This time two customers contacted the Global Technical Service (GTS) group with complaints of poor filterability of their lastest batches of a custom serum-free CHO medium, being used in large scale production of a commercial biopharmaceutical. This formulation contained about 70 components including soy hydrolysate. We put in a call to the Raw Material Detectives (RMD).
A review of the raw material lot numbers, used in the medium, by the GTS staff identified an association with a specific lot number of polyvinyl alcohol (PVA) CAS 9002-89-5 and poor filtering lots of medium. Thus a clue! The Raw Material Detectives started with the PVA vendor.
Editor's Note: PVA is used primarily in industries like food, paper, adhesives and textiles. Cell culture is a very minor use application by comparison -- i.e. and therefore has reduced influence on the manufacturer/ supplier. PVA is produced by the polymerization of vinyl acetate to produce polyvinyl acetate. This is then hydrolyzed with sodium hydroxide to yield PVA. Possible impurities in PVA would include sodium acetate, methanol, and methyl acetate. PVA should be readily soluble in water but requires heating. Degreee of hydrolysis , molecular weight, particle size distribution, and particle crystalline structure all affect solubility. Optimal solubility occurs at 87-89% hydrolysis, a fine particle size (< 80 mesh), 3.5-4.5 cps viscosity with a 4% solution at 20 C.
Vendor PVA lots were evaluated for D90 particle size, viscosity of 4% solution, and percent hydrolysis of PVA to look for correlation with customer poor filterability results. There was no obvious correlation.
SAFC worked with its customers to develop a scaled down Vmax assay that correlated with the filter difficulties seen at large scale. The following reference was developed.
- Vmax <1000 L/m2 were correlated with below average filterability
- Vmax >2000 L/m2 were correlated with above average filterability
- Vmax 1000-2000 L/m2 generally were above average but not 100%
Based on the above assay, it was decided to to analytically characterize multiple PVA lots with different filterability results in an attempt to explain the poor filtration characteristics.
PVA powders associated with know above and below average filtering lots at large scale were sent to Chemir Analytical Services (Maryland Heights, MO) for the following tests:
- Gel permeation chromatography - compares MW properties
- Glass capillary viscometry (GCV) - compares intrinsic viscosities
- Fourier transformation infrared spectroscopy (FTIR) - compares base chemistry
- Proton nuclear magnetic resonance spectroscopy (1H-NMR) - compares backbone chemistries
- Differential scanning calorimetry (DSC) - compares thermal properties
- X-ray particle diffraction (XRPD) - compares crystalline structures
- Laser diffraction - gives particle size determination
Four lots were sent for analysis from the current vendor - 4 above average filtering lots and 4 below average filtering lots. Additionally, the Raw Material Detectives came up with 2 addtional lots from 2 addtional vendors to also test.
Pretty slick set of analytical assays - huh? Unfortunately, there were no significant differences in the physical properties of the various PVA lots tested that were identified by any of these seven assays. WOW! Time to quit? Never. The Raw Material Detectives always get their Man (so to speak). Read the next post for the Rest of the Story.
Posted by Bruce Lehr June 14th 2010.