3.1. Subjects characteristics
Detailed group characteristics are summarized in Table 2. ET and PD - groups did not differ in terms of age or time since surgery. Mean age was 70.8±10.4 years (PD patients) and 64.4±9.9 years (ET patients, p=0.28), mean duration of DBS treatment was 174.8±102.7 days (PD patients) and 162.6±52.6 days (ET patients, p=0.79). ET patients showed a significant longer disease duration (PD: 6.8±3.1 years; ET: 33.6±18.7 years; p=0.002). The position of the electrodes did not differ significantly between PD and ET patients. Given the border between VIM and PSA at the level of the midcommissural line, all but one contact investigated were located in the PSA. The stimulation amplitude to induce ataxia tended to be higher in PD patients but the difference was not significant (PD 3.58±0.68 mA. ET 3.19±0.53 mA, p=0.06).
3.2Perioperative tremor scores in PD and ET patients
The course of pre- and postoperative tremor scores in the observation period of about 6 months differed between both patient groups (Fig. 1). In PD patients, there was a slight, but not significant dopaminergic tremor response in a preoperative L-Dopa challenge using suprathreshold morning dosage (UPDRS-tremor score: preoperatively MedOFF 10.4±2.2, MedON 8.6±2.4). The pre- and postoperative STIM OFF tremor scores at the time of the experiment were comparable (UPDRS III tremor score: pre-DBS Med OFF 10.4±2.2, post-DBS STIM OFF 10.14±1.35, p=0.10) (Fig. 1A) indicating stable tremor characteristics in the off condition in the observation period. In contrast, tremor scores of ET patients significantly worsened in the longitudinal time course with significant perioperative increase of tremor severity (FTMRS sum score: pre-DBS 12.79±4.49, post-DBS 15.79±3.58 (p=0.005; Fig. 1B). Noteworthy, the interval between pre- and postoperative assessments did not differ in PD and ET patients (Table 2).
3.3. Conventional and new stimulation modes of VIM/PSA-DBS in PD patients
In PD patients, new stimulation forms as short pulse stimulation and directional stimulation were compared to conventional omnidirectional 60µs VIM/PSA-DBS at two different contact levels, the most ventral ring contact and the adjacent dorsal, segmented contact (Fig. 2A).
The use of short pulse stimulation was equivalent in suppressing tremor compared to conventional VIM/PSA-DBS, but with less side effects at the ventral contact. GLM ANOVA with the factor stimulation mode (3 levels: 1. STIM OFF 2. oDBS60 3. oDBS30) revealed significant effects for the total tremor score (F=36.66, p=0.001) and ataxia (F=11.93, p=0.003). Paresthesia were not significantly different during the different STIM conditions (F=3.08, p=0.11).
The total tremor score was significantly improved by both oDBS60 (2.14±0.9; p=0.001) and oDBS30 (1.71±0.76, p=0.001) compared to OFF (12.86±4.6), there was no significant difference between both stimulation modes indicating comparable tremor improvement by both stimulation modes. Interestingly, the effect of DBS was divergent on the different tremor subtypes as tremor at rest, postural tremor, intention tremor and tremor while drawing Archimedes spiral (Fig 3A). A two factorial GLM ANOVA with the intrasubject factor 1. STIM condition and 2. tremor subtype revealed a highly significant effect for STIM (F=84.46, p<0.001) and tremor subtype (F=22.97, p<0.001) with significant interaction (F=12.63, p=0.001). Both stimulation modes improved significantly tremor at rest (oDBS60 and oDBS30 p<0.001), postural tremor (oDBS60 and oDBS30 p<0.001) and slightly tremor while drawing spirals (oDBS60 p=0.022, oDBS30 p=0.021) compared to STIM OFF, but not sufficiently intention tremor (oDBS60 and oDBS30 p=0.078). Thus, both stimulation modes improved tremor compared to STIM OFF, however with a restricted effect on intentional tremor – this pattern of specific tremor subtype suppression was similar between conventional or short-pulse stimulation.
Both stimulation modes differed in terms of side effects. The use of conventional oDBS60 (3.0±1.0) induced significant ataxia, which was not present in STIM OFF. Stimulation-induced ataxia was significantly less accentuated while oDBS30 (1.71±0.76) compared to oDBS60 (p=0.021) indicating a better side effect profile using short pulses (Fig 2B).
The use of directional stimulation revealed a similar pattern when comparing omnidirectional and directional stimulation at the adjacent dorsal contact using the same pulse width of 60µs. The total tremor score was significantly reduced (GLM ANOVA factor STIM condition F=29.41, p=0.003) by both stimulation forms (oDBS60 3.0±0.7, p=0.001; dDBS60 1.34±0.6, p=0.006) compared to STIM OFF (12.86±4.6) with no significant differences between both stimulation modes. Again, there was a tremor subtype specific effect of stimulation (GLM ANOVA factor STIM condition F=66.99, p=0.001, tremor subtype F=14.27, p=0.01) and the interaction (F=7.76, p=0.012; Fig 3A). The pattern of tremor improvement was similar between both stimulation modes. Directional stimulation at the best segment reduced tremor at rest (p=0.001), postural tremor (p=0.004) but there was no significant improvement of intentional or drawing tremor.
In terms of the side effect ataxia (GLM ANOVA STIM F=22.34, p=0.007), directional stimulation at the best segment (dDBS60 2.40±0.55) tended to induce less ataxia than omnidirectional DBS (oDBS60: 3.40±1.14), but the difference was not significant between both stimulation modes.
In summary, VIM/PSA-DBS, regardless of the stimulation mode, improved resting and postural tremor, but not intention tremor and slightly less drawing tremor in PD patients. Short-pulse and directional DBS were comparably efficient in the improvement of tremor compared to conventional DBS, but with less side effects as ataxia, particularly with short-pulse stimulation.
3.4. Comparison of VIM/PSA-DBS effects between PD and ET patients
3.4.1 Comparison of DBS effects with short pulse stimulation
The total FMTRS was significantly reduced by VIM/PSA-DBS at the ventral contact in both patient groups regardless of the pulse width. GLM ANOVA with the intrasubject factor stimulation condition (3 levels: 1. STIM OFF 2. STIM ON 60µs 3. STIM ON 30µs) and the intersubject factor disease condition (2 levels: 1. PD 2. ET) revealed a highly significant difference for the factor stimulation condition (F=113.31, p< 0.001), but no significant interaction between stimulation condition and disease (F=0.17, p=0.76). Comparable to PD patients, the total tremor score was significantly reduced in ET patients in both ON-condition (OFF: 15.79±3.58, oDBS60: 6.00±3.16, oDBS30 5.0±3.01, both p<0.0001) with no significant differences between STIM ON conditions.
However, the DBS effects were diverging in respect of disease condition and the effect on the different tremor entities as tremor at rest, postural tremor, intention tremor and tremor while drawing Archimedes spiral (Fig 3). A two-factorial GLM ANOVA with the intrasubject factor STIM condition and tremor subtype revealed a highly significant effect again for STIM (F=164.04, p<0.001) and tremor subtype (F=41.44, p<0.001), with significant interaction of tremor subtype, disease condition (F=34.05, p<0.001) and STIM condition (F=27.77, p<0.001) indicating a disease specific action of short pulse and conventional DBS on the different tremor subtypes. This finding could be replicated testing short pulse and conventional DBS at the more dorsal contact (STIM (F=67.75, p<0.001), revealing significant effects for tremor subtype (F=45.7, p<0.001) with significant interaction with disease condition (F=20.16, p<0.001) and with both, disease and STIM condition (F=13.97, p<0.001). In detail, the predominant tremor at rest and postural tremor in PD patients was completely suppressed by VIM/PSA-DBS, whereas intentional and action tremor, particularly drawing spirals, was less impacted by VIM/PSA-DBS, which was more prominent in ET patients.
When normalising the total tremor scores in STIM ON in relation to the STIM OFF condition with lower percentages indicating better tremor reduction, we observed differential effects of DBS in both patients groups. The relative tremor reduction was more pronounced in PD patients for both stimulation modes compared to ET. This disease-specific effect was significant for oDBS60 and oDBS30 (oDBS60: PD 19.66%±7.2, ET 40.46%±22.5, p=0.0003; oDBS30: PD 15.86%±10.63, ET 31.18%±22.64, p=0.0024 (Fig 4A).
Suprathreshold stimulation with 60µs at the ventral contact triggered ataxia as measured by the ICARS sum score (PD: OFF: 0.38±0.37, oDBS60: 2.86±1.07 p<0.0001; ET: OFF: 0.14±0.36, oDBS60: 2.86±1.35 p<0.0001). GLM ANOVA with the intrasubject factor stimulation condition (3 levels: 1. STIM OFF 2. STIM ON 60µs 3. STIM ON 30µs) and the intersubject factor disease condition (2 levels: 1. PD 2. ET) revealed a highly significant effect for the factor stimulation condition (F=36.86, p< 0.001), but no significant interaction between stimulation condition and disease (F=1.37, p=0.27). Post-hoc analyses revealed that ataxia induced by conventional 60µs DBS could be significantly reduced by usage of short-pulse stimulation with 30 µs in both patient groups (PD: 1.43±0.78 p=0.0001, ET: 2.29±0.64 p=0.0001) (Fig 1B). The same could be observed for stimulation effects at the second most dorsal contact (stimulation condition (F=33.58, p< 0.001), no significant interaction with disease condition (F=0.62, p=0.52).
Again, we compared disease conditions more in detail by calculating the relative difference between the ICARS sum score with oDBS60 and the new stimulation condition (DeltaICARS), with larger values indicating less ataxia with the new stimulation condition. The use of oDBS30 resulted in better ataxia reduction in PD patients compared to ET patients (PD: 1.83±1.11, ET 0.92±1.32 p=0.040; Fig 4B).
Paresthesia were assessed in both patient groups while application of the different stimulation modes. GLM ANOVA revealed a significant effect for the STIM condition at the ventral (F=7.99, p=0.006) and dorsal contact (F=8.53, p=0.003), but no significant interaction with disease condition indicating similar effects in both patient groups
3.4.2. Comparison of omnidirectional and directional stimulation in PD and ET patients
To evaluate and compare the effect of omnidirectional and segmented VIM/PSA-DBS in both patients groups, the stimulation effect with 60µs in the ring mode and at the best segment was compared for the total tremor score, the tremor subtypes and the potential side effects ataxia and paresthesia.
The total tremor score was improved by both stimulation modes compared to STIM OFF condition in both patient groups. GLM ANOVA revealed a significant effect for stimulation condition (F=60.442, p<0.001), but no significant interaction with disease condition. The different tremor subtypes were differentially impacted by DBS. There was a significant effect for the factor tremor subtype (F=39.71, p<0.001) with significant interaction with disease condition (F=19.9, p<0.001) as well as with stimulation mode and disease condition (F=14.25, p<0.001) revealing tremor-specific action of the different stimulation modes in both disease entities. Again, the predominant resting tremor in PD patients was improved by omni- and unidirectional stimulation. In ET patients, the postural and intention tremor was improved, but the predominant drawing tremor was less impacted by both stimulation modes.
Comparing relative changes of DBS compared to STIM OFF with lower values indicating better symptom reduction, the relative reduction of tremor was comparable in PD patients compared to ET (dDBS60: PD 19.42%±12.26, ET 38.44%±31.29, p=0.12; Fig 4A). In terms of the reduction of stimulation induced ataxia, the use of dDBS60 showed no significant difference between patient groups (DeltaICARS: dDBS60: PD 0.80±1.5, ET 1.86±0.86, p=0.11; Fig 4B).