CD19 CAR-T therapy and sepsis: dancing with the devil

Adoptive CAR–T-cell immunotherapy has emerged as a promising and novel therapeutic modality for a variety of cancers in recent years. On August 30, 2017, we witnessed the first US Food and Drug Administration approval of a gene therapy product, Kymriah (tisagenlecleucel), an autologous CD19 CAR-T product, and since then, a second CD19-based product, Yescarta (axicabtagene ciloleucel), has been approved. CD19 is universally expressed on most B-cell malignancies. Several groups have developed CD19 CAR–T cells and are at various stages of testing these products for efficacy and safety in patients. Although encouraging remissions are seen with different CD19 CAR-T products, complications such as cytokine release syndrome (CRS) and neurologic toxicities are getting more and more attention. More than 20 years ago, the observation of a sudden release of proinflammatory cytokines during graft-versus-host responses gave CRS its original name, “cytokine storm,” and, since then, CRS has been observed as part of many infectious diseases, ranging from streptococcal disease to influenza (reviewed in Tisoncik et al 2012).²  Thus, CAR therapy appears to engage the same devilish tools occurring in events more commonly seen in sepsis. With more patients receiving CD19 CAR-T treatment, there is increasing observation of infectious diseases associated with CAR-T treatment. Brentjens et al published the first report of sepsis as the potential culprit responsible for a chronic lymphocytic leukemia patient death 40 hours following CD19 CAR–T-cell infusion, although no identifiable pathogen was isolated at autopsy.³  Nevertheless, Kochenderfer et al recently reported low incidence of severe infection in the 4 diffuse large B-cell lymphoma patients with long-term CR after anti-CD19 CAR–T-cell therapy, despite long periods of B-cell aplasia and hypogammaglobulinemia.⁴  Nevertheless, it is reasonable to question the role played by sepsis either in setting the stage for CRS or for worsening its outcome. The authors of this latest study have nicely examined the incidence, onset, types of pathogens, and resolution of infection over time in relationship to CAR-T therapy.

There are several factors predisposing patients to infection while receiving CAR-T therapy. In addition to the pretreatment factors such as impaired immune function and tissue damage from prior chemotherapy regimens, these patients also experience cytopenia caused by the lymphodepletion preceding CAR–T-cell infusion, use of immunosuppressive drugs such as steroids and tocilizumab, possible intensive care unit stay, and subsequent hypogammaglobinemia due to prolonged B-cell aplasia. Intriguingly, this study shows the incidence of infections is comparable to reports from clinical trials using chemo/immunotherapies in similar patient populations. As expected, Hill et al show an increased risk of infections in CAR-treated patients with acute lymphoblastic leukemia, in those having >4 prior lines of antitumor regimens, and in those receiving the highest CAR–T-cell doses. Importantly, the multivariable analysis indicated that only the severity of CRS was a factor associated with infection after CAR-T therapy. Of interest, the incidence of infection was significantly reduced in patients receiving an optimized regimen to reduce the severity of CRS, and it remains unclear how severe CRS leads to infections. A recent study by Hay et al using the same patient population showed severe CRS is characterized by endothelial activation with increased angiopoietin-2 and von Willebrand factor in the blood.⁵  It is plausible that CRS-induced endothelial damage might initiate or facilitate infectious processes. Hill et al support this by the observation that in patients with grade 4 or above CRS, the majority (73%) of the infections occurred after the CRS severity grade had peaked. Reducing the severity of CRS should result in less endothelial damage, which in turn might contribute to lower incidence of infection.

With 2 commercial CD19 CAR-T products available and many more in development, the total number of patients receiving CD19 CAR-T cells is expected to increase exponentially in the near future. It is crucial to keep in mind that, although not all fevers are due to CRS, the infection rate was 21% during the first 28 days following CAR-T infusion in this study, and thus prophylaxis of infection applies here as it does in other immunosuppressed patients. It is possible that prophylaxis might eventually include other immune-modulating therapies, such as tocilizumab, steroids, or agents yet to be learned. However, the key to prevention of CRS, like that of irreversible sepsis, will require a better understanding of its pathogenesis. We suggest that management will require interventions based on control of proinflammatory and anti-inflammatory processes. For example, research such as that shown in a recent report on the effects of human resistin⁶  on the early events of sepsis could be extended to the search for improved prevention and management of CRS.

Given the short follow-up of this study, it is unknown if these CAR–T-treated patients retain an immunodeficiency similar to that of patients post–autologous transplant, requiring revaccination. Future studies incorporating evaluation of the host cellular immunity late after CAR-T treatment will shed light on this issue. A limitation of this study is that it is a single-center study. All patients received treatment using a defined CD19 CAR–T-cell product derived from the patient’s own T cells. Therefore, future studies examining the incidence of infection in patients receiving different CD19 CAR–T-cell products will be needed to determine if the findings in this study are generalizable. Until then, the devil we know is better than the devil we do not, and improvement in prevention and management of CSR after CAR–T-cell therapy will likely emerge side by side with improvements in control of sepsis.