Validity of western immunoblot band patterns in the serodiagnosis of Lyme borreliosis.

Serodiagnosis of Lyme disease is hampered by low specificity of the standard assays currently used. The Western immunoblot has therefore been proposed as a potential confirmatory test. For the present report, the method was evaluated by testing sera from patients with clinically defined early- and late-stage borreliosis. In early-stage borreliosis, the 41,000-molecular-weight flagellin protein (41K) of Borrelia burgdorferi was the major antigen detected by antibodies in sera, but the specificity of the reaction pattern was dependent on the intensity of the band. The evaluation of different interpretation rules based on a semiquantitative record of band intensities showed the highest specificity (96%) and a corresponding sensitivity of 78% if there was at least one distinct (optical density range, 0.2 to 0.4) immunoglobulin G and immunoglobulin M reaction with the 41K band. Blots of B. burgdorferi proteins were also probed with sera from patients who were diagnosed by clinical criteria as having stage III Lyme borreliosis and with a control group of sera from asymptomatic persons with positive antibody titers against B. burgdorferi in the standard assays. Reaction patterns were recorded densitometrically. Statistical analysis and graphical marker analysis revealed significant discriminating capacities and relatively high specificities, respectively, for the 94K, 30K, and 21K bands, whereas the 41K and 60K bands were not discriminative between the symptomatic and asymptomatic groups and were specific only at high intensity values. Different multiple-band rules were evaluated, revealing a low specificity for positivity definitions of the type "four or five bands present" if the rules were not confined to known major bands.