Sepsis in Patients with Febrile Neutropenia: A Clinical and Autopsy-Based Study

Introduction: Sepsis in febrile neutropenia (FN) is a life threatening condition, and a health problem of increasing proportions. Although multiple organ dysfunction syndrome (MODS) frequently precedes death in patients with sepsis, the ultimate mechanisms responsible for organ dysfunction and tissue damage in sepsis are yet to be determined. Currently, tissue damage is attributed to an exacerbated response of the immune and hemostatic systems, mediated by endothelial cells, platelets and neutrophils. Of note, recent evidence demonstrated that neutrophils, platelets and fibrin participate in this response by mediating neutrophil extracellular traps (NET) formation, and promoting the hemostatic containment of infectious foci. In animal models, down-regulation of NET formation, coagulation and platelet activation are usually associated with deficiencies in pathogen clearance. Unfortunately, activation of hemostasis and NET formation could potentially contribute to tissue damage by a process called "immunethrombosis". Although the increase of sepsis severity in patients with severe neutropenia is well described, the mechanisms of sepsis-associated tissue damage in the context of severe neutropenia/thrombocytopenia are yet to be determined.

Methods: In order to investigate the mechanisms of tissue damage in the context of severe neutropenia/thrombocytopenia, we compiled clinical data from two different prospective sepsis cohorts (A, neutropenic; n=129; and B, non-neutropenic; n=30) followed at our Institution. In addition, we reviewed histopathological data from 16 autopsies of individuals with hematological malignancies and septic shock from our institution (cohort C; n=16). H&E-stained slides from liver, kidneys and lungs were systematically analyzed by one investigator, and reviewed by 2 experienced pathologists, all of them blind to the presence or absence of neutropenia and thrombocytopenia. In each organ, we characterized (as present or absent) three main organic lesions: thrombi in microvessels, microorganism colonies, and inflammatory infiltrate (mononuclear and polymorphonuclear). Inflammation was graded as weak or intense, and only considered when there was no neoplastic infiltration.

Results: Median ages of patients from cohorts A and B were respectively 46.0 years (13-78), and 59.4 years (22-85); P<0.0001. In cohort A, neutrophil counts were < 100/mcl in 55.8% of patients and between 100-500/mcl in 42.6%. Platelet counts were also lower in cohort A (30,126 vs 213,933/mcl; P<0.0001). Median SOFA scores (at admission) were 4 (0-15) and 5 (0-17); P=0.3 in neutropenic and non-neutropenic patients respectively, and sepsis-related mortality was 22.5% and 10.3% in the same groups (P=0.19). Among patients with a higher SOFA score, mortality was higher in neutropenic patients (100% vs 33.3%; p=0.04). The frequencies of clinically-evident infection foci were 60% in cohort A and 100% in cohort B (P=0.0001). In contrast, positive blood cultures were present in 38% of neutropenic, but in only 3.3% of non-neutropenic patients (P<0.0001). The autopsy-based study included 10 patients with lymphoma and 6 with acute leukemia. The cause of death was septic shock in all of them, and three patients presented severe neutropenia (<500/mcl). The main histological findings are shown in table 1. The only neutropenic patient with microthrombi presented AML-M3 and leukostasis. Using H&E staining, no bacterial colonies were found in any slide.

Conclusions: as expected, septic patients with severe neutropenia presented a worse outcome compared to non-neutropenic patients in our cohort. In addition, the lower frequency of clinically-defined infectious foci, coupled with a strikingly higher frequency of positive blood cultures, suggest that severe neutropenia and thrombocytopenia could impair pathogen containment and clearance. Severe neutropenia/thrombocytopenia was compatible with inflammatory infiltrates and microvascular thrombosis in lungs, although the latter was only observed in a patient with leukostasis and promyelocityc acute leukemia.