Objective Vascular lesions of the posterolateral thalamus typically result in a somatosensory syndrome in which some patients develop central neuropathic post-stroke pain (CPSP). Damage to the spinothalamic tract terminus is assumed to be a prerequisite for thalamic CPSP. At the nuclear level, it remains a matter of debate whether the ventral posterolateral nucleus (VPL) or the posterior portion of the ventral medial nucleus (VMpo) constitutes the decisive lesion site. The hypothesis of the study was that lesion location in thalamic CPSP patients differs from that in thalamic stroke patients without pain, and the aim was to identify whether this difference comprises the VPL and/or the VMpo.
Design 30 patients with chronic thalamic stroke and a persistent contralateral somatosensory syndrome were examined. CPSP patients (n=18) were compared with non-pain control patients. By coregistration of a digitised thalamic atlas with T1 weighted MR images, lesion clusters were allocated to the thalamic nuclei.
Results VPL was affected in both groups, but CPSP lesion clusters comprised the more posterior, inferior and lateral parts of the VPL compared with controls. Additional partial involvement of the VMpo was seen in only three pain patients. In three other pain patients, lesions involved neither the VPL nor the VMpo, but mainly affected the anterior pulvinar.
Conclusion This study specifies the role of the VPL in thalamic CPSP and shows that the posterolateratal and inferior parts in particular are critically lesioned in pain patients. In this thalamic subregion, afferents of the spinothalamic tract are known to terminate. In contrast, the data do not support a pivotal impact of the VMpo on thalamic CPSP.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Any stroke affecting the somatosensory pathways from the medulla to the cortex may lead to chronic neuropathic central post-stroke pain (CPSP1). Lesions of the posterolateral part of the thalamus are associated with a considerable risk for the development of CPSP.2 This region comprises the principal somatosensory relay nuclei such as the ventral posterolateral nucleus (VPL) and the ventral posteromedial nucleus (VPM). Patients with posterolateral thalamic strokes present with a contralateral somatosensory syndrome often distributed to the hands and face.3 Notably, only about one-fifth of these patients develop thalamic central pain,4 ,5 a syndrome which was first been described over 100 years ago.6 ,7 CPSP usually evolves within weeks or months after stroke and can severely impair the quality of life.2 ,8 ,9
It is still controversial as to whether the development of thalamic CPSP depends on the lesion of the particular somatosensory nuclei.10 ,11 Evidence suggests that involvement of the spinothalamic tract (STT) is a necessary prerequisite for CPSP12; therefore, thalamic post-stroke pain most likely results from a lesion of the thalamic STT target nuclei. On the one hand, there is evidence that STT predominantly terminates in the VPL and VPM.13–16 Accordingly, imaging studies in CPSP patients have repeatedly linked posterolateral thalamic stroke to the involvement of the VPL and/or VPM1 ,4 ,5 ,17–19 although a digital atlas based assignment of stroke lesions to thalamic nuclei has not been done.
On the other hand, the posterior portion of the ventral medial nucleus (VMpo) has been suggested as a distinct and decisive thalamic terminus of STT projections from the dorsal horn lamina 1.20 ,21 Although it has been proposed that lesions to the VMpo are crucial for thalamic pain syndromes,22 evidence from imaging studies in patients with thalamic CPSP is lacking.
In order to reliably allocate stroke lesions to thalamic nuclei, atlas based mapping of MR images were used in two case reports.23 ,24 A systematic comparison of patients who suffer from thalamic pain with those with similar posterolateral thalamic infarcts but without pain has not been described previously.
Here we examined 30 patients in the chronic phase after stroke of the posterolateral and inferior portion of the thalamus with a corresponding contralateral somatosensory syndrome. Patients were assigned to a pain (CPSP) or a non-pain (control) group. We hypothesised that patients with thalamic stroke and CPSP differ from thalamic non-pain patients with respect to localisation of the stroke lesions within the thalamus and the particular thalamic nuclei involved. We aimed to characterise whether this difference involves the VPL and/or the presumed location of the VMpo.
Patients and methods
Patients and neurological examination
The study was approved by the local ethics committee. Written informed consent was obtained from each patient prior to study inclusion. Patients admitted to our hospital between 2000 and 2010 and registered at our inhouse stroke database with a diagnosis of chronic solitary ischaemic or haemorrhagic thalamic stroke accompanied by contralateral somatosensory symptoms were included in this analysis. During the first phase of the study, patients were recruited retrospectively from the database, while during the later phase, nine patients (four controls) were recruited as part of a larger and still ongoing prospective study. Exclusion criteria were: (a) lesions in other parts of the somatosensory pathway, (b) moderate to severe leucencephalopathy or (c) any other musculoskeletal, internal or neurological disorder potentially confounding the aetiology of their pain syndrome (ie, including other entities of post-stroke pain such as shoulder pain or spasticity). Each patient underwent a detailed history taking and neurological examination, with emphasis on testing of positive and negative somatosensory symptoms. Sensory evaluation comprised: light touch (cotton wool), temperature (test tubes filled with water at 20°C and 40°C), algesia (pinprick), vibration (128 Hz tuning fork), joint position sense, stereognosis and graphesthesia. Somatosensory positive signs assessed were: paraesthesia, dysaesthesia (spontaneous or evoked) and allodynia (dynamic touch, warm and cold temperature). All sensory modalities were tested at least in the face, the forearm, the palm and dorsum of the hand, the lower leg and the dorsum of the feet. Sensory modalities were categorised as not present, diminished, normal or increased, compared with the non-affected corresponding contralateral body site. Areas of maximum positive (evoked) sensory signs were recorded.
CPSP was defined here as spontaneous pain or allodynia which emerged within months after stroke, located within a distribution of a somatosensory deficit corresponding to a vascular thalamic lesion confirmed by MRI.2 ,9 Based on these criteria, patients were assigned either to the CPSP or (non-pain) control group.
MRI and localisation of thalamic lesions
In 21 patients, MRI was performed on a clinical 1.5 T whole body scanner (Vision; Siemens, Erlangen, Germany) using a standard head coil. Nine patients were examined in a 3 T clinical Scanner (TIM Trio; Siemens). To screen for thalamic and extrathalamic cerebral lesions, transversal T2 weighted images of the whole brain and TIRM (turbo inversion recovery magnitude) images of the diencephalon were acquired at 1.5 T. At 3 T, transversal whole brain images were acquired using a FLAIR sequence. In all patients a three-dimensional dataset at high resolution using a magnetisation prepared rapid gradient echo (MPRAGE) imaging sequence was acquired that served for the atlas to MRI coregistration (both: spatial resolution 1.0×1.0×1.0 mm3; at 1.5 T: TR=2480 ms, TE=4 ms, flip angle=12°; at 3 T: TR=1900 ms, TE=2.5 ms, flip angle=9°).
Coregistration with thalamic atlas and lesion group analysis
All thalamic lesions were delineated manually and voxelwise from MPRAGE volumes by three blinded raters (TK, SP, MEL) and saved as ‘regions of interest’ (ROIs) using MRIcro (http://www.mricro.com). MPRAGE datasets were coregistered with coronal section outlines of a digitised version of an anatomical atlas of the human brain.25 This was performed for the thalamus and its adjacent subcortical structures using custom written software. First, the anterior commissure and the intercommissural line were defined as reference axes. The images were then repetitively linearly transformed (translation, rotation, linear scaling) by the raters with respect to visually identifiable structural landmarks, as described previously (eg, caudate nucleus, wall of the third ventricle, putamen).26 If necessary, anatomical datasets were flipped from left to right to ensure that all lesions were located on the same (left) side. The transformation files created during this coregistration procedure were used to spatially transform ROI datasets. MPRAGE datasets as well as ROI images were subsequently re-sliced using SPM8 (Statistical Parametric Mapping; Wellcome Department for Cognitive Neurology, University College London, UK) running on Matlab (V.7.3, The Math-Works, Natick, Massachusetts, USA), to 1.0×1.0×1.0 mm3 and 0.25×0.25×0.25 mm3 isovoxel resolution, respectively.
All MPRAGE datasets were averaged to create an anatomical template for superposition with lesion clusters. Lesion clusters of CPSP and control patients were merged separately for each group, resulting in maps with relative frequency (in per cent) of lesions colour coded for each voxel. Lesion cluster maps of both groups were superimposed onto coronal section outlines of the digitised anatomical atlas25 in order to assess nuclear lesion topography. Because VMpo is not depicted in this atlas, two coronal sections from Blomqvist et al20 were digitised and coregistered with the thalamic atlas manually by four raters (TK, GJJ, SP, PB) to allow for overlay of group clusters.
Using Fisher's exact test of independence,27 error probability values were calculated voxelwise for both lesion clusters in order to determine thalamic subregions associated with CPSP.
A total of 30 thalamic stroke patients (mean age 65.6 (SD 10.4) years, eight women, one patient with a small haemorrhagic stroke restricted to the thalamus) were included in the final analysis. Clinical features of all patients are summarised in table 1. Eighteen patients had a chronic CPSP syndrome with a minimum duration of pain of 2 months. In all CPSP cases, pain was located contralateral to the ischaemic lesion and within the region of their initial somatosensory symptoms. The remaining 12 control patients did not report any pain symptoms but showed persistent non-painful somatosensory symptoms. Patient groups did not differ with respect to age, sex, National Institute of Health Stroke Scale or modified Rankin Scale score at the time of clinical examination. Somatosensory symptoms were present in all patients but alterations of algesia and thermaesthesia as well as positive somatosensory signs such as allodynia, paraesthesias and dysaesthesias were seen more often in CPSP patients. These differences were statistically significant for dysaesthesia and allodynia (Fisher's exact test, p<0.01).
None of the patients included had any lesion affecting other parts of the proposed somatosensory system or the cortical pain network. One patient (pain group) had one additional chronic (tubero-) thalamic lesion contralateral to the posterolateral thalamic stroke lesion. Mean volume of stroke lesions (±SD) was 179 mm3 (±160 mm3) for controls and 170 mm3 (±167 mm3) for the CPSP group, with no significant difference between the groups (Student's t test, p=0.44). In figure 1 stroke lesion clusters of pain and control patients are superimposed on the average T1 weighted image obtained from all patients. Both clusters showed overlap, but differed with respect to cluster boundaries. The group cluster of pain patients extended to more posterior, lateral and inferior parts of the thalamus (figure 1A,B). Lesion clusters were then overlayed onto coronal planes of the digitised thalamic atlas (figure 2),25 with voxelwise colour coding of relative lesion frequency. Both clusters comprised virtually identical thalamic nuclei: VPL, VPM, anterior pulvinar, medial pulvinar and lateral pulvinar, centromedian nucleus, internal part of the ventrolateral posterior nucleus (VLPI), dorsoanterior part of the medial geniculate nucleus, internal medullary lamina and medial dorsal nucleus. However, the CPSP lesion cluster extended more into posterior, lateral and inferior parts of the VPL/VPM and the anterior, medial and lateral pulvinar nucleus, whereas the control lesion cluster reached into more anterior parts of the thalamus such as the VLPI. The internal capsule was lesioned in two of the control patients but in none of the CPSP patients.
Voxels associated with CPSP are shown in figure 3 (p<0.05 and p<0.1, respectively, uncorrected for multiple comparisons). The most significant voxels associated with pain were located within the VPL. The respective voxels associated with the non-pain group were located in the VLPI (not shown).
In order to determine any potential involvement of the VMpo, lesion clusters were overlayed onto modified coronal sections of Blomqvist et al,20 which were realigned beforehand with the customised thalamic atlas. VMpo was partly affected in a minority of three CPSP patients (figure 4). In all patients with a lesion of the VMpo, the VPL and anterior pulvinar were also affected. In non-pain patients, no involvement of the VMpo was observed.
Single subject analysis revealed that in another three of the 18 CPSP patients, neither the VPM/VPL nor the VMpo was affected by their stroke. Enlarged coronal views of their lesions in relation to thalamic nuclei are given in figure 5. In all three patients the pulvinar nucleus was mainly affected.
In 30 patients with chronic posterolateral thalamic stroke and persistent somatosensory symptoms with or without CPSP, we investigated whether CPSP was associated with distinct lesion patterns compared with patients without CPSP. Lesions in pain patients extended to more posterior, inferior and lateral STT target parts of the thalamus, and predominantly occurred in the VPL/VPM and pulvinar nucleus. In a minority of CPSP patients the VMpo was additionally affected, and in three other patients, thalamic CPSP was associated with a pulvinar lesion.
Neuropathic pain is defined as ‘pain arising as a direct consequence of a lesion or disease affecting the somatosensory system’.28 The involvement of the STT is considered to be an essential pathophysiological precondition for the development of CPSP.12 ,29 CPSP patients most often present with alterations of algesia and thermaesthesia (modalities which are mediated via the STT), and with signs of somatosensory hypersensitivity.1 ,8 ,9 ,30 ,31 Accordingly, this pattern was also present in our CPSP patients compared with controls although this difference was significant only for allodynia and dysaesthesia.
Lesion topography, as observed in our patients, was similar to those described in previous reports and comprised two vascular territories: the thalamogeniculate arteries supply to the lateral thalamus, including among others the ventrocaudal nuclei (VPL, VPM) and the rostrolateral portion of the pulvinar.4 ,5 ,32 The posterior choroidal artery supplies the classical posterior thalamic territory.4 Previous studies have shown that strokes in both of these vascular territories result in contralateral somatosensory deficits, as was the case in our patients.
Location and spatial extension of stroke lesion clusters
Although lesion clusters in both patient groups overlapped in certain portions of the thalamus, the outer boundaries of the clusters differed between the groups (figure 1). The CPSP lesion cluster stretched into more posterior, lateral and inferior parts of the thalamus compared with controls. This pattern was most apparent when clusters were superimposed onto coronal sections of the thalamic atlas (figure 2). In terms of thalamic nuclei, the pain lesion clusters and their centres of lesion frequency comprised more posterior, inferior and lateral parts of the VPL/VPM and the pulvinar. Furthermore, voxels in these nuclear subregions were significantly associated with CPSP (figure 3).
The spinothalamic tract and thalamic pain
In contrast, several studies have provided evidence for the involvement of the VPL/VPM in stroke lesions in pain patients. It has been shown in non-human primates that the STT predominantly terminates in both somatosensory nuclei.13 ,14 ,16 Not only do neurons in these nuclei respond to diverse peripheral painful stimulation,33–37 but studies have also shown that perioperative microstimulation of these thalamic nuclei leads to painful sensations in patients.38 ,39 Several clinical imaging studies of thalamic CPSP patients have repeatedly identified inferolateral and posterior lesions that presumably involved the VPL/VPM.1 ,4 ,5 ,17–19 ,23 ,24 Given that all of our patients by definition had persistent somatosensory symptoms, one might expect that the VPL/VPM as the principal somatosensory nuclei should be affected in most patients, irrespective of post-stroke pain. In line with this, we found that the VPL/VPM was also lesioned in control patients. Thus one might ask whether lesions in the VPL/VPM play any role with regards to CPSP.
However, in our study, the intranuclear pattern of VPL/VPM lesions did indeed differ between patient groups. As mentioned previously, CPSP lesions comprised larger and especially more posterior, inferior and lateral parts of the VPL/VPM compared with controls. Taking a closer look at some of the histological studies, our finding is in congruence with some details of their data. It has been consistently demonstrated that the STT not only terminates in the VPL/VPM per se, but it does so predominantly and clustered in the more posterior, lateral and inferior parts.13–15 ,33 ,40 In patients receiving stereotactic thalamic surgery, intraoperatively evoked painful sensations were seen more frequently at stimulation sites in the posterior and inferior region of the ventrocaudal nucleus which comprises the VPL/VPM.37–39 In our opinion, this strongly suggests that the precise lesion location within the VPL/VPM has a significant impact on the development of CPSP.
In other words, when thalamic lesions comprise more posterior, lateral and inferior parts of these somatosensory nuclei, the spinothalamic afferents are more likely to be affected. Therefore, such lesions are more likely to result in thalamic post-stroke pain.
Is the VMpo affected in CPSP patients?
Craig et al suggested an alternative STT terminus within the thalamus. Monkey and human studies have demonstrated that the STT projections from the dorsal horn lamina 1 are conveyed specifically via the posterior portion of the ventromedial nucleus, termed the VMpo.20 ,21 The authors proposed that a lesion of the VMpo leads to disinhibition of a medial STT pathway projecting to the anterior cingulate cortex, and may function as a prerequisite for thalamic central pain.22 Yet imaging studies failed to show involvement of the VMpo in thalamic CPSP to date.23 ,24 Our data indicate that the pain lesion cluster reached into the lateral and dorsal parts of the VMpo, albeit with a low relative lesion frequency. To the best of our knowledge, this is the first study to show the presence of lesions in the VMpo in a small proportion of CPSP patients. However, of note, in all three patients with lesions of the VMpo, the VPL was also affected. Therefore, despite our observation, an exclusive or pivotal impact of the VMpo in CPSP cannot be derived here. It also has to be noted that the VMpo is not delineated in available human atlases. To overcome this, we coregistered the anatomical outlines of the VMpo20 to our atlas sections, a procedure which implies a residual spatial uncertainty.
The involvement of the anterior pulvinar
Remarkably, the CPSP lesion cluster affected large parts of the anterior and medial pulvinar nucleus, a region which is commonly regarded as a relay site in higher order visual processing (for review see Grieve et al41). One may speculate that the involvement of the pulvinar observed in this study is a mere collateral effect of the VPL/VPM lesions due to a shared vascular supply. In fact, parts of the anterior pulvinar were also affected in control patients, although to a lesser spatial extent. Nonetheless, a number of tracer studies in non-human primates support the view that the anterior pulvinar is indeed part of the thalamocortical somatosensory network. Spinothalamic pathways from lamina I terminate in the anterior pulvinar.13 Accordingly, connections from the anterior pulvinar to the primary as well as the secondary somatosensory cortex have been demonstrated.42 ,43 Clinical case studies of thalamic CPSP patients have reported an additional involvement of the anterior pulvinar alongside lesions of the VPL.5 ,23 Despite these reports, the anterior pulvinar is not considered to be essential for the aetiology of CPSP to date. Our data further support these findings and point to a more prominent role of the anterior pulvinar, at least in some CPSP patients. As an argument against a mere side effect of the vascular supply, three out of the 18 pain patients had thalamic lesions which completely spared the VPL/VPM and the presumed location of the VMpo, but extended largely into parts of the anterior pulvinar in each patient.
Alternative causes for thalamic central post-stroke pain
Apart from lesions of the aforementioned nuclei there are other potential contributors to the emergence of CPSP. Whereas accompanying lesions of the internal capsule were ruled out in our pain patients, involvement of the STT in terms of fibres of passage within the thalamus, especially in pain patients without lesions of the VPL or VPM, cannot be excluded. Although it is presumed that the ventral posterior inferior nucleus also receives STT input,44 the ventral posterior inferior nucleus was not significantly affected in either of the two patient groups, and thus did not play a crucial role in our data. Beyond this, there are other possible pathophysiological mechanisms secondary to the stroke lesion—for example, hyperexcitability or sensitisation of central nociceptive neurons or an imbalance between the lateral (sensory discriminative) pain system (including the somatosensory cortices and the insular) and the medial (affective emotional) pain system (including anterior cingulate gyrus) (for review see Klit et al2). Such alternative or complementary mechanisms for the development of CPSP have to be targeted in future studies.
Our study corroborates and refines the hypothesis that the VPL plays a major role in thalamic pain. Lesions comprising rather posterior, inferior and lateral parts of the VPL are more likely to be associated with the occurrence of CPSP—a finding in congruence with histological and microstimulation studies which show that spinothalamic afferents tend to terminate in this nuclear subregion. Although this is the first study to indicate partial lesioning of the VMpo in some patients with thalamic CPSP, a crucial impact of the VMpo in these patients cannot be confirmed. As an unexpected novel finding, the anterior pulvinar appears to be a further relevant lesion site for the emergence of thalamic post-stroke pain, at least in some CPSP patients.
The authors thank their colleagues Jochen Fiebach, Susanna Frank, Heike Israel, Claudia Kunze, Bianca Müller, Christian Nolte, Nadia Safar and Fabia Wegener, all colleagues from the Stroke Imaging Team and the Trial Team of the Centre for Stroke Research Berlin (CSB) and from the Department of Neurology, for their support and contribution in patient recruitment, study performance and data acquisition. The authors thank Anna Kufner for carefully editing and proof reading the final version of the manuscript. Finally, the authors thank all of the patients who participated in the study.
Funding The project has received funding from the German Federal Ministry of Education and Research (BMBF) via the grant Centre for Stroke Research Berlin (01 EO 0801) and as part of the Competence Net Stroke (01GI9902/4).
Competing interests None.
Ethics approval Ethics approval was provided by the local ethics committee, Charité-Universitätsmedizin Berlin.
Provenance and peer review Not commissioned; externally peer reviewed.