Carlos Lee works for Lyndra Therapeutics, Watertown, MA.

Carlos Lee

Questions often arise about the completeness and robustness of assay and content uniformity procedures in oral solid dosage formulations. Although FDA allows up to ± 10% API loading variability for brand-name formulations (and ± 20% for generics), low and/or highly variable results often trigger debate about their source. Analysts sometimes trace them to the formulation, while formulators may blame the analytical methods used to study them.

	A semi-automated sample preparation and extraction device was introduced to the market in 2015, with the goal of speeding up, simplifying, and providing more robust and complete extraction of APIs (and impurities) from solid oral dosage formulations. This article summarizes results of comprehensive equipment testing.

	Over the past 25 years, significant strides have been made in liquid chromatography. The advent of ultra-high pressure liquid chromatography (UHPLC) and the resurgence of supercritical fluid chromatography (SFC) have led to significant reductions in chromatographic run times. It is now standard for potency and purity testing based on liquid chromatography and SFC methods for pharmaceutical dosage formulations to take 2–30 minutes. Ten to 15 years ago, this testing required more than an hour.

	However, before content uniformity and assay/purity testing samples can be injected into chromatography equipment, they must be prepared, and API and impurities extracted. Typical sample preparation and extraction times are on the order of 30 minutes or more for immediate release (IR) type solid oral dosage formulations and can run between 4 and 24 hours for controlled release (CR) type formulations.

	A 1991 survey in LCGC (1) found that sample preparation and extraction accounted for two-thirds of the time that analytical scientists spent on testing and analyzing samples, and not much has changed in 25 years (2). Automated and semi-automated sample preparation and extraction techniques such as microwave assisted extraction (MAE) and accelerated solvent extraction (ASE), applicable to IR formulations, have helped reduce sample preparation and extraction times (3,4). However, most sample preparation for content uniformity and assay/purity testing is still done manually, and requires such steps as sonication, stirring, and mechanical shaking. In addition, results continue to reflect inefficient, lengthy, and incomplete extraction of APIs and impurities/degradation products from drug product matrices (5).

	Based on extensive testing, the PrepEngine (Distek, USA) allowed preparation of samples and API extraction from solid oral dosage formulations to be accomplished in less than 10 minutes. Testing evaluated variables such as processing speeds, extraction times, extraction volumes, and extraction solvents and their impact on the extent of recovery of APIs and impurities/degradation products. It also compared these values with results achieved via manual preparation and extraction procedures for the same formulations.

	Key steps in sample preparation and API extraction

	Three critical steps are required for effective and efficient sample preparation and extraction in pharmaceutical solid oral dosage formulations (6). First, the formulation undergoes dispersion or disintegration to produce small granules. Following the matrix’s increase in surface area, the API is solubilized in the dissolving solvent, then the undissolved excipients that remain in the resulting solution are removed by either centrifugation or filtration, providing a solution free of particles for chromatographic analysis.

	The PrepEngine was designed to focus on the two most critical and rate-limiting steps in the procedure: particle size reduction and solubilization of the API. Stainless steel or polypropylene blades within the device’s tall, cylindrically shaped PrepTubes serve as a homogenizer and allow for dispersion of solid oral dosage formulations. Simultaneously, the blades create intense vortexes within the tubes, which facilitates interaction between the API and the dissolving solvent. 

	Solid oral dosage formulations: One IR tablet dosage formulation, three CR tablet dosage formulations, and one softgel IR capsule solid oral dosage formulation were obtained from Pfizer Inc. (Groton, CT) and evaluated using the PrepEngine. Results were then compared with results from the manual validated sample preparation and extraction procedures that had been developed for these formulations. Apart from the IR tablet, the formulations that were tested were chosen for the complexity of their manual sample preparation and extraction procedures. Information about the various formulations is provided in 

.  Samples were manually prepared as described in 

and analyzed using the validated high-performance liquid chromatography (HPLC) methods developed for the compounds involved in this study.
	

	Sample preparation and extraction procedure

	In all but a few of the studies conducted, the extraction solvents that were developed for and used in the manual sample preparation and extraction procedures were retained. In cases where a sequential addition of diluents was called for, such as aqueous followed by organic, extraction was simplified by pre-mixing the diluents and using the mixture to extract the API and impurities/degradation products from the formulation. Extractions were performed on single, intact tablets as well as on a composite of tablets (typically five tablets were used per composite).

	The general procedure employed for sample preparation and extraction using the PrepEngine involved accurately transferring a specified volume of pre-mixed extraction solvent to the polypropylene PrepTubes. A Dosimat Plus titrator (Metrohm, USA) was used to deliver the dissolving solvent to the PrepTubes. After addition of the extraction solvent to the PrepTubes, a single intact tablet/capsule was weighed and transferred to the tubes. The tubes were then capped, and the samples were processed for specified times and speeds.

	With the PrepEngine, samples can be processed for up to four hours, with speeds ranging from 500 rpm to 6000 rpm. Following development work on single replicates to optimize conditions for extraction of the actives and impurities/degradation products from the various formulations, 10 replicates were evaluated in parallel under the optimized conditions. Composite samples involving up to five tablets were also evaluated.

	Initial studies with the PrepEngine were performed on Compound A, a 100-mgA (milligram active) immediate release tablet formulation. IR formulations are designed to rapidly disintegrate when exposed to aqueous conditions. Results following extraction of the active component of Compound A using the PrepEngine are shown in 

. Extractions were performed at 3000 rpm for two minutes, and provided potency results essentially identical to that obtained by the manual validated method. While the manual method, which used a reciprocal shaker to disperse and extract the active from Compound A, can provide complete extraction of the active, the PrepEngine was capable of performing the extraction in two minutes compared to the 30 minutes required by the manual method. The above corresponds to savings of nearly 93% in extraction time for the 100-mgA Compound A IR tablet. Furthermore, with the ability to perform extractions on up to 10 samples in parallel, the PrepEngine device has the potential to increase throughput in the laboratory, when compared to semi-automated approaches such as ASE and the tablet processing workstation (TPW). Additionally, no new degradation products and/or impurities were observed in the PrepEngine prepared samples, and no carryover issues were evident after cleaning and drying the PrepTubes.
	

	Compound B - CR (ECS tablet technology): Compound B is an 11-mgA CR tablet formulation based on the extrudable core system (ECS) technology developed in 2009 (7). ECS tablets are single layer core tablets with a semi-permeable coating. The active is delivered osmotically through a hole in the coating.

	The primary components in the 11-mgA tablet formulation include the API, and excipients such as copovidone, sorbitol, hydroxyethylcellulose, magnesium stearate, and cellulose acetate. As indicated in 

, the 5.5-hour manual sample preparation procedure for Compound B involved the sequential addition of aqueous and organic solvents, followed by filtration of the final solution through a 0.45 µm nylon filter. Using the pre-mixed manual method final diluent composition (85/15, v/v, 0.1% HClO4/acetonitrile), the PrepEngine extracted the active and degradants/impurities from Compound B tablets completely within 5 minutes, with processing speeds of 5000 rpm (

	Compound C and D (SCT technology): The most challenging controlled release formulation investigated in this study involved those based on the swellable core technology (SCT) (8,9). SCTs are osmotic pump bilayer tablets in which the active layer and osmotic or sweller layer is surrounded by a thick, hard, semi-permeable, polymer membrane. Once exposed to an aqueous environment, the osmotic layer absorbs water and swells. As the osmotic layers swells, it pushes onto the active layer, forcing the hydrogel suspension of active and excipients to exit through a small laser drilled hole in the coating. This osmotic pump mechanism provides drug delivery at a constant rate.

	As indicated in the sample preparation and extraction procedure described in 

, manual extraction of the active from Compound C, a 20-mgA SCT formulation, was extremely challenging and time-consuming. The 24 or more hours required to extract the active from the formulation is due primarily to the thick and hard polymer coating, and the polymer in the core of the tablet. Both serve to impact tablet wettability, thus decreasing the rate at which water penetrates the coating to initiate swelling and dispersion of the formulation. Additionally, the presence of polyethylene oxide (PEO) in the osmotic and drug layers facilitates gelling of the formulation, leading to the formation of a highly viscous suspension.

	Initial sample preparation and extraction studies on the 20-mgA Compound C SCT tablet focused on identifying more suitable dissolving solvents capable of facilitating rapid dispersion of the tablets while having no negative impact on the viscosity of the extraction solution. Results from exploratory work on Compound C are shown in 

. As shown, all extraction solvent systems containing acetonitrile and/or water did not facilitate complete dispersion of the SCT tablet. Increasing the processing speed or processing time did not have any appreciable impact on the dispersion rate when acetonitrile and/or water were present in the extraction solvent. The presence of acetonitrile (ACN) and/or water in the diluent system caused the bilayer tablets to swell and made them rubbery. The above made it difficult for the rapidly moving stainless steel blades to fully disperse the intact tablet. In sharp contrast, diluent systems containing alcoholic solvents such as methanol or ethanol and/or tetrahydrofuran (THF) facilitated complete dispersion of the SCT tablets. While solvent systems containing THF allowed for complete dispersion of the SCT tablets within single digit timeframes, they were not studied further because the presence of THF in the dissolving solvent led to chromatographic peak splitting issues due to the differences between the highly organic dissolving solvent and the highly aqueous initial mobile phase chromatographic conditions. Reducing the injection volume did not minimize the observed peak splitting issues.
	

	Methanol, when used alone, facilitated complete dispersion of the SCT tablets; however, the viscosity of the solution tended to slowly increase with time, likely due to the slow solubilization of PEO in the diluent. In contrast to methanol alone or THF-based solvent systems, ethanol by itself or in combination with methanol proved to be ideally suited for the 20-mgA Compound C SCT tablets.

	Not only did ethanol or ethanol/methanol mixtures provide rapid and complete dispersion of the SCT tablets, the viscosity of the resulting solutions remained visually unchanged (when compared to the viscosity of the sample diluent in the absence of the drug product matrix). The only noteworthy drawback with use of the above alcoholic solvent systems was the observation of peak splitting during chromatographic analysis when the injection volume was at or above 4 µL. The observed peak splitting issues were, however, easily overcome by reducing the injection volume to 2 µL. The nominal concentration was increased by a factor of two to maintain the same nominal on-column concentration.

	Following developmental work to optimize the processing speed and processing time, the 20 mgA Compound C SCT tablets were analyzed in replicates of 10 via the PrepEngine. As shown in 

, the PrepEngine quantitatively extracted the active ingredients from the 20 mgA Compound C SCT tablets within six minutes, when processed at 3000 rpm. The resulting chromatograms of the PrepEngine extracted samples were identical to those obtained via the manual method. Essentially identical results were obtained using ethanol or 50/50, v/v, ethanol/methanol extraction solvents.

	Sample solutions containing 100% organic solvents can sometimes lead to chromatographic peak splitting due to incompatibility between the sample solution and the mobile phase. Additionally, the presence of high levels of PEO and other high molecular weight excipients in the sample solution can affect HPLC column lifetime and compromise HPLC systems. Attempts were made to reduce this impact by assessing and introducing sample clean-up procedures following extraction with 50/50 ethanol/methanol. The sample clean-up step evaluated involved diluting the ethanol/methanol extracted solution (either pre- or post-centrifugation) with water to obtain a 40/40/20 (v/v/v) ethanol/methanol/water mixture, followed by the addition of 1 g of silica gel.

	The solution was vortexed and centrifuged, and a sample of the clear non-viscous supernatant was analyzed by HPLC. The purpose of adding approximately 20% water to the 50/50 ethanol/methanol extract was to make the solution more compatible with the initial mobile phase composition. Additionally, silica gel was added to the final solution mixture in order  to remove (by adsorption) PEO and other high molecular weight excipients from the solution matrix, thereby reducing the viscosity of the solution and minimizing its potential impact on column lifetime.

	Assay results following sample clean-up of the initial 50/50 ethanol/methanol extraction solvent are captured in 

. Results were essentially identical to that obtained when directly injecting the 50/50 ethanol/methanol based extraction solution at 2 µL, suggesting that the sample clean-up step had no impact on the overall extraction efficiency.
	

	Challenging manual sample preparation for SCT

	Results of tests on the 20-mgA Compound C SCT formulation were subsequently applied to Compound D, another SCT formulation. As indicated in 

, the manual sample preparation procedure for the 10-mgA Compound D SCT formulation was quite challenging, involving cutting the tablets in half to expose the inner core, followed by shaking on a reciprocal shaker for 4-hours using 100% ACN as the extraction diluent. A near 40-fold improvement in the rate of extraction was realized when the 10-mgA Compound D SCT formulation was extracted with the PrepEngine and 50/50 ethanol/methanol (

	The results following sample clean-up were also unchanged from those obtained with no sample clean-up. The results obtained for the two SCT formulations suggest that the use of alcoholic solvents such as 50/50 ethanol/methanol or 100% ethanol in combination with the PrepEngine might allow for a universal sample preparation and extraction approach for SCT formulations. The schematic representation of such an approach is captured in

. The universality of the proposed approach is, however, highly dependent on the solubility of the API of interest in alcoholic solvents.
	

	Compound E-softgel capsule: An interesting formulation evaluated with the PrepEngine was Compound E, a 20-mgA softgel capsule formulation. The primary excipients in this formulation included polyethylene glycol (PEG) and polysorbate 80. The manual sample preparation and extraction condition for this formulation was a bit atypical for a pharmaceutical dosage formulation. The procedure entailed shaking the sample in a water bath at elevated temperatures (45 ºC), for up to four hours.

	The capsule formulation was extracted in 50/50 potassium phosphate buffer, pH 6.8/methanol using 100-mL volumetric flasks. Initial exploratory work on the PrepEngine showed that Compound E could be extracted from the softgel capsule within three minutes at 3000 rpm with potency values between 97.8–98.7%. Within this timeframe, the softgel capsules were easily ruptured, allowing the active and excipients to be released from the capsule shell and interact with the diluent.

	Extending the process time pulverized the capsule shell

	Additional development work showed that extending the processing time from three to six minutes allowed for the capsule shell to be totally pulverized rather than simply ruptured, thereby providing for a more quantitative and more robust extraction procedure (

). Under the optimized PrepEngine condition, and utilizing the same diluent as that used in the manual sample preparation and extraction procedure, the PrepEngine was able to quantitatively extract the API and impurities/degradation products from Compound E within six minutes (as compared to the two to four hours required by the manual sample preparation procedure). No new degradation products and/or impurities were observed when Compound E was extracted with the PrepEngine.

	Testing showed that the PrepEngine facilitated rapid and quantitative extraction of APIs and degradation products/impurities from IR and CR pharmaceutical solid oral dosage formulations. Extraction efficiencies and precision were comparable to levels obtained with often more time-consuming and cumbersome manual sample preparation and extraction procedures. 

	With the capability to process up to 10 samples in parallel, throughput is excellent when compared to that of most semi-automated procedures such as accelerated solvent extraction and the tablet processing workstation. The device is simple and user friendly, and, when coupled with alcoholic solvents such as mixtures of ethanol and methanol or 100% ethanol, allows for the development of a universal sample preparation and extraction approach for SCT formulations.

	The authors would like to thank Timothy Graul of Pfizer, Groton, CT, for his support of this study and for his detailed review of the manuscript. The authors would also like to thank Raymond Chen, Julie Wood, Michele Guo, Gary Haggan, and other Pfizer colleagues who provided the samples that were used in the studies.

	1.  R.E. Majors, LC-GC 20 (12) (2002) 1098–1113.
	2. R.E. Majors, LC-GC 33 (11) (2015) 46–51.
	3. Lee, C. W. J. Gallo, W. Arikpo, V. Bobin, J. Pharm. and Biomed. Anal. 45, 2007, 565.
	4. Lee, C. W.,  J. Amer. Pharm. Rev., Sept/Oct. Issue, 2007.
	5. B. Nickerson, K. R. Lang, “Agitation and Particle Size Reduction Techniques,” in: B. Nickerson (Ed.), Sample Preparation of Pharmaceutical Dosage Forms, 2011, pp. 43–61. 
	6. D. Brannegan, C.W. Lee, J. Wang, L. Taylor, “Extraction Techniques Leveraging Elevated Temperatures and Pressures,” in: B. Nickerson (Ed.), Sample Preparation of Pharmaceutical Dosage Forms, 2011, pp. 93–128. 
	7. K. W. Waterman, B. C. MacDonald, M. C. Roy, J. Control. Rel. 134 (3) (2009) 201–206. 
	8. D. R. Swanson, B. L. Barclay, P. S. L. Wong, F. Theeuwes, Am. J. Med. 83 (6B) (1987) 3–9.
	9. A. G. Thombre, L. E. Appel, M. B. Chidlaw, P. D. Daugherity, F. Dumont, L. A. Evans, S. C. Sutton, J. Control. Rel. 94 (1) (2004) 75–89.

Solid Dosage Drug Development and Manufacturing

	When referring to this article, please cite it as C. Lee, A. Charles, B. Nickerson, G. Johnson, and J. Warzeka, “Speeding Sample Preparation and API Extraction from Solid Oral Dosage Formulations," 

Solid Dosage Drug Development and Manufacturing 

Carlos Lee, Beverly Nickerson, Gail Johnson, and John Warzeka 

all worked for Pfizer Worldwide Research and Development in Groton, CT during the performance of the study. 

was a graduate student at the University of Missouri’s Chemistry Department, Columbia, MO. 

(clee@lyndra.com) currently works for Lyndra Therapeutics, Watertown, MA.

To whom all correspondence should be addressed.

Articles

Laboratory testing found that a novel approach reduced the time required for sample preparation from hours to minutes. This article summarizes test methods and results.