Discussion

I. Surgical thalamic targets in the light of connectivity.

The unique feature of the atlas plates presented here is that unlike those in other stereotactic atlases of the human brain the outlines of thalamic nuclei shown in three standard stereotactic planes are from the same brain. This ensures high compatibility of the target coordinates in the three planes as well as facilitates co-registration of the 3D volume with individual MRI’s. Another important feature is that in addition to the cytoarchitecture the outlines of the motor thalamic nuclei are based on experimental data from tracing of subcortical connections in nonhuman primates and extrapolating these to human thalamus using an immunocytochemical marker that brings out specific features of each subcortical afferent territory ( Kultas-Ilinsky et al., 2011. For more references see  BIBLIOGRAPHY page).

During the early period of stereotactic interventions in the human thalamus discussion of neurosurgeons centered on the question of which of the three nuclei Vim, Vop or Voa of Hassler) is the most efficient for elimination of Parkinsonian tremor. The radiological coordinates of targets were compared and histological analyses of available postmortem samples were performed. The results were then related to clinical outcomes of the cases. Eventually consensus developed that lesions placed in Vim and/or Vop of Hassler provided the best clinical results for tremor of different origin while lesions placed more anteriorly in Hassler’s Voa, Lpo or  ventral anterior nucleus (VA) using Walker’s nomenclature  for the same area, are clinically more efficient for rigidity and torsion dystonia.

At present with the use of modern non-invasive visualization techniques (see review by Frey et al., 2022 [1]) the focus is on how to identify location of ‘Vim’ in brain MR images. Various approaches have been proposed. Still the remarkable resolution provided by modern MRI scanners and scanning protocols are not sufficient for resolving individual nuclei within the large area supplied by motor and somatosensory afferents to the thalamus. However, proposed creative image processing techniques can help in improving it (Corona et al. 2020 [2]). Moreover the use of tract density imaging for simultaneous tracing of connections from cortical and subcortical areas is also very helpful approach (Behrens et al., 2003 [3]; Middlebrooks et al., 2018 [4]; Akram et al. 2018 [5]; Akram et al., 2019 [6]; Coenen et al., 2020 [7]). Point of the overlap of fibers from the primary motor cortex and those from cerebellum that are destined to the thalamus, has been considered to be Vim and used as the surgical target. Results show that stimulating electrodes in this location are indeed very effective (Akram et al., 2018). At the same time comparison of different cases reveals a rather wide spread of effective points that do not all fit into probabilistic Vim. (Fereira et al., 2020 [8]. Besides individual variability used to explain these findings there is another important aspect here that has not attracted much attention. Namely, the problem may be semantic as the majority of the points are still confined to the same larger area that we call ventral VL (VLv).

In our thalamic maps VL refers to the entire cerebellar afferent territory that occupies a substantial part of the thalamus and extends from about 4.50mm medially to 18.00mm laterally whereas its narrow ventral part reaches the ac-pc plane only between 12.5 mm and 15.75 mm from the midline. Consequently Hassler’s Vim and most of the Vop are both part of the ventral VL. Thus Vim and Vop both being an integral part of the extensive cerebellar afferent receiving territory in the thalamus are histologically and functionally identical, and location wise correspond to our VLv. It is this ventral part of VL that over several decades has been proven to be the best target for elimination of tremor (Hassler et al., 1979 [9]; Lenz et al., 1995 [10]; Benabid et al., 1996 [11]; Hamani et al., 2006 [12]). The modern in vivo tract tracing techniques often show probabilistic Vim at the ventralmost boundary of the thalamus almost below grey matter [8]. At the same time the resolution of provided illustrations is not sufficient to resolve the anatomical details of the area therefore it is not clear whether the spot is in the grey or white matter.  In a recent study Neudorfer et al., 2022 [13] identified a subthalamic ‘hypointensity region’ in FGATIR MR images that proved to be clinically the best target for tremor elimination with DBS.  It should be kept in mind that cerebellothalamic fibers enter VL  at almost all its mediolateral coordinates but their highest density in the subthalamus is at the level of ventral VL and situated under it VPi (ventralposterior inferior nucleus a component of somatosensory territory). Thus, this subthalamic region with the highest concentration of cerebellothalamic fibers (erroneously labeled in the current literature as dentatorubro thalamic tract or DRT), as well as the ventral VL are both implicated as effective DBS targets. However, the exact anatomical structure responsible for the positive clinical results still needs to be determined. Comparison of mediolateral stereotactic coordinates of effective spots as well as establishing their positions relative to the ac-pc plane could be very useful for finding anatomical correlates.

In solving this dilemma of grey vs white matter the present atlas may be of help as it has been successfully co-registered with MNI space and it would be interesting to see whether it can be applied in individual surgical cases provided that fast and efficient user friendly software is designed. Together with algorithms, which show highly accurate registration of MNI space resources to individual patient brains (Ewert et al., 2018 [14]), the application of a standardized atlas as the present one bears potential in translational clinical neuroscience.

 

II. Some anatomical considerations in choosing targets for tremor correction.

As noted in the above part two brain regions proved to be effective surgical targets for alleviation of different origin tremors.

One of these targets is a relatively small, well-defined area in the ventral aspect of the VL (VLv) that roughly includes Hassler’s nuclei Vop, Vim and Voi. Another one is significantly larger, yet poorly defined subthalamic region recently referred to as the posterior subthalamic area (PSA) by Neudorfer et al. [15].

In the subthalamic space as such there are two nuclei: the subthalamic nucleus (STN) and zona incerta (ZI). STN that occupies the most ventral anterior position has been a popular target for deep brain stimulation and more recently in ultrasonic lesioning in movement disorders. ZI is located dorsally to STN. Its cells are anatomically and physiologically similar to those of the thalamic reticular nucleus but differ connectionally.  Between these two nuclei as well as dorsally to the ZI is the white matter that consists of different origin fibers.

Within the classical stereotactic coordinate system VLv is topographically always above the intercommissural (CA-CP) plane extending from 2 to 5 mm above it whereas PSA is always below it in the space occupied by white matter, posterior segment of ZI, and ventral posterior inferior nucleus (VPi) . The latter is a nerve cell poor and fiber rich entity, a part of the somatosensory nuclear complex. At the most lateral-posterior levels large part of it is below the CA-CP plane. Hence identifying the exact location and composition of the PSA in MR images is not a simple task  without additional landmarks. In practical terms images of histology based sagittal sections at known distances from the midline could be the most helpful for orienting in this area.

As discussed in Part I there is a consensus among clinicians and experimental neuroscientists that cerebellothalamic connections play a critical role in mechanisms of tremorogenesis. The most recent clinical MRI based studies have identified a hypointensity region in FGATIR images suggesting that it can be the region of concentration of cerebellothalamic fibers while comparative clinical analyses imply that it may be equally good if not better target than VLv for achieving positive effect on tremor using high frequency stimulation [13,15].

When determining the topography of ascending motor related pathways, especially in MRI, one should keep in mind that most of them pass through the subthalamic region, some as compact fiber bundles like pallidothalamic (for example lenticular fasciculus located at the top of STN) and others in an entirely diffuse manner, like nigrothalamic fibers [16]. Distribution patterns of cerebellothalamic fibers are in between of these two extremes since their concentration varies at different locations. As a compact bundle cerebellothalamic together with cerebellorubral fibers form the superior cerebellar peduncle. After decussation in the midbrain the cerebellothalamic fibers run around the red nucleus and ascend towards thalamus at different mediolateral levels, most of them reaching VL through its ventral aspect [17]. The bulk of medial lemniscal fibers enters thalamus posteriorly to the cerebellar fibers but at some levels the two ascend through the same area, i.e. VPi, maintaining the same topographical relationship.

Current state of art MRI research methods do allow visualization of dense fiber bundles (see Part I of Discussion for references) but when the same fiber systems are nearing their termination zones they fan out and become undetectable with these methods. This applies to both pallidothalamic and cerebellothalamic fibers that pass through the subthalamus. In this context placing a couple of classical stereotactic landmarks to MR images may be helpful in establishing the chosen target position in this area.  Such key reference would be the CA-CP line established on the MRI images with marked middle point. Determining the position of the probe relative to it and knowing the image distance from the midline may facilitate comparison of the probe position with that of the target structure as seen in matching atlas plates shown on this site. We believe that until the resolution of the imaging becomes sufficient for identification of individual structures  this extra step may facilitate anatomical verification of the targets in the clinical set-up.

References:

[1] https://doi.org/10.3389/fneur.2022.825178

[2] https://onlinelibrary.wiley.com/doi/10.1002/hbm.24933

[3] https://doi.org/10.1038/nn1075

[4] https://doi.org/10.1016/j.nicl.2018.10.009

[5] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5790021/pdf/main.pdf  

6] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6543124/

[7] https://doi: 10.1016/j.nicl.2020.102165

[8]  https://www.biorxiv.org/content/10.1101/2020.08.05.236679v1.full

[9]  Stereotaxis in Parkinson Syndrome, 1979, Heidelberg: Springer

[10]  https://doi.org/10.1002/mds.870100315

[11] https://doi.org/10.3171/jns.1996.84.2.0203

[12]  https://DOI:10.1227/01.NEU.0000192166.62017.C1

[13] https://doi.org/10.1002/ana.26326

[14] https://doi.org/10.1016/j.neuroimage.2018.09.061

[15] https://doi.org/10.1016/j.neurot.2023.e00313

[16] https://doi.org/10.1002/cne.902360304

[17] https://doi.org/10.1002/cne.902620303