Isolation of Pharmaceutical Impurities and Degradants Using Supercritical Fluid Chromatography

Isolation of trace impurities and degradants from mixtures containing primarily an API, or an API plus excipients, is often necessary for structure elucidation purposes. Using chromatographic methods for this purpose can be a slow and painstaking work. The authors demonstrate that using supercritical fluid chromatography (SFC) offers distinct advantages in speed and in clean isolation of the desired peaks. These advantages are derived from the rapidity of method development, the efficiency of the preparative fractionation, and the ease with which commonly used reversed-phase high-performance liquid chromatography data may be correlated with SFC. Case studies are used to illustrate these efficiencies.

Chromatographic isolation of degradants and impurities, whether from stressed lots of pharmaceutical compounds or directly from the API, is often required when their structures are unknown and/or reference standards cannot be prepared by synthesis (1). Isolation of these (often trace) components can be a painstaking process for analytical laboratories that involve, even in the simplest scenarios, many repeated injections of material on a chromatographic column to accumulate the trace component chromatographic peak. Often the peak of interest is profiled using a stability-indicating method designed for optimum resolution of all possible impurities and degradants (2, 3). Adapting this method to preparative scale consumes huge quantities of solvent, has a long cycle time that slows accumulation of the trace component, and requires that large amounts of accumulated solvent be removed to recover a few milligrams of the peak in quantity sufficient for structural analyses using nuclear magnetic resonance (NMR) and multiple and sequential mass spectrometry (MSⁿ ).

Using supercritical fluid chromatography (SFC) (4) in place of traditional reversed-phase and normal-phase high-performance liquid chromatography can greatly reduce the timelines for this process. Both the rapid method development cycle in SFC and the high efficiency of preparative SFC separations contribute to the reduced timelines (5). Solvent consumption is also reduced because SFC uses a mixed phase of solvent and recycled carbon dioxide (CO2). The accumulated fractions from SFC chromatography are highly concentrated compared with those collected with conventional liquid chromatography. In addition, the lability of compounds collected during the isolation process may be minimized in common SFC solvent systems and by the mild and fast evaporation conditions used for such highly concentrated fractions.

SFC methods, however, are not commonly developed as stability-indicating methods. SFC is a normal-phase chromatography, and most stability-indicating methods use reversed-phase high-performance liquid chromatography (RP-HPLC). SFC methods must be developed a priori for isolation work. Using SFC as an isolation technique requires some investment in equipment, often in columns; however, the benefits to the pharmaceutical development process can be significant for a laboratory accustomed to RP-HPLC approaches.

Methods and materials

All instrumentation used in this work was acquired from Waters Corp. The systems included Thar AMDS analytical SFC systems, Thar Prep80 preparative SFC systems, Waters Alliance 2795 HPLC systems, Waters 996 photodiode array detector, and Waters ZQ single quadrupole mass spectrometry (MS) detector. All solvents were HPLC grade. Buffers and solvent additives were supplied by Sigma-Aldrich. Column sources are indicated with the specific methods as further described.

Enrichment. Method A. The RP-HPLC method to identify and assign purity to the target impurity isolate used a Halo C18 4.6 × 150 mm 2.7-µ column (Mac-Mod Analytical). Solvent A was water (0.1% trifluoroacetic acid [TFA]). Solvent B was acetonitrile (0.1% TFA). The solvent gradient was 17% B over 12 min, 17–100% B over 3 min, 1-min hold, and 3-min recycle. The impurity at 9.0 min has an enhanced absorbance relative to the main peak when detected at 260 nm.

Method B. The intermediate SFC method to collect an enriched fraction used an (S,S) Whelk-O1, 30 × 250 mm, 5-µ column (Regis Technologies). The preparative method was isocratic, 10% methanol in CO2 at 80 g/min and 120 bar pressure; 300 mg of feed were injected every 2.1 min for a productivity rate of 8.5 g/h.

Method C. For final purification of the target from the enriched sample, a new SFC method was developed using a RegisPack 4.6 × 100 mm 5-µ column (Regis Technologies). The isocratic method used 13% methanol:isopropanol (50:50) in CO2; 240 mg of enriched fraction were processed in eight injections over 30 min.

Direct Isolation. Method D. The client's HPLC method was used to assess the stress degradation mixture, using a Waters XBridge Phenyl 4.6 × 150 mm 5-µ column. Solvent A was 10 mM ammonium formate in water adjusted to pH 4.0. Solvent B was 10 mM ammonium formate in methanol:water (90:10) adjusted to pH 4.0. The gradient elution at 1 mL/min used a solvent gradient 10–40% B over 30 min, 40–100% B over 5 min, and 4-min hold.

Isolation strategies

Two isolation strategies are commonly used in the authors' laboratory—enrichment and direct isolation. In the enrichment strategy, an SFC method is developed that appears to resolve the main peak (usually the active ingredient) from neighboring peaks, and the main peak is captured and removed to produce an enriched fraction containing all other compounds, including the desired degradants. This approach takes advantage of the speed with which SFC can process a feedstock solution of the sample and can be used even when the trace target peak(s) is/are not visible under preparative chromatography conditions due to limited detection range at high loads. Using the direct-isolation strategy, a specific SFC method is developed that resolves the desired trace peak, and that peak is collected as a purified isolate. This approach requires that the desired peak be visible with a specific detection signature (e.g., a specific UV absorbance wavelength or a mass not shared by other compounds in the mixture). Enrichment and direct isolation are frequently used together, and complex projects involving multiple isolates may require their sequential application. For example, once an enriched fraction containing the desired peak is obtained, the desired peak is often an abundant component in a fraction that becomes a small feedstock fraction for direct isolation.

In the authors' laboratories, a process termed "targeted isolation" is used to rapidly develop methods for purification of single peaks from complex mixtures. Targeted isolation involves linking specific information, such as the UV absorbance spectrum or mass spectral data acquired in RP-HPLC analysis, to normal phase SFC chromatograms. This approach is used to cross-correlate peaks as analysts rapidly develop SFC methods—initially, gradient elution and finally, isocratic methods, to be scaled to preparative chromatography. While this article does not discuss all the aspects of targeted isolation, the parts of the process used to "close in" rapidly on the peak of interest are discussed. Case studies that illustrate these strategies and discuss the merits and advantages of SFC technology in their application are presented.

Case study 1: enrichment

A client API contained two process impurities—one identified and one unknown. Both were visible by UV detection at 260 nm, a wavelength at which the API absorbs light less strongly. Thus, the relative absorbance chromatogram observed in the straightforward gradient RP-HPLC method (Method A, see Figure 1) was known to overrepresent the true abundance of both impurities. The unknown impurity had a relative absorbance of 0.5% and a true abundance estimated at less than 0.2%. Preparative isolation by scaling up Method A was rejected due to time, solvent consumption, and the excessive fraction volumes that would be accumulated in the isolation of this low-abundance peak. It was hoped that, by using SFC methods, the target impurity could be recovered more expediently in the milligram quantities desired for a complete or partial structure elucidation by 2D NMR and MSⁿ methods.

Figure 1: Reversed-phase high-performance liquid chromatography (UV 260 nm, Method A) of the API showing known and unknown impurities. The unknown impurity (0.5% by relative absorbance) is observed at 9.7 min. [QA: please define AU]

Under preparative SFC conditions with detection at 260 nm, the impurity peak signal was difficult to identify unambiguously. A preparative SFC method (Method B) was developed quickly and used to process several grams of API, with fractions collected before, during, and after the main peak eluted. These fractions were concentrated by rotary evaporation and analyzed using the RP-HPLC method for the presence of the desired peak. The desired peak was captured and enriched in a fraction collected immediately before the main peak, indicating that the chosen SFC method adequately resolved the peak from its neighbor. To accumulate the target peak in quantity sufficient for structural work, 50 g of the API were injected during a period of 7 h of automated stacked injection chromatography (see Figure 2) to produce a highly enriched fraction with a total mass of 240 mg (see Figure 3).

Figure 2: Preparative supercritical fluid chromatography (UV 260, Method B) chromatogram, showing the fractions collected to enrich the target impurity. The target elutes in fraction F1.

Method C, which involves direct isolation of the target peak from the enriched fraction, was developed by screening multiple SFC conditions in an automated method-development process. Typically, several hundred solvent and stationary phase pairings can be screened in an overnight series of gradient runs, and a promising separation may be converted within a few minutes to a preparative-scale isocratic process. In this case, the target peak was identified based on its relative abundance in the HPLC chromatogram in Figure 3a, and a method resolving the target completely from its near neighbors was chosen (see Figure 3b). The 240-mg enriched isolate was dissolved in the liquid cosolvent and processed in less than 1 h to isolate the desired compound. Evaporation of approximately 50 mL of collected solvent yielded 14 mg of the pure target peak.

Figure 3: a (left): Reversed-phase high-performance liquid chromatography (Method A) chromatogram of enriched fraction. b (right): Analytical supercritical fluid chromatography (Method C) chromatogram of the enriched target, showing remaining active and other impurities. The mass spectrum for target at m/z 385 is identical for the indicated peaks in each chromatogram.

The development of a controlled and well-characterized production process for a new drug is a complex task. The client producing this API estimated that the identification of this trace impurity, efforts to synthesize it based on various hypotheses about its source, attempts to remove it by chemical methods or through alternate synthetic approaches, and attempts to recover it through chromatography would take 12–14 months. Isolation of the compound using SFC required less than a week, and its structure was determined by NMR and MS.

Case study 2: direct isolation

A stress degradation product was partially purified by low-pressure chromatography to remove much of the remaining main drug peak. The degradant-enriched mixture had two significant peaks of interest (primary targets) and several of minor abundance (secondary targets), in addition to components not of interest to the chemistry team. The HPLC chromatogram of the mixture is shown in Figure 4. The goal for the isolation was to capture cleanly each of the primary targets for structure elucidation and to capture the secondary target peaks if it were not "overly difficult." Preparative RP-HPLC was undesirable due to poor loading and excessive processing time.

Figure 4: Reversed-phase high-performance liquid chromatography (UV detection 295 nm, Method D) of degradant-enriched sample, showing primary (light red) and secondary (light green) purification target peaks.

The development of an SFC method specific for a targeted minor component is often complicated by the lack of availability of an SFC method identifying the peak of interest. Frequently, stability-indicating methods identifying minor API components are RP-HPLC gradient-elution methods. Thus, the initial phase of method development is aimed at identifying the peak of interest in an SFC chromatogram, followed by an optimization phase to resolve that peak from its neighbors. The final isocratic method is scaled to a preparative column so that the peak may be accumulated by injection stacking.

Figure 5: Method development and fractionation tree for the recovery of the targets identified in Figure 4.

To facilitate the sometimes difficult process of correlating the peak in the RP-HPLC method with SFC, the authors previously developed a co-configured HPLC/MS and SFC/MS system (6) that uses the same mass spectrometer as a detector for either chromatography. Using this system, masses can be assigned to mixture components in both the RP-HPLC trace and the SFC chromatogram. This cross-correlation of orthogonal reversed-phase and normal-phase peak profiles, and the subsequent development of an appropriate preparative chromatography method to cleanly capture a target peak is part of the targeted isolation process. The standard HPLC method is still used to confirm the successful isolation of each component (see Figure 3).

This approach to separation is straightforward, but appears cumbersome at first glance. Due to the rapidity with which isocratic SFC methods may be developed, there is little need to spend time and effort to develop an optimal method first. A complex sample may be sectioned into several fractions (in this case four), which are then examined using the standard HPLC method for the presence of one or more of the targets. Once a fraction contains only a few components, the target peaks often can be selected based on relative abundance, mass (as measured by the MS detector), or by a specific UV signature. Highly specific methods are then developed to cleanly resolve the targets. At the end of a multicomponent isolation project, a method and fraction tree illustrates the work flow of the complete separation (see Figure 5). Individual isolates highlighted in the figures were assayed using Method D, and deemed sufficiently pure for structure elucidation.

Table I: Supercritical fluid chromatography methods used in the target peak isolation tree, numbered according to Figure 5.

The SFC methods needed to isolate the targets are detailed in Table I. The seeming complexity of the process is belied by the speed with which the work was accomplished—in this case, less than three weeks of effort during a period when concurrent project work also demanded laboratory resources. The rapidity derives from the efficiency of method development, processing, and recovery from liquid fractions when using SFC; all of these facilitate the rapid cycle time from peak to method to isolation.

Discussion

The two strategies take advantage of different key attributes of SFC. The enrichment strategy leverages the speed of processing in SFC versus HPLC, achieved through increased loading, narrow bands, and high linear solvent velocities. SFC has often been described as 3–10 times faster than HPLC (5).

The direct isolation strategy leverages the processing speed advantage as well, but in comparison with HPLC, the speed of method development offers a significant additional process enhancement. To achieve the same resolution in HPLC, one frequently uses a longer column with a higher plate count. In SFC, the linear mobile phase velocity is 8–20 times that of an equivalent HPLC method, as short columns and high flow rates are both commonly used in method development and these methods are arithmetically scaled to larger column formats (7). Moreover, this translates into extremely fast column equilibration times, meaning that as many as 20 different SFC methods may be evaluated for every one HPLC scouting run. In addition, it was observed that the highly concentrated fraction volumes in SFC offer a significant advantage, i.e., the ability to quickly remove solvent and analyze the isolate significantly speeds the cycle through processing and method development.

Finally, it was noted that the lack of oxygen and water in the SFC solvent environment limits many common degradative pathways that would otherwise prevent the capture of certain primary degradants. Less nucleophilic alcohols, such as isopropanol and the butanols, usually offer good selectivity and have been used when methanol reactivity is a concern. SFC may offer a greater chance of success when the goal is to isolate primary degradants intact without secondary decomposition.

Conclusions

SFC is, in many ways, an ideal tool for facilitating trace component capture and analysis. Scalable separations are developed rapidly, large amounts of feed may be rapidly processed, dual strategies may be used exclusively or combined, and milligram quantities of highly pure isolates can be accumulated for structural or biological assays. In addition, the solvent environment is much more benign than conventional chromatography approaches.

SFC can be superior to conventional chromatography approaches in many cases, and should be considered by laboratories. Given the complex nature and long timelines of these projects, and the scale of chromatography that must be undertaken to access trace components, it is worth the effort to identify the best approach to such challenging separations problems.

References

1. Handbook of Isolation and Characterization of Impurities in Pharmaceuticals, S. Ahuja and K. Alsante Eds., Separation Science and Technology Series Vol. 5 (Academic Press, San Diego, CA 2003).

2. M. Bakshi and S. Singh, J. Pharm. Biomed. Anal. 28 (3) 1011–1040 (2002).

3. V. Kumar, H. Bhutani, and S. Singh, J. Pharm. Biomed. Anal. 43 (2) 769–73 (2007).

4. G. Guiochon and A. Tarafder, J. Chromatogr. A 1218 (8) 1037–1114 (2011).

5. K. Anton and C. Berger, Supercritical Fluid Chromatography with Packed Columns (Marcel Dekker, New York, 1998).

6. J.P. Kiplinger and P. M. Lefebvre, "Improving Productivity in Preparative Supercritical Fluid Chromatography Separations," presentation at the International Conference on Supercritical Fluid Chromatography (New York, 2011).

7. Scale-Up and Optimization in Preparative Chromatography: Principles and Biopharmaceutical Applications, A.S. Rathore and A. Velayudhan Eds., Chromatographic Science Series Vol. 88 (Marcel Dekker, New York, 2003).

Jeffrey P. Kiplinger* is president, Paul M. Lefebvre is director of laboratory operations, Michael J. Rego is a staff scientist, and John H. Tipping is a staff scientist, all at Averica Discovery Services Inc., One Innovation Drive, Three Biotech, Worcester MA 01605 USA. jeff.kiplinger@avericadiscovery.com.

* To whom all correspondance should be addressed.

Submitted: June 28, 2012. Accepted: Aug. 28, 2012.