In-process gravimetric control validation of antineoplasic preparations by the accuracy profile and total error calculation analysis

Background
In-process gravimetric control is the method used for the double control of our preparations within our chemotherapy reconstitution unit.

Objective
Qualification of the method of analysis and determination of the rate of preparations that we will be able to control with a margin of error of 10% or 5% depending on the volume taken.

Methods
Tests were carried out in the isolator with a Secura^® Sartorius precision balance (reading accuracy of 0.1 mg) with an automatic calibration function that allow us to overcome weighing variations due to the flow of air in the enclosure. The total error was calculated from bias (trueness) and coefficient of variation (precision). For each type of syringe (1, 3, 5, 10, 20 and 50 mL), 5 different volumes were collected 3 times and on 2 different days. 2 types of significantly different density solvents were used (Water for Injection (WFI, Density = 1,000) and 50% Glucose (G50%, Density = 1,188)). The accuracy profiles for each sample were also calculated from these data. The maximum tolerated deviation from theoretical volume taken was set at 10% and then 5%. An inventory of the presence of datamatrix code and the density of products used routinely was also carried out.

Results
The calculated total error is between 2.04% and 3.64% for series with G50% (volumes withdrawn from 0.1 mL and 0.15 mL in 1 mL syringes, 1 mL and 1.4 mL in 3 mL excluded) and between -2.56%. and 4.11% for series with WFI (volumes of 0.1 mL, 0.15 mL and 0.5 mL in a 1 mL syringe and 1 mL in 3 mL excluded). The accuracy profiles show that the method will be usable with an uncertainty of 10% for volumes taken respectively from 0.5 to 60 mL and 3 to 60 mL for products of density comparable to WFI and G50%. For these 2 solvents, the uncertainty can be lowered to 5% from a volume of 17.5 mL. Of the 50 references used routinely we managed to recover the densities of 45 products. A datamatrix code is present on 35 secondary packaging (70%) and 28 vials (56%) but only 11 (22%) are exploitable (CIP code 13, lot number, expiry date). No solvent bag used had datamatrix code.

Discussion – Conclusion
The total error of our samples and the accuracy profiles indicate that the analytical method is reliable and accurate with an uncertainty of 10% for 93% of our preparations outside clinical trials. This uncertainty can be reduced to 5% for 64% of them, which is more commonly accepted in clinical practice. Only volumes smaller than 3 mL will remain visually controlled. To this quantitative analysis will be associated a qualitative analysis of vials and solvent bags through a datamatrix code whose feasibility in time will have to be evaluated.

See the poster