A broad variety of oculomotor alterations have been described in Parkinson's disease (PD) such as an increased latency of visually guided reactive saccades, reduced saccadic gain, impaired smooth pursuit and difficulties to inhibit unwarranted reactions [1]. Existing literature mainly focuses on different aspects of saccadic dysfunction and basal ganglia pathology without arriving at a generally accepted view. The substantia nigra and related brainstem areas are suggested to be crucially involved in various types of saccadic eye movements by mediating a dopamine-related descending input from frontal cortical areas to the superior colliculus [2], but are also thought to play a role in inhibiting inappropriate saccades [3]. To better understand the emergence of oculomotor alterations in PD, the effects of therapy on eye movements (dopaminergic treatment or deep brain stimulation of the subthalamic nucleus, STN-DBS) have been repeatedly examined. Some studies show no effect of dopaminergic medication [4], whereas other studies do show effects on latency, gain or amplitude of saccades (e.g. [5]). In general, dopaminergic treatment is supposed to be of little effect on ocular motor deficits in PD [6], in contrast to its benefit for limb motor function. Regarding the effects of STN-DBS on saccadic function, there exist only a few studies, so far. Most of these report a reduction of the latency of visually elicited reactive saccades (RS) [2] as well as an increase in gain [7, 8], whereas another study found no effect on latency and gain of RS [9] but an improvement of the amplitude of memory guided saccades [8]. Finally, a single case report suggests that saccadic intrusions in patients can be reduced by STN-DBS [10]. Since STN-DBS is generally accepted to be highly effective in reducing the levodopa sensitive parkinsonian motor symptoms [11], these observations suggest that oculomotor deficits in PD might to a considerable part be caused by a dysfunction of non-dopaminergic systems.

There is ample evidence that PD is a multisystem disorder with involvement of several brain regions and neurotransmitter pathways other than dopamine [12]. Hence the cause of numerous clinicopathological correlates of PD-related deficits such as cognitive dysfunction, dementia, depression, and behavioural or emotional dysfunction have to be searched outside the nigrostriatal system. Recently, these neuropsychological and psychiatric changes in PD have received considerable attention. They are supposed to arise either as a part of the disease itself but also as a collateral result of STN-DBS [13]. If non-dopaminergic deficits cause indeed some of the eye movement alterations in PD, it might be possible to demonstrate a correlation between such non-motor symptoms and oculomotor impairment in PD.

Since a number of methodological concerns complicate the interpretation and comparison of studies that examine the impact of dopaminergic therapy (different testing procedures, failure to counterbalance 'on' and 'off' order of testing, different dopaminergic medication and withdrawal regimes [14]), we chose to study PD patients with STN-DBS in the "on" and "off" condition and medication "on", but also PD patients with exclusively oral dopaminergic medication. Patients' eye movement were recorded by video-oculography (VOG) and analysed in relation to their demographic, clinical and neuropsychological data.

In this study we investigated whether an accurate assessment of oculomotor changes in PD patients - subdivided in a group with oral dopaminergic medication and a group of STN-DBS patients in the DBS "on" and "off" condition with additional oral dopaminergic medication - may allow a conclusion as to the pathomechanism of oculomotor dysfunction in PD. With regard to oculomotor function, we observed a reduced SPEM gain (typically caused by the release of predictive saccades, cf. Figure 1 in [18]), increased latency of RS (in PD-DA only) and a decreased number and gain of AVGS in all patient groups in comparison to controls. These results are in general agreement with the literature on oculomotor dysfunction in PD (i.e. [1]). However, none of the investigated oculomotor parameters exhibited significant differences between patients treated with dopamine only (group PD-DA) and those receiving STN-DBS (group PD-DBS-on). In DBS patients, the only effect of stimulation withdrawal was a just significant reduction of N30 whereas all other examined oculomotor parameters did not differ.

STN-DBS has been reported to significantly reduce the latency of visually triggered saccades [2, 7, 8], and to increase their gain [7, 8]. Similar to the present study, Rivaud-Péchoux et al. observed an only non-significant shortening of latency by STN-DBS but a reduction of errors in an antisaccades task. A convincing mechanistic explanation for the observed improvements does not seem to exist, so far. Since STN neurones are thought to excite crossed nigro-collicular cells, it is difficult to understand how a supposedly elevated STN activity in PD could facilitate (instead of prevent) the release of inappropriate saccades by the colliculus observed in many studies [4, 6] or how a reduction of STN overactivity by DBS could lead to the reduction of inappropriate saccades noted in other studies [10]. Sauleau et al. [7] invoke an altered control of the interaction between STN and substantia nigra pars reticulata (SNpr) in advanced PD which would be restored to normal via improved functioning of the parieto-collicular pathway. Temel et al. [19] suggest that the reduction of STN activity by DBS enhances the facilitation of the colliculus by interacting with the cortico-basal-collicular pathway. Yugeta et al. [8] invoke the more recent finding that PD pathology may be linked to the occurrence of abnormal oscillatory β-activity in STN. Similar to skeleto-motor performance [20], desynchronisation of this activity by STN-DBS would also be beneficial for oculomotor behaviour although its differential effect on reactive and unsolicited saccades remains to be explained.

Why do our measurements of saccadic latency not reproduce the beneficial effect of STN-DBS reported by others? With regard to latency and saccadic gain, the majority of our STN-DBS patients appear to have been too close to normal to exhibit any sizeable effect of STN-DBS. Moreover, the tendency for longer saccade latencies in PD as compared to CTL may in part have resulted from their dopaminergic medication that had been received two hours before eye movement recording [5].

We are unaware of studies that would have examined the effect of STN-DBS on the SPEM gain. The lack of a significant SPEM gain amelioration by STN-DBS in our patients suggests that the stimulation does not improve the inhibitory control of unwarranted saccades during pursuit (a main cause of SPEM gain reduction, [18, 21–26]). Experimental evidence for an amelioration of inhibitory control from previous studies by STN-DBS is mixed, at best. STN-DBS was found ineffective at reducing antisaccade errors [8, 9], but reportedly diminishes the frequency of premature saccades during a memory saccade task [8].

As a further paradigm which has not been studied during STN-DBS, we have examined the ability to rapidly saccade back and forth between two fixed targets. This paradigm was the only one to reveal a significant but small improvement under STN-DBS. This result is in line with the notion of the basal ganglia being particularly important for voluntary, endogenous saccades [3] in PD. In summary, our data suggest that STN-DBS can increase the frequency of voluntary saccades but does not improve SPEM gain in PD.

In contrast to the investigated eye movement parameters, the patients' skeletomotor system (UPDRS III scores) clearly improved with STN-DBS. The beneficial effect of STN-DBS on motor function is well established [11]. Nevertheless, if the observed eye movement alterations in PD resulted from a dysfunction of mainly dopamine-mediated mechanisms that can be relieved by STN-DBS, there should exist a correlation between STN-DBS-related improvement in UPDRS III and eye movement performance. We observed no such correlation, in particular when comparing the only eye movement parameter improving with DBS-STN (N30) to a skeletomotor analogue (UPDRS III subscore for rapid alternating hand movements). However, also authors who noted a SPEM improvement with dopamine agonists point out that this effect is much smaller than the concurrent UPDRS III improvement [22].

Also studies addressing the effect of dopamine on eye movements have produced ambiguous results (i.e. [4–6, 23]). Some of these studies warrant methodological concerns such as different medication and withdrawal regimes (for a review, [14]). Moreover, the long half-value period of the ubiquitously applied dopamine agonists makes it difficult to examine patients in a true medication-off state since the medication-free time has to be limited for ethical reasons and because it is unclear whether there is a direct link between the plasma level of medication and possible influences on oculomotor function.

Compared with skeleto-motor function, there may be not an as clear a link between pathologic STN oscillations and eye movement alterations as these oscillations mainly affect dorsal STN and less so the ventral, eye movement related STN [25] Therefore, the effect of STN-DBS on eye movement alterations may be limited. Taken all together, the above considerations are in line with growing evidence that pathologies outside the dopaminergic nigrostriatal pathway may also be responsible for oculomotor deficits in PD (i.e. [26]). In view of the occurrence of frontal lobe pathology in PD [18], an impact of non-dopaminergic alterations in frontal cortical areas on eye movement control in PD is conceivable [27]. In this regard Rieger et al. [28] have found the frontal eye field itself to be hypoactive when performing horizontal voluntary saccades in an fMRI study. This study [28] as well as data showing a statistically significant correlation between the impairment of smooth pursuit and premature saccades in a delayed saccade task in PD [21] strongly support for the assumption of a top-down frontostriatal pathomechanism of saccadic dysfunction in PD..

Available literature offers a mixed picture concerning the effects of therapy on eye movement control in PD, be it dopaminergic medication or high frequency deep brain stimulation. Some studies show no effect of dopaminergic medication [4], whereas others report effects on latency, gain or amplitude of saccades (i.e. [5]). In general, dopaminergic treatment has been found to be of little effect on the ocular motor deficits in PD [6]. STN-DBS has been reported to significantly reduce the latency of visually triggered saccades [2, 7, 8], and to increase their gain [7, 8]. If eye movement alterations in PD resulted from a dysfunction of dopamine-mediated mechanisms exclusively, motor performance in general should correlate with eye movement performance both in terms of disease progression and therapeutical response. The lack of such correlation in our data supports the idea of a mainly non-dopaminergic top-down frontostriatal pathomechanism of oculomotor dysfunction in PD.