A carregar...
Publicação
Modelação multivariada conjugada com espetroscopia NIR para a otimização de uma formulação contendo elevado teor em Vitamina C
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 3.92 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Este trabalho propõe o desenvolvimento de formulações de comprimidos de elevado teor em vitamina C através do uso de modelos matemáticos multivariados, a partir de dados correspondentes à composição das formulações, e da aquisição de espetros de infravermelho próximo (NIR). Inicialmente, foram desenvolvidos e produzidos 20 ensaios de comprimidos de formulações contendo celulose microcristalina (MCC), polivinilpirrolidona (PVP) e sorbitol (SBT) e quantidades padrão de vitamina C (Vit. C) (1000 mg) e de estearato de magnésio (MS) (15 mg). As 20 formulações produzidas foram de nidas de acordo com um desenho experimental de mistura. As formulações para apresentarem requisitos de qualidade, teriam de ter valores de friabilidade, dureza e desagregação conforme as seguintes especificações: friabilidade inferir a 1%, dureza superior a 10 kPa, e desagregação inferior a 30 minutos (ideal) ou 45 minutos (aceitável). Foram desenvolvidos modelos de regressão com base na regressão linear múltipla para previsão dos testes. O modelo para a friabilidade demonstrou que os ingredientes SBT, MCC e PVP contribuem para a sua diminuição. O modelo para a dureza, demonstrou que os ingredientes SBT, MCC e PVP contribuem para o seu aumento. O modelo para a desagregação demonstrou que a MCC e a PVP contribuem para a diminuição da desagregação, enquanto o SBT não tem significância estatística. A partir dos modelos criados, tentou-se criar uma formulação ideal para um objetivo de friabilidade (0,5%), dureza (15 kPa) e desagregação (15 min). Criou-se igualmente um design-space de acordo com estas especificações de qualidade. O melhor design-space encontrado neste trabalho apresentou valores de 2 000 DPMO (defeitos por milhão de oportunidades), e foi executado para um objetivo de friabilidade (0,5%), dureza (15 kPa) e desagregação (30 min).
Por último, a partir dos espetros NIR adquiridos, foram desenvolvidos e validados modelos PLS para estimativa de todos os ingredientes presentes nas formulações e para os testes de desagregação, friabilidade e dureza. Os modelos baseados nos espetros NIR para previsão da composição das formulações geraram valores muito bons em termos das figuras-de-mérito (RMSECV , R2cv e RER). O modelo para a PVP foi o que apresentou um desempenho menor. Para os testes de desagregação, dureza e friabilidade foram construídos dois modelos PLS, um apenas com base nos espetros NIR, e outro conjugando os espetros NIR com as composições das formulações. Verificou-se que dos três modelos construídos para a friabilidade, o que demonstrou melhor desempenho foi o modelo construído apenas com os espetros NIR. Já para a dureza, o modelo com melhor desempenho foi o modelo construído a partir da combinação das composições com os espetros NIR. Para a desagregação, o modelo com melhor desempenho foi o modelo construído por MLR a partir das composições. Em relação aos parâmetros, a dureza foi a que apresentou os modelos com melhor desempenho, seguida da desagregação e friabilidade.
This work proposes the development of high Vitamin C content tablets resourcing to of multivariate models based on formulations composition and NIR spectra obtained directly from tablets. Initially, 20 formulations were generated according to a mixture design and used to produce tablets. Formulations were based on a mixture of components at variable amounts (microcrystalline cellulose (MCC), polyvinylpyrrolidone (PVP) and sorbitol (SBT)) and components at fixed amounts (vitamin C (Vit. C) (1000 mg) and magnesium stearate (MS) (15 mg)). In order to comply with product quality requirements, the produced formulations should have values of friability, hardness and disintegration according to the following specifications: friability below 1%, hardness above 10 kPa, and disintegration time below 30 minutes (optimal) or 45 minutes (acceptable). Explanatory models based on multiple linear regression were estimated to predict the aforementioned properties from composition of variable components (MCC, PVP and SBT). The model for friability showed that the SBT, MCC and PVP ingredients, contribute all to a reduction. The model for hardness showed that the SBT, MCC and PVP ingredients, contribute to increase this property. The model for disintegration time showed that both MCC and PVP contribute for its reduction, while SBT was not statistically significant. From these models, an optimal formulation for a purpose of friability (0.5%), hardness (15 kPa) and disintegration (15 min) was estimated. Additionally, the definition of the design-spaces for this product was assessed. The best design-space yielded values of 2000 DPMO (defects per million opportunities), and was executed for a target of 0,5% friability, 15 kPa hardness and 30 min disintegration time. Finally, from the NIR spectra acquired directly from tablets in diffuse reflectance mode, PLS models were calibrated for the three propertiesf and for all ingredients present in the formulations. All models were found adequate according to the figures-of-merit RMSECV, R2cv and RER). Models for the estimation of PVP were those that achieved the loweste performance in terms of these figures-of-merit. For the disintegration, hardness, and friability two PLS models, one based solely on the NIR spectra, and a second based on the combination of NIR spectra and tablet compositions were performed. It was found that for the three models built for friability, the model that demonstrated the best performance, was the model constructed using NIR spectra only. For hardness, the model with the best performance was that built combining compositions with NIR spectra. For disintegration, the model with the best performance was constructed by MLR from the compositions. Regarding the analysed parameters, the hardness was the parameter that presented models with better performance, followed by disintegration time.
This work proposes the development of high Vitamin C content tablets resourcing to of multivariate models based on formulations composition and NIR spectra obtained directly from tablets. Initially, 20 formulations were generated according to a mixture design and used to produce tablets. Formulations were based on a mixture of components at variable amounts (microcrystalline cellulose (MCC), polyvinylpyrrolidone (PVP) and sorbitol (SBT)) and components at fixed amounts (vitamin C (Vit. C) (1000 mg) and magnesium stearate (MS) (15 mg)). In order to comply with product quality requirements, the produced formulations should have values of friability, hardness and disintegration according to the following specifications: friability below 1%, hardness above 10 kPa, and disintegration time below 30 minutes (optimal) or 45 minutes (acceptable). Explanatory models based on multiple linear regression were estimated to predict the aforementioned properties from composition of variable components (MCC, PVP and SBT). The model for friability showed that the SBT, MCC and PVP ingredients, contribute all to a reduction. The model for hardness showed that the SBT, MCC and PVP ingredients, contribute to increase this property. The model for disintegration time showed that both MCC and PVP contribute for its reduction, while SBT was not statistically significant. From these models, an optimal formulation for a purpose of friability (0.5%), hardness (15 kPa) and disintegration (15 min) was estimated. Additionally, the definition of the design-spaces for this product was assessed. The best design-space yielded values of 2000 DPMO (defects per million opportunities), and was executed for a target of 0,5% friability, 15 kPa hardness and 30 min disintegration time. Finally, from the NIR spectra acquired directly from tablets in diffuse reflectance mode, PLS models were calibrated for the three propertiesf and for all ingredients present in the formulations. All models were found adequate according to the figures-of-merit RMSECV, R2cv and RER). Models for the estimation of PVP were those that achieved the loweste performance in terms of these figures-of-merit. For the disintegration, hardness, and friability two PLS models, one based solely on the NIR spectra, and a second based on the combination of NIR spectra and tablet compositions were performed. It was found that for the three models built for friability, the model that demonstrated the best performance, was the model constructed using NIR spectra only. For hardness, the model with the best performance was that built combining compositions with NIR spectra. For disintegration, the model with the best performance was constructed by MLR from the compositions. Regarding the analysed parameters, the hardness was the parameter that presented models with better performance, followed by disintegration time.
Descrição
Tese de mestrado, Engenharia Farmacêutica, Universidade de Lisboa, Faculdade de Farmácia, 2015
Palavras-chave
Comprimidos de vitamina C Delineamento experimental Quimiometria Espetroscopia de infravermelho próximo Indústria Farmacêutica Teses de mestrado - 2015