MATERIAL AND METHODS

The patients were referred for a neuro-otological opinion because of one or more of the following problems: vertigo, dizziness, postural instability, or eye movement disorder. All patients in whom pDBN was seen on the Dix-Hallpike manoeuvre,⁶ or when the head was extended directly backwards from the sitting to the supine straight head hanging position were included. The nystagmus was detected by direct clinical observation with the patient's eyes in primary gaze fixating on the examiner's face. Patients with torsional or horizontal nystagmic components added to their pDBN were included. By convention, a left torsional nystagmus means that the upper pole of the eye beats to the patient's left shoulder. Patients with DBN in any position of gaze while upright were excluded.

All patients underwent a neurological, eye movement, and neuro-otological examination. Bithermal caloric testing, as described in Fitzgerald and Hallpike⁷ and complemented with Frenzel's glasses observation, was obtained in 35 patients. Direct current horizontal electro-oculography was conducted in 47 patients and included search for spontaneous/gaze evoked nystagmus, smooth pursuit, optokinetic and rotational testing and, when required, saccadic velocities. Techniques and normal data have been reported elsewhere.^8–13 As the clinical and instrumental examination of the oculomotor and vestibular system were in agreement the results are combined. Neuroimaging was available according to clinical needs; cranial MRI in 30 patients and CT in 16 patients. Four patients were re-examined to obtain computerised three dimensional video-oculography (SMI, Berlin; bandwith 50 Hz) during positional testing.

RESULTS

The 50 patients were divided into two groups: those with unequivocal CNS dysfunction (38 patients; mean age 51, range 17 to 75 years old) and those with no evidence of CNS abnormality on clinical, oculographic, or imaging investigations (12 patients; mean age 50, range 33 to 78 years old (idiopathic group)).

CNS group

Patients with CNS lesions had multiple system atrophy (n=13), various types of cerebellar degeneration (n=12), or miscellaneous disorders (n=13), including cerebrovascular disease (n=5), multiple sclerosis (n=2), hydrocephalus (n=2; one idiopathic and one secondary to a high midbrain tumour). They reported walking difficulties (n=34), falls (n=15), slurred speech (n=24) orthostatic intolerance (n=13), and urinary dysfunction (n=14) in various combinations. Cerebellar abnormalities (n=28) prevailed in the neurological and oculomotor examinations, followed by autonomic (n=11), pyramidal (n=8), and extrapyramidal system (n=7) findings. The pDBN was triggered by the Dix-Hallpike manoeuvre in 37 out of 38 patients; in one patient only the straight head-hanging manoeuvre was performed. The Dix-Hallpike manoeuvre was bilaterally positive in 35 patients and unilaterally in two. The direction of the nystagmus was usually purely down beat (n=30), or associated with a horizontal (n=7) or torsional component (n=1). The onset was immediate—that is, no latency (n=37). The nystagmus usually adapted—that is, positioning rather than positional nystagmus (n=35). It was usually short lived (<10 s), particularly in degenerative conditions, but was more sustained in three patients. On sitting upright, there was no nystagmus or, occasionally, a transient up beat nystagmus was seen (n=7). Habituation, tested in five patients, was negative (the nystagmus did not diminish on repeated positional testing). There was no significant vertigo during the pDBN, except in two patients with hydrocephalus.

Idiopathic group

The individual clinical data are shown in table 1. The presenting symptom was positional vertigo (n=12) but four patients also reported spontaneous dizziness/vertigo. Five patients had a history of migraine, two a history of middle ear surgery, and one a myopathic syndrome of unknown cause. One patient had a history of recurrent posterior canal benign paroxysmal positional vertigo (BPPV), in another, the positional vertigo started after head trauma. No patient was on medication with CNS or oculomotor effects. The CNS and oculomotor examinations were normal. All patients had MRI (n=8) or CT (n=4), which showed no brain abnormality. In two patients, imaging disclosed abnormalities of possible relevance in the middle ear; one a previous mastoidectomy with remaining inflammatory activity in the attic/antrum region, the other repeated middle ear surgery with remaining fluid within the left mastoid air cells and middle ear cavity.

The nystagmus was triggered by the Dix-Hallpike manoeuvre bilaterally in nine patients (75%) and unilaterally in one. In two patients with a typical history of positional vertigo but negative Dix-Hallpike manoeuvre, the straight head-hanging manoeuvre was performed and found positive (patients 2 and 7; see case report 2 and table 1). This manoeuvre was also tried in a third patient with a positive Dix-Hallpike manoeuvre and was positive; the nystagmus during right/left Hallpike and the head-hanging position was similar (see case report 1). The nystagmus had a brief latency of 2–5 seconds in seven patients and appeared immediately in the other five. The nystagmus adapted (<1 minute) in all patients. Habituation was investigated in nine patients and was found in seven (78%). Most patients reported vertigo during the positional manoeuvres (11/12; not clearly stated in the notes for patient 12).

In six patients, torsional nystagmic components were present during the positional procedures (patients 1–6, table 1). In two cases a unidirectional, rightwards torsional component was concurrent with the pDBN and documented by three dimensional video-oculography (see case reports 1 and 2 below). These patients were thought to have right anterior canal BPPV. In patient 1 the direction of the torsional component was the same during right and left Dix-Hallpike manoeuvre (fig 1), probably indicating bilateral triggering of the right anterior canal BPPV (fig 2). In a third patient the torsional component was also concurrent with the pDBN but the direction could not be ascertained clinically; the vertigo and nystagmus disappeared spontaneously on follow up. The most likely diagnosis was anterior canal BPPV (see case report 3). A fourth patient (patient 5), repeatedly seen over 10 years, had typical posterior canal BPPV, either on the right or left or sometimes bilateral. In some visits there was only pDBN and no torsional component was clinically noted. Despite the apparent absence of a torsional component during the pDBN (see discussion) this patient's most likely diagnosis was anterior and posterior canal BPPV. Patient 6 could show either a moderate right beating horizonto-torsional nystagmus or violent, transient pDBN with full habituation in which a torsional component was not noted; the second was thought to be anterior canal BPPV. In the remaining case (patient 4) the torsional nystagmus preceded the pDBN—that is, typical posterior canal BPPV followed by down beat while still in the Dix-Hallpike position. The main diagnosis was posterior canal BPPV; such reversal to a DBN has been previously reported in posterior canal BPPV and could be a rebound due to central adaptation.^14,15

In six patients no torsional component was seen during positioning (patients 7–12, table 1). This group comprised a patient with pDBN only provoked by the head-hanging manoeuvre (patient 7), two patients with unilateral middle ear disease (patients 8 and 9), a patient with typical posterior canal BPPV on follow up (patient 10), and two patients with vestibular migraine (patients 11 and 12). The presence of predisposing factors for peripheral vestibular disease and absence of CNS signs, added to the fact that a torsional component can be missed clinically (see case report 2 and discussion), suggested a labyrinthine aetiology. The more likely explanation for the pDBN in these patients was anterior canal BPPV.

Caloric testing was performed in 10 out of 12 patients; two patients had abnormal otoscopy precluding this test. Horizontal rotational responses were obtained in all patients. Combined rotational and caloric results were normal in eight patients. A mild canal paresis (12 and 16%) in the light was present in two patients (upper normal limit 9%).^12,13 Pure tone audiograms showed a bilateral sensorineural hearing loss in two, a unilateral sensorineural hearing loss in one (associated with canal paresis, see case 1 below), a conductive hearing loss in the two patients with clinical/imaging evidence of middle ear disease, and was normal in the remaining seven patients.

Four patients were re-examined to record the pDBN with three dimensional video-oculography but only succeeded in two (case report 1 and 2). In patient 5 we could not elicit nystagmus on the day and in patient 3 the pDBN had resolved (case report 3).

The following three case histories illustrate representative patients.

Patient 1

A 47 year old man complained of 4 years of brief positional vertigo on lying down, rising from bed, looking up, or bending forwards. He had a history of migraine. His CNS and eye movement examinations were normal. Brain CT was normal but MRI could not be obtained because of the patient's obesity. A right canal paresis of 16% in the light and a mild right low-middle frequency sensorineural hearing loss were found. Brain stem auditory evoked potentials, electro-oculography, and rotatory responses were normal.

The right Dix-Hallpike manoeuvre triggered positional vertigo and 2–3 torsional nystagmic beats to the right (upper pole of the eye beating to the right shoulder). With the left ear down, there was DBN with a latency of 2–3 seconds, duration of 10–15 seconds, and habituation on repeated testing; no torsional component was noted. He was prescribed self paced exercises^16,17 and 3 months later we could not elicit positional nystagmus. After a further 3 months an identical positional nystagmus was found and he was treated twice with a left Epley manoeuvre. When reviewed 3 months later the right Dix-Hallpike test showed a predominant pDBN with a small right beating torsional component. The left Dix-Hallpike manoeuvre produced only some pDBN. He was treated with a right Epley manoeuvre twice but 3 months later there was no change in the nystagmus.

He was seen 16 months later with essentially the same positional nystagmus and vertigo but both habituated on repeated testing. Video-oculography showed that during both right and left Dix-Hallpike manoeuvre the nystagmus was down-beating with a right torsional component (fig 1).The straight head-hanging manoeuvre also elicited a similar nystagmus. Slow positioning did not provoke vertigo or nystagmus in any position. Horizontal canal manoeuvres were negative.

The pDBN with a right beating torsional component was compatible with a right anterior canal BPPV. The hearing loss, normal brain stem auditory evoked potentials, and canal paresis suggested labyrinthine involvement on that side. The patient attended the laboratory weekly, four times, for treatment with the three dimensional rotator (SEGA R360) which we used successfully for treatment of posterior canal BPPV.¹⁸ We performed 360° and 450° forwards rotations, at 90°/s, with the head straight and turned 40° to the left or right to bring the anterior canal in the plane of rotation. These manoeuvres were carried out three times each. On stopping from the first rotation only, intense, brief DBN and vertigo was provoked; on repetition there was no nystagmus or vertigo. On weekly follow up, the Dix-Hallpike manoeuvre on the right was negative on two occasions but then became bilaterally positive again. The patient was subjectively moderately improved throughout but this was largely the result of avoiding the offending head positions.

Patient 2

A 46 year old man complained of 8 years of brief positional vertigo mainly on lying down, rising from bed, or looking up. He had experienced in the last 3 years some 10 episodes of dizziness lasting between 10 minutes to 2 hours. He had tension headaches but not migraine. Neurological and eye movement examinations were normal. Brain CT, MRI, caloric/rotatory responses, and a pure tone audiogram were normal.

The Dix-Hallpike manoeuvre was negative but the straight head-hanging manoeuvre triggered a violent but brief (<10 s) pDBN. The nystagmus had a latency of a few seconds, showing habituation. This pDBN was a persistent, isolated finding in the three occasions the patient was seen between 1992 and 1999. Video-oculography showed that the pDBN was intense during the first few seconds, with slow phase velocity of 87–108°/s, decaying rapidly to a stop after 10 seconds. A right beating torsional nystagmic component, undetected clinically but clearly visible on the analogue videotapes and oculographic traces, was documented. The slow phase velocity of the torsional component was about 30% of that of the vertical component. These findings were compatible with right anterior canal BPPV. Forward rotations with the head straight and turned 40° to either side in the three dimensional rotator were performed on his last visit but in a postal reply the patient reported no significant change.

Patient 3

A 46 year old woman with a history of migraine complained of some 20 episodes of recurrent vertigo which lasted 2 or 3 days in the period 1987 to 1993. It was not clear if the vertigo was spontaneous or positional. Neuro-otological examination in 1990 was normal but for the Dix-Hallpike manoeuvre, both right and left, which triggered a pDBN. The nystagmus was intense and brief and habituated partly. There was a small associated torsional component, the direction of which could not be clearly ascertained. Caloric/rotatory responses, pure tone audiometry, brainstem auditory evoked responses, and MRI were normal. The findings were compatible with anterior canal BPPV. The patient was reassessed in 1998 as part of this study. She had no vertigo nor other symptoms between 1993–98 apart from a Bell's palsy which spontaneously recovered. Examination, including all positional manoeuvres, was negative.

DISCUSSION

Several series of patients with DBN in the upright position have been published^1–5 but, to our knowledge, no series of patients with pDBN have appeared in the literature. In our 50 patients with pDBN, 75% had CNS disorders involving the cerebellum or brain stem but 25% had no evidence of a CNS abnormality.

pDBN in CNS disease

because pDBN is known to occur in CNS disease, particularly posterior fossa lesions^19–22 we will only concentrate on two new points of clinical relevance. Firstly, it is important to note that no patient in our series had an Arnold-Chiari malformation, one of the most common causes of spontaneous, constant DBN.^2–4 The reason for this discrepancy is not clear but it may indicate that the cerebellar flocculus—a site involved in the Arnold-Chiari malformation and proved to cause constant DBN in animal lesion studies²³—is not the site responsible for transient pDBN. Floccular lesions in monkeys can produce positional nystagmus but this is horizontal-rotatory.²⁴ By contrast, nodular lesions in cats²⁵ selectively cause transient pDBN of the type seen in our patients with damage to the CNS.

The second point of clinical relevance is the finding of pDBN in patients with multiple system atrophy, so far only reported previously in abstract form.²⁶ Multiple system atrophy has three main presentation types, one with atypical parkinsonism (striatonigral type), one mostly cerebellar (olivopontocerebellar type), and one with mainly autonomic nervous system features (Shy-Drager type).^27–29 Most of our patients with multiple system atrophy had overt cerebellar signs but five had a prominent extrapyramidal syndrome. Thus, the positional manoeuvre is of value in the differential diagnosis of the atypical parkinsonian syndrome. The presence of pDBN is likely to depend on the presence of prominent cerebellar pathology in multiple system atrophy but not in other parkinsonian syndromes.^29,30

pDBN without clinical evidence of CNS disease

The pDBN in our two groups of patients had different characteristics. In the CNS group the trend was that there was no latency or habituation of the nystagmus or significant vertigo during positioning. In the group with no evidence of CNS disease the pDBN tended to have opposite characteristics, usually accepted as due to labyrinthine disease.⁶ Further, in four patients who were repeatedly examined, the pDBN resolved spontaneously in one and there were no additional findings in the other three followed up for 4 to 10 years.

It is possible that these patients had a peripheral abnormality involving the anterior canal, similar to that causing BPPV. Currently, it is accepted that BPPV of the posterior and horizontal canals is due to semicircular canalithiasis—that is, freely moving debris within the endolymph.³¹ The nystagmus in canalithiasis is thought to be induced by the plunger effect of debris when a canal is rotated with respect to gravity and the debris sink downwards along the canal. Due to the roughly coplanar orientation of the vertical canals (left anterior with right posterior and vice versa), a left Dix-Hallpike manoeuvre for the left posterior canal rotates the right anterior canal backwards. In doing so, any debris lying on the ampullar region would travel ampulofugally on reaching the head dependent position, thus exciting anterior canal afferents^32,33 (fig 2). In turn, this should lead to pDBN with a torsional component beating towards the uppermost ear,^34,35 in this example to the right.

There have been occasional reports of suspected anterior canal BPPV but these are not always concordant. During particle repositioning treatment of posterior canal BPPV, transient pDBN has been found and attributed to invasion of the anterior canal by the canalicular debris.^18,36 The nystagmus has been reported as purely down beating^18,36 or associated with a torsional component.^36,37 The Dix-Hallpike manoeuvre has been reported to trigger the nystagmus on the same side³⁶ or on the opposite side^18,37 of the suspected anterior canal. In none of these patients, however, was the nystagmus recorded. The three dimensional oculographic trace in figure 1 combines, as expected for right anterior canal BPPV, pDBN with a torsional right beating component.

However, several features in our patients deserve discussion: (1) the rarity of anterior versus posterior canal BPPV, (2) the fact that the nystagmus was elicited bilaterally with the Dix-Hallpike manoeuvre in nine of 12 patients, (3) only triggered by the head-hanging manoeuvre in two further patients, and (4) the apparent lack of torsional nystagmic components in some patients.

Rarity of anterior canal BPPV

The rarity of anterior canal BPPV is assumed to be due to the posterior arm of the anterior canal descending directly into the common crus and vestibule. Thus, debris within the anterior canal should be self clearing. It is possible that the persistent pDBN in some of our patients represents the small proportion in whom there is difficulty clearing the anterior canal; this is also found in about 10% of patients with BPPV involving the posterior canal.³⁸ Why this happens is unknown but could be due to the conglomerate of debris being too large, continuous production of debris, failure of disaggregation, or a narrow common crus. The failure of the whole body rotational treatment in patients 1 and 2 could also be explained by these mechanisms. Cupulolithiasis, in which debris become fixed to the cupula rather than floating freely in the endolymph, is also a cause of refractory BPPV^39,40; the almost vertical orientation of the ampullary segment of the anterior canal could result in extensive contact between debris and the cupula, thus facilitating cupulolithiasis.³³ This possibility is, however, unlikely to explain our data as cupulolithiasis should not show adaptation and habituation.^41–43 For these reasons the underlying mechanism of persistent positional vertigo in our patients is more likely to be refractory canalithiasis.

Bilaterally positive Dix-Hallpike manoeuvre in unilateral anterior canal BPPV

Posterior canal BBPV is usually unilateral,³⁷ by contrast with our findings where the positional nystagmus was triggered bilaterally. There are, however, some important anatomical differences between the anterior and posterior canals which may explain why a bilateral Dix-Hallpike manoeuvre may trigger a unilateral anterior canal BPPV (figs 2 and 3).

In canalithiasis of both the anterior or posterior canal, the canaliths will collect in the ampullary segment of the respective canal with the head in the upright position (fig 2, top). On tipping the head back, the downward motion of the canaliths induces ampullo-fugal endolymph flow which stimulates the respective ampullae. However, the orientation of the ampullary segment of the anterior and posterior canal differ (fig 3, taken from in vitro CT data).⁴⁴ The initial ampullary segment of the anterior canal in the upright position of the head is roughly vertical (70° with respect to earth horizontal) whereas for the posterior canal it is roughly 20° below horizontal. In the case of the posterior canal a head back tilt of 120° in the plane of the canal takes the ampullary segment past vertical, inverting it (50° off vertical); this will cause downward motion of the debris (fig 2, bottom). By contrast, the ampullary arm of the contralateral posterior canal remains roughly horizontal (fig 2, bottom). Thus for right posterior canal BPPV a left Dix-Hallpike manoeuvre is not capable of provoking nystagmus because it does not orient the ampullary segment of the right posterior canal downwards.

The situation is different for anterior canal BPPV and as an example we consider right anterior canal BPPV. A left ear down Dix-Hallpike manoeuvre results in the cupula of the right anterior canal becoming superior, with the open end of the ampullary segment pointing downwards at an angle of about 40° off vertical. Note that, due to the almost vertical orientation of the ampullary segment of the anterior canal, an ipsilesional left Dix-Hallpike will also attain a position whereby this segment points downwards at about 40° off vertical (fig 2 bottom). The only difference in the trajectory of the two movements is that during a contralateral Dix-Hallpike manoeuvre the head rotates in the plane of the affected anterior canal whereas during the ipsilesional Dix-Hallpike manoeuvre the head rotates orthogonally to the plane of the canal. We suggest that this difference is unlikely to be critical for particle mobilisation. Evidence comes from the successful treatment of posterior canal BPPV with fast rotation backwards (heels over head rotation)^15,18 or forwards (Semont manoeuvre).⁴⁵ Also, mastoid vibration, which does not cause endolymph flow, can trigger BPPV after a slow head reorientation.⁴⁶ These findings indicate that endolymph flow is of secondary importance in particle mobilisation compared with fast head acceleration, gravity, and a suitable final head down position. A Dix-Hallpike manoeuvre on either side should therefore be able to provoke unilateral anterior canal BPPV.

Apparent lack of torsional nystagmic components

The finding that in six of 12 patients the nystagmus seemed to be purely down beat, as opposed to torsional down beating, could mean that these patients are a different group, and do not have anterior canal BPPV. In theory, a minute lesion of the nodulus ²⁵ could explain the pDBN without disrupting other eye movements and be missed on scanning. However, it can equally be argued that a small torsional nystagmic vector may not be detected clinically due to its short duration and the simultaneous large downbeat component. This was indeed the case in patient 2, where the torsional component was only detected by high magnification videoimaging, and in patient 3, in whom the torsional component was seen but its direction not ascertained. The torsional nystagmic vector is smaller for the anterior than the posterior canal because the posterior canals are placed at 56° from the sagittal plane whereas the anterior canals are at 41°.⁴⁷ Calculations of angular eye velocity vectors derived from known canal geometry show the existence of an upwards bias in vertical slow phase eye velocity.⁴⁸ Thus, more down beat than torsional nystagmus is expected from anterior canal BPPV and more torsional than up beat nystagmus in posterior canal BPPV.

Nystagmus-provocation by the straight head-hanging manoeuvre

In patient 1, with pDBN and a right torsional component during both the right and left Dix-Hallpike manoeuvres (fig 1), it is easy to accept that the straight head-hanging manoeuvre will also provoke a similar nystagmus. The beat direction of the torsional component would be compatible with right anterior canal BPPV. Assuming that the other two patients with pDBN provoked only by the head-hanging manoeuvre may also have anterior canal BPPV we speculate why this manoeuvre may be more effective than that of Dix-Hallpike.

As discussed above, for triggering anterior canal BPPV, strict rotation in the canal plane is of relatively less importance than a final low head down position. The Dix-Hallpike manoeuvre does not achieve such low vertex positions as the head has been previously rotated 40°-45° horizontally. Indeed, observation shows that the head reaches a more dependent position, by about 20°, during the head-hanging than during the Dix-Hallpike position. This additional 20° may be crucial for provoking anterior canal BPPV as only then the ampullary segment will approach a vertical down pointing position. By contrast, in posterior canal BPPV the canal moves past vertical even during conventional Dix-Hallpike manoeuvre.

CONCLUSIONS

In summary, pDBN is a valuable neurological sign often due to cerebellar disorder. It may help differentiate multiple system atrophy from other parkinsonian syndromes. Surprisingly, the Arnold-Chiari malformation does not seem to be a common aetiology. A smaller proportion of patients with pDBN do not show any evidence of CNS disease. It is possible that these patients have refractory anterior canalithiasis. The straight head-hanging manoeuvre should be performed in patients with a convincing history of positional vertigo but no nystagmus induced by posterior and horizontal canal manoeuvres. Video-oculography may be necessary for detecting the torsional nystagmic component added to the pDBN. Findings in audiovestibular tests and the presence of concurrent or successive BPPV of other canals can indirectly support the diagnosis of anterior canal BPPV.