Depression as a disease of white matter network disruption: Learning from Multiple Sclerosis

Abstract

Depression is a common and debilitating psychiatric disorder that is associated with substantial morbidity and mortality. For nearly 40 years, scientists have attempted to localize depression in the brain with neuroimaging. Despite concerted research, many meta-analyses of neuroimaging research in depression have not identified a single brain region that, when lesioned, causes depression. However, recent work using coordinate and lesion network mapping has identified a distributed network from functional magnetic resonance imaging that is relevant for depression. In this review, we propose that multiple sclerosis (MS) is a powerful model to study the relationship between perturbations to white matter brain networks and depression. We highlight recent successes in using lesion network mapping to find associations between depression and MS lesion location and burden in retrospective samples. Next, we describe why prospective, longitudinal studies that track the onset of white matter lesions in MS with the emergence and resolution of depression may provide a way to understand both the pathophysiology of depression in MS and network mechanisms of depression more broadly.

Depression is a highly prevalent and debilitating psychiatric disorder that contributes to substantial morbidity, mortality, and economic burden. An estimated 322 million individuals, or 4.4% of the global population, are affected by depression(1). Its global prevalence increased by 18.4% between 2005 and 2015(1). Depression is strongly associated with elevated suicide risk, accounting for approximately one million deaths each year and ranking as a leading cause of preventable death worldwide(2). In the United States, the annual cost of depression is estimated in the hundreds of billions of dollars, driven by both direct treatment expenses and loss of productivity, disability, and absenteeism(3-5). The World Health Organization has identified depression as the most burdensome illness during midlife, and projects it to become the leading contributor to global disease burden by 2030(6,7).

Approximately 51.8% of Americans live with chronic medical conditions, many of which commonly co-occur with depressive disorders(8,9). When depression co-occurs with a chronic illness, the health burden is greater than either condition alone(9). This interaction exhibits a bidirectional relationship, with chronic illness serving as a risk factor for depression, and depression contributing to the prevalence and severity of chronic illnesses through increased symptom burden, functional impairment, impaired self-care, and treatment nonadherence(9,10). Moreover, depression is associated with a twofold increase in all-cause mortality(6).

Given the profound burden of depression, one important goal in psychiatric research is identifying its underlying neurobiology. Despite decades of depression neuroimaging research in medically healthy populations, many meta-analyses have not identified a single location that, when injured or functioning abnormally, causes depression(11-14). In contrast, there is an extensive literature that has identify a functional brain network that is relevant for depression(15-17). However, less is known about the consequences of lesioning the structural backbone that supports it. Here, we propose using multiple sclerosis (MS) as a model to study the relationship between white matter network disruption and depression. MS is a common neurological disorder, affecting one million people in the US, and up to 50% of people with MS (pwMS) experience depression(18-21). Depression rates in MS are higher than in other chronic autoimmune diseases, suggesting that brain pathology may confer increased depression risk(22). MS is characterized by white matter lesions that impact neurotransmission and provides an opportunity to examine how brain network disruption contributes to depression. We describe how this approach may both elucidate the pathophysiology of depression in MS and also provide important insights into the network-based mechanisms underlying depression more broadly.

Neuroimaging in depression: from lesions to networks

For nearly 40 years, scientists have attempted to localize the neurobiological source of depression. With the introduction of positron emission tomography and magnetic resonance imaging (MRI) in the 1980 and 90s(23-27), neuroimaging became a cornerstone of depression research(28). To date, over 15,000 studies have contributed insights into the association between abnormal brain function and depression. From early work demonstrating that strokes in the left dorsolateral prefrontal cortex cause depression to a recent precision functional mapping study highlighting trait-based frontostriatal salience network expansion in depressed individuals(29), it is clear that depression is a heterogeneous condition related to abnormalities in distributed brain regions(30-32).

The challenges in localizing depression to a single brain region motivated scientists to examine whether depression better mapped to a distributed brain network. A common phenomenon in neurology is that post-stroke neuropsychiatric symptoms can emerge from heterogeneous stroke locations(33,34). A researcher team in Boston hypothesized that lesions to the same network, rather than location, produce similar symptoms(33,34). To relate heterogeneous lesions to neuropsychiatric symptoms, they developed a novel technique, lesion network mapping (LNM; Figure 1)(33,34). In LNM, a neuropsychiatric symptom is identified (e.g., peduncular hallucinosis), and a cohort of cases (e.g., persons with post-stroke peduncular hallucinosis) and controls (post-stroke patients without peduncular hallucinosis) is constructed. The strokes are next used as seeds in functional connectivity analyses in a healthy population to identify the network connected to the lesions. Lesion networks are compared between cases and controls to identify brain networks that, when injured, produce symptoms(33,34). Lesion networks have been identified across a range of syndromes (e.g., psychosis, addiction, free will)(16,33,35-37). In the last 5 years, this team has identified and replicated a lesion network map for depression(15,16). This map included key regions consistently identified across numerous localization studies, including the left dorsolateral prefrontal cortex, subgenual cingulate, and regions that comprise the frontoparietal network(14,30,38-40).

Figure 1. Using the Human Brain Connectome to Localize Symptoms from Focal Brain Lesions.

Lesions that cause the same symptom but occur in different brain locations (Panel A) can be overlaid on a map of anatomical connectivity (Panel B) or functional connectivity (Panel C) to determine whether they are part of the same connected brain network. With lesion network mapping, lesion locations from different patients that cause the same symptom are traced on a common atlas (Panel D, left column). Functional connectivity between each lesion location and the rest of the brain is computed with the use of the connectome (Panel D, middle column). Lesion network maps can then be overlapped to identify common connections (Panel D, right column). In this example, lesion locations that cause visual hallucinations are functionally connected to a part of the brain involved in visual imagery (red circles). Panel D is modified with permission from Boes et al.(33) From New England Journal of Medicine, Michael D. Fox, M.D., Ph.D., Mapping Symptoms to Brain Networks with the Human Connectome, Volume 379(23), Page No. 2237-2245. Copyright © (2018) Massachusetts Medical Society. Reprinted with permission from Massachusetts Medical Society.

The success of LNM in depression suggests several opportunities for further progress. For example, classic LNM successfully identifies associations in people who develop symptoms post-stroke. Prospective studies that track the emergence of depression following lesion onset could provide further causal evidence(41). Additionally, LNM identified creative ways to use brain pathology to learn about general network mechanisms of psychopathology. Expanding these models to include broader lesion types would enhance generalizability. Finally, LNM is scalable. Over the years, lesions have been identified via literature, via manual segmentation, and now with automated segmentation techniques. With each iteration, the sample sizes that can be analyzed have increased. While most studies currently do lesion segmentation on research-grade scans, leveraging clinical populations with high depression comorbidity that are imaged as part of routine care -like pwMS - would massively increase the scale of data available(42,43).

Why MS is a good model to study depression

Despite the potential utility of MS for understanding mechanisms of depression, many mental health researchers and clinicians may not be familiar with the condition. Below, we provide a primer on its presentation, pathophysiology, and clinical course.

MS Presentation

MS is a chronic, immune-mediated neurological disorder that is associated with substantial morbidity and reduced lifespan(44-46). The cause(s) of MS involve a combination of genetic risk factors and environmental triggers(44). MS symptoms are highly variable and are often associated with the location of injury in the central nervous system(47). They can be acute and resolve, or progressively worsen over time(44,46). While the best-known symptoms of MS are fatigue, visual problems, sensory disturbances, motor impairment, and cognitive dysfunction(48,49), up to 50% of pwMS will experience a lifetime depression(20,21). Importantly, pwMS identify depression as a substantial contributor to poor quality of life and impaired physical and social functioning(50,51).

MS Clinical Course and Pathobiology

MS was traditionally conceptualized as a disease of independent stages defined by either acute relapses of neurological symptoms or progressive symptom worsening, and categorized into three clinical phenotypes: relapse-remitting, primary progressive, and secondary progressive(52,53). However, clinical relapse biology and progressive disease are distinct pathophysiologic processes that co-occur(53). Relapse biology involves focal inflammatory demyelinating lesions that develop from autoreactive peripheral immune cell-mediated injury(44,46). Early in lesion formation, peripheral immune cells, including T cells and phagocytic cells, infiltrate across a disrupted blood-brain-barrier, leading to acute inflammatory demyelination and axonal injury(44,45,54). New lesion formation may be clinically silent or, depending on lesion location and injury severity, cause new neurological symptoms (i.e., clinical relapse) and slow recovery (i.e., remission)(42,47,55).

Clinical progression, in contrast, can be due to relapses that do not fully recover, or occur as the result of non-relapsing progressive MS biology. The latter involves more chronic and widespread processes of inflammation and neurodegeneration that are compartmentalized within the CNS, and result in gradual injury to neurons and glia(56). An important driver of the more chronic inflammation involved in non-relapsing progressive MS biology can be found within a subset of focal MS lesions that, rather than resolving following the acute phase, develop into chronic-active, or smoldering, lesions(57). Treatment of acute MS relapses may include a course of corticosteroids to expedite recovery, while prevention of new relapses requires long-term management with a disease-modifying treatment (DMT) that targets peripheral immune pathways(58,59). While DMTs prevent new lesion development, they are substantially less effective at slowing the non-relapsing progressive biology and do not reverse existing disability(60,61).

MS Clinical Monitoring

In contrast to the feasibility challenges (e.g., cost, recruitment time) associated with depression neuroimaging research in medically healthy populations (described above), MRI is the most common paraclinical tool for clinically monitoring MS(62). MRIs are usually performed at the time of presentation to establish a diagnosis via the McDonald Criteria (Box 1)(62). Following DMT initiation, MRIs are often repeated within a few months to establish a new baseline for subsequent monitoring of treatment efficacy(62,63). Thereafter, on average, pwMS undergo annual MRIs both to identify new disease activity (i.e., new lesion development) and for DMT safety monitoring(63). International guidelines for clinical MRI parameters in MS include research-grade 3-tesla (3T), 3-dimensional, high-resolution T2-weighted FLAIR and T1-weighted imaging, though 1.5T may be used(62). As such, the number of high-quality clinical scans is both exponentially higher than in research studies and covered by insurance or national health services, allowing for large-scale studies and novel study designs.

Box 1. 2017 Revision of the McDonald’s Criteria.

An overview of depression in MS

There are striking similarities between the pathophysiology and clinical course of MS and depression. Like MS-associated neurological symptoms, depression is hypothesized to be due to injured or abnormally functioning brain networks(64-69), and emerging studies have linked white matter lesions in MS to depression(15,16,42). Depression is also associated with inflammation, and it is well known that medications that boost inflammation (e.g., interferon) are associated with depression(70,71). Corticosteroids, used to treat inflammation in acute MS attacks, are also known to cause mood symptoms(72,73). With respect to course, depression, like MS, is often relapsing and progressive, with periods of stability interspersed with symptom worsening(74,75). The clinical course of both conditions provides many windows to examine MS – depression causality.

Beyond the biological and clinical parallels, neuropsychiatric literature has described relationships between MS and depression for over 150 years(76,77). Jean-Martin Charcot noted that pwMS had behavioral irregularities in 1877(76). In 1926, Cottrell and Wilson described “abnormal emotionality that is characteristic of disseminated sclerosis(77).” Depression can herald a future MS diagnosis, and studies have suggested that mood fluctuations may portend MS relapse(78,79). While modern epidemiologic studies of MS and depression have unequivocally identified a relationship between the two conditions(80,81), the underlying mechanisms are not fully understood.

Two primary constructs are commonly explored in MS-depression literature. The first suggests that depression in MS reflects a psychological reaction to the physical and emotional burden of having the condition. In this framework, depression is formulated as an “adjustment disorder” or as “demoralization(82,83).” The second construct conceptualizes depression as a consequence of brain disease and neuroinflammation, akin to post-stroke depression or “depression secondary to a medical condition(42,82-84).” While not covered in this review, others have explored whether there are shared genetic traits that contribute to vulnerability for both depression and MS, though these studies have been negative(85,86). Below, we will review evidence within each of these constructs.

As with other medical illnesses with high physical morbidity (e.g., cancer, inflammatory bowel diseases), the MS-depression psychological construct posits that the severity of depression is dictated by the severity and chronicity of the functional impairments from the underlying medical disease(83,87). Numerous studies linking depression to the experience of MS and poor quality of life have supported this relationship(82,83,87,88). Physical disability has been associated with depression severity(89,90) and is a moderator between depression and suicide(91,92). However, a subset of pwMS with severe neurological disease do not have depression(87,88), suggesting that depression might not simply be a consequence of the experience of having MS.

The discrepancy between physical disability and depression led some scientists to hypothesize that underlying brain disease was the primary driver of depression in pwMS(82,93). Given the widespread availability of MRI data in this population, numerous research groups designed studies to identify locations where depressed patients had more white matter lesions than non-depressed patients. For example, Pujol et al., 1997 contributed early findings showing that left arcuate fasciculus white matter lesions were associated with depression(94). In 2000, Bakshi et al. instead reported that lesions in the frontal and parietal lobes were associated with depression(95). Feinstein et al. then described associations between left medial inferior prefrontal cortex T2 lesions and depression(93). Others have attributed depression in MS to gray matter volume loss(96-98) or emotional processing dysfunction driven by abnormal functional connectivity between the ventrolateral prefrontal cortex and amygdala(99-101). However, two seminal reviews were unable to find a single location that, when lesioned in MS, reliably led to depressive symptoms(21,82). Given the negative findings of these reviews, the authors suggested that psychological mechanisms might be more relevant than lesion-based accounts.

As the broader neuroimaging community found success using LNM to map post-stroke neuropsychiatric symptoms to common functional networks(16,33,34), scientists revisited the hypothesis that MS brain injury causes neuropsychiatric symptoms, including depression. Expanding on stroke-based LNM, three cross-sectional studies evaluated associations between white matter lesions due to MS and psychopathology. Siddiqi, Kletenik, and colleagues showed that the fMRI signal in white matter lesions in depressed pwMS tended to be functionally connected to their 2021 depression network(84). In our previous work, we constructed the white matter backbone of Siddiqi et al.’s 2021 depression network (Figure 2) and developed a technique to explore white matter lesion-depression associations(16,42).We spatially normalized lesions to compare them to canonical white matter fascicles in template space. We then calculated the volume of streamlines that ran through the lesions to derive a measure of injury. MS lesions preferentially landed within the white matter depression network(42). Further, depressed pwMS were more likely to have lesions in the white matter depression network than outside it(42). In 2025, we showed that greater lesion burden to the uncinate fasciculus, the white matter backbone that connects key regions in anxiety circuitry (amygdala and orbitofrontal cortex), was associated with anxiety symptoms in MS(102). Others have similarly mapped white matter lesion burden in MS to cognitive deficits and physical disability(103-105).

Figure 2. White matter depression network construction.

A. Functional depression network from Siddiqi et al., 2021, Nature Human Behavior. B. White matter depression network. Reprinted from Biological Psychiatry, 95(12), Baller EB, Sweeney EM, Cieslak M, Robert-Fitzgerald T, Covitz SC, Martin ML, Schindler MK, Bar-Or A, Elahi A, Larsen BS, Manning AR, Markowitz CE, Perrone CM, Rautman V, Seitz MM, Detre JA, Fox MD, Shinohara RT, Satterthwaite TD, Mapping the Relationship of White Matter Lesions to Depression in Multiple Sclerosis, pp1072-1080, Copyright (2024), with permission from Elsevier (License No. 6071370804640)(42).

Importantly, risk factors for depression in non-MS patients are risk factors in MS depression. Poor social support and strained financial resources are associated with depression, with and without MS(106-108). Polygenic risk scores continue to predict depression in MS cohorts(86). Taken together, it is most likely that depression in MS is the product of multiple interacting factors, including brain pathology in both white and gray matter, psychological functioning, genetics, social support, and financial resources(107). However, to meaningfully tease apart whether a particular brain lesion is linked to depression, longitudinal studies that concurrently track lesion evolution with depression symptomatology are required.

Leveraging the clinical course of MS to better understand mechanisms of depression

The unique clinical course of MS creates numerous opportunities to examine associations between MS and depression. We will focus on focal inflammatory white matter lesions here (Box 2) and expand upon other pathobiology in the following section. As a first step, the electronic medical record (EMR) can be used to explore population-level relationships between depression and MS relatively cheaply and with large samples (Box 2A)(109). However, EMRs were designed primarily for billing, rather than research(110). Diagnoses can be inflated to increase reimbursement while essential data (i.e., EDSS) may be missing(111). Conversely, psychiatric symptoms are known to be underdiagnosed and undertreated in the general medical setting(112). As such, they are not always accurately represented in the EMR and require multi-step decision frameworks to derive meaningful phenotypes(19,42,102,113). Research from the EMR is most appropriate for identifying associations between MS lesions and lifetime psychopathology – but such data cannot characterize causality.

Box 2. Summary table of depression study approaches leveraging the clinical course of multiple sclerosis.

A) Electronic Medical Record; B) Prospective, Cross-Sectional; C) Longitudinal: Remission Focused. Blue lines represent the first MS episode; D) Longitudinal: Relapse-focused; E) Longitudinal: Ecological Momentary Assessment. For all plots, the x-axis represents the years since MS diagnosis, the y-axis represents MS symptom burden (e.g., physical, functional, cognitive), the yellow circle represents a clinical MRI scan, and the purple segment represents the time window over which depressive symptoms are captured.

Prospective, cross-sectional studies that concurrently assess depressive symptoms and brain pathology can overcome the limitations in EMR psychiatric phenotyping but require personnel to administer assessments (Box 2B). Importantly, depression can be evaluated concurrently with physical and neurological symptom burden, fatigue, and cognition, allowing researchers to model the degree to which functional limitations are driving depression. Ongoing protocols have demonstrated the feasibility of assessing individuals by conducting assessments virtually, in-person, and with mobile technology(43,114). While these studies are well-suited to characterize associations between active symptoms and brain lesions at a point in time, they cannot assess causal relationships between lesion burden and depression and are not anchored within the MS course.

While a truly causal study would require selectively lesioning a white matter pathway and testing for the emergence of depression, this is not possible in human research. However, leveraging the clinical course of MS with longitudinal models would provide additional information regarding causality by linking new-onset MS lesions to the emergence and resolution of depression. At the time of MS diagnosis, a patient usually presents to care because of a new neurological symptom (e.g., visual disturbance, weakness), which prompts an MRI and subsequent diagnosis(63,115). Contrast-enhancing lesions during an acute flare produce their characteristic appearance on MRI because of blood-brain-barrier disruption, allowing contrast to pass into inflamed tissue(45). Within the lesion, there is local edema, demyelination, and axonal injury. Neurological symptoms usually improve in the following weeks to months due to resolving edema and repair mechanisms within the lesion(116). These changes can be observed during repeat clinical imaging 3-6 months after initial diagnosis and the initiation of DMT(62). Assessing depression symptoms at diagnosis and at repeated imaging enables testing whether depression symptom improvement is associated with resolving white matter lesion edema (Box 2C). If repeat imaging is performed without contrast, any new lesion not present at the initial scan can be assumed to be a new lesion and mapped accordingly.

Given the psychological impact of receiving the MS diagnosis and beginning treatment, disentangling the resolution of depression versus psychological adjustment to the disorder is critical(117). For example, it is advisable to evaluate associations between lesion burden and the resolution of core symptoms of depression (e.g., anhedonia, guilt, suicidality) while accounting for adjustment disorder symptoms as assessed using validated assessments(118).

MS is a chronic illness, and while DMTs limit the amount of new white matter lesions that are formed, they do not affect chronic lesions that predate treatment initiation(119,120). Additionally, a subset of the chronic active lesions described above in the MS pathology section, termed slowly expanding lesions (SELs), can slowly enlarge/expand despite DMT treatment(121). As such, SELs can be identified on longitudinal MRI and tracked over time(122). Assessing depression at each imaging time point and testing whether worsening depression is associated with increases in SEL size, number, or burden could enhance confidence in the lesion – depression relationship (Box 2D). Finally, new lesions still form in people who cannot tolerate, do not respond to, or are nonadherent with DMTs. Using ecological momentary assessment with daily depression monitoring and associating it with new lesion development would also strengthen the causal hypothesis (Box 2E)(114,123). Though the lag between a change in depressive symptoms and the proximal MRI still poses a challenge in assessing causality, identifying associations may contribute to the development of protocols where an acute depression worsening prompts a clinical MRI.

As previously mentioned, an ongoing challenge across depression research in any medical population is disentangling the relationship between depression caused by the disease pathophysiology from the psychological impact of living with the disease. This is particularly challenging because hallmark depression symptoms, including poor sleep, poor energy, concentration difficulties, and appetite changes, are nonspecific and accompany many medical illnesses, including MS. Importantly, several MS-validated self-report measures are available to assess depression (while excluding somatic symptoms) as well as physical functioning(124-126). Designing studies that disentangle the association between brain network injury and depression while accounting for physical disability would not only increase our understanding of depression in MS but also significantly enrich the biopsychosocial formulation of depression in all contexts.

Frontiers in depression research that leverages MS pathobiology

In this review, we focused on using MS as a model to better understand depression in the context of the emergence and resolution of acute inflammatory lesions. However, MS is also associated with gray matter injury and metabolic dysfunction and thus provides additional opportunities to understand brain disease-depression relationships. Below, we will describe areas of potential interest.

With respect to gray matter injury, two common pathological processes in MS may be relevant for depression. First, gray matter atrophy is a well-known pathological feature of MS that has been associated with depression and should always be assessed and integrated into studies of white matter disease(96,97). It can be detected at the first presentation of MS and worsens over time(127,128). As gray matter changes have also been associated with both the emergence of geriatric depression and cognitive decline, longitudinal MS models could track this relationship(129,130). Second, focal demyelinating lesions can be found within the cortical gray matter, although they are smaller and more difficult to detect on clinical 3T MRI sequences(131). Importantly, increased cortical lesion burden has also been associated with worse cognitive performance(132,133). Moving forward, it is worth evaluating whether gray matter demyelinating lesions could also contribute to depression using ultra-high field MRI.

The MS field has recently recognized the clinical importance of paramagnetic rim lesions (PRLs) in progressive biology. PRLs are a subtype of chronically inflamed focal MS lesions, like SEL, but their assessment requires the acquisition of susceptibility-weighted MR sequences that are not currently part of routine MS care to capture the accumulation of iron-laden microglia and macrophages around the lesion edges(134). Higher PRL burden is strongly associated with worse cognitive and motor performance, and the chronic inflammatory component is thought to drive local brain atrophy(134-136). As such, PRLs may provide another depression model to study network disruption over time.

To meaningfully leverage MS as a use-case to study depression, it is vital to acknowledge that lesion impact is not uniform. The clinical course of a lesion involves both injury and repair, but these processes are poorly represented on clinical MRI sequences(137,138). Recent evidence also suggests that non-lesional, normal appearing white matter, on T2 FLAIR sequences may also have pathological properties(139,140).An exciting new research area in MS utilizes advanced MRI sequences to provide quantitative information, including the severity of tissue injury(141) as well as its metabolic milieu, which may be associated with tissue damage and repair(142). For example, ultra-high field MRI allows for imaging of low concentration molecules, including glutamate, with magnetic resonance spectroscopy (MRS) and glutamate chemical exchange saturation transfer (GluCEST)(143,144). Previous MS research has demonstrated that elevated GluCEST signal in the prefrontal cortex, an important part of the depression circuit, was associated with accumulated physical disability and poorer cognition(145). Combining high-resolution lesion burden quantification with metabolic measurements may help identify molecular mechanisms linking MS pathobiology with depression.

Though there has been convergence in psychiatry that there is a common functional brain network associated with depression, MS offers a natural experiment for exploring heterogeneity in depression (and general psychopathology), given the variability in lesion location, course, brain pathology, and associated symptom profiles. Studies that use dysconnectome features in combination with data-driven or machine-learning driven techniques to parse heterogeneity, and assess whether subtypes predict symptom burden, may be y informative(104). Insights gained from studying depression in MS should also be validated in other brain disorders.

Studying MS as a use-case for depression has the potential to inform our understanding of mechanisms of depression pathogenesis and treatment in, as well as outside of, MS. For MS specifically, the presence of new or evolving lesions in the depression network may prompt a referral to psychiatry for early intervention. As there is heterogeneity in response to standard antidepressants in MS, it would be worth exploring whether antidepressant efficacy in MS is dependent on the degree of white matter damage in the depression network. Therapeutic agents that enhance remyelination could also be studied as potential antidepressants. In depression outside of MS, confirming the importance of these white matter networks could inform personalized interventional psychiatric approaches to depression treatment, including targeted transcranial magnetic stimulation or deep brain stimulation. Additionally, findings from MS could inform studies that aim to link normative variation in white matter (measured with diffusion MRI) to dimensional symptoms of depression.

Conclusions

We have highlighted the potential impact of leveraging MS as a model to advance our neurobiological understanding of depression. MS provides a natural experiment where acute and chronic disease processes can be observed over time and correlated with symptoms of depression. Its clinical course and high depression comorbidity, combined with the availability of routinely acquired research-grade imaging, creates unparalleled opportunities for mechanistic research. Future work integrating prospective, longitudinal assessments of psychopathology with lesion network mapping techniques and advanced modeling tools in MS has significant potential to provide a mechanistic framework for understanding the network-level pathophysiology of depression.