2026 ASH®
Annual Meeting & Exposition December 12-15, 2026 ▪ New Orleans, Louisiana

Paper Details

Abstracts, authors, affiliations, and disclosures.

Saturday, December 12, 2026
05:30 p.m. - 07:30 p.m.
MCCNO - Hall F
Technology and Procedures, Gene Editing, Treatment Considerations, Biological Therapies, Bispecific Antibody Therapy, Artificial Intelligence (AI), Machine Learning, Biological Processes, Immune Mechanism, Clinical Practice (Health Services and Quality), Health Economics, Quality of Care, Diseases, Immunology, Autoimmune Disorders, GVHD, Immunodeficiency, Infectious Diseases, Viral, Lymphoid Malignancies, Lymphomas, Plasma Cell Disorders, Myeloid Malignancies, Non-Hodgkin Lymphoma, B Cell Lymphoma, Indolent Lymphoma, Aggressive Lymphoma, Research, Fundamental Science, Translational Research, Clinical Research, Epidemiology, Real-World Evidence, Study Population, Human, Adult
PAPER

1228 Serum proteomics identifies a blood testing panel to distinguish idiopathic multicentric castleman disease from clinically similar inflammatory diseases

Joseph Zinski, PhD, Bridget Austin, MS2, Melanie D Mumau, PhD3, Michael Vincent Gonzalez, PhD, MSc4, David Beck, MD, PhD5, Mei-Kay Wong6, Yasufumi Masaki, MD, PhD7, Jose Tomas Navarro, MD, PhD8, Amy Chadburn, M.D.9, Sabrina Racine Brzostek, M.D.10, Remi Sumiyoshi11, Jane Burns12, Chisato Shimizu13, Dirk Foell14, Sabrina Fuehner15, Philipp Faustmann16, Maureen Rischmueller17, Chandra Kirana18, Ramya Ramaswami, MBBS, MPH19, Kyndal Cook20, Scott Canna21, Jessica Lee22, Lachlan Smith23, Ciprian Ionut Tomuleasa, MD, PhD24, Adrian Bogdan Bogdan Tigu, PhD25, Joshua D Brandstadter, MD,PhD,MSc26, David Fajgenbaum, MD, MBA, MSc27
1 University of Pennsylvania Center for Cytokine Storm Treatment & Laboratory, PHILADELPHIA, Pennsylvania, United States
2 University of Pennsylvania Perelman School of Medicine, PHILADELPHIA, Pennsylvania, United States
3 University of Pennsylvania Center for Cytokine Storm Treatment & Laboratory, PHILADELPHIA, Pennsylvania, United States
4 University of Pennsylvania Center for Cytokine Storm Treatment & Laboratory, PHILADELPHIA, Pennsylvania, United States
5 NYU Grossman School of Medicine Center for Human Genetics and Genomics, New York, New York, United States
6 NYU Grossman School of Medicine Center for Human Genetics and Genomics, New York, New York, United States
7 Kanazawa Medical University, Uchinada, Ishikawa, Japan
8 Universitat Autonoma de Barcelona, Institut Catala d'Oncologia-Germans Trias i Pujol Hospital, Josep Carreras Leukaemia Research Institute, Barcelona, Badalona, Spain
9 Weill Cornell Medicine, New York, New York, United States
10 Weill Cornell Medicine, New York, New York, United States
11 Nagasaki University Graduate School of Biomedical Sciences, Nagasaki, Japan
12 University of California San Diego Kawasaki Disease Research Center, La Jolla, California, United States
13 University of California San Diego Kawasaki Disease Research Center, La Jolla, California, United States
14 University Hospital Muenster, Muenster, Germany
15 University Hospital Muenster, Muenster, Germany
16 University Hospital Tuebingen, Tuebingen, Germany
17 University of Adelaide The Queen Elizabeth Hospital, Adelaide, Australia
18 University of Adelaide The Queen Elizabeth Hospital, Adelaide, Australia
19 National Cancer Institute Center for Cancer Research, Bethesda, United States
20 National Institutes of Health, Bethesda, United States
21 University of Pennsylvania Perelman School of Medicine Children's Hospital of Philadelphia, Philadelphia, United States
22 University of Pennsylvania Perelman School of Medicine Children's Hospital of Philadelphia, Philadelphia, United States
23 University of Pennsylvania Perelman School of Medicine, Philadephia, United States
24 Iuliu Hatieganu University of Medicine and Pharmacy, Cluj-Napoca, Romania
25 Iuliu Hatieganu University of Medicine and Pharmacy, Cluj-Napoca, Romania
26 University of Pennsylvania Center for Cytokine Storm Treatment & Laboratory, PHILADELPHIA, Pennsylvania, United States
27 University of Pennsylvania Center for Cytokine Storm Treatment & Laboratory, PHILADELPHIA, Pennsylvania, United States

Background: Idiopathic multicentric Castleman disease (iMCD) is a potentially fatal hematologic disease complicated by diagnostic delays leading to additional morbidity. Diagnosis often requires exclusion of clinically similar infectious, autoimmune, and malignant conditions as well as excisional lymph node biopsy. Discordant pathologic interpretation can lead to further delay in addition to the inherent heterogeneity of clinical presentations defined by clinical subtypes: Thrombocytopenia, Anasarca, Fever, Reticulin fibrosis, Organomegaly [TAFRO], not otherwise specified [NOS], idiopathic plasmocytic lymphadenopathy [IPL]. We sought to develop a blood test capable of distinguishing all subtypes of iMCD from clinically similar inflammatory conditions.

Method: To determine serum biomarkers capable of distinguishing iMCD from its mimetics, we used SOMAmer aptamers to measure 7,294 serum analytes from 470 individuals: 71 from iMCD flares (including TAFRO, NOS, and IPL) and 399 comparator samples spanning healthy donors and twelve mimetic conditions, including: human herpesvirus-8-associated multicentric Castleman disease, Kawasaki disease, multisystem inflammatory syndrome in children, Sjögren's disease, systemic juvenile idiopathic arthritis and associated macrophage activation syndrome, IgG4-related disease, VEXAS syndrome, COVID19, inflammatory disease NOS, sepsis, and autoimmune lymphoproliferative syndrome. We corrected our data for age, sex, and collection site. As a positive control, canonical disease biomarkers were measured against healthy controls (eg. IL-18 in sJIA and MAS, CXCL10 in MIS-C and COVID19). While the expected biomarkers were up-regulated in each disease vs control, most were not up-regulated when compared to the entire cohort, supporting the need for a Machine Learning (ML) method to create biomarker panels to discern iMCD from the mimetic cohort. We employed multiple tree- and regression- based ML models to find robust biomarkers for iMCD vs mimetics and each subtype vs mimetics. From these hits we constructed panels of proteins with clinically available tests compatible with clinical deployment.

Result: We clustered serum proteomic profiles by pairwise concordance from health control and found most conditions clustered according to diagnosis except for iMCD, which split into two groupings (TAFRO and IPL with NOS represented in both). TAFRO clustered alongside inflammatory NOS, sepsis, COVID19, and HHV8+MCD while IPL clustered close to IgG4-RD. TAFRO could be distinguished by increased levels of proteins involved in cell migration, IL-6/JAK/STAT3 signaling, and IL-2/STAT5 signaling. 

Our ML analysis of TAFRO and IPL versus the mimetic conditions revealed numerous serum biomarkers that were distinctly up- or down- regulated in these subtypes. The angiogenesis pathway member VEGFA and its receptor KDR/VEGFR-2 emerged as the top biomarkers of the TAFRO subtype. Together they can distinguish TAFRO from mimetics with an AUC of 0.98. On its own, VEGFA can distinguish TAFRO from mimetics with an AUC of 0.93 (sensitivity: 0.85, specificity: 0.92). For IPL, the P53-induced protein TP53I13 can distinguish it from mimetics with an AUC of 0.91 (sensitivity: 0.85, specificity: 0.88). However, unlike VEGFA, TP53I13 has no clinically available lab test. No single protein measured by a clinically available lab test can distinguish IPL from its mimetics with an AUC above 0.85.

To identify clinically deployable panels for TAFRO and IPL, we restricted our analysis to elevated candidate biomarkers with clinically available blood tests. We determined that VEGFA and APOC3 together can distinguish TAFRO from mimetics with an AUC of 0.94, a specificity of 0.92 and a sensitivity of 0.88. A panel of 3 proteins, C3, TFRC, and C4BPA,  distinguishes IPL from mimetics with an AUC of 0.9, a sensitivity of 1, and a specificity of 0.73.

Conclusions: Our comparative study of iMCD serum protein levels could distinguish a hard-to-diagnose disease from clinically similar inflammatory conditions. Importantly, we identified a high-performing single biomarker, VEGFA, present in the serum of TAFRO patients and available as a clinically available test. By combining VEGFA with other biomarkers, we developed small panels of clinically testable serum proteins that can be deployed as an easy, reliable, first-line test for iMCD-TAFRO. We expect our results to facilitate more accurate iMCD diagnosis and advance investigations into its etiology.

Michael Vincent Gonzalez, PhD, MSc: No relevant conflicts of interest to declare.