Polymorphisms in inflammatory cytokines and Fcγ receptors in childhood chronic immune thrombocytopenic purpura: a pilot study

Authors


Stephen J. Chanock, M.D., Immunocompromised Host Section, Pediatric Oncology Branch, National Cancer Institute, Advanced Technology Center, 8717 Grovemont Circle, Gaithersburg, MD 20877, USA. E-mail: sc83a@nih.gov

Abstract

Inflammatory cytokines and low-affinity Fcγ receptor (FcγR) polymorphisms were investigated in 37 children with chronic immune thrombocytopenic purpura (cITP) and 218 controls. Genotype analysis included common variants in the regulatory regions of cytokines, TNF, LTA, IL1RN, IL1A, IL1B, IL4, IL6 and IL10, and structural variants of the low affinity FcγRs, FCGR2A, FCGR3A and FCGR3B. Associations were observed for TNF (P = 0·0032), LTA (P = 0·019), FCGR3A (P = 0·038) and FCGR3B (P = 0·0034). Two combinations of genotypes (TNF and FCGR3A;P = 0·0003, and LTA and FCGR3B;P = 0·011) were significantly associated with cITP. These results provide preliminary evidence that variant genotypes of FcγRs and cytokines contribute to cITP pathogenesis.

Immune thrombocytopenic purpura (ITP) in children is typically acute and self-limited. In approximately 20% of children, thrombocytopenia persists beyond 6 months. Childhood chronic ITP (cITP) often has an insidious onset, occurs in older children and is rarely associated with an underlying autoimmune disorder. The pathogenesis of ITP may involve antibody-mediated platelet destruction and defective immune complex clearance, perhaps mediated by low-affinity Fc gamma receptors (FcγRs); furthermore, a Th1-predominant cytokine response has been described in cITP (Imbach, 1994; Semple et al, 1996).

To investigate genetic factors in the pathophysiology of cITP, a pilot genetic association study was conducted. The distribution of common polymorphisms in 11 candidate genes was studied in a cohort of children with cITP. The genes included three structural variants in low affinity FcγRs (FCGR2A, FCGR3A and FCGR3B) and regulatory variants in eight cytokine genes (TNF, LTA, IL1A, IL1B, IL1RN, IL4, IL6 and IL10).

Patients and methods

The population consisted of 37 Caucasian children with early cITP [persistent thrombocytopenia (< 150 × 109/l) of 6–9 months duration] enrolled in a study sponsored by the Early Chronic ITP Study Group. Subjects with secondary causes of ITP were excluded. The anonymous use of banked samples was determined to be exempt from Institutional Review Board (IRB) approval by the Office of Human Subjects Research at the National Institutes of Health (NIH), Bethesda, MD, USA. Control samples from 218 Caucasian blood donors were collected under an NIH IRB-approved protocol (Lehrnbecher et al, 1999).

Genotypes were determined in duplicate (Table I), using polymerase chain reaction (PCR) assays based on published methods (Walley & Cookson, 1996; Turner et al, 1997; Foster et al, 2000; Lehrnbecher et al, 2000). The distribution of genotypes was compared between cases and controls using a chi-squared test or Mehta's version of Fisher's exact test (M), as appropriate. Significance of differences in the distribution of genotype of each gene was calculated using either the chi-squared test or Fisher's exact test (f). The odds ratio (OR) and its 95% confidence interval (CI) were calculated using the exact method.

Table I.  Distribution of variant genotypes of candidate cytokine genes and low-affinity Fcγ receptor genes in chronic ITP compared with control population.
Chronic ITP
IL1RN
(In 2 VNTR)
N = 37

IL1A
(−889)
N = 37

IL1B
(+3953)
N = 37

TNF
(−308)
N = 37

LTA
(NcoI)
N = 37

IL4
(−590)
N = 37

IL6
(−174)
N = 37

IL10
(−1082)
N = 36

IL10
(−592)
N = 36

FCGR2A
(H/R)
N = 36

FCGR3A
(V/F)
N = 37

FCGR3B
(NA1/NA2)
N = 37
  • *

    For IL1RN, a 3 

  • ×

    2 χ2 analysis was used in a comparison of 1/1 genotypes versus 1/2 + 1/3 genotypes versus 2/2 + other genotypes.

  • Genotypes of variant cytokines or FcγRs were determined for 37 children with cITP and compared with a control population. These included the following genes (with common designations for variants), TNF(−308, 1/2), LTA (NcoI, 1/2), IL1RN (intron 2 VNTR; alleles 2–5), IL1A (−889, C/T), IL1B (3953, C/T), IL4 (−590, G/T), IL6 (−174, C/G), IL-10 (−1082, A/G), IL-10 (−592, A/C), FCGR2A (131 H/R), FCGR3A (158 V/F) and FCGR3B (NA1/NA2). The evidence for association was investigated using a χ2 analysis (3 × 2 tables with two degrees of freedom), except when Mehta's version of Fisher's exact test was used (M); no adjustments for multiple comparisons are presented. Not all samples were amplified at all loci.

1/1 = 22
(59%)
C/C = 22
(59%)
C/C = 22
(59%)
1/1 = 33
(89%)
1/1 = 3
(8%)
G/G = 29
(78%)
C/C = 8
(22%)
A/A = 10
(28%)
A/A = 1
(3%)
R/R = 7
(19%)
F/F = 13
(35%)
1/1 = 11
(30%)
1/2 = 9
(24%)
T/C = 14
(38%)
T/C = 13
(35%)
1/2 = 3
(8%)
1/2 = 11
(30%)
G/T = 7
(19%)
G/C = 12
(32%)
A/G = 19
(53%)
A/C = 17
(47%)
H/R = 13
(36%)
V/F = 23
(62%)
1/2 = 14
(38%)
1/3 = 3
(8%)
T/T = 1
(3%)
T/T = 2
(5%)
2/2 = 1
(3%)
2/2 = 23
(62%)
T/T = 1
(3%)
G/G = 17
(46%)
G/G = 7
(19%)
C/C = 18
(50%)
H/H = 16
(44%)
V/V = 1
(3%)
2/2 = 12
(32%)
2/2 = 3
(8%)
Other = 0
Controls           
IL1RN
N = 219
IL1A
N = 202
IL1B
N = 198
TNF
N = 199
LTA
N = 186
IL4
N = 186
IL6
N = 201
IL10 (1082)
N = 90
IL10 (592)
N = 89
FCGR2A
N = 218
FCGR3A
N = 191
FCGR3B
N = 215
1/1 = 115
(53%)
C/C = 91
(45%)
C/C = 116
(59%)
1/1 = 142
(71%)
1/1 = 14
(8%)
G/G = 135
(73%)
C/C = 24
(12%)
A/A = 38
(42%)
A/A = 6
(7%)
R/R = 54
(25%)
F/F = 92
(48%)
1/1 = 21
(10%)
1/2 = 76
(35%)
T/C = 94
(47%)
T/C = 65
(33%)
1/2 = 57
(29%)
1/2 = 101
(54%)
G/T = 43
(23%)
G/C = 102
(51%)
A/G = 36
(40%)
A/C = 31
(35%)
H/R = 97
(44%)
V/F = 78
(41%)
1/2 = 101
(47%)
1/3 = 5
(2%)
T/T = 17
(8%)
T/T = 17
(9%)
2/2 = 0
(0%)
2/2 = 71
(38%)
T/T = 8
(4%)
G/G = 75
(37%)
G/G = 16
(18%)
C/C = 52
(58%)
H/H = 67
(31%)
V/V = 21
(11%)
2/2 = 93
(43%)
2/2 = 20
(9%)
Other (2/4 = 1
and 3/3 = 2)
P* = 0·72P = 0·20P = 0·80P = 0·0032 (M)P = 0·019P = 0·75P = 0·083P = 0·30P = 0·42 (M)P = 0·27P = 0·038P = 0·0034

Results

The distributions of genotypes in 37 children with cITP are presented in Table I. Strong associations with ITP were observed at four loci, namely, FCGR3A, FCGR3B, TNF and LTA. Of these, the FCGR3B locus is particularly interesting (P = 0·0034) because of the over-representation of the NA1 allele, which is associated with more robust binding of IgG compared with the NA2 allele (Salmon et al, 1990; Nagarajan et al, 1995). The NA1/NA1 genotype was observed in 30% of patients compared with 10% of controls (P = 0·0022(f), OR = 3·91; 95% CI: 1·51–9·61). In addition, a second low-affinity FcγR, FCGR3A, was informative in our analysis of cITP (P = 0·038). The heterozygous V/F genotype of FCGR3A was increased in cITP patients compared with controls (62% versus 41%, P = 0·017, OR = 2·38; 95% CI 1·09–5·32). In an analysis of polymorphisms in pro-inflammatory cytokines, an overall association was observed at both the TNF (P = 0·0031) and the LTA loci (P = 0·019). Analysis of the remaining six genes, IL1A, IL1B, IL1RN, IL4, IL6 or IL10, was not informative. For the TNF (−308) polymorphism, the 1/2 or 2/2 genotypes were observed in 11% of cITP patients compared with 29% of controls (P = 0·027(f), OR = 0·31; 95% CI 0·08–0·94), suggesting that allele 1 is over-represented in cITP. For LTA, genotype 2/2 was over-represented in patients compared with controls (62% versus 38%, P = 0·0070, OR = 2·66; 95% CI: 1·22–5·96).

In an exploratory analysis, the contribution of genotype combinations to cITP was examined (see Table II). Children with cITP were more likely to possess both the TNF (−308) 1/1 genotype and the FCGR3A V/F genotype (59% versus 29%, P = 0·0003, OR = 3·64; 95% CI: 1·65–8·13). A similar effect was observed for LTA and FCGR3A at-risk genotypes (43% versus 17%, P = 0·0004, OR = 3·78; 95% CI: 1·62–8·63). Similarly, 81% of children with cITP possessed one or both of the higher risk variant genotypes of FcγRs (FCGR3A V/F or FCGR3B NA1/NA1) versus 50% of controls (P = 0·0005, OR = 4·33; 95% CI: 1·74–12·20). Eighty-one percent of children with cITP had a FCGR3A genotype with at least one V allele or had the FCGR3A FF genotype but were homozygous for the FCGR3B NA1 genotype versus 60% of controls (P = 0·014, OR = 2·88; 95% CI: 1·16–8·15).

Table II.  Further analysis of the distribution of informative variant genotypes in chronic ITP cohort compared with control population.
FCGR3A V/F
+ TNF 1/1
FCGR3B 1/1 
+ TNF 1/1
FCGR3A V/F
+ LTA 2/2
FCGR3A V/F
+ FCGR3B 1/1
TNF 1/1 
+ LTA 2/2
FCGR3B 1/1 
+ LTA 2/2
  1. Combinations of informative genotypes (first row of each couplet) were investigated in a purely exploratory fashion in 37 children with cITP and compared with results from the control population. The evidence for association was investigated using a χ2 analysis (2 × 2 tables with two degrees of freedom) when the expected value of all cells was five or more; if less than five, then Fisher's exact test was used. No adjustments for multiple comparisons are presented.

Chronic ITP
22 (59%)9 (24%)16 (43%)4 (11%)22 (59%)6 (16%)
15 (41%)28 (76%)21 (57%)33 (89%)15 (41%)31 (84%)
Controls
52 (29%)14 (7%)29 (17%)3 (2%)63 (35%)7 (4%)
129 (71%)180 (93%)144 (83%)183 (98%)115 (65%)174 (96%)
P = 0·0003P = 0·0041 (f)P = 0·0004P = 0·0157(f)P = 0·0064P = 0·011(f)

Discussion

Polymorphisms in two low-affinity FcγR genes (FCGR3A and FCGR3B) and in two pro-inflammatory cytokine genes (TNF and LTA) were each associated with childhood cITP. Our data suggest an association between FcγR genotypes characterized by increased immunoglobulin binding activity (FCGR3A allele V and the FCGR3B allele NA1) and cITP (Salmon et al, 1990; Nagarajan et al, 1995; Koene et al, 1997). FCGR3A allele V has increased affinity for three separate IgG subclasses, IgG1, IgG3 and IgG4 (Koene et al, 1997). Similarly, the FCGR3B allele NA1 is associated with increased phagocytosis and enhancement of FcγRIIa activity compared with the isoform NA2 (Salmon et al, 1990; Bredius et al, 1994; Nagarajan et al, 1995). FCGR3A, which is highly homologous to FCGR3B, is found on natural killer (NK) cells and mononuclear phagocytes. The FCGR2A allele, R131, which has decreased affinity and binding for IgG2, was observed in 7 out of 36 (19%) cITP patients and in 54 out of 218 (25%) controls (P = 0·49). These results support the hypothesis that defects in immune complex handling contribute to ITP. Variant FcγR genes with decreased activity may provide protection against ITP, a situation somewhat analogous to the Fc receptor blockade that follows treatment with i.v. Ig.

The data also suggest that genotypes of TNF or LTA associated with decreased cytokine expression could protect against cITP. These genes are tandemly arranged within the MHC on chromosome 6p, and the TNF-308 allele 2 and LTA allele 1 haplotype is associated with increased TNF expression (Moffatt & Cookson, 1997).

In contrast to our results in children, the RR genotype of FCGR2A was reported to be increased in adult cITP (48% versus 18%; P = 0·005) (Williams et al, 1998). It is notable that the course of cITP varies in children and adults, with the main difference being the potential for spontaneous remission in children, independent of treatment. Whether these apparently conflicting results reflect statistical noise or true biological differences between cITP in children and adults remains to be determined.

A role for the effect of variants in proinflammatory cytokine genes in cITP is suggested by the observation that the transcriptionally more active allele of TNF (allele 2 of −308) and the closely linked LTA allele 1 are both less common in paediatric cITP patients than in healthy controls. Whether enhanced pro-inflammatory or strong Th1-type immune responses provide protection against aberrant antibody formation in cITP will be an interesting avenue for further research (Semple et al, 1996). It is important to consider that the informative sites described here could be in linkage disequilibrium with another neighbouring variant. Indeed, correlations with both HLA-DRw2 and HLA-DRB1*0410 and ITP have been reported, although others have failed to correlate HLA alleles with ITP overall (Karpatkin et al, 1979; Nomura et al, 1998). It is probable that additional genes may also be involved in the pathogenesis of ITP. Indeed, homozygous deletion of the Ig variable gene Humhv3005 has been reported to be more common in cITP patients (n = 44; P = 0·002) (Mo et al, 1996).

This study supports the hypothesis that genetic factors from each of two important immunological pathways contribute to the pathogenesis of childhood cITP, both individually and in combination. Although caution must be exercised in interpreting results derived from relatively small patient populations, we maintain that the strength of the observed associations warrants confirmation in a larger prospectively recruited cohort of children with ITP.

Acknowledgments

Femke Mol, John O'Mara, Marianne Subleski, Susan Samuels and James Taylor, M.D., are thanked for assisting with genotype assays. Control samples were collected with the assistance of Susan F. Leitman, M.D., in the Department of Transfusion Medicine, NIH.

Ancillary