(T0930-08-44) LC-PDA and a Multivariate Curve Resolution – Alternating Least Square Algorithm for Identification and Quantification of Co-eluting Impurities in Pharmaceutical Analysis

Purpose: In this poster we investigated the reliability of a multivariate curve resolution-alternating least squares (MCR-ALS) method applied to a matrix of data generated from HPLC coupled with photo-diode array detector (HPLC-PDA)¹ the identification and quantification of co-eluting impurities as low as 0.05 area%. We Further estimated the effect of Photodiode Array detector linearity degradation on the quantification performances.

Methods: Sample preparation: The DFBP and VP were dissolved in an acetonitrile and water mixture solution (acetonitrile/water = 45/55,v/v) to prepare a standard solution and mixture sample. A DFBP sample with relative concentration 100 was prepared and then respective standard solutions were prepared with VP relative concentrations of 0.05, 0.1, 0.2, and 1 (six individual samples for each condition). In addition, mixture samples were prepared with DFBP/VP relative concentration ratios of 100/0.05, 100/0.1, 100/0.2 and 100/1 (six individual samples for each conditions). Furthermore mixtures at relative concentration 0.05:0.05 and 0.05: 1 were also prepared (three individual samples for each condition). Chromatographic conditions: A Shimadzu Shim-pack XR-C8 reversed phase ultra-fast analysis column (3.0 × 50 mm, 2.2 um packing) was used to perform an isocratic analysis using an acetonitrile and water mixture (water/acetonitrile) as the mobile phase, which was ultrasonically degassed before use. Analysis was performed at a flowrate of 1.0 mL/min, column temperature of 45◦C, and PDA detection wavelength range of 190 to 700 nm. Each individual sample was analyzed using different chromatographic conditions; the % of mobile Phase B (acetonitrile) was changed in order to obtain different resolution values between co-eluting peaks. Figure 1 shows Rs=0.8 for impurity/main component co-elution (left panel, Model A) and Rs=0.8 for impurity/impurity co-elution (right panel, Model B).

Results: The algorithm was able to consistently identify 0.05% impurities when co-eluting with the main component (Rs ≥ 0.8) as well as when co-eluting with another impurity (Rs ≥0.5). In the case of two co-eluting impurities from 0.05% to 1% (Rs ≥ 0.5), the quantification error ranged from +10.6% to -16.7%. In the case of an impurity co-eluting with the main component (Rs ≥ 0.8), the quantification error was 4.4-8.9% for 1% impurity and 109-184% for 0.05% impurity. The precision was excellent for the range of 0.05-1.0% impurities with the RSD being 1.0-3.0% for 1% impurity and 4.0-14% for 0.05% impurity. The evaluation of the effect of linearity degradation, potentially introduced by the Photodiode array detector, on the quantification accuracy confirmed that any linearity value below 2.5 AU introduce a high imprecision when quantification is attempted on low level co-eluting impurities (Figure 2, simulated data).

Conclusion: This peak resolution by MCR-ALS approach provides fast and robust identification and quantification of co-eluting impurities and the factorization function seems a suitable method for improving the quantification accuracy for lower-level impurities. This software concept could therefore find a wide range of applications in pharmaceutical analyses.

References: 1. S. Arase et al. / J. Chromatogr. A 1469 (2016) 35–47

Fig. 1 – Deconvolution in Impurity/main component co-elution (Left). Impurity/impurity co-elution (Right).



Fig. 2 – Effect of linearity degradation on quantification accuracy for 0.05% impurity (peak area).