A lipid-immune network signature defines susceptibility to asparaginase-associated pancreatitis
Cheng-Yu Tsai, Na Bo, Thai Hoa Tran, Maisam Abu-El-Haija, Gayathri Swaminathan, Bomi Lee, Sudhir Ghandikota, Li Wen, Yves Théorêt, Steven D. Mittelman, Elena J. Ladas, Anil G. Jegga, Lewis B. Silverman, Ying Ding, Sohail Z. Husain
Cheng-Yu Tsai, Na Bo, Thai Hoa Tran, Maisam Abu-El-Haija, Gayathri Swaminathan, Bomi Lee, Sudhir Ghandikota, Li Wen, Yves Théorêt, Steven D. Mittelman, Elena J. Ladas, Anil G. Jegga, Lewis B. Silverman, Ying Ding, Sohail Z. Husain
View: Text | PDF
Clinical Research and Public Health Gastroenterology Hematology Oncology

A lipid-immune network signature defines susceptibility to asparaginase-associated pancreatitis

Abstract

BACKGROUND Asparaginase is essential for curing acute lymphoblastic leukemia (ALL), but its use is limited by asparaginase-associated pancreatitis (AAP), a severe and unpredictable toxicity lacking validated prospective biomarkers. We sought to define early systemic molecular features of susceptibility to AAP.METHODS We performed longitudinal lipidomic and proteomic profiling in two independent pediatric ALL cohorts (n = 161; 79 AAP cases, 82 controls) using paired blood samples collected before asparaginase exposure and at the end of induction therapy (including a single dose of asparaginase), thereby capturing pre-injury biology rather than consequences of pancreatitis. We applied differential abundance and network-based analyses and integrated lipid-cytokine associations using proteomics.RESULTS Across cohorts, we identified a reproducible lysophosphatidylcholine-centered (LPC-centered) signature characterized by attenuated induction therapy–associated LPC responses and disruption of LPC coregulation at the network level. Proteomic profiling revealed enrichment of cytokine signaling pathways, and integrative analyses demonstrated altered lipid-cytokine coupling, including a flip in association direction for LPC species and IL-18 between cases and controls. Although IL-18/LPC ratios did not differ globally, elevated postinduction IL-18/LPC ratios identified AAP risk within a protocol-defined very high-risk ALL subgroup (AUC = 0.81).CONCLUSION These findings support a systems-level model in which failure of coordinated lipid-immune responses under therapeutic stress confers vulnerability to AAP, providing a framework for validation and mitigation strategies.TRIAL REGISTRATION NCT00400946; NCT01574274; NCT03020030 (parent trials).FUNDING Servier Pharmaceuticals (IIT-95014-027-USA); SDRC (P30DK116074); Stanford SPARK; Fonds de Recherche du Québec – Santé; Fondation Charles-Bruneau; Leukemia & Lymphoma Society of Canada.

Authors

Cheng-Yu Tsai, Na Bo, Thai Hoa Tran, Maisam Abu-El-Haija, Gayathri Swaminathan, Bomi Lee, Sudhir Ghandikota, Li Wen, Yves Théorêt, Steven D. Mittelman, Elena J. Ladas, Anil G. Jegga, Lewis B. Silverman, Ying Ding, Sohail Z. Husain

×

Figure 2

Linear mixed-effects modeling identifies LPC species as key differentiators of AAP cases.

Options: View larger image (or click on image) Download as PowerPoint
Linear mixed-effects modeling identifies LPC species as key differentiat...
(A) Differential lipid species abundance between cases and controls at each time point using linear mixed-effects models adjusted for age, BMI, and initial ALL risk. After induction, 48 lipid species are significantly different, predominantly phospholipids. (B) Lipid class enrichment analysis of the 48 postinduction differential lipid species demonstrates strong overrepresentation of LPC, followed by PC, PE, and LPE. q < 0.05, Fisher’s exact test. (C) Within-subject postinduction to initial ratios reveal that 8 LPC species show significantly attenuated induction-associated increases in cases (n = 24) compared with controls (n = 26). (D) Composite LPC and LPE postinduction/initial ratios stratified by timing of pancreatitis onset show a graded pattern: highest in controls (n = 26), intermediate in non-early cases (n = 19), and lowest in early AAP cases (n = 5). *P < 0.05, **P < 0.01, linear trend test. (E) Multivariable logistic regression demonstrates that higher postinduction (26 controls and 24 cases) LPC and LPE concentrations are independently associated with reduced AAP risk. ORs are shown per 1-unit increase in log2 lipid concentration, adjusted for age, BMI, and initial ALL risk. *95% CI does not cross 1. The discovery cohort includes 26 cases and 26 controls. Unless otherwise indicated, analyses include all available samples from these participants at each time point.