ABSTRACT
Background Sepsis is the leading cause of death in hospitalized children worldwide. Despite its hypothesized immune-mediated mechanism, targeted immunotherapy for sepsis is not available for clinical use.
Objective To determine the association between cytometric, proteomic, bioenergetic, and metabolomic abnormalities and pathogen type in pediatric sepsis.
Methods Serial PBMC samples were obtained from 14 sepsis patients (34 samples) and 7 control patients for this pilot study. Flow cytometry was used to define immunophenotype, including T cell subset frequency and activation state, and assess intracellular cytokine production. Global immune dysfunction was assessed by TNF-α production capacity and monocyte HLA-DR expression. Mitochondrial function was assessed by bulk respirometry. Metabolites were measured by liquid chromatography-mass spectrometry. Results were compared by timepoint and pathogen type. For detailed Methods, please see the Methods section in this article’s Online Supplement.
Results Sepsis patients were older and had higher illness severity compared to controls; demographics were otherwise similar. Compared to controls, sepsis patients demonstrated global immune dysfunction, loss of peripheral of non-naïve CD4+ T cells, and reduced PBMC mitochondrial function. Metabolomic findings in sepsis patients were most pronounced at sepsis onset and included elevated uridine and 2-dehydrogluconate and depleted citrulline. Loss of peripheral non-naïve CD4+ T cells was associated with immune dysfunction and reduced cytokine production despite increased T cell activation. CD4+ T cell differentiation and corresponding pro- and anti-inflammatory cytokines varied by pathogen.
Conclusion Pediatric sepsis patients exhibit a complex, dynamic physiologic state characterized by immunometabolic dysregulation which varies by pathogen type.
CLINICAL IMPLICATIONS Comprehensive immune monitoring in critically-ill patients using small-volume samples yields novel insights into diverse T cell responses to pediatric sepsis, which may improve future personalized therapies for these high-risk patients.
Introduction
Pediatric sepsis is the leading cause of death in hospitalized children worldwide (1). As in adults, sepsis in children is characterized by concurrent pro- and anti-inflammatory states with dysregulation of the innate and adaptive immune response to infection (2, 3). Critical care for sepsis is limited to antibiotics, source control, and supportive care for organ dysfunction (4). Immunocompromised patients and those who develop immune suppression in the setting of sepsis both represent high-risk clinical phenotypes, with mortality rates that exceed 50% (5, 6). Despite a hypothesized immune-mediated mechanism, successful interventional trials of targeted immunomodulation in sepsis remain elusive (7).
Previous investigations have demonstrated that focused measures indicative of innate and adaptive immune dysfunction (8–10) are associated with organ dysfunction (11) and mortality (12) in pediatric sepsis, however comprehensive analyses of immunometabolic changes are lacking and could yield therapeutic insights for these high-risk patients. Advances in immune profiling have the potential to allow new insights into the role that lymphocytes play in shaping the immune response to sepsis (13–15). Here we report on a single-center prospective, observational study of immunometabolic function in children with sepsis in which we identified sepsis-associated immune dysregulation which varied by pathogen type through longitudinal cytometric, proteomic, bioenergetic, and metabolomic assays.
Results and Discussion
Study patients
Fourteen patients with septic shock and seven non-infected controls were included in our study. Serial blood samples were collected from patients with sepsis and processed for peripheral blood mononuclear cell (PBMC) and plasma; in total, 34 sepsis samples and 7 control samples were obtained. Sepsis samples were collected on days 1-2, 3-5, and 8-14 after sepsis recognition. Control samples were collected from neurosurgical patients in the pediatric intensive care unit (PICU) without infection or organ dysfunction. Detailed descriptions of patient recruitment are available online and have been reported previously (16, 17).
Patient demographics, clinical characteristics, and laboratory values are summarized in Table 1. As expected, sepsis patients had higher illness severity at admission compared to controls as assessed by the Pediatric Index of Mortality-2 (PIM-2) score (18), Pediatric Risk of Mortality-III (PRISM-III) score (19), and Pediatric Logistic Organ Dysfunction (PELOD) score (20). Sepsis patients were also older than controls, but other demographics were similar. Sepsis patients more frequently received endotracheal intubation, vasoactive infusion, and corticosteroids. There was one mortality in the sepsis cohort, none among controls.
Characteristics of sepsis patients and controls at PICU admission
Pediatric sepsis patients exhibit global immunometabolic dysregulation
To quantify differences in lymphocyte subsets between sepsis patients and controls, we first assessed the abundance of key T cell subsets by flow cytometry (Fig. 1a). Compared to controls, the abundance of non-naïve CD4+ T cells was significantly reduced in patients with bacterial sepsis (Fig. 1b). Loss of central memory CD4+ T cells (CD45RA−/CD27+) was the primary driver of this finding; CD4+ TEMRA cells (CD45RA+/CD27−) were increased in patients with sepsis regardless of pathogen type. Non-naïve CD8+ T cell subsets were similar between sepsis and controls.
Immunometabolic dysregulation in pediatric sepsis patients. Relative abundance of major T cell subsets differ between sepsis and control and are most pronounced for patients with bacterial sepsis. Global immune dysfunction and mitochondrial dysfunction present at sepsis onset resolves on longitudinal sampling. Metabolomics findings reflect increased reactive oxygen species, amino acid breakdown products, and citrulline deficiency.
We then assessed two global markers of immune dysfunction commonly used in critical care – ex vivo lipopolysaccharide (LPS)-stimulated tumor necrosis factor (TNF)-α production capacity and monocyte human leukocyte antigen (HLA)-DR expression. Immunoparalysis, defined by TNF-α production capacity <200pg/ml (8), was more common in sepsis patients than controls (39% vs 0%, p=0.08). Median monocyte HLA-DR was lower at sepsis onset compared to controls (75% vs 92%, p=0.02). Cohort heterogeneity by interquartile range decreased on serial samples, and markers of immune dysfunction generally improved with sepsis recovery (Fig. 1c).
We interrogated bulk PBMC mitochondrial function using an Oroboros Oxygraph to measure spare respiratory capacity (SRC), which represents the mitochondrial bioenergetic reserve available for cells to produce ATP in response to stress-induced increase in metabolic demand (21). Median SRC was lower at sepsis onset compared to controls (4.0 vs 8.4, p=0.01). SRC improved through time and varied by pathogen type (Fig. 1d). Because the proportion of lymphocyte subsets varied between sepsis patients and controls, we constructed a generalized linear model which found no association between the relative proportion of CD4+ and CD8+ subsets (naïve, central memory, effector memory, TEMRA) in the PBMC sample and mitochondrial SRC, suggesting that differences in SRC are not confounded by shifts in cell subtypes.
Finally, we assessed plasma metabolites by liquid chromatography-mass spectrometry (Fig. 1e). In this exploratory analysis, some amino acid breakdown products including uridine (p<0.001) and hydroxyphenylpyruvate (p=0.012) and intermediate markers of glucose metabolism including 2-dehydrogluconate (p<0.001) and D-gluconate (p=0.017) were markedly elevated in patients with sepsis across timepoints compared to controls. Conversely, citrulline was decreased in children with sepsis regardless of pathogen (p<0.001); citrulline deficiency has been previously associated with impaired small bowel microcirculation, nitric oxide production, and mortality in sepsis (22, 23).
Taken together, these data demonstrate that pediatric patients with sepsis develop global immunometabolic dysregulation assessed by lower TNF-α production capacity, monocyte HLA-DR expression, mitochondrial SRC, loss of peripheral non-naïve T cells, and metabolomic abnormalities.
T cell abundance and polarization in pediatric sepsis vary by pathogen type
Informed by the heterogeneity of immune abnormalities above, we then stratified T cell responses by pathogen type. Demographics and severity of illness did not vary by pathogen type. We assessed CD4+ T cell differentiation by measuring intracellular cytokines via flow cytometry after phorbol myristate acetate (PMA)/ionomycin stimulation and identified CD4+ T cell states based on resulting intracellular cytokine expression: Th1 (IFN-γ+), Th2 (IL13+), Th17 (IL-17α+), Tfh (IL-21+), Treg (FoxP3+). Th1 cells were reduced in bacterial sepsis and Th2 cells were reduced in viral sepsis compared to controls; Th17, Tfh, and Treg populations did not vary significantly (Fig. 2a). Th1 and Th2 lymphocyte abundance varied widely across serial samples (Fig. 2b). Among patients with bacterial sepsis, the proportion of IL2+ non-naïve CD4+ lymphocytes was lower at sepsis onset compared to controls (Fig. 2c), while the proportion of CD38+ cells did not vary by pathogen type.
T cell activation and differentiation varies by pathogen type in pediatric sepsis patients.
Both pro- and anti-inflammatory plasma cytokines were elevated at sepsis onset compared to controls and varied by pathogen type (Fig. 3). IL-6 and IL-18 were most elevated in bacterial sepsis; IFN-γ was low in this subgroup, consistent with reduced CD4+ Th1 cells in this population. IL-10 and IL-1RA were elevated in all sepsis patients, though IL-4 was only elevated in patients with viral sepsis. Among sepsis patients, cytokine values normalized through time as organ dysfunction resolved.
Plasma cytokine levels vary by pathogen type in pediatric sepsis patients.
Loss of peripheral non-naïve CD4+ T cells is associated with immune dysfunction and characteristic immunometabolic dysregulation
In our final analysis, we examined the association between the loss of peripheral non-naïve CD4+ T cells and markers of immune dysfunction in sepsis patients. The loss of peripheral non-naïve CD4+ T cells is associated with both TNF-α production capacity (R=0.35, p=0.03) and monocyte HLA-DR expression (R=0.59, p<0.001; Fig. 4a). The frequency non-naïve CD4+ T cells is positively associated with serum IFN-γ level (R=0.57, p<0.001) and negatively associated with CD38 expression on non-naïve CD4+ T cells (R=-0.48, p=0.001), suggesting that loss of peripheral non-naïve CD4+ T cells is associated with decreased T cell function despite evidence of T cell activation (Fig. 4b).
Loss of peripheral non-naïve CD4+ T cells is associated with functional, proteomic, and metabolomic abnormalities in pediatric sepsis patients.
Finally, we tested the association between plasma metabolites and peripheral non-naïve CD4+ lymphocytes. The loss of peripheral non-naïve CD4+ T cells is associated with markers of protein catabolism (hydroxyproline, uridine, phenylalanine; Fig. 4c). Conversely, low glutamine was associated with decreased peripheral non-naïve CD4+ T cells. Glutamine is an essential cofactor for T cell activation, and glutamine depletion has been associated with reduced lymphocyte proliferation, cytokine production, and lymphocyte apoptosis (24, 25).
Conclusions
We have demonstrated that pediatric sepsis is associated with characteristic immunometabolic dysregulation which vary by pathogen. Adaptive immune dysfunction in sepsis develops during a hyperinflammatory, catabolic state which is characterized by mitochondrial dysfunction and loss of peripheral non-naïve T cells. Heterogeneity in the immune response is partially explained by pathogen type, which appears to influence non-naïve T cell differentiation and function. These alternations in CD4+ T cell subset frequencies which vary by pathogen suggest that the dysregulated adaptive immune response to sepsis may be pathogen-specific.
While these findings are compelling, there are several limitations to our analysis of this pilot data. Our limited number of samples and cohort heterogeneity is susceptible to type II error. Comparison of sepsis patients to controls may bias toward the null if post-surgical patients have a mild inflammatory phenotype which overlaps with the immunometabolic phenotype of sepsis patients. Because steroid exposure was common in sepsis patients regardless of pathogen type, we cannot control for this important covariate in the present study. Because metabolic and mitochondrial measurements were performed in bulk, we cannot conclusively identify the association between these findings and specific immune cell subsets. Finally, we are unable to account for repeated measures within individuals due to limited sample size.
Through this pilot study, we have established the clinical feasibility of monitoring immune health in pediatric sepsis patients using small volume samples. Immune monitoring in pediatric critical illness is currently limited to inflammatory biomarkers and cytokine analysis. Deep immune profiling and functional testing of patient samples has the potential to identify mechanisms of immune dysfunction in pediatric sepsis, paving the way for personalized immunotherapy in critically ill children.
Data Availability
Due to the nature of this research, participants of this study did not agree for their data to be shared publicly, so supporting data is not available.
Acknowledgements
The authors thank Florin Tuluc and Jennifer Murry from the CHOP Flow Cytometry Core Laboratory and Fang Chen and Natalka Kengle from the Center for Immunotherapies at the University of Pennsylvania for their contributions to this study.
Footnotes
Sources of Funding: Financial support was provided by the Endowed Chair, Department of Anesthesiology and Critical Care, Children’s Hospital of Philadelphia and the University of Pennsylvania Perelman School of Medicine. Dr. Lindell is also supported by the Thrasher Research Fund #15351. Dr. Wallace is also supported by NIH grants NS021328, MH108592, and OD010944, U.S. Department of Defense grants W81XWH-16-1-0401 and W81XWH-21-1-0128, (PR202887.e002). Dr. Wherry is also supported by the Parker Institute for Cancer Immunotherapy which supports the cancer immunology program at UPenn, and by NIAID AI155577, AI115712, AI117950, AI108545, AI082630. Dr. Weiss is also supported by NIGMS K23GM110496 and NICHD R01HD102396. Dr. Henrickson is also supported by NIAID K08AI135091, the Burroughs Wellcome Fund CAMS, the Clinical Immunology Society, and the American Academy of Allergy, Asthma, and Immunology.
ABBREVIATIONS
- PBMC
- Peripheral blood mononuclear cells
- PIM-2
- Pediatric Index of Mortality-2
- PRISM-III
- Pediatric Risk of Mortality-III
- PELOD
- Pediatric Logistic Organ Dysfunction
- LPS
- Lipopolysaccharide
- HLA-DR
- Human leukocyte antigen - DR isotype
- SRC
- Spare respiratory capacity
- PMA
- Phorbol myristate acetate
- Th1 cell
- Type 1 CD4+ T helper cell
- Th2 cell
- Type 1 CD4+ T helper cell
- Th17 cell
- Type 1 CD4+ T helper cell
- Tfh cell
- T follicular helper cell
- Treg cell
- Regulatory T cell
- TNF-α
- Tumor necrosis factor α
- IFN-γ
- Interferon γ