HLH or sepsis: the truth is in the T cells

In this issue of Blood, Chaturvedi et al examined the T-cell phenotype in a cohort of patients with hemophagocytic lymphohistiocytosis (HLH) vs a cohort with sepsis. They found that, in HLH, there is expansion of a subset of peripheral blood T cells that express CD38 and HLA-DR.¹  This profile is most striking for CD8⁺ T cells, which also exhibit an effector memory phenotype and type 1 cytokine bias. HLH and other cytokine storm syndromes have emerged as areas of importance and interest in the hematology, as well as in the critical care, community. Additionally, there has been growing recognition of the phenotypic overlap between HLH and sepsis, with some sepsis patients even being described as having features of HLH or macrophage activation syndrome (MAS; defined as liver dysfunction, hyperferritinemia, and hematologic failure).²  However, phenotypic overlap does not necessarily mean pathobiological similarity.

CD8⁺ T cells have always been in the limelight when contemplating the pathogenesis of primary (genetic, inherited) HLH; however, CD8⁺ T cells have often been ignored in the study of the mechanisms involved in sepsis, with the exception of the pathologic immune system downregulation known as immunoparalysis that often follows severe sepsis.^2,3  In murine models of primary HLH, significant and prolonged CD8⁺ T-cell activation resulting from continued stimulation by activated antigen-presenting cells leads to the profound release of proinflammatory cytokines, such as interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α), interleukin-2 (IL-2), and IL-6.^4,5  Collectively, these cytokines mediate much of the morbidity and mortality associated with HLH.

Although immunologic data from murine models do not always translate fully in human diseases, the findings of Chaturvedi et al mirror what has been described in primary HLH murine models. Specifically, in the HLH patient samples examined, there was evidence of significant T-cell activation with increased proportions of a unique subset of CD38^high/HLA-DR⁺ cells, with this phenotype more striking for CD8⁺ cells vs CD4⁺ cells (see figure). These CD38^high/HLA-DR⁺ CD8⁺ T cells exhibited an effector memory phenotype marked by type 1 differentiation and increased release of IFN-γ and TNF-α. They were proliferative and exhibited reduced CD5⁺ expression, suggesting persistent antigen stimulation. Notably, CD38^high/HLA-DR⁺ CD8⁺ T cells were not restricted to the blood; they were present in other tissues, such as cerebrospinal fluid, lymph nodes, and bone marrow. Although the proportions of CD38^high/HLA-DR⁺ CD8⁺ cells did not correlate with clinical or laboratory markers of HLH activity at diagnosis, they decreased following HLH treatment, consistent with a reduction in disease activity. Strikingly, the proportions and phenotype of CD38^high/HLA-DR⁺ T cells in patients with sepsis were essentially no different from those in healthy controls.

These findings are congruent with previous findings of reduced HLA-DR expression on monocytes and T cells in sepsis, which has been associated with increased sepsis severity and, in some cases, the development of immunoparalysis.^3,6  Notably, when using a cutoff >7% circulating CD38^high/HLA-DR⁺ CD8⁺ T cells, HLH and early severe sepsis were accurately distinguished from each other with a sensitivity of 100% and specificity of 89%. The findings of Chaturvedi et al are compelling given the challenge of diagnosing HLH and distinguishing it from sepsis, particularly in patients in the critical care unit. With further clinical validation, the assessment of CD38^high/HLA-DR⁺ CD8⁺ T-cell proportions in the blood may enable a rapid and more accurate differentiation of HLH from sepsis. If so, the use of flow cytometry to characterize T cells could lead to earlier therapeutic interventions. Likewise, it will be a useful tool for assessing disease response. The information presented is relevant when considering frontline treatments for HLH, such as etoposide plus dexamethasone or anti-thymocyte globulin, which ablate activated T cells.^7-9  Their findings confirm that T-cell activation is a central feature of murine and human HLH, supporting the use of these agents in the treatment of HLH.

In an earlier study, Amman et al characterized T-cell activation within different HLH subgroups. They found elevated proportions of HLA-DR⁺ T cells in patients with primary or virally triggered secondary HLH.¹⁰  However, T-cell HLA-DR expression was not significantly upregulated in patients with nonvirally induced secondary HLH, mostly patients with MAS.¹⁰  These findings are not surprising given the predominance of innate immune system activation in MAS, which is similar to that observed in sepsis.⁴  These differences suggest that the characteristics of the T-cell activation in secondary HLH may vary depending on the underlying trigger. Thus, possible differences in T-cell HLA-DR expression and other markers, such as CD38 expression, should be evaluated further in different HLH subgroups.

Despite their clinical similarities, it is important to consider immunologic nuances between HLH and sepsis, especially as novel targeted therapies are developed and tested. In light of the study by Chaturvedi et al, therapies targeting T cells might be harmful if given to patients with early sepsis and might exacerbate the later development of immunoparalysis. On the other hand, therapies targeting cytokines released primarily by innate immune cells might be insufficient in the management of HLH. The question then arises, how does a provider marry the broadening knowledge of disease pathophysiology and therapeutic options when confronted with a patient presenting with hyperinflammation with an unclear underlying etiology? Flow cytometric assessment of CD38 and HLA-DR expression on CD8⁺ T cells might provide data to assist in making the correct diagnosis. Alternatively, one could consider initial treatment with agents that target cytokines more broadly, such as dexamethasone with or without the addition of a JAK inhibitor. Looking to the future, clinical trials should be designed to produce strong clinical and outcome data, as well as robust immunologic information that will better elucidate the underlying disease pathophysiology and direct the best targeted therapy.