Should the GCM2 gene be tested when screening for familial primary hyperparathyroidism?

Introduction

Primary hyperparathyroidism (PHPT) is a disorder characterized by hypercalcemia and elevated or inappropriate parathyroid hormone (PTH) secretion due to pararathyroid gland hyper-functioning. It can result from a single or multiglandular lesion, adenoma, hyperplasia or much less frequently carcinoma. The prevalence of this disease is estimated around 1–7 cases per 1000 persons (1). PHPT can be either sporadic or inherited. Among inherited PHPT (5–10% of cases), some belong to syndromic diseases such as multiple endocrine neoplasia type 1 (MEN1; due to mutations in MEN1 tumor suppressor gene), multiple endocrine neoplasia type 2A (MEN2A; due to mutations in RET proto-oncogene), multiple endocrine neoplasia type 4 (MEN4; due to mutations in CDKN1B proto-oncogene) or hyperparathyroidism-jaw tumor syndrome (HPT-JT; due to mutation in CDC73 = HRPT2). PHPT can also be non-syndromic and isolated. The cases of familial isolated PHPT (FIHP) can be included in incomplete expression of MEN1, MEN4, HPT-JT or associated with mutation in CASR, GNA11 or AP2S1 genes (2, 3). In recent years, activating mutations in GCM2, a gene previously known to be involved in inherited hypoparathyroidism, have been described in PHPT (4, 5). Cases of PHPT associated with GCM2 mutations in literature are characterized by multiple enlarged parathyroid glands, greater preoperative parathyroid hormone levels than patients without GCM2 mutations, persistent or recurrent PHPT after surgery and higher risk of parathyroid carcinoma (6). Currently, frequency of GCM2 mutation in FIHP is estimated to be around 18% in literature (5).

GCM2 is located in locus 6p34. It encodes for a 506 amino acid protein composed of several functional domains: a DNA-binding domain, 2 transactivation domains, a C-terminal conserved inhibitory domain (CCID) and a nuclear localization signal. This protein, which is exclusively expressed in parathyroid cells, acts as a transcription factor involved in parathyroid gland differentiation and development and in PTH production. Knockout mice for GCM2 are characterized by parathyroid agenesis and hypoparathyroidism (7). GCM2-inactivating mutations have been involved in hypoparathyroidism and more recently, activating germline mutations in this gene have been described in association with hyperparathyroidism, suggesting that GCM2 is a parathyroid proto-oncogene. These gain-of-function mutations are mostly located in the CCID domain (5).

Currently, genetic screening recommendation for PHPT in France do not include GCM2 analysis. The aim of our study was to investigate germline GCM2 sequence variations in a French cohort of patients with PHPT and to classify the identified variants according to ACMG-AMP guidelines (8, 9) in order to determine if this gene should be systematically included in PHPT gene screening decision tree. The deleterious effects of some sequence variants are controversial in literature and are still open for debate.

Material and methods

Results

Seven hundred thirteen patients with PHPT were included in this study. Excluding a very frequent variant identified in about 10% of the DNA samples and directly considered as a benign class 1 variant (c.-44T>C, rs16870746, 10.73% in gnomAD(2.1)), 85 patients carried at least one GCM2 variant.

To be more accurate, we identified 12 variants classified as class 1 or 2 variants (benign or likely benign) according to ACMG criteria (Table 1). Interestingly, some of these variants were cosegregating variants such as c.679A>G, p.(Ile227Val) associated with c.350C>T, p.(Ala117Val) and c.462G>A, p.(Lys154Lys) associated with c.607G>A, p.(Gly203Ser). These class 1 or 2 variants were not taken into account for genotype–phenotype correlation, as they were very frequent and considered as SNP.

Twelve sequence variations were classified as class 3 (uncertain significance) or 4 (likely pathogenic) variants (Table 2). No sequence variation was classified in class 5 (pathogenic). Forty-seven index cases harbored these classes of variants, which correspond to a frequency of 6.6% (47/713). More accurately, 44 patients were carriers of a class 3 variant and three patients carried a class 4 variant. Interestingly, all these sequence variations were substitutions. We identified no deletion or duplication in these classes of variants. Main clinical characteristics of the corresponding patients are presented in Table 3. For these patients, mean age of PHPT diagnosis was 49 years (min: 13; max: 82). Nine patients harbored parathyroid hyperplasia, 17 patients presented with parathyroid adenoma and anatomopathological data were not available for the other patients.

In parallel of the study of this cohort, we had the opportunity to perform genetic investigations in a family (genealogic tree in Fig. 1). The index case, a 28-year-old man, presented with a 20 mm parathyroid adenoma associated with a calcemia level of 4.15 mmol/L and a PTH level of 355 pg/mL. Our genetic analyses allowed to identify the c.1181A>C, p.(Tyr394Ser) class 4 variant. Interestingly, above the 13 relatives analyzed in this family, we identified this variant in 7 of them: four members carrying this variant at heterozygous state but without PHPT respectively at the age of 80 (Ca: 2.4 mmol/L and PTH: 30.2 pg/mL), 58 (Ca: 2.3 mmol/L and PTH: 6 pg/mL), 34 and 33. The three other members of this family harboring the variant had a symptomatic PHPT.

Genealogic tree of the family with GCM2 (NM_004752.3) c.1181A>C, p.(Tyr394Ser) variant.

Citation: European Journal of Endocrinology 182, 1; 10.1530/EJE-19-0641

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Genealogic tree of the family with GCM2 (NM_004752.3) c.1181A>C, p.(Tyr394Ser) variant.

Citation: European Journal of Endocrinology 182, 1; 10.1530/EJE-19-0641

• Download Figure
• Download figure as PowerPoint slide

Genealogic tree of the family with GCM2 (NM_004752.3) c.1181A>C, p.(Tyr394Ser) variant.

Citation: European Journal of Endocrinology 182, 1; 10.1530/EJE-19-0641

• Download Figure
• Download figure as PowerPoint slide

Discussion

In our study, 47 patients out of 713 were carriers of a class 3 or class 4 variant, none of them carried a class 5 variant. This represents 6.6% of the patients screened for PHPT in our population. Guan et al. (5), who selected kindreds with FIHP, found a frequency of 18% of patients with a GCM2 variant and without deleterious mutation on the current other genes involved in heritable hyperparathyroidism. This observed discrepancy can be due to selection of familial cases in their study and to prior exclusion of mutation in MEN1, CASR, and CDC73, while our cohort included all patients screened for heritable PHPT, including sporadic cases. Sporadic presentation in case of genetically determined PHPT is not so rare because PHPT can be asymptomatic over many years, knowledge of first- but especially second- or third-degree relative affected with PHPT is not always obvious, and incomplete penetrance of pathology with genetic abnormality is also to be considered. That is the reason why genetic screening is recommended in France for all patients of less than 50 years with PHPT, even with sporadic presentation. The mean age of PHPT diagnosis in patients carriers of a class 3 or 4 GCM2 variant in our study was 49 years, which stands between mean age of parathyroidectomy of family presenting PHPT patients (42.8 years) and sporadic presenting PHPT patients (59.2 years) in the study of El Lakis et al. (1).

Castellano et al. (10) reported a female-to-male ratio of 3:1 in PHPT patients. In our study, this ratio was 2:1 in patient carriers of a class 3 or 4 GCM2 variant. This more equilibrated female-to-male ratio is consistent with a genetic predisposition to the pathology and could thus be explained by the Mendelian transmission.

El Lakis et al. (1) also demonstrated that activating mutations of GCM2 were associated with a greater rate of multi-glandular diseases. In our study, some patients with class 3 or 4 variants harbored parathyroid hyperplasia or multiple glandular adenomas but the lack of anatomopathological information available for many patients did not allow us to confirm this point.

A great number of variants are still classified as class 3 variants (variant of uncertain significance). Indeed, even when in vitro functional data are available, their interpretation can remain tricky, as for the c.844T>G, p.(Tyr282Asp) variant. In our cohort, this variant has been identified in association with a deleterious variant once in MEN1 gene and once in CASR gene. As a transcription factor expressed in parathyroid cells and activating PTH transcription, GCM2 can contribute to development of PHPT but, may be, to a lesser extent than MEN1, CDKN1B or HRPT2 (11). Interaction with CASR expression in parathyroid cells has been described, as well as interaction with GATA3 and MafB for PTH expression, but some compensation mechanisms seem to exist (12, 13). This could explain the low penetrance observed, as shown with our family study.

The pathogenicity of p.(Tyr282Asp) variant is controversial in literature and its role in tumorigenesis is not proven (14). D’Agruma et al. (4) demonstrated that this variant was associated with hypercalcemia and exhibited increased transcriptional activity compared to the wild-type protein. In the same way, Marchiori et al. (15) showed that this variant was associated with increased serum Ca levels. On the contrary, Guan et al. (5) demonstrated that the transcriptional activity of this variant was similar to the wild-type sequence. This variant is not localized in the CCID domain where activating mutations have mostly been identified. The mutated position is not well conserved among species and in silico predictions mainly predict the variant as benign (six prediction software considered it as benign vs three pathogenic). Frequency of p.(Tyr282Asp) is currently of 1.23% in ESP database (European American). This frequency over 1% in population could tilt the balance in favor of the interpretation of this variant as a benign polymorphism, even if the threshold of 5% is not reached, which would constitute a stand-alone evidence of benign impact according to ACMG–AMP guidelines. Anyway, an enrichment of this sequence variation is noticed in patients with PHPT, as initially described by D’Agruma et al. (4). It is also the case in our cohort in comparison with the control population of gnomAD v2.1 database (30/1426 VS. 2549/282896, OR 2.3 (IC95: 1.6 to 3.4)), but too moderately to be considered as an argument for a pathogenic effect. Moreover, in our cohort, this variant has been identified in two patients with pathogenic variation in another gene: one patient had a pathogenic inactivating substitution in CASR (NM_000388.3 c.164C>T) and the other patient had a deletion in MEN1 (NM_130799.2 c.776del). Finally, the selected criteria for ACMG classification could be as follows: PS4 (The prevalence of the variant in affected individuals is significantly increased compared with the prevalence in controls) + BP4 (Multiple lines of computational evidence suggest no impact on gene or gene product (conservation, evolutionary, splicing impact, etc…)) resulting in classification as a class 3 variant, that is variant of uncertain significance. PP5 criteria (Reputable source recently reports variant as pathogenic, but the evidence is not available to the laboratory to perform an independent evaluation) was not applied because the functional effect of this variant is controversial in literature. All these elements could suggest that this sequence variation is a phenotype modifier variant, as it has been described for the frequent variant p.(Ala986Ser) in CASR, described as contributor to the pathogenesis but not as being responsible for the development of the pathology. Probably more experience and family studies will allow to classify more clearly this variant.

Among the 44 patients harboring a class 3 variant, excepting p.(Tyr282Asp) variant, 14 patients harbored another variant. This represents about 2% of our cohort which is not negligible. This highlights the necessity for carrying on the investigations on this gene.

Unlike p.(Tyr282Asp), the pathogenicity of the variant p.(Tyr394Ser), localized in the CCID domain, is well admitted in literature (14). It has been shown that this variant is enriched in Ashkenazi Jewish kindreds with PHPT (16). It was identified in three index cases in our cohort. The ages of onset were 28, 46 and 78 years. The youngest was a male who presented with severe hypercalcemia linked to unique adenoma and family history of PHPT, the second one was a female with recurrent parathyroid hyperplasia and the third one was a male with PHPT, but no more clinical information was available. Thus, heterogeneity of clinical and anatomopathological presentations are observed with this variant. Incomplete penetrance associated with this GCM2 sequence variation was also highlighted by the family study presented in this article, with two siblings of more than 55 years old presenting with normalcalciumand phosphate levels. Our observation is in agreement with the recent article of Riccardi et al. (17) who studied GCM2 variants localized in CCID in sporadic PHPT and concluded that the penetrance of these variants was low. We thus classified this variant as a class 4 variant (likely pathogenic).

Notwithstanding the limits of our study due to lack of some biological and anatomopathological data, we suggest that GCM2 sequencing should be included in the systematic genetic screening of PHPT for patients with a familial history of PHPT or exhibiting a sporadic presentation before the age of 50 years or with recurrent/persistent PHPT or with multiple gland parathyroid lesions. Next-generation sequencing facilitates this process (18). However, the interpretation of the sequence variations identified in GCM2 is quite difficult, due to the lack of functional studies available in routine diagnosis laboratory to determine the consequences of the variants, and the incomplete penetrance of likely pathogenic variants. This also highlights the importance of discussion between geneticists and clinicians in multidisciplinary concentration meetings and validation of variants by expert groups such as TENGEN in France, which achieves it for many years for MEN1 variants (19). So, in front of incomplete penetrance and variants of uncertain significance, what about management of patients harboring those variants? Follow-up of asymptomatic patients harboring such a variant is not excessively binding. They should just be monitored for calcemia and PTH to be able to detect the onset of the pathology, which can happen very late in life indeed never as the penetrance is incomplete.

To conclude, we propose to include this gene in panels for PHPT screening. The proposition of genetic screening for all patients of less than 50 years with PHPT appears to be still valid, as diagnosis of genetically determined PHPT may be younger for patients with HRPT2 or MEN1 mutation but older for patients with CASR or GCM2 mutation. Up till now, many GCM2 variants are classified as variants of uncertain significance. However, improvement in the future of our knowledge about this gene will help us in the tricky interpretation and classification of such variants.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of this study.

Funding

This work was supported by funding from CHU de Lille, Assistance Publique Hôpitaux de Marseille, Hospices Civils de Lyon, Assistance Publique Hôpitaux de Paris, France.