Serum Proteomics

Background:

The etiology and pathogenesis of iMCD and other cytokine storm disorders remain poorly understood, . Diagnosis in iMCD is further complicated by there being no objectiveclinical diagnostic biomarkers and a subjective review of lymph node by a hematopathologist is required, leading to long paths to diagnosis and administration of/ineffective treatments. Anti-IL-6 therapy is effective in ~34% of patients. Recent SOMAlogic proteomic study performed to identify biomarkers and novel therapeutic targets. Comparisons were performed between iMCD and related disorders (Hodgkin Lymphoma, Rheum Arthritis, HIV/HHV8+ MCD) and healthy controls. This has produced a number of valuable discoveries: 

  • Differentially expressed protein sets in iMCD vs. controls, iMCD vs. related diseases

  • up/down regulated pathways for each comparison revealed IL-6-JAK-STAT3

  • Also allowed identification of 7 analyte panel that can be used to predict whether a patient would go on to benefit from siltuximab treatment. (not available clinically, limited effectiveness currently)

  • CXCL13 is one of most elevated proteins in iMCD

    • May be a potential diagnostic biomarker or early indicator of response to anti-IL-6 therapy

Many shared clinical observations in autoimmune/infectious disease/malignancies may argue for shared proteins/pathways involved in disease pathology. Four overlapping inflammatory conditions are: systemic juvenile idiopathic arthritis (sJIA), a chronic inflammatory disorder of children, sJIA with lung disease (sJIA-LD), macrophage activation syndrome (MAS), a cytokine storm that occurs in patients with sJIA and other autoimmune conditions, and malignancy-associated hemophagocytic lymphohistiocytosis (M-HLH), which is very similar to MAS but occurs due to the presence of a malignancy instead of an autoimmune disease. A recent SOMAlogic study by the Canna Lab of sJIA and MAS found:

  • MAS and sJIA proteomic signatures overlapped substantially, including serum amyloid A (SAA) and S100A9

  • Heatshock and glycolitic enzymes were also elevated (HSP 70, 90 proteins, S100 genes, GPI, LDH, ENO1)

  • ICAM5, MMP7 in sJIA-LD that may be able to distinguish sJIA-LD from sJIA/MAS

Project Plan:

Disparate datasets will be compared using statistical methods to compare across 8 rare disease cohorts (iMCD, RA, HL, HHV8+MCD, sJIA, sJIA-LD, MAS, M-HLH). Z- score transformation compared to healthy donors (to try to address batch effect) will be evaluated as a method to standardize somalogic somascan protein quantification in serum from an external rare disease cohort. This will allow a comparison of the differentially abundant proteins in both published datasets and to draw conclusions about shared/diverse protein architecture in disease pathology. Being able to compare multiple different datasets using somascan technology in various diseases can show overlapping/distinct disease signatures between disorders that share clinical signs/symptoms and may explain disease specific biology.

Research Questions:

Overall goal: To discover insights into the biology of these hyperinflammatory conditions and identify potential biomarkers that could help to distinguish them from one another. 

Secondary goal: To serve as promising preliminary data to inform the development of the SPACE study.

Can two (or 3 once we have Jordan data) disparate somascan datasets be combined using Z-scores generated from comparisons to healthy donor datacompared? 

  1. Can we determine if there is a linear relationship between protein analytes from HC cohorts from Canna & Fajgenbaum datasets

  2. If they are comparable, what is an effective standardization strategy to begin comparing between datasets? 

  3. Can we visualize overall similarities/differences in our dataset using unbiased clustering? 

    1. Perform/Visualize PCA using UMAP or other unbiased clustering to identify overall transcriptomic similarity/differences in 7-8 separate disease states

    2. Can we recognize subclusters within single disease? Are there a subset of iMCD that are more similar to sJIA-MS? More dissimilar? 

    3. Are any similar/dissimilar findings potential drug targets? Involved in a specific pathway? 

  4. What proteins are most/least abundant in serum for each disease state compared to healthy donors? 

    1. Compare the most significantlytop 25 increased/decreased proteins from each disease

    2. Are there shared genes? What is their function? 

    3. Are there different genes? Do they explain the different clinical presentations? 

  5. What proteins are most/least abundant in serum for each disease state compared to the other 6-7 disease states? 

    1. Compare the most significantly increased/decreased proteins from each disease

    2. Are any genes sensitive and specific as diagnostic biomarkers that could distinguish one disease from another?

    3. Are there shared genes? What is their function? 

    4. Are there different genes? Do they explain the different clinical presentations? 

  6. Targeted analysis of CXCL13

    1. How does this perform as a differentiator between diseases?

  7. Targeted analysis of HSP, glycolytic proteins identified in Canna publication

    1. Thinking heatmap for visualization here? 

  8. Can this z-score transformation strategy be used to add Michael Jordan’s dataset? And potentially beyond, i.e. SPACE

    1. Need to evaluate only 1300 measured analytes

    2. Is this data one to one?