Session: 203. Lymphocytes and Acquired or Congenital Immunodeficiency Disorders: Poster III
Hematology Disease Topics & Pathways:
Research, Biological therapies, Translational Research, Diseases, Immune Disorders, Therapies, immunology, Biological Processes
To characterize the immune dysregulation in iMCD, our lab previously profiled serum analytes and circulating cell types in patient blood during active disease. Serum proteomic analyses identified tumor necrosis factor (TNF) signaling as a highly enriched pathway in both siltuximab responders and non-responders. Both immunophenotypic and serum proteomic data indicated that T cells, which can produce TNF, were activated in iMCD. However, whether TNF production by activated T cells was a key mechanism promoting iMCD was unclear.
To determine if T cells contribute to the characteristic cytokine storm in iMCD, we investigated the ability of T cells from iMCD patients to produce different inflammatory cytokines including TNF. Upon stimulation with phorbol myristate acetate and ionomycin (PMA/I), both naïve and non-naïve CD4+ T cells from iMCD patients (n=9) produced significantly more TNF compared to healthy controls (n=9). There was a >2-fold increase in the frequency of TNF-expressing naïve CD4+ T cells from iMCD patients versus healthy controls. We also found a significant, yet marginal, increase in the frequency of TNF+ non-naïve CD4+ T cells from iMCD patients. Although TNF and IFN-γ are both well-established markers of T cell activation, we did not observe a similar rise in interferon-gamma (IFN-γ) after stimulation in either CD4+ T cell subset. Taken together, these data indicate that CD4+ T cells in iMCD patients hyper-respond to stimulation specifically by producing excess TNF.
In parallel to flow cytometric, proteomic, and T cell stimulation data, we leveraged KGML-xDTD, a knowledge graph-based machine learning framework, to predict potential novel drug treatments for iMCD. Utilizing 3,659,165 nodes and 18,291,237 edges in RTX-KG2 from 70 public biomedical sources, we trained a random forest-based machine learning algorithm on true positive and true negative treatment relationships. The top three novel predicted treatments for iMCD included two TNF inhibitors, adalimumab and certolizumab pegol, and the B cell depleting agent, rituximab, the second most prescribed drug for iMCD after siltuximab. We then used the reinforcement learning module of KGML-xDTD to predict the mechanism linking adalimumab to iMCD, which included IL-4, IL-6, IL-8, IL-10, STAT3, and CD4. These data reveal a potential new treatment strategy and further implicate TNF by CD4-expressing T cells in iMCD pathogenesis.
In addition to our laboratory findings and predictive modeling data that suggest that TNF production by T cells promotes iMCD, we report the successful treatment of a highly refractory iMCD patient with off-label use of a TNF blocker. When a 50-year old iMCD patient was experiencing multi-organ system dysfunction and preparing for hospice care after not responding to IL-6 inhibition, IL-1 inhibition, Bruton’s tyrosine kinase (BTK) inhibition, chemotherapy, and autologous stem cell transplantation, we initiated treatment with adalimumab alongside BTK inhibition. Within 3 days of the first infusion, the patient’s symptoms and organ dysfunction began to improve and the patient has been in remission for over 6 months.
We utilized a translational research approach including experimental and unbiased machine learning approaches to identify TNF as a novel therapeutic target that we inhibited to treat a highly refractory iMCD patient. Together, our data suggest that over-production of TNF, in part by activated T cells, promotes iMCD pathogenesis and highlight that further research is needed into TNF inhibition as a potential treatment strategy for iMCD.
Disclosures: Fajgenbaum: Medidata, a Dassault Systemes company: Consultancy; EUSA Pharma/Recordati Rare Disease: Consultancy, Research Funding.
OffLabel Disclosure: Adalimumab for the treatment of iMCD
See more of: Oral and Poster Abstracts