Prediction of reversible IgG1 aggregation occurring in a size exclusion chromatography column is enabled through a model based approach.

One important aspect of antibody separation being studied today is aggregation, as this not only leads to a loss in yield, but aggregates can also be hazardous if injected into the body. The aim of this study was to determine whether the methodology applied in the previous study could be used to predict the aggregation of a different batch of IgG1, and to model the aggregation occurring in a SEC column. Aggregation was found to be reversible. The equilibrium parameter was found to be 272 M(-1) and the reaction kinetic parameter 1.33 × 10(-5) s(-1) , both within the 95% confidence interval of the results obtained in the previous work. The effective diffusivities were estimated to be 1.45 × 10(-13) and 1.90 10(-14) m(2) /s for the monomers and dimers, respectively. Good agreement was found between the new model and the chromatograms obtained in the SEC experiments. The model was also able to predict the decrease of dimers due to the dilution and separation in the SEC column during long retention times.

Prediction of IgG1 aggregation in solution.

Interest in monoclonal antibody aggregation is increasing as aggregates of biopharmaceuticals can cause an immunogenic response when injected into the body. In this work, a stoichiometric reaction model from concentration-time data is developed to predict the dimer ratio in stored antibody solutions over time. IgG1 was incubated at pH from 4.5 to 5.5, salt concentrations from 100 to 600 mmol/kg and protein concentrations of 10.6-26.3 g/L; samples were taken at intervals of 20 min to 5 h over time periods from 4 h to 7.6 days, and analyzed with size-exclusion chromatography. The experiments showed the formation of dimers from monomers, but no higher order aggregates. Dilution of samples containing dimers led to the reversal of the dimerization reaction. Measurements of the concentrations of each component were made by fitting exponentially modified Gaussian peaks to the chromatograms used to measure the concentrations of the different forms of protein. This stoichiometric reaction model was able to predict the formation of dimers by the antibody studied. The equilibrium constant was found to be dependent on the salt concentration, and the kinetic constant showed a dependence on the pH of the solution. The prediction of the aggregation leads to a possibility of optimizing the conditions in order to prevent the dimer formation and to maximize the monomer concentration.

Modelling and optimisation of preparative chromatographic purification of europium.

A model commonly used to describe the separation of biomolecules was used to simulate the harsh environment when eluting neodymium, samarium, europium and gadolinium with a hot acid. After calibration, the model was used to optimise the preparative separation of europium, as this is the most valuable of the four elements. A kinetic dispersive model with a Langmuir mobile phase modulator isotherm was used to describe the process. The equilibration constant, the stoichiometric coefficient and the column capacity for the components were calibrated. The model fitted the experimental observations well. Optimisation was achieved using a differential evolution method. As the two objective functions used in optimising the process, productivity and yield, are competing objectives, the result was not a single set point but a Pareto front.

Optimization of preparative chromatographic separation of multiple rare earth elements.

This work presents a method to optimize multi-product chromatographic systems with multiple objective functions. The system studied is a neodymium, samarium, europium, gadolinium mixture separated in an ion exchange chromatography step. A homogeneous Langmuir Mobile Phase Modified model is calibrated to fit the experiments, and then used to perform the optimization task. For the optimization a multi-objective Differential Evolution algorithm was used, with weighting based on relative value of the components to find optimal operation points along the Pareto front. The objectives of the Pareto front are weighted productivity and weighted yield with purity as an equality constraint. A prioritizing scheme based on relative values is applied for determining the pooling order. A simple rule of thumb for pooling strategy selection is presented. The multi-objective optimization gives a Pareto front which shows the rule of thumb, as a gap in one of the objective functions.