Compliant or non-compliant ? When the in-process controls and liberating analyticals controls are not aligned

Introduction
To ensure security and quality of chemotherapy masterful preparations, an in-process securing with gravimetric control and identification of each bottle by DataMatrix code was implemented and allows 100% control of our production (Diffusers and clinical trials included). Even if we demonstrated the accuracy of the weighing, the step of identification of each bottle done before the preparation doesn’t leave out that the sample was in a wrong bottle. A consistency control between our in-process monitoring (MIP) and the final analytical dosage (FAD) by QCprep was considerate.

Material and method
228 preparations, accepted by our MIP control and representative of our production
were sampled and dosed helped by a QCPrep. The analytical compliance was set to +/- 15% of errors of the dose. In case of discrepancy MIP/FAD, the non-compliances (NC) were analyzed on a case-by-case basis.

Results
13/228 preparations were identified as NC during FAD so 5,7%. One case was a qualitative variance (docetaxel instead of paclitaxel) and was there because of a formulation modification with an impact on the UV/Visible spectrum and so on the good recognition by the QCPrep. The 12 others cases were quantitative errors (Dosage of –22,61% to 18,71%) while the errors on weighing were only from - 2,88% to –6,77%.
The hypothesis considerate (and cumulative) were:
• Issue of homogenization; in agreement with the high under-dosage proportion (11/12) and feedbacks of centers managing the analytical control
• Difficulties to dose the cisplatin (4/12); the cisplatin that absorbs a little and on a specific way in the UV-Visible
• Low volume (<3mL) (3/12), gravimetric control difficult or high impact of dilution?
• Drugs to reconstitute (3/12): quality of initial dissolution and cumulative gravimetric steps

Conclusion
The 228 preparations dosages confirmed the security of our preparations process, especially on a qualitative point of view (only one NC because of an analytic control error). The quantitative discrepancies are explained because of analytic difficulties (homogenization of preparations, non-specific spectral analysis, results impacted by high dilutions) and not because of gravimetric difficulties, even if the case of drugs to reconstitute can insert cumulative weighing errors.
These results, completed by universality (diffusers and EC), low cost, low technical level and the possible intervention during preparations management validates our choice of parametric liberation for chemotherapies masterful preparations

See the poster