Midbrain lesion-induced disconjugate gaze: a unifying circuit mechanism of ocular alignment?

A 63-year-old man emergently referred himself after waking up with double vision and a sense that the environment was moving, an unsteady gait, dizziness, and a mild headache. He had no history of visual or ocular motor dysfunction and strabismus and reported normal vision in both eyes. Neuroophthalmological assessment (Fig. 1A) showed a rightward head tilt and a mild skew deviation with left eye hypertropia upon cover testing, consistent with a rightward ocular tilt reaction. There was a tonic horizontal misalignment: esodeviation with right eye turning inwards, greater in distance and upgaze, but without associated abduction deficit, consistent with non-paretic esotropia. Limited upgaze could be partly overcome by vertical vestibular-ocular reflex testing, suggesting a supranuclear etiology.

There were no other neurological abnormalities such as light-near-dissociation. Over the course of acute clinical management (5–7 h) including emergency MR-imaging, the initial HSSN pattern evolved into a predominantly upbeating pattern more pronounced in the right eye.

Binocular videooculography ~ 7h after admission confirmed limitation of vertical gaze range and saccades to ~ 5° from midline (Supplementary Fig. 1A).

In darkness, the horizontal channel showed a left beating spontaneous nystagmus with linear slow phases of 1–2.5°/s, superimposed on disconjugate, convergent drifts, more pronounced in the right eye, with a median frequency of 0.2Hz (Fig. 1B, purple box). The binocularly averaged vertical channel showed an upbeat nystagmus with linear slow phases and a median frequency of 0.7Hz (Fig. 1B, grey box I). Vertical and horizontal oscillation frequencies were uncorrelated in darkness.

During fixation in light, the horizontal channel showed convergent–divergent oscillations with mostly pendular waveforms and significantly higher frequency (3.7Hz vs. 0.2Hz, P = 0.03, Fig. 1B, purple box). The vertical oscillations shifted to pendular waveforms and showed a similar increase of oscillation frequency from 0.7 to 3.7Hz (P = 0.03, Fig. 1B, grey box II).

Gaze position affected vertical and horizontal oscillation frequency (gaze direction x nystagmus plane interaction, F(2) = 3.98, P = 0.004) and waveform (Fig. 1B, grey box III& IV). Left gaze position reduced horizontal and vertical oscillation frequency to 2.4Hz (P = 0.002) and 3.2Hz (P = 0.016), respectively. During fixation in light, vertical and horizontal oscillation frequencies were strongly correlated (left eye: rho = 0.67, P < 0.001, right eye: rho = 0.59, P = 0.0041, Fig. 1B).

Three days later, vertical gaze and saccade range was normalized and no ocular oscillations were detectable (Supplementary Fig. 1C).

Diffusion-weighted MRI revealed a solitary, acute ischemic lesion (~ 3.8 × 4.2mm) in the left meso-diencephalic junction (Fig. 1C). Topographical analysis revealed a lesion localization directly adjacent to the posterior commissure (PC), abutting the vertical saccade generator, rostral interstitial nucleus of the medial longitudinal fasciculus (riMLF)⁴ but not the vertical gaze integrator, interstitial nucleus of Cajal (INC) [1, 4, 7] (Fig. 1D–E).

In frontal-eyed animals, disconjugate eye movements serve a primary teleological purpose: precisely aligning objects in 3D space onto each eye’s fovea, producing the needed vertical and horizontal vergence [4] and thereby enabling depth perception. Disconjugate eye movements also support fixation and reduction of motion parallax during head movements in 3D space, via the translational vestibulo-ocular reflex (TVOR) [14].

Distinct midbrain regions have been associated with disconjugate horizontal [4‐6] or vertical [1, 2, 7, 8] eye movements. The dorsal midbrain has been implicated both in hemi-seesaw and convergence–retraction nystagmus [5, 6, 8]. Simultaneous occurrence however is rare and has only been reported with extensive subcortical lesions [8], limiting precise neuroanatomical conclusions. Here, MRI revealed a solitary focal midbrain lesion adjacent to the posterior commissure, hinting at a circumscribed neuroanatomical origin for both oscillations. The lesion abuts the riMLF, consistent with the vertical saccade deficit [4], but not the INC. The area surrounding the INC, but not the nucleus itself, is critical for disconjugate vertical nystagmus [7, 8]. Experimental lesions directly lateral to the posterior commissure cause disconjugate, 2–4 Hz seesaw and convergence–divergence oscillations which are modulated by vestibular and visual inputs [3, 7]. These findings closely correspond to our patient’s lesion location and mirror the observed disconjugate 3.7Hz oscillations. Moreover, vertically acting eye muscles relax during convergence, presumably via inhibitory INC projections crossing through the PC [4, 15]. Thus, this pathway’s lesion-induced disconnection may induce globe retraction via inappropriate eye muscle co-contraction [4, 5, 13].

Ocular and postural control depends on the fusion of visual and vestibular signals into a congruent percept of self in space [4, 16]. Defective visuo-vestibular integration results in fixation-dependent oscillations like (hemi-)seesaw and infantile nystagmus [8, 9], but has not yet been linked to CRN. Our data suggest a close relationship of HSSN and CRN and therefore suggest a common substrate for vertically and horizontally disconjugate eye movements. The shared effects of fixation and illumination imply retinoreceptive neural circuits of the midbrain pretectum and accessory optic system (AOS) [16].

The AOS is a primordial visual system which has been implicated in developmental strabismus and nystagmus [9‐11, 16]. Emerging theories propose a resurgence of AOS-related ocular motor patterns, which link disconjugate eye movements to atavistic balance reflexes in lateral-eyed species [9, 10, 16]. These encompass stereotypical patterns of esodeviation, vertical divergence and disconjugate fixation nystagmus [9, 10, 16], closely mimicked by this case. Additionally, binocular excyclotorsion co-occuring with convergent oscillations suggests inappropriate inferior oblique activation, yet another AOS-related phenomenon in developmental strabismus [16].

Densely interconnected via the posterior commissure, the pretectum and AOS also receives visual-related cortical inputs [17] and relays visual climbing fiber input relevant for vestibular-ocular reflex (VOR) tuning to the cerebellum via the inferior olive [9, 10]. A strategic lesion to AOS-related pretectal circuits deprives the cerebellum of visual-related inputs to match to vestibular signals—a nystagmogenic mechanism also proposed for oculopalatal tremor and developmental strabismus [4, 16].

The interplay between vision, translational VOR and vergence is paramount for depth perception during 3D head motion [4, 14]. Similar to a vestibular roll plane imbalance causing vertical skew deviation [4, 14], an imbalance in the translational fore-aft plane may cause esodeviation, which has been formalized as the “horizontal skew deviation” of developmental strabismus [16]. Skew deviations are associated with translational VOR deficits[14] and frequently accompany esodeviation and CRN [5, 6], like in our case. Finally, both oscillations’ striking gaze-position dependence suggests a vestibular coordinate framework, because vestibular nystagmus increases when gaze axis and culprit semicircular canal planes align [4]. This is mirrored by abnormal central vestibular findings in subjects with developmental strabismus [18].

Rooted in Occam’s razor, we argue that the observed parallels to the hallmarks of developmental strabismus—namely esodeviation with increase in upgaze, vertical divergence with inappropriate inferior oblique activation and fixation-dependent, disconjugate nystagmus—are not mere coincidence. Instead, we propose that this stereotypical pattern echoes a shared visuo-vestibular circuit mechanism with an essential hub region in the paracommissural midbrain. This fits with contemporary experimental results highlighting midbrain–cerebellar circuits as key drivers of developmental strabismus patterns [19]. Expanded circuit-level studies may further illuminate shared pathomechanisms and thereby potentially open new avenues for translational and therapeutic research.

Based on a single case report, generalizability is limited. We aimed to mitigate this by incorporating longitudinal assessments and employing granular phenotyping techniques facilitating comparisons to experimental studies. Due to technical constraints, vertical videooculographic recordings were only available in a binocularly averaged format, unable to fully capture the short-lived vertically disconjugate oscillations observed in the initial clinical presentation. However, we supply a high-resolution clinical video illustrating the clinical findings.