(T1330-11-61) NIR-Based Quality Analytics for Semisolid Extrusion 3D Drug Printing

Purpose: 3D drug printing is an advanced manufacturing technology that can produce small-scale personalized medicines as well as dose forms and release profiles not attainable by traditional manufacturing methods. In traditional manufacturing, quality attributes are measured on representative samples of the entire lot. This approach is not optimal for small-batch 3D printing, since printing errors may affect only a single tablet and would likely be missed by random sampling. Rapid, non-destructive, and reliable analytical methods would ensure consistent quality for each print. Quality analytics for 3D printed tablets using near-infrared (NIR) spectroscopy with principal component analysis (PCA) was utilized to distinguish key quality attributes, including right drug, right dose, and other quality measures such as moisture content. Thus, the objective of this study is to demonstrate that NIR coupled with PCA can rapidly identify print errors linked to variations in API content, moisture, or other quality attributes. This strategy will help to support 3D drug printing for point-of-care, personalized medicine, clinical trial supply, and pre-clinical animal testing.

Methods: 3D printed tablets were printed using the FABRX M3DIMAKER2 (FabRx Ltd., London, UK) with the semisolid extrusion (SSE) print head. The head was loaded with an extrudable paste composed of 0%, 20% 30% or 40% w/w hydrochlorothiazide (HCT) with lactose monohydrate, polyvinylpyrrolidone k90 and sodium carboxymethyl cellulose at 3:1:3 mass ratio.  The powder was mixed in a Turbula® mixer (Glen Mills). The 3D printing ink was based on the work of Suárez-González et. al. [1] 20 tablets were printed for each HCT level producing 80 total tablets.  Water was added at a 3:8 (powder to water by weight) ratio to obtain an ideal viscosity for SSE.  The 15 mm tablets (3.6 mm height) were printed (see Figure 1) with the following printer settings: 50 °C, three layers, 1.2 mm nozzle diameter, 2 perimeters, print speed = 10 mm/s, extrusion multiplier=15, fill density = 40%. Printed tablets were freeze dried (LabConco Freezone 4.5).  NIR measurements were collected in duplicate using the Metrohm RCA: 400‑2500 nm, 2 nm resolution, average 32 scans per sample. Spectra were preprocessed using the standard normal variate (SNV) [2] and mean centered prior to classification by PCA [3].

Results: Normalized NIR spectra of the 3D printed tablets are plotted in Figure 2 along with the NIR spectra of the individual components for comparison. The tablets with four different levels of HCT can be differentiated by the strong HCT absorbances at 2034 nm and 2250 nm. The NIR spectra clearly distinguishes the placebo tablets from the others, showing a missing peak for the API at 2034 nm. The PCA scores and loading plots for PC1 (57% explained variance) and PC2 (31% explained variance) are shown in Figure 3. The four different formulations are easily classified by PCA and all but two samples lie within the 95% confidence ellipses.  The two outlier samples had strong NIR baseline shifts, often linked to physical properties such as density/porosity. PC1 has high positive loadings in the HCT absorption regions. The negative loadings of PC1 are associated with the excipient absorption peaks. The API and excipient covariance are described by a single principal component because an increase in API content necessitates a decrease in excipient content. PC1 can be interpreted as the variation in HCT concentration, with NIR measurements of higher API tablets plotted further to the right in the loadings plot. The tablets containing the four different levels of HCT are grouped along PC1 and the groups are sufficiently separated for classification. PC2 has loadings near 1900 nm, the OH stretch region. PC2 describes variation in water content. NIR measurements for tablets having lower water content lie further to the top of the scores plot. For confirmation, three 20% HCT tablets were left in ambient room conditions to absorb moisture for nine days. The PC2 scores for all three tablets were -9.4, -9.9, and -9.6 (red points in Figure 3) confirming that PC2 is related to moisture content. In this case the increased moisture shifted these three samples outside the 95% confidence ellipses. The tilt observed in the ellipses representing the 20%, 30%, and 40% HCT groups suggests an inverse relationship between moisture content (PC2) and HCT% in the tablets (PC1). This inference is based on an increase in tablet weight resulting from higher moisture content, while the quantity of HCT remains constant. HCT and water content may be monitored and controlled for each printed tablet by placing upper and lower control limits for PC1 and PC2.

Conclusion: NIR coupled with PCA distinguishes the API concentration differences, moisture variation, and differentiates the placebo from API-containing tablets, which can be used to identify certain critical print errors. NIR with PCA shows potential as a quality tool for 3D drug printing at the point of care to support a pharmacist-friendly pass/fail model.

References: 1. Díaz-Torres, E.; SuárezGonzález, J.; Monzón-Rodríguez, C.N.; Santoveña-Estévez, A.; Fariña, J.B. Characterization and Validation of a New 3D Printing Ink for Reducing Therapeutic Gap in Pediatrics through Individualized Medicines. Pharmaceutics 2023, 15, 1642.
2. F. Menges "Spectragryph - optical spectroscopy software", Version 1.2.16.1, 2022
3. Hammer, Ø., Harper, D.A.T., Ryan, P.D. 2001. PAST: Paleontological statistics software package for education and data analysis. Palaeontologia Electronica 4(1): 9pp.

Acknowledgements: The authors gratefully acknowledge funding from SBIR Phase II funding from NIST, grant # 70NANB23H209.


3D printed HCT tablets (20%)


Normalized NIR spectra of 80 3D printed tablets grouped in color by HCT content (top) plotted for comparison with NIR spectra of the individual tablet components (bottom).


PCA scores (top) and loadings (bottom) plot for PC1 and PC2.