Изучение настроения и сознания у людей с множественным склерозом: перспективный протокол перекрестного исследования.

STRENGTHS AND LIMITATIONS OF THIS STUDY

We will leverage research-grade 3-tesla (3T) MRIs acquired as part of routine MS care and maximise scalability by using the Method for Inter-Modal Segmentation Analysis for automated white matter lesion segmentation.

Our study will include participants with medical comorbidities, creating a more representative population and more broadly applicable results.

We will obtain detailed clinical and cognitive assessments from each participant to evaluate the inter-relationship of mood symptoms, anxiety symptoms, and cognitive deficits and relate them to white matter disease.

The resolution of 3T MRI scans limits our ability to evaluate the ongoing biological processes present within white matter lesions, including myelin repair and chronic inflammation.

This is a single-centre study.

Introduction

Multiple sclerosis (MS) is an immune-mediated neurological disorder that affects one million people in the USA. 1 2 It is characterised by demyelinating lesions in the brain and spinal cord due to the immune-mediated attack of oligodendrocytes. 3 , 5 Depression is highly comorbid with MS, with up to 50% of people with MS reporting depression. 6 7 Depression comorbidity in MS contributes to excess morbidity and mortality; it is associated with suicide rates double that of the general population. 8 , 10 Rates of depression in MS are higher than depression comorbidity in other chronic autoimmune diseases, suggesting that the neural pathophysiology of MS may specifically confer increased depression risk. 11 Cognitive dysfunction is also frequently observed in this population, affecting approximately 40–60% of individuals with MS. 6 12 13 Notably, higher levels of depressive symptoms are also associated with more severe cognitive deficits. 14 , 16 Despite the clear overlap between the conditions, the underlying mechanisms driving this association remain unclear. 6 7 The scientific premise of this study is that investigating the relationship between white matter lesions and depression in MS may provide a way to understand depression in MS as well as the pathophysiology of depression more broadly.

MS is marked by substantial heterogeneity in its clinical presentation and underlying pathology, with many studies highlighting the contributions of white matter injury to neurological disability and disease progression. 17 , 20 Previous studies of idiopathic depression in healthy persons have described associations between white matter abnormalities and depressive symptoms in numerous cortical and subcortical white matter fibre bundles, though the directionality and strength of associations have been inconsistent. 21 22 However, research in depression consistently excludes participants with intracranial pathology, and thus results from these studies cannot be extrapolated to people living with MS. 21 , 23 Techniques that relate heterogeneous white matter disease to rigorously characterised depression symptoms are vital to fill in this critical knowledge gap.

In contrast to white matter disease, more work has been done relating heterogeneous grey matter disease to the emergence of depression. Abnormalities in grey matter, particularly in circuits involving the limbic system, anterior cingulate, and prefrontal regions, have been implicated in the pathophysiology of depression. These circuits are involved in emotional processing, regulation, and decision-making, and their dysfunction has been linked to depressive symptoms. 24 However, meta-analyses have not been able to identify a single region that, when injured, reliably causes depression. 25 To address this, Fox and colleagues developed lesion network mapping (LNM), 26 which has shown that strokes in different areas of the brain that cause the same clinical symptoms are part of connected functional networks. 27 , 29 In recent work, they extended this paradigm to depression. 30 , 32 The existence of this ‘depression network’ relies on the assumption that network hubs exist on the backbone of structural connectivity. Two recent retrospective studies have indeed identified associations between white matter lesions in tracts that connect hubs of the depression network and depression symptoms. 33 34 Prospective studies that test for the presence of these relationships would be informative.

The purpose of this study is to delineate the relationship between mood and cognition to white matter brain lesions in MS by performing structured assessments of persons undergoing 3T scanning as part of their clinical care. It will also shed light on the contributions of anxiety, fatigue, and physical functioning to depression in MS. Together, this project will provide critical knowledge regarding how brain network dysfunction contributes to depression, learning from the example of white matter lesions in MS.

Aim

This study aims to prospectively evaluate white matter disease as a predictor of depressive symptoms in adults with MS. 250 participants will be recruited for our prospective cross-sectional sample from the University of Pennsylvania Comprehensive MS Center. Participants will be assessed with structured instruments to measure the severity of depressive symptoms, anxiety symptoms, lifetime psychopathology and cognitive functioning. This study will assess the relationships among individual variation in white matter lesions and depressive symptoms, while accounting for cognitive functioning, and comorbid psychopathology. Measuring depression while accounting for cognition and physical functioning may help clarify whether MS-related depression represents a unique clinical profile or aligns with typical presentations of depression in non-MS populations.

Hypothesis

In a prospective sample, individual variation in depressive symptoms in participants with MS will be associated with white matter lesion burden affecting tracts connecting brain regions in known depression networks.

Methods and analysis

Overall study design

This is a prospective study that will include data collection. Over the span of the 5-year recruitment window (1 July 2023–30 June 2028), we will enroll 250 participants diagnosed with MS who are receiving neuroimaging under the University of Pennsylvania MS 3T clinical MRI protocol for an in-person or virtual research visit ( figure 1 ). Data analyses will take place after the recruitment phase is completed. Participants will undergo a comprehensive neuropsychiatric assessment to assess current and lifetime psychiatric symptoms, as well as cognitive functioning.

Participant flow chart. Participants will be screened in MS clinics and then verified for eligibility using the inclusion and exclusion criteria. During the visit, participants will be asked questions about their demographics, answer mood and anxiety questionnaires, and complete the Computerized Neurocognitive Battery (CNB) and the CAT-GOASSESS. MS, multiple sclerosis; CAT-GOASSESS, computerised adaptive version of the GOASSESS.

Symptom ratings

Participants will complete several symptom rating scales that will measure current levels of depression and anxiety ( table 1 ). This will include: Patient Health Questionnaire Adult (PHQ-9 Adult), 35 Beck Depression Inventory-II, 36 Patient Reported Outcomes Measurement Information System (PROMIS) Emotional Distress Depression Short (Adult), 37 , 39 Generalized Anxiety Disorder, 40 Beck Anxiety Inventory, 41 PROMIS Emotional Distress Anxiety Short (Adult) 37 , 39 and the Hospital Anxiety and Depression Scale (HADS). 42 Furthermore, to measure levels of fatigue and physical functioning, participants will complete the PROMIS Short Form v1.0–Fatigue–Multiple Sclerosis 8a 43 and the Patient Reported Outcomes Measurement Information System Neuro-QOL (PROMISnq) Short Form V.2.0–Physical Function–Multiple Sclerosis 15a. 44 The MS-specific PROMIS measures, which were developed collaboratively by both MS patients and neurologists, were selected to provide a comprehensive and patient-centred assessment of physical functioning and have been shown to correlate well with the Expanded Disability Status Scale (EDSS). Unlike the EDSS, which focuses primarily on ambulatory function, the PROMISnq 15a captures a broader range of symptoms that can significantly impact daily life. 45

Table 1 Participant reported outcomes assessed Scale administered Outcome assessed Patient Health Questionnaire Adult Depression Beck Depression Inventory II Depression PROMIS Emotional Distress Depression Short (Adult) Depression Generalized Anxiety Disorder Anxiety Beck Anxiety Inventory Anxiety PROMIS Emotional Distress Anxiety Short (Adult) Anxiety Hospital Anxiety and Depression Scale Anxiety and depression PROMIS V.1.0–Fatigue–Multiple Sclerosis 8a Fatigue PROMISnq V.2.0–Physical Function–Multiple Sclerosis 15a Physical function PROMIS, Patient Reported Outcomes Measurement Information System; PROMISnq, Patient Reported Outcomes Measurement Information System Neuro-QOL.

Participant reported outcomes assessed

PROMIS, Patient Reported Outcomes Measurement Information System; PROMISnq, Patient Reported Outcomes Measurement Information System Neuro-QOL.

Assessment of lifetime psychopathology

Participants will also complete the computerised adaptive version of the GOASSESS (CAT-GOASSESS), 46 a clinical assessment that provides a dimensional characterisation of lifetime psychopathology. Rather than generating clinical diagnoses, which tend to have overlapping symptoms, the CAT-GOASSESS produces transdiagnostic z-scores for five domains: mood and anxiety, phobias, externalising disorders, psychosis, and personality disorders ( table 2 ). 46 Of note, the psychosis domain reflects symptoms of both active and prodromal psychosis, with prodromal questions drawn from the Structured Interview for Prodromal Syndromes 47 and the Prevention through Risk Identification Management and Education screen. 48 Personality pathology screening questions were drawn from the Personality Inventory for DSM-5 – Short Form (PID-5-SF). 49

Table 2 Computerised adaptive version of the GOASSESS (CAT-GOASSESS) Domain Number of questions in item bank Average number of items administered Associated psychopathology Mood and anxiety 38 12 Major depressive episode, mania/hypomania, generalised anxiety, separation anxiety, social anxiety, panic, post-traumatic stress disorder Phobias 25 12 Agoraphobia, specific phobia Externalising disorders 25 12 Attention deficit/hyperactivity, oppositional defiant, conduct Psychosis 25 9 Psychosis, prodromal psychosis Personality 25 9 Negative affect, detachment, antagonism, disinhibition, psychoticism

Computerised adaptive version of the GOASSESS (CAT-GOASSESS)

Penn Computerized Neurocognitive Battery

Cognition will be assessed with a 1-hour Computerized Neurocognitive Battery (CNB) 50 using a version of the CNB that has been developed specifically for MS participants. It will assess 32 measures obtained from 17 neurocognitive tests of performance (15 for accuracy, 17 for speed). Domains include executive functioning, episodic memory, social cognition, complex reasoning, and sensorimotor speed. The MS CNB will include the Penn Conditional Exclusion Test, Penn Continuous Performance Task (Number Letter Version – Short), Letter-N-Back Test (2-Back Version), Penn Word Memory Test (plus delayed recall), Penn Face Memory Test (plus delayed recall), Visual Object Learning Test (Short Version+delayed recall), Penn Logical Reasoning Test (Short Version), Penn Matrix Analysis Test (Form A), Variable Short Penn Line Orientation Test (−15 V.2.00), Penn Emotion Recognition Test (40 Faces), Measured Emotion Differentiation Test (Form A), Psychomotor Vigilance Test (3-minute Version), Motor Praxis Test and the Short Penn Computerized Finger Tapping Test ( table 3 ).

Table 3 Computerized Neurocognitive Battery (CNB) Test Domain Accuracy score Speed score Conditional Exclusion Task Abstraction and mental flexibility Y Y Continuous Performance Task Attention Y Y N-Back Test Working memory Y Y Word Memory Test Verbal memory Y Y Word Memory Test - Delayed Version Verbal memory Y Y Face Memory Test Facial memory Y Y Face Memory Test - Delayed Version Facial memory Y Y Visual Object Learning Test Spatial memory Y Y Visual Object Learning Test - Delayed Version Spatial memory Y Y Logical Reasoning Test Verbal reasoning Y Y Matrix Analysis Test Nonverbal reasoning Y Y Line Orientation Test Spatial reasoning Y Y Emotion Recognition Test Emotion identification Y Y Measured Emotion Differentiation Test Emotion differentiation Y Y Psychomotor Vigilance Test Alertness and vigilance Y Y Motor Praxis Test Sensorimotor speed N Y Computerized Finger Tapping Test Motor speed N Y

Computerized Neurocognitive Battery (CNB)

The CNB was selected instead of the commonly used Symbol Digit Modalities Test (SDMT) due to its broad cognitive domain coverage. Additionally, it has been culturally adapted in Western and non-Western societies, translated into 12 languages to date, has been used to study adolescent development and healthy adult cognition in the human connectome project and to track cognition by the Army, Marines, and NASA. In a paediatric-onset MS population, it has also demonstrated improved detection of processing speed deficits as compared with the SDMT. 51 , 54

Medical record and MRI imaging data

Electronic medical record data will be obtained, with participant permission, via the University of Pennsylvania’s Data Analytic Center. We will obtain demographic data, International Classification of Diseases (ICD)−10 diagnoses, medication lists, and depression screens, including Patient Health Questionnaires (two question form, PHQ-2; and nine question form, PHQ-9). 35

3T MRI including 3D T1 magnetisation-prepared rapid acquisition gradient echo (TR (repetition time)=1.9 s, TE (echo time)=2.48 ms, FA=9°, acquisition time=4:18, 176 sagittal slices, resolution=1 mm 3 ) and 3D T2 FLAIR, fluid-attenuated inversion recovery (TR=5 s, TE=398 ms, FA (flip angle)=120°, acquisition time=5:02, 160 sagittal slices, resolution=1 mm 3 ) scans will be acquired as part of routine care and will be transferred directly from radiology to our server.

Eligibility and exclusion criteria

All participants will be (1) people aged 18–80 years, (2) able to provide signed informed consent, (3) proficient in English, (4) physically and cognitively able to participate in a cognitive assessment and complete questionnaires, (5) have an ICD-10 diagnosis of MS from one of the specialists in the University of Pennsylvania Comprehensive MS Center, and (6) receiving routine clinical 3T MRI under the University of Pennsylvania MS protocol from a University of Pennsylvania affiliated hospital.

Subjects will be excluded if (1) medical illness impedes participation, (2) active substance use presents the possibility of acute intoxication or withdrawal, (3) poor clinical image quality, or (4) history of a pervasive developmental disorder.

Outcome measures

Outcome measures will include volume of streamlines that intersect lesions at any point in their trajectory (‘injured streamlines’) within 77 canonical fascicles as well as disease burden within versus outside the white matter depression network (below). They will be related to the total scores from the symptom ratings, dimensional measures of psychopathology from the CAT-GOASSESS, and cognition from the CNB ( tables1 3 ).

Image processing and statistical analysis

Constructing the white matter depression network

In a series of studies using LNM, Fox et al showed that lesions in the brain that are functionally connected to hubs of the frontoparietal and dorsal attention systems reliably produced depressive symptoms and that neuromodulation of hubs in this network could alleviate depression. 30 , 32 As we have done previously, we will define the grey matter depression network as areas positively associated with depression. 30 33 To construct the white matter depression network, we previously identified fascicles that served as the structural backbone of a functional depression circuit map from Siddiqi et al . 30 We then constructed a binary map by applying a threshold of t >3.09 to identify voxels with a statistically significant positive association between depression symptoms and brain disease or stimulation. The white matter depression network was then created in DSI Studio by creating 77 canonical fascicles spanning cortical and subcortical regions from an atlas that was derived from a large sample of high-quality diffusion MRI. 31 32 We next calculated the volume (in voxels) that overlapped between the fascicle’s individual fibres, or streamlines, and the binarised functional depression network. Fascicles were ranked by their volume of overlap with the functional depression network. Fascicles with the highest degree of overlap (top 25%, 19 fascicles) were considered to be in the white matter depression network. The resulting white matter depression network included fascicles known to be commonly affected in MS, including the corpus callosum and thalamic projections. However, the superior longitudinal fasciculus and arcuate fasciculus, which support language and complex cognition, also comprised a substantial portion of the white matter network.

Lesion segmentation

The Method for Inter-Modal Segmentation Analysis (MIMoSA) will be used to perform automated lesion segmentation on the MRIs. 55 MIMoSA uses local covariance patterns across T1 and T2 FLAIR images within an individual subject to produce a probability map of potential white matter lesions. 55 These probability maps will be thresholded and binarised to produce a mask of lesions.

Streamline filtering

We will conduct analyses in template space using tools from DSI Studio, a tractography tool for mapping brain pathways and analysing tract-related metrics. 56 , 58 Specifically, we will project the lesions segmented with MIMoSA onto an expert-vetted, population-averaged atlas of the structural connectome derived from diffusion MRI data (n=842), as reported previously. 33 56 This template was achieved by creating a high-resolution template of diffusion patterns averaged across individual subjects and using tractography to generate 550,000 trajectories of representative white matter fascicles annotated by anatomical labels. The trajectories were subsequently clustered and labelled by a team of experienced neuroanatomists in order to conform to prior neuroanatomical knowledge. 56 For each individual, the volume of white matter tracts intersecting each lesion will be computed via streamline filtering in DSI Studio. 56 , 58 This will require identifying whether individual streamlines within a fascicle were impacted (ie, passed through a lesion) or spared (avoided a lesion). Delineating fascicles in a single diffusion MRI dataset is known to be error-prone, so we will compare spatially normalised lesions to 77 canonical fascicles in the template space. 59 Individual lesion maps will be normalised to the template space of the canonical fascicles (Montreal Neurological Institute 2009b Asymmetric template) using the T1-weighted-based transform calculated by antsRegistration. 60 , 62 Streamlines intersecting lesions at any point in their trajectory will be considered injured and isolated from the rest of the fascicle. The total volume (in voxels) occupied by injured streamlines will be calculated as the measure of disease burden in the fascicle.

Statistical analysis

Deriving psychopathology summary scores

We will obtain summary depression and anxiety scores in two possible ways. First, we will conduct exploratory factor analysis on all depression and anxiety scale scores. Based on substantial previous research, 63 64 we expect the two rotated factors to clearly represent (1) depression and (2) anxiety. Factor scores will be calculated from this model and used as the depression and anxiety summary scores. If the factor analysis does not produce a clearly separated depression/anxiety configuration, we will instead create mean composite scores for depression and anxiety. That is, all scale scores will be z-scaled and averaged within their domain to obtain summary scores.

Primary and secondary analyses

Next, for our primary analysis, we will evaluate the relationship of white matter lesion burden within and outside the depression network with depression symptomatology as summarised by the depression factor score. Our primary statistical model will use generalised additive models (R-package mgcv) 65 to evaluate depression severity as a function of lesion burden within the white matter depression network, covarying for age, sex, and physical functioning. Given that MS is a progressive illness that worsens with age, it is vital to account for age effects—which are often nonlinear over the lifespan. To do this, penalised splines will be used to flexibly model both linear and nonlinear ageing trajectories while avoiding overfitting by applying a penalty for increasing levels of nonlinearity, which is estimated from the data using restricted maximum likelihood.

Secondary analyses will include relating white matter lesion burden to measures of anxiety, cognition and lifetime psychopathology. Additional exploratory models will incorporate further covariates, including disease duration, anxiety severity and fatigue. We will also test for lesion burden by sex interactions. All analyses will account for multiple comparisons.

Reproducibility

Data will be managed by DataLad, a free and open-source distributed data management system that allows for the storage of large high-dimensional data and for data analysis to be tracked or undone, thus guaranteeing 100% reproducibility. 66 Furthermore, Curation of the Brain Imaging Data Structure (BIDS), 67 a Python-based software package that aids users in validating and managing the curation of neuroimaging datasets, will be used to identify heterogeneity in acquisition parameters and ensure all data are in valid BIDS format. All image processing and analysis will be containerised to ensure reproducibility. 67

Sample size power calculation

To define our sample size target, power calculations were conducted in PASS 68 assuming a 5% type I error rate and using two-sided hypothesis tests. Conservatively assuming 20% of subjects have insufficient data quality (final n=200), we expect to have 80% power to determine associations between depression scores and lesion burden in the depression network with small effect sizes of f 2 =0.04 or larger, adjusting for age, sex, and physical disability.

Patient and public involvement

Patients and the public were not involved in the design or conduct of this study. Individual results will be communicated to participants who wish to be notified.

Ethics and dissemination

Research ethics approval

Ethics approval has been obtained from the University of Pennsylvania Institutional Review Board.

Data access and dissemination

All processing pipelines used will be shared via a publicly available software repository on GitHub. As part of our efforts to maximise rigour and reproducibility, all analyses will use Jupyter or Rmarkdown notebooks, which will ensure that all interested parties will be able to fully reproduce all analyses. These publicly available notebooks include prose, code and results. In publication, presentation or data sharing resulting from this research, the participants will not be identified.

Footnotes

Funding: This work was supported by the National Institute of Mental Health Grant number K23MH133118, the Brain and Behavior Research Foundation NARSAD Grant number 31319, the National Institute of Mental Health Grant number R01MH112847, the National Institute of Mental Health Grant number R01MH123550, and the National Institute of Neurological Disorders and Stroke Grant number R01NS112274.

Prepublication history for this paper is available online. To view these files, please visit the journal online ( https://doi.org/10.1136/bmjopen-2024-094733 ).

Patient consent for publication: Not applicable.

Provenance and peer review: Not commissioned; externally peer reviewed.

Patient and public involvement: Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.

References