About

Matthew Baker is an Assistant Professor of Medicine in the Division of Immunology and Rheumatology at Stanford University. He is the Clinical Chief in his division and has established a clinical research program focused on clinical trials, epidemiological studies, and bench-to-bedside translational research. His research emphasizes sarcoidosis, IgG4-related disease, and rheumatoid arthritis, and he has designed and led investigator-initiated and industry-sponsored clinical trials in these areas. Dr. Baker is also the Co-Director of the Stanford Multidisciplinary Sarcoidosis Program, collaborating with team members to advance clinical care and research related to sarcoidosis. He received his bachelor's degree from Pomona College, his medical degree from Harvard Medical School, and his master's degree in Epidemiology and Clinical Research from Stanford University. His training includes completing an Internal Medicine residency at Massachusetts General Hospital and a Rheumatology fellowship at Stanford University. Dr. Baker is board-certified in Internal Medicine and Rheumatology by the American Board of Internal Medicine. His educational background also includes a fellowship in Rheumatology and Immunology at Stanford University, which he completed in 2016.

Research topics

• Medicine
• Internal medicine
• Neuroscience
• Pathology
• Physical therapy
• Biology
• Oncology
• Immunology
• Genetics
• Psychology

Selected publications

• Artificial Intelligence–Aided Lung Ultrasound Detection of Interstitial Lung Disease in Systemic Sclerosis and Inflammatory Myopathy

  Arthritis Care & Research · 2026-04-06

  article

  OBJECTIVE: Lung ultrasound (LUS) is a sensitive, low-cost, and radiation-free modality for interstitial lung disease (ILD) detection. We previously developed and validated LUS interpretation criteria in systemic sclerosis (SSc) and idiopathic inflammatory myopathy (IIM) showing excellent diagnostic performance and correlations with ILD severity. In this study, we applied deep learning to evaluate whether convolutional neural networks (CNNs) can accurately detect ILD and its severity on LUS. METHODS: Patients with SSc or IIM ± ILD and paired LUS and chest computed tomography (CT) were included. LUS images were labeled using CT results and human LUS-ILD 2024 (LUS-ILD-24) interpretation. Three pretrained CNN architectures (InceptionV3, ResNet-50, VGG-16) were fine-tuned via transfer learning, and a de novo lightweight architecture (LUS-Net) was developed. Model performance for ILD detection was assessed at image and patient levels using area under the curve (AUC), sensitivity, specificity, and agreement with expert interpretation. CNN outputs were correlated with pulmonary function tests (PFTs) and CT-based Computer-Aided Lung Informatics for Pathology Evaluation and Rating indices. Gradient-weighted Class Activation Mapping (Grad-CAM) visualized regions driving predictions. RESULTS: A total of 140 patients representing 3,920 LUS images were included and split into development (74) and independent test sets (66). VGG-16 achieved the best patient-level performance (AUC 0.972, sensitivity 97.4%, specificity 92.6%) showing strong correlations with PFTs and CT severity. Grad-CAM highlighted pleural features as the primary regions influencing model predictions. CNN performance matched or exceeded LUS-ILD-24 interpretation. CONCLUSION: Deep learning applied to LUS enables accurate ILD detection in connective tissue disease and can enhance expert interpretation. Explainable artificial intelligence suggests pleural features, even when B-lines are infrequent, are sufficient for reliable ILD recognition.

  PublisherDOI

• High Burden and Phenotype-Specific Variability of Pain Polypharmacy in Early Autoimmune Rheumatic Diseases: A 15-Year Real-World Analysis

  Research Square · 2026-04-01

  preprintOpen access
  PublisherDOI

• Phenotypes of Chronic Overlapping Pain Multimorbidity in Rheumatology: A Population‐Based Claims Study

  Arthritis Care & Research · 2026-05-18

  article

  OBJECTIVE: This study aims to characterize the burden and distinct patterns of chronic overlapping pain conditions (COPCs) multimorbidity in adults with autoimmune rheumatic diseases (ARDs). METHODS: We analyzed 2008-2021 data from the Merative MarketScan Commercial Claims and Encounters Database, identifying 149,742 patients with ARD (ankylosing spondylitis, psoriatic arthritis, rheumatoid arthritis, Sjögren's disease, systemic lupus erythematosus, systemic sclerosis) using ICD-9/10-CM codes. We assessed COPC (e.g., fibromyalgia, chronic low back pain, migraine, irritable bowel syndrome, chronic fatigue syndrome, temporomandibular disorders, and pelvic pain conditions) prevalence, multimorbidity patterns, and pairwise co-occurrence frequencies, calculating observed/expected ratios and odds ratios. RESULTS: Of 149,742 ARD patients, 57.5% had ≥1 COPC, with 22.7% experiencing multiple conditions. Chronic low back pain (41.4%) and fibromyalgia (21.1%) were most prevalent. Women and individuals aged 31-50 showed the highest COPC prevalence. Many COPC pairs co-occurred more frequently than expected by chance (e.g., fibromyalgia and chronic low back pain: observed/expected ratio 1.35; migraine and chronic low back pain: 1.42). Strongly associated dyads included urologic chronic pelvic pain and vulvodynia (OR 10.46). CONCLUSION: A substantial burden and distinct multimorbidity patterns of COPCs exist among ARD patients, suggesting shared pathophysiological mechanisms. Routine COPC screening and targeted non-opioid multimodal interventions, especially for common dyads, are crucial for improved pain management and reducing opioid-related risks in this population.

  PublisherDOI

• Expanding the spectrum of annexin A11 proteinopathy in frontotemporal lobar degeneration and motor neuron disease

  Acta Neuropathologica · 2025-11-12 · 1 citations

  articleOpen access

  Aggregation of TAR-DNA-binding protein 43 (TDP-43) is strongly associated with frontotemporal lobar degeneration (FTLD-TDP), motor neuron disease (MND-TDP), and overlap disorders like FTLD-MND. Three major forms of motor neuron disease are recognized and include primary lateral sclerosis (PLS), amyotrophic lateral sclerosis (ALS), and progressive muscular atrophy (PMA). Annexin A11 (ANXA11) is understood to aggregate in amyotrophic lateral sclerosis (ALS-TDP) associated with pathogenic variants in ANXA11, as well as in FTLD-TDP type C. Given these observations and recent reports of ANXA11 variants in patients with semantic variant frontotemporal dementia (svFTD) and FTD-MND presentations, we sought to characterize ANXA11 proteinopathy in an autopsy cohort of 379 cases diagnosed with a primary TDP-43 proteinopathy, including FTLD-TDP, FTLD-MND, and MND-TDP. Cases with FTLD-MND and MND-TDP were classified further into PLS, ALS, and PMA based on the relative loss of upper and lower motor neurons. ANXA11 proteinopathy was present in over 40% of FTLD-MND cases. Further, ANXA11 colocalized with TDP-43 in the pathologic inclusions of all FTLD-TDP type C cases, as well as 38 out of 40 FTLD-PLS cases (95%), of which 84% had TDP type B or an unclassifiable TDP-43 proteinopathy and 16% had TDP type C. Genetic analysis excluded pathogenic ANXA11 variants in all ANXA11-positive cases. We thus demonstrated two novel ANXA11 proteinopathies strongly associated with FTLD-PLS, but not with TDP type C or pathogenic ANXA11 variants. Given the emerging relationship between TDP-43 and ANXA11 in neurodegenerative disease, we propose that TDP-43 and ANXA11 proteinopathy (TAP) comprises a distinct group of molecular pathologies and define three TAP types based on key clinical and neuropathologic characteristics.

  PublisherOA PDFDOI

• Brain transcriptomics highlight abundant gene expression and splicing alterations in non-neuronal cells in aFTLD-U

  Acta Neuropathologica · 2025-08-10 · 2 citations

  articleOpen access

  Atypical frontotemporal lobar degeneration with ubiquitin-positive inclusions (aFTLD-U) is a rare cause of frontotemporal lobar degeneration (FTLD), characterized postmortem by neuronal inclusions of the FET family of proteins (FTLD-FET). The recent discovery of TAF15 amyloid filaments in aFTLD-U brains represents a significant step toward improved diagnostic and therapeutic strategies. However, our understanding of the etiology of this FTLD subtype remains limited, which severely hampers translational research efforts. To explore the transcriptomic changes in aFTLD-U, we performed bulk RNA sequencing on the frontal cortex tissue of 21 aFTLD-U patients and 20 control individuals. Cell-type deconvolution revealed loss of excitatory neurons and a higher proportion of astrocytes in aFTLD-U relative to controls. Differential gene expression and co-expression network analysis, adjusted for the shift in cell-type proportions, showed dysregulation of mitochondrial pathways, transcriptional regulators, and upregulation of the Sonic hedgehog (Shh) pathway, including the GLI1 transcription factor, in aFTLD-U. Overall, oligodendrocyte and astrocyte-enriched genes were significantly over-represented among the differentially expressed genes. Differential splicing analysis confirmed the dysregulation of non-neuronal cell types with significant splicing alterations, particularly in oligodendrocyte-enriched genes, including myelin basic protein (MBP), a crucial component of myelin. Immunohistochemistry in frontal cortex brain tissue also showed reduced myelin levels in aFTLD-U patients compared to controls. Together, these findings highlight a central role for glial cells, particularly astrocytes and oligodendrocytes, in the pathogenesis of aFTLD-U, with disruptions in mitochondrial activity, RNA metabolism, Shh signaling, and myelination as possible disease mechanisms. This study offers the first transcriptomic insight into aFTLD-U and presents new avenues for research into FTLD-FET.

  PublisherOA PDFDOI

• Epstein-Barr virus reprograms autoreactive B cells as antigen-presenting cells in systemic lupus erythematosus

  Science Translational Medicine · 2025-11-12 · 42 citations

  articleOpen access

  Systemic lupus erythematosus (SLE) is a systemic autoimmune disease characterized by antinuclear antibodies (ANAs). Epstein-Barr virus (EBV) infection has been epidemiologically associated with SLE, yet its role in pathogenesis remains incompletely defined. Here, we developed an EBV-specific single-cell RNA-sequencing platform and used it to demonstrate that EBV infection reprograms autoreactive antinuclear antigen B cells to drive autoimmunity in SLE. We demonstrated that, in SLE, EBV + B cells are predominantly CD27 + CD21 low memory B cells that are present at increased frequencies and express ZEB2 , TBX21 (T-bet), and antigen-presenting cell transcriptional pathways. Integrative analysis of chromatin immunoprecipitation sequencing (ChIP-seq), assay for transposase-accessible chromatin sequencing (ATAC-seq), and RNA polymerase II occupancy data revealed EBV nuclear antigen 2 (EBNA2) binding at the transcriptional start sites and regulatory regions of CD27 , ZEB2 , and TBX21 , as well as the antigen-presenting cell genes demonstrated to be up-regulated in SLE EBV + B cells. We expressed recombinant antibodies from SLE EBV + B cells and demonstrated that they bind prototypical SLE nuclear autoantigens, whereas those from healthy individuals do not. We further found that SLE EBV + B cells can serve as antigen-presenting cells to drive activation of T peripheral helper cells with concomitant activation of related EBV − antinuclear double-negative 2 B cells and plasmablasts. Our results provide a mechanistic basis for EBV being a driver of SLE through infecting and reprogramming nuclear antigen-reactive B cells to become activated antigen-presenting cells with the potential to promote systemic disease–driving autoimmune responses.

  PublisherOA PDFDOI

• Vagotomy and the incidence of rheumatoid arthritis and osteoarthritis: a Danish register-based study

  Arthritis Research & Therapy · 2025-05-16 · 3 citations

  articleOpen access1st authorCorresponding

  OBJECTIVES: Given the potential role of vagus nerve stimulation in treating rheumatoid arthritis (RA), we examined the incidence of RA and osteoarthritis (OA) in patients who underwent different forms of vagotomy that disparately affect the inflammatory reflex. METHODS: Using nationwide health registries, we constructed cohorts of patients in Denmark who underwent truncal or superselective vagotomy between 1977 and 1995 and comparison members from the general population matched 10:1 on birth year, sex, and calendar year. We identified incident RA or OA and used Cox proportional hazards models to compute adjusted hazard ratios (aHRs) and corresponding 95% CI. RESULTS: Our cohorts consisted of 2,260 truncal vagotomy patients matched with 22,610 comparators, and 3,810 superselective vagotomy patients matched with 38,090 comparators. The incidence rate (IR) of RA per 1,000 person-years (95% CI) in the truncal vagotomy cohort was 10.2 (6.5-15.3) versus 7.2 (6.1-8.4) in the matched comparison cohort. The aHR (95% CI) for RA development was 2.62 (1.47-4.67) in the truncal vagotomy cohort and 1.05 (0.51-2.17) in the superselective vagotomy cohort, with respect to comparison cohorts. The risk of developing OA was not significantly different for either vagotomy cohort compared with comparison cohorts. CONCLUSION: Truncal vagotomy was associated with an increased incidence of RA; this association was not observed with superselective vagotomy. No association with either form of vagotomy was seen with OA. These findings support the hypothesis that disruption of vagus nerve signaling impacts the inflammatory reflex and contributes to the development of RA.

  PublisherOA PDFDOI

• Clinical Practice Guideline for Evaluation and Management of Peripheral Nervous System Manifestations in Sjögren's Disease

  Arthritis Care & Research · 2025-12-02 · 1 citations

  articleOpen access

  OBJECTIVE: Sjögren's disease is an autoimmune disorder that can impact multiple organ systems, including the peripheral nervous system (PNS). PNS manifestations, which can exist concurrently, include mononeuropathies, polyneuropathies, and autonomic nervous system neuropathies. To help patients and providers in the decision-making process, we developed an evidence-based clinical practice guideline for the evaluation and management of PNS manifestations in patients with Sjögren's disease. METHODS: A Topic Review Group, comprising experts in rheumatology, neurology, and guideline methodology, developed Patient, Intervention, Comparison, and Outcome questions and conducted a systematic review to identify the current best evidence on management of PNS manifestations of Sjögren's disease. PubMed and Embase were searched for evidence published up to July 22, 2025. Literature screening, data extraction, and critical appraisal were performed in duplicate. Six case series, one retrospective cohort, and two prospective cohort studies lacking a comparison group met the inclusion criteria. RESULTS: We developed an aligned nomenclature of PNS terms that can be used across disciplines, 31 good practices for evaluation of suspected PNS manifestations, and 20 evidence-based treatment recommendations, the latter of which were rated as conditional or strong based on the Grading of Recommendations Assessment, Development, and Evaluation methodology. Because of the scarcity of high-level evidence, this guideline predominantly derives from expert opinion. CONCLUSION: This clinical practice guideline on PNS manifestations of Sjögren's disease provides clinicians a rigorous, evidence-based resource, developed through an expert consensus-based process, for the assessment, diagnosis, and treatment of peripheral neuropathy in patients with Sjögren's disease. Recommendations were rated as strong when the benefits significantly outweighed potential harms, creating a scenario in which most patients would prefer the advised action.

  PublisherOA PDFDOI

• Analysis of the splicing landscape of the frontal cortex in FTLD-TDP reveals subtype specific patterns and cryptic splicing

  Acta Neuropathologica · 2025-06-06 · 5 citations

  articleOpen access

  Dysregulation of TDP-43 as seen in TDP-43 proteinopathies leads to specific RNA splicing dysfunction. While discovery studies have explored novel TDP-43-driven splicing events in induced pluripotent stem cell (iPSC)-derived neurons and TDP-43 negative neuronal nuclei, transcriptome-wide investigations in frontotemporal lobar degeneration with TDP-43 aggregates (FTLD-TDP) brains remain unexplored. Such studies hold promise for identifying widespread novel and relevant splicing alterations in FTLD-TDP patient brains. We conducted the largest differential splicing analysis (DSA) using bulk short-read RNAseq data from frontal cortex (FCX) tissue of 127 FTLD-TDP (A, B, C, GRN and C9orf72 carriers) and 22 control subjects (Mayo Clinic Brain Bank), using Leafcutter. In addition, long-read bulk cDNA sequencing data were generated from FCX of 9 FTLD-TDP and 7 controls and human TARDBP wildtype and knock-down iPSC-derived neurons. Publicly available RNAseq data (MayoRNAseq, MSBB and ROSMAP studies) from Alzheimer's disease patients (AD) was also analyzed. Our DSA revealed extensive splicing alterations in FTLD-TDP patients with 1881 differentially spliced events, in 892 unique genes. When evaluating differences between FTLD-TDP subtypes, we found that C9orf72 repeat expansion carriers carried the most splicing alterations after accounting for differences in cell-type proportions. Focusing on cryptic splicing events, we identified STMN2 and ARHGAP32 as genes with the most abundant and differentially expressed cryptic exons between FTLD-TDP patients and controls in the brain, and we uncovered a set of 17 cryptic events consistently observed across studies, highlighting their potential relevance as biomarkers for TDP-43 proteinopathies. We also identified 16 cryptic events shared between FTLD-TDP and AD brains, suggesting potential common splicing dysregulation pathways in neurodegenerative diseases. Overall, this study provides a comprehensive map of splicing alterations in FTLD-TDP brains, revealing subtype-specific differences and identifying promising candidates for biomarker development and potential common pathogenic mechanisms between FTLD-TDP and AD.

  PublisherOA PDFDOI

• Validation of Lung Ultrasound Interpretation Criteria for Interstitial Lung Disease in Systemic Sclerosis and Inflammatory Myopathy

  Arthritis Care & Research · 2025-05-05 · 5 citations

  articleOpen access

  OBJECTIVE: Interstitial lung disease (ILD) has a high prevalence in patients with systemic sclerosis (SSc) and inflammatory myopathy (IM), and early identification reduces associated morbidity and mortality. We previously developed lung ultrasound (LUS) interpretation criteria for ILD detection in 2020 (LUS-ILD-20) showing excellent sensitivity and specificity in patients with SSc-ILD; herein, we sought to validate a revised LUS-ILD-24 in a large cohort with SSc and IM. METHODS: Patients meeting criteria for SSc and IM, with planned computed tomography (CT) chest imaging underwent LUS imaging interpreted with LUS-ILD-24 by three blinded readers. The sensitivity and specificity for LUS-ILD detection as noted on CT was analyzed for subgroups with SSc, IM, and possible incident ILD. Inter- and intrarater agreements were calculated. Correlations between LUS-ILD-24 severity, CT imaging severity, and pulmonary function tests were assessed. RESULTS: Ninety-five patients were included in the analyses. Sensitivity and specificity for ILD detection ranged from 92.4% to 95.5% and 82.8% to 86.2% across readers with similar accuracy in all subgroups. Inter- and intrareader reliability showed near perfect agreement (κ = 0.92 and κ = 0.90 to 1, respectively). LUS severity correlated with CT imaging severity and inversely correlated with diffusion capacity for carbon monoxide and forced vital capacity. CONCLUSION: We validated our revised LUS-ILD-24 in cohorts with SSc and IM and found excellent sensitivity, specificity, and reliability for detection of ILD identified on CT imaging. LUS severity correlated with CT and pulmonary function test markers of ILD severity. Validation of the revised LUS-ILD-24 supports the implementation of LUS imaging in screening algorithms for ILD in patients with SSc and IM.

  PublisherOA PDFDOI

Frequent coauthors

• Rosa Rademakers

  VIB-UAntwerp Center for Molecular Neurology

  382 shared

• Dennis W. Dickson

  Mayo Clinic in Florida

  273 shared

• Keith A. Josephs

  Mayo Clinic

  160 shared

• Bradley F. Boeve

  Mayo Clinic

  148 shared

• Zbigniew K. Wszołek

  WinnMed

  138 shared

• Neill R. Graff‐Radford

  WinnMed

  133 shared

• David S. Knopman

  Mayo Clinic

  127 shared

• Ronald C. Petersen

  Mayo Clinic in Florida

  118 shared