Statistical evaluation of numbers of animals to be used in vaccine potency testing: a practical approach

Some of the guidelines for potency testing of vaccines issued by regulatory bodies such as the European Pharmacopoeia (EP) and WHO are detailed and stringent (e.g. EP monograph for Newcastle Disease (ND) Vaccine (inactivated)), whereas others only stipulate that the number of animals used should be sufficient to meet the required accuracy (e.g. EP monograph for Hepatitis A vaccine (inactivated)). Simulation studies in our laboratory using historical ND potency test data indicated that the number of animals specified in the monograph is too high; a considerable reduction from 10 to seven animals per group does not substantially influence the precision of the results. Multipoint models (e.g. EP monograph for Tetanus Vaccine (adsorbed)) require at least three dilutions per vaccine for testing for response linearity. However, when historical data clearly show that in the range used the response curves are linear, it is superfluous to verify this in every test. Furthermore, linearity has little priority for a valid parallel line assay calculation. A simulation study using historical Diphtheria potency test data showed that calculations using two dilutions per vaccine in relatively small groups of animals produced results comparable to those obtained from the full assay. This procedure still enables calculation of the relative potency, in contrast to the 1 + 1 method, which gives only a pass or fail result, while the number of animals required is only slightly higher. This method could be applied in cases where the 1 + 1 method fails. In conclusion, by providing guidelines on methods in which proven consistency in production and testing may be taken into account, manufacturers are more stimulated to look for other (cheaper) ways to test the potency of a vaccine using less animals.