Identification of eight novel SDHB, SDHC, SDHD germline variants in Danish pheochromocytoma/paraganglioma patients

During the last 9 years, we have performed genetic screening of the entire coding region and the exon-intron boundaries of the SDHB, SDHC, and SDHD genes on genomic DNA from Danish PCC/PGL patients. Up until May 2015, we have screened 143 individuals and so far 18 germline variants have been identified, of which eight are novel (Table 1). Moreover, 14 well-known polymorphisms were identified, including six in SDHB, five in SDHC, and three in SDHD (Table 2).

Variants in the SDHB gene were the most frequent, with a total of nine germline variants of which four are novel. The variants included a p.Trp218Ter nonsense mutation, as well as three large genomic rearrangements (LGR) deleting exon 1 (c.-1-?_72+?del), exons 1–7 (c.-1-?_765+?del), and exons 4–5 (c.287-?_540+?del), of which the two latter are novel. Moreover, we identified a start-loss mutation (p.Met1?) and three missense variants (p.Cys68Tyr, p.Leu139Pro, and p.Cys189Tyr) in SDHB, which were all highly conserved and predicted to be pathogenic by A-GVGD, SIFT, or PolyPhen-2 (Table 1). None of the missense variants have been described before in the ExAC database or in the Danish population. Sequencing analysis of DNA from tumor tissue from the p.Leu139Pro carrier revealed loss of heterozygosity (LOH) (Fig. 1). Finally, we identified a synonymous variant in SDHB located in exon 4 (p.Ser100Ser). This variant is reported with a frequency of 0.15 % in the ExAC database.

Moreover, we identified six variants in the SDHC gene, including one nonsense mutation, one frameshift mutation, three missense variants, and one intronic variant. Four of the six variants are novel (Table 1). Both the nonsense mutation (p.Arg15Ter) and the frameshift mutation (c.191_207del17) result in premature stop codons. The p.Arg15Ter mutation is reported in one individual out of more than 60,000 (0.00082 %) in the ExAC database. Of the three SDHC missense variants (p.Arg50Cys, p.Gly81Val, and p.Leu166Val), only the p.Gly81Val variant is reported in one South Asian individual (0.00082 %) in the ExAC database, while none of the missense variants have been identified in the Danish population. In silico analysis predicted that the p.Arg50Cys and p.Gly81Val variants to be pathogenic, while p.Leu166Val was suggested to be benign. Moreover, we identified an intronic variant (c.78-19C > T) in SDHC, which is reported with a frequency of 0.021 % in the ExAC database.

Finally, we identified three SDHD missense variants (p.His50Arg, p.Asp92Tyr, and p.Tyr114Cys) all of which have previously been described. Two of the missense variants were identified in a single patient (p.His50Arg and p.Tyr114Cys). Only the p.His50Arg variant is reported with a frequency of 0.65 % in the ExAC database. In silico analysis predicted the p.Asp92Tyr and p.Tyr114Cys variants to be pathogenic, whereas the p.His50Arg variant was predicted to be benign.

Screening for SDHB, SDHC, and SDHD germline variants in Danish PCC and PGL patients revealed the identification of 18 different germline variants of which eight are novel.

Four of the nine variants identified in SDHB are novel, including two LGR. The deletion of exons 1–7 (c.-1-?_765+?del) was identified in a patient who suffered from a PGL at the age of 36, while the deletion of exons 4–5 (c.287-?_540+?del) was identified in a 16-year-old patient diagnosed with PGL. In addition, we found an exon 1 deletion in a patient with PGL and severe hypertension recognized at the age of 51. This deletion has previously been reported in patients with PGL [21, 22]. All LGRs in SDHB introduce frameshifts and premature stop codons and were classified as pathogenic (Class 5). The SDHB p.Trp218Ter nonsense mutation was identified in a patient with PGL at the age of 39 and was classified as pathogenic (Class 5) in accordance with a previous report [23]. Moreover, we identified a start-loss mutation (p.Met1?) in SDHB in a patient diagnosed with a PGL at the age of 29. This mutation destroys the methionine start codon and is thought to prevent SDHB from being translated from this position. Although an alternative downstream in-frame methionine start codon is present at codon 58, the mutation has previously been associated with the development of hereditary PGL [24]. Furthermore, immunohistochemical (IHC) staining shows loss of SDHB in association with this variant [25] and thus we classify this variant as likely pathogenic (Class 4). All three SDHB missense variants are predicted as pathogenic by in silico analysis (Table 1). The p.Cys68Tyr variant has previously been described in patients with PCC or PGL, where IHC analysis of tumor tissue showed no SDHB staining or LOH of the wild-type SDHB allele [26, 27]. Based on these findings, the variant is classified as likely pathogenic (Class 4). The p.Leu139Pro variant has not previously been reported. Sequencing analysis of DNA from FFPE tumor tissue showed LOH, indicating that the mutation is likely pathogenic (Class 4). The final missense variant (p.Cys189Tyr) has not previously been reported, although two different amino acid changes at this position (p.Cys189Arg and p.Cys189Phe) have been associated with hereditary PGL [23, 28]. We currently classify this variant as a variant of uncertain significance (VUS) (Class 3) since no co-segregation or functional studies have been performed. Finally, the synonymous variant p.Ser100Ser has previously been reported in patients with PGL [29, 30] but since no functional studies have been performed, we classify this variant as a VUS (Class 3).

The p.Arg15Ter nonsense mutation and the c.191_207del17 frameshift mutation in SDHC both result in premature stop codons and are thus classified as pathogenic (Class 5). The p.Arg15Ter variant has previously been reported [31‐33], whereas the frameshift mutation is novel. Moreover, we identified three missense variants in SDHC, of which p.Gly81Val and p.Leu166Val are novel. In silico analysis indicates that the former variant is pathogenic while the latter is predicted to be benign. The nucleotide substitution leading to p.Gly81Val is present on the first base in exon 5 and in silico splicing analysis indicates that this change could affect the splicing of exon 5. Recently, another change at position 81 (p.Gly81Arg) has been reported to affect splicing via the splice donor site [23]. The p.Gly81Arg variant was identified in the homozygous state in a patient who developed PCC at the age of 31. Homozygous pathogenic germline mutations usually result in a much more severe phenotype or syndrome, and since no functional studies have been performed at either the RNA or protein level, the p.Gly81Val variant is classified as a VUS (Class 3). The p.Leu166Val variant is located close to the stop codon in SDHC. Another variant at position 166 (p.Leu166Arg) has been reported in the LOVD database but is not described in the literature. Since no functional studies have been performed, we classify this variant as a VUS (Class 3). The p.Arg50Cys variant has previously been described, and in silico and functional analyses indicate that this variant has some effect on the SDH activity [32, 34]. Moreover, since the variant was identified in an offspring of the proband, who suffered from PGL at a young age (<25 years), the variant is classified as likely pathogenic (Class 4). Finally, we identified a novel intronic variant (c.78-19C > T) in SDHC in an elderly patient with PGL at age 75. In silico analysis does not predict any effect of this variant on splicing; however, since no functional studies have been performed, it is classified as a VUS (Class 3).

The missense variants identified in SDHD have all previously been described. The p.His50Arg variant is predicted to be benign by in silico analysis. The variant has been described several times in the literature and functional studies in yeast show that this variant has no effect on the phenotype [35]. In agreement with these data and taking the high frequency in the control population into consideration, we classify the variant as likely benign (Class 2). The p.His50Arg variant was identified alongside a pathogenic mutation p.Tyr114Cys, which is highly conserved and has been shown to co-segregate with disease [4, 34, 36‐38] and therefore are classified as pathogenic (Class 5). Finally, the p.Asp92Tyr is a well-known Dutch founder mutation and was indeed identified in a Dutch individual, who had inherited the mutation from his father. Functional analysis has shown that the mutation impairs the function of the SDHD protein [4, 36, 38‐42] and in agreement with these findings we classify this variant as pathogenic (Class 5).

In conclusion, we have examined the mutation spectrum associated with familial PCC and PGL in the Danish population. Our investigation uncovered 18 germline variants in SDHB, SDHC, and SDHD, of which eight are novel. Based on the literature, in silico data, and LOH analysis, we classify 12 variants as likely pathogenic/pathogenic, one variant as likely benign, while five variants are classified as variants of unknown clinical significance. Further co-segregation or functional studies are needed before these variants can be classified.

IHC, immunohistochemical; LOH, loss of heterozygosity; PCC, pheochromocytomas; PGL, paragangliomas; VUS, variant of uncertain significance