Prediction of rehabilitation induced motor recovery after stroke using a multi-dimensional and multi-modal approach

When the hell are you people going to actually deliver EXACT RECOVERY PROTOCOLS? Instead of this crapola of predicting failure to recover?

Prediction of rehabilitation induced motor recovery after stroke using a multi-dimensional and multi-modal approach

Silvia Salvalaggio^1,2*, Andrea Turolla^3,4, Martina Andò¹, Rita Barresi¹, Francesca Burgio¹, Pierpaolo Busan¹, Anna Maria Cortese⁵, Daniela D’Imperio¹, Laura Danesin¹, Giulio Ferrazzi⁶, Lorenza Maistrello¹, Eleonora Mascotto⁷, Ilaria Parrotta¹, Rachele Pezzetta¹, Elena Rigon¹, Anna Vedovato⁸, Sara Zago¹, Marco Zorzi^1,2,9, Giorgio Arcara¹, Dante Mantini^1,10 and Nicola Filippini^1*

• ¹IRCCS San Camillo Hospital, Venice, Italy
• ²Padova Neuroscience Center, Università degli Studi di Padova, Padua, Italy
• ³Department of Biomedical and Neuromotor Sciences (DIBINEM), Alma Mater Studiorum – Università di Bologna, Bologna, Italy
• ⁴Unit of Occupational Medicine, IRCCS Azienda Ospedaliero-Universitaria di Bologna, Bologna, Italy
• ⁵Department of Rehabilitation Medicine, AULSS 3 Serenissima, Venice, Italy
• ⁶Philips Healthcare, Milan, Italy
• ⁷Department of Physical Medicine and Rehabilitation, Venice Hospital, Venice, Italy
• ⁸General Hospital San Camillo of Treviso, Treviso, Italy
• ⁹Department of General Psychology, University of Padova, Padua, Italy
• ¹⁰Movement Control and Neuroplasticity Research Group, KU Leuven, Leuven, Belgium

Background: Stroke is a debilitating disease affecting millions of people worldwide. Despite the survival rate has significantly increased over the years, many stroke survivors are left with severe impairments impacting their quality of life. Rehabilitation programs have proved to be successful in improving the recovery process. However, a reliable model of sensorimotor recovery and a clear identification of predictive markers of rehabilitation-induced recovery are still needed. This article introduces the cross-modality protocols designed to investigate the rehabilitation treatment’s effect in a group of stroke survivors.

Methods/design: A total of 75 stroke patients, admitted at the IRCCS San Camillo rehabilitation Hospital in Venice (Italy), will be included in this study. Here, we describe the rehabilitation programs, clinical, neuropsychological, and physiological/imaging [including electroencephalography (EEG), transcranial magnetic stimulation (TMS), and magnetic resonance imaging (MRI) techniques] protocols set up for this study. Blood collection for the characterization of predictive biological biomarkers will also be taken. Measures derived from data acquired will be used as candidate predictors of motor recovery.

Discussion/summary: The integration of cutting-edge physiological and imaging techniques, with clinical and cognitive assessment, dose of rehabilitation and biological variables will provide a unique opportunity to define a predictive model of recovery in stroke patients. Taken together, the data acquired in this project will help to define a model of rehabilitation induced sensorimotor recovery, with the final aim of developing personalized treatments promoting the greatest chance of recovery of the compromised functions.

1. Introduction

Stroke is a cerebrovascular disease representing the third cause of death in high socio-demographic countries (Institute for Health Metrics and Evaluation [IHME], 2018). Improvements in prevention and treatment of the acute stage have significantly increased the survival rate. However, stroke remains a leading cause of severe long-term motor disability, affecting the quality of life of stroke survivors, limiting their return to a normal life, and representing a burden for their families. Worldwide there are over 33 million stroke survivors most of whom suffer from long-term disability (Feigin and Krishnamurthi, 2010).

Upper limb (UL) impairment represents the most impacting long-term disability caused by stroke. Among stroke survivors, motor impairment can be related to different aspects of movement, such as motor planning, learning, and control (Pollock et al., 2014). The aim of rehabilitation-mediated recovery is twofold: (i) minimize sequelae and (ii) improve the recovery of the affected limb(s).

Despite important advances in the medical and physical rehabilitation fields, neurological rehabilitation still lacks a reliable physiological model of UL sensorimotor recovery after stroke. For example, the magnitude of the motor recovery, occurring after rehabilitation treatment, that can be effectively attributed to the rehabilitation process itself is still largely unclear. Moreover, a precise identification and characterization of the key markers to be used as predictive factors of rehabilitation-induced recovery is needed (Hayward et al., 2017).

To date, both in clinical trials and current clinical practice, many treatment methods and assessment tools are available for quantification of the final outcome. Alongside the traditional therapeutic approaches (generally referred as “Conventional Therapy,” CT) based on neurodevelopmental principles, the so-called “innovative approaches” (e.g., robotics and virtual reality – VR) emerged recently providing an augmented environment with reinforced feedback to the patient. These technology-based methods have become popular, although current evidence on their efficacy is still under investigation (Feigin and Krishnamurthi, 2010; Bernhardt et al., 2016).

Evidence suggests that spontaneous recovery expresses its maximum effect from 3 to 6 months (Bernhardt et al., 2017), nevertheless recent results demonstrated that rehabilitation interventions can promote clinically significant improvements of motor outcomes even after this sensitive window, especially if high dose of therapy is provided (Daly et al., 2019; Ward et al., 2019). Indeed, motor improvements in UL recovery were found to be achieved after 90–300 h of rehabilitation, even in the chronic phase (Daly et al., 2019; Ward et al., 2019). Moreover, data available in the literature also emphasize the importance of the so-called “active ingredients” of rehabilitation (i.e., the specific elements which are assumed to be responsible for the treatment effect), but which are often not defined, classified, or measured (Ward et al., 2019).

During motor performance, there is an integration between motor and cognitive components (Barrett and Muzaffar, 2014; D’Imperio et al., 2021). For example, the control of sensorimotor aspects of motor actions requires attentional and cognitive demands, especially when performing complex movements. Consequently, cognitive abilities, such as attention, may play a key role, particularly in individuals recovering from stroke, as suggested by several studies (McDowd et al., 2003; Krakauer, 2006; Shafizadeh et al., 2017; VanGilder et al., 2020). This indicates that the identification of cognitive skills capable of guiding and predicting motor recovery could help in “a priori” patients stratification based on different recovery potentials.

Alongside motor recovery, prediction of the optimal level of functional improvement to be expected, is also a critical aspect in the stroke rehabilitation field. Indeed, prediction may be a useful guidance for setting rehabilitation goals and monitoring patient’s achievements over time (Piscitelli et al., 2018). Up to date, literature has mainly focused on prognostic factors (i.e., considering spontaneous recovery) rather than predictive ones (i.e., considering response to a rehabilitative intervention) (Clark, 2008). Coupar et al. in a systematic review examined potential factors for predicting UL recovery, such as: (i) preserved conduction and anatomical integrity of the cortico-spinal tract (CST) confirmed by motor evoked potentials (MEPs), and fractional anisotropy (FA), respectively; (ii) preserved sensation function; (iii) strong Shoulder Abduction and Finger Extension (SAFE) (Coupar et al., 2012; Stinear et al., 2017b). However, these studies did not take into account whether patients received rehabilitation or the dosage and modalities used, therefore, the impact of therapy and/or dosage on the accuracy of predictions has not yet been evaluated (Coupar et al., 2012; Stinear et al., 2017b). One study suggested that the best predictors of response to robotic treatment were preserved CST integrity, great ipsilesional motor cortex activation, and great interhemispheric connectivity, although the dosage used was still relatively low and not investigated as a factor potentially influencing the recovery prediction (Burke Quinlan et al., 2015). Likewise, improvement of motor functions after stroke can be associated with some changes in brain activity and connectivity as measured with electroencephalography (EEG): on inter-regional synchronization of neural oscillations (Nicolo et al., 2015; Romeo et al., 2021), on the evoked responses as measured in an oddball task (Naatanen et al., 2012), or in the brain oscillations evoked during gamma entrainment (Pellegrino et al., 2019). Some recent results also suggest that the aperiodic component of the power spectrum can have an important prognostic role in brain damaged patients (Maschke et al., 2023).

With regards to recovery prediction, data from the literature suggest that specific algorithms can be used to provide an accurate prediction of UL functional recovery. These algorithms are mostly based on clinical, imaging (magnetic resonance imaging – MRI) and neurophysiological (MEPs, collected with transcranial magnetic stimulation – TMS) measures. However, the validity of such models has been investigated only with assessments performed within a time window from 2 to 11 days (i.e., 2, 3, 5, 9, and 11 days) and UL prediction recovery at 3 months (Stinear, 2017). Moreover, factors such as: (i) variability in time of patient transfer to rehabilitation facilities, (ii) absence or delay in the acquisition of timely specific information (i.e., MEPs, FA, and National Institute of Health Stroke Scale – NIHSS), and (iii) different phases of recovery (i.e., subacute and chronic) when patients join the rehabilitation care, represent some of the most common confounders that can severely affect the predictive accuracy of the algorithms (Smania et al., 2009; Stinear, 2010, 2017; Stinear et al., 2012, 2017a). If only clinical outcome measures are available, predicting the effect of true recovery due to “treatment” over time is virtually impossible. Indeed, standalone clinical measures do not allow the thorough interpretation of the mechanisms underlying UL functional recovery (Stinear et al., 2007). For these reasons, imaging and physiological techniques, such as functional magnetic resonance imaging (fMRI) and TMS, have been previously used to accrue information on motor-related brain regions functionality following stroke recovery (Allman et al., 2016; Smith and Stinear, 2016). Similarly, diffusion tensor imaging (DTI) technique can be used to determine the anatomo-histological integrity of motor-related white matter (WM) pathways (Stinear et al., 2007). Recent preliminary studies have also suggested that measuring plasma levels of specific miRNAs (small RNA molecules that act as regulators of cell development, proliferation, cycling, and differentiation), and determining the specific genetic profile of stroke patients could provide useful biologically derived predictive markers of functional recovery (Chang et al., 2016; Edwardson et al., 2018; Dimyan et al., 2022).

Here, we report the protocol of the study “NeuroPro” (Investigation of NEUROphysiological substrates of UL sensorimotor impairment after stroke and PROgnosis of rehabilitation-induced recovery of motor function: longitudinal study), which is a longitudinal study employing a multi-dimensional and multi-modal approach to evaluate motor function and predict recovery in stroke inpatients in a clinical setting. We will combine clinical, biological, neurophysiological/imaging, neuropsychological, and rehabilitation measures with the final aim of developing a multidimensional predictive model of motor recovery after stroke. Whereas clinical, neuropsychological, and rehabilitation measures are usually acquired in ecological clinical settings, the possibility of performing this study in a rehabilitation research center, with both clinical and research compounds in the same institute, will allow a detailed characterization of participants (e.g., clinical profiling, operationalization of pharmacological treatment, and individual frailty levels). Moreover, the access to neurophysiological/imaging and biological facilities, will grant us the possibility to acquire a set of specific measures to develop a comprehensive and sensitive model of motor recovery, after rehabilitation treatments in stroke survivors.

More at link.