Skip Navigation

Human Granulocytic Anaplasmosis and Macrophage Activation

  1. J. Stephen Dumler1,
  2. Nicole C. Barat1,
  3. Christopher E. Barat2, and
  4. Johan S. Bakken3
  1. 1Division of Medical Microbiology, Department of Pathology, The Johns Hopkins University School of Medicine, Baltimore
  2. 2Villa Julie College, Stevenson, Maryland
  3. 3St. Luke's Infectious Disease Associates, Duluth, Minnesota
  1. Reprints or correspondence: Dr. J. Stephen Dumler, Division of Medical Microbiology, Dept. of Pathology, The Johns Hopkins University School of Medicine, 720 Rutland Ave., Ross 624, Baltimore, MD 21205 (sdumler{at}jhmi.edu).

  1. Presented in part: 43rd Annual Meeting of the Infectious Diseases Society of America, San Francisco, CA, October 2005 (abstract 184).

 
Next Section

Abstract

Patients with human granulocytic anaplasmosis present with fever, thrombocytopenia, leukopenia, and an elevated aspartate transaminase level. Clinical and histopathologic features of severe disease suggest macrophage activation. Twenty-nine patients with human granulocytic anaplasmosis had higher ferritin, interleukin-10, interleukin-12 p70, and interferon-γ levels than did control subjects matched for age and sex; severity correlated with triglyceride, ferritin, and interleukin-12 p70 levels. Several severely affected patients had cases that fulfilled macrophage activation syndrome diagnostic criteria. Macrophage activation and excessive cytokine production may belie tissue injury associated with Ananplasma phagocytophilum infection.

Human granulocytic anaplasmosis (HGA) is a tick-transmitted infection that is caused by Anaplasma phago-cytophilum [1]. Seroepidemiologic studies suggest the frequent occurrence and broad distribution of HGA, as well as the potential for frequent subclinical infections [1, 2]. When ill, most patients present with undifferentiated fever, accompanied with thrombocytopenia, leukopenia, and laboratory evidence of liver injury [1,2–3]. Complications are infrequent but include hepatitis, peripheral neuropathy, myocarditis, interstitial pneumonitis, adult respiratory distress syndrome, a shock-like syndrome, and death [1,2–3]. The pathophysiological basis for clinical manifestations is unknown.

In vitro HGA models reveal evidence that infection subverts neutrophil function to promote inflammation while impairing neutrophil antimicrobial defenses [1, 4]. Similarly, both mouse and equine models demonstrate the disease's inflammatory basis but support the concept that inflammatory injury is not driven by bacterial load but rather by the host innate immune and/or inflammatory response [5,6,7,8,9,10,11–12]. Model infections also provide several important observations: (1) the in vivo target of A. phagocytophilum is the neutrophil; (2) the majority of infected cells are in the bloodstream, not in the tissues; (3) the cause of thrombocytopenia and leukopenia is questionably related to infectious destruction of hematopoietic elements; (4) bone marrow suppression is not an evident hematologic abnormality; and (5) activated, sometimes hemophagocytic macrophages suggest that severe HGA possesses some attributes of macrophage activation disorders [1, 2, 7]. To investigate this, we assessed blood and serum indices that are considered to be part of macrophage activation and hemophagocytic syndromes and compared results from patients with HGA with those from control subjects matched for sex and age, for whom a diagnosis of HGA was excluded.

Previous SectionNext Section

Materials and Methods

Samples and patients Serum samples were obtained from a -20° C archive, including samples from patients from northwest Wisconsin and Minnesota who had manifestations suggestive of HGA and samples from all patients for whom a sufficient volume of a serum sample was obtained during the acute phase of infection. Definitive HGA identification was based on ⩾1 of the following laboratory features: bacterial inclusions (morulae) in neutrophils during the acute phase, a 4-fold change in antibody titer during the convalescence phase, A. phagocytophilum DNA detected by PCR in blood samples obtained during the acute phase, and positive blood culture results for A. phagocytophilum. To adequately control for the variables examined, control serum samples from patients for whom HGA was ruled out as the cause of fever on the basis of negative blood smear microscopic results, absence of A. phagocytophilum antibodies during the acute and convalescent phases, and, for many control subjects, a negative PCR result and/or a negative blood culture result, were obtained from the same archive, matching for age (± 2 years) and sex [13]. All serum samples for the current analyses were obtained during the acute phase of disease while patients were symptomatic (from 0 through 9 days; median days, 0) after presentation for acute onset of fever, and on median day 6 of illness (first and third quartiles, 3 and 9 days, respectively; range, 1–48 days). All control subjects presented with fever and a history of recent tick-exposure or a bite and received a nondiagnostic physical examination. Although final diagnoses were not established for all control subjects, most subjects had a self-limited illness of presumed viral etiology. All samples were obtained from the subjects after informed consent was received, in accordance with Duluth Clinic and St. Mary's Hospital Internal Review Board approved protocol; approval from The Johns Hopkins Medicine Internal Review Board was also obtained.

Severity index. For all patients with HGA, overall severity was determined on the basis of the following factors: (1) hospitalization, (2) intensive care unit admission, (3) a duration of hospitalization <3 days, (4) fever (temperature, <38.5° C), and (5) death. The sum of the number of severity factors was used without additional weighting as the severity index for each patient.

Laboratory tests. Serum samples were assessed for concentrations of ferritin and triglycerides at the Clinical Chemistry Laboratory of The Johns Hopkins Hospital. Serum cytokine concentrations, including IFN–γ, IL-10, IL-12 p70, TNF-α, and IL-1β, were determined using a 5-plex fluorescent microbead assay (BioRad Laboratories) on a Luminex100 (Luminex) in duplicate or triplicate.

Statistical analyses. The distributions of results of all tests were initially examined for normality using Kolmogorov-Smirnov/Lilliefor, Shapiro-Wilk, D'Agostino Skewness, D'Agostino Kurtosis, and D'Agostino Omnibus tests to determine the appropriate statistical comparisons. Because none of the results were found to be normally distributed, subsequent analyses relied on nonparametric tests. For laboratory tests, median levels were calculated for samples from patients with HGA and control subjects; the median values in these groups were compared for significance using the Wilcoxon matched pairs test and shown in boxplots that illustrate the first and third quartiles, as well as the minimum and maximum values as estimators of variability. For some leukocyte and platelet counts and hemoglobin concentrations, several data points were not available, and for these, comparisons were made using the Mann-Whitney U test for independent variables. For severity index assessments, correlation with laboratory and cytokine tests was determined by Spearman's rank correlation coefficient. P values <.05 were considered to be statistically significant.

Previous SectionNext Section

Results

Patients and control subjects. Sufficient serum samples for the studies were available for 29 patients with corroborated HGA and 29 control subjects matched for sex and age who lacked any diagnostic features of HGA. All patients and control subjects presented with fever and received nonspecific physical examinations after a median incubation period (available for 17 patients) of 7 days (first and third quartiles, 5 and 9 days, respectively; range, 1–21 days). The median month of presentation for patients with HGA was June, with 75% of cases identified during June and July; these findings were nearly identical to those for the control subjects (for whom the median month of presentation was June, with 75% of cases occurring from June through August). The median age of patients with HGA and control subjects was 67 years (age range, 16–91 years for patients with HGA vs. 18–87 years for control subjects). The ratio of males to females was 1.4 : 1. of the 29 patients with HGA, 23 were hospitalized, and 4 required admission to the intensive care unit; the median duration of hospitalization was 5 days (range, 3–14 days). A fever (temperature, <38.5° C) was observed in 23 patients, and the overall severity of disease was classified as very mild in 3 patients, mild in 3 patients, moderate in 7 patients, moderately severe in 13 patients, severe in 1 patient, and fatal in 2 patients. Similar clinical data and assessments were not available for control subjects. Although patients with HGA and control subjects were similar with regard to demographic data, epidemiologic history, and clinical presentation, as anticipated, they differed with regard to hematologic findings; compared with control subjects, patients with HGA had more-frequent leukopenia (median leukocyte level, 4.0 × 109 leukocytes/L in patients with HGA vs. 6.6 × 109 leukocytes/L in control subjects; P = .002, by Mann-Whitney U test) and thrombocytopenia (median platelet count, 65 × 109 platelets/L in patients with HGA vs. 203 × 109 platelets/L in control subjects; P < .001, by Mann-Whitney U test), but hemoglobin concentrations were not different (median hemoglobin concentration, 13.1 g/dL in patients with HGA vs. 13.5 g/dL in control subjects; P = .372, by Mann-Whitney U test).

The overall results of ferritin, triglyceride, and cytokine tests for patients with HGA and control subjects are shown in figure 1. Briefly, compared with control subjects, patients with HGA had significantly higher serum concentrations of ferritin (median ferritin concentration, 548 ng/mL for patients vs. 251 ng/mL for control subjects; P = .018), IFN-γ (median IFN-γ concentration, 129 pg/mL for patients vs. 4.7 pg/mL for control subjects; P = .001), IL-12 p70 (median IL-12 p70 concentration, 49 pg/mL for patients vs. 6.8 pg/mL for control subjects; P < .001), and IL-10 (median IL-10 concentration, 164 pg/mL for patients vs. 3.7 pg/mL for control subjects; P < .001). Triglyceride levels were not different between patients with HGA and control subjects (P = .114). Only very low levels of both TNF-α and IL-1β were found in the archived serum samples, for which differences between patients with HGA and control subjects were not identified. In contrast, only the serum ferritin level (r = .503; P = .005), triglyceride levels (r = .478; P = .009), IL-12 p70 level (r = .455; P = .013), and the ratio of IL-10 to IFN-γ (r = .413; P = .026) were significantly correlated with disease severity in patients with HGA (figure 2). There was a highly statistically significant relationship between the production of ⩾1 cytokine produced as part of a proinflammatory or inflammatory regulation response. Neither age nor incubation period were correlated with severity.

Figure 1
View larger version:
Figure 1

Median serum concentrations (bars) of cytokines, ferritin, and triglycerides in samples from 29 patients with proven human granulocytic anaplasmosis (HGA), compared with those in samples from control subjects matched for age and sex. P values (determined by Wilcoxon matched pairs test) for each comparison are shown at the top of the graph. The boxes represent the limits of first and third quartiles, and the lines represent the range of all values. All values, except those for ferritin, are indicated on the left axis.

Figure 2
View larger version:
Figure 2

Correlation of triglyceride, ferritin, and IL-12 p70 levels, the ratio of IL-10 to IFN-γ, and epidemiologic findings with severity in 29 patients with proven human granulocytic anaplasmosis (HGA; by Spearman rank correlation). Each point represents an individual patient. Severity of 0 is mild, and severity of 5 is death. The line represents the least squares estimate for each panel; r and P values are shown in each panel.

Previous SectionNext Section

Discussion

The diversity of clinical manifestations associated with A. phagocytophilum infection is not well understood [1]. Although it appears that most infected persons recover without developing substantial clinical signs, one-third to one-half of those who become ill will be hospitalized, and of these persons, 7% will require admission to an intensive care unit, and 0.5% will die [2, 3]. The most frequent complication that leads to hospitalization is fever, in conjunction with pancytopenia and some degree of hemodynamic instability [1,2–3]. The occurrence of a septic or toxic shock—like illness is a rare but well recognized complication of infections with Anaplasmataceae species, such as Ehrlichia chaffeensis and A. phagocytophilum, in humans [1,2–3, 7]. However, it is very difficult to explain these manifestations, because a small proportion of leukocytes is infected, platelets are not infected, and these bacteria lack lipopolysaccharide and peptidoglycan [14]. Moreover, inflammatory manifestations, such as respiratory distress (for which the pathologic basis is interstitial pneumonitis and alveolar damage) and hepatitis, generally occur early during the course of infection, before substantial development of adaptive immune response is likely to occur [1, 2, 7].

Recent advancements in the understanding of innate immune responses and the complexity of proinflammatory regulation illustrate the central role of mononuclear phagocytes, such as macrophages. Shock syndromes with infections that relate to inflammatory mediator release can occur in a number of situations, including pathogen-associated molecular pattern—mediated septic shock and infection-associated hemophagocytic syndromes, among others [15, 16]. Although the signaling and molecular pathogenesis of lipopolysaccharide-induced septic shock is increasingly understood, only limited progress has been made with regard to the understanding of hemophagocytic syndromes and the related macrophage activation syndromes [15,16–17]. The current belief is that the shock induced with the latter syndromes relates to induction of effectors from macrophages, including cytokines (e.g., TNF-α and IL-1β), nitric oxide, superoxide anions, hypochlorite, and degradative enzymes, among others, as well as a nonspecific activation of phagocytic activity that leads to the characteristic morphology [15, 16]. Macrophages are activated via production of cytokines (e.g., IL-12 and IFN-γ) from activated cytotoxic cells (e.g., CD8 T lymphocytes, NK cells, and perhaps NKT cells) [15,16–17]. These cytotoxic cells ordinarily also serve a regulatory function by cytotoxic deletion of some activated macrophages, which serves to dampen the overall response and production of damaging mediators. This activity is diminished in cases of hemophagocytic and macrophage activation syndromes for reasons that are not entirely clear but generally relate to NK or CD8 T cell cytopenia or dysfunctional perforin generation [15,16–17].

A small number of prior studies involving humans with HGA and animal models of A. phagocytophilum infection demonstrate increased macrophage tissue infiltration and hemophagocytosis, suggesting activation of these cells [5, 7]. In addition, the current study supports the concept that HGA and disease severity relate to macrophage activation by virtue of biomarkers of macrophage activation and innate immune response, including serum ferritin, IFN-γ, IL-12, and IL-10 levels, during active infection [15, 16]. Although, typically, the triglyceride levels are also elevated because of the enhanced macrophage lipoprotein lipase activity, that feature was not present in the patients with HGA to a degree greater than that in control subjects; moreover, samples were not collected strictly during times of fasting. The concurrence of IFN-γ, IL-12, and IL-10 suggests a pattern of innate immune triggering that would be anticipated to initiate and cyclically maintain or dampen macrophage activation syndromes. In contrast, significant correlations between HGA severity and ferritin, IL-12, and triglyceride levels, but not the IFN-γ level (figure 2), could reflect the downstream effects of macrophage activation, even though triglyceride levels were not different from those in control subjects.

As anticipated, IFN-γ and IL-12 p70 expression were well correlated (r = .44; P = .016). With this finding in mind, one must question why IFN-γ rank does not correlate with severity, as would be expected with macrophage activation. Our observations could partially be explained by the fact that the serum samples for which cytokine levels were determined were obtained at a single point in time, after the onset of symptoms (median period from onset of symptoms to sample collection, 6 days), whereas the severity index reflects the cumulative effect of events over the total course of infection, including before and after the time that sampling occurred. Because IFN-γ rank was negatively correlated with time after presentation (when the sample was obtained; r = -.445; P = .016), its sampling could misrepresent the degree of macrophage activation over the entire course of each patient's illness. Moreover, the paradoxical correlation of severity with the ratio of IL-10 to IFN-γ could reflect high levels of IL-10 simultaneously expressed as part of a compensating regulatory cytokine cascade mediated by macrophage activation, as well.

Although also anticipated with macrophage activation, neither TNF-α, nor IL-1β, were detected at biologically significant levels in patients or control subjects. However, both are notoriously labile in archived serum, and their absence here could be artifactual, although these data do confirm limited earlier studies that indicate, at minimum, a role of proinflammatory cytokines in HGA [6]. The significance of this point is that, if these cytokines are not present, the immunopathogenesis of severe HGA could differ from that increasingly described for Ehrlichia species in murine models of human monocytic ehrlichiosis, in which TNF-α derived from CD8 T lymphocytes plays a critical role [18]. Our observations are consistent with data emerging from murine HGA models indicating that, regardless of bacterial load, A. phagocytophilum triggers innate immune responses and production of Th1 cytokines that are able to activate macrophages, resulting in the production of macrophage-dependent effectors, such as nitric oxide [5, 10, 12]. It is compelling to note that, among the 5 most severely affected patients, all had fever, cytopenia, elevated IFN-γ levels, and hepatic injury; 4 had a ferritin level <5000 ng/mL; 2 had triglyceride levels < 3 mmol (266 mg/dL); and both patients who died had hemophagocytic macrophages in the spleen and other organs [7, 19], consistent with current diagnostic criteria for macrophage activation and hemophagocytic syndromes [15, 16, 20].

Taken together, these data provide evidence that infection with A. phagocytophilum may represent a manifestation of macrophage activation and that clinical severity is, in part, dependent upon the degree of macrophage activation. The data further show that most patients with HGA do not have substantial macrophage activation, which explains mild disease courses; however, most severe infections concurrently develop clinical, histopathologic, and laboratory findings with many features of macrophage activation and hemophagocytic syndromes. Likewise, it is also clear that HGA is not a classic macrophage activation or hemophagocytic syndrome, as is true for sepsis, in which hemophagocytosis is often recognized [21]. The specific immunological trigger for the process is not clear, nor are the mechanisms or other cellular participants that lead to the activation. Prime candidates for this process now must include cytotoxic cells, such as CD8 T lymphocytes, NK lymphocytes, and even NKT lymphocytes, as suggested by recent investigations involving animal models [9, 22].

Previous SectionNext Section

Acknowledgments

Financial support. National Institutes of Allergy and Infectious Diseases (R01 AI41213).

Potential conflicts of interest. All authors: no conflicts.

  • Received June 13, 2006.
  • Accepted March 29, 2007.
Previous Section
 

References

  1. ↵
    1. Dumler JS,
    2. Choi KS,
    3. Garcia-Garcia JC,
    4. et al
    . Human granulocytic anaplasmosis Anaplasma phagocytophilum. Emerg Infect Dis 2005;11:1828-34.
    MedlineWeb of Science
  2. ↵
    1. Bakken JS,
    2. Dumler JS
    . Human granulocytic ehrlichiosis. Clin Infect Dis 2000;31:554-60.
    Abstract/FREE Full Text
  3. ↵
    1. Demma LJ,
    2. Holman RC,
    3. McQuiston JH,
    4. Krebs JW,
    5. Swerdlow DL
    . Epidemiology of human ehrlichiosis anaplasmosis in the United States, 2001–2002. Am J Trop Med Hyg 2005;73:400-9.
    Abstract/FREE Full Text
  4. ↵
    1. Carlyon JA,
    2. Fikrig E
    . Mechanisms of evasion of neutrophil killing by Anaplasma phagocytophilum. Curr Opin Hematol 2006;13:28-33.
    MedlineWeb of Science
  5. ↵
    1. Martin ME,
    2. Caspersen K,
    3. Dumler JS
    . Immunopathology ehrlichial propagation are regulated by interferon-γ interleukin-10 in a murine model of human granulocytic ehrlichiosis. Am J Pathol 2001;158:1881-8.
    MedlineWeb of Science
  6. ↵
    1. Dumler JS,
    2. Trigiani ER,
    3. Bakken JS,
    4. Aguero-Rosenfeld ME,
    5. Wormser GP
    . Serum cytokine responses during acute human granulocytic ehrlichiosis. Clin Diagn Lab Immunol 2000;7:6-8.
    Medline
  7. ↵
    1. Lepidi H,
    2. Bunnell JE,
    3. Martin ME,
    4. Madigan JE,
    5. Stuen S,
    6. Dumler JS
    . Comparative pathology, immunohistology associated with clinical illness after Ehrlichia phagocytophila-group infections. Am J Trop Med Hyg 2000;62:29-37.
    Abstract
  8. ↵
    1. Choi KS,
    2. Scorpio DG,
    3. Barat NC,
    4. Dumler JS
    . Msp2 variation in Anaplasma phagocytophilum in vivo does not stimulate T cell immune responses or interferon-gamma production. FEMS Immunol Med Microbiol 2007;49:374-86.
    CrossRefMedlineWeb of Science
  9. ↵
    1. Choi KS,
    2. Webb T,
    3. Oelke M,
    4. Scorpio DG,
    5. Dumler JS
    . Differential innate immune cell activation proinflammatory response in A. phagocytophilum infection. Infect Immun 2007;75:3124-30.
    Abstract/FREE Full Text
  10. ↵
    1. Browning MD,
    2. Garyu JW,
    3. Dumler JS,
    4. Scorpio DG
    . Role of reactive nitrogen species in development of hepatic injury in a C57bl/6 mouse model of human granulocytic anaplasmosis. Comp Med 2006;56:55-62.
    MedlineWeb of Science
  11. ↵
    1. Scorpio DG,
    2. Von Loewenich FD,
    3. Bogdan C,
    4. Dumler JS
    . Innate immune tissue injury murine HGA: tissue injury in the murine model of granulocytic anaplasmosis relates to host innate immune response not pathogen load. Ann N Y Acad Sci 2005;1063:425-8.
    CrossRefMedlineWeb of Science
  12. ↵
    1. Scorpio DG,
    2. von Loewenich FD,
    3. Gobel H,
    4. Bogdan C,
    5. Dumler JS
    . Innate immune response to Anaplasma phagocytophilum contributes to hepatic injury. Clin Vaccine Immunol 2006;13:806-9.
    Abstract/FREE Full Text
  13. ↵
    1. Bakken JS,
    2. Aguero-Rosenfeld ME,
    3. Tilden RL,
    4. et al
    . Serial measurements of hematologic counts during the active phase of human granulocytic ehrlichiosis. Clin Infect Dis 2001;32:862-70.
    Abstract/FREE Full Text
  14. ↵
    1. Lin M,
    2. Rikihisa Y
    . Ehrlichia chaffeensis Anaplasma phagocytophilum lack genes for lipid A biosynthesis incorporate cholesterol for their survival. Infect Immun 2003;71:5324-31.
    Abstract/FREE Full Text
  15. ↵
    1. Menasche G,
    2. Feldmann J,
    3. Fischer A,
    4. de Saint Basile G
    . Primary hemophagocytic syndromes point to a direct link between lymphocyte cytotoxicity homeostasis. Immunol Rev 2005;203:165-79.
    CrossRefMedlineWeb of Science
  16. ↵
    1. Grom AA
    . Natural killer cell dysfunction: a common pathway in systemic-onset juvenile rheumatoid arthritis, macrophage activation syndrome, hemophagocytic lymphohistiocytosis? Arthritis Rheum 2004;50:689-98.
    CrossRefMedlineWeb of Science
  17. ↵
    1. Jordan MB,
    2. Hildeman D,
    3. Kappler J,
    4. Marrack P
    . An animal model of hemophagocytic lymphohistiocytosis (HLH): CD8+ T cells interferon gamma are essential for the disorder. Blood 2004;104:735-43.
    Abstract/FREE Full Text
  18. ↵
    1. Ismail N,
    2. Soong L,
    3. McBride JW,
    4. et al
    . Overproduction of TNF-α by CD8+ type 1 cells down-regulation of IFN-γ production by CD4+ Th1 cells contribute to toxic shock-like syndrome in an animal model of fatal monocytotropic ehrlichiosis. J Immunol 2004;172:1786-800.
    Abstract/FREE Full Text
  19. ↵
    1. Walker DH,
    2. Dumler JS
    . Human monocytic granulocytic ehrlichioses. Discovery diagnosis of emerging tick-borne infections the critical role of the pathologist. Arch Pathol Lab Med 1997;121:785-91.
    MedlineWeb of Science
  20. ↵
    1. Janka G,
    2. zur Stadt U
    . Familial acquired hemophagocytic lymphohistiocytosis. Hematology Am Soc Hematol Educ Program 2005;2005:82-8.
    Abstract/FREE Full Text
  21. ↵
    1. Emmenegger U,
    2. Schaer DJ,
    3. Larroche C,
    4. Neftel KA
    . Haemophagocytic syndromes in adults: current concepts challenges ahead. Swiss Med Wkly 2005;135:299-314.
    Medline
  22. ↵
    1. Mattner J,
    2. Debord KL,
    3. Ismail N,
    4. et al
    . Exogenous endogenous glycolipid antigens activate NKT cells during microbial infections. Nature 2005;434:525-9.
    CrossRefMedline
| Table of Contents

This Article

  1. Clin Infect Dis. (2007) 45 (2): 199-204. doi: 10.1086/518834
  1. AbstractFree
  2. » Full Text (HTML)Free
  3. Full Text (PDF)Free

Classifications

Share

  1. Email this article

Navigate This Article

Current Issue

  1. March 1, 2012 54 (5)
  1. Clinical Infectious Diseases
  1. Alert me to new issues

Most