A blood gas analyzer for parenteral nutrition bags analysis? Performance evaluation and method validation

Background
The parenteral nutrition (PN) unit of our hospital produces binary PN bags. Until recently, release control of electrolyte content was carried out using capillary electrophoresis, but the supplier discontinued its activity. After comparing different instruments, we selected a blood gas analyzer (BGA), originally designed for human biological matrices.

Objective
The aim of this study was to validate this dosage method.

Materials and Methods
The selected BGA allows measurement of sodium (Na), potassium (K), and calcium (Ca) by potentiometry.
Standard PN bags with formulas “G6” and “G13” were tested.
According to ICH Q2 (R1) guidelines, the following validation criteria were assessed:
• Specificity: evaluated using pure solutions of each electrolyte at concentrations equivalent to those in G6. A six-point calibration curve was prepared in quintuplicate. The slopes of linear regression (LR) lines (electrolyte alone versus in the matrix) were compared using Student’s t-test.
• Repeatability and intermediate precision (IP): assays were repeated six times on one G6 and one G13 bag. IP was evaluated on the same sample in the same laboratory with different operators and days.
• Linearity: determined from a mixed cation solution with six concentration levels. LR line was established and a Fisher test comparing variances between theoretical and measured concentrations was performed.
• Lower limit of quantification (LLOQ): calculated for each electrolyte.
• Accuracy and trueness: evaluated through confidence intervals (CI) of mean recoveries (R).

Results
The specificity did not show any significant difference in slopes for Na and K (t _exp < t _{5%, 5 ddl}, P_value > 0,05.). Specificity for Ca could not be determined.
Repeatability gave coefficients of variation (CV) < 2%, and the FI of CV were <4% for all electrolytes
Linearity showed correlation coefficients of the regression line R² > 0.989. Fisher tests were significant for each electrolyte with pvalue > 0.05, confirming linearity.
The LLOQ was 0.53 mmol/L for Na and 0.49 mmol/L for K. It could not be determined for Ca.
For Accuracy and trueness, the CIs of recoveries included the target value of 100%.

Discussion – Conclusion
These results demonstrate the absence of a matrix effect for Na and K.
Low CV values confirm repeatability and intermediate precision.
Fisher tests validated linearity between theoretical and measured concentrations for all electrolytes.
Accuracy and trueness were confirmed.
The method is therefore validated for quality control of both standard and individualized PN bags.
Specificity and LLOQ for Ca could not be established, likely due to its higher concentration compared with human biological matrices, requiring further investigation.

Keywords: Method validation, Parenteral nutrition, Quality control

See the poster