Autoimmunity predominates in a large South African cohort with Addison's disease of mainly European descent despite long-standing disease and is associated with HLA DQB*0201.

OBJECTIVE

We sought to determine whether autoimmunity is the predominant cause of Addison's disease in South Africa and whether human leucocyte antigen (HLA) DQ association exists.

DESIGN

We compiled a national registry of patients from primary care, referral centres and private practices.

PATIENTS: A total of 144 patients, 94 of European descent, 34 Mixed Ancestry, 5 Asian and 11 Black Africans (mean age 45.9 years, range 2.7-88 years; mean duration of disease 13.1 years, range 0-50 years) and controls were matched for gender and ethnicity. All potential causes were investigated.

RESULTS

Fifty one per cent of cases (74 patients) were autoimmune in aetiology. Either 21-hydroxylase autoantibodies (72 patients, 50% of entire patient group) or adrenocortical autoantibodies (35 patients, 24%) were present, while 23% of patients had both. None of the Asian (n = 5) or Black (n = 11) patients had evidence of autoimmune disease. Overall 8% of patients had tuberculosis, 4% adrenoleucodystrophy, 1% adrenocorticotrophic hormone resistance syndrome and 6% X-linked adrenal hypoplasia. In those with autoimmune disease primary hypothyroidism (47%), premature ovarian failure (8%) and type 1 diabetes (7%) were the most prevalent accompanying autoimmune conditions. HLA DQB1*0201 alleles predominated in the autoimmune group (DQB1*0201: 65%vs 43% of controls P = 0.017) with the *0201/*0302 heterozygous genotype being the most prevalent (28%vs 8%P = 0.02).

CONCLUSIONS

While autoimmunity accounts for at least half of patients with Addison's disease in South Africa and is associated with HLA DQB1*0201, none of the Black Africans or Asians in this cohort had adrenal autoantibodies. Moreover, 21-hydroxylase autoantibodies were detectable in a higher proportion than adrenocortical autoantibodies, especially in those patients with a long history after disease onset.