Evolution of Solid-State Nmr in Pharmaceutical Analysis

Abstract

The technique of solid-state NMR, as applied to pharmaceutical analysis, has evolved from a sparingly used technique into an integrated, information-rich technique that uses a variety of sophisticated 1D and 2D experiments. In this article, several key developments in the field are reviewed, highlighting the increasing sophistication of the analyses being utilized and the detailed results obtained, in addition to future directions and developments.

Figure 1. Timeline of the evolution in solid-state NMR of pharmaceutical analysis over the past 15 years, focusing on two key drivers: the commercial availability and technical establishment of new technologies and the applications that became demonstrable in practice with these technologies.

The solid-state structure of a pharmaceutical product can influence its properties, manufacturability and stability [1]. Since these characteristics can impact product performance and patient safety, regulatory agencies require control over solid-state properties in many situations. In the quality-by-design environment that is emphasized in modern pharmaceutical development, increased understanding of products gained by solid-state analysis is very useful when developing an active pharmaceutical ingredient (API) or a formulation process for solid-dose products, including oral, inhaled and injected products [1,2]. Solid-state NMR (SSNMR) spectroscopy of pharmaceutical materials has evolved from a sparingly used technique into an important component of pharmaceutical solid-state analysis activities over the past decade, largely owing to its unique capabilities. Furthermore, continued innovation and application are occurring in the field that are likely to continue this trend. In this work, a summary of the path to the current status of applied pharmaceutical SSNMR is provided. The latest developments in the field are summarized, focusing on the utility of the technique during the development of a drug from the candidate stage through to a marketed, approved product. The dominant use of SSNMR in the pharmaceutical industry occurs during development, particularly during process development for API and drug products for human clinical trials. However, several examples are also given of the utility of SSNMR during the discovery stage, prior to candidate selection. Finally, future directions and developments in the field of pharmaceutical SSNMR are considered.

Solid-state NMR analysis of the ¹³C nucleus at natural abundance, using the crosspolarization with magic-angle spinning (CP-MAS) method, has been traditionally used to provide ‘fingerprint’ characterization of solid crystalline drugs and is widely used for this purpose today [3–5]. In this role, SSNMR serves primarily as an orthogonal technique to the more routine and widely available techniques of vibrational spectroscopy (e.g., mid-infrared and Raman spectroscopy) and x-ray powder diffraction (XRPD). This is a very useful role, particularly since ¹³C SSNMR often offers the highest resolution of any technique between solid forms of interest, such as polymorphs, hydrates and solvates. However, the limited use of SSNMR as an orthogonal technique has restricted its overall acceptance, due to cost–benefit concerns, as an SSNMR instrument currently has a high installation and operational cost and lower throughput. Typically, a mid-field 400-MHz SSNMR system costs five- to ten-times more than an infrared or Raman spectrometer and two- to three-times more than an XRPD instrument. It is the additional capabilities of SSNMR that ultimately help justify its cost, including its ability to analyze complex solid-phase mixtures such as drug products and the detailed information content accessible from a wide range of nuclei using a variety of specialized 1D and 2D experiments. In these roles, it is superior to all other solid-state analysis techniques currently available that operate on powdered samples. Most of these advanced capabilities can be implemented on lower-cost 400- and 500- MHz systems, although higher fields are required in some cases, as discussed below. The flexibility of SSNMR allows a range of experiments to be sequentially performed on the same sample to extract out a variety of information beyond the conventional 1D spectrum. It also offers high spectral information content and benefits from extensive interpretability of the data using the vast, well-established empirical knowledge of NMR chemical shifts from solution-state NMR, as well as state-of-the-art theoretical calculations of NMR chemical shielding and quadrupolar interactions [6–9]. This latter feature allows most SSNMR spectra to be ‘assigned’ so that each resonance is linked to an individual atom in the chemical structure of the drug molecule, as with solution-state NMR spectra of small- to mid-size molecules. Even in cases where resonances are overlapped in both large and small molecules, useful group assignments of SSNMR spectra can be made.

The evolution of pharmaceutical SSNMR over the past 15 years, from its use as a fingerprint technique to its current status, is sketched in a timeline shown in Figure 1. Figure 1 necessarily focuses on several key technologies that led to significantly increased pharmaceutical applications and does not attempt to include every advance made in the field over this time period. Over the time span plotted in Figure 1, there has also been a concurrent trend towards higher static magnetic fields. Generally, ¹H frequencies of instruments used for pharmaceutical SSNMR were in the 300–400 MHz range in the mid-1990s, but as of 2010 are more typically in the range of 500 to 700 MHz (and even beyond).

Developments between 1995 & 2005

Between 1995 and 2005, the traditional ‘fingerprint’ role was expanded with applications to a variety of different drug systems that displayed polymorphism, hydration and solvation as well as different salt formation behavior [10–18]. A feature of much of this work was spectral interpretation via careful interpretation of subtle spectral effects, such as hydrogen bonding and the use of more advanced 1D SSNMR experiments for spectral editing to enhance and confirm spectral assignments. This interpretation ultimately allows for useful structural conclusions. Several approaches were frequently employed in these studies, ranging from spectral editing to highlighting quaternary and methyl carbons to the investigation of solid-state molecular dynamics via ¹³C spectral lineshapes and ¹³C and ¹H spin-lattice (T₁) and other relaxation times. The techniques used in these studies were well established by the early 1990s. The assessment of phase purity, an important routine use of SSNMR, was also demonstrated in many of these examples. SSNMR compares favorably to the indexing of XRPD data for phase purity assessment, as phase impurities are usually immediately detected in an assigned ¹³C spectrum. Beyond ¹³C SSNMR, ¹⁵N CP-MAS analyses also appeared in many of these basic applications, owing to its sensitivity to hydrogen bonding both as a donor and an acceptor and to protonation effects in salts and zwitterions [11,15,16,18].

Quantitative analysis methods were also demonstrated during this timeframe, illustrating the general applicability of SSNMR in these scenarios [19,20]. SSNMR is particularly useful early in development, when there is little time or material for a standardized, calibrated quantitative method and a standardless quantitative output is desired [19]. When development progresses and calibration standards can be prepared, XRPD and vibrational spectroscopy are more commonly used for quantitative analysis once a firm understanding of the solid form situation is gained, except in unusual cases (e.g., in drug products containing small amounts of drug, where ¹⁹F SSNMR might provide the only quantitative approach) [19].

The development of triply tuned ‘HFX’ probes capable of simultaneously irradiation of ¹H, ¹⁹F and an X nucleus (e.g., ¹³C) under high-power conditions also occurred between 1998 and 2005. While the development of HFX probes was certainly of interest to the wider SSNMR community, this technology had a disproportionately high impact on pharmaceuticals, given the frequent occurrence of fluorine in drugs and the emphasis given to fluorine in modern medicinal chemistry [21,22]. Reliable, stable designs of HFX probes, first reported in 1998 [23], achieved widespread commercial availability, with spinning rates up to 35 kHz, and had a substantial installed base by 2005. Pharmaceutical applications then became commonplace for both drug substance and drug product, the latter benefiting from the high sensitivity of ¹⁹F SSNMR [19,24–26]. Although not apparent from the number of published reports, ¹⁹F SSNMR is heavily used during pharmaceutical development by many organizations.

Solid-state NMR has proven to be particularly successful in the characterization of amorphous forms, including samples that contain both amorphous and crystalline content produced by a multitude of methods, including desolvation and milling [3,27,28]. This is to be expected, given the limitations of the other high-resolution technique (XRPD) in this area and the limitations of lower-resolution vibrational spectroscopic approaches. Various studies of amorphous forms have revealed interesting structural features of the amorphous state using 1D ¹³C SSNMR experiments [3].

During the period of 1995–2005, the use of SSNMR to analyze drug products also increased markedly [19,24–26,28–32]. The analysis of complex phase mixtures, such as many common solid oral-dosage forms, is an area where both the spectral resolution and experimental flexibility of SSNMR allow it to make a major contribution. This may seem surprising in light of the often quoted insensitivity of NMR. However, modern SSNMR systems have more than sufficient sensitivity to detect the drug in typical pharmaceutical formulations at 5–10% w/w levels using ¹³C detection [26,29]. Both solid and semi-solid formulations can be readily analyzed [19]. Developments in instrumentation, probes and higher static field strengths have contributed to the increasing use of SSNMR in drug product analysis. In particular, the advent of digital signal generation and detection and reliable high-power solid-state amplifiers has led to greater sensitivity, sharper lineshapes from high-power decoupling and higher quality spectral baselines for quantitative analysis [33].

Developments since 2005

Since 2005, increased use of 2D SSNMR and other advanced methods has been observed in a number of published articles. The ability of 2D SSNMR methods to observe interactions between nuclei via the greater resolution and connectivity information in a 2D spectrum has allowed for more detailed structural analysis of pharmaceutical materials. Early work in this field was pioneered by Reutzel-Edens and co-workers using detailed interpretation of 1D SSNMR spectra in conjunction with crystallographic data [31]. The advent of homonuclear ¹H decoupling using the frequency-switched Lee–Goldburg method led to dipolar-based heteronuclear correlation (HETCOR) methods, which remove and reintroduce dipolar interactions to allow for detection of through-space ¹H and ¹³C internuclear interactions within a few angstroms [34]. An early example presented for a pharmaceutical hydrate illustrated the ability of this technique to selectively detect most hydrogen sites in the crystal structure [35]. Use of ¹H SSNMR has led to greater access to structural information using dipolar correlation, including applications of sensitive inverse-detected experiments [36,37]. The increasing usage of advanced experiments in SSNMR studies is apparent in recent reviews of the field [19,28]. This work has culminated in the ability to determine crystal structures using SSNMR [38–43]. Methods to date have been based on spin-diffusion buildup rates that yield ¹H-¹H distances, chemical shift tensor measurements that are linked to structure via density functional theory simulations of chemical shielding, and other experiments that make use of through-space dipolar interactions to detect internuclear proximity. SSNMR approaches to crystal structure solution generally make use of relatively sophisticated software; a critical element of many studies has been the computational prediction of chemical shielding values from crystal structures for comparison and refinement against experimental chemical shift tensor values [9,40–42]. Methods using molecular clusters replicate short-range effects in the crystal lattice, while other approaches, such as the gauge including projector augmented waves (GIPAW) method and the embedded-ion method (EIM), replicate longer range effects in crystals [9,40–42]. Most SSNMR approaches are limited to molecules that are considerably smaller than typical drugs; a solution to this is to combine XRPD and SSNMR-based solution methods; an example crystal structure of a real pharmaceutical molecule solved using this approach with commercial software has been demonstrated [41]. It should be noted that the determination of crystal structures using SSNMR and XRPD methods is a relatively narrow area of demand in practice, as many systems are amenable to the ‘gold standard’ of single-crystal x-ray diffraction (SCXRD), particularly with continual advances at the frontiers of this field, which have driven down crystal size and quality requirements [43].

A broader area of potential focus for the more advanced 2D SSNMR approaches is the characterization of more complex systems, either to avoid the need for SCXRD entirely (by offering a faster means of answering a structural question) or to obtain structural data otherwise not available because of the complexity of the system. For example, the analysis of pharmaceutical cocrystals by SSNMR methods is greatly enhanced using a combination of techniques, including ¹H fast MAS, ¹H-¹H dipolar correlation and ¹H-¹³H and ¹H-¹⁹F dipolar CP-HETCOR experiments [44]. Cocrystals are currently an exciting area of research in both industry and academia as part of efforts to improve the solubility and physical properties of drugs [45]. Similar applications have been demonstrated for complex disordered structures, including amorphous dispersions [Pham TN, Watson SA, Edwards AJ, Chavda M, Clawson JS, Strohmeier M, Vogt FG. Solid-state NMR analysis of amorphous solid dispersions. (Manuscript Submitted)]. Some of these techniques rely on spin diffusion, a well-known phenomena owing to its appearance in diverse fields from solution-state protein NMR to SSNMR of semi-crystalline polymers. In the context of pharmaceuticals, the detection of spin diffusion over a 5–10-Å range, easily accomplished with modern 2D experiments, allows for SSNMR to directly confirm the formation of a glass solution in a single sample without the need for difficult-to-prepare amorphous reference materials, answering one of the most important challenges in the characterization of amorphous dispersions [Pham TN, Watson SA, Edwards AJ, Chavda M, Clawson JS, Strohmeier M, Vogt FG. Solid-state NMR analysis of amorphous solid dispersions. (Manuscript Submitted)]. The advanced 2D SSNMR approaches also enable detailed structural analyses in concert with SCXRD for challenging systems such as composite crystals containing both salt and cocrystal environments [46], disordered crystalline solid solutions [47,48] and molecular conformation in amorphous dispersions [28,30].

The discussion of SSNMR capabilities to this point has been limited to nuclei with a spin of ½, such as ¹H, ¹³C, ¹⁵N, ¹⁹F and ³¹P. Although these nuclei continue to sit at the core of most pharmaceutical SSNMR analyses, increased attention has been focused on studies of quadrupolar nuclei in pharmaceutical solids. Activity in this growing field has been driven by the availability of higher static magnetic fields, reducing second-order quadrupolar broadening effects, as well as improved hardware and pulse sequences. As higher fields have become available, previously inaccessible nuclei have been targeted; for example, detailed ³⁵Cl SSNMR studies of hydrochloride pharmaceutical salts have recently been reported [49]. Quadrupolar nuclei that were previously thought to be inaccessible due to large quadrupolar broadening and low sensitivity can also be addressed. For example, a recent ⁶⁵Cu SSNMR study of blue copper protein azurin showed the ability of this technique to probe the electronic state of copper sites in proteins [50]. ²³Na represents a more common, well-established quadrupolar nucleus of great interest in both pharmaceuticals and excipients. The ability to perform ¹H-²³Na HETCOR experiments offers insight into the hydrogen environment around this nucleus [35]. Owing to their presence in water, quadrupolar ²H and ¹⁷O SSNMR studies have found use in the study of complex hydrate systems, both to probe mobility and to understand structure [32,35,48,51]. The use of SSNMR to track moisture benefits from its unique ability to sensitively detect water isotopes and observe the physical state of water (e.g., surface vs bound), in addition to studies of crystalline hydrates. This allows studies of moisture ingress across semi-permeable membranes, such as dry micronized powder sealed in a device for inhalation or for studies of moisture uptake into an amorphous dispersion. ²H MAS experiments can also determine where water has come into contact with exchangeable groups via chemical exchange, even after the moisture is removed [32,35,48]. Once a physical understanding of the water environment is gained using SSNMR, more efficient infrared, Raman and MS methods may be to enhance the speed of analysis for moisture tracking.

In the last 5 years, the active field of SSNMR of biomolecules has demonstrated a number of potential applications to drug-discovery activities [52–56]. As noted previously, any industrial applications in this field would represent a shift into drug-discovery applications from the traditional drug-development role served by SSNMR in the pharmaceutical industry. This work also highlights the potential applications of increasingly high-end SSNMR instrumentation, normally procured for pharmaceutical development, to pharmaceutical drug discovery. Examples include the structural analysis of amyloid proteins [52], membrane proteins [53] and drug–receptor interactions between small molecules and proteins [54,55]. Unlike the other applications of SSNMR described, these applications generally require extensive isotopic labeling of the molecules involved. Finally, recent experimental developments utilizing J-coupling and multidimensional SSNMR have opened up new avenues for even more detailed structural analysis of large biomolecules [56].

Future perspective

Looking beyond 2010, several SSNMR applications for pharmaceutical development are expected to continue evolving. First, new experimental techniques are likely to open more territory for structural and dynamic studies. For example, recent pulse sequence developments have created the opportunity to work with the quadrupolar ¹⁴N nucleus [57,58], which is much more abundant than the spin-half ¹⁵N isotope. The use of ¹H SSNMR to probe formulations has been demonstrated and is expected to continue, providing sensitivity for low drug loads comparable to ¹⁹F SSNMR [37] and allowing studies of the structure of nanoparticle delivery systems [59]. Furthermore, as the advanced techniques described above become more embedded, they will likely find new applications to more complex systems. One example is the characterization of biopharmaceutical solids produced by lyophilization. Although these products are normally for intravenous use, it is still desirable to understand both mobility and structure of the solid material [60], which can include a number of excipients in addition to a biomolecule (typically a protein, but potentially a peptide, oligonucleotide or antibody fragment). It is also important to understand the interactions between excipients and the biomolecule, as well as the solid-state form of the excipients, to ensure there is no potential impact on stability or reconstitution [60]. Finally, increased applications of SSNMR in drug discovery may become more commonplace as methods are refined and more difficult targets are explored. The continued evolution of pharmaceutical SSNMR is thus likely to follow the trend of the past 15 years, with technology and applications advancing in concert to yield ever greater detail about the solid-state structures of these systems.

Financial & competing interests disclosure

The author is an employee of GlaxoSmithKline plc. and owns stock in this corporation. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.

Additional information

Funding

The author is an employee of GlaxoSmithKline plc. and owns stock in this corporation. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.