Carryover contamination in routine HPLC analytical control within a sterile injectable compounding unit: impact on analytical reliability and implementation of a corrective cleaning procedure
30 September 2026
F. Kérourédan 1, J.M. Bernadou 1, S. Dubois 1, A. Berroneau 1, G. Bouguéon 1,21. Pharmaceutical Technology Department, Bordeaux University Hospital, Pessac, France
2. ARNA laboratory, ChemBioPharm, INSERM U1212, UMR 5320 CNRS, University of Bordeaux, France
Introduction
Within our production unit (65,000 preparations/year), a part of sterile injectable preparations (≈10,000; 12 different molecules) undergoes quality control by liquid chromatography (HPLC) coupled with UV-DAD detection, using either column-based separative methods or flow injection analysis (FIA). In the context of increasing production activity, an analytical bias was observed, associated with carryover contamination and overestimation of assay results. This issue compromises reliability of routine analytical control and exposes to the release of out-of-specifications (OOS) preparations.
Materials and Methods
Assays were performed continuously within production workflow (acceptance limits: ±15% of theoretical concentration). Successive analytical sequences of different suspected chemicals (cytarabine, paclitaxel, and 5-fluorouracil [5-FU]) were analyzed over a 7-month period. When carryover was identified, a cleaning procedure was performed, consisting of manual external cleaning of the injection needle followed by an automated wash cycle using a 50:50 (v/v) water-acetonitrile solution. Cleaning efficiency was evaluated using a predefined acceptance value: residual peak height <2 mAU on the chromatogram after water injection.
Results
Carryover contamination was identified through OOS paclitaxel assay results (>115%). Over the study period, 26 episodes leading to assay overestimation were identified (µ=119.4% [115.2-122.9%]). The effect increased over the course of the day, and following the analysis of cytarabine vials, suggesting carryover contamination involving the autosampler system. Following implementation of the cleaning procedure, residual peak heights were < 2 mAU in 100% of evaluated cases, and paclitaxel assay results returned within analytical specifications (µ=100.7% [98.5–103.8%]).
Discussion
Repeated analysis of different chemicals appears to promote accumulation on metallic and polymeric surfaces within the injection system. This observation is consistent with a surface adsorption process occurring in the autosampler system followed by subsequent desorption and release. The variable physicochemical properties of molecules analyzed, including cytarabine (hydrophilic), paclitaxel (highly lipophilic), and 5-FU’s ability to crystallize on surfaces, may contribute to this mechanism. HPLC automated wash cycles appear insufficient to prevent carryover, whereas implementation of a standardized cleaning procedure represents a simple and effective corrective measure to control this contamination risk.
Conclusion
This study highlights the impact of carryover contamination on assay overestimation in HPLC analyses. The implementation of a targeted cleaning procedure restored analytical compliance and improved reliability of routine HPLC control of injectable sterile preparations.