Introduction

The hereditary motor and sensory neuropathies (HMSNs) are a clinically heterogeneous group of peripheral neuropathies, characterized by slowly progressive weakness and atrophy of the distal limb muscles [1]. The prevalence of all types has been estimated at 1 in 10,000 [2]. HMSN type I or Charcot-Marie-Tooth disease type 1 (CMT1) is the most common form. Clinically, CMT1 is characterized by pes cavus, reduced or absent deep-tendon reflexes, and hypertrophic nerves. The age of onset of the symptoms is the first or second decade, with a considerable variation among CMT1 patients ranging from almost no symptoms to severe weakness, atrophy, and foot deformity. However, all CMT1 patients have severely reduced nerve conduction velocities (NCVs) and segmental de- and remyelination on nerve biopsy.

Positional cloning has shown that CMT1 is genetically heterogeneous with at least four distinct loci. The major autosomal dominant subtype CMT1A is linked to chromosome 17p112 [3, 4], a minor subtype CMT1B is linked to chromosome 1 in the region 1q22-q23 [5, 6] and a third, still unassigned subtype CMT1C is not linked to either of these loci [7]. An X-linked dominant locus was mapped to chromosome Xq13 [8]. In the majority of the CMT1A patients, the disease is associated with a tandem DNA duplication of 1.5 Mb [9, 10]. This duplication is also the cause of the disease in the majority of the sporadic CMT1 cases [11, 12]. The peripheral myelin protein 22 gene (PMP22) was found to be located within the CMT1A duplication, suggesting that overexpression of this gene causes the CMT1A disease phenotype [13–16]. Point mutations in PMP22 in nonduplicated CMT1A patients confirmed the direct role of the gene in the CMT1A disease process [17–20]. The myelin protein zero gene (MPZ), the major myelin gene of the peripheral nerve, has been assigned to chromosome 1 in the region where CMT1B was previously mapped by linkage analysis studies [5]. Several distinct mutations in MPZ cosegregating with the disease identified MPZ as the CMT1B gene [21–26]. The gene encoding connexin 32 (Cx32), a gap junction protein, was found to be located in the CMTX candidate region at Xq13.1. Mutations were found in 24 out of 27 X-linked CMT1 families [27–30]. The expression of Cx32 in the peripheral nerve was not known before [27].

Hereditary neuropathy with liability to pressure palsies (HNPP), also called tomaculous neuropathy, is characterized by periodic episodes of numbness and palsies that follow relatively minor compression or trauma to the peripheral nerves. Electrophysiology may reveal reduced motor and sensory NCVs in clinically affected patients and asymptomatic carriers. On nerve biopsy, tomaculae or sausage-like structures are present. HNPP is inherited as an autosomal dominant trait [31]. The disease is usually associated with a deletion in chromosome 17p11.2 [32]. In the vast majority of cases, the deletion comprises all markers that are duplicated in CMT1A. Furthermore, the duplication and the deletion arise from the same unequal crossing-over event at the CMT1A-REP site, a repeat sequence flanking the CMT1A region [33, 34]. Therefore, the CMT1A duplication and the HNPP deletion are reciprocal mutations. A decreased dosage of PMP22 is a possible cause of HNPP. This hypothesis is supported by the fact that a nondeleted HNPP patient was identified who carried a 2-bp deletion in PMP22 causing early termination of transcription [35].

To estimate the frequencies of the CMT1A duplication and the HNPP deletion in Europe, and to make an inventory of the different mutations in PMP22, MPZ, and Cx32, the data of several European research centers were pooled. A first European Neuromuscular Center (ENMC)-sponsored workshop on hereditary motor and sensory neuropathies was organized in Baarn, The Netherlands, in May 1991 [36]. Here it was decided to establish diagnostic criteria for HMSN type I, and to define the requirements for individuals to be included in a linkage analysis [36]. Briefly, these criteria are: (1) slowly progressive symmetrical muscle wasting and weakness, predominantly of the distal part of the lower limbs; (2) severely decreased motor median NCVs (≤30 m/s), absence or marked decrease of sensory nerve action potentials (SNAPs) in the lower limbs and/or a sensory nerve biopsy consistent with a diagnosis of demyelinating neuropathy, and (3) pedigree consistent with autosomal dominant inheritance. The low cutoff NCV value of 30 m/s was used since such low NCV values have been observed only in CMT1 patients, ensuring that only families with a CMT1 phenotype would be ascertained in the linkage analysis studies. At this meeting, a European CMT consortium group was also formed. A second ENMC-sponsored workshop was organized in Den Dolder, The Netherlands, in December 1992 [37]. Members of the CMT consortium group involved in DNA diagnosis of CMT patients were sent a questionnaire to assess the frequency and the size of the 17p11.2 duplication in their patients with a CMT1 phenotype, to determine the frequency of new mutation cases in sporadic CMT1 patients, and to evaluate possible screening methods for the CMT1 duplication. Only data obtained on patients with a well-defined phenotype were included. Selection criteria for the familial cases were as defined at the first workshop. The clinical and electrophysiological criteria were the same for the sporadic CMT1 patients. In addition, both parents had to be clinically and electrophysiologically normal and analyzed for the duplication, and paternity of the patients had to be confirmed. Data from 20 laboratories in 11 countries were retained in the final analysis, resulting in a duplication frequency of 84.6% of the familial CMT1 cases (total of 273 families), and a de novo duplication frequency of 93% (total of 28 patients) [37]. In 9 of the de novo duplications, the parental origin could be deduced: all 9 were paternally derived [12].

At the 26th annual meeting of the European Society of Human Genetics in Paris, France, in June 1994, a meeting of the European CMT consortium group was organized to update the data on the duplication frequency in CMT1 patients, to make a first inventory of the type and frequency of mutations in PMP22, MPZ, and Cx32, and to estimate the deletion frequency in patients with HNPP. All members of the CMT consortium group were asked to complete a questionnaire concerning duplication, deletion, and mutation screening of CMT1 and HNPP patients. The CMT1 patients had to fulfil all clinical and electrophysiological criteria as defined at the first workshop. However, since we wanted to ascertain all CMT’l patients, a cutoff value of 38 m/s for the motor median nerve was used, as proposed by Harding and Thomas [38] on the basis of motor median NCV measurements in 170 CMT patients [38]. Furthermore, no restrictions were made with respect to patients’ family history, since we also wanted to assess the duplication/deletion frequencies in clinically isolated and genetically sporadic cases. For HNPP patients, no consensus diagnosis was available and thus each center used their own diagnostic criteria. We received the results from 28 centers, the data were compiled and are discussed in this paper.

Methods

Duplication and Deletion Screening

The most common technique used in the different participating centers to detect the duplication was the scoring of three alleles or dosage differences on Southern blot hybridization: intensities of heterozygous RFLP alleles were visually compared one versus the other. Subclones [39] of the following markers, free of repetitive sequences, were provided by the European CMT consortium group: pVAW409R1, pVAW409R3, pVAW412R3, or pEW401. One or more of these markers, or the PMP22 probe p132G8Rl [13], were used in the analysis. The scoring of triple alleles was also done using the microsatellite markers RM11-GT and Mfd41 [10, 40]. In a few laboratories, the duplication was identified by quantification of alleles of fluorescent-labeled microsatellite markers RM11-GT, Mfd41, AFM191xh12, AFM317yg1, and AFM200yb12, using ABI GENESCAN 672 software (Applied Biosystems, Foster City, Calif., USA) [41, 42]. A few centers used additional techniques: fluorescence in situ hybridization (FISH) or pulsed-field gel electrophoresis (PFGE). FISH analysis detects three spots of a marker located within the CMT1A region compared to two spots of a marker located outside the duplication [10]. With PFGE, a novel junction fragment of 500 kb can be detected in DNA digested with different rare cutter restriction enzymes such as SacII, FspI, or AseI, and hybridized with the marker pVAW409R3 or probes from the CMT1A-REP region [10, 15, 33, 39].

Similar methods were used to detect the deletion in HNPP patients: scoring of loss of alleles on Southern blot hybridization or microsatellite markers and FISH analysis. In this study, PFGE was not used in the diagnosis of HNPP, although it is now possible to detect junction fragments in HNPP patients with markers from the CMT1A-REP region [19, 34, 43, 44].

Mutation Analysis

The techniques used for mutation screening in CMT1 and HNPP were single-stranded conformation polymorphism (SSCP) analysis, heteroduplex analysis or direct sequencing of the coding regions of the myelin genes PMP22, MPZ, or Cx32. SSCP and heteroduplex analysis do not detect all the mutations. The sensitivity of SSCP and heteroduplex analysis is estimated at 80% [45]. Direct sequencing should reveal all mutations localized in the coding region of the genes. However, since this technique is very time-consuming and expensive, it was not routinely used in the DNA diagnostic laboratories.

Results

CMT1A Duplication Frequency

A total of 881 unrelated CMT1 patients were tested for the presence of the CMT1A duplication by one or more of the techniques mentioned above. 819 patients were informative for the duplication analysis, i.e. 579 patients were duplicated and 240 patients were not. The remaining 62 patients were uninformative. These data resulted in a duplication frequency of 70.7% (table 1). Part of the duplication screening results of the individual participating centers have been published elsewhere [9, 39, 46–54]. Only 6 out of 579 (1.0%) patients were found to have a smaller duplication. One of these cases was described by Palau et al. [12]. The CMT1A duplication frequency varied significantly between the different centers. The highest frequency was 100.0%, the lowest 34.3% (table 1). In principle, the duplication frequencies could be biased towards higher frequencies if some of the centers would have included related CMT1 patients. However, to the best of our knowledge, the patients under study were unrelated, since the different centers contributed only 1 patient sample per CMT1 family. Also, the differences in frequency did not seem to reflect a different ethnic origin, since high and low frequencies occurred in the same country. However, in more isolated populations, like northern Sweden, the low duplication frequency (37.5%) could be caused by a relatively higher frequency of recessive CMT1 cases [51].

Table 1 Patients tested for the CMT1A duplication and the HNPP deletion
Full size table

The duplication frequency of the familial cases, i.e. cases with at least one other known CMT1 patient in the family, was 75.9% (477/628). If we selected only the proven autosomal dominant cases, the duplication frequency was 85.2% (table 1). When we also included the dominant cases, i.e. cases belonging to families without male-to-male transmission, the duplication frequency decreased to 78.4%, because in the latter group, the duplication frequency was only 63.1% (111/176). An explanation for this low duplication frequency could be that this group of dominant cases also comprises patients with X-linked CMT1.

In clinically isolated cases, i.e. cases which had no family history of the disease or cases for which data on family history were not available, it was 53.4% (102/191). If only genetically sporadic cases were considered, i.e. cases with both parents clinically and electrophysiologically normal, analyzed for the duplication and paternity confirmed, a de novo duplication was observed in 76.5% of the cases (table 1). Some of the individual participating centers have published their data elsewhere [11, 12, 55]. The de novo duplication cases represent 6.7% of the total number of duplicated CMT1 patients (39/579).

HNPP Deletion Frequency

162 HNPP patients were screened for the deletion. 156 patients were informative for the deletion analysis, i.e., 131 patients had the deletion, 25 were not deleted, while the remaining 6 patients were not informative. These data resulted in an overall deletion frequency of 84.0% in the HNPP patients. Deletion screening results of some individual participating centers have been published [44, 56–59]. 5 out of 131 (3.8%) of the deleted HNPP patients had a smaller deletion; 1 of these patients is described by Chapon et al. [submitted].

The deletion frequency in the familial cases was 86.1% (105/122), that in the proven autosomal dominant cases was 87.6% (table 1), and in the dominant cases 89.5% (17/19). These comparable deletion frequencies suggest that an X-linked dominant form of HNPP is rather unlikely.

The deletion frequency in the clinically isolated cases was 76.5% (26/34). In 85.7% of the genetically sporadic cases, a de novo deletion was present (table 1). The de novo deletion cases represent 4.6% (6/131) of the total number of deleted HNPP patients.

Mutations in Myelin Genes

Mutation screening of the PMP22, MPZ, and Cx32 genes was performed in 13 of the participating centers in 40.8% (98), 44.6% (107) and 15.0% (36), respectively, of the nonduplicated CMT1 patients. In 3 out of 98 (4.1%) patients, a sequence variation in PMP22 was detected by SSCP analysis. In one of these patients, the mutation has been identified by sequencing (table 2). A PMP22 mutation was also identified by sequencing in a CMT1 patient with no altered SSCP pattern [20]. In MPZ, 14 sequence variations were observed by SSCP or heteroduplex analysis in 107 patients (13.1%). 13 MPZ mutations have already been identified by sequencing (table 2). In addition, 2 MPZ mutations were found in patients not included in table 1, since they did not fulfil the inclusion criteria (no NCV values were available, table 2). Screening of the Cx32 gene by SSCP analysis and sequencing revealed 10 mutations in 36 patients (27.8%, table 2). A further 17 mutations in Cx32 were identified, 6 in patients not fulfilling all inclusion criteria of this study, 10 in proven X-linked families, and 1 in a patient not informative for the CMT1A duplication (table 2).

Table 2 CMT1 and HNPP mutations
Full size table

Eight out of the 25 (32.0%) nondeleted HNPP cases were screened for mutations in the PMP22 and MPZ genes. SSCP analysis revealed one sequence variation in exon 1 of PMP22 segregating with the disease in an autosomal dominant HNPP family. Sequencing identified a splice donor site mutation [Bort, pers. commun.].

Discussion

In this European collaboration study on CMT1 and HNPP mutation frequencies, 70.7% of 819 unrelated CMT1 patients had the CMT1A duplication. A similar duplication frequency of 68% was found in a smaller study of 63 unrelated CMT1 patients from the USA [60]. The duplication frequencies in the familial cases was 75.9%, in the clinically isolated cases 53.4%. The duplication frequency in the autosomal cases was 85.2%, that in genetically sporadic cases 76.5%. In 84.0% of the 156 unrelated HNPP patients, the 17p11.2 deletion was present. The nonavailability of strict diagnostic criteria for HNPP hampered the evaluation of the results from the different centers. Consequently, the frequencies varied widely among the different centers. The deletion frequency in the familial cases was 86.1%, in the clinically isolated cases 76.5%. The deletion frequency in the autosomal cases was 87.6%, that in genetically sporadic cases 85.7%. Only 1.0% of the duplicated CMT1 patients had a smaller duplication, while 3.8% of the HNPP patients had a smaller deletion. These numbers are possibly an underestimation of the real number of smaller duplications/deletions, since not all smaller duplications/deletions may be recognized if not all markers of the CMT1A region were analyzed. However, a recent report demonstrates that the same 500-kb junction fragment was found in 512 CMT1A duplication patients [61], confirming that a smaller duplication/deletion seems to be a very rare event in CMT1A and HNPP disease. The fact that the duplication and the deletion comprised exactly the same region in the vast majority of the CMT1 and HNPP patients supported the hypothesis that the duplication/deletion arises from an unequal crossing-over event at the CMT1A-REP site, the repeat sequence flanking the CMT1A/HNPP region [33, 34]. The high frequency of de novo duplication/deletion cases in CMT1A and HNPP suggests a relative high mutation rate at 17p11.2. The high duplication/deletion frequencies in CMT1/HNPP patients indicate that the duplication/deletion screening is an important molecular genetic tool allowing a molecular diagnosis in the majority of these patients.

In 7.0% (62/881) of the CMT1 patients and 3.7% (6/162) of the HNPP patients, the duplication/deletion screening was not informative. The informativeness of the duplication/deletion screening depends on the detection method. The FISH and PFGE methods are 100% informative, but are not routinely used in most centers. In the Southern blot hybridization and microsatellite analysis, the informativeness depends on the heterozygosity of the alleles. A method to partially solve this problem is to use a nonduplicated reference marker in the Southern blot hybridization to compare dosages in homozygous patients. In general, it is difficult to quantify PCR products, particularly when radioactive labeling is used. Therefore, in microsatellite analysis, the presence of the duplication is most reliably diagnosed if three alleles are present [60]. However, automated analysis of fluorescence-labeled microsatellite markers allows the identification of the duplication by quantification of the alleles [41, 42].

Mutation screening of the CMT1 myelin genes in non-duplicated CMT1 patients is not yet a routine technique in most laboratories. Only part of the nonduplicated CMT1 cases were screened by SSCP analysis and/or direct sequencing. Therefore an exact estimation of the frequency of CMT1 mutations cannot be made. Based on the total number of mutations in the CMT1 genes reported here and by others, it is clear that mutations in Cx32 are more frequent than MPZ mutations, and that they are both much more frequent than PMP22 mutations. Therefore, for diagnostic purposes, nonduplicated CMT1 patients should be screened first for mutations in Cx32 (unless male-to-male transmission occurred in the family) and then in MPZ, before screening for PMP22 mutations. Eight of the 25 nondeleted HNPP patients were tested for mutations in PMP22 and/or MPZ. In 1 patient, a splice donor site mutation in PMP22 was identified [Bort, pers. commun.]. Since not all HNPP patients had a deletion or a mutation in PMP22 and since linkage analysis in some of the nondeleted families excluded a large part of chromosome 17p [62], the data confirmed that HNPP is genetically heterogeneous with a locus on 17p11.2 and at least one other locus still to be identified.