From Table 4, it is evident that the

immunoassays from laboratory 7 are giving lower estimated potencies for all three samples A – C. Laboratory 2 has estimates that are higher than other laboratories for samples A and B, but for sample C they are in agreement with the other laboratories. Apart from these results, all laboratories appear to be giving consistent results and are in reasonable agreement. The within-laboratory, between-assay, variability is shown in Table 4, as %GCVs. These represent good within laboratory repeatability, with all GCVs less than 10%, and the majority being less than 5%. There was greater variability between estimates from individual plates within Panobinostat datasheet assays in some laboratories (data not shown). This appeared to result from possible plate effects (variation in response across different rows or columns of the plate). Because a balanced layout was used, varying the position of the samples across different plates, consistent results were obtained when the individual plate estimates were combined to give single assay estimates. However, it does emphasise the need to be aware of potential plate effects, and the importance PLX-4720 manufacturer of using a suitable experimental layout across plates. Samples A and B are duplicates of the same material (86/500). The average within-assay % differences in potency

estimates between duplicates are shown in Table 5. All but one of the laboratories are achieving average agreement within 10%, with the majority being within 5%. The overall geometric means of the laboratory means, along with between-laboratory %GCVs and the range of potency estimates are shown in Table 4. The overall trimmed mean (excluding the highest

and lowest laboratory estimates) are shown in Table 6. For the candidate standard 86/500, there is very little difference between the overall mean and the trimmed mean. The effects of the low results from laboratory 7 and the high results from laboratory 2 Ibrutinib solubility dmso on the overall mean cancel each other out. The combined overall mean for samples A and B is 202 IU based on all laboratories, or 203 IU based on the trimmed mean of the central 8 laboratories. For sample C, the potency estimates are around 20% higher than for A and B, at 236 IU and 242 IU for the overall and trimmed means respectively. Table 7 shows the overall means based on the 6 laboratories performing bioassay only. For the candidate standard 86/500 the mean is a little higher at 211 IU compared to the 201 or 203 IU from the overall or trimmed means of all laboratories. This is because restricting the calculation to the bioassays alone has the effect of removing the low results from the immunoassay of laboratory 7, but including the high results from the bioassay of laboratory 2. For sample C, there is little difference between the trimmed mean of all laboratories and the overall mean of the bioassays alone.