Hereditary Auditory, Vestibular, Motor, and Sensory Neuropathy in a Slovenian Roma (Gypsy) Kindred

Members of a Roma (Gypsy) family with hereditary motor and sensory peripheral neuropathy (HMSN) and concomitant auditory and vestibular cranial neuropathies were identified in Kocevje, Slovenia. The illness begins in childhood with a severe and progressive motor disability and the deafness is delayed until the second decade. There are no symptoms of vestibular dysfunction. The family structure is consistent with an autosomal recessive pattern of inheritance and the genetic locus for the disorder is linked to the same region of chromosome 8q24 as other Roma families with HMSN and deafness from Lom, Bulgaria (HMSN-Lom). The present study shows that the deafness is caused by a neuropathy of the auditory nerve with preserved measures of cochlear outer hair cell function (otoacoustic emissions and cochlear microphonics) but absent neural components of auditory brainstem potentials. The hearing loss affects speech comprehension out of proportion to the pure tone loss. Vestibular testing showed absence of caloric responses. Physiological and neuropathological studies of peripheral nerves were compatible with the nerve disorder contemporaneously affecting Schwann cells and axons resulting in both slowed nerve conduction and axonal loss. Genetic linkage studies suggest a refinement of the 8q24 critical region containing the HMSN-Lom locus that affects peripheral motor and sensory nerves as well as the cranial auditory and vestibular nerves.

Auditory neuropathy is a disorder of hearing caused by an impairment of auditory nerve function in the presence of intact cochlear outer hair cell function. [1][2][3] The patients have absent or profoundly abnormal auditory brainstem responses beginning with wave I, the component generated by distal portions of the VIII nerve. They also have preserved cochlear receptor functions represented by microphonic potentials generated by both inner and outer hair cells 4,5 and preserved motility of cochlear outer hair cells resulting in faint sounds, otoacoustic emissions, recorded within the ear canal. 6,7 Patients with auditory neuropathy have elevations of pure tone thresholds that can vary from mild to severe. 1,8 However, auditory perceptions dependent on the processing of temporal cues of acoustic signals are particularly affected with markedly impaired speech perception that is out of proportion to the pure tone loss and impaired localization of sound sources. 1 Sev-eral of these patients have an associated hereditary motor and sensory neuropathy (HMSN). 2 Recently, Gypsy families originating from Lom, Bulgaria, 9,10 northern Italy, 11 and a non-Gypsy Bulgarian patient 12 were described with HMSN and deafness. Sural nerve biopsy revealed both a demyelinating disorder and a significant loss of large axons. The hearing loss was described as "sensorineural" in type. The patients had absent auditory brainstem potentials or abnormalities beginning with wave I. No measures of cochlear hair cell function (cochlear microphonic potentials and otoacoustic emissions) were obtained to ascertain if cochlear receptor activities were preserved. The mode of inheritance of the disorder in the Bulgarian and Italian kindred was autosomal recessive and the locus was mapped to the long arm of chromosome 8 (8q24). The disorder has been designated "HMSN-Lom. " We have seen a family of Gypsy extraction in Slo-venia with individuals affected by progressive motor and sensory nerve degeneration before the age of 10 years, and a hearing loss developing in their teens, with a mode of inheritance that also appears to be autosomal recessive. The results presented in this study show that the hearing loss is caused by a neuropathy of the auditory nerve and that cochlear outer hair cell receptor functions are preserved. The affected patients also have a concomitant asymptomatic vestibular impairment. We also show that the gene associated with the disorder is located in the same region as the locus for the Bulgarian and Italian families (HMSN-Lom) and we suggest refinement of the interval on chromosome 8q24 that contains the gene. In addition, our analyses indicate that the disorder in this family is most likely genetically identical to HMSN-Lom, described in Bulgarian and Italian Gypsies.

Subjects and Methods Subjects
The family lives in Kocevje, which is about 60 km from Ljubljana, the capital of Slovenia. We studied 3 living affected individuals (2 males and 1 female) and obtained information from interviews of the elders of the family for members spanning five generations (Fig 1). There appeared to have been 4 affected individuals (2 females and 2 males) who developed difficulties in walking when they were children and who died as children or young adults. The family elders did not have information as to the hearing in these deceased family members. We were unable to ascertain whether these individuals had ever undergone medical examinations. The 3 living affected individuals developed a progressive motor and sensory impairment of the limbs beginning in the first decade of life and a hearing loss manifested in the second decade. The patients did not have symptoms of vestibular dysfunction. The pedigree structure is consistent with an autosomal recessive pattern of inheritance, and no consanguineous marriages were reported. The Slovenian Ethics Committee approved this study and informed consent was obtained from Patient III-9 who underwent sural nerve biopsy.

Audiometric Testing
Auditory functions were measured by using standard clinical procedures including pure tone audiometry, speech compre-hension, tympanometry, middle ear muscle reflexes, otoacoustic emissions to transient stimuli, and distortion products.

Vestibular Testing
Caloric stimulation with warm and cold water was used to obtain oculovestibular reflexes, while recording extraocular movements.

Neurophysiological Testing
Brainstem auditory evoked potentials were tested by using ipsilateral earlobe to vertex recordings with click stimuli presented from TDH-39P ear phones at both slow (2/sec) and rapid (10/sec) stimulus rates. Condensation and rarefaction click stimuli were tested separately at 100 dB, normalized hearing level (nHL). Averages were made from 2,000 trials.
Visual evoked potentials to alternating checkerboard stimulation were tested to monocular stimulation.
Somatosensory evoked potentials to median nerve stimulation were recorded, and averages to 246 trials were made. Somatosensory evoked potentials were absent in all 3 affected individuals.

Genetic Marker Typing
Blood samples were drawn from 19 family members and DNA was obtained by the ethanol extraction method. A set of eight microsatellite markers (D8S558, D8S378, D8S529, 326CA2, 189CA17, 474CA1, D8S256, and D8S1462) in the 8q24 region was typed in 10 members, spanning three generations, including the 3 living affected individuals. Only the samples from those family members that could provide information for linkage analysis (parents of affected individuals, affected individuals, and their siblings) were genotyped. DNA samples of HMSN-Lom patients from various Gypsy groups, representing typical as well as rare haplotypes, were run in parallel with controls. The genotyping protocol was similar for each marker and primers were obtained from Research Genetics. In brief, the polymerase chain reaction mixture included 40 ng of DNA, 50 ng of each primer, one of which was end-labeled with [␣-32 P]ATP, 200 M dNTPs, and 1 U of Taq polymerase, together with the optimal concentration of magnesium chloride (3-5 mM). The polymerase chain reaction times and temperatures for each of the 35 cycles included denaturation for 20 seconds at 94°C, annealing for 20 seconds at about 55°C, and extension for 20 seconds at 72°C. A 5-minute final extension at 72°C was then performed. Electrophoresis was performed on a 6% denaturing polyacrylamide-sequencing gel for about 2 hours using 1 L of the polymerase chain reaction product. The DNA fragments were visualized by overnight autoradiography. The program LODLINK 13 was used for LOD score calculations.

Neurological Findings
The 3 affected individuals (Patient III-8, age 33 years; Patient III-9, age 31 years; and Patient IV-11, age 22 years) had muscle atrophy typical of HMSN phenotypes but with the addition of significant hearing loss. The first motor symptoms of the disease appeared at the age of 4, 8, and 5 years, respectively, and hearing loss developed later at 10, 13, and 15 years, respectively. They all had marked difficulties in walking by their late teens. The mental state was normal without signs of cognitive impairments. Facial muscle strength tested by maintained eye closure against resistance was reduced. The patients could hear sounds but could only understand speech by using lip reading. There was marked atrophy of the distal limb muscles, with claw deformities of the hands and equinovarus deformities of the feet. Patient IV-11 had polydactyly in the hands. There was less severe weakness or atrophy of neck or proximal muscles. Deep tendon reflexes were absent in all patients. They were wheelchair bound. Patient IV-11 was able to walk only a few steps independently but with orthopedic foot support, and his gait was very ataxic. Patients III-8 and III-9 had scoliosis. Sensory examination showed distal hypesthesia to touch and pin and absent sense of vibration in the toes and ankles. Patient IV-11 had ulcers of the skin of the feet.

Neurophysiological Findings
The Table contains the results of the tests assessing peripheral nerve (nerve conduction velocity [NCV]), muscle (electromyogram), hearing (audiogram and word comprehension), auditory pathway (otoacoustic emissions, cochlear microphonics, and auditory brainstem potentials), vestibular tests (electronystagmogram [ENG]), and visual and somatosensory evoked potentials for the 3 affected family members.
Motor NCVs tested within 2 years of the onset of their symptoms in the affected individuals were either absent (lower limbs) or markedly slowed (upper limbs), attesting to the rapid course of the neuropathy. Median NCVs in the forearm were less than 20 m/sec and terminal motor latencies in the abductor pollicis brevis were especially prolonged (23 msec). When tested 7 or 8 years later, no muscle contractions of the intrinsic hand muscles could be elicited to median or ulnar nerve stimulation. However, slowed median nerve conduction (7 m/sec) could still be shown in the nerve segment above the elbow in Patient IV-11. Sensory nerve action potentials were absent bilaterally in upper and lower limbs, in 1983, in Patients III-8 and III-9, and in 1987 in Patient IV-11. Electromyography (using needle electrode) of muscles in the arms and legs was performed in 2 of the affected individuals during their first decade and showed signs of denervation. The mother (Patient II-9) of 2 of the affected individuals in Generation III had normal NCVs.
Audiograms tested between 10 and 20 years after the onset of the motor signs showed a pure tone loss without air-bone gap that varied across subjects from mild to profound. The configuration of the audiogram was flat and the loss varied from mild to severe, with 1 of the patients showing an island of mild loss in the 1,500-to 2,000-Hz range in just the right ear. Speech (phonetically balanced words) identification scores were 0% bilaterally in 2 patients (Patients III-8 and III-9), and in Patient IV-11, 20% in the left ear and as high as 70% at 60 dB SL in the right ear. Discrimination with this left ear showed a rapid decrement as the intensity was raised further (20% at 65 dB SL), a phenomenon called "speech intelligibility rollover," which occurs in patients with auditory nerve but not cochlear hearing disorders (Patient IV-11). Tympanograms were normal and acoustic middle ear muscle reflexes were absent. Transient and distortion product otoacoustic emissions were present bilaterally in all. Brainstem auditory evoked potentials did not contain neural components but did show cochlear microphonics in 2 of the patients (Patient III-9 and IV-11). The pure tone audiogram, word comprehension scores, transient evoked otoacoustic emissions, and auditory brainstem potentials from the right ear of the youngest affected family member (Patient IV-11) are shown in Figure 2.
Vestibular testing showed absence of caloric responsiveness in the 3 affected individuals.
Visual evoked potentials were of normal latency in Patients III-8 and IV-11.
Somatosensory evoked potentials were absent in all 3 affected individuals.

Neuropathological Findings
Light microscopy of longitudinal paraffin sections tested with monoclonal antibodies against NGF-rec showed strong expression of NGF-rec on Schwann cells, reflecting the altered relationship between Schwann cells and axons (Fig 3A). Light microscopy of semithin cross sections revealed extreme devastation of myelinated nerve fibers of all types. Only a few myelinated fibers were left and were difficult to detect (see Fig 3B). "Onion bulb" formations were not evident.
Electron microscopic examination revealed that preserved myelinated nerve fibers were surrounded with or in the vicinity of nonmyelinated nerve fibers, creating "pseudo-onion bulb" formations (see Fig 3C). Neither myelinated nerve fiber degeneration nor demyelination was detected. Only rare Schwann cells contained debris of myelin or lipids in their cytoplasm. Nonmyelinated nerve fibers were composed of two to five flattened thin axons surrounded with Schwann cell cytoplasmic processes that were focally reduplicated (see Fig 3C). Pseudo-onion bulb formation with nonmyelinated fiber in the center was observed occasionally (see Fig  3D). There were very few endoneural tubes containing Schwann cells without axons (denervated Schwann cells). Collagen pockets were too numerous, with respect to the age of the patient. The amount of endoneural collagen was prominently increased. Considerable effort was made to obtain a single teased myelinated nerve fiber but, unfortunately, without success.

Genetic Analyses
No recombination was found between the disease locus and any of the eight markers. These markers span a distance of approximately 3 cM, 9 and the most likely order from centRomare (cen) to telomere (qter) is cen-D8S558-D8S529-D8S378-326CA2-189CA17-474CA1-D8S256-D8S1462-qter. 14 Haplotypes were constructed and are shown in Figure 4. The LOD score between the disease locus and the region is 1.53 at a recombination fraction of zero. All of these markers are also tightly linked to the HMSN-Lom locus in the Bulgarian family described by Kalaydjieva and co-workers. 9 Thus, the disease locus is most likely to be the same in the two families. As shown in Figure 4, the 3 affected individuals are homozygous for the three markers in the interval closest to the telomere from 474CA1 to D8S1462, and this haplotype (2-8-3) is identical to the most common HMSN-Lom haplotype in European Gypsies. 9 The affected individual in the fourth generation (Patient IV-11) is also homozygous for the region from D8S558 to 189CA17. However, this latter haplotype (4-6-4-2-3) has not been observed previously among HMSN-Lom or normal Gypsy alleles. 9 The 2 affected individuals in the third generation (Patients III-8 and III-9) are heterozygous in the D8S558-189CA17 region for this 4-6-4-2-3 haplotype and the 6-2-8-4-4 haplotype; the latter is the common HMSN-Lom haplotype. These findings suggest that 189CA17 may be a flanking centromeric marker for the disease gene.

Discussion
We have studied the clinical course, physiological and pathological findings, and genetic locus of a progressive motor and sensory disorder of the limbs in a family of Gypsy extraction in Slovenia that closely resembles a disorder in Bulgarian and Italian Gypsy families (HMSN-Lom) described by Kalaydjieva and coworkers. 9,10 Affected members in these three families have a juvenile onset of motor disability and a delay of about 10 years before the appearance of hearing impairment. A major and early symptom of the hearing disorder shown for the Slovenian family is an impairment of speech perception that is out of proportion to the elevation of pure tone thresholds. The results from the present study show that the hearing loss is caused by a disorder of auditory nerve function in the presence of preserved cochlear outer hair cell functions. The patients have an absence of all neural components of the auditory brainstem potentials beginning with wave I, the component generated by activity of the distal portion of the VIII nerve within the cochlea. 15 Oto- acoustic emissions, a test that measures faint sounds within the ear canal produced by movements of healthy outer hair cells, 6 were normal, thus providing evidence that cochlear outer hair cell functions were intact. Otoacoustic emissions are lost when outer hair cells are damaged. 7 Two of the 3 patients also had evidence of cochlear microphonic potentials, which are receptor potentials generated by both inner and outer hair cells. 4,5 The 1 patient who did not show cochlear microphonics did have preserved otoacoustic emissions. All of these results suggest that the hearing loss in these patients is neural and not sensorineural in type, as cochlear hair cell functions are preserved but auditory nerve responses are abnormal. Almost all patients with auditory neuropathy from a wide variety of causes have a loss of acoustic brainstem reflexes governing both middle ear muscle contractions 1 and olivocochlear bundle suppression of otoacoustic emissions 3 without changes in middle ear ossicle or tympanic membrane mobility. Thus, the absence of middle ear muscle acoustic reflexes in patients in both this study and the Italian and Bulgarian Roma lineages 9,10 is not a sign of a conductive hearing loss but, rather, evidence of an abnormality of acoustically activated brainstem reflexes. The finding of an asymptomatic bilateral vestibular disorder in affected patients in the present study may provide a basis for the observation of nystagmus in several of the patients from the Bulgarian lineage, 10 although none of our patients had a disorder of extraocular movements. Our previous observations of abnormal or absent caloric responses in patients with auditory neuropathy caused by various causes 2 led us to investigate the caloric vestibulocular reflexes in the affected individuals in this study. The absence of caloric responsiveness described in the present study is most likely the result of a neuropathy of the vestibular portion of the VIII nerve rather than a receptor disorder, in keeping with the basic disease process affecting sensory and motor nerves. Vestibular neuropathies have also been described in patients with HMSN in the past. 16 The absence of symptoms of vestibular disorder in our patients may be the result of the gradual occurrence of a bilateral vestibular disorder, allowing the development of mechanisms that compensate for altered vestibular inputs.
The peripheral nerve abnormalities found in the family of this report are similar to those described in a Gypsy family from Bulgaria 10 and for a non-Gypsy Bulgarian patient. 12 The neuropathy is classified as demyelinating because of the markedly slowed conduction velocities. However, axonal loss is also present early in the disorder as evidenced by (1) early loss of nerve fibers in sural nerve biopsies, (2) clinical evidence of profound muscle atrophy, and (3) reduced amplitude of compound muscle action potentials to maximal electrical stimulation of motor nerve fibers. The sensory loss was profound in the feet with absent appreciation of vibration, position, and pin. One patient had skin ulcers on the feet. The sensory changes were compatible with the sural nerve biopsy findings of loss of all but a few fibers. In "typical" HMSN I, primary demyelination is considered the major pathological process 17,18 with secondary axonal loss noted to be an essential feature of the disorder. 19 The role of demyelination as the primary pathogenic mechanism is unclear, because asymptomatic patients without objective clinical evidence of disease can have marked reductions of motor NCVs. 20 Moreover, studies of adult patients over time show that NCVs can remain essentially unchanged despite deterioration of motor and sensory functions. 18,21,22 It is likely, therefore, that axonal loss is the major contributor to clinical disability in HMSN. 17,18,22 In the patients with HMSN-Lom, the relative contribution of axonal or myelin pathology to the clinical disorder is not known. 10 Sural nerve biopsies from both the Bulgarian and Slovenian families taken 30 years after the onset of the disorder show that all nerve fibers (axons and myelin) are lost. However, early in the course of the illness, onion bulbs and hypertrophic changes can be found 10 that then apparently "disappear" at later stages of the disease. The loss of onion bulbs at later stages of the disease has also been described in HMSN IA patients with severe axonal loss. 23 In the patients presented in this study, we found clinical, neurophysiological, and neuropathological evidence of demyelination, severe axonal loss, and an altered relationship between the Schwann cells and axons. We did not find onion bulb formations, and other authors have proposed that primary demyelination can be accompanied by secondary axonal loss without typical accompanying onion bulb formation. 24,25 We suggest that in affected members of these kindred, the pathological disorder affects both the axon and the myelin sheath contemporaneously quite early in the disease, with evidence in the first decade of life of both nerve conduction slowing (demyelination) and limb muscle atrophy (axonal loss). Demyelination in patients with HMSN IA is uniform throughout the axonal length 27 and is not believed to block conduction or show significant desynchronization of transmission in peripheral nerve. 28 However, for auditory processes even a slight degree of temporal desynchronization accompanying demyelination, either primary or secondary to axonal disease of the VIII nerve, would likely have significant adverse effects on auditory function. The auditory system is especially adapted for encoding temporal features in the microsecond range of environmental acoustic signals. Patients with auditory neuropathy are particularly impaired on tasks testing temporal resolving capacities such as sound localization, word comprehension, and detection of small gaps in steady sound. 1 Impaired temporal synchrony of auditory nerve fibers is also likely to be responsible for the failure to detect averaged auditory brainstem potentials to acoustic signals that are clearly perceptible. We have modeled the effects of introducing a slight temporal jitter in the averaging process (Ͻ1 msec) for auditory brainstem potentials in healthy subjects and shown a marked attenuation of averaged auditory brainstem potentials beginning with wave I. 1 The expression of the pathological process in auditory/vestibular nerves in these patients may be related to specific anatomical features of these nerves compared with other cranial and peripheral nerves. 29 The portion of the auditory/vestibular nerve that is distal to the arachnoid is myelinated by Schwann cells and the central portions are myelinated by oliogodendroglia. Auditory nerve fibers in humans are relatively homogeneous, averaging 6 m (range, 3-11 m), and vestibular nerves are slightly larger (mean diameter, 9 m; range, 3-15 m). Morphometric analyses of sural nerves from affected young individuals revealed a loss of myelinated fibers greater than 5 m 10 and a loss of all nerve fibers at later stages of the illness. Symptoms of auditory nerve involvement occurred about 10 years after the limbs were first affected, perhaps reflecting the predominantly small-fiber composition (Ͻ5 m) of the auditory nerve.
HMSN neuropathies are a heterogeneous group of disorders, with various cranial and even autonomic nerves being affected along with the peripheral disorder. These disorders involve optic nerves, 30 phrenic and vagal nerves, 31 facial and vestibular nerves, 16 trigeminal nerve, 32 and auditory nerve. [33][34][35] Other sensory nerve involvement (olfaction and taste) has not been specifically commented on. The results of the genetic analysis for the Slovenian family combined with the results for the Bulgarian and Italian families 9,11 suggest that the region of chromosome 8 containing the disease locus can be refined to the interval between 189CA17 and D8S256. This interpretation is based on the assumption that a recombination event between 189CA17 and 474CA1 created the two different haplotypes in the affected individuals, Patients III-8 and III-9 (6-8-2-4-4-2-8-3 and 4-6-4-2-3-2-8-3; see Fig 4). It was surprising to find that the third affected individual, Patient IV-11, is homozygous for a haplotype (4-6-4-2-3-2-8-3; see Fig 4) that has not been observed in affected Bulgarian Gypsies, because the fathers (Patients II-10 and III-12) of the 3 affected individuals are not known to be related. We will genotype additional DNA samples from the Slovenian Gypsy population, to determine the relative frequencies of the two haplotypes associated with the disease allele and to determine their origin relative to the separation of the Slovenian and Bulgarian gypsies, which took place in the 18th century. 36 This study provides a detailed characterization of HMSN in affected members of a Slovenian Gypsy family and suggests that the disease locus is the same as in a Bulgarian Gypsy family. In addition, results suggest that the critical region containing the HMSN-Lom locus can be refined. A BAC contig of this region is being completed, 14 to assist in the isolation of the gene that is defective in individuals with hereditary auditory, vestibular, motor and sensory neuropathy.