Genotype-Phenotype Correlation in Fibrous Dysplasia/McCune-Albright Syndrome

Abstract

Context

Fibrous dysplasia/McCune-Albright syndrome (FD/MAS) is a rare bone and endocrine disorder resulting in fractures, pain, and disability. There are no targeted or effective therapies to alter the disease course. Disease arises from somatic gain-of-function variants at the R201 codon in GNAS, replacing arginine by either cysteine or histidine. The relative pathogenicity of these variants is not fully understood.

Objective

This work aimed 1) to determine whether the most common GNAS variants (R201C and R201H) are associated with a specific clinical phenotype, and 2) to determine the prevalence of the most common GNAS variants in a large patient cohort.

Methods

This retrospective cross-sectional analysis measured the correlation between genotype and phenotype characterized by clinical, biochemical, and radiographic data.

Results

Sixty-one individuals were genotyped using DNA extracted from tissue or circulating cell-free DNA. Twenty-two patients (36.1%) had the R201C variant, and 39 (63.9%) had the R201H variant. FD skeletal disease burden, hypophosphatemia prevalence, fracture incidence, and ambulation status were similar between the 2 groups. There was no difference in the prevalence of endocrinopathies, ultrasonographic gonadal or thyroid abnormalities, or pancreatic involvement. There was a nonsignificant association of cancer with the R201H variant.

Conclusion

There is no clear genotype-phenotype correlation in patients with the most common FD/MAS pathogenic variants. The predominance of the R201H variant observed in our cohort and reported in the literature indicates it is likely responsible for a larger burden of disease in the overall population of patients with FD/MAS, which may have important implications for the future development of targeted therapies.

Fibrous dysplasia/McCune-Albright syndrome (FD/MAS) is a rare mosaic disorder with broad clinical manifestations, including skin hyperpigmentation, skeletal fibrous dysplasia, and hyperfunctioning endocrinopathies (1, 2). The syndrome is caused by gain-of-function missense pathogenic variants in GNAS, which encodes the ubiquitously expressed activating G protein subunit Gα _S (3). Vertical transmission has not been reported; therefore, germline gain-of-function pathogenic variants in GNAS are thought to be embryonic lethal. Instead, sporadic postzygotic pathogenic variants lead to mosaic expression of constitutively active Gα _S in only a subset of tissues (4).

The clinical spectrum in FD/MAS is broad, and the specific manifestations depend on the tissues that bear the pathogenic variant. Because G protein–coupled receptor activation regulates production of several hormones, protracted Gα _S signaling in endocrine organs can lead to precocious puberty, hyperthyroidism, growth hormone (GH) excess, or Cushing syndrome. Structural changes are commonly observed on ultrasound of affected thyroid glands, ovaries, and testes (1, 2, 5-11) (Fig. 1B and 1D). Hyperpigmented macules with characteristic jagged borders result from increased tyrosinase gene expression and melanin production in affected melanocytes (12) (Fig. 1C). Altered signaling in bone leads to inappropriate proliferation of immature osteoblast progenitor cells, leading to localized FD lesions, which are structurally unsound and prone to fracture and deformity (13, 14) (Fig. 1A). Local expansion of these lesions may impinge on neighboring structures, such as cranial nerves, contributing to pain and loss of function. A sufficiently large burden of skeletal disease leads to excess production of the hormone fibroblast growth factor 23, causing variable degrees of renal phosphate wasting and frank hypophosphatemia, which may result in rickets/osteomalacia (15, 16). Hepatobiliary and pancreatic manifestations, including hepatic adenomas, pancreatitis, intraductal papillary mucinous neoplasms (IPMNs), and cancer have also been reported in individuals with FD/MAS (17) (Fig. 1E). GNAS was initially described as a weak oncogene (18), and several cancers have been reported in association with FD/MAS, including osteosarcoma, breast, thyroid, and testicular carcinoma (5, 19-21).

Figure 1.

Representative images of clinical features evaluated in patients with fibrous dysplasia/McCune-Albright syndrome. A, ¹⁸F-sodium fluoride (NaF)–positron emission tomography/computed tomography scan showing increased tracer uptake in areas of fibrous dysplasia, including the skull, ribs, and extremities (red arrows). Skeletal burden score is quantified on a scale from 0 to 75 using previously established and validated methodology (22). This patient has a burden score of 27. B, Thyroid ultrasound showing typical heterogeneous hyperechoic and hypoechoic areas. C, Hyperpigmented macules classically appear with irregular borders and reflect along the midline of the body. D, Testicular ultrasound shows diffuse microlithiasis. E, Magnetic resonance cholangiopancreatography shows a T2 hyperintense lesion within the inferior portion of the pancreatic head (red arrow), consistent with an intraductal papillary mucinous neoplasm.

The phenotypic diversity of FD/MAS has been attributed to differential developmental timing of the GNAS pathogenic variant, with variants that occur earlier in embryogenesis leading to broader tissue involvement and more severe disease. However, most patients who come to clinical attention have disease in tissues derived from 2 or more germ layers, which must result from pathogenic variants occurring prior to gastrulation. Moreover, since GNAS is an imprinted locus and the most common FD/MAS-associated pathogenic variants are C-to-T transitions in a CG dinucleotide, it has been proposed that these variants arise during a massive methylation event that occurs between the morula and blastocyst stages of embryogenesis (23). Given this narrow developmental window, it is likely that other factors that affect the proliferation, survival, migration, and metabolic activity of these cells contribute to the highly variable disease phenotype.

Almost all the known FD/MAS-associated pathogenic variants occur at the conserved R201 (CGT) codon of exon 8, resulting in replacement of the arginine by either a cysteine or histidine residue (24-26). Pathogenic variants in the Q227 codon of exon 9 have also been reported in a small number of patients (25). It is unclear whether different pathogenic variants alter the catalytic or signaling activity of Gα _S to a different extent, since only a few in vitro studies report a direct comparison between variants. Landis et al (18) demonstrated about 2.5-fold higher basal adenylyl cyclase activity in a human GH–secreting adenoma bearing the R201H variant compared to adenomas with the R201C or the much rarer Q227R variants. However, in indirect in vitro assays performed in the same study, all 3 pathogenic variants decreased the GTPase activity of rat Gα _S to a similar degree (~ 30×). Bhattacharyya et al (27) reported higher cyclic adenosine 5′-monophosphate production in cell lines expressing the R201H variant compared to those expressing the R201C variant. However, in a study by Celi et al (8), both variants increased the basal transcriptional activity of the D2 deiodinase promoter to the same degree compared to wild-type Gα _S. The relative inherent pathogenicity of these variants has not been previously explored in animal models or clinical studies.

Understanding the relative pathogenicity of disease-causing variants in FD/MAS is important to identify prognostic markers, and to develop therapies targeting Gα _s, which should prioritize variants that cause a larger burden of disease. The purpose of this study was to 1) determine whether the most common pathogenic variants in GNAS (R201C and R201H) are associated with a specific clinical phenotype, and 2) to determine the prevalence of the most common pathogenic variants in GNAS in a large cohort.

Materials and Methods

Study Design

Patients were evaluated at the National Institutes of Health between 1999 and 2020, as part of a long-standing FD/MAS natural history study (NCT00001727). The study was approved by the institutional review board of the National Institute of Dental and Craniofacial Research (NIDCR). Informed written consent/assent was obtained from all participants and/or their guardians for clinical and genetic testing. The diagnosis of FD/MAS was established based on clinical parameters, using previously reported guidelines (1, 28). Clinical data were obtained through history, physical examination, and physiatry evaluation. Biochemical tests and imaging studies, including bone-age x-ray and thyroid and pelvic ultrasound, were performed per protocol to assess for the presence of associated endocrinopathies (Fig. 1B and 1C). FD/MAS-associated pancreatic involvement was assessed by detailed gastrointestinal history and biochemical markers of pancreatitis (amylase and lipase). Additionally, contrast-enhanced magnetic resonance imaging and magnetic resonance cholangiopancreatography for detection of IPMNs were performed in patients evaluated after 2013 who could tolerate the test without sedation (Fig. 1E). Skeletal FD burden was determined by whole-body ⁹⁹Tc-methyl diphosphonate and/or ¹⁸F-sodium fluoride (NaF) positron emission tomography/computed tomography bone scans and quantified using a previously reported validated scoring method (22, 29) (Fig. 1A). Retrospective analysis was performed to assess the clinical severity of FD and prevalence of extraskeletal manifestations in patients with detected GNAS variants.

Variant Detection Techniques

Dysplastic bone and thyroid tissue specimens were collected under the natural history study protocol from patients undergoing orthopedic surgery, bone biopsy, or thyroidectomy as part of their clinical care. Serum was collected at the time of routine blood sampling for biochemical testing. Genomic DNA was isolated from frozen tissue specimens using the DNeasy Blood & Tissue Kit and from paraffin-embedded bone specimens using the QIAamp DNA formalin-fixed paraffin-embedded (FFPE) Tissue Kit. Cell-free DNA was isolated from fresh or frozen serum using the QIAamp MinElute circulating cell-free DNA (ccfDNA) Kit (all products by Qiagen). Genotyping was performed by competitive allele-specific TaqMan polymerase chain reaction (PCR) (Applied Biosystems) using primers designed to amplify the R201C and R201H alleles of GNAS. Data were analyzed using the QuantStudio Design & Analysis Software, version 1.4, and MutationDetector Software (both products by Thermo Fisher).

Statistical Analysis

Statistical analyses were performed using GraphPad Prism version 7.0c for Mac OS. Results were expressed using median (interquartile range) (IQR) or odds ratio (OR), as described. Comparison between groups was performed using Mann-Whitney and Fisher exact tests for continuous and categorical variables, respectively. A P value less than .05 was considered significant.

Results

Study Population

A total of 240 patients with FD/MAS were evaluated as part of the natural history study. Serum samples were collected from all participants on a prospective basis, and surgical specimens were collected as waste material from clinically indicated procedures. Twenty-six individuals had surgical specimens available for genotyping and were included in the analyses. To preserve availability of serum samples for future studies, samples were chosen from a subset of 65 patients based on the following criteria: 1) the availability of one or more stored serum sample, and 2) the availability of 4-mL or more total stored serum. Six participants had genotyping performed both in tissue and serum to evaluate consistency between techniques. Demographic characteristics of individuals in whom genotyping was attempted was similar to the overall natural history study cohort (age 20.6 [IQR, 12.3-34.5] vs 19.7 [9.9-34.6], P = .207, sex (male/female) 39/61 vs 37/63, P = .890, race/ethnicity: White 77% vs 74%, Black/African American 2% vs 7%, Hispanic/Latino 8% vs 8%, Asian 9% vs 6%, other/unknown 3% vs 5%). Genotyping was performed and gain-of-function GNAS pathogenic variants were detected in affected tissues or ccfDNA of 61 patients (Fig. 2). The 6 participants in whom both were available had the same variant identified in tissue and blood.

Figure 2.

Flow diagram. The diagnosis of fibrous dysplasia/McCune-Albright syndrome (FD/MAS) was made clinically based on the presence of 2 or more clinical manifestations of MAS and/or the presence of histologically confirmed FD. *Two patients with no pathogenic variant detected in circulating cell-free DNA (ccfDNA) had a variant detected in the biopsy specimen. †In 6 patients, an identical R201 variant was detected both from the tissue biopsy specimen and ccfDNA.

Table 1 summarizes the demographic characteristics of the study population. There was an almost 2-fold predominance of R201H variants compared to R201C in genotyped patients (63.9% R201H vs 36.1% R201C). There were no significant differences in sex, age at diagnosis, or age at last encounter between the 2 groups. The racial distribution differed between the groups, with a higher proportion of patients with the R201C variant identifying as White and African American and a higher proportion of patients with the R201H variant identifying as Asian.

Table 1.

Patient characteristics

Genotype	Total detectable R201 mutations	R201C	R201H	P C vs H
Total patients, n (%)	61 (100%)	22 (36%)	39 (63%)
Sex, % M/F	44/56	59/41	36/64	.066
Age at diagnosis, y, median (IQR)	3 (2-5)	4 (2-5)	3 (2-7)	.267
Age at last encounter, y, median (IQR)	20 (13-34)	22 (16-44)	20 (12-34)	.408
Race/Ethnicity, %
- White/Caucasian	74%	86%	67%
- Black/African American	2%	5%	0
- Hispanic/Latinx	10%	9%	10%
- Asian	11%	0	18%
- Other/Unknown	3%	0	5%

Abbreviations: C, R201C; F, female; H, R201H; IQR, interquartile range; M, male.

Severity of Skeletal Disease

FD was present in all genotyped patients, and there was no significant difference between individuals with R201C and R201H variants in the location or magnitude of skeletal disease (Fig. 3A and 3B), as assessed by a validated skeletal burden scoring system (median skeletal burden score 55.4 vs 44.4, P = .255). The prevalence of hypophosphatemia (54.5% vs 41.0%, OR 0.58, P = .423) and severity of renal phosphate wasting were similar between the 2 groups (Fig. 3C). There was no association between bone turnover marker levels (alkaline phosphatase, osteocalcin, and N-terminal telopeptides) and genotype (data not shown); however, these levels correlated positively with skeletal FD burden and negatively with age, as previously reported (30). Clinical outcomes, such as lifetime fracture incidence (0.29 vs 0.22 fractures/year, P = .939) and ambulation status (50.0% vs 38.5% requiring assistive devices, OR 0.63, P = .428) were equivalent in patients with different variants (Fig. 3D and 3E).

Figure 3.

Relative prevalence and clinical severity of fibrous dysplasia (FD) in genotyped individuals bearing the R201C or R201H pathogenic variant. A, Prevalence of FD in the craniofacial, axial, and appendicular skeleton. B, FD skeletal burden score. C, Prevalence of hypophosphatemia requiring treatment. D, Prevalence of long-bone fractures, expressed as number of lifetime fractures/age in years. Solid bars represent the median incidence for each group. Error bars represent the interquartile range. E, Ambulation status at last assessment. Solid bars represent the proportion of patients requiring assistive ambulation devices (eg, crutches, walker, wheelchair) in each group. OR, odds ratio.

Endocrine Manifestations

Involvement of all FD/MAS-affected endocrine tissues occurred in both genotype groups (Table 2). The prevalence of ovarian (cysts) and testicular (macroorchidism, microlithiasis, or focal lesions) ultrasonographic abnormalities did not differ between groups (see Table 2). Precocious puberty in male (OR 0.56; CI, 0.09-3.23, P = .648) and female patients (OR 2.10; CI, 0.31-11.76, P = .591) occurred with similar prevalence in both groups. The prevalence of ultrasonographic thyroid abnormalities (OR 2.06; CI, 0.59-6.81, P = .322) and hyperthyroidism (OR 1.03; CI, 0.38-2.90, P > .999) was equivalent between the 2 genotypes. Likewise, other hyperfunctioning endocrinopathies, including GH excess (OR 0.84; CI, 0.27-2.52, P = .778) and neonatal hypercortisolism (OR 0.83; CI, 0.16-5.01, P > .999) occurred with similar prevalence.

Table 2.

Relative prevalence of extraskeletal manifestations in patients with R201C and R201H mutations

Clinical manifestation	R201C, % (n)	R201H, % (n)	Odds ratio
Precocious puberty (females)	77.8 (7/9)	88.0 (22/25)	2.10, P = .591
Precocious puberty (males)	23.1 (3/13)	14.3 (2/14)	0.56, P = .648
US ovarian abnormalities	88.9 (8/9)	92.0 (23/25)	1.44, P > .999
US testicular abnormalities	92.3 (12/13)	78.6 (11/14)	0.31, P = .596
Hyperthyroidism	45.5 (10/22)	46.2 (18/39)	1.03, P > .999
US thyroid abnormalities	72.7 (16/22)	84.6 (33/39)	2.06, P = .322
GH excess	31.8 (7/22)	28.1 (11/39)	0.84, P = .778
Cushing syndrome	9.1 (2/22)	7.7 (3/39)	0.83, P > .999
Pancreatic involvement	36.4 (8/22)	20.5 (8/39)	0.45, P = .229
Cancer	0 (0/22)	10.3 (4/39)	Infinity, P = .287

Abbreviations: GH, growth hormone; US, ultrasonographic.

Other Extraskeletal Manifestations

Pancreatic involvement, manifesting with either pancreatitis or intraductal papillary mucinous neoplasms, occurred in a minority of individuals with FD/MAS with no significant association with either genotype (OR 0.45; CI, 0.15-1.33, P = .229). Cancers, including osteosarcoma (n = 1), leukemia (n = 1), breast carcinoma (n = 1), and papillary thyroid carcinoma (n = 1), occurred only in patients with the R201H variant. However, the overall prevalence of cancer was low (6.6%), and the association did not reach statistical significance (P = .287; see Table 2).

Discussion

Findings from this large study demonstrate that there is no genotype-phenotype correlation between the 2 predominant disease-causing variants in FD/MAS. R201H variants occur at a higher frequency than R201C variants, indicating that R201H variants likely account for an overall larger measure of disease burden in the FD/MAS population.

We used a sensitive variant detection technique and multiple DNA sources to genotype a large group of patients with FD/MAS. The striking predominance of R201H to R201C variants in the study cohort contrasts with our previous supposition that the variants are equally prevalent (31, 32). We found a similar variant distribution in a systematic review of the published literature (Supplementary Table 1) (33), which suggests that this asymmetry is not due to sampling error. Since R201C and R201H variants in GNAS are invariably encoded by the same codons (TGT and CAT, respectively), which are thought to arise by deamination of the same GC dinucleotide on opposite DNA strands (23), it is difficult to envision a biochemical mechanism that would favor the occurrence of one variant over the other. The R201H variant may confer a selective advantage by enhancing the proliferative or survival capacity of affected cells in vivo; however, this is not supported by in vitro studies, which have not consistently demonstrated a difference in the catalytic or signaling activities of R201C- and R201H-variant Gα _S (8, 18, 27). However, because ccfDNA presumably arises from apoptosis of variant-bearing cells, the exaggerated predominance of R201H in patients genotyped by cell-free DNA (71% H) compared to those genotyped from biopsy specimens (56% H) could potentially reflect a higher prevalence or increased apoptosis of variant cells in individuals with the R201H variant.

Retrospective analysis of clinical data obtained as part of a long-standing natural history study revealed no evidence of a genotype-phenotype correlation in the severity of FD or the prevalence of extraskeletal manifestations. Therefore, as molecular confirmation of the FD/MAS diagnosis becomes more feasible with the development of sensitive techniques, the prognostic utility of genetic testing remains limited. However, the higher prevalence of the R201H variant relative to R201C suggests it is responsible for a larger burden of disease in the overall population of patients with FD/MAS. In addition to R201C inhibitors, which are being explored (34), R201H inhibitors should potentially be prioritized as a target for selective Gα _S inhibitor development. Novel therapeutic approaches are particularly critical for FD, because there are no available treatments that effectively and sustainably control its progression.

The genotyping method used in this study was designed to detect only the 2 most common FD/MAS-associated pathogenic variants, R201C and R201H. Thus, it is possible that pathogenic variants at the Q227 position were missed in genotyped individuals without a detectable variant (n = 24). Q227 variants have been reported in only one FD/MAS genotyping study (25) and account for about 1.2% of detectable variants in studies that evaluated both exons 8 and 9 of GNAS (see Supplementary Table 1) (33). It is also possible that other pathogenic variants at the R201 position, such as L and S substitutions, may have been missed (see Supplementary Table 1) (33). Although we cannot exclude the possibility that these rarer variants are intrinsically associated with a specific clinical phenotype, given their exceedingly low prevalence, they likely affect only a small minority of patients and therefore would not significantly confound this genotype-phenotype correlation analysis. Additional studies are needed to identify and phenotype individuals with FD/MAS resulting from rarer pathogenic variants.

Although efforts were made to include patients with a broad range of disease severity, several factors may contribute to the underrepresentation of those with mild disease in the genotyped group. Tissue specimens were obtained only in the setting of clinically indicated surgeries (eg, orthopedic procedures, thyroidectomy), which generally occur in more severe disease and likely resulted in selective genotyping of more severely affected individuals. Although serum for cell-free DNA was obtained from all patients presenting for clinical care, regardless of disease severity, the sensitivity of variant detection by this technique depends on variant prevalence, which is presumed to be higher in patients with more extensive disease. In addition, in patients with FD/MAS only a subset of cells bears the pathogenic variant, and even affected tissues are “mosaics” of wild-type and variant cells. Variability in variant burden between individual patients and/or tissues may therefore confound any inherent differences in variant pathogenicity. This speaks to important unanswered questions regarding the utility of the cell-free DNA technique in FD/MAS, including the following: 1) whether there is an association of quantitative variant burden from blood and tissue with clinical disease severity, in which case cell-free DNA may potentially serve as a prognostic biomarker, and 2) the diagnostic sensitivity of cell-free DNA, in particular for patients with limited disease involvement who often present a diagnostic challenge. While these questions are beyond the scope of the present paper, work is actively ongoing to address these questions and better define the role of the cell-free DNA technique in FD/MAS.

Genomic imprinting adds another layer of complexity to the evaluation of genotype-phenotype correlation in FD/MAS. GNAS is a complex locus that encodes 5 primary transcripts, whose expression is regulated by distinct promoters with differential methylation on maternal and paternal alleles. Gα _S expression is biallelic in bone (35, 36); therefore, parental origin of the variant is unlikely to significantly affect the variability of FD burden between individuals. However, in isolated GH-secreting pituitary tumors, thyroid gland, and gonads, Gα _S is expressed predominantly from the maternal allele (37-40). In a small study of imprinting in FD/MAS, activating GNAS variants have been detected only on the maternal allele in patients with acromegaly and on either allele in patients with precocious puberty, hyperthyroidism, or FD (35). Therefore, based on the available evidence, parental origin of activating GNAS variants would not be expected to significantly affect the severity of the most common endocrinopathies in MAS. Unfortunately, given the low prevalence of the variant allele in cell-free DNA and the limited availability of tissue specimens, the materials available in this study were not sufficient to allow for assessment of variant GNAS imprinting status.

Potential ethnic differences in variant frequency may represent another confounder. A higher proportion of R201C variants were detected in White and Black/African American individuals, and a higher proportion of R201H variants were detected in individuals identifying as Asian or multiracial. However, this study was biased toward a majority-White population, and few reports in the literature include information on ethnicity in genotyped patients with FD/MAS. Additional studies including more diverse cohorts are therefore needed to determine whether these results accurately reflect ethnic differences in the FD/MAS population.

The prevalence of cancer in genotyped patients was 6.6%, representing 4 cancers of varying types, 3 of which (breast, thyroid, and bone) are known to be associated with FD/MAS. The overall malignancy risk in patients with FD/MAS has not been firmly established but appears to be at least mildly increased in comparison to the general population (41, 42). This likely reflects the oncogenic potential of GNAS variants, which are known to play a role in tumorigenesis (43). In patients with FD/MAS malignancy risks are likely multifactorial, depending on tissue type and other genetic, clinical, and environmental factors. Additional studies with larger numbers of affected individuals are needed to determine the relationship between GNAS pathogenic variants and malignancies.

In conclusion, this retrospective analysis of a large cohort of patients with FD/MAS did not reveal any association between genotype and clinical disease severity. While genetic testing may be useful diagnostically in unclear cases, it does not inform evaluation of disease prognosis. However, the identification of R201H as the predominant pathogenic variant will be an important factor in guiding the development of targeted therapies for FD, which continues to lack effective treatment.

Glossary

Abbreviations

ccfDNA

circulating cell-free DNA

FD

fibrous dysplasia

GH

growth hormone

IPMN

intraductal papillary mucinous neoplasm

IQR

interquartile range

MAS

McCune-Albright syndrome

NIDCR

National Institute of Dental and Craniofacial Research

OR

odds ratio

Additional Information

Disclosures: The NIDCR receives funding from Amgen, Inc, for an investigator-sponsored study of denosumab in fibrous dysplasia.

Data Availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.