PREDICTION OF THE IN-VIVO BIOLOGICAL-ACTIVITY OF HUMAN RECOMBINANT FOLLICLE-STIMULATING-HORMONE USING QUANTITATIVE ISOELECTRIC-FOCUSING

Currently, the in vivo biopotency of commercial preparations containin g the glycoprotein follicle-stimulating hormone (FSH) is declared on t he basis of the ovarian weight augmentation assay as described in the British, European and United States Pharmacopoeias. Human FSH contains approximately 35% (W/W) carbohydrate which introduces considerable mi croheterogeneity (isohormones). Analysis oi isohormones of recombinant FSH has revealed a relation between the isoelectric point (pi) and th e in vivo bioactivity. isohormones in the range of pi 3.5 are 100- to 200-fold more potent in the in vivo bioassay than isohormones with a p i of 5.5-6.0. These data suggest that quantification of the isohormone profile should enable us to predict the in vivo bioactivity. Thus, is ohormones of recFSH samples were separated by isoelectric focusing (IE F), visualized by Coomassie brilliant blue G250 staining and quantifie d using densitometry. The data from 21 samples were compared with the in vivo bioassay data using partial least square (PLS) regression. A c lose correlation was found using a model with 2 PLS factors (correlati on coefficient (r)=0.95, standard error of estimation s(d): 1.02 IU/mu g protein). In addition, ordinary least square (OLS) regression revea led a similar correlation between the fraction of isohormones between pl 3.9 and 4.9 and the in vivo bioactivity: r=0.95, s(d)=1.04 IU/mu g protein. Thus, an increase in the acidic isohormone fraction results i n an increase in the in vivo bioactivity. The reverse is true for the amount oi isohormones focusing between pi 5.1 and 5.7. An increase of this fraction results in a decrease of the in vivo bioactivity. These data are consistent with what might be expected from the in vivo bioas say data of the isohormones. The OLS model was subsequently validated using the guidelines of the European Centre for Validation of Alternat ive Methods (ECVAM) using 10 samples of recFSH that had not been used for the calibration. The relative standard deviation (RSD) of the mean difference between experimental and predicted in vivo bioactivity was approximately 6%. A Student's t-test performed on the experimental an d predicted bioactivity data indicated that the predicted bioactivitie s do not deviate significantly from the experimental in vivo bioactivi ty data (P < 0.05). These results demonstrate that the IEF scanning da ta can be used to predict the in vivo bioactivity with reasonable accu racy. This may be the first step towards replacing the in vivo bioassa y for highly purified FSH by a physicochemical alternative. In general , quantitative charge-based separation methods like chromatofocusing, high performance capillary electrophoresis and ion exchange chromatogr aphy may also be considered as alternatives. Finally, quantitative cha rge profiling may prove to be as important for the estimation of the p otency of other therapeutic glycoproteins like luteinizing hormone (LH ), chorionic gonadotropin (CG), thyroid-stimulating hormone (TSH) and erythropoietin (EPO). (C) 1997 The International Association of Biolog ical Standardization.