Neutrophil-erythrocyte rosettes in autoimmune hemolytic anemia


  • Sol Schulman,

    Corresponding author
    1. Harvard Medical School, Boston, Massachusetts
    2. Department of Hematology, Massachusetts General Hospital, Boston, Massachusetts
    • Correspondence to: Sol Schulman, Harvard Medical School, 25 Shattuck Street, Boston, MA 02115. E-mail:

    Search for more papers by this author
  • Mark M. Awad,

    1. Department of Hematology, Massachusetts General Hospital, Boston, Massachusetts
    2. Dana-Farber Cancer Institute, Boston, Massachusetts
    Search for more papers by this author
  • David J. Kuter

    1. Harvard Medical School, Boston, Massachusetts
    2. Department of Hematology, Massachusetts General Hospital, Boston, Massachusetts
    Search for more papers by this author

  • Conflict of interest: Nothing to report.

Leukocyte erythrocyte rosettes can be a striking manifestation of autoimmune hemolytic anemia (AIHA). Many details of rosette formation in vivo are unknown because the phenomenon is encountered so infrequently, but in vitro studies suggest that rosettes are more likely to form with cells of the monocyte lineage than with other leukocytes [1-3]. Here, we present a rare case of severe autoimmune hemolytic anemia with neutrophil-erythrocyte rosettes on peripheral blood smear (1).

Figure 1.

Neutrophil-erythrocyte rosettes in a patient with autoimmune hemolytic anemia. A: Peripheral blood smear demonstrating rosettes around all neutrophils (50×). B: Peripheral blood smear emphasizing morphologic features of three rosettes at high power (100×). C: Peripheral blood smear highlighting the specificity of the rosettes for neutrophils, with two neutrophils surrounded by erythrocytes in close proximity to an uninvolved monocyte (100×).

A 19-year-old female presented to an outside hospital with low grade temperatures and fatigue. Examination was notable for jaundice, scleral icterus, II/VI systolic ejection murmur best heard at the left lower sternal border, and mild splenomegaly. After she was noted to have a hematocrit of 11%, haptoglobin below 10 mg/dL, lactate dehydrogenase of 269 U/L, and reticulocyte count of 15%, she was transfused four units of packed red blood cells (RBC) to a hematocrit of 29% and discharged on prednisone 60 mg daily. Six days later, she returned after an episode of post-micturition syncope and again had a hematocrit of 11% with total bilirubin of 9.6 mg/dL and direct bilirubin of 0.4 mg/dL. She was transfused one additional unit of RBCs and transferred to our hospital. Direct antiglobulin test (DAT) was weakly positive for IgG and 2+ for complement. While an eluate from her erythrocytes demonstrated a strong IgG pan-agglutinin, a plasma antibody panel reacted with K+ cells but did not reveal a pan-agglutinin, consistent with a low titer but high affinity autoantibody. An anti-i cold agglutinin with low thermal amplitude was also present. Antinuclear antibody (ANA) test and an extensive infectious workup were negative. After initially requiring supportive transfusion and prolonged prednisone therapy at 60 mg/day, she eventually tolerated a slow prednisone taper without return of clinically significant hemolysis.

Leukocyte erythrocyte rosettes can form via interaction of surface Fc receptors with IgG1 or IgG3 decorated erythrocytes. Such rosettes have been proposed to represent a physiologic intermediate in extravascular red blood cell destruction and strongly predict clinical AIHA [1, 4, 5]. Indeed, a defunct laboratory test for rosette formation between patient erythrocytes and sheep monocytes (the “monocyte monolayer assay”) had many limitations, but greater predictive value for clinically significant hemolysis than the standard Coombs' test [1, 5].

A limited number of reports describe neutrophil-erythrocyte rosettes [6, 7], as observed in this patient's peripheral blood smear (1A,B). This may be explained by in vitro studies demonstrating markedly different thresholds for rosette formation by monocytes and neutrophils, at approximately 200 and 5000 molecules of IgG per RBC, respectively [1] (for comparison, most DAT assays are standardized to become positive when there are 500 IgG molecules per RBC). These studies demonstrate that while monocyte-erythrocyte rosette formation is less efficient with IgG1 than IgG3, neutrophil-erythrocyte rosette formation with IgG1 is negligible even at very high titers [3]. We predict but are unable to prove that our patient's AIHA was due to one or more IgG3 antibodies, consistent with the high affinity but low titer antibody detected. In one prior report, neutrophil-erythrocyte rosette formation was IgG-mediated but temperature-dependent [6], and in another the rosettes turned out to be an artifact of EDTA tube collection that was not reproducible with other anticoagulants or finger-prick sampling [7], both important caveats to consider. The absence of rosettes around other leukocytes, particularly monocytes, is quite remarkable (1C).