Article Text


Genotype-phenotype correlation in hereditary multiple exostoses


Hereditary multiple exostoses (HME) is a genetically heterogeneous autosomal dominant disorder characterised by the development of bony protuberances mainly located on the long bones. Three HME loci have been mapped to chromosomes 8q24 (EXT1), 11p11-13 (EXT2), and 19p (EXT3). The EXT1and EXT2 genes encode glycosyltransferases involved in biosynthesis of heparan sulphate proteoglycans. Here we report on a clinical survey and mutation analysis of 42 HME French families and show that EXT1 andEXT2 accounted for more than 90% of HME cases in our series. Among them, 27/42 cases were accounted for byEXT1 (64%, four nonsense, 19 frameshift, three missense, and one splice site mutations) and 9/42 cases were accounted for by EXT2 (21%, four nonsense, two frameshift, two missense, and one splice site mutation). Overall, 31/36 mutations were expected to cause loss of protein function (86%). The most severe forms of the disease and malignant transformation of exostoses to chondrosarcomas were associated withEXT1 mutations. These findings provide the first genotype-phenotype correlation in HME and will, it is hoped, facilitate the clinical management of these patients.

Statistics from

Hereditary multiple exostoses (HME, MIM 133700) is an autosomal dominant condition characterised by the development of multiple osteochondromas (exostoses) from the metaepiphyseal areas of the long bones.1 The exostoses are either sessile (broad based) or pedunculated and they vary widely in size and number. They can be present at birth and continue to appear and grow throughout childhood and into puberty. Exostoses are usually located in the metaphyseal region of the proximal humerus, distal radius and ulna, proximal tibia, and distal femur, but can also be found on the pelvis and scapula. They can cause multiple and severe complications including pain, restricted range of joint movement, deformities and shortening of the long bones, and nerve or blood vessel compression. The most severe complication is malignant transformation of an exostosis into a chondrosarcoma, which occurs in 0.5-2% of cases.2-4 In addition, intra- and interfamilial clinical variability of HME has long been recognised.2 3

HME is genetically heterogeneous as threeEXT loci have been identified so far.EXT1 (MIM 133700) has been mapped to chromosome 8q23-24,5 EXT2 (MIM 133701) to chromosome 11p11-p12,6 andEXT3 (MIM 600209) to chromosome 19p.7 Linkage studies have identifiedEXT1 and EXT2 as the two major disease loci in HME families, whileEXT3 appears to be a minor locus.8 Three additional loci designatedEXTL1, EXTL2, andEXTL3 have been identified and mapped to chromosomes 1 (1p36, 1p11-p12) and 8 (8p12). However, no HME family has yet been linked to these loci.9-11 Loss of heterozygosity (LOH) at the EXT1,EXT2, and EXT3loci has been observed among patients withEXT related and unrelated chondrosarcomas suggesting that the EXT genes are tumour suppressors.12-15

So far, only two EXT genes have been cloned, namely EXT1 andEXT2. 16-18 Both genes encode glycosyltransferases harbouringN-acetyl-glucosamine transferase (GlcNac-T) and D-glucuronic acid transferase (GlcA-T) activities that catalyse the biosynthesis of heparan sulphate proteoglycans (HSPG).19 20 Recent studies suggest that EXT1 and EXT2 may interact in vivo to form a functional oligomeric complex in the Golgi apparatus where the heparan sulphate polymerisation is thought to occur.21

A number of mutations have been reported in HME but the cause of its clinical variability remains unclear.22 23 Very few studies have concerned the correlation of genotype to phenotype.24 We have carried out an extensive mutation analysis of the EXT1 andEXT2 genes in a large series of HME families and investigated whether specific clinical features could be associated with either locus. Here, we show that EXT1mutations caused the most severe forms of the disease and degeneration of exostosis into chondrosarcoma only occurred inEXT1 patients.

Patients and methods


A total of 42 HME French families (representing 217 affected subjects) were investigated. In order to document the severity of the disease and intrafamilial variability, a questionnaire was sent to each patient. Patients were asked to fill in the questionnaires with the help of their physician. In most cases clinical andx ray data were re-examined in collaboration with the physicians and orthopaedic surgeons to verify the accuracy of the answers. No discordance between the answers and our observations were noticed. For evaluating the functional rating, a modification of the scale established by the Musculoskeletal Tumor Society was used.2 25 The overall functional grade was based on five factors (including pain) and four categories of deformities, namely brachymetacarpy, bowing of the forearm, shortening of the forearm, varus or valgus angulation of the knees, and shortening of limbs. For an overall functional rating of “fair”, at least one of the five factors had to have had that rating (table 1). For each patient, the severity of the disease was evaluated from five factors: age at onset, number of exostoses at the time of evaluation, location of the exostoses (vertebral), stature, and functional rating (table 2). For each family, the degree of severity was evaluated from the most affected members.

Table 1

Evaluation of the functional rating

Table 2

Evaluation of the severity of the phenotype


Blood samples were obtained with the written consent of patients and unaffected relatives. The whole set of families was tested by using microsatellite DNA markers closely linked to EXT1(D8S199, D8S198, D8S85), EXT2(D11S905, D11S903, D11S544), and EXT3 loci (D19S840, D19S413, D19S221).8 For mutation detection in probands, the whole coding sequence of theEXT1 and EXT2genes was amplified26 and analysed by single strand conformation polymorphism (SSCP) either on Hydrolink MDE gels or GeneGel Excel 12.5/24 gels. The Hoefer Automated Gelstainer was used for gel silver staining (Pharmacia Biotech). Amplification products showing abnormal patterns of migration were reamplified and sequenced by using the fluorescent dideoxy terminator method on an automatic sequencer ABI.


Fisher's exact test was used for the analyses; p values equal to or less than 0.05 were considered to be significant.



Linkage analyses confirmed the existence of at least threeEXT loci on chromosomes 8, 11, and 19. Among the 42 families, 29 showed strong linkage to EXT1(69%), nine to EXT2 (21%), and one to EXT3 (3%). Interestingly, several pedigrees (7%) gave negative lod scores with markers of chromosomes 8, 11, and 19. Taken together, these results suggest thatEXT1, EXT2, andEXT3 account for about 90% of cases but that at least one additional locus could account for HME. An abnormal pattern of migration and a mutant genotype was found in 36/38 patients. The mutations identified in EXT1 andEXT2 are summarised in table 3. No mutation was detected in two families linked to EXT1despite SSCP analysis and extensive sequencing of allEXT1 exons.

Table 3

Subgroups of the severe type

Among the 36 mutant alleles reported here, 27 (75%) were found inEXT1. Mutations identified included four nonsense mutations, 19 frameshift deletions or insertions, three missense and one splice site mutation. These mutations were almost randomly distributed over the first nine exons. Most of them were novel, except for five previously reported mutations23(table 4). Ninety percent of them caused a premature termination of the EXT1 protein. Among the nine mutations found in theEXT2 gene, four were nonsense mutations, two were frameshift deletions or insertions, one altered a splice site, and two were missense mutations. Most of theEXT2 mutations were novel mutations, randomly distributed over the first eight exons.

Table 4

Summary of EXT gene mutations, clinical data


Based on the phenotypic expression of the disease, two groups of families were recognised. The first group, which included families with severely affected members, was referred to as group S. In the second group (group M), affected members had a moderate phenotype. Group S could be subdivided into four distinct clinical subgroups (table 3). In type IS, the number of exostoses ranged from 10 to 25 with no vertebral location and a height above the 10th centile. Type IIS was associated with a larger number of exostoses (>25). Patients in the type IIIS group had vertebral exostoses and patients belonging to type IVS group had very short stature (equal to or below the 3rd centile).

Clinical data were available for 42 pedigrees and 110 questionnaires from affected subjects were collected (table 4). Among the 42 families, seven belonged to group M (type M). The 35 other families belonged to group S. Among them, 10 had the characteristics of type IS, five of type IIS, eight of type IIIS, two of type IVS, and six families could not be assigned a specific type, as the patients exhibited characteristics of several types. Interestingly, chondrosarcoma was observed in nine patients from seven families, which all belong to group S.

Intrafamilial variability was observed in most families, supporting the previously reported variable expressivity of HME.28 In four families (A1-12, A6-19, B3-32, B3-33), clinical assessment of the affected subjects was in favour of anticipation. The disease showed a gradient of severity ranging from a mild phenotype in the grandparents (one to two exostoses) to a severe form in the probands (17-30 exostoses).


The relationship between genotypes and clinical features are shown in fig 1. Most of the families with a moderate phenotype (type M) were associated with EXT2 mutations, while all type IIS, IIIS, or IVS families except one (B3-32) were associated with EXT1 mutations (table 4). A statistical analysis of the data confirmed that type S (IS, IIS, IIIS, IVS) was significantly associated with EXT1 mutations while type M was associated with EXT2mutations (p=0.0129, Fischer's exact test). Interestingly, the type IS phenotype was associated with either EXT1 orEXT2 mutations, without hotspots or a specific type of mutation in one or the other gene. No correlation was found between the type IIS or type IIIS phenotypes and location of the mutations in the EXT1 gene. By contrast, mutations associated with type IVS (patients with very short stature) were consistently located in the first EXT1exon. It is worth noting that chondrosarcomas were restricted to patients carrying EXT1mutations.

Figure 1

Genotype-phenotype correlation. Distribution of EXT1 and EXT2 mutations among moderate (M) and severe (IS, IIS, IIIS, IVS) types of HME (data from only one person (proband) in each family were used).

The phenotypic variability of HME was noted in 2/3 of the familial forms caused by EXT1 mutations. By contrast, members of families harbouring an EXT2mutation showed clinical homogeneity of the disease except for two families. The family linked to EXT3 belonged to group M. Interestingly, incomplete penetrance was noted in that family, especially in females. This was confirmed byx ray analysis of all family members. No specific phenotype was observed in three families unlinked to the threeEXT loci.


In order to determine the respective contribution of the threeEXT loci in HME, we investigated their distribution among 42 families. Thirty nine families (93%) proved to be linked to one of the three loci but three families showed no linkage to any of them. No linkage to one of the threeEXT-like loci has been reported so far in HME and no EXT-like gene mutation has been identified.27 All these observations support the existence of at least one additional (EXT4) locus in HME.

Mutation analysis showed that EXT1 mutations were involved in 64% of HME families whileEXT2 mutations occurred in 21% of families. These findings differ from previous studies in white patients.23 This may be because of a more efficient detection of mutations which allowed us to detect 95% of them. The failure to detect mutations in two EXT1linked families suggests that mutations might be located either within the promoter or elsewhere outside the coding region.

Mutations within the EXT1 gene were found to be evenly distributed over the first nine exons, while the last exon, encoding part of the conserved carboxy-terminal domain, harboured significantly fewer mutations. Most EXT2mutations appeared to be novel and randomly distributed over the first eight exons. As observed in EXT1, no mutation was identified in the carboxy-terminal region of the protein. The EXT mutations reported here are expected to cause premature translation termination and loss of function as mutated gene products lack one of their catalytic domains putatively located at the C-terminal end of the protein.21

It is worth noting that most EXT1 mutations reported here (19/27) were novel mutations, but the commonEXT1 mutations were also detected. Hence, the 1 bp deletion (1469delT) previously detected in seven families23 was also identified in two of our families. This deletion occurred at the end of a polycytosine tract, known to be a mutation hot spot. Similarly, the previously reported R340C mutation was found in three patients. This amino acid substitution is likely to alter proper functioning of the EXT1 protein. Remarkably, the corresponding arginine residue was substituted for lysine in a Chinese hamster ovary (CHO) cell mutant lacking glycosyltransferase activity.28 Furthermore, McCormick et al 21 have shown that anEXT1 construct containing the R340C mutation was enzymatically inactive in vitro.21

The genotype-phenotype analyses reported here suggested a strong correlation between EXT1 mutations and the severity of HME phenotypes. Statistical analyses firmly established a significant correlation between EXT1mutations and the severe phenotypes (type S). Hence, large numbers of exostoses, short stature, and/or vertebral location were consistently associated with EXT1 mutations. These particularly severe cases were not accounted for by a cluster of specific mutations. Moreover, malignant degeneration of exostoses into chondrosarcomas was also observed in families carryingEXT1 mutations. In keeping with this, it has recently been shown that hereditary osteochondromas and secondary chondrosarcomas were associated with LOH at theEXT1 locus.14 By contrast to a previous report,15 we did not observe loss of the secondEXT allele in the somatic DNA of the benign osteochondromas that we have studied. According to the tumour suppressor model,29 we speculate that singleEXT1 germline mutations would be the first hit causing multiple benign osteochondromas while inactivation of the second allele would trigger malignant transformation. Our preliminary results on somatic DNA from two secondary chondrosarcomas also indicate LOH at the EXT1 locus (not shown).

Similarly, moderate phenotypes were frequently associated withEXT2 mutations, while the group of severe type IS phenotype resulted from mutations in either gene. The intrafamilial clinical variability that we have observed remains unclear at present. One could speculate that mutations or polymorphisms in a modifier gene may modulate the disease phenotype.

Recent studies have shown that the EXT1 andEXT2 gene products exert both GlcA-T and GlcNAc-T activities in vitro19 and biochemical analyses have shown that the two EXT proteins can interact to form homo/hetero-oligomers in vivo.30 Indeed, formation of a stable EXT1-EXT2 complex in vitro is required to elicit maximal GlcA-T activity.21 Recent experiments indicate that EXT1 and EXT2 cannot substitute for each other in transfected cells. However, while EXT1 alone exhibited significant Glc-T and GlcNAc-T activity in transfected cells, EXT2 alone had no glycosyltransferase activity and was unable to compensate for the abrogation of GlcA-T activity caused by a R340C mutation in EXT1. Whether a functional EXT1 protein could partially restore GlcA-T activity in EXT2 mutant cells is still unknown. Such difference in the glycosyltransferase activities of the two EXT enzymes could explain the lesser severity of HME phenotypes caused byEXT2 mutations. Alternatively, EXT2 may play a non-essential role in heparan sulphate synthesis, acting as a chaperone for EXT1.28

In conclusion, our study provides what we believe to be the first evidence of genotype-phenotype correlation in HME, asEXT1 mutations are more likely to trigger malignant degeneration of exostoses. It is hoped that this study will improve the management of patients.


Accession numbers and URLs for data in this article are as follows: GenBank, forEXT1 genomic sequence (135994) and forEXT2 genomic sequence (344921). Online Mendelian Inheritance in Man (OMIM), EXT1 (MIM 133700),EXT2 (MIM 133701), andEXT3 (MIM 600209).

We are deeply grateful to the HME patients and their families for participation in this study. We would like to thank the following physicians for collaboration: P Calvas, M P Cordier, V Cormier-Daire, M M Freund, D Gaillard, B Gilbert, A Joannard, D Lacombe, B Leheup, J F Mallet, P Maroteaux, A Nivelon, S Odent, M O Peter, N Philip, H Plauchu, G Plessis, J C Pouliquen, and D Rosenberg. The authors are grateful to P Margaritte-Jeannin for statistical analyses and W Van Hul for sending EXT primers. This work was supported by a grant from ARC (Association pour la Recherche sur le Cancer).


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.