New pharmaceutical applications involving Near Infrared spectroscopy as a PAT compliant process analyzer

Taking into account the « Process Analytical Technology » concept as described by the FDA in 2004, the main objective of the present work was to study and by the way to confirm the potentiality of NIR spectroscopy for the study of various types of pharmaceutical formulations. Parallel to the development of quantitative NIR applications, a secondary objective was to investigate whether the accuracy profile approach could be in any cases extended to the validation of NIR quantitative methods.In the first place, a qualitative NIR method was developed to identify routinely through polyethylene bags the API raw materials being dispatched in a pharmaceutical manufacturing line. The results showed that chemical but also grade and provider identification could be achieved. Further, as a specific case study of API raw material polymorphism, the interchangeability between NIR spectroscopy and X-Ray powder diffraction was demonstrated for the quantification of pure crystalline form II of fluconazole in binary polymorphic mixture (form II + III). It was then concluded from this part of our work that such methods could consist in the very first PAT checkpoint of any pharmaceutical manufacturing process allowing to easily avoid any further misuse of these raw materials. Later on, NIR as a PAT tool to control pharmaceutical formulations was considered. A NIR method for the API determination in a low dose syrup was first developed (2 % (w/v)). From the validation results, the accuracy profile enabled to visualize the accuracy of the NIR method all over the considered API range, it guaranteed that at least 95 % of the actual and future results will be included within the ± 5 % acceptance limits. Further, the validated method allowed successfully to monitor the API concentration during the mixing between lab-scale samples. Further, following the pharmaceutical pellets process flow chart, an «off/at-line» method able to determine the API content of the non-coated pellets was developed and validated. The method robustness was successfully tested with 2 new API concentration levels, blends of validation batches and industrial batches. The interchangeability of the NIR method with the reference HPLC method was further demonstrated. Prior to the packaging steps, the coated pellets moisture content specifications were evaluated with a developed and validated «off/at-line» NIR method while the feasibility of an «in-line» NIR method for API evaluation was investigated. NIR spectroscopy was able to qualitatively assess the API content of coated pellets in a great variety of particles streams. One quantitative NIR method fitted for one specific particle stream was developed and validated: the validation results demonstrated that accurate quantitative measurements from moving particles could be achieved with NIR spectroscopy.The last chapter of the work was focused on the development of PAT tools to assess 3 critical quality attributes of silicone-based drug reservoirs. NIR spectroscopy successfully allowed the real time monitoring of the crosslinking process while Raman mapping enabled to evaluate the API distribution within the drug reservoirs. Finally, a NIR method able to determine the API content of the reservoirs was successfully developed and validated. The accuracy profile demonstrated that the NIR method could efficiently replace the actual HPLC reference analysis involving a very time consuming sample preparation step. As a conclusion, the interest of NIR spectroscopy as a PAT tool was confirmed and its reliability as a PAT process analyzer for the evaluation of multiple critical quality attributes of various types of pharmaceutical formulations was clearly demonstrated by means of several practical examples. The present work also showed that the use of the accuracy profile could be extended without any problem to the validation of NIR quantitative methods and this profile could be a very interesting decision tool in the framework of the selection of the most appropriate calibration model.