Spectrum of Germline RET variants identified by targeted sequencing and associated Multiple Endocrine Neoplasia type 2 susceptibility in China

Background Germline RET mutations and variants are involved in development of multiple endocrine neoplasia type 2 (MEN2). The present study investigated a spectrum of RET variants, analyzed genotype-phenotype relationships, and evaluated their effect on the MEN2 phenotype in Han Chinese patients. Methods Targeted sequencing detected germline RET variants in 697 individuals, including 245 MEN2, 120 sporadic medullary thyroid cancer (MTC), and 15 pheochromocytoma (PHEO) patients and their 493 relatives. In silico analyses and classifications following ACMG-2015 were performed. Demographic, clinical variant types, and endocrine neoplasia molecular diagnosis records were also analyzed. Results Nineteen different RET mutations (18 point and 1 del/ins mutations) in 214 patients with MEN2A (97.7%) or MEN2B (2.3%) were found, of which exon 11/10 mutations accounted for 79% (169/214). Nineteen compound mutations were found in 31 patients with MEN2A. Twenty-three variants (18 single and 5 double base substitution/compound variants) non-classification were also found. Of these, 17 (3 of pathogenic, 10 of uncertain significance, 2 of likely benign and 2 as benign) were found in 31 patients with MTC/PHEO. The remaining 6 variants (4 of uncertain significance and 2 of likely benign) found in 8 carriers had no evidence of MEN2. The entire cohort showed MEN2A-related PHEO, all occurring in exons 11/10, particularly at C634. Kaplan-Meier curves showed age-dependent penetration rates of MTC and PHEO, and occurrence rates of PHEO in patients with exon 11 mutations were all higher than those within exon 10; these bilateral PHEO were always associated with exon 11 mutations (all P < 0.05). While patient offspring had PHEO, parents with MEN2A had none, the frequency was approximately 10%. Interestingly, at least 6.8% of families were adoptive. Also, 3 non-hotspot RET variants (R114H, T278N, and D489N) appeared with high frequency. Conversely, polymorphism S836S was absent. Conclusions These data are largely consistent with current evidence-based recommendations in the clinical practice guidelines. Diversity of RET variants or carriers may involve a different natural disease course. Further large-scale targeted sequencing studies will serve as an accurate and cost-effective approach to investigating MEN2 genotype-phenotype correlations for discovery of rare or unknown variants of RET.

In the present study, a spectrum of germline RET variants were investigated in patients in our hospital using targeted genes and next-generation sequencing (targeted sequencing). Specific analysis of genotype-phenotype relationships, and the possible effects of rare RET variants on the MEN2 phenotype was assessed.

Participants
From 2011 to 2020, 206 index cases of histopathologicallydiagnosed MTC and/or PHEO were subjected to genetic screening at the 903rd PLA Hospital and Zhejiang Cancer Hospital (Hangzhou, China). After confirming RET pathogenic variants in 73 MEN2 patients or rare variants in 118 cases of sporadic MTC (sMTC), and 15 cases of sporadic PHEO (sPHEO) patients were performed. Relatives of patients were also assessed for family studies. All individuals were subjected to clinical examinations, biochemical/imaging examinations, and genetics screening according to the published criteria [1,7] and/or underwent surgical treatment. Data of 697 individuals were summarized in Fig. 1. Most of the study population was from Zhejiang Province, and a few were from Shanghai, Guangdong, Jiangsu, Anhui, Fujian, Guangxi, Sichuan, Hubei, Henan, Shandong, Shanxi, Xinjiang, and other provinces and cities in China. The study protocol was approved by the Ethics Committee of the 903rd PLA Hospital, and written informed consent was obtained from all study subjects or their legal guardians.
RET screening using targeted sequencing Targeted sequencing using an Illumina DNA-HiSeq 2000 Analyzer was performed as previously described [7,8,39]. Briefly, genomic DNA was extracted from EDTAanticoagulated peripheral blood and then used for target capture, enrichment, and elution. A custom capture array (NimbleGen; Roche) was designed to capture all exons and flanking splice junctions, as well as the immediately adjacent intron sequences for 10 genes (RET, NF1, MAX, TMEM127, VHL, SDHA, SDHB, SDHC, SDHD, and SDHAF2) associated with hereditary PHEO diseases following GeneReviews (NCBI). The results were followed by Sanger sequencing with an ABI Prism 3700 automatic sequencer (Perkin-Elmer, Fremont, CA, USA).

Molecular characterization confirmation
The entire germline RET coding region, including some variable splicing and flanking regions, was analyzed (Genbank RefSeq: NM_020975.4). According to American College of Medical Genetics and Genomics (ACMG) and the Association for Molecular Pathology (AMP) Guidelines (hereafter referred to as ACMG-2015), MEN2 Database, and ATA-2015 recommendations [1,10,40,41], three specific terms were used to classify or define RET single variants: "pathogenic variant" or "mutation" refers to those variants affecting RET structure and function and well-defined as causing MEN2; conversely, SNPs occurring in greater than 1% of a population or self-verified not to alter cellular functional effect were considered "benign." RET SNPs observed included A45A (rs1800858), A432A (rs1800860), G691S (rs1799939), L769L (rs1800861), S836S (rs1800862), and S904S (rs1800863); a "variant of unknown significance" (VUS) were those considered "likely pathogenic," to have "uncertain significance" (UCS), and "likely benign." Moreover, RET compound mutations or variants were limited to the presence of at least one pathogenic variant and at least one other simultaneous concomitant mutation (s) or VUS (defined mutation) or a combination of two or more VUS (as VUS). However, compound mutations/variants were not classified by the ATA-2015.

In silico analyses
In silico analyses were performed using public databases and web-based software with three different bioinformatics algorithms: Sorting Intolerant From Tolerant (SIFT), Polymorphism Phenotyping-2 HDIV (PolyPhen-2 HDIV), and Mendelian Clinically Applicable Pathogenicity (M-CAP) for missense prediction of protein functional significance of the RET variants. Seventeen different RET pathogenic point mutations were also matched within exons 10, 11, and 13-16, and a G691S SNP within exon 11 as controls reported previously [7].

Statistical analysis
All data were analyzed with SPSS version 20.0 (SPSS Inc., Chicago, IL, USA). Measurement data confirming ordinal or normal distribution was summarized and reported as the mean ± standard deviation, mean (range), or both, and comparisons of quantitative variables were made with a Student's t-test. The frequency of occurrence, percentages, and comparisons of enumeration variables were assessed by the χ 2 or Fisher's exact test. Age-dependent penetrance of MEN2-related MTC/ PHEO was assessed with Kaplan-Meier curves, and a log-rank test was used for comparisons between curves. The test level was set α = 0.05, and a P < 0.05 was considered statistically significant.

RET screening and clinical information
A total of 697 individuals were subjected to RET screening by targeted sequencing. Of them, 73 were index cases, and their 399 relatives were used for family studies (i.e., 245 patients with RET mutations and 227 of their relatives without RET mutations); others included 118 sMTC, 15 sPHEO, and their 92 relatives with or without rare variants (Fig. 1) Fig. 2a).
Of these 18 single variants, 10 missense variants associated with MTC, 3 (I788S, S409Y, and A604S) had not been previously described in the Database of SNP (dbSNP), 1000 Genomes Project (1000 GP), Exome Aggregation Consortium (ExAC), or genome Aggregation Database exomes (gnomAD exomes) ( Table 2). The I788S (c.2363 T > G), was considered a "damaging" variant by three different in silico analyses, and a patient presenting with MTC alone was diagnosed at 43 years-old. Variant S409Y (c.1226C > A) also qualified as "damaging" or "possibly damaging," and 5/14 carriers were diagnosed with MTC at a mean age of 57.6 years (range, 41-75 years). The latter, A604S (c.1810G > T), qualified as "tolerated," "possibly damaging," or "damaging" depending on the 3 algorithms used; 2 patients from 2 different families were diagnosed with MTC at 44 and 46 years-old, respectively. Of the remaining 7 missense variants associated with MTC listed in these databases (Table 2), 2 variants [M918V (c.2752A > G) and L1018F (c.3052C > T)] qualified as "damaging" or "possibly damaging," and 2 patients were diagnosed with MTC at 69 and 46 years-old, respectively. The other 5 variants qualified as being either "damaging" or "benign;" 2 of these 8 patients with MTC were diagnosed before age 40. Three variants, R114H (rs76397662), T278N (rs35118262), and D489N (rs9282834), and the following compound mutations described, appeared at a relatively high frequency (Tables 2, and 3). Also, the frequencies in East Asian populations listed in the 1000 GP, ExAC, and gnomAD exomes databases were relatively higher than that of other ethnic populations (https://www.pubvar.com/ variant/10-43597793-G-A; 43,600,607-C-A; 43,606,856-G-A). Interestingly, two synonymous variants, P679P (c.2037C > T) in 2 patients with MTC were diagnosed at 26 and 72 years, and P841P (c.2523G > A) in 1 patients with MTC was diagnosed at 39 years-old, respectively, were found in 10 carriers from 4 families. The remaining 6 single variants, 4 missense and 2 synonymous, were found in 8 carriers who had no evidence of MEN2-related clinical manifestation (Table 2). Further, InterVar Classify System which mainly consists of automatically interpretation by 28 criteria and manual adjustment by users to re-interpret the clinical significance was used for classifying all these 18 single variants according to the consensus recommendation of the ACMG-2015 (http://wintervar.wglab.org/). Two variants (S409Y and M918V) could be classified as pathogenic, 10 of UCS, 4 of likely benign, and 2 as benign (  (Table 1). While these ages did not significantly differ (t = 0.826, P = 0.410), there seems to be a trend towards younger onset (Table 3).
The other 10 patients with exon 10 compound mutations all had MTC alone; 6 carried trans C618R/T278N, C618Y/R114H, or C618Y/A1105V mutations and were diagnosed at a mean age of 36.3 years (range, 12-59 years). Another male patient with C618R/A639T mutation was diagnosed with MTC at age 38. His son and daughter, carrying a C618R/A639T mutation, respectively, were respectively diagnosed with MTC at 15 and 25 years-old, however, his wife had non-C618R or A639T, meaning the C618R/A639T mutation was cis. In addition, an isolated female patient with a 620S/R114H mutation was diagnosed at age 25. The MTC diagnostic ages of 10 patients with exon 10 compound mutations (32.1 ± 14.61 years) and 65 patients with single mutations (40.48 ± 17.80 years) were not significantly different (t = 1.401, P = 0.166). The remaining 7 patients presenting with MTC alone and trans L790F or S891A compound mutations were diagnosed at a mean age of 41.8 years (range, 16-66 years). This included 3 patients carrying an S891A/A1068 mutation diagnosed at 25, 33, and 40 years, respectively.

Additional information and RET polymorphisms
A total of 328 carriers with 61 different RET variants including 38 pathogenic and 23 variants were found (Tables 1, 2 and 3; Fig. 1). With the exception of 6 RET variants in 8 carriers that exhibited no evidence of MTC/PHEO, 55 different RET variants (38 pathogenic and 17 variants) in 276 patients with confirmed or suspected MEN2 are shown in Fig. 1 and Fig. 3, and distribution of pathogenic variants frequencies is summarized in Table 4. Interestingly, 5 offspring of individuals with MEN2A-related unilateral PHEO were diagnosed at 18,21,26,29, and 37 years. So far, their father or mother with MEN2A showed no clinical, biochemical, or imaging manifestations of PHEO. The frequency in all 51 patients with MEN2A-related PHEO was approximately 10% (Tables 1, 3, and 5). Moreover, family studies unexpectedly found that 5 index patients (probands) belonging to 5 families were adopted as an orphan or abandoned child (3 C634Y, 1 C634S, and 1 C634W mutations), while their foster parents and/or siblings without RET mutations consistently showed no evidence of MEN2A. As a consequence, 6.8% of the 73 RET-defined families were actually adoptive families (Tables 1 and 3) in addition to 1 patient with an S409Y mutation that was not classified by the ATA-2015 which was also adopted. The presence of 5 RET SNPs (A45A, A432A, G691S, L769L, and S904S) and absence of S836S (rs1800862) was observed in all 697 individuals included, similar to those shown in East Asian populations listed in the 1000 GP, ExAC, and gnomAD exomes databases. The S836S mutation frequency was lower than that of other ethnic populations (https://www.pubvar.com/

Discussion
To our knowledge, the present study is the first to analyze the distribution of RET mutations/variants in confirmed or suspected MEN2 patients by targeted sequencing in an ethnic Han Chinese cohort (Table 1; Fig.  1). With the exception of 93 patients defined as sMTC and 14 classified as non-MEN2-related PHEO, 55 different pathogenic or variants of RET were found in 276 patients with confirmed or suspected MEN2 (Tables 1, 2 and 3; Fig. 1). C634 in exon 11 was the most frequently mutated codon. However, there are differences in the frequency of these mutations due to sample size, geography, and ethnicity. For example, C634 has been found in approximately 30.6-43.2% of European-Americans [14][15][16][17] and 37.1-72.5% of Asian populations [11,43]. Moreover, the G533C mutation, with a prevalence of 36.2% in Greece, has also appeared to cluster in Brazilians living in the US and Mediterranean countries that might be extrapolated to originate from Greece, but rarely reported in other ethnic groups [ Table 4] [13][14][15][16][17][43][44][45]. In the current series, the high prevalence of families with single mutations that were carriers of C634Y (33.8%) was presumably caused by a founder Table 4 Distribution of germline RET pathogenic mutation frequencies observed in China and in 5 major published studies, BrasMEN (17), France (16), Germany (15), ItaMEN (18) and Greece (43). The table references to Lebeault et al. (16) and Rui et al. (17) reported previously and complements our data RET  V292M, Refer in particular to the V292M/R67H/R982C *Our these 82 families included 72 families with single missense mutation or compound mutations, 1 family with del/ins mutation and other 9 families respectively with V292M/R67H/R982C, S409Y or M918V considered as pathogenic mutations effect. Conversely, the frequency of families with V804M (4.9%~6.2%) was lower than that reported in France and Italy but similar to that in Greece [Tables 1 and 4] [14,16,18]. These results implied that the spectrum of RET mutations identified in the present study is quite different from that found in other countries [ Fig. 3; Table 4] [14][15][16][17][43][44][45].
Over the past 25 years, new insight into the natural course of disease and genotype-phenotype data caused a paradigm shift in management of MEN2. Identification of asymptomatic carriers of RET mutations in exons 8, 10, 11, and 13-16 by a routine procedure, followed by appropriate screening through pedigree investigation, early diagnosis, and timely prophylactic treatment is essential to improving the likelihood of good outcomes [1,9,[21][22][23][24]. In present study, 88.8% of 276 patients presenting with MEN2 carried RET-defined mutations, of which 87.3% had single mutations, and 12.7% had compound mutations. However, 31 (11.2%) patients with suspected MEN2 carried RET variants mostly located in non-hotspots and were not classified by the ATA-2015 ( Fig. 1; Tables 1, 2 and 3), which is consistent with recent reports of more non-cysteine-linked codon mutations and rare RET variants [14,20]. Nonetheless, these results provide a novel insight into MEN2 development.
Nineteen single mutations of RET in 214 patients with MEN2 were found herein (Table 1; Figs. 1, 2, and 3). Of these, 18 were involved in the development of MEN2Arelated MTC in 94.8% of 194 carriers with available clinical information. Of which 98 patients with C634 mutations (ATA-H) all presented with MTC. In accordance with the ATA-2015 recommendation, prophylactic thyroidectomy should be performed before age 5, whereas 10/96 patients with an ATA-MOD mutation and normal basal serum calcitonin levels should be actively followed up and monitored [P = 0.001] [1,4,7,18]. Also, 87.1% of the 194 patients carrying mutations in exons 11/10, and 24.7% of patients presenting with PHEO only associated with cysteine mutations in these two exons, predominantly in C634 mutations with a combined PHEO was 43.9%, and the youngest PHEO diagnostic age was 18. The evidence seems to suggest that clinical follow-up of MEN2A cases with non-cysteine mutations could be simplified by eliminating the annual or extending times interval evaluation of PHEO, which is still recommended by guidelines. Of course, it supports that mandatory in patients with cysteine mutations, in particular at C634 [1,[14][15][16][17][18]36]. Further analysis of the estimated cumulative frequency of MTC or PHEO by current Kaplan-Meier curves showed that penetration rates of MTC and PHEO were higher in patients with mutations in exon 11 versus 10 (P = 0.041 and P = 0.000, respectively), and PHEO incidence rates in patients with mutations in exon 11 were higher than in those with mutations in exon 10 (P = 0.000). Moreover, these bilateral PHEO were always associated with exon 11 mutations (P = 0.014) and lack of non-cysteine RET mutations (Table 1; Fig. 2). In contrast, the mean age at PHEO diagnosis was not significantly different between patients with exon 11 and 10 mutations (P = 0.215), though those with exon 10 mutations occurred relatively later 7.6 years. Of note, 5 offspring of individuals with MEN2A-related PHEO, but their father or mother with MEN2A had consistently no evidence of PHEO. These unexpected findings indicate that patients presenting with the same genetic alteration can vary significantly in clinical phenotype. It is speculated that the natural history of MEN2A-related PHEO could be influenced by genetic or environmental modifying factors, but continued research to confirm or refute this hypothesis is necessary, especially with respect to RET variants/SNPs seemingly not involved in the development of PHEO in these specific 5 patients [1,16,27,30,36,38] (Table 5). Moreover, in the present study, exception with a genotype-phenotype relationship between MEN2A-related CLA and RET C611Y [45], the CLA and HPTH mainly involving C634, in contrast to be classically reported that CLA develops in up to 9% of carriers, HPHT in 20-30%, and HD in 7%, suggesting those prevalence rates in Chinese or Asian populations were lower than that in European-Americans [1,11,43]. Even so, CLA might present earlier, prior to the onset of clinical symptoms, facilitating early recognition of individuals at risk of MEN2A-specific tumors [1,46]. The remaining 1 single mutation involved in the development of MEN2B in 5 patients, all of whom were affected by the M918T mutation that it is known to present a unique physical appearance characteristic of extraendocrine signs. However, the frequency and prevalence rate of the de novo M918T mutation in the present study were lower than those reported previously [60% versus 90 and 2.3% versus 5%, respectively] [1,47]. Unexpectedly, 1 patient with MEN2B presented with mixed medullary-follicular carcinoma in conjunction with special immunostaining features, which may be unique biological behavior and a relatively favorable prognosis relative to other MEN2B-related pure MTC [48].
Herein, 17 rare RET variants were found in 75 carriers present in 31 patients with suspected MEN2 mostly diagnosed after age 40 (Table 2; Fig. 3), and the majority of these variants have uncertain biological significance. However, it is interesting that an I788S variant found in a 43-year-old patient with MTC was predicted to be "damaging", while the substitution is considered to be of UCS according to ACMG-2015 guidelines. In addition, a 46-year-old patient with MTC was reported to have the synonymous heterozygous variant I788I [c.2364C > T] [49], implying that I788S may be a UCS or a potentially pathogenic mutation. Two patients diagnosed at 44 and 46 years, respectively, had an A604S variant classified as UCS depending on the 3 algorithms used and following the ACMG-2015. An important finding of the present series was the discovery of the S409Y variant classified as "damaging" or "possibly damaging" whose functional tests had a low oncogenic potential. Meanwhile, cosegregation with MTC in at least 6 patients of 4 families affected by S409Y causing the disease has been definitively confirmed [7]. The S409Y variant is considered a pathogenic variant by the ACMG-2015. The M918V mutation has been shown to have low oncogenic potential by in vitro testing [50] and has been associated with a moderate risk of MTC in a recent study of multiple families, none of whom presented clinical features of MEN2B [16,51]. Herein, the patient with M918V was diagnosed with MTC alone at age 69, further supporting that M918V may induce in MTC, and is classified as a pathogenic mutation by the ACMG-2015. Thus, S409Y and M918V variants should be classified as ATA-MOD mutations associated with familial MTC [ Fig. 3; Table 4] [7,16,51]. In contrast, a 41-old-year female patient carrying R600Q had MTC, but her father carrying R600Q had no abnormality, similar to a previous report [52], classifying this variant as UCS. Interestingly, a relatively high frequency of R114H, T278N, and D489N variants was revealed in the current series and in other East Asian populations, mainly associated with HD, but rarely reported to play a role in MTC or in tumorigenesis [53][54][55][56]. Herein, patients with these 3 variants had no family history and alterations that were more like sMTC. Currently, limited evidence is available regarding whether or not these 3 variants are "benign". Five patients from 4 families were found to have the cis V292M/R67H/R982C compound mutation. One had isolated MTC at 70 yearsold, and 1 had C cell hyperplasia at 44 years-old. Of the other 3 from 3 families, 2 had MTC at ages 43 and 47, and 1 had PHEO alone at age 26. Meanwhile, another study reported an Italian MEN2A patient with V292M presented at the age of 44 with PHEO and MTC, and in vitro assays showing a low-grade transforming potential associated with V292M have been reported previously [57]. In this study, the V292M/R67H/R982C alteration was considered to have weaker pathogenicity [ Fig. 3; Table 4] [13,25,57]. There are no family studies available on the other variants diagnosed before age 40 and rare variants (A1068L and L19delC/A1068L), so their influence on the clinical course of MTC is not yet clear. Similarly, the fact that the remaining 6 variants (4 of UCS and 2 of likely benign) were found in 8 carriers even though none of them had MEN2 remains to be further clarified (Table 2). Nonetheless, accurate characterization of the pathogenic role of these variants, including new family studies, and the correlation between genotype-phenotype would be of great relevance for patients. Clinicians should be prudent in trusting in silico results and choose appropriate treatment approaches for patients with susceptibility variants.
Regarding the discovery of the 19 compound mutations in 31 patients with MEN2 (Table 3; Figs. 1, and 3), all major mutations were classified as either ATA-H or ATA-MOD, whereas secondary variants were not classified by the ATA-2015. The mean age at MTC diagnosis of those with compound mutations showed a trend towards younger ages relative to those with single mutations, and there seems to be no need to change the risk grade. However, compound mutations C634Y/V292M/ R67H/R982C and C634F/V292M/R67H/R982C seemed to lead to higher MTC aggressiveness or clinical staging than C634Y or C634F and V292M/R67H/R982C [25,42], while the compound mutations S891A/A1068L and C618R/A639T may predispose to a relatively earlier MTC diagnostic age than that of S891A and C618R. Thus, an additive synergistic effect is speculated to exist. In contrast, 2 patients with S891A/R525W mutations showed a relatively later MTC diagnostic age than those with S891A described here and elsewhere [37,58]. These results reflect the fact that the combined secondary variant has a duality or diversity of effects on disease genesis, although there is insufficient clinical data in the present study to support a change in risk category. Since 12 patients with concomitant secondary variants R114H, T278N, and D489N presented with a relatively wide range of diagnostic ages, the clinical disease course remains uncertain and inconsistent. A cautious approach might be to remain aware of the possible concomitant presence of compound mutations when faced with an abnormal natural disease course or unusual clinical features; however, appropriate validation is needed to avoid misinterpretation and irreversible clinical consequences [25,29,30,37,38,42].
Approximately, 5.6-9% of patients with MEN2A are de novo RET mutation, and that almost always arises from the paternal allele [59]. However, there are no MEN2A patients carrying de novo mutations in the present series. Screening for germline RET mutations in new patients with MTC regardless of their family history is recommended and may help not only to discover de novo families, but also correctly diagnose approximately 1-7% of patients with presumed sMTC that actually have MEN2 [1,14,19], such as patients with S409Y, M918V, and C609R, C634W in the present series. Unexpectedly, however, at least 6.8% of families in the current study were found to be adoptive, reflecting a thoughtprovoking socioeconomic phenomenon-relatively high abandonment and/or adoptive rates of children with MEN2A, more likely to occur in emerging countries [1,19,59]. Controversially, data of adopted children was usually protected and managed by various legislative and social regulations, the worldwide adoptive rate might need further clarifying. Integrate accurate genealogical information, RET testing, and statement records of proband patients, which could have a cautious and clinical diagnosis of being adoptive. Considering public policy standpoints and social ethics, cost-effective haplotype analysis seems unnecessary. Nevertheless, an accurate diagnosis depends on various molecular biology methods, and haplotype analysis is still worth performing to determine lineage, facilitating a biological definition for adoptive families [1,25,59]. Additionally, the SNP S836S was absent in the present study, similar to observations in East Asian populations showing its frequency was lower than that of other ethnic populations.

Conclusions
In conclusion, the present data are largely consistent with the current evidence-based recommendations in the European-American clinical practice guidelines. Since the genetic, socioeconomic, and environmental backgrounds of Asian populations is very different, diversity in RET variants or population carriers may result in a different natural disease course. Further large-scale studies using targeted sequencing should be conducted as it is a rapid, accurate and cost-effective approach for study of genotype-phenotype correlations and discovery of rare or unknown variants of RET.