**3. Discussion**

Since the important role of *RET* alterations in the pathogenesis of MTC has been ascertained, it is considered appropriate to perform the analysis of this gene in all individuals diagnosed with primary CCH, MTC or MEN2, either sporadic or familial [3]. The identification of positive subjects is in fact important for the management of patients and especially of their healthy relatives, who could benefit from specific surveillance programs and/or prophylactic treatments [43,44]. The aim of this study was to critically analyze our experience with *RET* testing.

Among the 72 individuals affected by MTC or CCH who underwent *RET* analysis at our laboratory between 2002 and 2020, 23 (31.9%) were found to carry a germline *RET* variant. As expected, a positive family history increased the chance of finding a variant: out of 16 subjects who had positive family history, 15 had a detectable *RET* variant. In the only familial case where the analysis was negative (subject 51), the patient had been treated at 32 years of age for primary CCH and her father had died at 43 years of age for MTC; given the strong suspicion of an underlying genetic cause, a second-level analysis was performed using the NGS multi-gene panel for endocrine tumors, but no variants were found. Despite the negative test results, the early age of onset and the family history are suspicious for an undetected MTC-predisposing gene defect.

Germline *RET* variants were identified in eight of 56 subjects with apparently sporadic MTC (14.3%, slightly higher than the 4–10% reported in the literature [26,45–47]), confirming the importance of screening*RET* in all cases of MTC, even when the family history is negative [48]. Indeed, testing allowed us to reclassify as hereditary a fraction of apparently sporadic cases and led us to extend genetic testing to 18 relatives, eight of whom (44.4%), belonging to three distinct families, were found to carry the variant (p.Val804Met in all cases) and were therefore able to benefit from specific surveillance.

Among patients with CCH, 1/3 (33.3%) carried a *RET* variant (of MOD risk level); actually, she was an asymptomatic patient who was found to have inherited the familial variant and subsequently undertook the surveillance program that led to the diagnosis of CCH. On one hand, this may have represented a successful instance of early diagnosis, provided that CCH is a precursor condition for MTC; on the other hand, we cannot be sure that this benign finding would eventually evolve into a malignant condition and this may have been, conversely, a case of overdiagnosis and overtreatment.

Taking into account clinical features and genetic test results shows that wild-type *RET* patients and carriers of MOD risk level variants only differ by family history (*p* < 0.001), but not by clinical characteristics such as mean age at diagnosis of MTC and cancer stage. This is also supported by the fact that among 40 individuals found to carry variants assigned to MOD risk level or the p.Ser904Phe variant (excluding those for whom we have no clinical information), 22 had not developed primary CCH or MTC at the time of the last follow-up (55.0%): 4/9 of those over 60 years of age (44.4%), 8/18 of those between 40 and 60 years of age (44.4%), 8/11 of those between 20 and 40 years of age (72.7%) and 2/2 of those under 20 years of age (100.0%). Several lines of evidence have suggested that the aggressiveness of MTC does not depend on the presence, absence or type of *RET* variant (which mainly affects the age at onset of the disease) but on the stage and the age at diagnosis of the disease [20], which are the strongest predictors of survival for patients with MTC; therefore, our data further support the appropriateness of the non-invasive screening recommended by current guidelines for healthy carriers of MOD risk level variants.

For carriers of variants with unclear associated risk, however, until the genotype–phenotype correlations are clarified, the attempt to assess the risk and respective proper management of healthy carriers only relies on the clinical history of carriers—which, however, can provide meaningful information only when large families with many characterized members are available—or on any significant biomolecular evidence.

Thus, in the 91-O-03 family, the availability of several genetically and clinically characterized members allowed us to provide evidence that the p.Ser904Phe variant is highly penetrant (7/10 of carriers developed MTC, 70%) but leads to the development of slowly-progressing MTC at relatively advanced age (average age at diagnosis: 46.3 years), suggesting that recommended screening for carriers of lower risk mutations is appropriate for healthy carriers of this variant.

Moreover, in the case of the p.Asp631\_Leu633delinsGlu variant, the clinical picture of the patient and the "de novo" origin, associated with the functional studies on a very similar variant [41,42], provide convincing evidence in favor of the pathogenicity of p.Asp631\_Leu633delinsGlu and of its assignment to the HST/H risk level category. Of note, clinical manifestations in this patient were consistent with MEN2B, a phenotype that has been reported to be associated in 95% of cases with the p.Met918Thr variant and in 2–3% of cases with the p.Ala883Phe variant [49]. Both these variants affect residues located in the substrate specificity pocket of the central catalytic core of the tyrosine kinase domain and likely cause RET activation by altering its substrate specificity [50]. Conversely, p.Asp631\_Leu633delinsGlu affects residues located at a great distance in a different domain (the cysteine-rich extracellular domain). Mutations in this domain are expected to cause RET activation by inducing its disulfide-linked dimerization and are generally associated with a MEN2A/FMTC phenotype, which gives a new perspective in the view of elucidating molecular mechanisms leading to the more severe MEN2B phenotype. Rarely, a MEN2B-like phenotype has been described in patients carrying two *RET* variants (bi-allelic or *in-cis* on the same allele) [51,52]. It can be hypothesized that, although different, all these defects result in a particularly intense RET activation and that the higher the activation level, the more severe is the phenotype.

Among the families with MOD risk level variants, the case of the 228-O-18 family, carrying the p.Leu790Phe variant, is worthy of consideration. In this family, the p.Leu790Phe variant was found in nine individuals, four of whom developed MTC; intriguingly, three of these individuals were also affected by neurofibromatosis 1 (NF1), caused by a mutation of the *NF1* gene inherited from the other parental branch, and one developed a PCC. The p.Leu790Phe variant is classified at the lowest risk level and is generally associated exclusively with MTC [53]; it is therefore possible that, in this individual, the risk of developing a PCC was greater than that of the ordinary carriers of this variant, due to the co-presence of the mutation of *NF1*, another gene whose alterations are associated with an increased risk of developing PCC.

One limitation of this study is that most patients underwent the analysis of selected *RET* exons through Sanger sequencing, which is expected to be less sensitive if compared to whole-gene NGS-based analysis. However, since all the variants identified were found using the Sanger method, and the percentage of individuals is in line with data previously reported in the literature (even slightly higher for sporadic cases), we can conclude that this testing approach demonstrated satisfactory accuracy in finding *RET* variants, supporting the evidence that most clinically relevant variants reside in known mutational hotspots [54].
