Background: Hypersensitivity to abacavir is a well-characterized clinical syndrome occurring in 4% of patients.  Features of the syndrome are consistent with an immunologic process possibly influenced by genetic factors. 90% of cases occur within the first 6 weeks of therapy.

Methods: A retrospective case-control proof of principle study enrolled patients to compare selected genetic polymorphisms in HIV-infected subjects who developed symptoms consistent with hypersensitivity (cases) to subjects who did not (controls). All subjects participated in clinical trials with abacavir. All cases had clinical histories compatible with hypersensitivity to abacavir, and all controls had received abacavir for a minimum of 6 weeks. Cases and controls (1:2 ratio) were matched by race, gender, and when possible CD4+ cell count, age, and antiviral therapy.  Conditional logistic regression and recursive partitioning analyses were performed.

Results: Evaluable data are available on allele frequencies for 114 candidate gene markers (200 subjects) and HLA A, B, and DR (197 subjects).  Overall, 92% of subjects were male, 74% Caucasian, and 14% were black.   59% of cases could be matched to 1 or more controls; 41% were unmatched. Multiple polymorphisms were identified within the HLA region associated with susceptibility to hypersensitivity.  Analysis identified a known polymorphism of TNF-α at position –238A among 25 of 58 (43%) cases compared to 7 of 99 (7%) controls (p<0.001). No data for TNF-α were available from 43 subjects.  In addition, HLA-B57 was present in 39 of 84 (46%) cases vs 4 of 113 (3.5%) controls (p<0.001). The association of the TNF-α polymorphism was almost completely accounted for by the HLA-B57 association.  No statistical model using a combination of genetic markers, including polymorphisms for abacavir drug metabolizing enzymes was more predictive than a model using HLA-B57 alone.

Conclusions: These data support the role of genetic factors in hypersensitivity reactions, but a causal relationship to polymorphisms within the HLA region is not established.  HLA typing as a prognostic screening tool is problematic due to false positive rate, lack of data in many key demographic groups.  As HLA-B57 was not fully predictive, HLA typing must not be used as a diagnostic test when symptoms suggest a hypersensitivity reaction.  The HLA status of the patient does not alter the clinical management of a hypersensitivity reaction.