Raw Material Detectives - Case #1 The Crystal Shift

It was a dark and gloomy night.

Our Customer contacted the Global Technical Service (GTS) group with complaints of poor product performance. The GTS squad logged the report into the complaint handling system.

The customer, a large, US-based biopharmaceutical company, reported that they were having problems with their custom CHO medium - and were observing growth and productivity fluctuations from lot to lot within their commercial process.

An investigation began with GTS and quality assurance in a cross functional team that included members from supply chain management (SCM) and cell sciences and development (R&D). A review of batch records listing raw material lots and vendors in the batches of media narrowed the list of potential culprits to 6 raw materials.

After extensive collaborative trouble-shooting, the Customer and GTS concluded that there likely was a problem with the L-tyrosine from a specific vendor that was included in the formulation. Suspect was L-tyrosine 2Na H₂O.

The Investigative Team lined up 3 lots of L-tyrosine 2Na H₂O from 3 vendors for interrogation. Vendor 1 was suspected as the supplier of the problematic material. Vendor 2 and 3 were alternative qualified vendors for our line up. We also included a positive control. CSD began its testing.

Analytical Testing

The suspect lot of L-tyrosine was tested side by side with lots from two other qualified vendors via several analytical methods. The testing was performed to assure product identity and to allow the detection of impurities.

Reverse phase-HPLC (RP-HPLC)

Size-exclusion chromatography (SEC)

Ultra-performance LC-MS (UPLC-MS)

No significant differences were detected between the suspect amino acid, comparison vendor samples and controls via these methods. No obvious impurities were detected.

Batch Record - Process Review

The manufacturing process flow diagrams were examined from the 3 vendors - suspect and alternatives. Our first important clue - there was 1 process step difference from the others with the suspect vendor.

The Investigative Team decided to perform biological testing between the 3 different vendors to see if the suspect vendor's material would show biological variance.

Biological Testing

CSD tested 3 lots of L-tyrosine 2Na H₂0 from each of the 3 vendors in CHO-GS cells producing a control MAb. CSD plotted growth curves and measured MAb productivity.

Cell growth curves generated with lots 1 & 2 of L-tyrosine from Vendor 1 at best reached only 88-90% of the average peak cell density seen for Vendors 2 & 3. The worst performing lot 3 from Vendor 1 only reached 55% of the peak cell density seen for Vendors 2 & 3, and thus was also signifcantly worse performing than lots 1 & 2 from same suspect Vendor 1.

Download Tyrosine Cell Culture Data

Similarly, IgG productivity of CHO-GS cultures grown on medium containing suspect Vendor 1 lots of tyrosine was 12-63% lower than cultures grown on medium incorporating L-tyrosine from Vendors 2 & 3.

Download Tyrosine Productivity Data

Filtration Studies

Since the biological variance of the suspect material had been confirmed, one logical explanaton for the poor performance was a solubility problem with the L-tyrosine from Vendor 1 as compared to the other two vendors.

The Investigative Team decided to run filtration studies. Lots of L-tyrosine from the various vendors were filtered and the post-filtration amino acid concentration was calculated versus the expected yield. Vendor 1 material did not filter as well. When cells were grown in medium formulated with different lots of L-tyrosine from the 3 vendors, there was a direct correlation found between the amino acid concentration remaining and the productivity titers in cell culture assays. The suspect L-tyrosine was not making it through the filters.

Download Tyrosine Filtration Data

Physical Property Analysis

SAFC owns a company called Pharmorphix that provides expert solid state chemistry services and specializes in pharmaceutics. CSD decided to have them analyze 3 lots of the L-tyrosine 2Na H₂O from suspect Vendor 1(bad lots) and Vendor 2 (good lots).

SAFC's Pharmorphix group performed 3 different analytical techniques:

X-ray particle diffraction (XRPD) to determine crystal structure.

Differential scanning chromatography (DSC) to determine melting points of the crystals.

Ion chromatography to allow quantitation of the Na⁺ ions.

Results for the XRPD analysis are shown below. These results reveal there were distinct crystal forms between lots from Vendor 1 and Vendor 2. What's more? Vendor 1 had different crystal forms between its own lots and multiple crystal forms existed in 1 lot - the very poor performing lot 3. DSC also pointed to the suspect L-tyrosine having melting points that varied from the "good lots" and exhibited multiple melting points within the population of 3 suspect Vendor 1 lots. Suspect 1 was different.

Download Tyrosine XRPD Data

Case Conclusion

Evidence revealed differences in crystal structure between good lots of tyrosine and suspect Vendor 1 bad lots of tyrosine. These structure differences likely led to the poorer solubility characteristics of Vendor 1 material. This in turn led to poorer filterability and lower concetrations of Vendor 1 tyrosine in the final medium post-filtration. This resulted in the poorer growth and productivity observed in the CHO GS cells.

Why the difference in crystal structure? Recall the batch record review. Suspect Vendor 1 was using spray drying to prepare its final tyrosine product. Vendors 2 & 3 were using tray drying. We believe this led to the variablity in crystal structure seen with the Vendor 1 material.

Corrective Action

SAFC Supply Chain Management and Quality Assurance shared the analytical findings of the CSD team with Vendor 1. Vendor 1 was busted.

Vendor 1 at present has been removed as a Qualified Vendor of L-tyrosine to SAFC. Vendor 1 would have to go back through a qualification process with at least 3 lots of new material before it could potentially be qualified again.

Case File: 04272010-01