Management of Cytokine Release Syndrome

Introduction

Chimeric antigen receptor (CAR) T-cell therapy has been shown to have antitumor activity in a variety of cancers including pediatric and adult acute lymphoblastic leukemia (ALL), ^{, ,} diffuse large B-cell non-Hodgkin's lymphoma (NHL), ^, multiple myeloma, and chronic lymphocytic leukemia (CLL), ^, leading to the commercial approval of two CD19 CAR T-cell therapies, tisagenlecleucel and axicabtagene ciloleucel.

Cytokine release syndrome (CRS) is a constellation of symptoms resulting from supraphysiologic cytokine production that can begin within hours of infusion and last up to 30 days after the administration of CAR T-cell therapy. ^, Fever, hypotension, tachycardia, tachypnea, hypoxia and capillary leak, hypoalbuminemia, coagulopathy, and, in rare cases, multiorgan failure characterize CRS. ^, Presentation of CRS may vary depending on which CAR T-cell therapy or dose is given, on patient characteristics, and on disease burden. The symptoms associated with CRS are caused by a robust, rapid release of cytokines and chemokines from proliferating CAR T-cells as well as from the host immune response.

In a recent systematic review and meta-analysis of CAR T-cell trials in hematologic malignancies and solid tumors, more than 50% of recipients experienced CRS. CRS prevalence did not differ between the hematologic cancers treated, costimulatory domains, or phase of the clinical trial. While CRS can be mild in some patients, in others the reactions can be life-threatening or even result in death. While this study did not find an association between CRS incidence and severity and tumor efficacy, several single studies have identified correlations between response and CRS incidence and severity within their patient population. ^, Since severity of CRS can be mitigated with anticytokine therapies, such as tocilizumab or corticosteroids, and since the timing of such intervention has changed over time from later to earlier in the course, broad studies of correlations between CRS severity and outcome may not be valid. Grading of CRS was not uniform across all studies further complicating analysis.

With the anticipated adoption of the recently published CRS consensus definition and grading criteria sponsored by the American Society for Transplantation and Cellular Therapy (ASTCT), future analyses across trials and products will be more robust. However, the field has yet to establish consensus management guidelines for CRS, which would allow for even better intraproduct comparisons. The ASTCT and other groups are actively pursuing a CRS management consensus, but at this time, it is not available. Therefore, this chapter will dissect the common clinical manifestations of CRS and suggest current best management approaches where data exist or expert opinion aligns. Importantly, practitioners should remain current with the latest data and practice in CRS management, as this is a rapidly evolving field. Furthermore, every clinical situation is unique, so practitioners should exercise their best judgment when interpreting or implementing suggestions contained herein. To begin, one should understand the pathophysiology and risk factors for CRS.

Pathophysiology of Cytokine Release Syndrome

CRS begins with the engagement of CAR T-cells to its corresponding antigen on both malignant and nonmalignant, antigen-expressing cells, releasing a large amount of inflammatory cytokines. It is important to note that cytokine production is not restricted to CAR T-cells. CRS may occur with other agents that target and engage T and/or B-cells, including rituximab and blinatumomab. ^,

Early insight into the pathophysiology of CRS was obtained when six healthy individuals were administered TGN1412—a superagonist monoclonal antibody to CD28. Within 90 minutes of the infusion, all volunteers developed a systemic inflammatory response accompanied by headache, nausea, diarrhea, and hypotension and progressed rapidly (12–16 hours) to multiorgan failure requiring aggressive critical care support for recovery. Severe hypotension requiring vasopressors, capillary leak syndrome resulting in respiratory failure and acute kidney injury, coagulopathy, and neurological manifestations were all reported. C-reactive protein (CRP) was elevated, and retrospective analysis revealed marked elevations in IFN-γ, TNF-α, IL-6, IL-10, IL-2, and IL-1β, among others. High-dose corticosteroids were administered, and all patients eventually recovered.

Correlative studies in patients who develop CRS after CAR T-cell therapy have similar patterns of supraphysiologic cytokine production. The direct and indirect effects of such robust cytokine release are highly complex and have not been systematically evaluated. Though a small handful of molecules involved are shown in Table 5.1 , this list is by no means exhaustive. The end result of this inflammatory milieu is endothelial injury and capillary leak, which contribute to the predominant clinical manifestations ( Fig. 5.1 ). Such effects have recently been recapitulated in animal models.

Murine Models of Cytokine Release Syndrome

A humanized, leukemia-bearing mouse model developed by Norelli et al. demonstrates features of CRS when treated with murine-derived human CAR T-cells. Mice experience weight loss, systemic inflammation, fever, and elevated IL-6, TNF-alpha, IL-1, and IL-10, all reminiscent of CRS in humans. As suspected, IL-1 and IL-6 were produced by monocytes in animals rather than the CAR T-cells themselves. Indeed, if monocytes are eliminated in this model, the clinical manifestations of CRS did not occur. Administration of tocilizumab, a monoclonal antibody that blocks the IL-6 receptor, also prevented CRS though it did not prevent CAR T-cell–associated neurotoxicity. Interestingly, anakinra, an IL-1 receptor small molecule antagonist, prevented both CRS and neurotoxicity in this mouse model providing strong evidence for a clinical trial of this agent in managing CAR toxicities.

T. Giavridis et al. reported a mouse model with high intraperitoneal lymphoma burden, which develops CRS within 2–3 days after CD19 CAR T-cell infusion. Interestingly, the same mice when engrafted with leukemia in their marrow did not develop CRS. Mice present with fatigue and weight loss, and if untreated, this syndrome resulted in mortality. IL-6, IL-1, and nitric oxide were produced extensively by host macrophages not CAR T-cells, and tocilizumab or anakinra abrogated CRS symptoms and prevented mortality, similar findings to Norelli's group. In contrast, Giavridis did not observe neurologic dysfunction in these animals, which do not bear a humanized immune system, suggesting one is required for neurotoxicity.

Nonhuman Primate Models of Cytokine Release Syndrome

ROR1, a tyrosine kinase receptor, is a potential candidate for CAR T-cell therapy since it is expressed on the surface of many hematologic malignancies and may have a role in the survival of tumor cells. Berger and colleagues evaluated ROR1-directed CAR T-cells in nonhuman primates ( Macaca mulatta ) and did not see any increased toxicity to low-level, transcript-expressing ROR1 organs of the macaques. CAR T-cells were thought to be activated as IFN-gamma and IL-6 levels in circulation increased after CAR T-cell infusion. However, only one of two animals with B-cells that highly expressed ROR1 developed fever 12 days after infusion of the ROR1 CAR T-cells but no other signs of CRS. Though data are limited, this study, which did not use lymphodepletion prior to CAR T-cell infusion, suggests that nonhuman primates may be an adequate model to study CRS.

A different group infused three M. mulatta nonhuman primates with CD20-targeted CAR T-cells after lymphodepletion with cyclophosphamide. CAR T-cells proliferated and were detectable for up to 43 days after the infusion. Approximately 5–7 days after CAR T-cell administration, subjects developed symptoms of CRS (fever and weight loss), which coincided with the proliferation and activation of the CAR T-cells. Similar to humans, CRP, ferritin, IL-6, IL-8, and IL-1β, among others, were elevated in CD20 CAR recipients but not controls and correlated to symptoms. Interestingly, elevated IL-6, IL-2, granulocyte-macrophage colony-stimulating factor (GM-CSF), and VEGF were found in the CSF of animals that also developed neurotoxicity.

These mouse and nonhuman primate models as well as correlative studies from patients receiving CAR T-cell therapy on clinical trials have improved our knowledge of the pathophysiology of CRS. Cytokines from host monocytes and macrophages as well as endothelial activation may contribute to CRS ( Table 5.1 ). Clearly, IL-6 plays a key role in toxicity, as blockade of its signaling with tocilizumab can mitigate the clinical symptoms of CRS, but IL-6 and tocilizumab are not the only possible areas for intervention. As more preclinical data are developed, clinical trials using other CRS inhibitors, such as anakinra, will be important to advance the field.

Important Biomarkers of Cytokine Release Syndrome

IL-6 is one of the central mediators of CRS. This cytokine can cause both proinflammatory and anti-inflammatory responses depending on its interaction with gp130 and the IL-6 receptor (IL-6R), respectively. CD130 (gp130) is expressed broadly on nonhematologic tissues, and the IL-6R is present on macrophages, neutrophils, hepatocytes, and T-cells. IL-6, when present at low concentrations, interacts primarily with the high-affinity, membrane-bound IL-6R and mediates anti-inflammatory properties. A soluble form of the IL-6R (sIL-6R) also exists. During periods of higher IL-6 concentration (e.g., present during CRS), IL-6 will form a complex with the lower-affinity sIL-6R, which can then activate the trans signaling pathway through its binding to gp130 independent of the membrane-bound IL-6R, resulting in a proinflammatory response in a wider array of tissues ( Fig. 5.2 ). Understanding this pathophysiology has led to important therapeutic implications of using IL-6R inhibitors like tocilizumab ( Fig. 5.2 ) Indeed, serum IL-6 levels have been shown to correlate with CRS severity.

Other groups have retrospectively mined clinical trials of CD19 CAR T-cells for correlations with other cytokines or chemokines in an effort to better characterize and perhaps predict severe CRS. Correlative studies performed by Hay et al. found that patients who developed > grade 4 CRS had higher levels of IFN-gamma, IL-6, IL-8, IL-10, IL-15, monocyte chemoattractant protein (MCP-1), tumor necrosis factor receptor p55, and macrophage inflammatory protein 1β within 36 hours after CAR T-cell infusion ( Table 5.1 ). For those patients with milder CRS, the levels took longer to increase. When compared with CRP, ferritin, or other cytokines, MCP-1 performed better at predicting patients with ≥ grade 4 CRS. In another clinical trial, serum from 51 patients who received CD19 CAR T-cell therapy was analyzed serially for 43 cytokines that might predict CRS and compared with patients who had severe CRS to those who did not. Twenty-four cytokines were associated with severe CRS, including IFN-gamma, IL-6, IL-8, sIL2Rα, sgp130, sIL6R, MCP1, MIP1α, and GM-CSF. ( Table 5.1 )

The group at the Fred Hutchinson Cancer Research Center (FHCRC) made additional key discoveries. Analysis of samples from patients treated with CD19 CAR T-cell therapy at their center suggested that endothelial dysfunction may play a role in the hypotension, capillary leak syndrome, and coagulopathy associated with CRS. An examination of angiopoietin-1 and angiopoietin-2 as well as von Willebrand factor levels in patients who received CAR T-cell infusions revealed that endothelial cell activation was seen in CRS ( Table 5.1 ). Patients with severe CRS had higher levels of endothelium-activating cytokines and their serum activated human umbilical vein endothelial cells in the lab to a higher degree than controls. This led to the binding of von Willebrand factor and platelets. In some patients, the endothelial activation was even seen prior to the lymphodepleting chemotherapy for CAR T-cell therapy. A postmortem examination of a patient with fatal neurotoxicity after CAR T therapy revealed that the patient had had multifocal thrombotic microangiopathy.

Much attention has been focused on IL-6 and CRP as possible biomarkers for real-time applications. Measuring IL-6 in real time is not widely available as a CLIA certified test and is expensive, which limits its usefulness as a biomarker at the present time. IL-6 signaling results in secretion of CRP from hepatocytes, and measuring CRP is, of course, readily available and inexpensive. However, CRP elevations are not specific to CRS, and changes in CRP tend to lag behind the clinical picture by as much as 24 hours. Therefore, CRP should not be relied upon to diagnose or manage CRS. Interestingly, CRP has been found to increase shedding of sIL-6R from neutrophils, possibly creating a feed-forward circuit of widespread cellular activation.

Risk Factors for Cytokine Release Syndrome

Several clinical risk factors may predispose a patient to severe CRS ( Table 5.2 ). In a phase I dose escalation trial of CD19 CAR T-cell therapy in children and young adults with ALL, both CRS severity and probability of tumor response correlated with higher CAR T-cell expansion in blood after infusion. Not surprisingly, patients receiving higher CAR T-cell doses had more severe CRS ^, ; thus, determining the ideal CAR T-cell dose to be infused (i.e., the biologically active dose rather than the maximally tolerated dose) is important to reducing CRS impact in CAR T-cell recipients. Along these lines, better selection of CAR T-cell subsets to be infused (i.e., selecting for central memory T-cells) may also be an important consideration in the management of CRS.

Patients with higher disease burden tend to have more severe CRS. ^, This suggests that disease cytoreduction prior to treatment with CAR T-cells may reduce the incidence of severe CRS, though data are not yet available supporting this hypothesis and many patients are already refractory to conventional therapy, making this variable often difficult to modify.

In a study by Hay et al., authors looked at risk factors for CRS in 133 adult patients who received CD19 CAR T-cells for NHL, CLL, and B-cell ALL ( Table 5.2 ). Approximately 70% of the patients had CRS, and only 10 patients (3.8%) each had grade 4 or 5 CRS. By multivariate analysis, patient risk factors associated with CRS include higher CD19+ tumor burden in the bone marrow and severe thrombocytopenia. CAR T product risk factors for any grade CRS included using bulk CD8+ T-cells without selection of the central memory subset and higher CAR T-cell dose infused. The addition of fludarabine to the lymphodepleting chemotherapy regimen also increased the risk of developing severe CRS. Higher CAR T-cell dose and fludarabine/cyclophosphamide lymphodepleting chemotherapy were associated with grade 4 or higher CRS. In a study of 32 patients with NHL treated with CD19 CAR T-cells by the same group, patients who were given fludarabine with their cyclophosphamide lymphodepleting regimen had increased expansion and persistence of CAR T-cells and higher complete response rates. However, lymphodepleting chemotherapy without fludarabine was determined to be less effective.

Clinical Presentation of Cytokine Release Syndrome

Very early attempts to define and grade CRS described its constellation of symptoms including fever, hypotension, hypoxia, neurotoxicity and other end organ damage. Transaminitis, for example, was specifically addressed. With continued experience, it became clear that these definitions and grading systems needed to be modified. CAR-associated neurotoxicity, now termed immune effector cell (IEC)–associated neurotoxicity syndrome (ICANS), can occur outside the temporal context of the other features of CRS suggesting a different pathophysiology. The most recent definition of CRS, arrived by a consensus of a large number of experienced investigators from all aspects of the CAR T-cell space, specifically identifies neurotoxicity and CRS as two distinct entities. The ASTCT consensus manuscript defines CRS as,

a supraphysiologic response following any immune therapy that results in the activation or engagement of endogenous or infused T-cells and/or other immune effector cells. Symptoms can be progressive, must include fever at the onset, and may include hypotension, capillary leak (hypoxia) and end organ dysfunction.

Many early studies of CD19 CAR T-cells began to characterize the clinical course of patients experiencing CRS. It should be noted that most of these studies used varying criteria for grading CRS, as discussed in more detail below, so direct comparisons of one product or experience to another across these trials are impossible. However, we can begin to understand many key features of CRS from these studies.

First, for CRS to be diagnosed, symptoms must begin in a reasonable temporal association to the therapy given. Usually, this occurs rather rapidly and is limited in its duration. In a phase 2 study of axicabtagene ciloleucel in 111 patients with refractory NHL, 93% of patients experienced any grade CRS with 13% of patients experiencing grade ≥3 CRS with all patients in this latter group experiencing fever, hypoxia, and hypotension ( Table 5.3 ). Median time to onset was 2 days with a range of 1–12 days and resolution taking a median of 8 days. In a separate study, of 75 pediatric patients with ALL treated with tisagenlecleucel, CRS of any grade occurred in 77% with a median time of onset of 3 days, range of 1–22 days, and a median duration of 8 days. 46% of patients in this study experienced grade 3 or higher CRS ( Table 5.3 ).

This timecourse is not unique to CD19 CAR T-cells. A phase I CAR T-cell study targeting B-cell maturation antigen in multiple myeloma reported no grade 4 or higher CRS and only two recipients with grade 3 CRS despite 76% of the 33 treated patients experiencing any grade CRS ( Table 5.3 ). Despite the lower CRS grade, median onset was 2 days, and the median duration was 5 days.

It should be noted that very early onset of fever has been associated with more severe CRS in some studies. For example, patients receiving CD19 CAR T-cells at FHCRC who developed grades 4–5 CRS had fever earlier (within 0.4 vs. 3.9 days), a higher temperature peak (40.4°C vs. 39.4°C), and longer duration (2.5 vs. 4.4 days) compared with those who developed grades 1–3 CRS.

In addition to fever, hypotension, and hypoxia, patients with severe CRS can present with a host of other symptoms ( Fig. 5.1 ). More common symptoms include tachycardia, tachypnea, hypoalbuminemia, and capillary leak syndrome, resulting in weight gain, respiratory insufficiency, and ultimately impaired hemodynamics. ^, Arrhythmias particularly in older patients have been noted. Many patients with grades 3–4 CRS will go on to develop organ dysfunction including acute kidney injury, coagulopathy, diarrhea, and hepatic injury manifested by transaminitis, hyperbilirubinemia, and elevated alkaline phosphatase. Given the severity of CRS in many patients and the potential for long-term morbidity, identification, prevention, and treatment strategies for CRS are needed.