Mass Spectrometry in Drug Discovery and Development. Fall 10
More recently it has become one of the foundation technologies in high throughput omic analyses. Approximately 27 companies manufacture and supply a broad range of Mass Spectrometry equipment. Annual sales, service and ancillary efforts involving such instrumentation constitute a $3.3 billion global market.
The use of Mass Spectrometry in the pharmaceutical sector associated with the Drug Discovery and Development process is rich and varied. Many of the initial efforts were associated with online high performance liquid chromatography-mass spectrometry in drug metabolism, pharmacokinetic and pharmacodynamic studies. There have been numerous innovative efforts to apply various mass spectrometric techniques in early drug discovery, preclinical and clinical development, as well as in Phase 0 studies using Accelerator Mass Spectrometry. Today there is a re-evaluation and refocusing on how to efficiently adopt, adapt and use modern Mass Spectrometry instrumentation in the Drug Discovery and Development process.
Productivity in the pharmaceutical sector continues to undergo withering scrutiny1-8. In a recent supplement of Drug Discovery World (DDW Insights, Summer 2010), a panoply of pharmaceutical, scientific, technology and business experts opined on the necessity of improving drug safety and efficacy as well as significantly reducing costs associated with the drug discovery and development (DDD) process9. A number of these authors suggested specifically that an increasing role of technologies would enhance productivity and directly address critical DDD issues. However, Naylor and others have argued that the haphazard adoption and untrammeled use of technology has had limited impact on productivity and costs3,10-12, and “…has been overhyped as a possible cure for the industry’s productivity woes”3. This is due primarily to the poor understanding of new technology introduction into the DDD process. It is important to recognise that a complex array of factors must be considered in any technology adoption and includes the Technology Development Cycle (TDC), Technology Hype Cycle (THC), and the Technology Assimilation and Innovation Adoption Curves (TAC and TIAC respectively). These processes have been described in detail elsewhere, and inadequate consideration of their consequences leads to poor integration and misuse of new technologies in the DDD process3.
Cursory consideration of Mass Spectrometry (MS) suggests it should not be subject to the same vagaries as other ‘new technologies’ when utilised in the DDD process. MS is a mature analytical technology predicated on the ability to separate charged analytes (ions) based on their mass to charge (m/z) ratio, and was first demonstrated by J.J. Thompson in 191313. The first commercially viable MS instrument was developed by A.J. Dempster in 191813. In addition five Nobel Prizes have been awarded to J.J. Thompson (1906), F. Aston (1922), E.O. Lawrence (1939), jointly to H.G. Dehmelt and W. Paul (1989), and jointly to J.B. Fenn and K. Tanaka (2002) in areas pertaining to the development of MS over the past century13. Today it is widely used in basic research across a diverse array of disciplines as well as industries and specialty areas including food, environmental, forensic, toxicological and geochemical sciences as well as all aspects of the health and life sciences14. In particular MS is now the foundation technology of high throughput omic analyses for proteomics, metabolomics and to a lesser extent some aspects of genomics such as SNP determination14.
There are numerous examples of MS facilitating the DDD process and historically they include pharmacokinetic and pharmacodynamic analyses as well as Phase I and II drug metabolism studies15- 18. However the overall impact of MS in the pharmaceutical sector over the past 15 years has sometimes been uneven and unpredictable. An immediate question that arises is how can a mature, 100 year-old technology not be utilised in an efficient manner? In this paper we will attempt to address the issue as well as describe in some detail the role of MS in the DDD process.
Mass Spectrometry has emerged as a powerful analytical tool applied to the health life sciences and the pharmaceutical sector14,15-18. The essence of a modern MS platform consists of a sample introduction port and/or online separation device, a source region coupled to the mass spectrometer and a detection system, all under workstation control complete with software packages to assist in data acquisition and interpretation as shown in Figure 1A. More extensive reviews and details are available elsewhere14,19 but for convenience, a glossary of common terms used in MS and discussed in this article are provided in the Glossary Box.
Online separation chromatography-MS
The use of direct sample introduction or online chromatographic-MS analysis is determined by the sample complexity. The analysis of complex biological mixtures encountered in the DDD process is greatly aided by additional stages of separation prior to analysis by MS (Figure 1A). There are a myriad of such online chromatography technologies but three are commonly used in the pharmaceutical sector and include Gas Chromatography (GC), Reversed Phase High Performance Liquid Chromatography (HPLC) or Capillary Electrophoresis (CE). Currently, a powerful new gas phase separation technology, Ion Mobility Spectrometry (IMS), is being used in the separation of complex biological-derived peptide mixtures. The development of IMS-MS has been pioneered by David Clemmer at Indiana University and Waters Corporation recently introduced the SYNAPT™ commercial MS instrument.
GC-MS is used in conjunction with either Electron Impact (EI) or Chemical Ionisation (CI) as the primary ionisation source. Complex mixtures of organic molecules are separated in a capillary tube coated with a suitable hydrophobic material using a carrier gas as the mobile phase to transport the individual molecules through the capillary and into the mass spectrometer.
The dynamic and individual interactions of molecules with the capillary surface induce separation as they traverse the capillary. The separated molecules are eluted directly into the source region where ionisation occurs followed by subsequent separation in the actual mass spectrometer. HPLC: Reversed-phase high performance liquid chromatography-MS is routinely used in the pharmaceutical sector since it is well suited to separation of biological mixtures and can be readily interfaced to an Electrospray Ionisation (ESI) ion source. HPLC separates analytes based upon their hydrophobicity. The complex biological mixture is initially adsorbed, from an aqueous solution, in a narrow band at the beginning of a column packed with a stationary support. A liquid mobile phase is then pumped through the column, while the hydrophobic nature is increased by adding an organic solvent (typically acetonitrile). This causes peptides to differentially migrate through the column as a function of their hydrophobicity. When the HPLC is connected to MS, the MS continually analyses the HPLC effluent and detects the peptides as they elute from the HPLC column. Great strides in HPLC-MS have occurred over the past decade through miniaturisation of both the HPLC technique and ESI interfaces, since for every 50% reduction in column diameter, a four-fold increase in sensitivity occurs.
Capillary Electrophoresis: CE-MS actually consists of a family of CE techniques that include capillary zone electrophoresis (CZE), isotachophoresis (ITP), isoelectricfocusing (IEF), micellar electrokinetic chromatography (MEKC), affinity electrophoresis (ACE) and capillary electrochromatography (CEC). All have been used coupled to a mass spectrometer20. Analyte mixtures are separated in a capillary tube containing a conductive liquid medium subjected to a high voltage. Individual analytes are separated, to a first approximation based on their charge (or partial charge) to mass ratio and directly sprayed into the ESI source.
IMS: Ion Mobility Spectrometry is a separation technique where ions are subject to an applied electrical potential gradient in the presence of a neutral carrier gas such as argon. Analyte ions separate as a function of ‘drift time’ based on their ionic radius to charge of the ion. Since separations in the IMS drift tube take on the order of milliseconds and MS analysis occurs at a much faster rate, then it is possible to sample analyte ions from the IMS analysis in real time to obtain an IMS-MS spectrum rich in information content.
Basic principles of MS
A mass spectrometer is an analytical instrument that separates ions based on their m/z ratio and determines the molecular weight of elemental, chemical and biological compounds to a high degree of precision and accuracy (~10-3-10-6%) as well as sensitivity (detection of 10-9-10-21 moles of sample required). A simple schematic is shown in Figure 1A. A limitation of mass spectrometry is that compounds can only be analysed in the gas phase, either as negatively- or positively-charged ions. Hence the source region serves as both the sample inlet and ionisation chamber (Figure 1A). In the pharmaceutical sector commonly used ionisation techniques include EI, CI, Atmospheric Pressure Chemical ionisation (APCI) and Inductively Coupled Plasma (ICP) ionisation. These ionisation approaches are used to analyse small organic chemicals and metabolites (EI, CI and APCI) or elements and electrolytes (ICP).
More recently workers in pharmaceutical companies have focused primarily on using two common ‘soft’ ionisation techniques known as Electrospray Ionisation (ESI) and Matrix-Assisted Laser Desorption Ionisation (MALDI) for the analyses of biologically derived molecules. Soft ionisation refers to the ability to ionise and volatilise thermally labile compounds, such as peptides proteins, oligosaccharides, drug metabolites and other chemically fragile moieties, without inducing fragmentation or decomposition. The ESI process generates charged, micro-droplets containing analyte ions. Gentle evaporation of the droplets in the source region ultimately results in a chargetransfer from the water droplet surface to the analyte, leading to the creation of gas phase ions. Such ions are detected as a series of multiply charged ions (Figure 1B). In order to determine the molecular weight (Mr) of the compound, a simple algorithm ‘transforms’ this ion series into a single value Mr. In the case of MALDI, the analyte is mixed and co-crystallised with a photoactive organic acid matrix, which readily absorbs energy from laser irradiation. Hence, when the target containing analyte and matrix is placed in the source region and subjected to laser bombardment, analytes are projected into the gas phase, typically as singly charged ions (Figure 1C). Different ionisation methods can be used with MS analysers. However MALDI is most commonly associated with a timeof- flight (TOF) analyser (MALDI-TOF-MS). ESI is frequently coupled to a quadrupole, ion-trap, magnetic sector, quadrupole-TOF analyser, Fourier transform ion cyclotron resonance mass spectrometer (FTMS) or Orbitrap. In all cases, the MS analyser functions to separate the ions produced in the source region, based on their m/z ratio, which are then counted at a detection device, typically an electron multiplier or ion counting detector, as delineated in Figure 1A.
Structural information-tandem Mass Spectrometry or MSn
Soft ionisation techniques such as ESI, or its various analogues such as nanospray-ESI or picospray- ESI, and MALDI do not induce any significant fragmentation of compounds in the source region. Therefore, in order to acquire structural information on a biological compound of interest, such as a peptide sequence, it is necessary to induce fragmentation of the molecules/ions in the mass spectrometer. This can be achieved using tandem mass spectrometry (MS-MS), or multiples of MS (MSn), and this is shown schematically in Figure 2A. For example, a complex mixture of peptides is individually ionised in the source region using ESI or MALDI. These peptides are then further separated, based on their m/z ratio, by Mass Spectrometer 1 and detected at Detector 1 (Figure 2). Subsequently, one peptide, eg MH+=1801 is selected and subjected to fragmentation, and commonly referred to as collision-induced dissociation (CID) (Figure 2B). All the ions having an m/z 1801 are allowed into the collision cell, which is filled with an inert gas such as xenon or argon. The resulting collisions between ions and gas cause fragmentation of the peptide. These fragment ions (also known as product ions) are separated on their m/z values in Mass Spectrometer 2 and detected at Detector 2. For peptides the fragment ions represent loss of individual amino acids, and hence can provide valuable sequence/structural information as shown schematically in Figure 2C. It should be noted that there a number of methods now available to induce fragmentation and they include Electron Capture Dissociation (ECD), Electron Transfer Dissociation (ETD), Surface Induced Dissociation (SID), and Infrared Multiphoton Dissociation (IRMPD) and they are discussed in detail elsewhere14,19.
As shown in Figure 1A, data acquisition, representation and many aspects of data interpretation are under workstation control complete with a suite of software tools and some of the offerings provided by the MS instrument companies as detailed in Table 3 discussed below.
Data output: The most common form of data output from a modern MS instrument is in the form of a mass spectrum schematically shown in Figure 1B or 1C. The x-axis designates the m/z value of the ions detected and the y-axis indicates the relative ion abundance compared to the largest (base) peak. If an on-line chromatography-MS analysis is undertaken then the data representation can also be in the form of a mass chromatogram. In this case the x-axis is time (typically in minutes for GC, HPLC or CE and milliseconds for IMS) and the yaxis is still relative ion abundance. In addition there is now software commonly available to represent the data in three-dimensional plots where the x-axis is m/z, the y-axis is relative ion abundance and the z-axis is typically time.
Data analysis and interpretation: This broad and complex topic has received much attention and undergone massive development over the past decade. It is beyond the scope of this article, but the interested reader is referred to detailed reviews on the subject published elsewhere14. Suffice it to say that the over-interpretation of data and a poor understanding of the data that a MS instrument can produce has led to an over-extension of MS capability and this is discussed in more detail later.
Mass Spectrometry companies
The global market for MS instrumentation has grown rapidly over the past decade into a multibillion annual sales market. In 2009 global sales, instrument service contracts and ancillary services were approximately $3.3 billion21. The sector has continued strong growth prospects and is projected to experience annual growth of 8-10% through 2012. At least 27 different companies manufacture and sell a wide range of MS instrumentation. The global distribution breaks down in terms of numbers into 14 USA, nine European, three Asian (all Japan) and one Australian-based companies They range in size and market cap from the giant conglomerate Thermo-Fisher Scientific Inc (Waltham, MA, USA) with annual corporate revenues of $10.11 billion (2009) to Vitalea Sciences (Davis, CA, USA) which manufactures custom-built Accelerator-MS instruments. All these companies are listed in Table 1 along with contact website information and the date the company was either founded or created.
Bullish sales and growth have led to a flurry of mergers and acquisitions (M&A) over the past decade in this vibrant MS market. This was preceded by much more modest M&A activity. For example, in 1989 Shimadzu (Kyoto, Japan) acquired one of the pioneers of modern mass spectrometry, Kratos Analytical, based in Manchester UK. Then in 1997 Waters Corporation (Milford, MA, USA) bought Micromass Ltd (Manchester, UK), another early MS leader, for $178 million in cash. In the 00s (2000 and beyond) several strategies have emerged as companies seek to either enter into or leverage existing MS capability. In the former case, Biorad (Hercules, CA, USA), a major reagent and instrument supplier for the life sciences, purchased Ciphergen’s SELDI proteomics platform in 2006 for $20 million cash and $3 million equity. A good example of the latter situation was the acquisition of Ionspec by then Varian Corporation in 2006 for $17.3 million in cash and assumed debt. This was subsequently followed by the acquisition of Varian into Agilent Technologies (Santa Clara, CA, USA) in 2009 for $1.5 billion. As part of the deal and stipulated by the monopoly concerns, Varian also sold its ICPMS and GC-MS line of instruments to Bruker Daltonics (Billerica, MA, USA) for an undisclosed sum. In the same year Applied Biosystems merged with the life sciences reagent behemoth Invitrogen to create a new entity, Life Technologies Inc (Carlsbad, CA, USA), in a deal worth $6.7 billion in cash and stock. As part of the deal structure, Life Technologies Inc sold the MDS Sciex MS instrument division to Danaher (Washington DC, USA) for $1.1 billion to create AB Sciex. All the M&A activity of the past decade in the MS instrumentation sector is summarised in Table 2.
The MS manufacturing companies listed in Table 1 produce a diverse array of MS instruments designed to fulfill a variety of analytical demands. In addition they provide a plethora of software tools that enable facile data acquisition, processing, analysis and interpretation. All of this is captured and summarised in Table 3. Many of these companies produce instruments required for specific types of analysis including surface analysis and material sciences using Secondary Ion Mass spectrometry (SIMS) (including ASI, BME, CAMECA, Comstock Inc and Hiden Analytical); environmental analysis (including GSG, Kore Technologies, MKS Instruments, NU Instruments and Spectromat Massenspektrometer GmbH) or isotope ratio analysis (Monitor Instruments and Sercon). The pharmaceutical sector is primarily served by only 33% of the MS companies which produce MS instrumentation. The list includes Agilent Technologies, Bruker Daltonics Corporation, Danaher/AB Sciex, Hitachi High Tech, JEOL, Perkin Elmer Inc, Shimadzu, Thermo Fisher Sciences Inc and Waters Corporation. Companies such as Biorad and Leco Corporation (St Joseph, MI, USA), which offer an interesting range of MS instrument capabilities that would be of value to pharmaceutical companies, have not yet penetrated this lucrative sector. Finally, Vitalea Sciences custom-manufactures AMS instruments used in the burgeoning field of Phase 0 clinical studies and mass balance analyses in ADME Toxicology.
The MS companies which offer products to the pharmaceutical sector produce a diverse and powerful array of instrumentation that potentially meets all DDD needs, as summarised in Table 3. Single stage HPLC-MS instruments are routinely used in PK/PD and drug metabolism studies and are used primarily as a detector for the HPLC instrument. HPLC-MS/MS tandem instruments continue to garner significant technological development and are considered the workhorses of MS in drug discovery and pre-clinical studies. MALDITOF- MS instrumentation continues to play a significant role in drug discovery but the significant growth patterns of the early 2000 period have long subsided. FT-MS continues to attract considerable interest from pharmaceutical companies as they struggle to employ the superior capabilities of such instrumentation, whereas there is minimal interest and growth in the use of magnetic sector instruments in DDD. More recently there has been increasing demand for the new IMS-MS instrument offered by Waters Corporation.
Mass Spectrometry in DDD
Mass Spectrometry has evolved into a widely used technology across a number of diverse industrial sectors14. In a parallel process there continues to be rapid new developments in sample introduction, online chromatography-MS, creation of multifaceted separation approaches such as HPLC-nanoESIQ- IMS-TOF-MS/MS (eg Waters SYNAPT™ instrument), enhanced performance mass analysers and detectors and ever more sophisticated software for data analysis and interpretation. Pharmaceutical companies have attempted to utilise these improved technologies in their complex, multifaceted DDD process. Indeed, Papac and Shahrokh have argued that “Mass Spectrometry has significantly altered how the pharmaceutical and the biotechnology industries discover new therapeutics and develop them into safe and marketable drugs”15. There is no doubt that MS has been successfully utilised in a variety of different roles primarily in the Discovery and Preclinical phases of the DDD process. For example, in the late 1990s and early 2000 period combinatorial libraries and high throughput screening were in vogue. MS was used in a variety of different ways to characterise compounds from libraries, determine purity and in conjunction with bioassays perform high throughput screening. New generation HPLC-MS/MS and CE-MS/MS instruments have significantly enhanced throughput capability in the PK, PD and metabolism studies of NCEs. In addition the use of Accelerator-MS is providing the opportunity to carry out Phase 0 clinical trials22. Since Accelerator-MS has such exquisite sensitivity the in vivo metabolic distribution of a drug can be readily determined with minimal exposure levels of the radioactive compound. MS has been heavily used in the arena of omic analysis and biological therapeutics, but with more mixed indicators of success. In the case of simple QC and inprocess monitoring, as well as the structural determination of recombinant protein variants and degradation products, MS has been a useful tool. All of this is summarised in Table 4 and has been discussed in more detail elsewhere14,15-18.
The hype surrounding the advent of proteomic and metabolomic profiling using MS-based platforms has infiltrated the pharmaceutical sector over the past decade2. This led to an outlay of considerable capital by all the major pharmaceutical companies to purchase the latest MS technologies as well as build in-house technical expertise. Unfortunately, this expenditure was not translated into significant breakthroughs in early discovery efforts predicated on MS analyses. Armed with the clarity of hindsight it is interesting to observe that an interesting phenomenon occurred that explains the poor return on MS investment for proteomic and metabolomic analyses. The MS technologies, were evolving and constantly improving predicated on the efforts of the MS manufacturers during the past 10 years. Nevertheless, proteomic and metabolomic analyses were performed on relatively mature MS platform technologies. It was not the performance characteristics of the MS platform(s) that were poorly understood. Unfortunately, it was the application of MS to proteomic and metabolomic analyses that was not adequately understood and hence subject to the Hype Cycle. This led to poorly thought through experimental protocols and mismatched expectations of MS capability versus the quality and usefulness of data output. For example, in the differential analysis of blood samples (plasma or serum) from a control group versus a specific disease state cohort, little or no consideration was given to the well-known dynamic range limitations of MS. It should not have been surprising to learn that irrespective of the disease state being investigated, the same high abundant proteins typically associated with the inflammatory response were always found to be the differentiators between control and diseased populations. MS companies will continue to innovate and develop better instruments as well as firmware and software packages that control MS hardware and aid in data output and interpretation. However, the pharmaceutical sector is wisely re-evaluating and refocusing the role of MS in certain areas of the DDD process. In the case of early discovery using proteomic and metabolomic approaches, there needs to be a specific biological question being asked that requires a defined answer. Understanding the limitations and capabilities of the instrumentation, firmware and software in concert with the biological question is imperative for expectations to be met and satisfied. The days of fishing expeditions in the DDD process employing MS platform technologies are hopefully over.
MS predicated platform technologies will continue to play a significant role in specific areas of discovery and preclinical processes. Well-defined processes that require specific information content will continue to be served well by employing MS. The role of MS in clinical development will continue to be severely limited. However, by determining a specific need, for example in the newly emerging area of Companion Diagnostics7 and avoiding the Hype Cycle associated with application of MS to the problem, one can foresee future roles for MS in the clinical development process.
Finally, it is important for pharmaceutical companies to understand the limitations of current technologies and software. Much of the MS platform technologies are designed to serve two customer bases with very different expectations. The basic research community apply different standards to the interpretation of data compared to pharmaceutical companies which ultimately must satisfy stringent regulatory requirements imposed on them by FDA or EMA. This can sometimes lead to a mismatch of expectations in terms of MS capability. For example, many of the commercially available MS/MS data analysis programs used for peptide sequencing determination and protein identification, such as SEQUEST, Mascot, X! Tandem and others are much less than 100% accurate. Often times this level of inaccuracy may be tolerable for the basic researcher, but for the pharmaceutical company, such inaccuracies waste time, money and may jeopardise the progress of a NCE. Pharmaceutical companies need to understand such problems and limitations at the outset and seek ways to alleviate such shortcomings. In summary, the role of MS in the DDD process is assured provided realistic expectations are set and practical heads determine the decision making process.
Dr Naylor acknowledges the support of the National Institutes of Health through an SBIR Grant (5R44-HL082382-03) in partial support of this work.
Dr Stephen Naylor is a Founder, Chairman of the Board and CEO of D2D Inc, a start-up Companion Diagnostics Company. He is also Founder, President and Chairman of MaiHealth Inc, a molecular bioprofile diagnostics company in the health and wellness sector, as well as Founder, Chairman and former CEO of Predictive Physiology and Medicine Inc, a personalised medicine company offering a series of information content tools to consumers.
Paul T. Babcock is a public health professional working for the Marion County Health Department based in Indianapolis (Indiana, USA). He has a Master’s in Public Affairs with a focus on drug control and delivery policy and is involved in a series of early stage start-up ventures.