual exclusivity of KIF5B-RET and EGFR alterations suggested that KIF5B-RET was a strong driver mutation. For the seven patients harboring concurrent EGFR and non-KIF5B-RET fusion, the fusion types of RET included CCNYL2-RET (n = 1), PCM1-RET (n = 1), CCDC6-RET (n = 3), and NCOA4-RET (n = 2). Among them, sequencing samples of six patients were plasma. We observed that overall allelic fraction (AF) ratio of first-generation EGFR-TKI sensitizing mutations was higher than non-KIF5B-RET in each of the six patients (AF ratio of EGFR/RET = 2.4) (Fig. 4), indicating that non-KIF5B-RET fusion might function as a potential acquired resistance mechanism to EGFR tyrosine kinase inhibitors. The RET rearrangement may exist as a minor clone with EGFR-sensitive alterations and expanded while the EGFR-sensitive alterations were inhibited by EGFR-TKI.

Figure 4.

Relative allelic fraction of EGFR in seven RET-rearranged patients. The y-axis indicates EGFR relative allelic fraction that was normalized to RET fusion in each individual patient.

Clinical Outcomes of an EGFR and RET Fusion Concurrent Patient Treated With Cabozantinib

Several tyrosine kinase inhibitors (TKIs), such as vandetanib, cabozantinib, and sunitinib, have been proven with anti-RET activity. Cabozantinib inhibits a broad range of tyrosine kinases and displayed a 28% ORR, and median PFS and OS of 5.5 months and 9.9 months, respectively, for patients with advanced RET-rearranged NSCLC³³.

In our cohort, clinical outcomes of cabozantinib were available for six RET-rearranged patients. The median treatment period of cabozantinib for these patients was 5 months. Among them, one patient was identified harboring concurrent RET fusion and EGFR mutation. It was a 65-year-old male patient diagnosed with stage IV lung adenocarcinoma with bone metastasis. He was treated with gefitinib and osimertinib as the first- and second-line therapy, and achieved stable disease after the two lines of treatment, with progression-free survival of 8 months and 4 months, respectively. After development of resistance to osimertinib, NGS revealed that this patient had concurrent CCNYL2-RET fusion and EGFR L858R. Combination of osimertinib and cabozantinib was used to treat this patient after the positive detection of the two concurrent alterations. He achieved stable disease (SD) with a tumor shrinkage of 13% 1 month after treatment initiation. Finally, he experienced disease progression after a PFS of 5 months.

DISCUSSION

In this study, we retrospectively analyzed molecular profiling data of 6,125 Chinese lung cancer patients and identified 84 patients harboring RET rearrangements, accounting for 1.4% of this cohort. This ratio was in agreement with a previous study about RET fusion²⁴. We investigated the distinct clinical characteristics of RET fusion patients. Furthermore, we analyzed the fusion partners, demonstrated their molecular pattern, and investigated the mutual exclusivity of RET fusion with other concomitant gene alterations.

Several studies have investigated the correlation of RET fusion and clinical demographics in lung cancer, and most of them revealed that RET rearrangements were more prone to occur in lung adenocarcinomas.³⁴ In our study, we observed similar results that most of RET ⁺ patients were lung adenocarcinomas. However, there were discrepancies among previous studies about other factors such as gender and age. Michels et al. reported that rearrangements of RET occurred with a high proportion of men (59% vs. 41%) and median age of 62 years in a European cohort¹⁴. Another study carried out in Japanese patients revealed that RET fusion was not associated with gender (p = 0.524) but significantly correlated with younger age (57.5 years)¹³. A meta-analysis reported that RET fusions were identified at significantly high frequency in younger (<60 years) females¹². In our Chinese cohort, we observed that RET rearrangement had a tendency to occur in females with a median age of 58 years at diagnosis. One explanation is that this discrepancy may have resulted from differences in ethnicity, lifestyles, environmental factors, or molecular heterogeneity. Therefore, studies are needed to further investigate the underlying relationship of RET fusion patients and these factors.

KIF5B was the most frequently appeared partner of RET fusion and accounted for 53.8% of all the RET fusions identified in our study. For KIF5B-RET, the most common variant was K15-R12, occurring in 71.9% of all KIF5B-RET fusions. The occurring frequencies in this study was similar with previous literature^11,21–23. However, no consistent conclusion was obtained in terms of clinical outcomes to RET inhibitors across different RET fusion types¹⁰. Further studies need to address this issue to provide more guidance for patients at high risks so that optimal treatment strategy may be implemented.

The mutual exclusivity of RET fusion and other molecular alterations has been poorly investigated, and no definitive conclusion has been obtained to date. Some studies reported that RET fusion was exclusive with other gene alterations. However, Song et al. reported several concomitant genomic alterations, including EGFR, MAP2K1, CTNNB1, and AKT1, occurred in 4 of 11 RET-rearranged NSCLC patients, with a frequency of 36.4%.²⁵ Similar co-occurrence frequency of genetic alterations was also found in another study in 10 of 22 RET ⁺ patients (45.5%), consisting of eight TP53, one MET amplification, one CTNNB11, and one EGFR rare mutation L833F¹⁴. In our study, we also identified several RET fusion co-occurring mutations, and EGFR oncogenic driver mutation was identified in seven RET ⁺ patients. It was reported that patients harboring both RET fusion and EGFR mutation were resistant to EGFR TKIs^26,35,36, which suggested that RET fusion maybe one of the mechanisms that contributed to resistance of EGFR TKI. This study provided the basis for the hypothesis that an actionable driver mutation could function as an acquired resistant mechanism for another actionable driver alteration.

To date, chemotherapy is still the standard first-line treatment for RET-rearranged patients. Several RET inhibitors have been developed, but the overall outcomes to RET inhibitors were inferior to targeted therapies in other lung cancer oncogenic mutations like ALK and ROS1. Cabozantinib displayed an overall response rate (ORR) of 28% and median PFS of 5.5 months in a phase II clinical trial of RET-rearranged lung cancers (n = 26)³³. The ORR to vandetanib was reported as 18% (n = 19) and to lenvatinib was 16% (n = 25) in Korean patients. Several reasons are attributed to less sensitivity of RET inhibitors. One reason was that the RET inhibitors could lead to toxicities due to the activity to VEFGR kinase; thus, the clinical uses of these inhibitors were often reduced to 70% at a suboptimal dose³³. Therefore, more potent and selective RET inhibitors that do not target VEFGR are needed to increase the sensitivity and reduce the off-target toxicity. Moreover, combinatorial treatment strategy is another approach to be taken into account.

There were still some limitations in this study. First, we only analyzed the prevalence of RET fusion in terms of gender, age, and histology, and other clinical characteristics like smoking status, tumor size, and metastatic status were not included due to incompleteness of clinical records. Second, owing to the lack of response and survival information, we did not perform survival analysis to interrogate the clinical outcomes across different RET variants. Last, but not the least, further validations are needed to support the hypothesis that non-KIF5B-RET may serve as an acquired resistance mechanism for another driver mutation.

The development of next-generation sequencing greatly improved the molecular diagnosis of cancer. The knowledge of specific clinical features associated with RET fusion can guide patients at high risk for precise diagnosis and treatment strategy. We further propose that studies to be carried out include more clinical feature analysis in RET fusion-positive patients and prognostic prediction evaluation among different RET fusions variants.