Statistical procedure for evaluating the sensitivity of Limulus amoebocyte lysate by using a reference lysate

A study was designed to estimate variability of the Limulus amoebocyte lysate test by comparing a reference lysate against itself. Three technicians performed parallel tests, i.e., titrated side by side, the contents of two vials of reference lysate on 4 different days using 24 vials of the United States reference lysate and 12 vials of the United States reference endotoxin. Each parallel test was replicated three times. From the sensitivity endpoints, ratios were calculated for each parallel test. These ratios were converted to the logarithm for estimating variability among technicians and among vials of endotoxin. By using the overall variability of log ratios, a statistical procedure was developed to evaluate the sensitivity of each lot of licensed lysate submitted to the Bureau of Biologics for release.