Abstract

Sera from 31 Legionnaires' disease (LD) survivors of the Philadelphia outbreak, 31 Legionnaire (L) controls, and 300 additional controls were examined for the presence of specific antibodies to five antigen preparations of Legionella pneumophila (serogroup 1) to determine the effect of antigen preparation on the sensitivity and specificity of the indirect immunofluorescence test. Diagnostic levels were determined for each antigen at the upper limit of normal value (ULNV) titre, which established the titre not exceeded by 85% of controls.

Antigens were prepared from formalin-killed L. pneumophila suspended in egg yolk sac (EYS) (LPF:EYS) or bovine serum albumin (BSA) (LPF:BSA); and from heat-killed organisms suspended in EYS (LPH:EYS) or BSA (LPH:BSA). Antigen was also supplied by the Center for Disease Control (CDC:AG).

Although there was wide variation in the sensitivity of the antigens, at the ULNV level all antigens tested could be used to differentiate LD survivors from L controls (p<0.001; X² test). Formalin treatment resulted in the most specific antigen by eliminating titres in L controls. The results of the X² test, comparing LD survivors with L controls, ranked the antigens in the following ascending order of sensitivity: LPH:BSA 15·3, <CDC:AG 22·8, <LPH:EYS 24·2, <LPF:BSA 45, <LPF:EYS 51. Moreover, when differences in positive results among LD survivors were compared, statistically significant differences were found when LPF:EYS (p<0·01; X² test) and LPF:BSA (p<0·025; X² test) were compared with CDC:AG, LPH:EYS, and LPH:BSA antigens. Titres in 300 additional controls paralleled those found in L controls.

It was concluded that formalin treatment of L. pneumophila resulted in a sensitive antigen, which increased the number of positive tests in survivors while decreasing false-positive tests in controls. It should be considered for use in routine testing programmes for diagnosing LD.