Hello, the topic of this session is phenotyping sepsis-induced multiple organ dysfunction syndrome. My name is Julie Fitzgerald. I work in the Pediatric Intensive Care Unit at the Children's Hospital of Philadelphia, and my disclosures are noted on the slide. The objectives for this talk are to review the CAPCORN and National Institutes of Health study, Biomarker Phenotyping of Pediatric Sepsis and Multiple Organ Failure, or PHENOMS, outline the proposed sepsis-induced multiple organ failure phenotypes, including thrombocytopenia-associated multiple organ failure, sequential multiple organ failure, and immunoparalysis, and finally to explore the macrophage activation syndrome. First, we will start with a case scenario as an illustration of identification of a sepsis-induced multiple organ failure phenotype, the management that may be directed at the source of hyperinflammation in this phenotype, and successful resolution of the sepsis episode. In this case, a 13-year-old boy is admitted in respiratory failure with septic shock secondary to influenza and methicillin-resistant staph aureus pneumonia. After 18 hours in the Pediatric Intensive Care Unit, he has progressed to multiple organ failure, overt DIC with purpura fulminans. He's developed a fixed and dilated right pupil and later confirmed to have intracranial hemorrhages. He's been treated with escalating vasopressors and inotropes, escalating ventilator support, continuous FFP, bicarbon fusions, and he's received over 250 mLs per kilo in fluid and blood product resuscitation. By 26 hours in the Pediatric ICU, therapeutic plasma exchange has been started to target hyperinflammation as well as continuous renal replacement therapy. By 48 hours, he has been de-escalated to treatment with only one inotrope, his ventilatory requirement has stabilized, and his DIC has been reversed. Ultimately, the patient received five days of plasma exchange therapy, greater than one week of high-frequency oscillatory ventilation for ARDS, and four weeks of renal replacement therapy. Two months later, he is discharged to home, and five years later, he graduated from high school with a tennis varsity letter. It is well described that increasing number of organ failures correlate with higher mortality rates in children with severe sepsis. This graph shows data from an international point prevalence study of severe sepsis in Pediatric ICUs. The maximum number of concurrent organ dysfunctions in the enrolled patients is shown on the x-axis, and the proportion of patients with mortality at hospital discharge is shown on the y-axis. As the number of concurrent organ failures rises, the mortality rate shows a corresponding rise, with mortality of less than 20% for two organ failures and mortality of over 80% for six organ failures. One question is what drives these increasing numbers of organ failures in children with sepsis? It is likely that a state of uncontrolled hyperinflammation is a strong contributor. As part of the infection process, a pathogen interacts with host immune cells through the recognition of the pathogen's pathogen-associated molecular pattern by the host pattern recognition receptors. This leads to gene expression via second messenger systems, then synthesis and release of inflammatory mediators. In turn, this can lead to the systemic inflammatory response syndrome, sepsis, and septic shock. This manifests clinically as signs of host distress and ultimately death of host cells and host tissues. As these cells and tissues die, damage-associated molecular patterns are released, which are also recognized by the host pattern recognition receptors, and the process is perpetuated. In both pediatric and adult sepsis trials, many promising therapeutics have failed to show benefit, possibly because of a broad application of these therapies to a heterogeneous disease process. The aim of precision medicine in sepsis is to take into account individual patient variability in the development of strategies to prevent, diagnose, and treat sepsis. The goal is to identify phenotypes or endotypes of disease that can be used to enrich a study population that may be more likely to benefit from a therapy, and to test therapies that target biologically plausible disease pathways in that population. Different strategies may be used to enrich a study population. Phenotypes may be recognized by a patient's particular clinical presentation and disease course. A phenotype is an observable characteristic of a disease, including clinical signs and symptoms and biochemical properties. Endotypes are identified using molecular tests. An endotype may include characteristics of molecular, cellular, immunological, or genetic tests, and these provide insight into the underlying pathology of the disease process. An endotype defines a distinct functional or pathobiological mechanism. Endotypes are likely present within phenotypic clusters. In diseases, there may be environmental triggers. Specifically, in sepsis, this trigger would be an infection and or toxin exposure. The body responds to this trigger through gene transcription and protein expression. And the variety of host responses through these mechanisms may manifest as different phenotypes. In sepsis, some of these described phenotypes are macrophage activation syndrome, or MAS, immunoparalysis, thrombocytopenia-associated multiple organ failure, or TAMOF, and sequential multiple organ failure. A patient's genotype and endotype contribute to the observed phenotype. This diagram shows the framework for proposed pediatric sepsis phenotypes that will be covered in this talk. First, patients are identified with septic shock and ideally resuscitated according to pediatric surviving sepsis campaign guidelines, consisting of early antibiotics, hemodynamic support, and source control. Many of these patients will go on to have resolution of septic shock without additional organ failure. A subset will have single organ failure. Another subset will go on to have multiple organ failure. As previously discussed, mortality increases incrementally with increasing number of organ failures. Those with multiple organ failure may be classified as having specific phenotypes, shown in these boxes. There are broad general descriptors of the phenotypes in each box, describing the underlying mechanisms for each phenotype. The phenotypes are TAMOF, characterized by microvascular thrombosis, immunoparalysis, and sequential multiple organ failure, which is defined as respiratory distress or respiratory failure, followed by liver dysfunction. There can be significant overlap across these different phenotypes. And then the ultimate common final pathway of hyperinflammation driving these phenotypes can result in macrophage activation syndrome. This figure shows more detail of each of these phenotypes. TAMOF is characterized by thrombocytopenia, acute kidney injury, and low ADAMTS-13 level. Clinically, patients will have hemolysis, liver dysfunction, and may have mutations in inhibitory complement or ADAMTS-13 signaling. Immune paralysis is characterized by decreased monocyte HLA-DR levels or decreased whole blood TNF response to LPS stimulation. Immunosuppression and congenital immunodeficiencies are associated with this. Sequential multi-organ failure is characterized by respiratory followed by liver dysfunction and elevated soluble FAS ligand levels. This is associated with immune suppression in the setting of transplantation with EBV infection or in viral infections in the setting of certain immunodeficiencies. A multicenter study was performed to evaluate the hypothesis that these inflammation phenotypes in pediatric sepsis are associated with increased risk of MAS and increased risk of death. This table shows the definitions used for each phenotype and for MAS in this study. With the phenotype in the first column, the clinical criteria for the phenotype in the second column, and the laboratory values used in assigning the phenotype in the third column. These are similar to what was presented on the previous slide. 401 patients in nine pediatric ICUs were enrolled in this study. Enrolled children had to have sepsis and at least one organ failure as well as an arterial or central venous blood-drilling catheter. This is a table from the manuscript reporting the results of this study and shows the association of phenotype with relative risk of mortality. Patients with single organ failure in the first row made up about 28% of enrolled patients and had a low risk of ICU mortality. The mortality rate for this group was about 1%. Patients with at least two organ failures were classified as multiple organ failure, then further divided into those with multiple organ failure who did not have one of the three phenotypes, which was about 47% of the study cohort, and those who did have one of the three phenotypes. And these patients made up one quarter of the cohort. As expected, mortality was higher in those with multi-organ failure compared to those with single organ failure at 15% versus 1%. Those with an inflammatory phenotype and multiple organ failure had a higher mortality with 24% mortality in this group compared to 10% in those with multiple organ failure without one of the phenotypes. The relative risk of mortality after adjustment for important covariates was over two-fold higher in those with an inflammatory phenotype compared to those with multiple organ failure without an inflammatory phenotype, which was the reference group. They also looked at patients with each phenotype and compared to the reference group, and each phenotype individually showed an increased risk of mortality. Finally, those who met criteria for MAS had the highest mortality at 46%. Thus, the investigators proved their hypothesis that these inflammatory phenotypes in pediatric sepsis identify patients with a higher risk of mortality. The next steps will be to design and implement interventional trials in these subpopulations that target the underlying mechanisms driving these states. The investigators looked at the timing of onset of these phenotypes compared to timing of onset of multiple organ failure. Most patients with multiple organ failure met this criteria on the first day of sepsis, while the median time to onset of TAMOF was four days, to sequential multiple organ failure was six days, and to immune paralysis was five days. The median time to meeting MAS criteria was shorter than the phenotypes at three days. Based on these data, targeted therapies would need to be implemented only once a phenotype is identifiable. This provides some time for multiple organ failure to resolve in those patients who will not go on to develop a phenotype, and some time for a clinician to identify the phenotype and determine which targeted therapy might be appropriate for that particular patient. Now we will delve deeper into each of these phenotypes, starting with immune paralysis. It is well described that in systemic inflammation and multiple organ failure, there is a balance between pro and anti-inflammatory mediators, and disruption of that balance may contribute to disease pathogenesis. Immune paralysis occurs when the anti-inflammatory response predominates. Patients with immune paralysis are unable to clear infections and succumb to unremitting infection. This slide shows different definitions and lab cutoffs for the presence of immune paralysis. I'll pause for you to take a moment to read these definitions. This slide shows data from an interventional study. In this study, children with at least three organ failures and evidence of immune paralysis based on TNF response were randomized to receive treatment with GM-CSF or usual care. A low TNF response or immune paralysis was associated with more nosocomial infections and higher mortality. GM-CSF therapy facilitated recovery of TNF response and prevented nosocomial infections in the enrolled patients. In this multicenter observational study during the 2009 H1N1 influenza season, in the panel in the upper left, non-survivors had lower ex vivo TNF alpha production in response to LPS stimulation. The cumulative incidence curve in the upper right panel shows that patients with higher TNF production were more likely to survive and to be discharged sooner from the ICU. The bottom left panel shows that TNF production was lowest in those with Staph aureus co-infection, highlighted by the red box. Finally, the bottom right panel shows that non-survivors, denoted by filled circles, had both lower TNF production on the x-axis and lower absolute monocyte counts on the y-axis. Taken together, these studies show that decreased immune responses are seen in patients with immune paralysis. This is associated with lower survival and GM-CSF may restore immune function and prevent secondary infection. Thus, it may be a good target for larger interventional trials in children with sepsis and immune paralysis. In TAMOF, thrombocytopenia is thought to be due to consumptive coagulopathy in the setting of DIC, TTP, HUS, or thrombotic microangiopathic hemolytic anemia. Patients with TAMOF will have elevated LDH, acute kidney injury, may have schistocytes on peripheral blood smear, and have a decreased ADAMTS-13 level. DIC is more common in this spectrum of disease processes and is related to infection triggering leukocytes to release tissue factor, which then leads to fibrin-predominant thrombi. TTP is much less common and is related to ADAMTS-13 deficiency and generation of ultra-large von Willebrand factor multimer platelet aggregates, causing platelet von Willebrand factor-predominant thrombi. In this framework, DIC leads to the majority of TAMOF, with TTP, HUS, and other causes contributing in lower frequencies. A primate study investigated ADAMTS-13 deficiency. In this study, monoclonal antibodies that functionally inhibit ADAMTS-13 were sufficient to trigger TTP in a dose-dependent manner. Specific inflammatory mediators have been shown to inhibit ADAMTS-13 in humans, such as immunoglobulin G, IL-6, plasma-free hemoglobin, and shiga toxins. Mediators that can cleave and inhibit ADAMTS-13 are granulocyte elastase, plasmin, thrombin, and neutrophil peptides. So these inflammatory mediators could lead to functional inhibition of ADAMTS-13 and are likely to contribute to the development of TTP. These figures are from a single-center study of children with TAMOF. The Venn diagram shows that most of the patients in this study had low ADAMTS-13 activity, and many had ultra-large von Willebrand-factor multimers and elevated PT levels. On the right are slides of tissue from brain, kidney, and lung from a representative patient stained with von Willebrand-factor polyclonal antibody in the lower panels. The top panels are negative controls. You can see von Willebrand-factor-rich microvascular thrombi staining in each organ of this patient in the bottom panels. This was a multi-center observational study of the use of therapeutic plasma exchange, noted as TPE in these figures, for children with TAMOF. In the first panel, the PILOD score, a measure of organ dysfunction, decreased more over time in children treated with plasma exchange than in those who were not treated with plasma exchange. In the table, the red arrow points to the data showing that plasma exchange was associated with reduced 28-day mortality by propensity score weighting analysis in this population. This provides evidence of the potential efficacy of a specific treatment, plasma exchange, for patients with sepsis and TAMOF. Moving on in the three phenotypes, sequential multiple organ failure is respiratory failure followed by hepatic failure with viral lymphoproliferative disease. This is the least common of the three phenotypes. Common findings in this phenotype are elevated ALT, elevated creatinine, and elevated soluble FAS and soluble Fas-ligand levels. This phenotype is driven by a perturbation in the immune system's ability to perform activation-induced cell death leading to lymphoproliferation. It is also characterized by an inability to perform DNA viral infection killing, leading to uncontrolled viremia. Fas-ligand bound to cells normally mediates apoptotic pathways. Soluble Fas-ligand decreases the normal pathways of viral and activated immune cell death. Furthermore, soluble Fas-ligand mediates hepatic cell injury and death. Thus, increased soluble Fas-ligand leads to lymphoproliferation and increased cellular reservoir for DNA viruses such as EBV and to liver injury. In the setting of transplant patients, treatments are typically decreasing immunosuppression. In the setting of PTLD, antibodies to CD20, such as rituximab, are used to decrease the lymphocyte reservoir for EBV. Now we will discuss the final common pathway of hyperinflammation in sepsis, macrophage activation syndrome. MAS combines features of hepatobiliary dysfunction and DIC. And patients with MAS will have elevated ALT, elevated bilirubin, elevated INR, and elevated ferritin, as well as thrombocytopenia or sometimes pancytopenia. This is the normal host response to a new antigen, either from infection, malignancy, or in the setting of immune response. In this response, the antigen is recognized by antigen-presenting cells, which present the antigen to cytotoxic T lymphocytes. The cytotoxic T lymphocytes then produce inflammatory mediators to generate a systemic immune response to the antigen. At the same time, the cytotoxic T lymphocytes use a negative feedback pathway to destroy or downregulate the antigen-presenting cells to dampen the ongoing inflammatory response generated by the presentation of the antigen. The immune system then responds to the inflammatory mediators and eliminates the antigen, leading to patient recovery in the setting of infection. It is proposed that in pathologic states of immune activation, either acquired or genetic causes will interfere with either the normal process of antigen presentation or the negative feedback control of antigen-presenting cells. The resulting systemic inflammatory response is then uncontrolled and leads to organ injury and MAS. The common themes in this framework are persistence of the antigen and the antigen presentation, persistence of activated antigen-presenting cells, and a postulated process of viral priming followed by bacterial infection. Evidence from mouse models support this viral priming idea. In mice who are given a viral challenge followed by exposure to LPS, the mice exhibit a syndrome similar to MAS or secondary HLH, characterized by high ferritin, high IL-2 levels, hemophagocytosis, and high mortality rates. When the antigens are given in the reverse order, in other words, if the mice are given an LPS challenge followed by a viral challenge, then these mice do not exhibit this MAS phenotype. Pathologic immune activation may be expressed along a continuum of risk. First, there is classical HLH, hemophagocytic lymphohistiocytosis, in which there is a null genetic mutation leading to no activity of a certain aspect of the normal immune response pathway. This disease process is treated according to standard primary HLH treatment protocols. There are also hypomorphic genetic mutations that lead to lower levels of activity, but not absolutely no activity of part of this pathway. These patients typically present later with a milder form of HLH. Finally, there may be polygenic mutations that result in this spectrum of disease phenotype only after being triggered by certain infections, malignancies, or autoimmune exposures. This diagram shows different pathways that may contribute to MAS or secondary HLH in sepsis. Starting on the left, tissue factor can promote endothelial dysfunction and consumptive coagulopathy. Hemolysis or transfusion can lead to free hemoglobin-haptoglobin complexes that activate macrophages and monocytes through CD163. Macrophages themselves release tissue factor, soluble CD163, ferritin, IL-6, IL-1, and IL-18. These make up the inflammasome and positively feedback the cycle of inflammation. Decreased natural killer cells or cytotoxic T lymphocytes resulting from sepsis allows further macrophage and lymphocyte proliferation, again driving this cycle. So what about possible directed treatments for sepsis-related MAS? This study was a secondary post-hoc analysis of a trial of IL-1 receptor blockade in adults with sepsis that originally did not demonstrate evidence for benefit in the larger enrolled patient population. The investigators in this particular study reanalyzed the results looking at subgroups of patients who fulfilled criteria for hepatobiliary dysfunction and DIC. The red arrows point to the lines in the Kaplan-Meier curve for survival in the subgroups of patients with hepatobiliary dysfunction and DIC who did in the black solid line and did not in the black dashed line receive treatment with IL-1 receptor antagonists. Treatment with this showed an almost two-fold lower mortality, 35% mortality in those who received the IL-1 receptor blocker versus 65% mortality in those who did not receive the IL-1 receptor blocker. This provides promising evidence that IL-1 receptor blockade may be a plausible therapy for patients with sepsis-induced MAS. Let's circle back to the earlier diagram of the different sepsis hyperinflammation phenotypes. Based on data presented in this talk, as well as several other studies, potential therapeutic targets have emerged for evaluation as treatments in these different phenotypes. These are noted here and include plasma exchange for TAMOF, IVIG and antiviral therapies for sequential multiple organ failure, decreasing immunosuppression and administering GMCSF for immunoparalysis, and steroids IVIG and IL-1 receptor blockade in MAS. These need to be studied in rigorous multicenter trials to prove their efficacy in these populations but are promising based on the evidence to date. There are several clinical trials planned or in process for these phenotypes in adults with sepsis. Many are listed here on this slide for your reference. There are also several pediatric trials planned, in process, or recently completed, with results pending, including GM-CSF for immune paralysis and IL-1 receptor blockade in hyperinflammation. Thank you for attending this course. To summarize the key points of this talk, children with sepsis and multiple organ failure with at least one of the discussed hyperinflammation phenotypes have increased risk of mortality compared to those with multiple organ failure without an inflammation phenotype, making this a key group for further study of targeted therapeutics based on the underlying pathology of the phenotype. I look forward to joining you during the live question and answer session to answer any questions on this talk.

sepsis-induced MODS

children

phenotypes

treatment options

clinical trials

inflammatory phenotype