AUTHORS
Jack Stylli, MD, Lukas Ronner, MD MSCR2, Kaitlyn C. Housel, BS MS3, Mika Esperanza, BS4, Matthew Erlich, BS5, Todd A. Miano, PharmD PhD6, Caroline A.G. Ittner, PhD7, Michael G.S. Shashaty, MD MSCE8, John P. Reilly, MD MSCE9, Catherine Lai, MD, MPH10, Nuala J. Meyer, MD MS11
AFFILIATIONS
1 Hospital of The University of Pennsylvania, Philadelphia, United States
2 Hospital of The University of Pennsylvania, Philadelphia, United States
3 University of Pennsylvania Perelman School of Medicine, Philadelphia, United States
4 University of Pennsylvania Perelman School of Medicine, Philadelphia, United States
5 University of Pennsylvania Perelman School of Medicine, Philadelphia, United States
6 University of Pennsylvania Perelman School of Medicine, Philadelphia, United States
7 University of Pennsylvania Perelman School of Medicine, Philadelphia, United States
8 University of Pennsylvania Perelman School of Medicine, Philadelphia, United States
9 University of Pennsylvania Perelman School of Medicine, Philadelphia, United States
10 Hospital of The University of Pennsylvania, Philadelphia, United States
11 University of Pennsylvania Perelman School of Medicine, Philadelphia, United States
ABSTRACT
Background: Hyperinflammatory and hypoinflammatory subphenotypes, identified using circulating cytokines and clinical variables, have been reproducibly described in critically ill patients with sepsis and acute respiratory distress syndrome and are associated with distinct outcomes and treatment responses. Patients with acute leukemia admitted to the intensive care unit (ICU) with sepsis frequently exhibit the hyperinflammatory subphenotype. Prior work suggests that subphenotype expression in acute myeloid leukemia (AML) varies with leukemia biology, but inflammatory responses in acute lymphoblastic leukemia (ALL) are poorly characterized. We compared admission cytokine profiles, inflammatory subphenotypes, and clinical outcomes among ICU patients with AML, ALL, and no hematologic malignancy.
Methods: The Molecular Epidemiology of SepsiS in the ICU (MESSI) study is a prospective cohort of patients admitted to medical ICUs with suspected infection and at least one organ failure. Plasma IL-1 receptor antagonist (IL-1RA), IL-1β, IL-6, IL-8, IL-10, and soluble TNF receptor-1 (sTNFR1) were measured on ICU admission. We identified patients enrolled from 2013–2022 with AML (n=127) or ALL (n=36) and contemporaneous patients without hematologic malignancy (n=891). Hyperinflammatory subphenotype was assigned using a previously validated parsimonious classifier. Standardized mean differences (SMDs) compared baseline characteristics, sepsis-related complications, and subphenotype assignment; SMDs ≥0.200 were considered meaningful. Multivariable logistic regression adjusted for age, sex, APACHE III score, chronic renal insufficiency, and congestive heart failure assessed associations between malignancy, subphenotype, and organ failures. Cox regression with the same covariates evaluated 30-day survival from ICU admission.
Results: Compared with patients with AML, patients with ALL were younger (median 41 vs 65 years; SMD 1.28), had lower illness severity (median APACHE III 75 vs 98; SMD 0.43), were more frequently neutropenic (71.4% vs 59.7%; SMD 0.25), more often in remission (20.0% vs 1.6%), and less often receiving induction therapy (14.3% vs 36.6%). The hyperinflammatory subphenotype was assigned to 61.5% of patients with ALL, 70.4% with AML, and 44.9% without hematologic malignancy. After adjustment, AML was independently associated with hyperinflammatory classification compared with no hematologic malignancy (odds ratio [OR] 2.61, 95% CI 1.64–4.23; p<0.001), while the association for ALL did not reach statistical significance (OR 2.39, 95% CI 0.997–5.95; p=0.054). Compared with AML, ALL demonstrated lower IL-1RA (median 683 vs 3,596 pg/mL; SMD 0.35) and higher IL-10 (23.2 vs 13.2 pg/mL; SMD 0.40), with similar concentrations of IL-6, IL-8, and sTNFR1. ARDS incidence was similar (47.2% vs 52.0%; SMD 0.10), whereas acute kidney injury (47.2% vs 64.6%; SMD 0.36) and 30-day mortality from ICU admission (30.6% vs 52.8%; SMD 0.46) were lower in ALL. In time-to-event analysis, AML (hazard ratio [HR] 1.32, 95% CI 1.06–1.66; p=0.015) and the hyperinflammatory subphenotype (HR 1.23, 95% CI 1.05–1.45; p=0.011) were independently associated with worse 30-day survival from ICU admission, whereas ALL was not (HR 1.43, 95% CI 0.89–2.28; p=0.139). Exploratory phenotype-stratified survival curves were nearly overlapping in ALL; however, this comparison included only approximately 22 hyperinflammatory and 14 hypoinflammatory patients.
Conclusions: Critically ill patients with ALL frequently exhibited the hyperinflammatory subphenotype but had a distinct cytokine profile and lower observed mortality than patients with AML. The lack of survival separation between inflammatory subphenotypes in this small ALL cohort should be interpreted cautiously. Despite the high prevalence of hyperinflammatory classification,30 day survival from ICU admission in ALL was substantially better than in AML, suggesting that inflammatory subphenotypes may not carry identical biological or prognostic significance across acute leukemia types. Larger ALL cohorts are needed to determine whether these differences reflect leukemia biology, treatment phase, or host characteristics.
DISCLOSURES
Jack Stylli, MD: No relevant conflicts of interest to declare.