Current-dependent Ocular Tilt Reaction in STN-DBS: evidence for a putative incerto-interstitial pathway?


 Objective: To report a patient with Parkinson’s disease presenting with a combined vestibular, oculomotor and postural syndrome dependent of deep brain stimulation (DBS) of the subthalamic nucleus. Methods: In a systematic monopolar review, eye, head and trunk position in roll and pitch plane were documented as a function of stimulation amplitude and field direction. Repeat ocular coherence tomography was used to estimate ocular torsion. The interstitial nucleus of Cajal (INC), zona incerta (ZI) and ascending vestibular fiber tracts were segmented on MRI using both individual and normative structural and connectomic data. Thresholded symptom-associated volumes of tissue activated (VTA) were calculated based on documented stimulation parameters. Results: Ipsilateral ocular tilt reaction and body lateropulsion as well as contralateral torsional nystagmus were elicited by the right electrode in a current-dependent manner and subsided after DBS deactivation. With increasing currents, binocular tonic upgaze and subsequently body retropulsion could be elicited, consistent with an irritative effect on the Interstitial Nucleus of Cajal (INC). Symptom-associated VTA was found to overlap with the dorsal zona incerta (dZI) and the lateral ipsilateral vestibulothalamic tract (IVTT), while lying in close proximity to the medial IVTT and rather distant to the INC proper. As described in non-human primates, a ZI-to-INC, “incerto-interstitial” tract (IIT) with contact to the medial-uppermost portion of the VTA could be traced. By ways of directional current steering laterally to both tracts, therapeutic response could be preserved while vestibular side effects were minimized. Conclusion: Unilateral stimulation of mesencephalic vestibular-related circuitry induces an ipsilateral vestibular, oculomotor and postural roll-plane syndrome, which converts into a combined pitch-plane syndrome, when functional activation expands to the bilateral INC. The phenomenology of the roll-plane syndrome in this patient points to an activation of INC neurons by DBS, hypothetically via a potentially aberrant incerto-interstitial pathway. Directional current steering proved useful in managing this rare side effect.


Introduction
Deep brain stimulation (DBS) is an established treatment in movement disorders. Connectomic approaches using "volume of tissue activated" (VTA) models shape understanding of DBS action on functional networks and may help optimizing treatment e cacy 1 . In single cases, subtle vestibuloperceptive signs like deviations of subjective visual vertical (SVV) have been reported after DBS 2 .
Vestibular-oculomotor (VOM) and postural side effects, however, are highly unusual, likely due to stimulation targets lying remote of VOM re ex arcs.

History
Since DBS implantation (Boston Scienti c® Gevia/ Cartesia) one year prior, a 58-year-old patient reported to perceive "poles tilted like the Tower of Pisa". He noticed a rightward head tilt and falling to the right but denied vertigo, diplopia, or other focal neurological signs.

DBS programming
Monopolar review (130Hz, 60µs) revealed that OTR was in uenced exclusively and reproducibly by contact 14 (facing supero-postero-medially) of the right lead. OTR magnitude could be modulated linearly starting from 5° * rightward SVV deviation at 1.8mA to approximately 10° at 2.2mA and 25° at 4.5mA ( Fig. 1B) accompanied by increasing rightward lateropulsion, skew deviation, and a torsional nystagmus beating to the left (Supplementary Video). Upon further increase, binocular tonic upgaze and diagonal lateropulsion evolved into retropulsion, requiring the exploration to be stopped immediately. The patient felt "being drawn rightwards". Anterolateral current steering on the same level reduced vestibular effects (7° SVV) while preserving antiparkinsonian effects (UPDRS-III 12 points).

Imaging and VTA modelling
Cranial CT disclosed no intracranial pathologies but an obtuse and caudal trajectory of the right electrode. Lead localization and VTA modelling (Fig. 2) showed VOM-syndrome related VTA extending supero-postero-medially ( Fig. 1B, C, D).

Discussion
Albeit the VTA overlapping with the iVTT (Fig. 1C, D), the evolving VOM-syndrome and degree of SVV deviation 2 argue against the iVTT as its sole correlate. The oculomotor and postural ndings can more likely be related to the INC, which is involved in eye and head coordination in the roll and pitch plane 4 .
Unilateral damage of ascending vestibular projections to the INC at midbrain level induce a contralesional OTR, while damage of the INC per se additionally results in ipsilesional torsional nystagmus 5 . In contrast, stimulation of the riMLF seems unlikely, since it would generate an ipsilateral torsional nystagmus 5 .
Given the binocularity of oculomotor signs in absence of double vision, 3 rd and 4 th cranial nerve involvement can be excluded. Hence, we reason that the current-dependent OTR ipsilateral to the stimulation site may be generated by modulation of projections targeting the INC, while contralateral torsional nystagmus at higher currents speaks for additional activation of integrator neurons directly in the INC 5 . Given the neuroanatomical modelling, we hypothesize INC modulation in this patient through a potentially aberrant "incerto-interstitial" projection, rst described in primates 6 (Fig. 1C). Moreover, simultaneous modulation of medial and lateral iVTT axons may then contribute to the enormous SVV deviation, in line with theoretical considerations of tissue-dependent differential effects of DBS 7 .
Increasing upgaze and body retropulsion implicates bilateral INC involvement. Given the topographical relations, direct current spread over the midline can be excluded as a mechanism. With dense posterior commissure projections interconnecting the bilateral INC, synaptic mechanisms for signal transmission to the contralateral INC could be assumed. The transition of ocular and postural symptoms from roll to pitch plane with increase of stimulation currents is consistent with the "double roll = pitch" framework 4 .
These case-based ndings should of course be interpreted with caution and con rmed in prospective studies, potentially correlating ne-grain oculomotor assays to VTA locations. Nonetheless, this case has three relevant aspects: 1) It gives mechanistic insight into the functional anatomy of plane-speci c control of eye, head and body position in the midbrain, 2) allows inferences about modes of action of DBS in sensorimotor circuits and 3) underscores the utility of directional DBS for optimizing treatment e cacy.

Declarations
Ethics approval and consent to participate and disclose: Patient gave informed written consent to consent, disclose and for publication of this retrospective study.
Data availability: Non-identi able data is available upon reasonable request to the corresponding author.

Competing interests:
Authors declare no con icts of interest pertaining to the manuscript. MF obtained and analyzed patient data, conceptualized, drafted and revised the manuscript. HD analyzed patient data and drafted the manuscript. JR analyzed patient data. AZ analyzed patient data and revised the manuscript. PC obtained patient data. MR and JV obtained patient data and revised the manuscript. artifacts and their rotational orientation determined by Brainlab algorithm. Volume of tissue activated (VTA) were calculated based on the stimulation parameters and con guration. Extraction and normalization of the VTAs from the patients native space into the common MNI; ICBM 2009b NLIN asymmetric brain space (Fonov et al., 2009(Fonov et al., , 2011 as done by an in-house built toolbox in Matlab environment (in part publicly available at: https://github.com/JonasRoothans/ArenaToolbox), which makes use of the uni ed tissue segmentation (Ashburner & Friston, 2005) in SPM12 (statistical Parametric Mapping, http://www. l.ion.ucl.ac.uk/spm/software/spm12/).

Segmentation of the ipsilateral vestibulothalamic tract (iVTT) and tract from Interstitial Nucleus of Cajal (INC) to Zona incerta.
To reconstruct the ipsilateral vestibulothalamic tract, we segmented the vestibular nucleus in Mango (Multi-Image analysis GUI, http://ric.uthscsa.edu/mango/) on the fractional anisotropy template of the MNI; ICBM 2009b NLIN asymmetric brain space in a similar manner as previously described , while depending on the posterior, medial and lateral anatomical boundaries for segmenting the vestibular nucleus in the shape of a square prism with the long dimension along the vertical axis and otherwise equal sides of 5 mm . We then performed ber-tracking in DSI studio using the HCP1021 template (http://brain.labsolver.org/diffusion-mri-templates/hcp-842-hcp-1021). The template represents the averaged voxel-based spin distribution functions of subjects from the human connectome project allowing for ber reconstruction based on the regions of interest and tracking parameters (Van Essen et al., 2012;Yeh et al., 2010;Yeh & Tseng, 2011) The right vestibular nucleus was set as a seed and the right thalamus from the automated anatomical labelling atlas 3 (Rolls et al., 2020) as a region of interest. We used the following parameters: 100,000 seeds, QA threshold: 0.25, angular threshold of 52 degrees and using Euler's deterministic ber-tracking algorithm with a step size of 0.5 mm. For visualization purposes the INC was segmented on a T1 structural template (MNI;ICBM 2009b NLIN asymmetric) in Mango with aid of relations to neighboring neuroanatomical structures as depicted in Allen human brain atlas (http://www.atlas.brain-map.org) with the help of HCP842 tractography atlas (Yeh et al., 2018) where appropriate. The zona incerta was acquired from supplementary les of published literature (Lau et al., 2020). The INC was used as seed and zona incerta as a region of interest for ber-tracking in DSI studio using the HCP1021 template with the following parameters: 100,000 seeds, QA threshold: 0.20, angular threshold of 52 degrees and using Euler's deterministic ber-tracking algorithm with a step size of 0.5 mm. The tracts and anatomical structures of interest were then visualized in DSI studio superimposed on a high resolution joint template from T1 & T2 7T MRIs of an ex-vivo brain normalized to MNI space (Edlow et al., 2019).