Multiple sclerosis (MS) is a chronic inflammatory demyelinating and neurodegenerative disorder of the central nervous system. Ongoing inflammation causes neuronal damage in MS patients (Kotelnikova et al., 2017). Clinical, radiological, and histological classification systems have been proposed (Pitt et al., 2022). Histological evidence of inflammation is incorporated to categorize lesions into early active, late active, smoldering, inactive, and shadow plaques (Frischer et al., 2015). Refinement of these classification systems to incorporate recently developed radiologic techniques for non-invasive evidence of inflammation may further guide clinical management and elucidation of disease pathogenesis.

To date, a commonly recognized classification in the clinic is to differentiate active (acute) from inactive lesions. Introducing gadolinium (Gd)-based contrast agents, Gd contrast-enhanced MRI can identify acute MS lesions (Tourdias et al., 2012), because the inflammatory infiltration led by lymphocytes and microglia can affect blood-brain-barrier (BBB) functions, allowing extravasation of gadolinium-based contrast agents (Lassmann et al., 2007). However, the gadolinium enhancement of acute lesions usually only lasts for weeks (Mahad et al., 2015). There is evidence suggesting that the gadolinium-enhanced lesions would turn into non-gadolinium-enhanced lesions even with inflammation continuously proceeding (Maggi et al., 2014). Moreover, the potential risk of Gd accumulation requires clinicians to make judicious use of this imaging method (Wattjes et al., 2021). Clinical practitioners may be able to make more personalized treatments and monitor the therapeutic efficacy of treatment by reconsidering the interpretation of lesion activity, especially for Gd non-enhanced (Gd-negative) lesions.

There has been a significant amount of research dedicated to the recognition of smoldering lesions or detecting inflammatory changes in MS lesions. Quantitative susceptibility mapping (QSM) could reflect iron deposition when the relative lesion susceptibility value is above zero (Wisnieff et al., 2015). The susceptibility of MS lesions increases significantly when gadolinium-enhanced lesions evolve into non-gadolinium-enhanced lesions (Zhang et al., 2016b). In addition to iron, reactive oxygen species (ROS) are also reported to be key elements of inflammation (Dunham et al., 2017). The proton exchange rate (k_ex) MRI approach (Shaghaghi et al., 2019), derived from chemical exchange saturation transfer (CEST) imaging, has recently been proposed as a non-invasive approach to assess tissue oxidative stress since endogenous ROS have been shown to promote in vivo k_ex in tissue (Tain et al., 2019; Shaghaghi and Cai, 2022). Our team has previously conducted a preliminary study on k_ex MRI showing its potential to further characterize Gd-negative MS lesions (Ye et al., 2020). However, the contrast mechanism of k_ex-enhanced MRI as a surrogate biomarker for ROS has not been fully validated. Given that iron deposition catalyzes Fenton reaction that produces hydroxyl ROS (Tain et al., 2019), the correlation between QSM and k_ex MRI may help to validate k_ex MRI as an emerging inflammatory imaging contrast for MS stratification.

In the present study, we will compare k_ex MRI with the relatively well-established QSM in all MS lesions and investigate if the combination of k_ex MRI and QSM helps to stratify Gd-negative MS lesions based on lesion inflammatory activity.

In the present study, we further classified Gd^– lesions into three sub-patterns with different levels of k_ex and susceptibility elevation. Both k_ex and susceptibility elevation of Pattern I were higher than Pattern II. Furthermore, the patient with longer disease duration tended to have a lower percentage of lesions with Pattern I.

Multiple pathological processes occur during the course of MS, but the disruption of the BBB (Gaitan et al., 2011) the iron redistribution, and the oxidative damage (Dunham et al., 2017) are among the crucial inflammatory mechanisms. Assessing their correlation could be valuable for gaining a deeper understanding of lesion evolution. In the present study, three MR imaging sequences (Gd-enhanced, QSM, and k_ex MRI) were used to compare the structural damage or the molecular changes from these three mechanisms, respectively. Gd enhancement usually suggests acute inflammation and consequential BBB disruption. The degree of axon demyelination and iron deposition could be reflected by susceptibility, while k_ex has been demonstrated to be able to reflect the overproduction of ROS due to oxidative stress (Tain et al., 2019).

On such premises, Patterns I & II could be regarded as radiologically active for at least one sequence suggesting the existence of inflammation, while Pattern III (which are Gd^–, k_ex^–, and QSM^–) might be considered radiologically inactive because inflammation was not detected by either QSM or k_ex MRI. Although the inconsistency of k_ex elevation and QSM hyperintensity did exist, over 70% of all MS lesions were consistently positive on both sequences, which is likely because iron can also serve as the catalyst for the Fenton reactions producing ROS (Tain et al., 2018). This finding suggests that the inflammation due to iron redistribution and oxygen burst can persist regardless of BBB repair. Therefore, interpreting all Gd^– lesions as inactive is inaccurate.

Ring enhancement, compared with nodular enhancement, has been considered to result in longer and more severe tissue damage (Davis et al., 2010). Patients with ring-enhancing lesions were also reported to be to reach a EDSS of 4.0-6.0 in shorter time (Llufriu et al., 2010). Ring hyperintensity on QSM, as a sign of iron-containing cells surrounding the lesion edge, has also been related to more severe demyelination and chronic lesion activity/expansion (Absinta et al., 2013; Zhang et al., 2016a). Of the Gd-enhanced lesions (the acute lesions), lesion k_ex elevation was higher in either ring Gd⁺ lesions or ring QSM⁺ lesions than in nodular lesions, which implies that k_ex elevation is positively correlated with the severity of lesion inflammation though further pathologic evidence based on animal experiments are pending for verification. Endogenous ROS of MS is reported to come from the oxygen burst in microglia, inducing mitochondrial dysfunction and subsequent histotoxic hypoxia (Fischer et al., 2012). This further results in energy deficiency and ionic imbalance in oligodendrocytes, axons, and neurons (Mahad et al., 2015). When the respiratory burst leads to cellular degradation, the ferrous iron released from microglia and macrophages may in turn amplify inflammatory oxidative injury and neurodegeneration (Haider, 2015). The mild positive correlation between susceptibility and k_ex elevation we found may just reflect the interplay of iron content and ROS in MS lesions.

Based on the potential mutual reinforcement of iron deposition and oxygen damage, Pattern I is more likely to be inflammatory than Pattern II. For patients with relapsing-remitting courses, the inflammatory status of lesions might not always stay at the same level; instead, it could drop from a higher level to a lower one. The change in the percentage of Pattern I lesions with disease duration might reflect such evolution of chronic inflammation. As for the change in the percentage of Pattern II lesions, it may represent the combination of two different evolutionary trends, since Pattern II consists of two subtypes: IIa (which are k_ex⁺ and QSM^–) and IIb (which are k_ex^– and QSM⁺). Comparing to the other two patterns, a relatively larger standard deviation of k_ex elevation in Pattern II might also reflect the highly heterogenous nature of it. In the present study, the two subtypes were integrated into one pattern because of the possibly similar level of inflammation, which was lower than Pattern I but higher than Patter III. The difference between k_ex MRI and QSM reflects their differences in contrast mechanism. Nevertheless, attributing k_ex MRI contrast to oxidative stress and ROS production and comparing iron-induced inflammation requires further validation with biochemical analyses.

Lesion susceptibility elevation has generally been reported to remain relatively stable for months or years (Stuber et al., 2016), but lesion k_ex elevation may possess a more volatile trait since if the balance of pro- and antioxidants has not been achieved, ROS may not cease to fluctuate. Unlike the other two imaging sequences, the more susceptible ROS production seems to make the k_ex map a more sensitive method for inflammation indication and curative response monitoring. By enhancing the sensitivity and specificity, the combination of k_ex and QSM MRI could help to further stratify the lesion inflammatory status and detect smoldering lesions after the acute stage. In combination with QSM, k_ex MRI is promising for detection of lesion inflammation or activity.

Clinically, MS patients with a slow or moderate disease course typically receive a moderately effective and very safe medication, whereas patients who demonstrate highly active or rapidly evolving disease are candidates for highly effective but potentially less safe treatment (Thompson et al., 2018b). The time lost in the wait-and-see approach delays highly critical intervention for patients who are later identified as rapidly progressing. Therefore, an improved ability to risk-stratify MS lesion activity with k_ex MRI would help predict prognosis and guide treatment. The clinicians might also utilize non-invasive longitudinal k_ex MRI to monitor the therapeutic efficacy of treatment. For example, a drop of Pattern I percentage in all MS lesions might suggest good treatment response. With an extended lesion classification, a more personalized treatment is facilitated.

This study presents a preliminary exploration of the feasibility of QSM and k_ex MRI in MS lesion categorization based on retrospective, cross-sectional, single-center data. The sample size is relatively small, and the small lesion amount affected the statistical analysis of the two subtypes of Pattern II as well as Pattern III. To conform with k_ex MRI, the image analysis was limited to one slice of image for each case and was also restricted by the spatial resolution of Z-spectral MRI. The errors on lesion registration and kex measurement may exist because there are small lesions which are challenging to differentiate from potential ischemic lesions. Moreover, further validation with pathological studies on preclinical animal models is needed. Research with larger samples, prospective and longitudinal designs, and invasive biochemical validation should be carried out.

In conclusion, the study extended the classification of MS lesions by the combination of three MRI sequences (Gd-enhanced, QSM, and k_ex MRI). Underlying inflammation of Gd^– lesions was further visualized by QSM and k_ex MRI, suggesting their utility in assessing the inflammation status of lesions previously considered inactive with Gd-enhanced MRI.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Institutional Review Board of Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

HL: analysis and interpretation of data, drafting and revision of the manuscript for content, including medical writing for content, and study concept or design. ZC: processing, analysis and interpretation of data, drafting of the manuscript for content, and including medical writing for content. HY and QC: major role in the acquisition of data and analysis of data. YZ: acquisition of data. MS and SL: revision of the manuscript for content and including medical writing for content. WC: study concept or design, acquisition and interpretation of data, and revision of the manuscript for content. KC: study concept or design, interpretation of data, and revision of the manuscript for content. All authors contributed to the article and approved the submitted version.