Extraction of corticospinal tract microstructural properties in chronic stroke

Highlights

• A variety of corticospinal tract reconstruction methods and properties are used post-stroke.

• Comparison of commonly used methods could help to identify a unified method for the field.

• Fractional anisotropy discriminated stroke severity.

• Location of tract reconstruction differentially related to motor impairment as compared to function.

Abstract

Background

Information about the structural integrity of the corticospinal tract (CST) from diffusion-weighted imaging can improve our ability to understand motor outcomes in people with upper limb impairment after stroke, especially those with severe impairment. Yet, there is no consensus on which method of CST generation most accurately represents function and impairment in individuals with chronic stroke.

New method

The aim of the study was to compare different methods of CST reconstruction and resulting microstructural properties, as well as the relationship between these properties and motor function and impairment. Fifteen individuals with mild-moderate impairment and 15 with severe impairment who were in the chronic phase post-stroke underwent a diffusion-weighted imaging scan and motor function and impairment assessments.

Results

Different relationships existed between reconstruction methods, microstructural properties, and impairment and function. In severe stroke, fractional anisotropy (FA) emerged over and above apparent diffusion coefficient (ADC) and tract number to index CST integrity; FA correlated with impairment and function, whereas ADC and tract number did not correlate. No significant differences between methods or microstructural properties were found in mild-moderate stroke.

Comparison with existing methods

Our study demonstrates that CST reconstruction method influences the extraction of microstructural integrity in individuals with chronic severe stroke, with FA appearing to be the most representative method. A similar line of investigation is warranted earlier post-stroke.

Conclusion

Differences in this data set highlight the need to establish a common methodology for CST reconstruction and analysis which may eliminate discrepancies in interpreting DWI and enhance biomarker use post-stroke for motor function.

Introduction

Brain based biomarkers may be useful to understand upper limb (UL) outcome and recovery after stroke and help to inform the stratification of people into rehabilitation interventions (Boyd et al., 2017). Most often motor outcome is explored using clinical assessments (Winters et al., 2015). However, this information has limited utility when considering individuals with more severe UL impairment (Hayward et al., 2017a, Hayward et al., 2017b), as these individuals expose clinical tests to floor effects (Winters et al., 2016).

Diffusion-weighted imaging (DWI) enables reconstruction of white matter tracts of the brain including the corticospinal tract (CST). The CST is the primary motor output pathway that has been implicated in motor outcome in the chronic phase post-stroke (Ward 2015). To index the CST, a region of interest (ROI) analysis is typically performed to generate and extract tract information (Lindberg et al., 2007; Stinear et al., 2007; Tang et al., 2010; Thomalla et al., 2004; Ward et al., 2006). Prior reports of CST integrity after stroke have chosen ROIs based on neuroanatomical landmarks drawn around the tract at the following anatomical landmarks: base of the corona radiata (CR), posterior limb of the internal capsule at the level of the anterior commissure (PLIC), and the pyramidal column in the pons (PONS) (Burke Quinlan et al., 2015; Sterr et al., 2014; Stinear et al., 2007; Werring et al., 2000). However, relationships between CST integrity and motor impairment reported across studies have been inconsistent (Groisser et al., 2014; Marshall et al., 2009; Puig et al., 2011; Zhu et al., 2010), even when controlling for time post-stroke, severity and age (Thomalla et al., 2004). It is possible that inconsistencies may arise as a result of different ROI-combinations used to generate the CST (Byblow et al., 2015; Jang et al., 2005; Liang et al., 2007; Park et al., 2013; Schaechter et al., 2008; Werring et al., 2000).

Another factor that may influence CST reconstruction is the method of tract reconstruction. There are two ways tracts can be isolated: 1) extract only tracts that are constrained between the ROIs, defined as the SEGMENTED tract method (Sterr et al., 2014; Stinear et al., 2007), or 2) extract all tracts that extend both upstream, in between and downstream of the selected ROIs, defined as the WHOLE tract method (Groisser et al., 2014; Hayward et al., 2017a, Hayward et al., 2017b). It remains to be determined if tract reconstruction approach yields different outcomes.

Once the ROI-combination and tract reconstruction methods have been selected, three microstructural properties may be explored: fractional anisotropy (FA), apparent diffusion coefficient (ADC), and estimated tract number. FA and ADC assesses how freely water molecules are able to diffuse and are indicative of the integrity of white matter microstructure, whereas tract number approximates the number of fibers present (Jones et al., 2013). These microstructural properties provide quantitative information regarding the integrity of the lesioned and non-lesioned CST tracts. FA, ADC, and tract number have all been used as an index of white matter degeneration in chronic stroke (Boespflug et al., 2014; Jang et al., 2005; Moller et al., 2007; Stinear et al., 2007; Thomalla et al., 2004)_. In individuals with chronic stroke, higher ipsilesional FA correlates with better motor function (Schaechter et al., 2009), higher ADC and a decrease in fiber number reflects damage to the CST tracts are reported but relationship with motor function remains largely unknown (Lutsep et al., 1997; Maraka et al., 2014). However, the microstructural properties have not been uniformly assessed in a chronic stroke population to determine the impact of different methods of tract reconstruction.

Without a standardized method for extracting information regarding the CST, it is difficult to establish the utility of the CST as a biomarker of motor outcome both within and across studies (Hayward et al., 2017a, Hayward et al., 2017b). As such, the current study aimed to explore CST ROI-combination and tract reconstruction approaches for commonly used microstructural properties of FA, ADC, and tract number. Given that severity of impairment may influence the resulting output (Stinear et al., 2007; Bigourdan et al., 2016; Hayward et al., 2017a, Hayward et al., 2017b), we included a group with mild-moderate upper limb impairment and a group with severe impairment. We hypothesized that 1) ROI-combination ‘CR + PLIC’ would implicate CST injury more consistently as compared to other ROI-combinations, 2) ‘segment’ tract reconstruction method would limit the amount of spurious fibers included within the extraction process, and therefore more accurately represent CST microstructural integrity, and 3) that microstructural integrity would be more intact in individuals with mild-moderate as compared to severe impairment. Next, we sought to determine if extracted microstructural properties were representative of functional ability of people with chronic upper limb (UL) impairment after stroke. We hypothesized that ‘CR + PLIC’ would be correlated more strongly with function and impairment as compared to other ROI locations.

Next, we sought to determine if extracted microstructural properties were representative of functional ability of people with chronic upper limb (UL) impairment after stroke. We hypothesized that ‘CR + PLIC’ would be correlated more strongly with functional ability and impairment as compared to other ROI locations.