Category: Formulation and Quality
Purpose: The fast and efﬁcient separation of complex mixtures consisting of multiple components including impurities as well as principal drug substances remains a challenging application for liquid chromatography in the ﬁeld of pharmaceutical analysis. In some cases, complete separation of those complicated components is difficult even after a long time method optimization for improving separation. A novel data analysis technique named i-PDeA II (Intelligent Peak Deconvolution Analysis II) was developed for extracting two or more target peaks from unseparated peaks using data from three-dimensional photodiode array (PDA) detector with the multivariate curve resolution-alternating least squares (MCR-ALS) technique. It also employed an expectation-maximization (EM) algorithm with a bidirectional exponentially modiﬁed Gaussian (BEMG) model function as a constraint for chromatograms and a huge number of PDA spectra aligned with the time axis. The i-PDeA II function can automatically extract respective peak profiles and absorption spectra from unseparated elution band by merely specifying the wavelength range and the elution time interval. Obtained results can be directly used for spectrum identification and quantitative determination. Here we confirmed the applicability of i-PDeA II for various incomplete separations of two and three component-system.
Methods: As two-components system deconvolution, we tested the incomplete separation of cytidine and adenosine monophosphate monosodium n-hydrate (AMP) in reversed phase chromatography. The obtained deconvolution results of peak areas and spectra were evaluated by comparing the results from individual analyses of cytidine and AMP. As three-components system, we tested the separation of the positional isomer mixtures of methylacetophenone (MAP) that show usually similar retention times and a small spectral difference. In this experiment, the MAP isomers eluted in the order, o-MAP, p-MAP, and then m-MAP. In addition to preparing mixture solutions with relative concentration ratios of 100/100/100 and 100/1/100, (with concentration ratios indicated in the elution order), standard solutions of the o-MAP, p-MAP, and m-MAP components individually, which were used to prepare the mixture solutions, were also measured to confirm true values. For the 100/100/100 mixture solution, the peak resolution between o-MAP and p-MAP was 0.83 and the resolution between p-MAP and m-MAP was 0.84. The similarity between respective components in the UV spectral data obtained from the peak apex for each standard sample with a relative concentration 100 was determined to be 0.8410 for o-MAP/p-MAP, 0.9123 for p-MAP/m-MAP, and 0.9809 for o-MAP/m-MAP. Additionally, we applied i-PDeA II for the simultaneous analysis of synthetic polymer and its additives. Furthermore, the same algorithm was applied to the ultra-high speed separation of caffeine, ethylparaben, and acetophenone that showed a lack of separation to improve quantitative reliability.
Results: Cytidine and AMP were difficult to fully separate due to weak retention by the reversed phase chromatographic conditions. But the errors with respect to true values for ATP and uracil peak areas after deconvolution were −2.14% and −0.45%, respectively, which is quite small. The i-PDeA II function provided less than±1.0% error between true and simulated peak area values at Rs of 0.6, three-two-one abundance ratio of 100:100:100, using simulation data for a three-component mixture with spectra similarity of each combination was around 0.9 at peak apex. Unseparated three polymer additives were determined with GPC column separation that is not normally sufficient for small molecule separation. The peak area errors for 12 seconds analysis of the three components were dramatically improved as approximately 13% to 5%.
Conclusion: The peak deconvolution algorithm i-PDeA Ⅱ developed in this research offers faster analysis, streamlining of sample pretreatment, and the ability to obtain quantitative results based on the simple settings of time range and wavelength range. These features are especially significant for applications such as drug research and development. Through the experiments, we confirmed following conclusion: i-PDeA II includes an algorithm for reliable chromatographic peak deconvolution that was developed by applying the MCR-ALS to PDA data; Ultra-fast and accurate quantitative analysis is possible at any desired wavelength even in case of incomplete chromatographic separation; i-PDeA II can be applied to the analysis of isomers with identical molecular weights resulting in an alternative to the MS detection; reliable spectral data analysis can be obtained even after peak deconvolution.
Davide Vecchietti– Kyoto, Kyoto, Japan
Yoshiyuki Watabe– Kyoto, Kyoto, Japan
Yoshiyuki Watabe– Kyoto, Kyoto, Japan
Kosuke Nakajima– Kyoto, Kyoto, Japan
Yoshihiro Hayakawa– Kyoto, Kyoto, Japan
Kyoko Watanabe– Kyoto, Kyoto, Japan
Keiko Matsumoto– Kyoto, Kyoto, Japan
Toshinobu Yanagisawa– Kyoto, Kyoto, Japan