Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease characterized by the progressive death of upper motor neurons (UMNs) of the primary motor cortex and corticospinal tract (CST), in conjunction with lower motor neurons (LMNs) associated with the anterior horns. Despite the recognition of UMN and LMN involvement as a characteristic signature, the mean diagnostic delay among ALS patients is around 12 months, primarily due to patients being misdiagnosed with more common diseases that might mimic the early stages of ALS [1
]. The long diagnostic delay underscores the dearth of robust and sensitive clinical biomarkers in ALS. Clinical indicators of disease status, such as Revised ALS Functional Rating Scale (ALSFRS-R), electromyography (EMG), and muscle strength tests, may be confounded by inter-rater variability and/or low sensitivity to ALS; these factors are further compounded by the clinical heterogeneity of disease onset and progression. The efficacy of ALSFRS-R as a measure of clinical outcome can also be affected by the underreporting of functional impairment severity [2
]. Hence, there is an urgent need to establish robust, non-invasive, and quantitative biomarkers that can serve as early and specific diagnostic and prognostic indicators of disease.
Although current literature on motor neuron dysfunction in ALS is extensive, there is a continuing debate about whether motor neuron death is of a forward (UMN spreading to LMN) or backward (LMN spreading to UMN) nature [3
]. Irrespective of the progression pathway, neuromuscular junction degeneration leads to skeletal muscle denervation and is known to accompany clinical symptom onset. Symptomatically, around 75% of ALS patients present limb muscle weakness, while others present a bulbar onset [4
To date, the clinical role of magnetic resonance imaging (MRI) in ALS has primarily been limited to exclusion of other neurodegenerative diseases that present similar symptoms to ALS [6
]. Advanced quantitative imaging methods, while promising, have predominantly focused on evaluating singular pathologic characteristics, such as the disrupted fiber tracks of UMNs [7
]. Clinically, LMN dysfunction is primarily assessed using electromyography [9
], electrical impedance myography [10
], muscle ultrasonography [12
] or muscle biopsies [13
]. Previous imaging studies have employed semi-quantitative or quantitative approaches to characterize the LMN pathways of ALS disease progression [10
]. One of the earliest studies of ALS muscle demonstrated that while edema regions exhibit relative T1
signal increases, fatty infiltration causes relative T1
to decrease and relative T2
to increase [11
]. Another study used the relative T2
signal approach, wherein the authors additionally implemented a whole-body imaging protocol for their study [14
]. They demonstrated that for ALS patients with bulbar, upper extremity, or lower extremity onset, the affected muscle regions exhibited increased relative T2
signal when compared with healthy muscle. More recently, diffusion-weighted imaging (DWI) has been added to this whole-body approach [14
] and used to longitudinally evaluate patients over 12 months. The authors concluded that relative T2
signal, in comparison with DWI, was most effective at detecting longitudinal changes in leg muscle groups. A more quantitative approach was implemented when evaluating skeletal muscle differences in spinal and bulbar muscular atrophy (SBMA) and ALS patient populations when compared with healthy controls [15
]. Fatty infiltration due to atrophy occurring from muscle denervation was quantified using m-Dixon-based fat fraction measures, and a semi-quantitative short tau inversion recovery (STIR) imaging approach was employed to evaluate edema arising from denervation in thigh, calf and tongue muscles. The fat fraction measures were more sensitive to differentiating SBMA muscle from healthy muscle, while ALS muscles were better identified using the STIR-based approach.
While longitudinal studies have established the potential for muscle imaging to interrogate motor neuron disease progression using relative signal changes and quantitative measures of fatty infiltration [16
], there remains a need for robust, quantitative, and sensitive imaging biomarkers to characterize LMN dysfunction, and in particular, ones that interrogate ALS-associated myofiber pathology. Biomarkers that are sensitive to myofiber architecture could enable more robust detection of disease progression and therapy response, as compared with downstream and indirect surrogates of disease status, such as edema. Contrast-enhanced MRI techniques have been successfully used to characterize tissue pathophysiology. We recently showed that, by simultaneously quantifying T1
changes associated with the dynamic contrast agent passage, a unique parameter that reflects cellular microstructure can be quantified—a technique previously termed relaxivity contrast imaging (RCI) [17
]. In particular, the contrast agent’s transverse relaxivity at tracer equilibrium, or TRATE, is an RCI parameter that is shown to be predominantly sensitive to cellular microstructure [18
]. Given the microstructural changes in muscle myofibers that accompany, and potentially precede, muscle degeneration and atrophy, the purpose of this study is to provide the first evaluation of TRATE as a quantitative, noninvasive muscle imaging biomarker for ALS characterization and progression.
In this work, we have demonstrated a quantitative approach to consistently differentiate ALS-affected calf muscle from healthy muscle by comparing TRATE, fat fraction, and relaxometry derived T2 values. A major highlight of this work is that among all quantitative imaging and clinical metrics, only TRATE demonstrated consistent changes between (a) ALS and healthy muscle, and (b) the two timepoints of data acquisition for every single ALS patient. These findings support the hypothesis that RCI is sensitive to ALS-induced aberrations in muscle myofiber architectural features (e.g., reduced fiber diameter, density, atypia).
The use of a multi-echo dynamic gradient echo sequence enabled the simultaneous quantification of T1
relaxation times [17
]. The T1
changes enable quantification of the local contrast agent concentration. The T2*
changes depend on the contrast agent concentration but, more importantly, reflect contrast-agent-induced magnetic field perturbations resulting from the compartmentalization of the agent within tissue compartments (e.g., extracellular space surrounding myofibers). A fundamental characteristic of the resulting perturbations is their dependence on the geometry of the compartment containing the contrast agent. In muscle, the contrast agent can leak out of blood vessels and distribute around muscle fibers. The microscopic interaction of contrast agent and water in each compartment leads to the observed T1
changes. While this interaction will shorten local T2
values, the more predominant effect on the observed T2*
changes originates from the mesoscopic magnetic field perturbations that occur as the contrast agent is compartmentalized around the myofibers. Accordingly, the differences in the dynamic ∆R2
* and ∆R1
, from which Ct
is computed, can be attributed to their dissimilar contrast mechanisms. In this patient cohort, relatively minor changes to Ct
were observed between ALS and healthy controls, indicating a larger influence of ∆R2
* on TRATE. The muscle-associated ∆R2
* changes reflect the local fiber properties, such as fiber density, geometry, organization, heterogeneity and size. Increased muscle atrophy as a result of the muscle fiber denervation process with ALS disease progression is expected to influence the muscle fiber density and diameter, which our prior computation studies predict should lead to reduced TRATE values [18
], as observed herein when comparing healthy versus ALS muscle and a given ALS muscle across time. TRATE is also inversely affected by contrast agent concentration. However, we observed that the contrast agent concentration evolution over time was comparable in both healthy and ALS-affected muscle and hence had minimal effect on TRATE. This also suggests that DCE-MRI alone may not be able to detect ALS-induced changes in muscle microstructure.
Another important observation was that relative change in TRATE, fat fraction and T2
between ALS and healthy muscles varied across different muscle groups. This was consistent with the heterogeneity in relative T2
measures observed across different muscle groups, as reported previously [15
]. The heterogeneity could be attributed to varying degrees of muscle atrophy within individual ALS patients and the respective contrast mechanisms of each imaging approach. It has been hypothesized that some of the more active muscle groups such as the Tibialis Anterior have a higher atrophy rate [25
]. At the time of imaging, the extent of muscle atrophy for each muscle group could be different due to the time elapsed between disease onset and imaging timepoint(s) in the study population. Additionally, this phenomenon could also be influenced by a multitude of other factors, such as age, exercise, muscle fiber orientation, and muscle volume.
This work also provides a preliminary evaluation of RCI-based TRATE as a biomarker of ALS disease progression. TRATE consistently declined in ALS patient lower extremity muscles, while the clinical measures remained stable or even improved slightly. TRATE reduced by a greater amount over time in the peroneus longus muscles than the tibialis anterior muscles. The tibialis anterior changes could have occurred earlier in the disease progression, while the larger peroneus longus changes could have occurred during the imaging time points of the study. Future studies will evaluate earlier stages of the disease to more systematically characterize longitudinal TRATE changes in each muscle group. While it is not uncommon to observe unchanged ALSFRS-R scores in patients with slowly progressing ALS after a period of 6 months, the results suggest that TRATE may outperform clinical measures, which can have significant variability, as a measure of disease progression. It has to be noted that only a subset of the ALS patient population was scanned at the 6-month timepoint. If a larger data set can confirm this, TRATE may become a useful measure in clinical trials to detect treatment effects better than some clinical measures.
There are a few limitations to this study. Primarily, the patient and control population sizes are small. However, future longitudinal studies will aim to increase patient population size. Secondly, the muscle regions of interest were manually segmented and could likely benefit from automated/semi-automated muscle segmentation tools for improved consistency [26
]. Due to scan times being influenced by the ability of ALS patients to lie still in the MRI scanner, additional limits had to be imposed on the image quality with respect to spatial and temporal resolution, scan time for each protocol and number of scans. Thirdly, longitudinal measurement of TRATE in limb muscle does not inform us of upper motor neuron dysfunction in ALS, which occurs simultaneously with lower motor neuron dysfunction and also affects the clinical measures we are comparing our imaging markers with.
Further work is required to validate RCI’s potential as a biomarker for patients with ALS. To systematically characterize differences between healthy and ALS muscle, it would be informative to characterize TRATE across relevant age groups. Additional information could be obtained from an RCI performance comparison between sporadic ALS and familial forms of ALS. Some ALS gene mutations can impact muscle, which may more dramatically impact TRATE measures. While this study focused on imaging muscles in the leg, it will be important evaluate RCI’s potential in other relevant muscle groups (e.g., bulbar imaging). To establish its utility as a biomarker, the repeatability of RCI needs to be established, and its sensitivity to early therapeutic response should be compared with other markers of myofiber microstructure (e.g., electromyography, electric impendence myography and muscle ultrasonography [9
]). Preclinical and computational studies could also shed light on the myofiber microstructural features that contribute to the observed TRATE changes and help guide future clinical interpretation.