Diagnosis and management of primary autoimmune neutropenia in children: insights for clinicians

Abstract

Autoimmune neutropenia of infancy (AIN), also called primary autoimmune neutropenia, is a disease in which antibodies recognize membrane antigens of neutrophils, mostly located on immunoglobulin G (IgG) Fc receptor type 3b (FcγIIIb receptor), causing their peripheral destruction. It is the most frequent type of neutropenia in children under 3–4 years of age and in most cases shows a benign, self-limited course. The diagnosis is based on evidence of indirect antineutrophil antibodies, whose detection frequently remains difficult. In this review we have analyzed the literature regarding AIN and present our personal experience in diagnosis and management.

Introduction

Neutropenia is a relatively frequent hematological abnormality in childhood and in Caucasians after the first year of life it is typically classified as severe (neutrophils <0.5 × 10⁹/l), moderate (between 0.5 and 1.0 × 10⁹/l) and mild (between 1.0 and 1.5 × 10⁹/l) [Dinauer, 2003]. In western populations the lower limit of neutrophils is 1.5 × 10⁹/l from >1 year to adulthood whereas it is1.0 × 10⁹/l in children from 2 weeks to 1 year of life [Dinauer, 2003]. It should also be noted that American Blacks [Reed and Diehl, 1991], Blacks of South African ancestry [Shoenfeld et al. 1988], Mexican-Americans [Hsieh et al. 2007], Caribbean Blacks [Bain, 1996], Yemenite Jews and some Arab populations [Shoenfeld et al. 1988; Hsieh et al. 2007; Bain, 1996; Weingarten et al. 1993] may have low normal limits of absolute neutrophil count (ANC), inferior than those observed in Caucasians. In newborns until 2 weeks of life there is a great variability in the ANC, related to sex (females have higher ANC than males), gestational age, type of delivery and possible intrauterine growth retardation [Schmutz et al. 2008; Wirbelauer et al. 2010].

Autoimmune neutropenia of infancy/childhood is a relatively frequent cause of neutropenia in children: the median age at diagnosis is 7–9 months [Lalezari et al. 1986; Bux et al. 1998; Wang et al. 2009]. The historical incidence is reported to be 1 out of 100,000 children under 10 years of age [Lyall et al. 1992] but, due to the benign course of the disease, there is clear evidence of underreporting, highlighted by frequent fortuitous findings (8–27% of all cases) [Bux et al. 1998; Audrain et al. 2011]. In our experience, diagnosis as consequence of a blood count planned for other reasons (i.e. surgery or pallor) is at least 30% of the total, and it is our opinion that the unexpected finding of a neutropenic child below 3–4 years of age most likely unveils a diagnosis of autoimmune neutropenia of infancy. There is no clear sex difference in incidence rate between males and females [Bux et al. 1998; Wang et al. 2009]. Most patients recover by 4–5 years of age and in about 90% resolution occurs before 2 years of duration [Bux et al. 1998]. Serious infections occur only in about 12–20% of affected children [Bux et al. 1998; Fioredda et al. 2013]. A mild associated leukopenia is possible and about a quarter of children present monocytosis [Bux et al. 1992b, 1998]. Based on the above considerations, autoimmune neutropenia of infancy looks very different from autoimmune neutropenia of older children and adulthood, since the latter is characterized by a more severe clinical course, by a higher frequency in females and by a far lower tendency to spontaneous recovery; it is also frequently associated with other autoimmune disorders [Bussel and Abboud, 1987; Capsoni et al. 2005].

Table 1 shows human neutrophil antigens (HNAs); the 11 antigens described to date have been identified on five polymorphic proteins of the granulocyte membrane. HNA-1 (FcγIIIb receptor), exclusively expressed on neutrophils, is the most immunogenic glycoprotein on the granulocyte membrane and has four isoforms, encoded by at least three alleles. HNA-1a and HNA-1b constitute the principal antigens implicated in autoimmune neutropenia of infancy (AIN). Their frequency varies among genetically different populations (Table 2) and the same phenomenon is described in all HNA antigens. On average, 4% of individuals express HNA-1a, HNA-1b and HNA-1c [Steffensen et al. 1999] on the neutrophil surface and about 0.1–2% of the general population are HNA-1a-1b-1c null [Hessner et al. 1996; Steffensen et al. 1999; Hauck et al. 2011; Porretti et al. 2012]. The target of the autoantibodies is related to the various expression level of specific neutrophils antigens among people of different ethnic background and also to the phase of the disease, since it seems that specificity against HNA-1a or HNA-1b appears later in time [Bruin et al. 2005].

Table 1.

Human neutrophil antigens (HNAs).

Glycoprotein	Antigen	Old nomenclature
FcγRIIIb (CD16b)	HNA-1a	NA1
	HNA-1b	NA2
	HNA-1c	NA3-SH
	HNA-1d
Gp 58–64 (CD177)	HNA-2a	NB1
Choline transporter-like protein 2	HNA-3a	5b
	HNA-3b	5ba
CD11b	HNA-4a	MART
	HNA-4bw
CD11a	HNA-5a	OND
	HNA-5bw

Table 2.

HNA-1a and HNA-1b frequency.

	HNA-1a	HNA-1b
Western Japan	–	99.5%
Asian Indians	30%	70%
Black (USA)	31%	69%
White (USA)	37%	63%
Turkey	42%	56%
Italy	49%	84%
Tunisian	52%	86%
Hispanic (USA)	53%	47%
Native Americans (USA)	55%	45%
Brazil	65%	83%
Chinese	91%	54%

Based on Hessner et al. [1996], Steffensen et al. [1999], Abid et al. [2000], Kissel et al. [2000], Taniguchi et al. [2002], Norcia et al. [2009], Hauck et al. [2011] and Porretti et al. [2012].

HNA, human neutrophil antigen.

Diagnosis

The granulocyte-specific antibodies causing neutropenia are directed against the cell surface membrane and have no relationship with antineutrophil cytoplasmic antibodies (ANCAs) in the vast majority of cases. It is important to highlight that there are frequent difficulties in the detection of autoantibodies causing AIN due to the labile nature of granulocytes (not storable without activation and consequent autolysis) and, in this field, expertise is restricted to a limited number of specialized laboratories [Lucas et al. 2013]. The direct granulocyte test (to discover antibodies anchored on the patient’s neutrophils) has an elevated number of false positives, principally due to an unspecific binding of immunoglobulin G (IgG) immune complexes to the Fcγ receptors or an insufficient number of isolated granulocytes. This is why, if a direct test is positive, it cannot be considered proof that granulocyte-specific antibodies are really present [Verheugt et al. 1978; Badolato et al. 2004]. On the contrary, a contingent negative direct test may be useful to presumably exclude AIN, since false negatives are extremely rare. The indirect test (detecting free granulocyte-specific antibodies in the patient’s serum after reaction with donor neutrophils) presents a very low false positive rate. However, there is a significant frequency of false negatives [Bux et al. 1998]. The low levels of autoantibodies and the possible lack of coverage of the full HNA pattern in the neutrophil suspension are the major determinants of the limited sensitivity of the indirect test. To reduce the false negative rate of indirect testing, the neutrophil panel should include at least one homozygous HNA-1a/1a and one HNA-1b/1b test cell, since the difficulty of autoantibodies binding to granulocytes coming from HNA-1a/1b heterozygous donors is well known. In addition this limit could be amplified by the variability of HNA-1 neutrophil surface expression associated with gene copy number variation [Willcocks et al. 2008]. The availability of selected homozygous donors is currently limited. For example, in Italy the antigen frequency of HNA-1a is about half that of HNA-1b [Porretti et al. 2012], and so it is unlikely to find neutrophils homozygous for HNA-1a expression in an indirect test performed using unselected donors, but at the same time, it is troublesome to have available homozygous HNA-1a individuals every time an assay has to be performed.

With the goal of increasing the possibility of detecting autoantibodies, reports from the International Granulocyte Immunology Workshop [Lucas and Carrington, 1990; Lucas et al. 2013] suggest the use of a combination of the granulocyte immunofluorescence test (GIFT) and the more time-consuming granulocyte agglutination test (GAT): GIFT is generally more sensitive, except for some antigens, such as HNA-3a (the major culprit in transfusion related acute lung injury), more readily detectable in GAT but not usually involved in cases of AIN. In both GIFT and GAT it is impossible to identify the specificity of a granulocyte reactive antibody, even though this is not fundamental for the ‘simple’ diagnosis of AIN. Thanks to a fluorescein isothiocyanate (FITC) conjugated antibody, in GIFT it is possible to identify antibodies bound to the neutrophil surface either by flow cytometry or by fluorescence microscopy. Only microscopic evaluation is able to recognize fluorescence signals originating from damaged cells as false positives and to distinguish the fluorescence reaction of antigen-bound antibodies from that coming from nonspecific adsorption of immune complexes. However microscope detection is more time-consuming, requires a lot of expertise, and consequently has high interobserver variability. Flow cytometry overcomes these pitfalls but still has some drawbacks:

1. Sometimes interpretation is made difficult by high background reactivity.

2. Sometimes the assessment of antibody specificities is limited by the presence of immune complexes or anti human leukocyte antigen (anti-HLA) antibodies that may also bind to granulocytes.

For all these reasons some reference laboratories [Lucas and Carrington, 1990; Lucas et al. 2013] also use monoclonal antibody immobilization of granulocyte antigens (MAIGA), a technique that allows the simultaneous determination of antibody specificity directed at a variety of antigens. Currently MAIGA is a laborious and time-consuming test, not suitable for antibody screening and rarely used in routine practice. Moreover it seems that GIFT is more sensitive than MAIGA [Bruin et al. 2005], and consequently since new techniques such as mixed passive hemagglutination [Han et al. 2006], transfected cell lines expressing neutrophil antigens [Yasui et al. 2007, 2008; Woźniak et al. 2012], soluble recombinant HNA production by transfection of insect cells [Bayat et al. 2009; Werth et al. 2013] or the microbead test [Fromont et al. 2010] have not yet definitely proved their efficacy or are not yet commercially available, the majority of laboratories worldwide analyse patient sera by the flow cytometry assay.

In keeping with this the diagnostic guidelines published by the Marrow Failure Study Group of the Associazione Italiana Onco-Ematologia Pediatrica (AIEOP) [Fioredda et al. 2011] rely on indirect antineutrophil antibody detection by flow cytometry immunofluorescence for the diagnosis of AIN. Obviously the sensitivity of indirect testing further increases on repeated determinations, and so in case the first test is negative but there is still a clinical picture consistent with AIN, the Italian guidelines recommend repeating the test up to four or more times over a time-span of 4–6 months or longer. If all four tests are negative and no maturational arrest can be found in bone marrow, the final diagnosis of ‘idiopathic neutropenia’ will be established. Due to the limited sensitivity of the indirect test, particularly if nongenotyped donors are used, it is common not to achieve a firm diagnosis of autoimmune neutropenia and many of these children are called ‘chronic/idiopathic benign neutropenia of infancy’ (or similar), but in the vast majority they are infancy autoimmune neutropenias. In these cases a definitive diagnosis will be reached only ‘ex post’, that is, after the highly frequent spontaneous recovery and after having excluded a postinfection neutropenia.

A frequent doubt is whether to perform a bone marrow evaluation. In most cases bone marrow is not informative [Bux et al. 1998], even though normal/hyper cellular marrow generally with a late maturational arrest may support the diagnosis. The presence of autoantibodies against myeloid precursors is the theoretical basis on which it is possible to explain the occasional appearance of myeloid hypoplasia [Currie et al. 1987; Hartman et al. 1994]. In our opinion it is possible to postpone the procedure until after at least the result of the first autoantibody assay if the child’s age is typical, if there is no other associated cytopenia, if there are no severe infections in the medical record, and if there are no clinical or laboratory criteria for suspicion of leukemia.

Differential diagnosis

There are many conditions that may enter in differential diagnosis (DD) with AIN, also in the newborn, even though AIN is unusual before 1 month of age [Bux et al. 1998]: there are only three reported patients with such an early appearance [Calhoun et al. 2001; Lejkowski et al. 2003]. We have observed some other cases (unpublished data), thus confirming that a presentation before 1 month of life, although infrequent, is not impossible.

The differential diagnosis of AIN includes:

• Infection-induced neutropenia (probably the most common type of neutropenia in newborns and infants). In addition to the different clinical pictures and the contribution of cultural, serological and genetic methods for identification of infectious agents, an efficient tool for DD is, in the newborn, a ratio of immature to total neutrophils higher than 0.2–0.3 (immatures, or bands, are intended neutrophils with a single-lobed nucleus and forms less mature than bands) that it is reported to have a strong predictive negative value [Engle and Rosenfeld,1984; Hornik et al. 2012]. The duration is usually less than 1–3 months [Sheen et al. 2009; Alexandropoulou et al. 2013] and therefore, a diagnosis of infection-related neutropenia can be considered appropriate if after 8–12 weeks ANC returns to normal and no other cause has become evident. Nevertheless it must be emphasized that some infections, i.e. those caused by human immunodeficiency virus (HIV) or human hepatitis C virus (HCV), may be associated with long-lasting neutropenia [Sheehan et al. 2013; Shi et al. 2014].

• Pregnancy or delivery-related neutropenias. Neutropenia is present in 67% of infants with asphyxia [Engle and Rosenfeld, 1984], 50% of those whose mothers had pregnancy-induced hypertension [Engle and Rosenfeld, 1984], 50% of newborns affected by Rh hemolytic disease [Koenig and Christensen, 1989; Blanco and Johnston, 2012], and in the donor twin in the twin–twin transfusion syndrome [Koenig et al. 1991]. In these cases the key element is the knowledge of this association, since the clinical picture is sufficient to establish a correct diagnosis.

• Alloimmune neonatal neutropenia (ANN) is another condition that must be considered if the appearance of neutropenia is in the first days of life: there is a fetomaternal granulocyte mismatch and an alloimmunization of the mother against the neutrophil antigens. Severe infections can be present in about 20% of patients [Bux et al. 1992a; Rodwell et al. 1996; Curtis et al. 2005]. A positive crossmatch between maternal sera and paternal neutrophils confirms the diagnosis.

• Neonatal alloimmune neutropenia secondary to maternal autoimmune neutropenia: there are few reports in literature [van Leeuwen et al. 1983; Kameoka et al. 2001; Davoren et al. 2004; Fung et al. 2005]. The differential diagnosis is simple if the type of neutropenia in the mother has been assessed correctly.

• Severe congenital neutropenias (SCN) are a heterogeneous group of diseases frequently complicated by malformations, dysmorfic features and metabolic problems. The issue is broad and falls outside the scope of this review. However, it is useful to note that SCN is far rarer (1–2 per million neonates) than AIN, that ANC is very often less than 0.5 × 10⁹/l, that there is almost constant need for granulocyte colony stimulating factor (G-CSF) administration, that bone marrow picture typically shows a blockage at the promyelocyte level and the infectious load is far higher than that observed in AIN [Fioredda et al. 2013]. All this makes differential diagnosis easy in the vast majority of cases [Kostmann, 1975; Bohn et al. 2007].

• Autoimmune neutropenia associated with other autoimmune diseases (often called secondary autoimmune neutropenia). In an important paper by Marie Bruin and colleagues on fewer than 30 patients [Bruin et al. 1999], the main differences between primary and secondary autoimmune neutropenia were the age of appearance (much later in the secondary type) and the spontaneous recovery (virtually impossible in the long term in secondary autoimmune neutropenia). Evans syndrome and autoimmune thrombocytopenia are the most frequently associated diseases in childhood, but neutropenia can complicate other autoimmune disorders such as systemic lupus erythematosus even though in some cases the mechanisms of cell depletion remain obscure and cannot be clearly ascribed to autoimmunity. It must be emphasized that neutropenia sometimes appears years before other nonhematologic manifestations of the underlying disease; for example, we observed a girl suffering from apparent ‘idiopathic neutropenia and leukopenia’ for 6 years and in whom antineutrophil antibodies were found only after an increase in antinuclear antibodies (ANAs).

• Neutropenia associated with immunodeficiency. Neutropenia can be associated with a deficit of both innate and acquired immunity but in most cases the mechanism is not autoimmune. Two immunodeficiencies in which neutropenia is mainly autoimmune are HypeIgM syndrome [Seyama et al. 1998; Jasinska et al. 2013] where neutropenia is observed in 60% of patients, and Good syndrome [Notarangelo, 2010], characterized by the association of hypogammaglobulinemia with thymoma, where autoimmune cytopenias (especially red cell aplasia and/or neutropenia) are common. Neutropenia occurs also in a minority of patients with common variable immunodeficiency in which may remain the only symptom for long time.

• Neutropenia secondary to drug administration. Drugs are the most common cause of an apparently idiopathic severe neutropenia. Few cases [Meyer et al. 1999] of ‘drug induced agranulocytosis’ (DIA) are definitely autoimmune and direct damage to the myeloid precursors probably plays a role in the majority of patients. Unfortunately, it is very difficult to prove the effective pathogenesis since serum samples should be tested in the absence and the presence of the drug, but this method lacks standardization. It is important to note that only about 10% of cases have been reported amongst children and young adults, and the vast majority of patients are older than 60 years of age [Andrès and Maloisel, 2008]. The duration of drug exposure before onset of agranulocytosis is extremely variable (2–60 days) and for 71% of patients it is over 1 month [Andersohn et al. 2007]. The time between appearance of agranulocytosis and normalization of ANC is 4–24 days [Andersohn et al. 2007]. In contrast to AIN, hypoplastic bone marrow is found in about two-thirds of cases [Andersohn et al. 2007]. In childhood the most frequently implicated class of medications is that of anti-epileptic agents, with frequent reports of leukopenia and/or neutropenia associated to carbamazepine [Asadi-Pooya and Ghetmiri, 2006] or valproate [Rahman et al. 2009] administration.

• Autoimmune neutropenia associated with neoplasm. It is very rare and most reports are of thymic neoplasms [Coudurier et al. 2008] and Hodgkin’s lymphoma [Lechner and Chen, 2010], but association with many other neoplasms has been reported [Carrington et al. 1990; Nakabe et al. 2002; Lamy and Loughran, 2003; White et al. 2003].

• More rarely neutropenia can be related to vitamin B12, folate or copper deficiency whose decreased serum levels orient the diagnosis [Sarode et al. 1989; Gabreyes et al. 2013; Atay et al. 2014].

Some elements of differential diagnosis are summarized in Table 3.

Table 3.

Elements for differential diagnosis.

	Infections	Splenomegaly/hepatomegaly	Growth retardation	Anemia and/or thrombocytopenia	Typical aspects
AIN	Mild	No	No	No	Age and fortuitous finding
Leukemia	–	Yes	No	Yes	LDH ↑
Secondary autoimmune neutropenia	Mild/moderate	Sometimes*	No	Possible*	Other autoimmunity
SDS	Mild/moderate	Yes	Yes	Frequent	Pancreatic Insufficiency
GSDIb	Moderate	Hepatomegaly	Yes	No	Hypoglycemia, ketoacidosis
Pearson Syndrome	Severe	Possible hepatomegaly	Yes	Yes	Lactate ↑, acidosis
SCN (Elane, Hax1 …)	Severe	No	Occasional	No	Possible dysmorphisms/ malformations
CN	Mild and Cyclic	No	No	No	Aphthous stomatitis
Post infection neutropenia	–	Possible	No	Possible (usually mild)	Medical history
Drug induced neutropenia	Moderate/severe	No	No	No	Medical history

^*Especially in Evans syndrome.

AIN, autoimmune neutropenia of infancy; CN, cyclic neutropenia; GSDIb, glycogen storage disease type Ib; LDH, lactate dehydrogenase; SCN, severe congenital neutropenia; SDS, Shwachman–Diamond syndrome.

Management

Severe infectious complications are overall less frequent in AIN than in genetic neutropenias [Fioredda et al. 2013]. Consequently, AIN children should have a normal lifestyle. In case of fever, we recommend that parents call the family doctor to examine the child. We suggest general practitioners make early use of antibiotics such as amoxicillin ± clavulanate. If an underlying immunodeficiency is ruled out, vaccinations with live vaccines can be performed. If an immunodeficiency it is not excluded then killed vaccines are preferred. Some authors suggest an antibiotic prophylaxis with trimethoprim/sulfamethoxazole that seemed efficient in reducing the frequency of infections [Kobayashi et al. 2003]. But since severe infections are very rare and the wide use of antibiotic may trigger microbial resistance, the Italian Marrow Failure Study Group does not recommend routine antibiotic prophylaxis [Fioredda et al. 2012], with the only possible exception of children living very far away from the nearest hospital. Granulocyte-colony stimulating factor (G-CSF) at an initial dose of 1–2 µg/kg/day should be used only in the presence of serious infections and/or in the case of surgery. If the goal of ANC >1.0 × 10⁹/l (without exceeding 5.0 × 10⁹/l) is reached within 5–7 days, this dose can be maintained; otherwise G-CSF should be increased to 1–2 µg/kg/day every 5–7 days [Fioredda et al. 2012]. In our view other therapeutic options, such as intravenous immunoglobulin, steroids, cyclosporine or anti-CD20 do not have a role in childhood.

Finally, although AIN usually does not require strong medical intervention, we suggest enrolling patients in a monitoring programme until the disease is finally resolved (Table 4). This, in addition to establish more accurately the occurrence of the resolution of the disease, enables also to intercept hidden disorders presenting as or masked by AIN. Since neutropenia may be the presenting symptom of a broader immunodeficiency fully appearing at a later stage, we recommend to re-assess for immunodeficiency or other autoimmune diseases patients whose AIN does not resolve within 3 years since the diagnosis or after 5 years of age.

Table 4.

AIN work-up.

Investigation	Time
Immunoglobulin dosage and lymphocyte subpopulations study	At diagnosis. Consider repeating in case of persistence of neutropenia 3 years since the diagnosis or after 5 years of age.
Full blood count	Every 1–3 months.
Bone marrow aspiration	In the presence of clinical signs suggesting a diagnosis re-evaluation.
Enlarged panel of autoimmunity	In case of persistence of neutropenia 3 years since the diagnosis or after 5 years of age.

AIN, autoimmune neutropenia of infancy.