Short-afferent inhibition and cognitive impairment in Parkinson's disease: A quantitative review and challenges

Depósito de investigación de la Universidad de Sevilla https://idus.us.es/ Esta es la versión aceptada del artículo publicado en: This is an accepted manuscript of a paper published in: Neuroscience Letters (2020): 6 February DOI: 10.1016/j.neulet.2018.06.048 Copyright: © 2018 Published by Elsevier B.V. El acceso a la versión publicada del artículo puede requerir la suscripción de la revista. Access to the published version may require subscription. “This is an Accepted Manuscript of an article published by Elsevier in Neuroscience Letters on February 6, 2020, available at: https://doi.org10.1016/j.neulet.2018.06.048.” 1 Short-afferent inhibition and cognitive impairment in Parkinson's disease: A Quantitative Review and Challenges Juan Francisco Martin-Rodriguez1, & Pablo Mir1,2 1Unidad de Trastornos del Movimiento, Servicio de Neurología y Neurofisiología Clínica, Instituto de Biomedicina de Sevilla, Hospital Universitario Virgen del Rocío/CSIC/Universidad de Sevilla, Seville, Spain. 2Centro de Investigación Biomédica en Red sobre Enfermedades Neurodegenerativas (CIBERNED), Madrid, Spain. Corresponding Author: Pablo Mir, MD, Ph.D., Unidad de Trastornos del Movimiento, Instituto de Biomedicina de Sevilla (IBiS), Hospital Universitario Virgen del Rocío, Avda. Manuel Siurot s/n, 41013 Seville, Spain. Tel.: +34-955923039; Fax: +34-955923101.; E-mail: [email protected] Abstract Traditionally, Parkinson’s disease (PD) has been considered a single neurotransmitter (dopaminergic) disease. However, research over the past 20 years has shed light on the involvement of multiple neurotransmission systems, in particular, the cholinergic system. Research has mainly focused on the role of this system in the pathophysiology of PD and its implications in the development of motor and non-motor disorders. Short-latency sensory afferent inhibition (SAI), investigates sensori-motor integration, and has emerged as a putative neurophysiological marker of cholinergic function in the human brain. In this quantitative review, a moderate-tosevere reduction in SAI was observed in PD patients. Furthermore, through moderator analysis, the impairment of SAI was shown to be associated with disease duration and therapeutic state. Patients under dopaminergic agents (“on” state) displayed worse SAI than those after dopaminergic agent withdrawal (“off”). We further assess the potential value of SAI as a marker of cognitive impairment in PD, and its association with four specific cognitive domains. This analysis revealed that patients with cognitive impairment displayed significantly lower levels of SAI than those without cognitive impairment. To conclude, a set of challenges to be addressed before SAI can be validated as a useful clinical tool in PD are presented. Keywords: Short afferent inhibition; Parkinson’s disease; Dopaminergic therapy; Cognitive impairment; Meta-analysis. 2 1. Introduction Parkinson’s Disease (PD) is the second most prevalent neurodegenerative, following Alzheimer’s Disease (AD). While PD is primarily a movement disorder, cognitive deficits often emerge as it progresses [53]. The neurobiological mechanisms of PDassociated cognitive impairments are not fully understood, however, cholinergic dysfunction in PD has been identified as a major predetermining factor for these complications [5, 53], a neurodegenerative feature that PD and AD share [33]. As in AD, the continual association of reduced BF and decreased acetylcholine binding indicates that cholinergic dysfunction seems to lead to decreased cognitive performance and impaired cognition [5, 7, 14, 37]. Cognitive performance in PD patients worsens with the use of anticholinergic drugs, however, is improved by cholinesterase inhibitors [19,44,55]. Cholinergic dysfunction may be responsible for the transition from normal PD to PD with dementia (PD-D); reductions in cholinergic markers are consistently more severe in PD-D than in PD [14, 25, 29, 31, 51]. Postmortem observations in cases of PD without dementia suggest that the neuronal loss from cholinergic BF may appear early in the disease [9, 11, 30, 44]. Opposing to the observed pattern in AD, these results suggest that cholinergic cell loss precedes fibre degeneration in PD. In fact, reduced BF volume in patients with PD and mild cognitive impairment (MCI) has been shown to predict subsequent progression to PD-D [31]. Identifying biomarkers for the early detection of cholinergic dysfunction has become a promising field of research over the last decade, inasmuch as it would enhance diagnostic accuracy, identify early-stage disease patients who may benefit from disease modifying therapies, and monitor disease progression along with the effects of treatment [13]. Furthermore, it is necessary to include proxy measures of cholinergic system integrity in order to gain physiological or clinical markers for the identification of patient sub-groups who experience greater cholinergic denervation and transmission. Short-latency afferent inhibition (SAI) is a measure of central cholinergic activity [16] and is induced through the conditioning of a cortical magnetic stimulus by electrically stimulating contralateral peripheral hand nerves. This is done at an inter-stimulus interval of ~20ms [56]. It has been suggested that SAI depends on neural interactions within the cerebral cortex [56] and is considered to be a measure of sensorimotor interaction [4, 47, 52]; the stimulation of peripheral nerves would 3 activate the cholinergic inputs that, in turn, would modulate the excitability of cortical interneurons [21]. However, the exact mechanisms underlying this type of inhibition have not been fully elucidated. The main supporting evidence that implies SAI results from cholinergic transmission has derived from its suppression through use of scopolamine, a muscarinic cholinergic antagonist, in healthy young-adults [16]. SAI is suppressed in AD and can be improved through treatment with acetylcholinesterase inhibitors [18]. SAI remains normal in frontotemporal dementia [19]. Nevertheless, other neurotransmitters may also be involved in SAI. For instance, the benzodiazepine Lorazepam can reduce SAI through GABA-A receptor activation [17]. The specific relationship between SAI and the GABA-B receptor system remains elusive, although interactions with long-interval intracortical inhibition have been observed [57]. Numerous studies have exploited the strong cholinergic involvement in SAI to explain the pathophysiological mechanisms of motor and non-motor symptoms in PD. Many patients with gait disturbances, visual hallucinations, REM-sleep behaviour disorder, dysphagia, and olfactory impairment (for a recent review see [41]), have displayed SAI abnormalities. Interestingly, some of these alterations are closely related to the presence of cognitive impairment and dementia in PD [12, 32, 46]. The matter of whether SAI is a specific biomarker of cholinergic dysfunction in PD is still a matter of debate [13]. The reported specificity associated with SAI for the detection of symptoms related to cholinergic dysfunction is suboptimal, something that could be extended to all TMS-derived measures [10]. Additionally, the deficit of SAI in PD is not consistently reported in the literature. 1.1. Purpose of and justification for the present review The study of cholinergic deficits is emerging as a key research area for our understanding of the pathophysiological mechanisms of certain motor and non-motor deficits in PD. The measurement of SAI is frequently used as a non-invasive measure of cortical cholinergic dysfunction in PD. However, the study of SAI in this population still yield controversial results. Some studies suggest that this controversy may be mediated by certain variables associated with PE, which modulate or may alter this type of measure. Using a meta-analytical approach, the present study attempts to quantify the degree of uncertainty associated with the measurement of SAI in PD. This analysis also intends to quantitatively study the impact of certain 4 clinical variables on the measurement of SAI. We estimate that this approach may shed light on the clinical utility of the measurement of SAI, as well as increase our understanding of the neurophysiological mechanisms related to cholinergic deficit in PD. A secondary goal of this quantitative review is to highlight the range of studies using SAI as a marker of cognitive impairment in PD, thus reinforcing the utility of SAI to study non-motor deficits of PD. To conclude, a set of challenges to be addressed before SAI can be validated as a useful clinical tool in PD are presented. 2. Methods We conducted a Pubmed and PsyINFO search to identify all studies that compared SAI between PD patients and healthy controls. We used the keywords “short afferent inhibition”, “SAI” and “Parkinson”, in the following search string: (Short afferent inhibition [tiab] OR SAI [tiab]) [AND] Parkinson*. Hedges’ weighted effect sizes (g) and other relevant data were extracted from the studies. Further, we contacted the corresponding authors of studies to obtain additional data. We performed separated meta-analyses comparing 1) PD patients vs. healthy controls 2) patients with cognitive impairment vs patients with preserved cognitive functions. Meta-analyses were performed using the random effects model, calculating both Q-statistics and I2 as indicators of heterogeneity. Further, Subgroup analyses and mixed-effect singlecovariate meta-regression were conducted to identify the association between clinical and methodological study-level covariates and SAI measures in PD. More detailed information about this search strategy can be found in the Supplementary methods. 3. Results and discussion 3.1. SAI dysfunction in PD patients The average effects and heterogeneity of SAI in PD, as compared with healthy subjects across the literature, is visually displayed in Fig. 1. Overall, PD patients showed a moderate-to-large decrease in SAI, but this effect was highly variable between studies (above 85%, [26]). Results from publication bias analysis conducted with the Egger’s test (z = -0.42; P = 0.66) along with results from the trim and fill approach converged in supporting that this data accurately represents the existent literature regarding SAI in PD. The fail-safe N calculations indicated that 485 studies averaging null results would be necessary to bring P to a non-significant value. 5 According to Rosenthal [48] this fail-safe number would be resistant to sampling bias. In the next section, we address heterogeneity found in the SAI effect size. 3.1. Moderator analysis To further explain heterogeneity of findings, moderator analyses with 6 continuous moderators (i.e. age, percentage of males in the PD sample, disease duration, MiniMental State Examination score – MMSE –, severity of motor symptoms as assessed by the Unified Parkinson’s Disease Rating Scale – UPDRS – in both on and off medication states, and finally, the total levodopa equivalent daily dose – LEDD –) were conducted. These were assessed using meta-regression. Moderator analyses of categorical variables such as the shape of the coil (figure-of-eight vs. others), TMS manufacturer (Magstim vs. MagPro), ipsior contralateral application of the SAI protocol, and whether the SAI assessment was performed in OFF or ON (dopaminergic therapy) 1 were also conducted. These analyses yielded negative results for the variables age (p = 0.15), percentage of males (p = 0.86), UPDRSmotor score on (p = 0.44), off (p = 0.20), and total LEDD (p = 0.92). These results suggest that SAI does not reflect the severity of motor impairment or dopaminergic therapy dosage in PD. On the other hand, we have not been able to confirm that the same conversion factors were used in all studies, so the latter assumption should be considered carefully. Unfortunately, the dosages of Levodopa and dopamine agonists (mg/d) administered were not consistently available in reviewed studies, so meta-regression analysis could not be performed for these variables. Meta-regression also revealed that disease severity (B = -0.11, p = 0.005; N = 20 studies) had significant negative impact on SAI (fig. 2A), thus suggesting that activity of inhibitory cholinergic systems deteriorates as the disease progresses. This significant correlation has not been noted by previous studies [49], which may be due to the fact that some correlational studies failed to exclude patients with long-term disease duration (≥ 10 years). Five different effect sizes in patients with a disease duration of more than 10 years were identified in our meta-analysis, thereby enabling the study of SAI associated with long-term PD. Moreover, the MMSE score was found to have a significant impact (B = 0.23, p = 0.039; N = 9 studies) on the 1 Other variables such as the method of choice of the interestimular intervals between the electric stimulus and the magnetic pulse (based on N20 or setting fixed timing), stimulated nerve (median nerve stimulation vs others), muscle recorded for the measurement of the evoked motor potential (first dorsal interosseous, abductor pollicis brevis or abductor digiti minimi), the use of (audio, visual or audio-visual) feedback to assist in maintaining complete relaxation, or intensity of the electrical stimulus were also collected. Nevertheless, the impact of these variables could not be analysed because the protocol used for SAI in most of the articles included in the meta-analysis presented few variations with respect to the original protocol of Tokimura et al. (2000) 6 obtained effect size for SAI (fig. 2B). However, all but one of the effect sizes found were obtained in patients with MMSE scores above 26. Only Celebi et al. [11] reported effect sizes in a sample of PD patients with mild to moderate dementia, indicating a significant negative correlation between MMSE and SAI. Following the exclusion of this study from out meta-regression, the effect size turned out to be nonsignificant (p=0.31). SAI was moderately impaired when testing both the more and less affected hemisphere (more affected side Hedges’ g = -0.75, p < 0.001, N = 17 studies; less affected g = -0.45, p = 0.025, N = 6 studies). There was no significant difference between the effect sizes found when testing SAI in the hemisphere contralateral/ipsilateral to the more affected side (omnibus test, QM = 4.26, p = 0.12; fig. 2C). A separate meta-analysis was carried out to assess differences in SAI between healthy controls and PD patients on and off medication. Normal SAI was found in patients off medication (g = -0.06, p = 0.935, N = 10 studies), whereas a significantly large impairment in SAI was associated with patients in the ‘on’ state (g = -0.95, p < 0.001, N = 17 studies). A significant difference in the effect size between ‘on’ and ‘off’ samples (QM = 12.03, p < 0.001) was also observed, as shown in fig. 2D. This effect of medication on SAI was determined to be robust following introduction of the affected side variable (B = -0.99, p = 0.035). Additionally, introducing the interaction between medication state and affected side did not significantly improve the performance of the model (likelihood-ratio test = 1.47; p = 0.22). Finally, neither the TMS manufactures (QM = 0.02, p = 0.886), nor the shape of the coil used (QM = 0.46, p = 0.499) significantly moderated SAI effect sizes. The previously documented interaction between the hemisphere most affected and dopaminergic treatment has been interpreted as being a consequence of advanced disease state [49]. However, our meta-analysis could not support this statement as the aforementioned factors did not show interdependence in explaining the impairment of SAI in PD. Nevertheless, this should be taken with caution in light of the relatively few studies conducted on the less affected side. Another posed hypothesis suggests that SAI is pathologically increased in PD and normal levels are restored through administration of levodopa [15]. This implies increased cholinergic muscarinic activity within the cerebral cortex contralateral to the most affected side. Again, our moderator analysis could not support this hypothesis considering SAI was found to be significantly impaired when testing in both the more and less affected hemispheres. A striking question which arose from our moderator analysis took into 7 account the fact that patients in ‘on’ medication states showed significantly impaired SAI, while this was found to be normal in the ‘off’ medication state. Previous research has suggested that dopaminergic medication could lead to decreases in central processing or integration of sensory signals in PD patients. For instance, it has been shown that dopaminergic medication could worsen proprioception and postural control [8, 42]. In a study combining peripheral electrical tactile stimulation and TMS, Palomar et al. [43] showed that PD ‘on’ patients did not exhibit the expected modulation of perception of sensory stimuli elicited by paired pulse TMS in the primary somatosensory cortex. Conversely, the expected modulatory effect was normal in PD patients in the ‘off’ medication state. These findings point towards altered somatosensory processing in PD patients on dopaminergic medication. Considering SAI to be a measure of sensorimotor integration, it is tempting to speculate that levodopa treatment interferes with this process by altering intra-cortical mechanisms of sensory input processing in the somatosensory cortex. 3.2. Short afferent inhibition as a marker of cognitive impairment in patients with PD A literature search identified 8 studies to provide quantitative or qualitative data regarding cognitive domains – namely, attention, executive functions, memory, and visuospatial/visuoperceptual abilities. From these studies, 13 independent effect sizes of SAI were extracted and used to compare patients with cognitive impairment and normal cognition. SAI was severely impaired in patients with cognitive impairment (g = -1.66; p < 0.0001), as opposed to the small effect size observed in patients with normal cognitive functions (g = -0.15; p = 0.676). A significant difference in effect size between the cognitively impaired and normal cognition group was observed (QM = 17.27, p < 0.0002; fig. 3A). This effect remained statistically significant after correcting by disease duration (B = -1.91; p < 0.0001). However, the ‘cognitive impairment’ variable could not be corrected by the dopaminergic state or the most affected side due to the lack of effects for this comparison in patients in ‘on’ or ‘off’ states. The association between altered SAI and the presence of specific cognitive deficits is illustrated in fig. 3B. Deficits in the 4 domains analyzed were shown to be associated with significant changes in SAI, with considerable overlapping of effect sizes; visuospatial deficits (median Fisher’s z = -2.84) followed by deficits related with 8 executive functioning (z = -2.67) were associated with a more severe alteration of SAI. It is believed that evaluating SAI mainly provides information about cortical cholinergic tone, being that the cholinergic projections of the BF are major contributors to this tone [36]. Its distribution creates widespread innervation in the cortex, and it is therefore unsurprising that varied cognitive functions are affected by a decrease in cortical cholinergic tone. Our data agree with this observation, as alterations in SAI were associated with the degree of deficit found in all tested cognitive domains. Despite this, two cognitive domains (visuospatial/visuoperceptive abilities and executive functions) showed a stronger association with the degree of SAI impairment, suggesting that these domains are more sensitive to cholinergic dysfunction. Indeed, Bohnen et al. [6] reported poor performance in tasks assessing frontal lobe function and visuospatial ability, likely associated with the cortical cholinergic denervation found in patients with PD and dementia. More severe deficits in visuospatial ability and executive function have also been linked to the presence of visual hallucinations [2] and freezing of gait [46], which are both associated with cholinergic dysfunction [36, 41, 45]. Interestingly, it has been suggested that both frontal lobe and visuospatial dysfunction could be prognostic factors of subcortical dementia in PD [1, 50]. However, these results should be taken with caution; effect sizes assessing attention in PD patients were underrepresented in our analysis in comparison with the other domains. According to recent studies, a closer relationship between attentional deficits and the capacity for sensorimotor inhibition should be expected considering the role that the cortical cholinergic system plays in top-down and bottom-up attentional control processes [2, 28]. Therefore, our analysis could be underestimating the relationship between SAI and deficits in attentional functions. 4. Weakness and limitations A number of limitations in the present review should be acknowledged. To address our research questions, we attempted to identify those studies that exclusively used SAI to assess PD patients. However, we cannot know whether the studies reviewed include data from other clinical entities showing high degrees of overlap with PD, such as dementia with Lewy bodies (DLB). Reduced cholinergic function is also seen in DLB and is correlated with cognitive decline [55]. Regrettably, the studies included in the present meta-analysis are lacking in essential information, including the clinical course or presence of associated/supporting clinical features, helping to evaluate the impact that the inclusion of misdiagnosed cases could have in our analysis. However, we believe that the occurrence of data from patients with DBL minimally impacted our results, considering only one of the studies [11] included in the meta- 15 [45] S. Perez-Lloret, F.J. Barrantes, Deficits in cholinergic neurotransmission and their clinical correlates in Parkinson's disease, NPJ Parkinson's disease 2 (2016) 16001. [46] M. Picillo, R. Dubbioso, R. Iodice, A. Iavarone, C. Pisciotta, E. Spina, L. Santoro, P. Barone, M. Amboni, F. Manganelli, Short-latency afferent inhibition in patients with Parkinson's disease and freezing of gait, Journal of neural transmission 122 (2015) 1533-1540. [47] T. Raij, J. Karhu, D. Kicic, P. Lioumis, P. Julkunen, F.H. Lin, J. Ahveninen, R.J. Ilmoniemi, J.P. Makela, M. Hamalainen, B.R. Rosen, J.W. Belliveau, Parallel input makes the brain run faster, NeuroImage 40 (2008) 1792-1797. [48] R. Rosenthal, The file drawer problem and tolerance for null results, Psychological bulletin 86 (1979) 638. [49] A. Sailer, G.F. Molnar, G. Paradiso, C.A. Gunraj, A.E. Lang, R. Chen, Short and long latency afferent inhibition in Parkinson's disease, Brain : a journal of neurology 126 (2003) 1883-1894. [50] G. Santangelo, L. Trojano, C. Vitale, M. Ianniciello, M. Amboni, D. Grossi, P. Barone, A neuropsychological longitudinal study in Parkinson's patients with and without hallucinations, Movement disorders : official journal of the Movement Disorder Society 22 (2007) 2418-2425. [51] J. Shin, S. Choi, J.E. Lee, H.S. Lee, Y.H. Sohn, P.H. Lee, Subcortical white matter hyperintensities within the cholinergic pathways of Parkinson's disease patients according to cognitive status, Journal of neurology, neurosurgery, and psychiatry 83 (2012) 315-321. [52] L. Spieser, H.B. Meziane, M. Bonnard, Cortical mechanisms underlying stretch reflex adaptation to intention: a combined EEG-TMS study, NeuroImage 52 (2010) 316325. [53] P. Svenningsson, E. Westman, C. Ballard, D. Aarsland, Cognitive impairment in patients with Parkinson's disease: diagnosis, biomarkers, and treatment, The Lancet. Neurology 11 (2012) 697-707. [54] S.J. Teipel, W.H. Flatz, H. Heinsen, A.L. Bokde, S.O. Schoenberg, S. Stockel, O. Dietrich, M.F. Reiser, H.J. Moller, H. Hampel, Measurement of basal forebrain atrophy in Alzheimer's disease using MRI, Brain : a journal of neurology 128 (2005) 2626-2644. [55] P. Tiraboschi, L.A. Hansen, M. Alford, M.N. Sabbagh, B. Schoos, E. Masliah, L.J. Thal, J. Corey-Bloom, Cholinergic dysfunction in diseases with Lewy bodies, Neurology 54 (2000) 407-411. [56] H. Tokimura, V. Di Lazzaro, Y. Tokimura, A. Oliviero, P. Profice, A. Insola, P. Mazzone, P. Tonali, J.C. Rothwell, Short latency inhibition of human hand motor cortex by somatosensory input from the hand, The Journal of physiology 523 Pt 2 (2000) 503513. [57] K. Udupa, Z. Ni, C. Gunraj, R. Chen, Interactions between short latency afferent inhibition and long interval intracortical inhibition, Experimental brain research 199 (2009) 177-183. [58] A.J. Yarnall, L. Rochester, M.R. Baker, R. David, T.K. Khoo, G.W. Duncan, B. Galna, D.J. Burn, Short latency afferent inhibition: a biomarker for mild cognitive impairment in Parkinson's disease?, Movement disorders : official journal of the Movement Disorder Society 28 (2013) 1285-1288. [59] L. Zaborszky, L. Hoemke, H. Mohlberg, A. Schleicher, K. Amunts, K. Zilles, Stereotaxic probabilistic maps of the magnocellular cell groups in human basal forebrain, NeuroImage 42 (2008) 1127-1141. 16 FIGURE LEGENDS Fig. 1. Forest plot depicting the effect size (Hedges’ g) and associated 95% confidence interval for each study in the meta-analysis that compared SAI between PD patients and healthy controls. Negative values of the Hedges’ g indicate less SAI in PD patients (less inhibitory effect on the motor evoked potential). Effect sizes are grouped according to whether they belong to samples of patients with normal cognitive functioning or patients with cognitive impairment. Below each subgroup, a summary polygon shows the results when fitting a random-effects model (RE) for the subgroup analysis. At the bottom, summary effects calculated using RE model for all studies, along with heterogeneity estimates (Q-statistic and I2-statistic). A significant Q value indicates a lack of homogeneity of findings among studies; the proportion of observed variance that reflects real differences in effect sizes was estimated by I2. Fig. 2. Visualizing continuous and categorical moderator variables of SAI in PD patients. Hedges’ g effect size (Effect size of SAI) is on the y-axes and disease duration in years (panel A), MMSE score (panel B), affected side (panel C) and medication state (panel D) are on the x-axes. Each point (either circle or triangle) represents a study and the size of the point represents the study weight (inverse of variance), where larger points are larger sample size studies and are therefore more precise estimates of the population ES. Fig. 3. Analysis of SAI and its relationship with specific cognitive deficits in PD. Patients with deficits in at least one cognitive domain display less SAI than those with intact cognitive functions (Panel A). SAI impairment is associated with cognitive deficits in the four major cognitive domains (Panel B).