Genomic landscape of 891 RET fusions detected across diverse solid tumor types

In this study, we report the clinicopathologic and genomic profiles of 891 patients with RET fusion driven advanced solid tumors. All patient samples were tested using a tissue-based DNA hybrid capture next generation sequencing (NGS) assay and a subset of the samples were liquid biopsies tested using a liquid-based hybrid capture NGS assay. RET fusions were found in 523 patients with NSCLC and in 368 patients with other solid tumors. The two tumor types with the highest number of RET fusion were lung adenocarcinoma and thyroid papillary carcinoma, and they had a prevalence rate 1.14% (455/39,922) and 9.09% (109/1199), respectively. A total of 61 novel fusions were discovered in this pan-tumor cohort. The concordance of RET fusion detection across tumor types among tissue and liquid-based NGS was 100% (8/8) in patients with greater than 1% composite tumor fraction (cTF). Herein, we present the clinicopathologic and genomic landscape of a large cohort of RET fusion positive tumors and we observed that liquid biopsy-based NGS is highly sensitive for RET fusions at cTF ≥1%.


INTRODUCTION
Rearranged during transfection (RET), located near the centromere on the long arm of chromosome 10 (10q11. 21), is a protooncogene that encodes for a single-pass transmembrane glycoprotein receptor tyrosine kinase (RTK) 1 . RET plays a vital role in the embryonic development of the human enteric nervous system and genitourinary tract and is essential for the normal development of cells [2][3][4] . Somatic RET gene alterations, including short variants and fusions, act as pathogenic driver alterations in approximately 2% of solid tumors. RET fusions occur in approximately 1% of non-small cell lung cancer (NSCLC) cases and are generally mutually exclusive to other primary driver variants and rearrangements. RET fusion-positive lung adenocarcinomas are associated with poor differentiation, solid sub-type, and smaller T stage (≤3 cm) with N2 disease 5 . RET fusion-positive NSCLC represents a rare, but clinically actionable, driver alteration class of tumor 6,7 . In addition, RET alterations play an essential role in thyroid cancer initiation and progression. RET fusions occur in approximately 10% of papillary thyroid carcinomas. While RET short variant mutations are pathognomonic of 98% hereditary and 50% of sporadic medullary thyroid cancers (MTC), they are rarely reported in other tumor types.
Early attempts at RET-targeted precision therapy relied on multikinase inhibitors such as cabozantinib, vandetanib, and lenvatinib [8][9][10][11][12] . A combination of modest clinical activities with ORR, mPFS, and mOS ranging from 16%-47%, 4.5-7.3 months, and 9.9-11.6 months, and significant toxicity from off-target activities dampened enthusiasm for further development. More recently a new class of RET-selective inhibitors (selpercatinib and pralsetinib) that potently inhibit both wild-type and RET-activated (both point mutations and fusions) tumors 13 . Selpercatinib received its accelerated approval by the FDA in May 2020 for metastatic RET-fusion+ NSCLC and papillary thyroid cancers and RET-mutant medullary thyroid cancer, based on the LIBRETTO-001 trial evaluating its activity in RET + advanced solid tumors [14][15][16] .
Similarly, pralsetinib accelerated approval for RET fusion-positive NSCLC came in September 2020 and in December 2020 for thyroid cancer based on the ARROW trial [17][18][19] . Furthermore, in August 2022, Subbiah et al., published their study examining the pancancer efficacy of pralsetnib in patients with RET fusions from the phase 1/2 ARROW trial 20 . Here, they observed an overall response rate of 57% (study cohort of 29 patients across 12 tumor types) and concluded that responses were observed regardless of tumor types in their study cohort. Most recently (October 2022), in a tumor-agnostic population (n = 41, LIBRETTO-001 trial), meaningful clinical activity (objective response rate was 43.9%) in the RET fusion positive cohort was shown 21 .
Currently there is a limited understanding in the genomic landscape of NSCLC and other solid tumors harboring RET fusions. Numerous prior studies on RET gene alterations are constrained by a small sample size where observed distribution frequencies have not reached statistical significance [22][23][24][25][26][27][28][29] . Here, we report the comprehensive molecular portfolio of RET-altered cancers among 523 patients with NSCLC and 368 patients with other solid tumors (excluding NSCLC). Our study describes the clinicopathologic and genomic features of RET fusion-positive and negative cohorts among NSCLC and other solid tumors.
The NSCLC RET fusion-positive cohort was significantly younger (median age = 64 vs 68; P < 0.001), had a higher female:male ratio (1.27 vs 1.02; P = 0.012) and had a higher frequency of specimens obtained from metastatic sites vs nonmetastatic sites (52% vs 43%; P = 0.002) when compared to the NSCLC RET fusion-negative cohort ( Table 2). In the NSCLC cohort, patients with Central and South American, East Asian, and South Asian ancestry were more highly represented in the RET fusions-positive subset vs the RET fusion-negative subset (8.6% vs 5.1%, 11% vs 4.3%, 1.9% vs 0.6%, respectively, P < 0.001). Lastly, there was a significant decrease in the tobacco smoking mutational signature in the NSCLC RET fusion-positive cohort when compared to the NSCLC RET fusionnegative cohort (0.6% vs 13%; P < 0.001) 30 .
For the other solid tumors RET fusion-positive cohort, we used NSCLC RET fusion positive cohort for inter-cohort comparison. Among 368 other solid tumors RET fusion-positive cases patients were significantly younger compared to NSCLC RET fusion-positive cases (median age = 61 vs 64; P < 0.001). In comparison to the NSCLC RET fusion-positive cohort, the other solid tumors RET fusion-positive cohort had a significantly higher prevalence of Central and South American and East Asian (8.6% vs 16%, 11% vs 3.8%, respectively, P < 0.01; Table 1). In addition, we examined the RET fusion positive to the RET fusion negative papillary thyroid carcinoma (PTC) cohorts and found that the age of the RET fusion positive PTC cohort was significantly younger then the RET fusion negative PTC cohort (33 vs 62 years old <0.001) (Supplementary Table 3). Lastly, we saw a trend in the different direction for colon adenocarcinoma, although a smaller absolute value difference (66 vs 60 years old, p < 0.001) (Supplementary Table 4).   Table 2). The prevalence of RET fusions also varied in subtypes of b NSCLC and c thyroid carcinoma. The two tumor types with the highest number of RET fusion were lung adenocarcinoma and thyroid papillary carcinoma and they had a prevalence rate 1.14% and 9.09%, respectively. NOS not otherwise specified. Number on top of bar indicates total number of RET fusion cases.
V Parimi et al.
On the other hand, more than half of the other solid tumors cohort was composed of RET fusions with gene fusion partners NCOA4 (32.6%) and CCDC6 (29.9%). Of note, the most common fusion in papillary thyroid carcinoma was CCDC6-RET and NCOA4-RET (41.3% and 35.8%, respectively). KIF5B-RET fusions were highly specific for NSCLC compared to other solid tumors (66% vs 6.3%; P < 0.001). In contrast NCOA4-RET (32.6% vs 2.9%, P < 0.001) and CCDC6-RET (30% vs 18.2%; P = 0.002) fusions was frequently seen among other solid tumors. In addition, 61 novel RET gene fusion partners were identified across both cohorts of patients (Table 3  and Supplementary Table 5). We examined the primary break point regions in RET gene fusions. Among all the RET fusion-positive cohorts, the RET gene breakpoints were mainly clustered in the intron 11 (87%) followed by intron 10 (5%) and exon 11 (4.8%) (Fig. 2). There is no significant difference in the distribution of RET break point regions between NSCLC and other solid tumors. Similarly, in liquid biopsies, RET gene breakpoints were mainly clustered in intron 11 in both the NSCLC and other solid tumors RET fusion-positive cohorts ( Supplementary Fig. 2).

Prevalence of tissue and liquid RET fusions detection
Among 891 total RET fusion positive cases, twenty-three cases were also tested with a liquid NGS assay. The median interval between specimen collection was 75 days, of which 10 out of 23 (43.5%) patients had a liquid assay performed after initial tissuebased NGS assay, 11 out of 23 (47.8%) had liquid assay as a primary comprehensive molecular NGS assay followed by tissuebased NGS and 2 of 23 (8.7%) had liquid and solid-based NGS at the same time point. Among 23 tissue RET fusion positive patients [20 (lung adenocarcinoma/NSCLC/lung cancer-NOS), 2 (carcinoma of unknown primary), 1 (prostatic adenocarcinoma)] with both tissue and liquid NGS, 14 (61%) patients had RET fusion detected on liquid assay. The concordance of tissue and liquid testing stratified by cTF is as follows: 100% among 2 patients with greater than 10% cTF, 100% among 8 patients greater than 1% cTF, and 40% (6) among 15 patients with less than 1% cTF (Fig. 4). Of the cases with cTF <1% and with detection of RET fusions with a liquid biopsy, the lowest cTF value was 0.27%. Lastly, among 9 patients with RET fusion positive tissue NGS but negative on liquid (all cTF <1%), 6 patients had no known somatic gene alterations detected in liquid and 3 patients had gene alterations (TP53, BRCA1, JAK2 and CHEK2) with less than 0.5% variant allele frequency in liquid consistent with cTF <1%.

DISCUSSION
To the best of our knowledge, this study represents the largest single cohort of patients with RET fusion-positive solid tumors characterized by a DNA hybrid capture-based molecular assay. We observed that with a well-designed DNA based assay, the sensitivity of detecting RET fusions is comparable with the prevalence rates of The Cancer Genome Atlas (TCGA) network, which used a variety of molecular profiling techniques including: exome and whole genome DNA sequencing, RNA sequencing, miRNA sequencing, SNP arrays, DNA methylation arrays, and reverse phase protein arrays 31 . Specifically, the TCGA network yielded RET fusion positive PTC samples (n = 496) at 6.8%, whereas the RET fusion prevalence in our study was 9% (1208 sequenced PTC samples) 31 . Similarly, comprehensive molecular profiling of 229 lung adenocarcinoma by TCGA showed 2 samples (0.87%) with RET fusion in comparison to 1.14% RET fusion-positive among 39,922 lung adenocarcinomas in our study 26 . In addition, to the best of our knowledge, we have reported 61 novel RET fusions with intergenic or intragenic gene partners not yet reported in the literature 32,33 . These data point to the high sensitivity of detecting RET fusions by a well-designed DNA assay.
We defined the clinicopathologic and genomic landscape of this large cohort of tumors driven by RET fusions. Consistent with our findings, multiple studies have suggested that RET fusions in lung cancer correlate with adenocarcinoma histology, younger age, never smoker status and advanced disease 5,34,35 . A novel observation in the RET fusion-positive NSCLC patients was an enrichment of Central and South American, East Asian, and South Asian patients when compared to the RET fusion-negative NSCLC cohort. As expected, we observed that like other driver gene fusions (e.g. ALK and ROS1), the majority of RET fusions are mutually exclusive with other primary driver alterations and the distribution of most common RET fusion partners [KIF5B 66%, CCDC6 18%, and others 16%] in NSCLC is similar to the existing RET registry studies on RET alterations 8 . Our findings also indicate no differences in the RET fusion breakpoints among NSCLC and other solid tumors. As previously described in RET fusion-positive CRC and similar to other fusions involving ALK, ROS1, and NTRK, we saw a high rate of MSI-H in patients with CRC driven by RET fusions 36,37 . Lastly, TMB is lower but the PD-L1 expression trended higher in the NSCLC RET fusion-positive cohort when compared to the NSCLC RET fusion negative cohort, which suggests for further efficacy evaluation of ICPI in RET fusion positive NSCLC.
Recently, liquid biopsy has emerged as an important tool for genomic profiling to guide clinical management of advanced NSCLC and other solid tumors. Studies describing somatic RET alterations detected using liquid NGS assays are rare 38 . In this context, the liquid assay utilized in this study detected 100% (8/8) of RET fusions among RET fusion-positive patients by tissue testing with cTF ≥1% and 40% (6/15) among RET fusion-positive patients by tissue testing with cTF <1%. While the number of patients with paired tissue and liquid testing was limited, this data suggests that when cTF is ≥1%, liquid biopsy can reliably detect RET fusions, and that when cTF is <1% RET fusion detection is still possible but negative results are less reliable. In this cohort, we detected a RET fusion in a case with a cTF value as low as 0.27%, suggesting that the assay was able to detect fusion even with very low amount of tumor shed.
This study has a few limitations. First, although this is the largest study to date to analyze co-occurring genomic alterations among RET-positive solid tumors, the cohort lacks full clinical annotation including therapeutic and systematic clinical follow-up information, stage of disease, smoking status, and reported race (though we infer the smoking status with the tobacco signature and race through the genetic ancestry of the patients). With additional clinical data, we could better characterize the efficacy of RETinhibitors for various RET fusions, especially the novel fusions discovered. Furthermore, a small proportion of patients with RET fusion-positive NSCLC were also found to have other driver alterations, such as EGFR and KRAS. However, acquired RET fusions have been described as a mechanism of resistance to targeted therapies, such as EGFR inhibitors and without complete clinical annotation, it is difficult to determine if these were de novo alterations or acquired in the setting of targeted therapy. In addition, this RET fusion-positive study cohort is representative of primarily clinically advanced solid tumors and may not be representative of tumors in other clinical settings.
In conclusion, we present the clinicopathologic and genomic landscape of a large cohort of RET fusion positive tumors, including the discovery of 61 novel fusions, detected by a DNA tissue-based NGS assay. In addition, we observed that liquid biopsy-based NGS is highly sensitive for RET fusions at cTF ≥1%.

Patient cohort
A review of the Foundation Medicine research database was performed on patients that were tested with FoundationOne® or FoundationOne®CDx assays between August 2014 and December 2020 to review all patients whose tumor tissue harbored RET fusions. In addition, we queried the database to examine all patients tested with FoundationOne®LiquidCDx with RET fusions detected by tissue biopsy between August 2020-December 2021. This study was approved by the Western Institutional Review Board Protocol (No. 20152817) and the IRB granted a waiver of informed consent under 45 CFR § 46.116 based on review and determination that this research meets the following requirements: (i) the research involves no more than minimal risk to the subjects; (ii) the research could not practicably be carried out without the requested waiver; (iii) the waiver will not adversely affect the rights and welfare of the subjects. All patient cases in this study were sent to Foundation Medicine Inc. for comprehensive genomic profiling (CGP) during routine clinical care. Manual review of accompanying pathology reports was performed to extract demographic information of the patients and site of specimen.
Tissue DNA sequencing assay FoundationOne®CDx/FoundationOne® are tissue-based next generation sequencing (NGS) assays that uses a hybrid capture methodology and is performed in a Clinical Laboratory Improvement Amendments (CLIA)-certified and College of American Pathologists (CAP)-accredited laboratory (Foundation Medicine, Cambridge, MA and Morrisville, NC). FoundationOne®CDx/Foun-dationOne® detects base substitutions, insertion and deletion alterations (indels), and copy number alterations (CNAs) in up to 324 genes and select gene rearrangements as previously described 39 . Each sample is reviewed by a board-certified pathologist to assessed for % tumor nuclei/tumor volume adequacy and to assign a diagnosis to the sample. As previously described, the tumor mutational burden (TMB) was determined on up to 1.1 Mb of sequenced DNA and assessment of microsatellite instability (MSI) 40 was performed from DNA sequencing on up to 114 loci 41,42 . TMB-high (H) was defined as ≥10 mutations/ Megabase for the purposes of this study. As research use only, tobacco mutational signature was called as described by Zehir et al. 30 , and genetic ancestry was assessed to be of predominately African, European, Central and South American, South Asian, or East Asian genetic ancestry as previously described 43 .
Liquid DNA sequencing assay FoundationOne®LiquidCDx is a liquid biopsy CGP assay that utilizes a hybrid capture methodology and is performed in a CLIAcertified and CAP-accredited laboratory (Foundation Medicine, Cambridge, MA). Similar to FoundationOne®CDx/FoundationOne®, FoundationOne®LiquidCDx detects base substitutions, insertion, and deletion alterations (indels), and copy number alterations (CNAs) in up to 324 genes and select gene rearrangements 44 . An investigational composite tumor fraction (cTF), which merges two methods for estimation of tumor fraction (TF) was utilized as previously described [45][46][47] .

RET fusion case selection
For this study, we included all RET fusions as detected by the FoundationOne®CDx/FoundationOne® and FoundationOne Liquid CDx assays (Foundation Medicine, Inc., Cambridge, MA). All rearrangements without a fusion partner were excluded from the analysis and only cases where the kinase domain of RET was preserved were included in this study.

Statistical analysis
We explored the clinical, pathologic, biomarker, and genomic differences between the different cohorts using Fisher's exact test or χ 2 test for categorical variables and Wilcox rank-sum test for continuous variables. P-value was adjusted for multiple comparisons using the Bonferroni method and p < 0.05 was considered significant 49 .

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.

DATA AVAILABILITY
The authors declare that all relevant aggregate data supporting the findings of this study are available within the article and its supplementary information files. In accordance with the Health Insurance Portability and Accountability Act, we do not have IRB approval or patient consent to share individualized patient genomic data, which contains potentially identifying or sensitive patient information and cannot be reported in a public data repository. Foundation Medicine is committed to collaborative data analysis and has well established and widely used mechanisms by which qualified researchers can query our core genomic database of >500,000 deidentified sequenced cancers. More information and mechanisms for data access can be obtained by contacting the corresponding author or the Foundation Medicine Data Governance Council at data.governance.council@foundationmedicine.com.