*2.3. Families with Unclassified Variants*

Twelve individuals were carriers of variants for which the associated risk has not been clearly established. Overall, five VUS were found: one in multiple individuals from a single family, while the others were identified in individual subjects. Details on the variants are shown in Table 3.


**Table 3.** Frequency and predictions of unclassified variants identified.

<sup>a</sup> From gnomAD exomes; <sup>b</sup> From GERP (Genomic Evolutionary Rate Profiling http://mendel.stanford.edu/SidowLab/ downloads/gerp/). Abbreviations: P = pathogenic; LP = likely pathogenic; NP = not present; D = deleterious; B = benign; LB = likely benign; CI = conflicting interpretations; US = uncertain significance.

The p.Ser904Phe variant was found in a family with several members affected by MTC (Figure 1): eight individuals were tested and found to carry the variant, while two other relatives were obligate carriers. Out of 10 carriers, seven developed slowly progressing MTC at an average age of 46.3 years and none manifested other *RET*-related problems. p.Ser904Phe is a rare variant previously reported in one family with father and son affected by adult-onset MTC [25]; therefore, the associated risk is still unclear. Cosci et al. [26], through in silico and in vitro analyses, showed that the variant has relatively high transforming activity but low aggressiveness and suggested to assign the variant to the lowest ATA risk level A. Consistently, the segregation of the variant in our family and the clinical history of carriers showed that, although highly penetrant, this variant causes late-onset, slowly progressing MTC, leading us to hypothesize that the screening recommended for carriers of lowest-risk mutations may be appropriate for healthy carriers of the p.Ser904Phe variant.

The p.Tyr791Phe variant was found in a patient with MTC diagnosed at 22 years of age and a negative family history for endocrine diseases. The variant, first reported in patients with Hirschsprung disease [27], MTC [28,29] and PCC [30], involves a highly conserved amino acid and used to be regarded as pathogenic based on in silico predictions. More recently, the evidence that the variant has similar frequencies in affected and unaffected subjects [31,32] is more common in the population than expected for a disease-causing variant [33,34], fails to co-segregate with the disease in some families [35,36] and co-occurred with a pathogenic variant in some patients [37,38] led researchers to reconsider it as likely benign. In our patient, however, the young age at MTC diagnosis raises the suspicion that this variant may have, to some extent, favored the development of MTC, possibly interacting with other factors in a multifactorial context, or that she carries a pathogenic variant undetected by the multigene test performed.

**Figure 1.** Family tree of family 91-O-03. We report the age at diagnosis of MTC and the age at death or at the last follow-up of patients evaluated at our clinic and/or who developed MTC; in red, we indicate the result of *RET* analysis ("+" = carrier of p.Ser904Phe variant; "(+)" = obligate carrier of the variant; "wt" = testing negative).

The p.Lys710Arg variant was found in a single 76-year-old patient who had PHPT and elevated serum calcitonin, who underwent a multigene analysis for hyperparathyroidism. This variant, very rare in population databases, has never been reported in patients affected by conditions known to be *RET*-related. The lysine residue substituted by arginine at codon 710 of the protein is highly conserved, but there is a small physicochemical difference between the two amino acids; consequently, computational predictors give conflicting results on the potential impact of this missense change, which is currently classified as of uncertain significance. As the clinical picture of our patient is not strongly suggestive of a *RET*-related condition, it is likely that this variant did not play a significant role in his disorder.

The p.Ser649Leu variant was found in a patient with MTC diagnosed at 59 years of age and a negative family history for endocrine diseases who also carried the common p.Val804Met variant, belonging to the MOD risk level. Her unaffected daughter was found to carry only the p.Val804Met variant, demonstrating that the two variants are "in trans" in the proband. The variant is rare in population databases and multiple lines of computational evidence reported by Varsome support its deleterious effect on the gene or gene product, but conflicting interpretations regarding its pathogenicity are present in the literature [36,39]. The evidence that, in our patient, the p.Ser649Leu variant is present "in trans" with a known pathogenic variant is against its pathogenicity, although an additive effect in combination with p.Val804Met cannot be excluded.

The p.Asp631\_Leu633delinsGlu (c.1893\_1898delCGAGCT) variant causes an in-frame deletion of two amino acids; the nomenclature reflects the fact that the deletion starts at the third base of the Asp631 codon and extends through Glu632 up to the second base of Leu633, resulting in deletion of Glu632 and Leu633 and a change of the Asp631 codon into glutamic acid. The deletion was identified in a female child with neonatal onset of abdominal distension, constipation and vomiting, with subsequent growth retardation, who was diagnosed with abdominal-pelvic plexiform ganglioneuroma when she was 2 years old. This patient also displayed lesions consistent with neurofibromas at the buttocks, mild dysmorphic features and nodules at the upper lip and was diagnosed with MTC and parathyroid adenoma when she was 7 years old. This clinical picture, resembling the MEN2B phenotype, led us to perform *RET* testing. The p.Asp631\_Leu633delinsGlu variant is not reported in population databases, or in the medical literature; therefore, its clinical impact was completely unknown. Segregation analysis undertaken in the family demonstrated that it occurred "de novo", supporting its pathogenicity. Generally, *RET* defects associated with MEN2 and FMTC are typically gain of function, while deletions of part of the gene are mostly expected to

cause loss of function; however, cases of *RET* deletions associated with MEN2 have been reported [40]. Moreover, Borganzone et al. studied a similar somatic alteration of *RET*, p.Glu632\_Leu633del (c.1894\_1899delGAGCTG), and demonstrated that this in-frame deletion reduces the spacing between two Cysteine residues, causing ligand-independent constitutive dimerization and activation of RET. Remarkably, RET activation was even greater in this case compared to activation induced by the frequent mutation p.Cys634Arg [41,42]. Although in our variant the deletion is shifted by 1 bp compared to the somatic mutation described by Borganzone et al., it results in the deletion of two amino acids in the same location; thus, the final effect—that is, the constitutive activation of RET signaling—is likely to be the same. Collectively taken, the "de novo" origin in our patient, her clinical phenotype and the functional data support the hypothesis that the p.Asp631\_Leu633delinsGlu variant is causative of MEN2B and should be assigned to the HST/H risk class.
