Papers

Association of mutations in mannose binding protein gene with childhood infection in consecutive hospital series

BMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7089.1229 (Published 26 April 1997) Cite this as: BMJ 1997;314:1229
  1. John A Summerfield, professor, department of medicinea,
  2. Michiko Sumiya, lecturer, department of medicinea,
  3. Michael Levin, professor, department of paediatricsa,
  4. Malcolm W Turnerb
  1. a Imperial College School of Medicine at St Mary's, London W2 1NY
  2. b Immunobiology Unit, Institute of Child Health, London WC1N 1EH
  1. Correspondence to: Professor Summerfield
  • Accepted 24 January 1997

Abstract

Objective: To determine the extent to which mutations in the mannose binding protein gene predispose to childhood infection.

Design: Clinical details and genotype of mannose binding protein determined in consecutive children attending a paediatric department.

Setting: Inner city hospital paediatric service in London.

Subjects: 617 children attending hospital between October 1993 and August 1995.

Main outcome measure: Infection as the cause for attendance or admission in relation to mutations in the mannose binding protein gene.

Results: The prevalence of mutations in the mannose binding protein gene in children with infection (146/345) was about twice that in children without infection (64/272) (P<0.0001). Increased susceptibility to infection was found in both heterozygotic and homozygotic children. 13 out of 17 children homozygotic for variant alleles presented with strikingly severe infections, including 6 with septicaemia.

Conclusions: The findings suggest that mutations in the mannose binding protein gene are an important risk factor for infections in children. Screening for such mutations should be included in the investigation of severe or frequent infections.

Key messages

  • Mutations in the mannose binding protein gene, which cause a common opsonic defect, are strongly associated with children presenting to hospital with infection

  • The mutations increase susceptibility to infection in children who are heterozygotic or homozygotic for the mutations

  • Children homozygotic for mannose binding protein gene mutations usually present with severe infections

  • Investigation of severe or frequent infections should include screening for mannose binding protein gene mutations

Introduction

Mannose binding protein is a calcium dependent lectin that plays an important part in innate immunity.1 A common opsonic defect, associated with low serum concentrations of mannose binding protein,2 is caused by mutations in codons 54 and 57 of the collagen domain that impair assembly of mannose binding protein homopolymer.3 4 5 A third, less common, mutation has been identified in codon 52.6 Repeated bacterial and fungal infections associated with these mutations have been reported.3 7 8 Both these mutations and childhood infections are common, but there are no data on the extent to which such mutations predispose to childhood infectious disease. We examined a consecutive series of children attending a hospital paediatric service to determine whether mutations in the genes for mannose binding protein are an important risk factor for infection.

Methods

Patients and methods–Blood samples were obtained from consecutive children attending St Mary's Hospital. All were venesected for clinical reasons and surplus blood was spotted on to blotting paper (Guthrie cards). Samples were collected between October 1993 and August 1995. Diagnoses were confirmed from the notes and the clinical diagnostic codes after discharge. The children were classified, without reference to the mannose binding protein genotyping, as to whether the presenting illness was an infection according to the International Classification of Diseases, ninth revision. The study was approved by the ethics committee of Parkside Health Authority. Guthrie cards were autoclaved at 120°C for 7 minutes and stored at room temperature. Genotypes were determined by sequence specific oligonucleotide hybridisation.9 Homozygosity was confirmed by DNA sequencing.

Statistical analysis–Calculations before the study showed that 150 children per group would enable us to detect a difference of 20% in the infected group with a power of 95% and significance of 5%. Prevalences of the mutations were determined by using the Hardy-Weinberg equation. Differences were evaluated with the χ2 test, odds ratios, and 95% confidence intervals.

Results

A total of 345 children were admitted with infection; 272 children, who acted as controls, were admitted with various other diagnoses (table 1). Six hundred and ninety blood samples were collected. Data from 617 children (89%) were complete (table 2). Ages ranged from 0 to 18 years, and 58% (357) were male. The expected prevalence of the codon 52 mutation (0.04) was observed. The prevalences of codon 54 (0.1) and codon 57 (0.05) mutations were lower than expected, but when we corrected for the ethnic composition of the sample (African and Caribbean 21%; white, Asian, and Oriental 79%) the expected prevalences for codon 54 in the Eurasian group (0.12) and for codon 57 in Afro-Caribbeans (0.22) were obtained. One child was homozygotic for codon 57 mutation. Three were homozygotic for codon 54 mutation, whereas the Hardy-Weinberg equation predicted that the number would be six. Fourteen were phenotypically homozygotic, having combinations of mutant 52, 54, or 57 alleles. Thus 17 (3%) were homozygotic for mutations in the mannose binding protein gene.

Table 1

Diagnoses in 272 control children without infections

View this table:
Table 2

Details of genotypes for mutations in mannose binding protein gene in children presenting with infections and those without infections

View this table:

We examined the association of mutant gene alleles for mannose binding protein with infection. Of the 272 children without infections (controls), 64 carried variant alleles and 208 carried normal genes (wild type). In contrast, of the 345 children with infections, 146 carried variant alleles and 199 were wild type (table 2). The increased prevalence of variant alleles in infected children was highly significant (odds ratio 2.4; 95% confidence interval 1.7 to 3.4; P<0.0001). The ethnic composition of the infected and control groups was similar. When the ethnic groups were analysed separately a significant excess of variant alleles in infected children remained (data not shown). Table 3) gives details of the infections in heterozygotic children.

Table 3

Infections in 133 children heterozygotic for mutations in mannose binding protein gene

View this table:

We examined whether variant alleles increased susceptibility to infection in both heterozygotic and homozygotic children (table 2): 60 controls were heterozygotic for variant alleles and 208 were wild type, whereas 133 with infection were heterozygotic for variant alleles and 199 were wild type (2.3; 1.6 to 3.4; P<0.0001). An increased prevalence of infection in heterozygotic children was observed when mutation data for codon 52 (3.0; 1.4 to 6.3; P=0.002), codon 54 (2.2; 1.4 to 3.5; P=0.0009), and codon 57 (2.1; 1.1 to 4.2; P=0.04) were analysed separately. The number of homozygotic children was smaller, but a significant difference was observed. Four controls were homozygotic for variant alleles compared with 13 children in the infection group (3.4; 1.0 to 11.4; P=0.048). Children homozygotic for the mutation presented with strikingly severe infections, including six with septicaemia (table 4).

Table 4

Clinical details of children homozygotic for mutations in mannose binding protein gene

View this table:

We examined whether the risk of infection conferred by variant alleles was related to age. Variant alleles conferred significant susceptibility to infection in children aged less than 6 months (3.0; 1.0 to 9.0; P=0.05), in those aged 6-18 months (4.0; 1.0 to 17.0; P=0.03), and those aged over 18 months (2.4; 1.5 to 3.7: P=0.0001) (fig 1).

Fig 1
Fig 1

Prevalence of gene mutations in mannose binding protein in infected and control children from different age groups. Prevalence of mutations was significantly greater in infected children under 6 months (P=0.05), between 6 and 18 months (P=0.03), and over 18 months (P=0.0001)

Discussion

Mannose binding protein is an important component of innate or natural immunity.1 Mutations in the mannose binding protein gene have been associated with recurrent infections.3 7 8 Both childhood infections and mutations in this gene are common, but there are no data on the roles that the variant alleles have in the susceptibility of children to infection. Our study used concurrent controls and was not referenced to a historically selected adult control group. We obtained 345 samples from children with infections (56%) and 272 samples from non-infected (control) children (44%) from the same population. Most (89%) of the 617 samples were obtained in general paediatric wards and clinics, but many of the 51 (11%) with meningococcaemia or the 16 with HIV infection or AIDS reflected referrals to St Mary's as a centre for paediatric infectious disease.

The data show a striking and highly significant association of infection with mutant mannose binding protein genotypes (table 2). Mutant mannose binding protein alleles were present in about twice as many children with infections as in control children. This was not due to differences in the ethnic composition of the two groups as the association was observed when we analysed ethnic groups separately. This increased risk of infection is similar to that observed in an earlier study of children with an opsonic defect,10 which is the functional consequence of mannose binding protein mutations.3 4 In contrast, another study did not find that children who were heterozygotic for mutant mannose binding protein gene alleles were at increased risk of infection.8 This discrepancy may be due to the use of an adult control group in that study rather than concurrent controls matched for age.

The prevalence of homozygotic alleles for mutant mannose binding protein gene in children was 3%, and most (13/17) had combinations of mutant 52, 54, or 57 alleles. Only three were homozygotic for codon 54 (0.005%), whereas the Hardy-Weinberg equation predicted that six should be. Our sample was not large enough to determine whether the difference was significant, but the data are consistent with other reports that populations are depleted of people homozygotic for codon 54.4 6 10 The reasons for this are obscure. It may be relevant that 13 of the 17 children homozygotic for variant alleles presented with serious infections, including six with septicaemia (table 4). Four of the homozygotic children had meningococcaemia, and this association is under further investigation. The association of severe tonsillitis and otitis media with homozygosity for mutant mannose binding protein alleles has been noted elsewhere.8

Low concentrations of mannose binding protein may be a particular risk factor between the ages of 6 and 18 months.11 The data from our study, however, indicate that mannose binding protein mutations confer a significantly increased risk of infection at all ages in childhood (fig 1). It seems reasonable to conclude from this and other reports7 8 that mannose binding protein gene mutations confer a lifelong risk of infection.

In conclusion, these data show that in a hospital population, children, whether heterozygotic or homozygotic for mannose binding protein gene mutations, are at increased risk of infection. The infections in the homozygotic children were particularly serious, suggesting that screening for these mutations should be included in the investigation of severe or frequent infections. Prospective controlled studies in the community are now needed to define the risk that mannose binding protein gene mutations confer.

Acknowledgments

We thank Dr Jane Wadsworth and Ms Angela Wade for statistical advice, the staff of the department of paediatrics, St Mary's Hospital, and Ms Caroline O'Loughlin for help with sample collection. We thank Dr Martin Hibberd for analysis of some samples.

Funding: Action Research.

Conflict of interest: None.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
View Abstract