Andrew Johnson is analytical specialist with Gaylord Chemical Co, LLC.

Andrew Johnson

Dimethyl sulfoxide (DMSO) is a small-molecule pharmaceutical ingredient with regulated uses as both an excipient and an API (1–2). Its capabilities as a transdermal penetration enhancer and its value as cryopreservation media were first recognized in the early 1960s. In recent years, this versatile substance has been incorporated into pharmaceutical products that require a high degree of microbiological quality.

Recently, DMSO has been used in autologous cell therapies. In 2017, Novartis brought the first FDA-approved gene therapy to market (Kymriah, Tisagenlecleucel) and Kite Pharma’s Yescarta (Axicabtagene Ciloleucel) was approved in the United States as a T-cell treatment for lymphoma. Both products contain DMSO and are delivered by intravenous infusion. Sterility of the finished drug product is paramount in such treatments.

Other types of DMSO-containing parenteral products include long-acting injectable (LAI) formulations and certain combination drug/device products. In LAI applications, DMSO usually acts as a solvent to deliver a biopolymer/drug combination via subcutaneous/intramuscular injection. Another drug product in which DMSO sterility is critical is the Gvoke Hypopen (Xeris Pharmaceuticals), which was approved by FDA in 2019 and uses DMSO as a delivery solvent for glucagon peptide.

While excipients are by definition inactive substances, they are potential contributors to a pharmaceutical product’s overall bioburden. Hair and other foreign matter may be contributed to the product during filling. There may the potential for cross contamination with materials of animal origin. Additionally, excipients are not chemically inert to every method of sterilization. Even simple small molecules may undergo chemical reaction when irradiated or thermally stressed. An understanding of an excipient’s unique chemistry is helpful when developing a sterilization protocol for a finished pharmaceutical formulation.

), manufactured as a bulk pharmaceutical ingredient by Gaylord Chemical Company, is not supplied as a sterile product. Increasing interest in suitable sterilization techniques compelled the authors to evaluate the impact of established sterilization techniques on the chemical product quality of the ingredient.

Sterilization methods may be categorized as physical or chemical in nature. Typical chemical techniques may involve programmed exposure to biocidal gases such as ethylene oxide (EO) or treatment with peracetic acid in its liquid phase. DMSO is a readily oxidized substance, known to produce dimethyl sulfone (CAS [67-71-0]) in the presence of oxidants such as hydrogen peroxide. Further, it is not inert to electrophilic substances such as EO and peracids (3). Gases and liquid sterilants are most applicable for the treatment of hard surfaces (medical device components, surgical tools) and are less useful with liquid or solid pharmaceutical formulations. For these reasons, chemical sterilization methods were not evaluated in this study.

Physical methods such as thermal sterilization, irradiation, and filtration are more suitable options for attaining sterility of dimethyl sulfoxide products. DMSO poses technical challenges for these methods, however. It is a strong solvent that can dissolve/degrade polymeric filtration media. DMSO is known to thermally decompose under some conditions. The impact of irradiation on DMSO quality has not been well documented. Another interesting option is the use of extremely high pressure to terminally sterilize pharmaceutical compositions (4). Although this method is likely suitable for use with DMSO from a physicochemical stability standpoint, it is the authors’ impression that this technique has not been broadly applied to the terminal sterilization of pharmaceutical finished dosages.

After considering the established and practical options, four sterilization techniques were tested to evaluate their effects on the chemical stability of DMSO. These were e-beam irradiation, gamma irradiation, thermal sterilization using dry heat, and aseptic membrane filtration sterilization. Although DMSO has a relatively high flashpoint (95 ˚C, open cup), the service provider would not sterilize DMSO samples in an autoclave or using the moist heat method, due to safety concerns.

Product samples of Procipient (dimethyl sulfoxide 

) were provided to respected sterilization service providers, which applied the sterilization method to these samples. The treated samples were then returned to the Gaylord lab for chemical analysis, based on the set of stability indicating assays underlying the product’s International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use long-term stability program. 

 diagrams this process. An unopened control sample for subsequent sterility assurance was tested by another service provider, along with the various treated samples. The additional details of the sample plan are given in 

. Additional experimental details follow.

. The service provider (Steritek) used an electron beam accelerator rated at 10 MeV @ 20 kW. Prior to irradiation of the DMSO samples, a dose distribution study was performed using six vials of Water, USP. These samples were processed such that opposite sides of the water samples were irradiated. Calibrated dosimeters (B3WINDOSE, GEX Corporation) were used to develop dosing recommendations for the DMSO product samples. A target dose of 49–53 KGy was recommended; a delivered dose of 51 KGy was delivered. Samples were irradiated in 250 mL Type II amber glass bottles.

. DMSO samples were irradiated using a Cobalt 60 irradiation field. A dose of 30–60 KGy was specified by the service provider (Steris); a delivered dose of 30.7–37.05 kGy was delivered for an exposure time of 834 minutes; final dosage 36.96 kGy. The DMSO samples were processed in 1L Type II amber glass bottles.

. Processing of DMSO samples was performed in a hot air oven set to a processing temperature of 160 °C. Two biological indicator vials and three thermocouple vials were placed in the oven during testing. Biological indicators and thermocouples were not placed in the DMSO samples themselves. After processing, the biological indicators were cultured. No growth was seen. All samples were sealed in 100 cc amber glass Type I vials using 20 mm Teflon-coated chlorobutyl stoppers and 200 mm aluminum crimp seals.

. Sterilization by aseptic filtration is the most commonly used method of sterilization for DMSO. It has been proven in the industry to be an acceptable sterilization method for solutions or liquids that may not be sterilized in a final container (5). Aseptic filtration of DMSO is widely used in research, clinical, and regulated products. Because this is a widely used and an industry standard, the authors deemed it suitable to test one sample lot for comparative purposes.

Sterility was performed in compliance with FDA regulations 21 

 Parts 210, 211 and 820, following Nelson Laboratories Standard Test Protocol: STP0077 Rev 22. DMSO samples were processed through a closed-membrane filtration system in a cleanroom environment; an industry standard (6). DMSO product was passed through a 0.22 μm polytetrafluoroethylene (PTFE) membrane and filter housings that are DMSO compatible. Following filtration, 100 mL of product was incubated for 14 days; incubation temperature of 20–25 °C soybean casein digest broth (SCDB)/30–35 °C fluid thioglycolate medium (THIO). This test verifies the product has been successfully filtered sterilized.

Bioburden and endotoxin tests were performed to validate the effectiveness of the different sterility methods. Even though Procipient is not supplied as a sterile product, the manufacturer’s historical data has shown no adverse changes in both bacterial endotoxin tests (BET) and bioburden tests performed internally. Results for BET and bioburden were compared with the non-sterilized samples; no growth was shown in any of the processed samples. The BET utilized was the kinetic turbimetric method (Nelson Labs LAL110); for bioburden both aerobic bacteria and fungi (Nelson Labs BIO220) and a sample suitability per organism (Nelson Labs BIO930) was performed. Results of all samples submitted were <0.2 EU/mL (1:40 dilution) for BET and <1 Anaerobic/<1 fungal colony forming units/1 mL for bioburden.

A basic sterility assurance test was performed by incubating for 14 days at 20–25 °C SCDB/30–35 °C THIO. No growth was detected in both mediums.

Of the four sterilization methods evaluated, only one (aseptic filtration) met all chemical quality requirements. Clearly, E-beam irradiation, gamma irradiation, and dry heat sterilization negatively impact the quality of DMSO. A summary of the chemical stability results is given in 

The results from sterility assurance testing on all samples were consistent, and positive across all four sterility test methods. For all samples, no growth was reported in either soybean casein digest broth (SCBD) or fluid thioglycolate medium. For BET testing, < 0.2 EU/mL was reported for sterility test methods, at a 1:40 dilution. Bioburden testing reported < 1 Anaerobic and < 1 fungal colony forming units (CFU)/1 mL.

Based on the above, the preferred method of the sterilization of DMSO is sterile filtration. The chemical quality data on the DMSO sample that had been sterilized by this method were indistinguishable from the untreated (and analytically pure) material. Aseptic filtration is, in fact, a proven DMSO sterilization technique in use today; there are “DMSO-compatible” filter media, filter housings, etc., available from the manufacturers of filtration products.

For many simple applications (i.e., sterilization of DMSO for use as cryomedia), aseptic filtration is ideal. As mentioned above, the materials of construction of the filter, peristaltic pump tubing, O-rings, etc., must be scrutinized. Unless verified through proper leachables/extractables testing, the use of specific perfluorinated materials (PTFE, Teflon, Kalrez) and some polyolefins (high density polyethylene, high density polypropylene) are prescribed. When pure DMSO is being processed, some polymers used in medical applications should be scrutinized, notably polyvinylchloride.

This presents a challenge for emerging technologies in which the finished pharmaceutical product is a viscous solution. Of course, this includes the use of DMSO as a delivery system for implantable medical polymers. One option is to simply sterilize the DMSO component and drug/polymer components individually, for later combination at the point of use.

Heat sterilization of a DMSO-based pharmaceutical formulation would seem ideal as a means to address this need. Clearly, however, thermal stress can impact the chemical quality of the DMSO component in such DMSO/polymer/drug products. The mechanism of DMSO’s thermal degradation has been described in an industrial setting; the mechanism of decomposition should be relevant under conditions of thermal sterilization (7). Traces of thiochemical degradation byproducts (dimethyl sulfide, etc.) are responsible for elevated ultraviolet (UV) levels of the product, failing related substance test results, etc. This suggests that—while not practical under the conditions evaluated in this work—process parameters that allow the thermal sterilization may be identifiable for some pharmaceutical formulations.

Irradiative sterilization of DMSO-containing pharmaceuticals is not straightforward. Gamma irradiation of the DMSO samples in this study failed all of the stability-indicating test methods. 

 compares related substance data for untreated and gamma irradiated samples. The samples (clear, odorless, and water-white prior to irradiation) were tinged and odorous post-irradiation. Of particular interest are the elevated and failing values for the non-volatile residue test. This suggests that higher molecular degradation products may form during the gamma irradiation process. Based on the reported decomposition behavior of DMSO, the residue generated during gamma irradiation is likely paraform, but this has not been verified analytically (7).

Chemical quality results from the E beam-irradiated samples were somewhat more positive. Although the UV absorbance and gas chromatographic-based test results failed, product color was adequate. Importantly, the E beam-irradiated samples passed the non-volatile residue test, suggesting that no high molecular weight decomposition byproducts were formed during sterilization. There is perhaps the potential for method as a DMSO sterilization technique, but more work is needed to validate this supposition.

This work ultimately recommends aseptic filtration as the most suitable sterilization method for DMSO. Our evaluation found that the other relevant sterilization techniques should be expected to negatively impact the chemical stability of the DMSO component in a finished pharmaceutical dose. In the event that aseptic filtration is not suitable for other components in the finished formulation, the DMSO component may be simply sterilized before use in the final product and combined with the other ingredients. The inherent stability of this versatile excipient is significant; the manufacturer of Procipient has documented a retest date of four years under USP controlled room temperature conditions.

Although it is not a completely inert material, the sterilization of DMSO is widely practiced. In cases in which E beam, gamma irradiation, or thermal sterilization are desirable, a feasibility study is probably worthwhile. The conditions chosen for this study, while representative of common sterilization parameters, may not be required in all applications.

All excipients should be evaluated for their stability under the conditions of sterilization to include potential interactions with other inactive ingredient, the formulation’s package, and the drug substance itself. The authors hope this work will serve as useful a starting point for pharmaceutical formulators and medical device designers as they consider the impact that a sterilization technique may have on the DMSO component of their products.

Pharm. Tech. APIs, Excipients, and Manufacturing Supplement

, s30–s35 (2016).
2. A.S. McKim, R.T. Strub, 

. 32 (5) 74-85 (2008).
3. P. Amels, H. Elias, K.J. Wanniwuis, 

. 93, 2537-2544 (1997).
4. A. Rodriguez, et. al., “High-Pressure Sterilization to Terminally Sterilize Pharmaceutical Preparations and Medical Products,” US Patent No. 8,617.467 B2, Dec. 31, 2014.
5. WHO, WHO Expert Committee on Specifications for Pharmaceutical Preparations–WHO Technical Report Series, No. 902 (36th Report, 2002).
6. FDA, 

Guidance for Industry, Sterile Drug Products Produced by Aseptic Processing–Current Good Manufacturing Practice

Lucelys Caton is a technical service scientist, Ashley Jones is a quality control chemist, and Andrew Johnson is a quality control manager, all with Gaylord Chemical Company.

When referring to this article, please cite it as L Caton, A. Jones, A. Johnson, “Suitable Sterility Methods for Dimethyl Sulfoxide 

Articles

Dimethyl sulfoxide is increasingly used in high-risk parenteral and medical device applications that must be manufactured as sterile products in their finished form. A study evaluated the effects four sterilization techniques have on the product quality of this ingredient. 

Suitable Sterility Methods for Dimethyl Sulfoxide USP, PhEur

	Submitted: Dec. 11, 2018.
	Accepted: Jan. 23, 2019.

Karl Fischer (KF) analysis is a fast and effective way to determine the water content in many excipients and APIs. Dimethyl sulfoxide (DMSO) is a hygroscopic pharmaceutical ingredient. For many years, its water content has been tested using the volumetric KF method, even though the accuracy of results using this method can be affected by reaction between the KF reagent and the DMSO analyte. 

In 2018, pharmaceutical companies petitioned the 

to change the DSMO testing monograph, which prompted the authors to compare different methods and instruments for this application. 

Spiking experiments that assessed water recovery were used to evaluate the interference for several reagent and technique combinations. Finally, method linearity, precision, and accuracy were assessed for each reagent and technique. This work resulted in developing a new technique and analytical instrumentation that would produce water standards in DMSO that were suitable for method development.

In Karl Fischer analysis, sulfur trioxide can react with DMSO, invalidating test results. The authors evaluated different instruments and methods, described in this article, to minimize the impact on results.

)-grade dimethyl sulfoxide (DMSO) has become increasingly common in topical and parenteral drug delivery systems as a way to make APIs more soluble. A wide variety of applications are currently on the market with multiple routes of administration (1,2). Some of its other applications include acting as a reaction solvent for the synthesis of APIs and intermediates as well as use as an API itself.

	The US and European compendial methods both specify that volumetric Karl Fischer (KF) be used for water determination in DMSO (3,4). Unfortunately, there are many drawbacks to using volumetric KF with DMSO. For one thing, an intermediate product, sulfur trioxide, reacts with DMSO in the presence of iodide. This redox reaction reduces DMSO to dimethyl sulfide (DMS), generating iodine in the process. This iodine then titrates water in the sample, which results in low recovery. This particular interference has been observed by a number of device manufacturers and has been further discussed in technical papers (5,6). Some manufacturers have also observed the same interference when using coulometric KF (7).

	Other issues have also become apparent when analyzing anhydrous DMSO samples. Due to the hygroscopic nature of DMSO, exposure to environmental moisture results in an uptake of water. This interference can lead to skewed results when calculating the recovery factor that is necessary for volumetric KF. The burette for volumetric KF is designed to titrate samples with a water content greater than 1% w/w. Device manufacturers for volumetric KF suggest that 10–90% of the burette volume be used for optimal usage. For anhydrous DMSO samples, the amount of burette volume used can be as low as 0.3%, leading to accuracy and precision errors.

	Each analytical method was evaluated over a wide range of concentrations. Because of the extremely hygroscopic nature of DMSO, special techniques were developed to prepare and handle DMSO standards that contained less than 50 ppm water.

	Septum-lined sample bottles were used to allow for the addition of water spikes and subsequent withdrawal of material without having to remove the container cap. Activated molecular sieves (8) were then used to dry down the DMSO samples to help achieve a lower initial water content.Once dried, the DMSO was transferred to a thoroughly dried sample bottle. To prevent exposure to environmental moisture during transfer, an apparatus was set up to transfer the dry material. It consisted of a stainless-steel cannula and a nitrogen purge, as shown in 

. All water and stock spikes were prepared using a gas-tight syringe (9) in the same bottle to control the risk of environmental moisture associated with transferring the dry DMSO into multiple bottles. A 1% w/w water stock solution was also used to control the error inherent to weighing small volumes of water.
	

	To perform recovery calculations, an initial water assay of the DMSO sample was required. This calculation was accomplished using the method of standard addition (10).

	The volumetric KF method (11) was then assessed using data from validation exercises that had been performed in 2016 and 2018. Low recovery was exhibited in both validations but could be corrected with the use of a recovery factor. Tests were run using different commercially available Karl Fischer reagents (12) and results were compared. These experiments showed that the use of methanol-based reagents improved both accuracy and precision.

	In a pharmaceutical environment wherein accurate water measurements are needed as a means of process control, it is beneficial to be able to make multiple injections without having to change out the solution. An experiment was carried out in which numerous injections of DMSO were made into a titration vessel without emptying the titration cell and filling with fresh solution. When comparing the cumulative volume of DMSO in the titration vessel against water concentration values returned by the instrument, two trends became apparent: A downward trend in the results was seen when using higher concentration material (

) and an upward trend for lower concentration material (

	The method works relatively well for DMSO that contains higher concentrations of water.  However, using a recovery factor is not sufficient when using KF analysis on extremely dry product samples. The authors had considered using a Karl Fischer oven module to assess water in DMSO samples, while preventing DMSO from entering the titration cell. Instrument manufacturers suggested, instead, that using an oven would result in the same problems, so the decision was made to consider using a gas chromatograph (GC) equipped with a thermal conductivity detector (GC-TCD) and an ionic liquid-based GC column (13, 14, and 15) as an alternative to volumetric KF.

	A GC-based method offered an attractive alternative to a KF titrator for several reasons. First, if it worked, the GC-TCD analysis would avoid the complicated interferences observed between DMSO and components of the KF reagent. In addition, a GC is generally a more robust instrument than a KF titrator. It requires less maintenance and fewer consumables and can be used to analyze for more than one impurity at a time. GC analysis is also less technique dependent, making it ideal for in-process testing.

	The authors found that the GC-TCD method worked relatively well at higher water concentrations, as shown in 

. At concentrations lower than 300 ppm, however, the GC-TCD signal was not intense enough to generate the area counts that were required to reproduce data precisely, which led to linearity problems. However, percent recovery values were deemed suitable across the testing range, based on an assessment of triplicate injections. Using fewer injections resulted in low precision at concentrations less than 300 ppm, causing recovery to vary widely. However, the need for triplicate injections resulted in a significant runtime for each sample analysis.
	

	The authors then considered the coulometric KF (15) method for this application, especially after learning that the approach had been mentioned in a 2018 petition to the 

to change the DMSO testing monograph. Spiking experiments (16) were used to assess the method’s capabilities (17).

	Excellent recovery and precision were observed, even for concentrations as low as 35 ppm up to around 100 ppm, without the use of a recovery factor. They were also seen for concentrations of up to 1000 ppm with the use of a recovery factor ( shown in 

). Manufacturer reviews suggested that a side reaction would occur, but using the recovery factor prevented this problem from taking place. However, coulometric KF does not seem to have the same precision issues that volumetric KF has at lower concentrations.
	

	When numerous injections of DMSO were made into the titration vessel without emptying the titration cell and filling it with fresh solution, there were no significant differences in the results (

). This was observed at both low and high concentrations. It seems that the DMSO/KF reagent interference has no effect on the trend in results after numerous injections for coulometric KF. In fact, the only potential downside of using coulometric KF is that it is not designed to titrate greater than 1 % w/w water. For DMSO samples with a water content greater than 1% w/w, volumetric KF can instead be used with a recovery factor. When a recovery factor is used at a concentration > 1 % w/w, results show a high degree of accuracy and precision.

	Of the three methods that were evaluated for water determination in DMSO, the coulometric KF method proved to be the most effective at analyzing samples that contained less than 1% w/w water. Even though the DMSO/KF reagent interference is still present, improved recovery, precision, and a significantly lower limit of quantitation is observed when a recovery factor is applied. Its short run time, as well as the ability to support numerous injections of DMSO without significant differences in results, makes this method ideal for in-process testing.

	Volumetric KF and GC-TCD are both suitable methods for DMSO samples that contain a water concentration greater than 1% w/w. A recovery factor should be applied when using volumetric KF due to the DMSO/KF reagent interference. Making numerous DMSO injections into a volumetric KF titration vessel is not recommended due to a decrease in recovery at high concentrations and an increase at low concentrations. In addition, the need for triplicate injections when using the GC-TCD method causes an increase in run time and makes it less ideal for in-process testing.

2016 Supplement (3), 30-35 (2016).
	2. A.S. McKim and R.T. Strub, 

p. 3792 (US Pharmacopeial Convention, Rockville, MD, 2017).
	4. EDQM, 

Volume 2 (EDQM, Strasbourg, France, 2017), p. 2280.
	5. W. Fischer, S. Beil, D. Krenn, 

(1), 266-268 (1995).
	6. Metrohm, Application Work AW KF DE5-0377-072012: Dimethyl Sulfoxide as Solubilizer in KF Titration (2011).
	7. Hydranal, Laboratory Report L 141: Water Determination in Dimethyl Sulfoxide (2017).
	8. 3Å molecular sieves were used (Catalog # 208574-5KG). The DMSO stood over the molecular sieves for >72 hours.
	9. Gas-Tight Syringe: 2.5 mL Hamilton #1002 PTFE Luer Lock
	10. J. M. Miller and J. B. Crowther, 

Analytical Chemistry in a GMP Environment: A Practical Guide,

pp.390-391 (Wiley, New York, NY, 2000).
	11. E. Scholz, “Titration Techniques,” in 

Karl Fischer Titration: Determination of Water 

(Springer-Verlag, New York, NY, 1984), pp. 15-19.
	12. Comparison of KF reagents: Hydranal KetoSolver (Catalog # 34738) and Hydranal Composite 2 (Catalog # 34806); Aquastar Solvent KN (Catalog # AX-1699EN-6) and Aquastar Composite 2 (Catalog # AX1698B-5); Aquastar CombiMethanol (Catalog # M1880092545) and Aquastar CombiTitrant 2 (Catalog # M1880022545)
	13. Y. Zhang, et al., 

24 (10), 516-529 (2011).
	14. J. M. Miller and J. B. Crowther, 

Analytical Chemistry in a GMP Environment: A Practical Guide

(Wiley, New York, NY, 2000), pp.236-240.
	15. Watercol 1900 Capillary Column 30m x 0.25mm x 0.2µm film thickness (Catalog # C21615010)
	16. E. Scholz, “Coulometric Determination,” in 

Karl Fischer Titration: Determination of Water

(Springer-Verlag, New York, NY, 1984), pp. 19-21.
	17. Coulometric Titrator: Mettler Toledo C20X (Catalog # 51105512); Coulometric Reagent: Aquastar CombiCoulomat Fritless (Catalog # 1092570500)


	Volume 43, Number 5
	May 2019
	Pages: 44–48

	When referring to this article, please cite it as A. Johnson and A. Jones, "Testing for Water in DMSO: Exploring Alternatives to Volumetric Karl Fischer Analysis," 

 is analytical specialist, ajohnson@gaylord.com and 

 is analytical chemist, ajones@gaylord.com both with Gaylord Chemical Co, LLC.