Short title:
CSU Transcriptomic signature and benralizumab treatment
Corresponding Author:
Jonathan A. Bernstein, M.D.
University of Cincinnati College of Medicine
Cincinnati, Ohio 45267-0563
Email:
Bernstja@ucmail.uc.edu
Phone: 513-235-7971
Fax: 513-558-3799
To the Editor,
We previously reported the efficacy of benralizumab (anti–IL5Rα
biologic) in controlling CSU refractory to second generation
H1-antihistamines (SGHA).1,2 Herein, we sought to
identify CSU-relevant genes and pathways and their regulation by
benralizumab.
CSU patients (N=10; unresponsive to up to 4x FDA-approved doses of SGHA)
were enrolled in a single-blind, uniform-dose, multi-treatment trial
(benralizumab 30mg × 3 treatments; Figure s1). Pre- and
post-benralizumab skin and blood samples were analyzed by RNAseq. We
identified a CSU signature using machine learning (ML) on an available
CSU dataset (GSE72540) which was validated using our
samples.3 Pre-, post-Rx symptom scores, and
corresponding blood eosinophil counts are summarized in Table-S1.
Comparison between lesional vs non-lesional CSU skin indicated
upregulation of multiple interleukins (IL-6, IL-20, IL-1β), chemokines
(CCL2, CCL3, CCL8, CCL13, CCL18 and CCL21) and adhesion molecules
(CD209, ICAM, CD69), while keratin variants (KRT25, KRT26, KRT27, KRT85)
were downregulated. Multiple enriched networks were identified with
IL1β, IL6, CCL2 and CCL3 as major hub genes (Figure-1A). Machine
learning and hierarchical clustering showed that a minimum of 9 genes
(OSM, PTGS2, UTP14A, ZFP36, STAT3, CCL2, RPF2, IL6 and LILRB3) were
required to stratify lesional from non-lesional samples with
> 90% predictive efficacy (Figure-1B). Compared to healthy
control skin, lesional CSU skin showed upregulation of adhesion
molecules (SELE, ICAM1, CD69), low-affinity IgE-receptor (FcER2) and
CC-chemokines (CCL2, CCL3, CCL4). Methods and further results are
summarized in the on-line supplement.
To determine the effect of benralizumab, the transcripts (Visit 2
pre-treatment vs Visit 5 end of study) were arranged by
significance (p-values) and the top 500 genes (fold-change value
> 2.0; ~ top 2% of all detected
transcripts) were used for downstream analysis.4,5 A
significantly enriched treatment-relevant network showed IL6, TNF and
IL1A as major hub-genes. Other relevant transcripts in the network were
CD44, mir-24, IFNg, IL4, S100A10 and S100A12. Relevant differentially
expressed genes (DEGs) were related to chemokeines/cytokines (CCR1,
CCL2, IL6; log2FC -2.04, -2.24, -2.94 respectively), inflammation
(PLA2G2A, PTGS2; log2FC -3.19, -1.02), adhesion (SELE;
log2FC = -3.7) and barrier function (AQP4;
Log2FC 4.27). Focusing on the three complete treatment
responders, the same set of hub genes (IL6, TNF and IL1A) in addition to
IL22, IL13, collagen-related genes (COL1A1, COL4A2) and NFkB gene
members were identified (Figure-1C, D). In addition, PLA2G2A
(Phospholipase A2, Group IIA), THBS1 (Thrombospondin-1) and SOCS3
(Suppressor Of Cytokine Signaling 3) were identified as significant DEGs
(FDR-adjusted p-values < 0.05; log2FC values
-3.3, -1.9 and -2.5 respectively). The functional significance of these
genes in regulating mast cell function, inflammation and cytokine
function have previusly been described.6-8 Moreover,
benralizumab responders also demonstrated a significant downregulation
of PTGDS2 (log2FC = -1.59), a major driver of cutaneous
inflammation. Baseline transcriptomic differences between CSU responder
and non-responder skin indicated that the major significantly
upregulated genes (log2 FC in perentheses) in
benralizumab-resistant CSU might be FOS family members of
proto-oncogenes FOSB (6.3), Activating Transcription Factor 3 ATF3
(4.5), and nuclear factor subfamily 4 members A1 and A2 (4.4 and 3.92,
respectively). Enrichment analysis identified platelet activities in
addition to glucocorticoid and phosphatase activities in the
non-responsive versus responsive CSU lesional skin. Finally, blood
transcriptomes pre vs post-benralizumab treatment demonstrated
downregulation of transcripts associated with IL5 signaling (e.g., IL5RA
and SIGLEC8 log2FC -2.79 and SIGLEC8 -5.47 respectively;
FDR-adjusted p-value < 0.01). When this analysis focused on
complete responders to benralizumab, a similar group of transcripts
(IL5RA, IL1RL1, RPL34, ALOX15, SIGLEC8 and S100A8) related to IL5
signaling and inflammatory pathways were downregulated. Collectively,
benralizumab was associated with blood IL5 downregulation and reversal
of the cutaneous CSU transcriptomic signature. Benralizumab-refractory
CSU could represent a distinct endotype warranting further
investigation.
Debajyoti Ghosh, MSc., PhD1
Rashid Karim PhD2
Xiang Zhang PhD3
Umesh Singh MD, PhD1
and
Jonathan A. Bernstein, M.D1
1 Division of Rheumatology, Allergy and Immunology,
Department of Internal Medicine, University of Cincinnati College of
Medicine
2Department of Electrical Engineering and Computer
Science, University of Cincinnati College of Engineering and Applied
Science
3Department of Environmental & Public Health
Sciences, University of Cincinnati College of Medicine.
AUTHOR CONTRIBUTIONS
DG and JB designed the study. DG, RK, XZ, US conducted the study, JB
supervised the work. DG and JB wrote the manuscript. All authors
contributed to the manuscript.
ETHICS STATEMENT
The study protocol was approved by the University of Cincinnati Internal
Review Board.
CONFLICTS OF INTEREST
Dr. Bernstein reports grants and consultant fees from AstraZeneca,
Sanofi-Regeneron, Novartis, Genentech, Amgen, Celldex, Allakos during
the conduct of the study. In addition, he has been a speaker for
AstraZeneca, Sanofi-Regeneron, Novartis and Genentech. Outside this work
he has received grants and consultant fees from Shire/Takeda, CSL
Behring, Pharming, Biocryst, Kalvista, Ionis, Escient, ONO, Incyte,
Cycle, Biomarin and Blueprint Medicine. He is a member of the Joint Task
Force for the AAAAI and ACAAI, a member of the CU UCARE guideline
committee and a UCARE and ACARE center of excellence, and a member of
the MAB of the HAEA organization. All other authors have nothing to
disclose.