MDSA potency and MDSA confidence limits (C.L.) were calculated using Huber’s robust regression; the resulting slopes and the validity criteria are also shown. Test criteria valid in the unweighted parallel line model and changed into invalid when
applying Huber’s weighted method are marked as “a”, representing non-linearity (significance level p 0.05), and “b” indicating non-parallelism (p 0.05). In the column “NIH potency” the batch potency values obtained by the manufacturers are presented. The confidence limits (NIH C.L.) were recalculated from the data of manufacturers’ batch protocols using CombiStats