Blend uniformity end-point determination using near-infrared spectroscopy and multivariate calibration

Area of Science:

• Pharmaceutical Manufacturing
• Analytical Chemistry
• Spectroscopy

Background:

• Ensuring blend uniformity is critical for pharmaceutical solid dosage forms to guarantee consistent drug delivery and therapeutic efficacy.
• Traditional methods for assessing blend uniformity can be time-consuming and may not provide real-time process monitoring.
• Near-infrared (NIR) spectroscopy offers a rapid, non-destructive analytical technique suitable for in-line process monitoring.

Purpose of the Study:

• To develop and validate a multivariate calibration model using NIR spectroscopy for determining the blend uniformity end-point in pharmaceutical manufacturing.
• To enable real-time prediction of active pharmaceutical ingredient (API) content during the blending process.
• To assess the robustness of the NIR-based model against variations in manufacturing parameters, such as water content during granulation.

Main Methods:

• A partial least-squares (PLS) multivariate calibration model was developed using 21 off-line, static calibration samples with varied concentrations of API and excipients.
• A micro-electrical-mechanical-system (MEMS) based NIR spectrometer was employed for spectral data acquisition.
• NIR spectra were preprocessed using standard normal variate (SNV) and second derivative Savitsky-Golay filtering to minimize spectral variations between static and dynamic measurements.

Main Results:

• The real-time API-NIR (%) predictions from the PLS model on 67 production batches ranged from 93.7% to 104.8% (SD: 0.5%–1.8%), confirming blend homogeneity.
• These results were corroborated by content uniformity analysis using High-Performance Liquid Chromatography (HPLC) on manufactured tablets (95.4%–101.3%, SD: 0.5%–2.1%).
• The model demonstrated reliable performance even with off-target batches affected by granulation water content, with API-NIR (%) predictions between 94.6%–103.5% (SD: 0.7%–1.9%).

Conclusions:

• A multivariate NIR-based PLS calibration model provides an effective and accurate method for real-time determination of blend uniformity end-point in pharmaceutical manufacturing.
• This approach allows for robust process monitoring and control, ensuring consistent product quality and potentially reducing manufacturing cycle times.
• The developed systematic approach using off-target data enhances the reliability of blend uniformity assessment and process optimization.