Low-intensity vestibular noise stimulation improves postural symptoms in progressive supranuclear palsy

Sixteen patients (age 70.7 ± 7.8 years, 6 females) with a probable PSP according to the MDS-PSP criteria [24] participated in the study (for detailed patient characteristics, see Table 1). [¹⁸F]PI-2620 PET imaging was performed to depict tau-deposits as an in vivo biomarker supporting the clinical diagnosis [8]. Each patient underwent a complete physical, neurological and neuro-otological examination by an expert neurologist, including a clinical assessment of vestibuloocular (head impulse test) and vestibulospinal (Romberg’s test) function (AZ). Symptom severity was scored using the Progressive Supranuclear Palsy Rating Scale (PSPRS) [20], while patients were taking their regular medication (i.e., L-dopa at a mean daily dose of 518 ± 254 mg in 10 patients). Patients had a mild-to-moderate disease severity with a PSPRS of 29.8 ± 11.8. Fourteen age-matched healthy controls (age 68.9 ± 6.5 years, 9 females) were included in the study to establish normative data. All participants gave written informed consent prior to study inclusion.

[¹⁸F]PI-2620 PET imaging was performed in a full dynamic setting (0–60 min post injection) with a Siemens Biograph or mCT PET/CT scanner (Siemens Healthineers, Erlangen, Germany) at the Department of Nuclear Medicine, LMU Munich as described earlier [8, 57]. The multilinear reference tissue model 2 (MRTM2) in PMOD version 3.9 (PMOD Inc) was used to calculate distribution volume ratio images (DVR; DVR = non-displaceable binding potential + 1) of each full dynamic data set. The cerebellum excluding the dentate nucleus and the central cerebellar white matter as well as the superior and the posterior cerebellar layers (thickness in z direction = 1.5 cm each) served as a reference region.

Two expert readers (M.B. and M.Z.) performed a dichotomous visual read of DVR maps. In addition, [¹⁸F]PI-2620 DVR values were obtained in the following PSP target regions predefined by Brainnetome atlas [14]: frontal lobe, thalamus, globus pallidus, and putamen (left- and right-sided, respectively). Patient data were compared to a normative in-house data set derived from a matched healthy control group of the same scanners.

Vestibular noise stimulation (i.e., nGVS) was applied via a pair of 4.0 cm × 6.0 cm Ag–AgCl electrodes attached bilaterally over the left and right mastoid process. Zero-mean Gaussian white noise stimulation with a frequency range of 0–30 Hz and varying peak amplitudes of 0–0.7 mA was delivered by a mobile constant current stimulator (neuroConn^®, Illmenau, Germany).

Body sway was recorded for 30 s on a posturographic force plate (Kistler, 9261A, Kistler Group, Winterthur, Switzerland) at 40 Hz while patients were standing with their eyes closed (Fig. 1A). This procedure was repeated eight times, while patients were stimulated with a different amplitude of nGVS (ranging from 0 to 0.7 mA, in a randomized order) in each trial. Patients were blinded to the exact stimulation order. Between trials, patients were given a short break to recover.

For each stance trial, mean sway velocity was calculated as the primary output measures based on the recorded radial center-of-pressure (CoP) trajectory using the formula \(SV = 1/T \times {\sum }_{i}\left|{r}_{i+1}-{r}_{i}\right|, [{\text{mm}}/{\text{s}}]\), where \(T\) is the total trial duration (i.e., 30 s) and \({r}_{i}\) is the radial CoP distance of the ith sample. For further analysis, sway velocity measures from the eight stance trials were normalized to sway velocity obtained during 0 mA stimulation (i.e., baseline condition).

To determine whether SR-like dynamics were present in the balance responses of patients to varying nGVS levels, we tested three increasingly rigorous criteria that built on one another: (1) The first criterion tested whether body sway of patients improved for at least one particular nGVS level compared to baseline condition (i.e., 0 mA nGVS). (2) The second criterion was based on a visual inspection of response dynamics of body sway across increasing nGVS level by three experienced human raters (i.e., MW, DP, and AZ). Each rater had to evaluate whether (in addition to the fulfillment of the first criterion) nGVS-amplitude-dependent changes of body sway in individual patients were further compatible with a bell-shaped response curve with improvements of performance at intermediate stimulation intensities that is indicative of the presence of SR. This evaluation was based on a plot of the normalized nGVS-dependent changes in body sway and a concomitant plot of a theoretical SR curve that was fit on the data using a goodness-of-fit statistic [2, 18] (Fig. 1B). The applied equation fit represents an adapted version of the originally proposed SR model by Benzi [3], including a piecewise, linear masking effect to model cases where nGVS effects at high amplitudes may have detrimental effects on the performance metric [58]. The criterion was met if at least two of the raters identified the presence of SR-like dynamics. (3) The third criterion additionally evaluated whether improvements at intermediate nGVS levels were greater than the minimal clinically important difference (MCID; defined as half the standard deviation for normative data [60]) for changes in body sway velocity. MCID for sway velocity was 2.3 mm/s calculated based on the posturographic recordings of the 14 age-matched healthy individuals standing with eyes closed for 30 s.