Thalamic infarctions cause side-specific suppression of vestibular cortex activations

Abstract

H₂O¹⁵-PET was performed during caloric vestibular stimulation of the right and left external ears in eight right-handed patients with acute unilateral infarctions or haemorrhages of the posterolateral thalamus (four right, four left). The posterolateral thalamus is the relay station for ipsi- and contralateral ascending vestibular input to the multiple multisensory vestibular cortex areas. The aim of this study was to evaluate the differential effects of unilateral vestibular thalamic lesions on thalamo-cortical projections, right hemispheric dominance and reciprocal inhibitory visual-vestibular interaction, as well as perceptual and ocular motor consequences during caloric irrigation. The major findings of the group analyses of the patients with right-sided and those with left-sided lesions were as follows: (i) activation of the multisensory vestibular temporo-parietal cortex was significantly reduced in the hemisphere ipsilateral to the thalamic lesion when the ipsilesional or contralesional ear was stimulated; (ii) activation of multisensory vestibular cortex areas of the hemisphere contralateral to the irrigated ipsilesional ear was also diminished; and (iii) the right hemispheric dominance in right-handers described above was preserved in those with right and left thalamic lesions. Simultaneous deactivations were often restricted to only one hemisphere—the one contralateral to the stimulation and contralateral to the vestibular cortex areas activated. There was, however, one area in the inferior insula which was also activated by either right or left ear stimulation in the hemisphere ipsilateral to the lesion. This supports the assumption that there is a bilateral direct ascending vestibular projection from the vestibular nuclei to the inferior part of the insula, which bypasses the posterolateral thalamus and is stronger in the right hemisphere. The cortical asymmetry of the pattern of activation during horizontal semicircular canal stimulation by calorics was not associated with a significant direction-specific asymmetry of caloric nystagmus or perceived body motion. Thus, the data demonstrate the functional importance of the posterolateral thalamus as a unique relay station for vestibular input to the cortex, of the dominance of the right hemisphere in right-handedness, and of ipsilateral ascending pathways. Furthermore, the normal interaction between the two sensory systems—the vestibular and the visual—appears to be impaired.

Introduction

Vestibular pathways run bilaterally from the eighth nerve and the vestibular nuclei through ascending fibres such as the medial longitudinal fascicle, the ascending Deiters' tract or the brachium conjunctivum to the ocular motor nuclei and the supranuclear integration centres in the pontine and rostral mesencephalic brainstem. These fibres represent the vestibulo-ocular reflex (VOR), which transmits information from the vestibular endorgans to the ocular motor nuclei and the supranuclear coordination centres for eye and head integration. From there, they reach several multisensory vestibular cortex areas through projections via the posterolateral thalamic subnuclei to mediate the cortical functions of the vestibular system such as the perception of verticality and self-motion (Brandt and Dieterich, 1999).

The posterolateral thalamus—containing the subnuclei ventrocaudalis externus (Vce), ventro-oralis intermedius (Vim), dorsocaudalis (Dc), ventrocaudalis internus (Vci) and nucleus ventroposterior lateralis (VPLo) [human nomenclature of Hassler (1959) in the thalamus atlas of Van Buren and Borke (1972)]—is the afferent relay station for multiple multisensory vestibular cortex areas, as determined by vestibular stimulation in animal experiments (Sans et al., 1970; Deecke et al., 1973, 1974, 1977; Büttner and Henn, 1976). Via the subnuclei of this relay station, vestibular information in animals reaches several separate and distinct cortex areas such as the parieto-insular vestibular cortex (PIVC) in the posterior insula, adjacent retroinsular areas and the granular insular region (Grüsser et al., 1982, 1990a, b; Guldin and Grüsser, 1996), the visual temporal sylvian area VTS posterior to PIVC (Guldin and Grüsser 1996, 1998), area 3aV in the central sulcus (Schwarz et al., 1973; Ödkvist et al., 1974; Büttner and Lang, 1979), parts of area 7 in the inferior parietal lobe [Brodmann area (BA) 40] (Ventre and Faugier-Grimaud, 1986; Faugier-Grimaud and Ventre, 1989; Ventre-Dominey et al., 2003), and probably area 2v at the tip of the intraparietal sulcus (Fredrickson et al., 1966; Schwarz and Fredrickson, 1971; Büttner and Buettner, 1978). The oral portion of the VPLo and, to a lesser extent, the inferior portion (VPI) seem to play the most important role in the projection sites of ascending vestibular thalamocortical pathways in monkeys, according to neurophysiological and anterograde tracer studies that used labelling in area 3a (Büttner and Lang, 1979).

All the above-mentioned cortical areas are multisensory and respond not only to rotational vestibular but also to somatosensory and in part visual motion stimuli. Tracer studies in monkeys have shown that, anatomically, these cortical areas are closely connected to each other (Guldin et al., 1993; Guldin and Grüsser, 1996). Such studies in monkeys and cats also identified the multiple projections from the thalamus to several vestibular cortical representations, mainly the PIVC, area 7ant (2v) and 3aV (Akbarian et al., 1992, 1993, 1994; Guldin and Grüsser, 1996). The PIVC is the dominant multisensory vestibular cortex area, which was considered to be a ‘core region’ (Guldin and Grüsser, 1996). It receives its main input from the vestibular parts of the ventroposterior complex (Vce, Vci, nucleus ventro-oralis externus (Voe), Vim) and the medial pulvinar (Pum). The ‘proprioceptive vestibular area 3aV’ receives its major thalamic projection from the VPLo.

In the past 10 years, functional imaging studies using vestibular, somatosensory and visual optokinetic stimulation have reported evidence of similar locations and connections of these bilaterally organized multisensory vestibular cortical areas in humans. A complex network of areas predominantly in the temporo-insular and temporo-parietal cortex was found in both hemispheres (Bucher et al., 1998; Lobel et al., 1998; Bense et al., 2001; Suzuki et al., 2001; Fasold et al., 2002; Emri et al., 2003). The activation pattern of this network in both hemispheres, which was only recently identified in humans during vestibular stimulation, is not symmetrical but determined by three factors that are additive (Bense et al., 2003, 2004; Dieterich et al., 2003). The most important factor was the subject's handedness. The activation pattern revealed a dominance of the non-dominant hemisphere (i.e. the activation was stronger in temporo-insular-parietal areas of the right hemisphere in right-handers and of the left hemisphere in left-handers). Secondly, the side of the stimulated vestibular endorgan was another important determinant. Stronger activation was observed within the hemisphere ipsilateral to the stimulated ear (i.e. stronger activation within the right hemisphere than within the left during stimulation of the right vestibular endorgan). Thirdly, the direction of the induced vestibular symptoms was also relevant. Stronger activation occurred within the hemisphere ipsilateral to the fast phase of vestibular caloric nystagmus.

Since there is a vestibular thalamo-cortical network in both hemispheres, the question arose as to the consequences of a unilateral lesion of the ‘vestibular relay station’ in the posterolateral thalamus. Thus, the aim of this ¹⁵O-labelled H₂O bolus PET study was to analyse the differential effects of unilateral caloric vestibular stimulation (right or left ear irrigation with warm water at 44°C) on the cortical and subcortical activation patterns of both hemispheres in patients suffering from an acute or subacute unilateral stroke of the posterolateral thalamus. Drawing on the three factors mentioned above that determine the activation pattern in healthy right-handers and the bilaterality of the pathways, we posed the following questions to elucidate data collected from two groups of right-handed patients: one group with a lesion of the right posterolateral thalamus and another group with a lesion of the left posterolateral thalamus.

• Does an infarction of the right posterolateral thalamus influence the activation pattern during caloric irrigation of the right ear only within the right hemisphere or within both hemispheres?

• Does an infarction of the right posterolateral thalamus influence the activation pattern during caloric irrigation of the left ear or does a normal pattern occur? If the pattern is affected, is it affected only within the right hemisphere?

• Does an infarction of the left thalamus induce similar effects on the activation pattern as a right-sided infarction or are the effects less significant, since the left hemisphere is the non-dominant hemisphere for the vestibular cortical system?

• Which areas of the multisensory vestibular cortical network are influenced by the lesion—only the areas in the insula or also areas in the temporo-parietal and frontal cortex?

• Does the posterolateral thalamic lesion influence the clinical parameters of the VOR, e.g. the ocular motor (caloric nystagmus) and perceptual components?

• Is there a correlation between cortical activation pattern and ocular motor and perceptual parameters? In other words, if there is no activation in the vestibular cortical areas of one hemisphere, does this cause a reduction of caloric nystagmus?

Patients and methods

Patients

Eight right-handed patients (four females, 23–62 years old, mean age 46 years; four males, 59–66 years old, mean age 63 years) presenting with a vascular lesion of the posterolateral thalamus participated in the study (Table 1) after giving their informed written consent in accordance with the Helsinki Declaration. The study was approved by the local Ethics Committee as well as by the radiation protection authorities of the Ludwig–Maximilians University and Technical University Munich. The aetiologies of the lesions were ischaemic infarctions in six (three right, three left) and circumscribed haemorrhages in two patients (one right, one left). All patients had CT and/or MRI scans to determine the aetiology, the exact site and the extent of the lesion. Patients were excluded from the study if they had received pharmacological medication (analgesic, anticonvulsant or sedative), had deficits from earlier brainstem events prior to the onset of disease, showed acute multiple lesions of the brainstem and/or the hemispheres, and if they were unable to keep in a supine position for at least three hours.

All patients were examined neurologically, neuro-ophthalmologically and by electro-oculography. Furthermore, to measure the tonic vestibular signs, all underwent orthoptic examination, which included measurements of the subjective visual vertical (SVV) and of tonic ocular torsion by a scanning laser ophthalmoscope (for methods and normal values, see Dieterich and Brandt, 1993a). Static SVV was determined binocularly while the patients sat upright with their head fixed in front of a hemispheric dome (60 cm in diameter and covered with random dots). A central test edge had to be adjusted (means of 10 adjustments) from a random offset position to the subjective vertical, with the hemispheric dome stationary. Under these conditions, the normal range is ±2.5°. Ocular torsion was defined as the mean of four fundus photographs taken by the scanning laser ophthalmoscope with the head upright and fixed in the sitting position. Otoneurological examination and electronystagmography showed that no patient had peripheral vestibular dysfunction. The Laterality Quotient for right-handedness was +100 in six patients and +60 in the remaining two according to the 10-item inventory of the Edinburgh test (Oldfield, 1971; Salmaso and Longoni, 1985).

For group analyses with PET, the patients were subdivided into two groups of four patients, one subgroup with right-sided lesions (two females, two males; age 58–66 years; mean age 61 years) and the second subgroup with left-sided lesions (two females, two males; age 23–63 years; mean age 47 years). This was necessary because the activation pattern in normal subjects is dependent on handedness. The H₂O¹⁵-PET scans were performed between day 4 and day 55 after lesion onset (day 10 was the mean time).

Caloric irrigation and electro-oculography in PET

During all conditions of PET data acquisition, the subject lay supine with eyes closed. Vestibular stimulation was performed by irrigating the right or left external ear canals with 100 ml of warm water at 44°C for ∼50 s. A 100-ml syringe was used in conjunction with a flexible plastic tube placed in the external auditory meatus of each ear. The head was slightly elevated (20°) for optimum stimulation of the horizontal canal. To monitor the effects, the horizontal DC electro-oculogram was recorded with the help of two silver–silver chloride electrodes placed at the lateral canthi of each eye (ground electrode over the glabella).

For calibration purposes, the subject was asked to look at targets 0°, 20° to the left and 20° to the right. The calibration saccades recorded were used for the analysis of the slow phase velocity (SPV) of caloric nystagmus, which was determined at the end of caloric irrigation and during the scanning period. After calibration, the subjects kept their eyes closed continuously during the control condition, and during and after caloric irrigation. After the scans, the subjects were asked to comment on their perception of motion. The subjective strength of the vestibular sensation was quantified by determining the perceived intensity of body rotation on a visual analogue scale (0–5). There were three stimulus conditions applied in random order with the eyes closed: (A) caloric irrigation of the right ear; (B) caloric irrigation of the left ear, and (C) the rest condition without vestibular stimulation.

PET scanning and image reconstruction

PET measurements were performed using a Siemens 951 R/31 PET scanner (CTI Knoxville, TN, USA) in 3D mode with a total axial field of view of 10.5 cm and no interplane dead space under standard resting conditions. Attenuation was corrected using a transmission scan with an external ⁶⁸Ge/⁶⁸Ga ring source obtained prior to the tracer injection. A semibolus injection of 7.5 mCi/run H₂O¹⁵ was administered intravenously over 35 s using an infusion pump.

A dynamic acquisition protocol was performed starting with the infusion. It consisted of a sequence of eight short-duration frames: one 15-s frame followed by seven 10-s frames, covering a total scan time of 85 s. After corrections for randoms, dead time and scatter, images were reconstructed by filtered back-projection with a Hanning filter (cut-off frequency 0.4 cycles/projection element), resulting in 31 slices with a 128 × 128 pixel matrix (pixel size 2.0 mm) and interplane separation of 3.375 mm. Since time-activity curves of the whole brain showed initial tracer appearance during scan 4 and its maximum between scans 4 and 8 in all subjects, frames 4–8 were added to a single frame consisting of 50 s for further analysis. Each condition was repeated up to four times (typical configuration ABCCBAACBBCA). The initial stimulation condition was thus varied in random order. To avoid the influence of non-vestibular effects of caloric stimulation (somatosensory and acoustic sensations), scans that were sampled for 50 s were started 25 s after irrigation ended. Previous tests had revealed that nystagmus is still prominent at this time (for details see Wenzel et al., 1996; Dieterich et al., 2003).

Statistical PET analysis

Tracer counts were normalized proportionally to the global cerebral activity, which was arbitrarily set to 1000, in order to perform analysis on relative tissue regional cerebral blood flow (rCBF) activity (Fox and Raichle, 1984). An automated program (NEUROSTAT; University of Michigan, Ann Arbor, MI, USA) was used to co-register, reslice and transform the image arrays into the stereotactic space of Talairach and Tournoux (1988) as described previously (Minoshima et al., 1993, 1994). To eliminate individual differences in gyral anatomy, these images were further smoothed with a three-dimensional Gaussian filter to give an effective resolution of ∼12 mm (full width half maximum). Repeated control and activation images (four scans for each condition) were averaged for each within a subject by calculating the global mean for each voxel. Differences between control and activation images were then averaged across subjects and were expressed as voxel-by-voxel t-statistic values using a pooled variance estimated from the whole brain grey matter (Worsley et al., 1992). Since the resulting t-statistic map is known to approximate closely to a standard Gaussian distribution (Worsley et al., 1992), these values were described as Z-scores. To determine a threshold for significant activation on the resulting t-map, we calculated the image smoothness (Friston et al., 1991) and estimated a statistical threshold at a one-tail (positive) probability of P = 0.05 using a statistical model that adjusts multiple comparisons and inherent correlation of neighbouring pixels (Montreal threshold) (Worsley et al., 1992). Statistical parametric maps were defined as ‘activation maps’ using the contrasts A versus C and B versus C, and ‘deactivation maps’ using the contrasts C versus A and C versus B.

For the areas attributed to multisensory vestibular functions, for which we had a theory-driven, a priori hypothesis, a Z-score >3 in the respective regions was considered representative of a significant change in regional CBF (rCBF). This corresponds to t-values that, without correction for multiple comparisons, achieve a probability of P < 0.0001 (Kosslyn et al., 1994). Beside the temporo-insular and the temporo-parietal cortex, the activated areas included in the analysis based on the a priori hypothesis were: the anterior insula and adjacent inferior frontal gyrus; putamen and caudate nucleus; precuneus; precentral gyrus; medial, superior and diagonal frontal gyri; anterior cingulum; hippocampus; thalamus; and cerebellum (Lobel et al., 1998; Bense et al., 2001; Suzuki et al., 2001; Fasold et al., 2002; Dieterich et al., 2003; Emri et al., 2003).

Results

Psychophysical and ocular motor data

None of the patients complained of vertigo, dizziness or oscillopsia. All but one had mild contralateral hemiparesis and one had contralateral hemiataxia (see Table 1). Five of the eight patients had moderate difficulties in stance and gait, and exhibited a tendency to fall to one side. Seven patients presented with abnormal tilts of the SVV amounting to 4.2–10.1°, with a mean of 5.9° (normal range ±2.5°; Dieterich and Brandt, 1993a). The direction of SVV tilt was ipsilateral in three patients and contralateral in four patients (Table 2). The one patient (BS) with normal SVV measurements had a longer period after disease onset of 55 days, whereas the other seven patients had a range of 4 to 35 days (mean 18 days) after disease onset. From earlier studies, it is known that SVV tilts in thalamic lesions recover spontaneously within 30 to 40 days (Dieterich and Brandt, 1993b), thus explaining the normal data of patient BS.

Measurements of tonic ocular torsion of both eyes were available in seven patients; all were within the normal range.

Caloric irrigation in PET

Warm-water vestibular stimulation of one ear caused a sensation of being tilted toward the other ear and a caloric nystagmus to the contralateral ear (direction of the slow phase). Measurements of intensity of motion sensation and caloric nystagmus were made during each stimulation period in PET, i.e. four times in each patient for each ear. The mean intensity of motion sensation was 1.0 for all patients when the ipsilesional ear was stimulated and 1.5 when the contralesional ear was stimulated (for details, see Table 2). The difference was not significant between the ipsilesional and the contralesional ear, nor for left- and right-sided lesions (paired t-test: 0.067) (Fig. 1). Mean slow phase velocity of the caloric nystagmus was 27.0°/s during stimulation of the ipsilesional ear and 36.1°/s during stimulation of the contralesional ear. This trend to a stronger effect if the ear contralateral to the lesion was stimulated also showed no statistical significance as determined by one-way ANOVA (analysis of variance).

PET: cortical areas activated during caloric irrigation

Patients with lesions of the left posterolateral thalamus

In the four patients with left-sided lesions, stimulation of the ipsilesional left ear caused activations within the left hemisphere located in the caudate nucleus, the lingual gyrus (BA 18), the anterior and medial parts of the insula, the subcallosal gyrus (BA 25), the medial temporal gyrus (BA 21/37), and the superior frontal gyrus (BA 10) (Table 3; Figs 2 and 3). The only activations of the right side were found in the cerebellar vermis and the anterior cingulate gyrus. Thus, the activation pattern within the left hemisphere was reduced and impaired, showing no activity of the posterior insula, superior temporal gyrus, or inferior and superior parietal lobe. Instead, there were activations of areas within the left visual cortex (BA 18), which are deactivated in normal subjects. Furthermore, apart from very small activations in the anterior cingulate gyrus and cerebellar vermis (both of which belong functionally to the activations within the left hemisphere), simultaneous activations of the right hemisphere were completely lacking.

Caloric irrigation of the contralesional right ear induced widespread activations only within the right hemisphere. The largest clusters were located in the posterior (Z-score = 4.88) and anterior (Z-score = 4.75) insula and adjacent superior temporal gyrus (BA 22), followed by activations in the inferior/medial frontal gyrus (BA 45/46), caudate nucleus, paramedian thalamus, hippocampus, cerebellar vermis, inferior frontal gyrus (BA 44/45), inferior/medial frontal gyrus (BA 10, 47), inferior and superior parietal lobule (BA 40), superior frontal gyrus (BA 9/10) and medial/superior frontal gyrus (BA 8) (Table 3; Figs 1 and 3).

This activation pattern of the right hemisphere represents the typical one seen in healthy right-handed volunteers when identical methods are used for vestibular stimulation of the right ear only for the right hemisphere (for comparison, see Dieterich et al., 2003). The only activated area within the left hemisphere of the patients was found in the superior frontal gyrus/anterior cingulate gyrus (BA 11). The latter is typically located contralateral to significant activations of multisensory vestibular cortex areas and therefore belongs to the circuitry of vestibular cortical areas (Guldin and Grüsser, 1996) of the right hemisphere. Such a severely reduced activation within the left hemisphere (functionally no activation at all) was not found in healthy volunteers subjected to identical stimulation conditions. Thus, the activation pattern during stimulation of the non-affected right side appeared normal for the right hemisphere, but was completely missing for the affected left hemisphere.

These widespread and strong activations within the right hemisphere allow the further conclusion that the diminished activation patterns during the other stimulation conditions were not caused artificially by too few patients or an inappropriate patient selection; they seem to be reliable results.

Patients with lesions of the right posterolateral thalamus

Warm-water vestibular stimulation of the ipsilesional right ear caused regional blood flow increases of the right hemisphere in the following areas: the medial temporal gyrus adjacent to the medial occipital gyrus (BA 19/37; area MT/V5); the hippocampus/inferior temporal gyrus (BA 34); the amygdaloid body; posterior cingulate gyrus (BA 23/29); the inferior frontal gyrus; and the cerebellum. No activations were found in the insular region of the right hemisphere. Activations of the left hemisphere were located in the posterior insula and retroinsular region, reaching superiorly into the inferior parietal lobule (Z-score = 4.0) and more inferiorly into the putamen, the medial temporal gyrus adjacent to the medial occipital gyrus (BA 37/39; upper parts adjacent to MT/V5), and the superior temporal gyrus (BA 22) (Table 3; Figs 2 and 4). Thus, compared with healthy volunteers (Dieterich et al., 2003), the activation pattern of the right hemisphere was severely reduced, while that of the left hemisphere was moderately reduced, showing some but not all of the multisensory vestibular areas.

Caloric irrigation of the contralesional left ear led to activations only of the hippocampus, the inferior posterior insula and the gyrus rectus (BA 11) within the right hemisphere, and to activations of the anterior and posterior insulae and the adjacent inferior frontal gyrus, putamen, subcallosal gyrus (BA 25) and superior temporal gyrus (BA 47/38) of the left hemisphere (Table 3; Figs 2 and 4). Thus, the activation pattern included a few of the areas within the left temporo-insular region known from healthy subjects, but it lacked others in the upper parts of the left temporo-parietal cortex, and was severely reduced within the right hemisphere.

PET: Cortical areas deactivated during caloric irrigation

Patients with lesions of the right posterolateral thalamus

During caloric irrigation of the ipsilesional right ear areas of rCBF decrease (i.e. deactivations) were located in the left lingual gyrus (BA 18/19; Z-score = 4.53) and left cuneus (BA 18), the medial temporal gyrus (BA 21) bilaterally, and the culmen of the right cerebellar vermis (Table 4). Thus, deactivations within the visual cortex did not appear bilaterally, which is also the case in healthy volunteers (Wenzel et al., 1996; Bense et al., 2001), but mainly in the hemisphere contralateral to the lesion.

Caloric irrigation of the contralesional left ear led to deactivations of the posterior cingulate gyrus (BA 29) and the precuneus (BA 31) of the left hemisphere, whereas more deactivations were seen within the right hemisphere, e.g. in the superior temporal gyrus (BA 38), medial temporal and medial occipital gyrus (BA 19), and culmen of the cerebellar vermis (Table 4).

Patients with lesions of the left posterolateral thalamus

During caloric irrigation of the ipsilesional left ear, deactivations were located in the hippocampus (BA 28/35) and visual cortex of only the contralateral right hemisphere [the hippocampal gyrus, lingual gyrus (BA 18) and medial occipital gyrus (BA 19/18)] (Table 4). Caloric irrigation of the contralesional right ear led to deactivations primarily within the left hemisphere only; they were located in the medial and inferior frontal gyri, caudate nucleus, lingual gyrus, anterior and medial part of the insula, medial temporal gyrus and inferior parietal lobule (Table 4).

Discussion

Vestibular dysfunction in patients with thalamic lesions

We know from patients with acute unilateral thalamic infarctions that only lesions of the posterolateral region cause transient vestibular signs and symptoms such as perceptual deficits with ipsi- or contralateral tilts of the subjective visual vertical and corresponding deviations of stance and gait. However, they cause no ocular motor deficits (Dieterich and Brandt, 1993; 2001). These signs and symptoms of vestibular imbalance are probably identical with the syndrome called earlier ‘thalamic astasia’, a condition of irresistible falls without paresis or sensory or cerebellar signs (Masdeu and Gorelick, 1988). Like these earlier findings of a tonic vestibular deficit in patients with posterolateral infarctions, our patients also presented with ipsi- or contralateral tilts of the SVV and postural imbalance. In addition, seven of eight patients had mild contralateral hemiparesis and one contralateral hemiataxia, signs of which were due to ischaemia or oedema of the adjacent internal capsule, which is regularly seen in combination with the vestibular signs (Barth et al., 2001; Bogousslavsky and Caplan, 2001; Pullicino, 2001). Thus, this combination of signs and symptoms was compatible with a lesion of the vestibular relay station within the posterolateral thalamus, the function of which was described earlier in humans and animals.

Neurophysiological data and tracer studies of the posterolateral thalamus in cats and monkeys describe these subnuclei as a relay station for vestibular cortex areas (Sans et al., 1970; Deecke et al., 1973, 1974; Büttner and Henn, 1976; Büttner and Lang, 1979), since vestibular stimulation of the animals elicited corresponding responses in their neurons. Most ascending vestibular fibres in the subnuclei such as VPLo were in close contact, intermingled with proprioceptive relay cells, and also responded to optokinetic stimuli (Büttner and Lang, 1979); this finding argues for an integration of vestibular and proprioceptive information at the thalamic level. Electrical stimulation of the posterolateral thalamic subnucleus Vim (nucleus ventro-oralis intermedius corresponding to VPLo) in humans elicited a rotation or spinning of the body, head or eyes either counterclockwise (more often) or clockwise (Hassler, 1959; Tasker and Organ, 1971; Tasker et al., 1982). These signs during stimulation of thalamic neurons were in agreement with the human ‘vestibular thalamic deficits’ described earlier in patients with posterolateral thalamic infarctions (Dieterich and Brandt, 1993b) as well as the patients in our current study. However, anatomical differences and divergent nomenclature make it difficult to correlate human and animal thalamic structures directly. Jones (1990) compared monkey and human data and proposed a carefully revised nomenclature for the ventral nuclei, according to which the ‘vestibular’ thalamic nuclei in humans were considered to be the ventral parts of nucleus ventrolateralis, pars posterior (VLp), nucleus ventroposterior lateralis, pars posterior (VPLp) and nucleus ventroposterior inferior (VPI) (for details, see Dieterich and Brandt, 1993b). The preferably affected subnuclei in the current study were the VPLp and dorsal and ventral parts of the VLp.

The data from the current PET study involve various aspects of vestibular connectivity, vestibular–visual interactions and interhemispheric interactions of the cortex (Fig. 5). These are discussed separately as to their anatomy and function below.

Differential effects of unilateral caloric irrigation in patients with unilateral vestibular thalamic lesions

A unilateral lesion of the vestibular relay station caused the following effects within the multisensory cortical network of both hemispheres involved in the processing of vestibular information during caloric stimulation. Our major findings of the group analyses of patients with right-sided and left-sided thalamic lesions in H₂O¹⁵-PET were fourfold.

• The activation of multisensory vestibular cortex areas was significantly reduced in the ipsilesional hemisphere when the ear ipsilateral to the thalamic lesion was stimulated.

• The activation of multisensory vestibular cortex areas of the hemisphere contralateral to the irrigated ipsilesional ear was also diminished but to a lesser extent.

• Right hemispheric dominance was preserved in the patients with right-sided and left-sided lesions.

• Compared with the deactivation pattern in healthy volunteers during vestibular stimulation, which is always located bilaterally and mainly in the visual and somatosensory cortex areas (Wenzel et al., 1996; Bense et al., 2001), the deactivations in our study were remarkable. They were reduced and predominantly located within only one of the hemispheres, the one contralateral to the stimulated ear and therefore contralateral to the activations, which were mainly ipsilateral to the stimulated ear. Accordingly, posterolateral thalamic lesions in our patients not only affected the patterns of activation in both hemispheres, but also the patterns of simultaneous deactivation.

During irrigation of the ipsilesional ear, the activations of contralateral vestibular cortex areas were most often diminished. Single patients (e.g. BS) even showed no activation of the two hemispheres during vestibular stimulation of the ear ipsilateral to the lesion site (Fig. 2), and caloric nystagmus and perception of rotation were either absent or minimal. Furthermore, the activation within the left hemisphere in patients with left-sided thalamic lesions was significantly diminished when the contralateral right ear was stimulated, although the activation pattern within the right hemisphere appeared normal. The hemispheric asymmetry of the cortical activation pattern during vestibular stimulation of the horizontal semicircular canals, however, was not associated with a significant directional asymmetry of caloric nystagmus or motion perception for the entire group (t-tests not significant). There was only a trend of the caloric nystagmus to be stronger during stimulation of the ear contralateral to the lesion side. Any further functional interpretation should proceed cautiously as only eight patients were involved in this study.

There was a striking contrast between the significant hemispheric differences in the mediation of vestibular input (activations) and the minimal vestibular signs and symptoms of the patients. One explanation could be that the calorically induced vestibular nystagmus appears to be mainly mediated by a subthalamic brainstem VOR circuitry and the vestibular cerebellum rather than by thalamo-cortical structures. This was also reflected by the absence of spontaneous vestibular nystagmus and rotational vertigo in our patients with acute and subacute lesions of the vestibular thalamus. The unaffected perception of body motion may be associated with the bilateral activation of visual motion-sensitive cortical areas in the temporo-occipital cortex (V5), which were largely activated independently of the lesions in both hemispheres (Fig. 3A).

The right hemispheric dominance in right-handers described above was preserved in those with right-sided or left-sided thalamic lesions. The absence of activation in both hemispheres was more pronounced in patients with right-sided than with left-sided lesions. Thus, the data demonstrate both the functional importance of the dominance of ipsilateral ascending pathways and the vestibular dominance of the non-dominant hemisphere (here right hemisphere in right-handers) (Fig. 5). Both factors were shown to be relevant for the processing of vestibular information.

Our data on the patients with posterolateral thalamic lesions fit the concept recently reported for healthy right-handers and left-handers during caloric vestibular stimulation (Dieterich et al., 2003). The importance of ipsilateral projections for the bilateral ascending vestibular pathways was a basic, and especially novel, finding in this lesion study, which further elucidated the role of the projections in normal subjects. If there had been an acute disruption of one projection of the bilateral pathways, theoretically one would expect all information to instead travel immediately via the preserved projection, even if it were the weaker one under normal conditions. However, this did not seem to be the case in our patients.

An exceptional finding was the preserved activation of an area in the inferior part of the insula (Fig. 1b, 4A: Z = −4) which, in normal volunteers, is confluent with the activation of the posterior insula, e.g. the PIVC. This inferior insula was also activated in the hemisphere ipsilateral to the lesion when the ipsilesional ear was irrigated. This supports the assumption that there is a bilateral direct vestibular projection from the vestibular nuclei to the inferior part of the insula, which bypasses the posterolateral thalamus (Fig. 5).

Alternatively, non-thalamic transcallosal interhemispheric transmission might be involved. This is less likely, however, because in such a case one would expect activation of all vestibular cortical areas within the lesioned hemisphere due to their intimate connections as described in the Introduction. The functional significance of this preserved area requires further investigations, especially to determine if it contributes to the relatively mild and transient vestibular symptoms of these patients.

Caloric irrigation in healthy volunteers not only activates vestibular cortex areas but at the same time deactivates visual cortex areas bilaterally (Wenzel et al., 1996; Bense et al., 2001). In the current study, deactivations of visual cortex areas were most often found in only one hemisphere, namely in the hemisphere contralateral to the stimulation and contralateral to activated vestibular cortex areas. Thus, the normal interaction between the two sensory systems, the vestibular and the visual, described earlier as a reciprocal inhibitory interaction (Brandt et al., 1998a), is impaired in our patients. Since the activations were mainly located in one hemisphere ipsilateral to the stimulated ear, the deactivations were located in the other hemisphere contralateral to the activations. One can thus speculate that the inhibitory interaction between the visual and the vestibular systems is organized by pathways that cross the hemispheres and which were preserved in our patients. Further functional interpretations appear premature at this stage.

Particular importance of ipsilateral ascending pathways for vestibular function

Animal studies have identified a network of cortical and subcortical multisensory areas responsible for the processing of vestibular information, which has strong bilateral connections to an ‘inner vestibular circle’, i.e. PIVC, area 2v (7ant), and area 3aV (Akbarian et al., 1993; Guldin and Grüsser, 1996). The vestibular nuclei in the medullary brainstem receive monosynaptic input from all these cortex regions, and their cortico-vestibular projections run bilaterally to the ipsilateral and contralateral vestibular nuclei. Interestingly, the contribution of ipsilateral and contralateral pathways varies for the different cortex areas (Guldin and Grüsser, 1996). The areas of the posterior insular region—PIVC, VTS and granular insula—generally project to and from the ipsilateral vestibular nuclei, whereas the premotor area 6, the somatosensory areas 3aV and 3aH, and the area 2v have more pronounced projections to the contralateral vestibular nuclei (Akbarian et al., 1993; 1994; Guldin et al., 1993). Thus, the ipsilateral projection to and from the PIVC, adjacent VTS, and the granular insula can explain why the side of the stimulated ear is important in healthy volunteers as well as in our patients with posterolateral thalamic lesions (Fig. 5).

Particular importance of handedness for vestibular function

The other highly relevant factor for the processing of vestibular information is the dominance of the non-dominant hemisphere as defined recently in detail in right-handed and left-handed healthy volunteers during identical stimulation conditions (Dieterich et al., 2003). The cortex areas with vestibular input were activated bilaterally in all right-handers, but there was a significant preponderance of the non-dominant right hemisphere. This dominance of the non-dominant hemisphere was even more pronounced in the healthy volunteers with strong left-handedness. A significant right hemispheric dominance for vestibular cortex areas had been assumed earlier in right-handers in the temporo-insular-parietal region with optokinetic stimulation (Dieterich et al., 1998; Brandt and Dieterich, 1999) and vestibular stimulation (Bense et al., 2001; Suzuki et al., 2001; Fasold et al., 2002). Furthermore, in right-handed healthy volunteers who performed a line bisection task with and without galvanic stimulation of the right or left vestibular nerve, the relevant cortical areas for the processing of vestibular information were shown to be the posterior insula bilaterally, but right more than left. The relevant cortical areas for egocentric perception of space and body orientation could be ascribed to the parietal lobule and the prefrontal cortex, both only within the right hemisphere (Fink et al., 2003). Our study on right-handed patients with posterolateral thalamic lesions underlines the dominance of the non-dominant right hemisphere; this dominance is maintained even when damage occurs within the bilateral cortical network at the thalamic level.

Interhemispheric connections

The fact that caloric irrigation of the unaffected right ear of patients with left-sided thalamic lesions led to a ‘normal’ activation pattern within the right hemisphere but no activation within the lesioned left hemisphere, except for the anterior cingulum in the frontal lobe (Fig. 1A), argues for the thalamus being a gate to the whole hemisphere. This finding could mean that there are no interhemispheric connections outside the thalamus between the temporal, insular and parietal cortices or between larger parts of the frontal cortex. However, there is one exception, i.e. the activation of the anterior cingulum in the superior frontal lobe. Activation occurs regularly within the hemisphere contralateral to the stimulated ear (Fig. 3A), even if this hemisphere has the lesioned thalamus (Fig. 1A). This exception can probably be attributed to crossed thalamo-cortical projections to the frontal cortex, which were shown in tracer studies in mice to represent bilateral components of the fronto-thalamic system (Carretta et al., 1996). Furthermore, the anterior cingulum belongs to the circuitry of vestibular cortical areas as described by Guldin and Grüsser (1996) in monkeys.

The only relevant interhemispheric projection in our study appeared between both temporo-occipital areas representing the motion-sensitive area MT/V5, i.e. between parts of the visual cortex bilaterally. During calorics of the right ear of patients with right-sided thalamic lesions, a significant activation occurred within the posterior insula and the parietal lobule within the left hemisphere. This was associated with an activation of temporo-occipital areas (MT/V5) bilaterally but not with an activation of vestibular areas within the ipsilateral right hemisphere (Fig. 4A). The activation of MT/V5 was unexpected, since there was no visual motion stimulation in the patients, who kept their eyes closed in darkness during caloric irrigation. A direct callosal connection between right and left middle temporal/medial superior temporal areas (MT/V5) was assumed (Brandt et al., 1998b), because sustained extrastriate cortical activation of MT/V5 without visual awareness had been found in functional MRI (fMRI) studies of hemianopic patients (Brandt et al., 1998b; Goebel et al., 2001). The bilateral MT/MST activation may be based on direct ipsilateral or transcallosal visual input. Transcallosal connections between areas MT/MST (V5) have been described in monkeys (Maunsell and van Essen, 1987). On the other hand, ipsilateral visual information can bypass V1 along the superior colliculus-pulvinar route and the lateral geniculate body nucleus route to the prestriate cortex as described in monkeys (Fries, 1981; Girard et al., 1992). However, these latter pathways were not involved in our study since there was no visual stimulation during caloric irrigation with the eyes closed. Thus, apart from the transcallosal projection between both temporo-occipital areas, there seem to be no further relevant interhemispheric connections of the vestibular system that could allow for compensation of the posterolateral thalamic lesions.

In a series of studies, we have tried to elucidate the pathways and function of the vestibular system by imaging cortical activity in humans with acute unilateral vestibular lesions at the vestibular nerve (acute vestibular failure by vestibular neuritis: Bense et al., 2004), the vestibular nuclei (Wallenberg's syndrome by medullary infarction: Dieterich et al., 2005) and the posterolateral thalamus level. Further studies are required on strategic lesions of the vestibular cerebellum, the medial longitudinal fasciculus (MLF) and vestibular cortex areas, especially of the PIVC in the posterior insula, in order to construct a more conclusive model of central vestibular disorders and their repair by compensation and substitution over time.

We wish to thank Judy Benson for her critical reading of the manuscript. This work was supported by the Wilhelm Sander-Stiftung and the German Research Foundation (DFG: Di 379/4–2; Br 639/6–2).

References

Author notes

Departments of 1Neurology and 2Nuclear Medicine, Johannes Gutenberg University, Mainz, 3Department of Neurology, Ludwig Maximilians University and 4Department of Nuclear Medicine, Technical University, Munich, Germany