Utilisation of Analytical Method Greenness Score to drive sustainable chromatographic method development

Abstract

Sustainability considerations are gaining prominence within the pharmaceutical industry, driven primarily by increased awareness of environmental impacts associated with pharmaceutical development and manufacturing. Green and sustainable analytical chemistry is pivotal in minimising the environmental footprint of pharmaceutical processes, ensuring safer and more efficient methods of drug development and production. Critical to this is the ability to measure the impact of chromatography on the environment. One advancement that facilitates this is the Analytical Method Greenness Score (AMGS), a comprehensive metric developed by the American Chemical Society's Green Chemistry Institute, in collaboration with industry partners. This tool evaluates the environmental impact of chromatographic analytical methods across various dimensions, including the energy consumed in the production and disposal of solvents used, their safety/toxicity and more uniquely, instrument energy consumption. By integrating AMGS into routine analytical procedures, organisations can systematically improve their sustainability profiles, reduce hazardous waste, and promote the development of greener alternatives. This article underscores the critical role of AMGS in fostering more sustainable analytical test methods. It outlines how AstraZeneca have utilised this readily available tool to assess the current status, influence standard practices and create internal tools that trend data as a mode of continuous process verification. Through this work, we highlight the importance of adopting such metrics for long-term ecological and economic benefits.

---

Green foundation

1. We report the use of the Analytical Method Greenness Score (AMGS) to provide holistic and strategic insight within AstraZeneca for the chromatographic methods within our pharmaceutical drug project portfolio. We have used this approach to identify sustainable chromatographic method development and redevelopment opportunities, influence standard practices and create software tools that trend data as a mode of continuous process verification. In summation of our work, we provide general guidance on how an analyst can reduce AMGS method scores to produce more sustainable chromatographic methods.

2. We outline a qualitative and quantitative approach for the assessment of chromatographic methods at the departmental level through to the drug project and individual method level. We highlight general methods which have poor sustainability credentials and discuss how these might be improved through specific examples. We also provide quantitative insight into scores for the instrument energy, solvent EHS and solvent energy scores calculated by the AMGS.

3. The AMGS tool has several constraints, which if addressed, would add value to the calculation, for example, additional analytical techniques, impact of chromatographic mobile phase and sample diluent additives. This would assist with understanding method granularity and provide additional strategic insight to organisations.

Introduction

In recent years the drive towards sustainability has become increasingly important in the pharmaceutical industry. The field of green chemistry and the seminal ‘12 principles’ were first proposed by Anastas and Warner in 1998.¹ These principles have been widely adopted by many industries where synthetic process chemistry is undertaken and are well embedded as concepts to reduce the environmental impact (improve ‘greenness’) of synthetic transformations and increase colleague safety. It is less known that soon after the 12 principles were published, Anastas provided his first thoughts on green analytical chemistry² and outlined several areas that could benefit from sustainable thinking, for example the automation of sample preparation, at-point analytical measurements, use of experimental design (to reduce number of experiments undertaken) and the use of standardised (‘generic’) methods.

While the synthetic and process chemistry communities adopted green principles readily, the analytical community were slower to do this. In part, this was due to the large disparity in scale of organic solvents used in process versus analytical chemistry and analytical chemists having no standardised framework to introduce green principles into their work. In 2013, Gałuszka and coworkers³ proposed a set of green analytical chemistry principles, which acted as a catalyst for more focussed thinking in the field – leading to several other publications on this topic from Gałuszka and colleagues, e.g., ref. 4–6. The concept of green analytical chemistry (GAC) extends the principles of green chemistry¹ to analytical processes, emphasising the reduction of hazardous substances, energy consumption, and waste generation. Several papers touching on these concepts have been published as green practices in analytical chemistry have become more established e.g. ref. 7–11.

In the pharmaceutical industry, analytical methods are fundamental to the quality control and regulatory compliance from starting materials to drug substances and products. Analytical measurements are the foundation of a company's position in development and submission of new medicines, making it essential that they perform according to regulatory body guidelines.¹² Beyond their regulatory role, analytical methods are also pivotal to the advancement of sustainability, whether by enabling the assessment of processes or by directly adopting greener approaches in analytical methods. The United Nations’ 17 Sustainability Development Goals,¹³ represent an urgent call to action by all countries. In combination with other socio-economic needs, the development goals aim to tackle climate change and preserve oceans and forests. This has provided an additional driver to develop and implement greener chemical processes and analytical methods within the pharmaceutical industry to minimise their environmental footprint. It should be mandated however, that the quality of an analytical measurement should never be compromised at the expense of making a method greener. This is especially important in the pharmaceutical industry, where patient safety is paramount.

At AstraZeneca, the commitment to suitable analytical chemistry is exemplified by our ambition to achieve carbon zero status for analytical laboratories by 2030. Achieving this ambitious goal requires a holistic examination of all aspects of drug substance manufacturing, including the often overlooked yet significant role of analytical chemistry. The environmental impact of even a single analytical method can be substantial when scaled across global manufacturing networks.

A case study of rosuvastatin calcium, a widely used generic drug, illustrates this point. Across its manufacturing process, there are 9 isolated intermediates, and each batch undergoes approximately 25 liquid chromatography (LC) analyses. These include tests for end-of-reaction, release, stability studies, and other quality control checkpoints throughout the drug product lifecycle. With an average of 14 injections per analysis and a typical monograph method flow rate of 0.75 mL min⁻¹ over a 70-minute runtime, each batch consumes approximately 18 L of mobile phase. When scaled to an estimated 1000 batches produced globally each year, this results in the consumption and disposal of approximately 18 000 L of mobile phase annually for the chromatographic analysis of a single active pharmaceutical ingredient (API).

The widespread perception that analytical methods have an insignificant impact on environmental life cycle analyses is both pervasive and damaging. This mindset, if left unchallenged, risks exacerbating the environmental burden associated with pharmaceutical manufacturing. While an individual LC method may appear routine, its cumulative environmental cost underscores the urgent need for more sustainable approaches to analytical method design and execution.

This example highlights the critical role of analytical chemistry in driving progress toward sustainability goals. By rethinking and redesigning analytical methods with a focus on reducing waste and minimising environmental impact, the pharmaceutical industry can make significant strides in balancing environmental stewardship with the delivery of high-quality medicines.

One of the fundamental questions in relation to analytical methods is how to measure their greenness in a consistent and scientifically rigorous way. Over the last ca. two decades, several approaches have been proposed, and these were summarised in a publication by Sajid et al. in 2022.¹⁴ The evaluation of an analytical method's greenness involves assessing various parameters such as solvent and reagent usage, instrument energy requirements, waste production, and the potential for recycling and reusability. To facilitate this evaluation, several tools and metrics have been developed, each with its own criteria and assessment protocols. There have been several papers recently published highlighting the merits of each approach.^15–17 Among the most widely cited tools are the Analytical Eco-Scale,⁴ GAPI⁵ and AGREE¹⁸ which are commonly used for the general assessment of analytical procedures.¹⁴ However, it is important to note that these tools were not specifically designed for specific analytical techniques such as LC, and their assessment models remain relatively coarse for chromatography.¹⁹ By incorporating these tools into the development and selection of analytical methods, analysts can make informed decisions that align with organisational sustainability goals while ensuring the robustness and reliability of their analytical processes.

The Analytical Eco-Scale approach⁴ provides a semi-quantitative evaluation of the environmental impact of analytical methods. This tool assigns penalty points for various parameters, including the type and amount of reagents used, energy consumption, and waste generated. The total eco-scale score is calculated by subtracting the penalty points from a base value of 100. A higher score indicates a greener method. This approach is advantageous for its simplicity and ability to provide a clear, numerical indication of an analytical method's greenness. The eco-scale is particularly effective in identifying specific areas where improvements can be made to enhance method greenness.

The Green Analytical Procedure Index (GAPI) offers a more detailed evaluation by incorporating a holistic approach similar to AGREE.⁵ GAPI uses a pictogram consisting of five pentagrams, each representing different stages of an analytical procedure, from sample collection to waste management. Each pentagram is divided into three or four sections that are color-coded based on their environmental impact: green for low impact, yellow for moderate impact, and red for high impact. This visual and comprehensive representation allows for a quick yet detailed assessment of the greenness of an analytical method. GAPI is particularly useful for its detailed breakdown of each step in the analytical process, providing insights into specific areas that require improvement.

Another popular tool is Analytical GREEnness (AGREE), which offers a comprehensive evaluation of analytical methods based on the principles of green chemistry.¹⁸ AGREE provides a visual and numerical representation of the greenness of an analytical method through a radar chart, which includes twelve criteria such as energy consumption, solvent toxicity, waste generation, and safety. Each criterion is rated on a scale from 0 to 1, with higher values indicating better performance. The overall greenness score is the average of these ratings. AGREE's strength lies in its comprehensive, visual, and easily interpretable format, which facilitates communication of the method's environmental impact. It is also applicable across multiple analytical techniques, not just chromatographic. An on-line calculator is also available to help users interested in this approach.²⁰

Other notable approaches include Life Cycle Assessment (LCA) which is a more extensive and detailed approach for evaluating the environmental impact of a process.²¹ While it was first employed for calculating the impact of chemical reactions, it has more recently been applied to analytical methods²² and sample preparation.²³ LCA involves a cradle-to-grave analysis, assessing the environmental burdens associated with all stages of a product's life cycle, from raw material extraction to disposal, but its complexity and data-intensive nature can be a limitation. For preparative chromatographic separations used to isolate organic compounds, Process Mass Intensity (PMI) has also been used.²⁴ This approach compares the amount of material fed into the chromatographic purification versus the amount of material recovered. The lower the value the better the fraction recovery and therefore provides an assessment of the sustainability of the method.

Recently a tool called the Analytical Method Greenness Score (AMGS) was developed by a cross pharmaceutical company collaboration supported by the American Chemical Society Green Chemistry Institute (ACS GCI) analytical team.²⁵ AMGS is currently limited to LC and Supercritical Fluid Chromatography (SFC) with Ultraviolet (UV) or Mass Spectrometry (MS) detection. Each factor is assigned a score, and the cumulative score represents the overall greenness of the method. This scoring system enables easy comparison between different methods and highlights areas for potential improvement. The AMGS is particularly valuable for its straightforward and transparent approach and is being widely adopted in several major pharmaceutical companies (e.g., AstraZeneca, BMS, GSK, Merck, Takeda) as well as other institutions.^26,27 An online calculator for this tool is freely available on the ACS GCI website.²⁸

The AMGS framework considers multiple factors, summarised into three key terms:

• Solvent energy – type and volume of solvents used (for both the chromatographic method and the sample preparation) in terms of their cumulative energy demand (CED) for production and incineration.

• Solvent EHS – Environmental Health and Safety (EHS) impact of solvents used.

• Instrument Energy – average energy utilised by an instrument over the course of an analysis.

The tool calculates greenness scores based on eqn (1) where a score closer to zero represents increased method greenness or a method with decreasing environmental impact.

AMGS formula:

represents the combined mass m_s of solvent n used for preparing standards and samples; represents the mass m, of solvent n used by the instrument during gradient mobile phase elution at a specified flow rate; S_n is the safety index of solvent n; values range from 0.01 for methanol to 2.72 for acetonitrile based on reactivity, explosion hazard and acute toxicity as unit-less values; H_n is the health index of solvent n, these values range from 0.01 for methanol to 1.2 for acetonitrile based on handling, irritation and chronic toxicity (unit-less values); E_n is the solvent environmental index score for solvent n1 based on a given solvent's persistence, air and water hazard, e.g. 3.37 for methanol and 3.44 for acetonitrile (unit-less values); CED refers to the cumulative energy demand for each particular solvent n based on production and solvent incineration for disposal, measured as the unit kg-solvent per Mjoule-equivalent; E_i refers to the instrument energy consumption or tabulated measured values determined for each instrument type selected e.g. 0.64 for HPLC and 1.22 for SFC-MS in energy units of kW hours; R is the number of sample injections required for a given analysis set, including suitability standards, reference standards and all-inclusive blanks.

In this paper we discuss how we have incorporated the AMGS tool at AstraZeneca to grade chromatographic methods and provide insight into which individual methods (and methods across specific therapeutic drug projects) are least sustainable – and would, therefore, benefit from redevelopment. Secondly, we will outline how this understanding of what types of methods are least green (e.g. assay, impurities, trace methods etc.) has led to new approaches and thinking in developing green chromatographic methods by design. We also highlight development of our own internal AMGS Streamlit app (an open-source Python package) which reuses metadata from chromatographic analysis in our corporate electronic laboratory notebook, chromatographic data software and quality documentation system to simplify and standardize the analytical chemist's approach to calculating and visualising AMGS across a large number of methods. Finally, we discuss our thoughts on future direction for sustainable method assessment.

Experimental

Instrumentation

Ultra-High-Performance Liquid Chromatography (UHPLC), High-Performance Liquid Chromatography (HPLC), Supercritical Fluid Chromatography (SFC), and Liquid Chromatography-Mass Spectrometry (LC-MS) methods performed on Agilent 1260 and 1290 Infinity series LC instruments (Santa Clara, US) or Waters Acquity Classic and H-Class series LC and UPC² SFC instruments (Milford, US) instrumentation with either binary or quaternary pumps and either single-wavelength or multi-wavelength Ultraviolet (UV) detectors (and MS detectors for LC-MS methods). It is assumed that the instrument energy assessed by ACS GCI²⁵ is equivalent to the systems utilised at AstraZeneca.

Chemicals and reagents

All chromatographic methods captured utilised analytical grade chemicals and reagents or above, largely sourced from VWR (Leicestershire, UK). Organic solvents primarily include acetonitrile, methanol, ethanol and isopropanol. pH additives, although not utilised in the calculation of AMGS, were captured indirectly through the AstraZeneca internal tool. The majority of pH additives were trifluoroacetic acid, phosphate salts, ammonium acetate or ammonium hydroxide. Beverage grade carbon dioxide is utilised in SFC analytical methods.

Analytes utilised within this study are proprietary pharmaceutical molecules. The data presented consists of the analysis of primarily small molecules (96%) as well as some proprietary oligonucleotides (3%) and peptides (1%).

Methods

183 chromatographic methods analysed. ‘Number of analytes of interest:’ field in the AMGS ACS calculator²⁸ is assumed to be the number of analytes the method is specific for. Sample preparation volumes are as stated in the individual method documents. Exemplar methods and calculations using the ACS on-line calculator are provided in the SI. Fig. 1 depicts the analytical workflow utilised at AstraZeneca in relation to understanding greenness of methods using the AMGS calculator.

Fig. 1 AMGS analytical strategy implemented at AstraZeneca. *e.g., greener techniques, reduce diluent volumes, reduce column volumes.

Results and discussion

In late 2023, AstraZeneca initiated the calculation of Analytical Method Greenness Scores (AMGS) for all applicable late-stage drug substance analytical methods utilising LC and SFC via the ACS AMGS website.²⁸ This initiative was subsequently expanded across various departments and sites worldwide, enabling the creation of a comprehensive global assessment of the greenness of late-stage development LC and SFC methods, encompassing both drug substances, their pre-cursors, and products. As a result, a global company baseline for chromatographic analytical ‘greenness’ was established. This facilitated extensive data analysis, enabling the identification of key insights and the establishment of targeted goals for future improvements. The process revealed that HPLC accounted for 64% of late-stage methods, with only 34% employing UHPLC; a mode of analysis offering faster analysis times, higher resolution, and reduced solvent consumption, making it a more efficient and environmentally friendly option. Additionally, only 1% of methods utilise SFC, which is often considered a more sustainable option due to its lower solvent usage and reduced waste generation (Fig. 2).²⁹ Breaking down the data collected, valuable insights into inter- and intra-departmental variations, test methods, drug projects, technique choices, and detailed AMGS score breakdowns have been attained.

Fig. 2 Technique breakdown across all late-stage UHPLC-UV, HPLC-UV, SFC-UV and LC-MS analytical release methods at AstraZeneca for both drug substance and drug products.

Departmental insights

A multi-departmental assessment encompassing late-stage research and development departments at AstraZeneca was conducted to evaluate sustainability differences across the company. Notably, there were significant differences in the average AMGS score between drug substance and drug product departments. The average score (Fig. 3) for drug product methods was equal to 1422 (including Content Uniformity (CU) and dissolution methods), whereas for drug substance, this value was 45% less (representing a significant increase in overall sustainability). However, of the methods assessed, two-thirds were drug substance methods, hence though drug product methods have a higher average AMGS, they represent a smaller proportion of the global AstraZeneca chromatographic method portfolio. The higher average AMGS for drug product methods was primarily attributed to the different test methods employed and greater sample preparation required (e.g., larger number of dilution steps due to higher initial sample concentrations of formulations) versus that of drug substance.

Fig. 3 Average AMGS for drug substance (chemical development) and drug product (all other departments) analytical methods alongside number of methods by department at AstraZeneca.

The AMGS scoring illustrated that CU methods were of particular interest. CU is a key test used to ensure label claim consistency across a drug product batch, governed by USP Uniformity of Dosage Units (UoDU).³⁰ UoDU can be assessed by either CU, where a minimum of 10 individual dosage unit preparations are assayed, or weight variation, where assay is assumed uniform across the batch and label claim is calculated from individual tablet weights. Although weight variation has the sustainability benefit of avoiding multiple sample preparations, it is not applicable to all dosage forms (e.g., liquid parenterals) – unlike CU, and hence, CU is currently still widely used within the pharmaceutical industry. In this case, the large solvent volumes required to prepare a minimum of 10 samples, contributes heavily to significantly higher scores for Solvent Energy (Fig. 4). Additionally, elevated scores for Solvent EHS are obtained due to the larger amounts of less green solvents, resulting in a higher overall AMGS score.

Fig. 4 Average AMGS breakdown across test methods. CU = content uniformity, ID = identity, Org imps = organic impurities, DDU = delivered dose uniformity, PSD = particle size distribution, IPC = in-process control.

This information has provided AstraZeneca with an understanding of one of the key greenness offenders affecting the total AMGS average. Drug product CU methods are often combined with stability-indicating organic degradant analysis, adopting the same chromatographic conditions (differing in the number of sample replicates and preparation schemes), often leading to long analysis times to quantify a single analyte, which based on eqn (1) yields a high AMGS, particularly due to the number of analytes. Opportunities to validate fast, generic methods for these purposes are being explored to deliver greenness improvements. CU methods had an average AMGS of 2915 which highlights the substantial negative impact these test methods have on the overall dataset (Fig. 4), despite comprising only 18% of the drug product data.

Test methods

Quantitative and qualitative pharmaceutical methods usually include a series of standard test injections to determine system suitability. These injections, containing test compounds, are typically used to ensure sufficient chromatographic resolution (e.g., drug substance impurity methods) or acceptable area reproducibility (system precision) prior to analysis of the test samples. As such, the resulting method greenness can vary greatly. The most common test methods, combined assay and organic impurities along with combined assay, organic impurities and identity contributed to ∼30% of the database. The breakdown of the AMGS of these test methods expressed Solvent Energy, Instrument Energy and Solvent EHS to represent 42%, 35% and 23% respectively.

Based on this information, a deeper review of the methods was undertaken. This revealed that how a method is documented, can have a large contribution to the AMGS. Many methods utilised excessive sample preparation volumes for calibration, recovery, limit of quantification (LOQ) standards, resolution and retention standard injections and sample solutions when describing preparation guidance. For example, a drug substance assay and impurities LC method in the database, detailed a total standard/solution preparation volume of 1030 mL which gave an overall AMGS of 834.

However, scaling down these preparations by a factor of 5, gives a recalculated AMGS of 325, a 61% reduction (Fig. 5), just by simply writing a method with greenness in mind whilst maintaining sufficient method accuracy and precision. A good practice is to state the minimum diluent required to prepare all solutions for a given test method, instead of providing a generic preparation for 1 L of diluent, as often, less than 0.5 L of diluent is required during sample preparation for drug substance methods.

Fig. 5 A UHPLC-UV method's AMGS breakdown, comparing its original preparation and an alternative scaled down preparation, illustrating the impact of minimising example preparation volumes that are written in methods.

CU methods score significantly higher compared to any other test method, a difference that can largely be explained by the extensive sample preparation they require relative to other methods. While dissolution methods also score highly, the underlying factors driving their scores differ significantly, as reflected in the distinct AMGS breakdowns (Fig. 4). CU methods (as a percentage) utilise minimal Instrument Energy and Solvent EHS only represents 27%, whereas Solvent Energy dominates, making up 59% of the AMGS figure. This is unsurprising, considering CU testing typically involves large sample preparation volumes with organic solvents requiring energy intensive production and disposal. In contrast, for dissolution assay methods, the AMGS breakdown is 2%, 30%, and 68% for Solvent Energy, Instrument Energy and Solvent EHS, respectively. Dissolution utilises water principally in its sample preparation (i.e., the dissolution media), but organic solvents are used in its mobile phases as with any typical LC method, and of the dissolution methods assessed, the vast majority were HPLC instead of UHPLC. The latter point helps to explains the higher instrument energy, not because the in-use power consumption of HPLC versus UHPLC is higher (they are actually very similar in value²⁵), but HPLC typically has longer analysis times relative to UHPLC, which therefore results in larger Instrument Energy. It is also contributed to by the substantial number of injections required for a typical sequence, 51, whereas a typical assay and organic impurities method requires only 17 injections. As for the minimal Solvent Energy and high Solvent EHS this data appears anomalous because, due to the large amounts of water utilised, a low Solvent EHS would be expected, along with potentially a high Solvent Energy to account for the energy barrier associated with the incineration of contaminated water. Further investigation is required to fully understand this data. AMGS dissolution data is also influenced by uncaptured data. Only LC dissolution methods have been captured, not UV finish dissolution methods using on- or off-line spectrophotometry (as AMGS only accounts for LC and SFC methods), which are far greener. Dissolution using a UV finish is current standard practice for AstraZeneca internal methods, and the AMGS dataset contains LC dissolution data, mainly as a result of historic Contract Manufacturing Organisations (CMO) methods, and UV interference of other components.

As a general point on test methods, efforts should be made to combine multiple test methods into one where possible. For example, an LC assay method, LC organic impurities method and LC identity method can often be developed as a trio combination of test methods, reducing the number of methods by two, providing easy, effective sustainability and time savings.

AMGS breakdown

Reviewing the AMGS data across all departments led to an interesting finding: Solvent Energy contributes 43% to the total AstraZeneca AMGS, whereas Solvent EHS and Instrument Energy only contribute 29% and 28%, respectively (Fig. 6).

Fig. 6 AMGS breakdown across AstraZeneca late-stage analytical methods.

Fig. 6 highlights that historically, many of AstraZeneca's chromatographic methods have been developed with little consideration of their environmental impact, utilising large preparation volumes. However, AMGS highlights the benefits of well-crafted methods and thorough sample preparation. These scores can encourage users to design and optimise methods with reduced sample preparation volumes from their first iteration, throughout development and into commercial quality control. This effectively brings sustainability to the forefront of every analyst's mind, fostering a culture of environmentally conscious practices. CU test methods help complete the picture as to why Solvent Energy dominates due to the very large preparation volumes required of solvents requiring intensive energy production and disposal.

Drug projects

The implementation of AMGS not only provided enough data to represent the late-stage greenness of LC and SFC analytical methods at AstraZeneca as a whole, but also sufficient data to detail the inter- and intra-drug project differences. Inter-drug projects data provided useful insights into the age of projects and technology available at the time of development – a strong correlation was observed between older drug projects and the number of HPLC versus UHPLC methods utilised. Fig. 7 details the AMGS totals across multiple late-stage drug projects at AstraZeneca. Fig. 8 compares Project A, which has the highest total AMGS across all AstraZeneca late-stage drug projects, against Project F which has ∼70% lower AMGS sum than Project A.

Fig. 7 Total AMGS summed across all relevant analytical methods for each AstraZeneca drug project.

Fig. 8 Project H's utilised analytical techniques breakdown (left) and LC breakdown (right) with UV detection.

The difference lies in the LC technique used: Project A relies entirely on HPLC, while Project F uses only 21% HPLC. This highlights how AMGS aids in distinguishing the greenness between HPLC and UHPLC, where a simple method translation can switch to the superior UHPLC for better greenness, efficiency, and speed.³¹ Project A has been in late-stage development for over 20 years, whereas Project F only reached this stage in the last 4 years. A decade ago, AstraZeneca's LC footprint was dominated by HPLC systems, leading to most chromatographic methods being developed using HPLC. Recent capital investments have made UHPLC the dominant system in research and development laboratories globally, with HPLC now a minority. This shift was due to UHPLC's superior capabilities, offering faster analysis and higher peak capacity.³¹ Consequently, Project F developed methods on UHPLC, due to their increased availability and enhanced performance rather than due to sustainability choices initially. AstraZeneca is now well-positioned to develop Reversed Phase Liquid Chromatography (RPLC) methods using UHPLC, highlighting the impact of investment decisions in this sustainability transformation.

Further data analysis illustrated other differences within LC methods – in particular RPLC versus Normal Phase Liquid Chromatography (NPLC). Fig. 8 compares Project H's technique utility, a drug project which entered late-stage development 5 years ago.

By utilising UHPLC for 53% of the LC methods and inclusion of SFC methods, its total AMGS decreases relative to projects that predominantly utilise HPLC. However, despite Project H utilising 35% HPLC methods its overall AMGS (9468) is still lower than Project F's (11444) which utilises UHPLC more than project H. This lower score can partly be attributed to the use of two SFC methods for enantiomeric purity analysis at both the crude (non-processed API) and API stage, which each have a very low AMGS of 44. SFC doesn't just offer orthogonal selectivity but utilises more benign solvents – carbon dioxide – and uses less organic solvent (which constitutes at most, 50% of the mobile phase). This leads to lower greenhouse gas emissions and waste generation. Due to the physical properties of the carbon dioxide, the flow path has lower viscosity compared to liquid flow paths in LC, which allow for higher flow rates, and thus faster separations being achieved,³¹ resulting in less instrument energy being consumed over the course of a typical analysis. The combination of these factors results in reduced waste and reduced greenhouse gas emissions for SFC analysis relative to LC. AMGS has been built to help underline the points above, and the very low AMGS for SFC reflects this. Project H's SFC methods were developed internally, however 20% of its LC methods are NPLC, and are a result of outsourced method development. NPLC, involves the use of higher volumes of environmentally unfriendly and unsafe solvents,²⁹ as well as HPLC systems, resulting in long run times which in turn lead to greater instrument and solvent energy usage, yielding higher AMGS. When compared with Project B with an AMGS total of 23 488, and all its LC methods utilising RPLC, this can appear to be misleading.

What AMGS doesn't account for, in this setting, is the control strategy (a comprehensive plan designed to ensure that a pharmaceutical product consistently meets its quality requirements³²) or the overall synthetic route of a drug project. More steps in a route will typically result in more chromatographic testing required. An intelligently designed risk-based approach involving minimalist testing, will result in an overall lower AMGS for the project – the best way to be sustainable in analytical chemistry is to minimise testing and remove methods which do not offer additional value regarding material quality. For instance, in a five-stage synthesis, a potential mutagenic impurity (PMI) may be present at stage 1 and one approach to ensure control of this impurity, is to test at each of the five stages for this impurity during manufacture, which would provide absolute understanding of how the impurity purges (i.e., is removed or reduced to a level considered non-toxic) through the process. A more sustainable and faster approach would be to generate understanding of the impurity during development and only test once at the most suitable stage during manufacture e.g., if the PMI purges sufficiently by stage 3, a chromatographic test at stage 3 solely, will demonstrate control of the impurity, and also reduce the amount of testing by 80%. In this context, the number of steps in the synthetic route for Project B is greater than H, which results in a greater amount of testing. It also utilises HPLC instead of UHPLC for 95% of its methods, leading to an overall larger AMGS total relative to Project G.

Intra-drug project AMGS data highlights how different testing requirements in a synthetic route result in very different AMGS breakdowns. Fig. 9 demonstrates that API and crude testing result in very similar ratios of Solvent EHS, Solvent Energy and Instrument Energy, and this is to be expected when considering that most of the testing performed at crude are repeated at the API stage, utilising the same methods (crude to API formation is often just a purification step or salt formation). For intermediates and Regulatory Starting Materials (RSMs), though their AMGS breakdown is comparable to one another, the Instrument Energy percentage is greatly reduced relative to API and crude, yet the Solvent Energy percentage dominates by over 40%. One plausible explanation for this, is due to the average number of components analysed being higher at API and crude stage. As it is the last/penultimate stage of substance control, these methods typically require longer chromatographic analysis to achieve suitable specificity, resulting in greater amounts of Instrument Energy for the number of components requiring control. Contrastingly, at the intermediate and RSM stages, often fewer components are analysed, so quicker chromatographic methods can be developed with suitable specificity, requiring less Instrument Energy, resulting in Solvent Energy (as well as Solvent EHS) being the predominating AMGS factors, due to the volume and nature of solvents used. A similar trend can be observed for drug product methods, where the total AMGS is dominated by Solvent EHS and Solvent Energy by similar amounts. This is likely due to the heavy skewing that CU and dissolution methods contribute in terms of their high AMGS which when their AMGS breakdown is viewed together, Solvent EHS and Solvent Energy dominate, contrasting with the drug product average assay, organic impurities and identity method that have a more equal weighting between all three factors.

Fig. 9 Normalised AMGS breakdown as an average across all methods across a synthetic route.

This level of insight was not available prior to AMGS and has allowed for targeted, granular greenness improvements, instead of relying on broad and unspecific goals. Now AstraZeneca can focus on specific activities to achieve analytcial method greenness e.g., for intermediate and RSM methods, efforts should be focussed on the use of greener solvents such as methanol instead of acetonitrile, and reduction in solvent volumes, or where possible, a switch to a greener separation technique altogether.

An example of switching to a greener organic modifier has been enacted in the commercial space. Replacing the organic acetonitrile mobile phase for bio-renewable ethanol has led to a 35% reduction in analysis time, a 40% reduction in instrument energy and 87% reduction in organic solvent volume required. The summation of these factors culminates in a total 76% reduction in AMGS (245 reduced to 59), whilst vastly improving performance and sensitivity of the analytical method. A demonstration of the ease of applicability of bio-renewable solvents in method development is shown in Fig. 10. A direct comparison of acetonitrile, methanol and (bio)ethanol resolving power for a small molecule pharmaceutical intermediate and its associated impurities is illustrated. While the (bio)ethanol method is longer (due to viscosity constraints on the flow rate), the total volume of organic solvent used in each analysis is comparable and the use of (bio)ethanol leads to lower AMGS scores.

Fig. 10 Example comparison of resolution in UHPLC-UV method development for different organic solvents, utilising a BEH C18 column. Chromatography conditions have been optimised based on solvent properties e.g., a lower flow rate was used with the more viscous ethanol analysis. Compounds A–F are proprietary pharmaceutical small molecules with a molecular weight range of 94–321 g mol⁻¹, a log D range of 0.06–5.02 and contain a combination of phenolic, benzoic acid, primary aniline, ketone, amide, sulphide and carbamate functional groups. Column: waters acquity BEH C18, 2.1 × 50 mm, 1.7 μm. Column temperature: 40 °C. Ethanol analytical method: flow rate 0.4 mL min⁻¹, gradient: 0 min 5% ethanol, 6.3 min 90% ethanol, 7.1 min 90% ethanol. Acetonitrile analytical method: flow rate 0.8 mL min⁻¹, gradient: 0 min 5% acetonitrile, 4.5 min 90% acetonitrile, 4.9 min 90% acetonitrile. Methanol analytical method: flow rate 0.6 mL min⁻¹, gradient: 0 min 5% methanol, 4 min 90% methanol, 4.5 min 90% methanol.

However, for API and crude methods, efforts would be better focussed on minimising analysis time by reducing the size of column dimensions and utilising smaller particle sizes, for example.

Technique choice

The primary analytical techniques utilised have been briefly discussed earlier, but further discussion of the relative merits of each technique regarding ‘greenness’ can be exemplified. Scaling of HPLC methods to UHPLC typically offers ∼80% solvent savings due to the higher chromatographic efficiency per unit time and lower absolute volumes of mobile phase used,³³ leading to significant reductions in an AMGS. Fig. 11 exemplifies this, demonstrating a 67% reduction in AMGS (66% reduction in Solvent Energy, 62% reduction in Solvent EHS and 71% reduction in Instrument Energy) by simply translating a method between the two techniques. The allowance for scaling from HPLC to UHPLC in the harmonized USP General Chapter: Chromatography³⁴ has significant implications for analytical laboratories. By providing guidelines for method scaling, it allows for the potential enhancement of analytical performance. This means that laboratories can take advantage of the improved resolution, sensitivity, and speed offered by UHPLC without needing to completely redevelop methods. However, it's important to note that while scaling can offer these benefits, it also requires careful consideration and verification of various factors such as the impact on retention order, resolution, peak capacity, and method robustness. Therefore, the allowance for scaling in the USP General Chapter³⁴ provides a framework for leveraging the advantages of UHPLC while ensuring the reliability and accuracy of analytical results.

Fig. 11 AMGS of an organic impurities analytical method originally developed utilising HPLC-UV but translated to a UHPLC-UV method utilising LC Method Transfer Assistant (ACD Labs). Compounds A–J are proprietary pharmaceutical small molecules with a molecular weight range of 535–1140 g mol⁻¹, a log D range of 4.76–11.20 and contain a combination of ester, primary alcohol, ether, aromatic halogen and ketone functional groups. Mobile phase A: water : methanol : acetonitrile : TFA (40 : 30 : 30 : 0.05), mobile phase B: methanol : acetonitrile : TFA (50 : 50 : 0.05), column temperature: 30 °C. HPLC specific analytical method conditions: column YMC Hydrosphere C18, 4.6 × 150 mm, 3 μm. Flow-rate: 1 mL min⁻¹. Injection volume: 10 μL. UHPLC specific analytical method conditions: column waters acquity CSH C18, 2.1 × 100 mm, 1.7 μm. Flow-rate: 0.36 mL min⁻¹. Injection volume: 2 μL.

Another technique which has nearly completely replaced NPLC chiral method design and optimisation in early-stage development at AstraZeneca, is SFC. Providing faster separations, reduced solvent usage, and alternative selectivity, SFC offers a greener alternative to NPLC,²⁹ which was historically often used for chiral separations at AstraZeneca. NPLC methods are often long and use large amounts of toxic solvents such as liquid alkanes and halogenated solvents. Chiral LC methods represent 9% of the AstraZeneca AMGS database (Fig. 12) and there is an opportunity to replace these with SFC methods to yield both throughput and sustainability savings.

Fig. 12 Technique breakdown across all late-stage LC (RPLC and NPLC) and SFC analytical methods with UV detection, captured at AstraZeneca.

SFC typically utilises a non-aqueous mobile phase system, which can be very beneficial for hydrolytically unstable compounds. Boronate esters are such a class of compounds, which form their boronic acid equivalents in the presence of acid or water. They are often found in drug synthetic routes as part of a Suzuki–Miyaura coupling.³⁵ However, RPLC analysis of these compounds has been an enduring challenge, as the aqueous mobile phases can result in the on-column formation of both the boronate ester and boronic acid leading to challenges in accurate quantification. SFC offers an ideal alternative, providing a non-aqueous system, simplifying the development process greatly, as only the boronate ester needs to be quantified (efforts still need to be made to understand rate of hydrolysis in methanol, a commonly used organic modifier in SFC analysis, but this is typically much lower than water).³⁶Fig. 13 illustrates an example of a boronate ester method which was originally developed utilising HPLC but later redeveloped to a SFC method, which resulted in an 86% improvement in AMGS (98% reduction in Solvent Energy, 83% reduction in Solvent EHS and 50% reduction in Instrument Energy).

Fig. 13 AMGS breakdown of a boronate ester intermediate method, originally developed utilising HPLC, but later redeveloped to an SFC method.

SFC, however, is not without its challenges. Whilst SFC technology is available internally, when manufacturing and testing of the development chemistry is outsourced to CMOs, the majority of capable suppliers often do not have SFC instrumentation. This creates a bottleneck in sustainable chromatographic method development and requires investment by CMOs in expertise and instrumentation of sustainable technologies such as SFC to bring about a transformation in capability.

Investigating alternative approaches to assay solid-oral dosage form drug products can result in very low AMGS. There is an established trend within the pharmaceutical industry to adopt spectroscopic approaches, where feasible. Although this process is inherently method- and analyte-specific, techniques such as near-Infrared Spectroscopy require no sample preparation, other than dispensing the sample onto the probe,³⁷ vastly reducing solvent consumption. In addition, as the analysis is performed in less than a second, the instrument energy is negligible, and data can be generated in real time. This approach, based on the AMGS formula, would result in an AMGS close to zero. Examples such as this illustrate how substantial improvements in both environmental sustainability and analytical efficiency can be realized by critically evaluating the requirements defined by the analytical target profile, and by carefully considering when chromatographic methods are truly necessary versus utilizing spectroscopic techniques, when appropriate.³⁷

Limitations and future requirements for AMGS

The AMGS metric, while a useful tool for assessing the sustainability of analytical methods, has several limitations. One significant shortcoming is its current inability to account for pH additives, which are crucial in evaluating the environmental impact of a method. Additionally, AMGS is limited in scope, as it can only calculate scores for LC and SFC methods. The relative nature of the AMGS unitless numerical value introduces ambiguity regarding its interpretation, often leading to the question: what defines a “good” AMGS? Furthermore, the website tool lacks a database, hindering the ability to track trends or conduct data analysis. Addressing these limitations is crucial for enhancing the metric's effectiveness and clarity in future applications.

Our work has led to the development of an internal AMGS app hosted through Streamlit (open-source Python library), which replicates the AMGS ACS calculation and scoring system, with some additional benefits over the ACS GCI hosted interface. Firstly, the app reads metadata across multiple AstraZeneca systems; our chromatography data system, electronic lab notebooks and quality documentation system, which greatly speeds up the rate of AMGS calculation, driving compliance and adoption of the tool. It also reduces user flexibility and minimises ‘free-text’ fields, improving the accuracy and integrity of the data collected. In addition, it has its own database with integrated visuals allowing for live tracking of trends across departments, drug projects, method versions, techniques, sites etc., providing visibility of the analytical method greenness impact, at both a granular and high-level. This will allow AstraZeneca to quantify the impact of chromatographic analytical methods on a global level and across a product's lifecycle.

Recommended best practices for green analytical methods (and low AMGS)

1. Technique choice. UHPLC and SFC offer avenues for both greener and faster separations with orthogonal selectivity to one another, providing a quality technique platform for challenging separations. Use of HPLC, in particular NPLC should be avoided wherever possible. Further still – is chromatography essential for your analysis, can other techniques be utilised e.g., infrared/Raman spectroscopy, MS?

2. Solvent choice. Utilise greener solvents wherever possible – investigate the impact of using ethanol or methanol (possibly from renewable sources) instead of acetonitrile as the LC organic modifier. Making use of SFC greatly reduces the use of environmentally unfriendly solvents, primarily due to use of carbon dioxide within the mobile phase instead.

3. pH additive utilised. Critically review the logD and pKa values of all components being analysed in a method. Is there another suitable pH region to analyse at that facilitates use of a more environmentally benign reagent.

4. Smaller column dimensions and stationary phase particle sizes. Faster separations, less solvent used, reduced instrument energy consumption – a simple change with a large impact when pharmacopeial guidance is followed.

5. Write the method appropriately. Do not prepare excessive diluent or mobile phase volumes. Calculate the required amount for a typical analysis and add a suitable, non-excessive, overage.

6. Faster gradients. Optimise methods to the most suitable, robust design space but with length of gradient in mind too – can the gradient time be shortened, and system suitability criteria still be met?

7. Control strategy. The best approach to be sustainable in analytical chemistry is to perform no testing at all. Review each test you perform and assess the impact of the data it informs and whether this can be combined in another test, or removed altogether.

What is a good AMGS? Based on the data collected at AstraZeneca, we propose the three categories below to provide greater meaning to the relative score:

• Good – AMGS < 200

• Average – 200 ≤ AMGS ≤ 600

• Poor – AMGS > 600

Conclusions

The introduction of AMGS at AstraZeneca has allowed for a comprehensive assessment of the sustainability of chromatographic methods, offering valuable insights for improvement. The global implementation has provided detailed analysis of the greenness of late-stage methods, revealing significant differences between drug substance and drug product aligned departments. The study has highlighted the impact of various test methods and the potential for redevelopment to enhance sustainability.

Although other analytical sustainability metrics exist, the strength of AMGS lies in its simplicity, broad applicability and non-subjectivity, positioning it as the preferred analytical sustainability metric at AstraZeneca. Users can compute a score for any LC or SFC method across various industries in less than 10 minutes, ensuring high score repeatability. This means that if another user calculates a score for the same method, they are likely to obtain the same result, due to the minimal ambiguity and subjectivity in the values required by the calculator, thereby enhancing the accuracy of AMGS data – a benefit that other GAC metrics struggle to meet.¹⁹ A metric intended for widespread use must be straightforward, easy to comprehend, and objective.

While acknowledging the current limitations of AMGS, including the lack of consideration for pH additives and its relative scoring system, ongoing efforts to address these limitations are underway. At AstraZeneca, we are continually developing the tool further to enhance its capacity to promote environmental sustainability. Fundamentally, AMGS acts as a guiding framework for advancing sustainable practices and should be employed in conjunction with rigorous scientific evaluation and critical analysis.

This work has highlighted key areas for AstraZeneca to target for greenness improvements while fostering a culture of sustainability across its global analytical departments. Sustainability is now embedded at the very outset of each method development process, ensuring that greener principles are integrated from the earliest stages of analytical design, without compromising on method performance. Additionally, AMGS supports in identification and prioritisation of commercial methods which could benefit from redevelopment. Together, these factors are driving a positive sustainability impact across all stages of the drug development lifecycle.

In summary, this research demonstrates the importance of integrating sustainability metrics, such as AMGS, into pharmaceutical analytical processes. The insights gained from AMGS, have the potential to guide the industry towards greener practices and drive a culture of sustainability, ultimately contributing to a more environmentally conscious pharmaceutical sector.

Author contributions

Francis Power: investigation, methodology, authoring original draft, review and editing. Paul Ferguson: authoring original draft, review and editing. Abi Herbert: conceptualisation, investigation, authoring original draft, review and editing. Sara Ryan: investigation, authoring original draft. Matt Osborne: ideation, supervision, guidance, authoring, review & editing. Louie Trezise: methodology.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data for this article, including calculations of AMGS, have been produced using the publicly available AMGS calculator that can is available at https://www.acsgcipr.org/amgs/. See supplementary information (SI) for example calculations.

Supplementary information is available. See DOI: https://doi.org/10.1039/d5gc01574j.

Acknowledgements

We would like to thank Malin Wallberg and Thomas Rardin for providing data coordination for AMGS at their respective development sites. We would also like to thank Alexander Blanazs for providing valuable data input.

References

1. P. T. Anastas and J. C. Warner, Green Chemistry: Theory and Practice, Oxford University Press, New York, 1998.
2. P. T. Anastas, Crit. Rev. Anal. Chem., 1999, 29, 167–175,  DOI:10.1080/10408349891199356.
3. A. Gałuszka, Z. Migaszewski and J. Namiesnik, TrAC, Trends Anal. Chem., 2013, 50, 78–84,  DOI:10.1016/j.trac.2013.04.010.
4. A. Gałuszka, Z. M. Migaszewski, P. Konieczka and J. Namiesnik, TrAC, Trends Anal. Chem., 2012, 37, 61–72,  DOI:10.1016/J.TRAC.2012.03.013.
5. J. Płotka-Wasylka, Talanta, 2018, 181, 204–209,  DOI:10.1016/j.talanta.2018.01.013.
6. F. Pena-Pereira, W. Wojnowski and M. Tobiszewski, Anal. Chem., 2020, 92, 10076–10082,  DOI:10.1021/acs.analchem.0c01887.
7. M. Yabr, et al. , Molecules, 2018, 23(5), 1065,  DOI:10.3390/molecules23051065.
8. N. Nakov, et al., Green Strategies toward Eco-Friendly HPLC Methods in Pharma Analysis in High Performance Liquid Chromatography – Recent Advances and Applications, in IntechOpen, ed. O. Nunez, S. Sentellas, M. Granados and J. Saurina, 2023,  DOI:10.5772/intechopen.110035.
9. S. Ražić, et al. , Analyst, 2023, 148, 3130–3152,  10.1039/D3AN00498H.
10. M. Mehta, D. Mehta and R. Mashru, Futur. J. Pharm. Sci., 2024, 10, 83,  DOI:10.1186/s43094-024-00658-6.
11. S. E. Deeb, Molecules, 2024, 29, 3205,  DOI:10.3390/molecules29133205.
12. ICH Q2(R2) Analytical validation, https://www.ich.org/page/quality-guidelines (accessed December 2024).
13. United Nations – Sustainable Development Goals, https://sdgs.un.org/goals (accessed December 2024).
14. M. Sajid and J. Płotka-Wasylka, Talanta, 2022, 238, 123046,  DOI:10.1016/j.talanta.2021.123046.
15. L. P. Kowtharapu, et al. , TrAC, Trends Anal. Chem., 2023, 166, 117196,  DOI:10.1016/j.trac.2023.117196.
16. M. Shi, et al. , TrAC, Trends Anal. Chem., 2023, 166, 117211,  DOI:10.1016/j.trac.2023.117211.
17. L. Yin, et al. , J. Pharm. Anal., 2024, 14, 101013,  DOI:10.1016/j.jpha.2024.101013.
18. F. Pena-Pereira, et al. , Anal. Chem., 2020, 92(2), 10076–10082,  DOI:10.1021/acs.analchem.0c01887.
19. L. A. Yahya, et al. , Anal. Chim. Acta, 2025, 1357, 344052,  DOI:10.1016/j.aca.2025.344052.
20. AGREE Analytical Greenness Calculator, https://mostwiedzy.pl/en/wojciech-wojnowski,174235-1/AGREE? (accessed August 2024).
21. D. Kralisch, D. Ott and D. Gericke, Green Chem., 2015, 17, 123–145,  10.1039/C4GC01153H.
22. B. Raccary, et al. , TrAC, Trends Anal. Chem., 2022, 147, 116525,  DOI:10.1016/j.trac.2022.116525.
23. B. Raccary, et al. , Adv. Samp. Prep., 2022, 1, 100009,  DOI:10.1016/j.sampre.2022.100009.
24. J. Kingston, et al., in Applications of preparative SFC in the pharmaceutical industry in practical application of supercritical fluid chromatography for pharmaceutical research and development, ed. M. Hicks and P. Ferguson, Elsevier Academic Press, San Diego, USA, 2022.
25. M. B. Hicks, et al. , Green Chem., 2019, 21, 1816,  10.1039/C8GC03875A.
26. H. K. Chanduluru and A. Sugumaran, RSC Adv., 2022, 12, 6683,  10.1039/D1RA08858K.
27. T. T. Handlovic, et al. , Green Chem., 2024, 26, 760,  10.1039/D3GC03005A.
28. ACS: About the AMGS Calculator, https://acsgcipr.org/tools/about-the-amgs-calculator/(accessed September 2024).
29. C. J. Welch, et al. , TrAC, Trends Anal. Chem., 2010, 29, 667–680,  DOI:10.1016/j.trac.2010.03.008.
30. USP 〈905〉 Uniformity of dosage units, https://www.usp.org/(accessed December 2024).
31. J. Płotka, et al. , J. Chromatogr., A, 2013, 1307, 1–20,  DOI:10.1016/j.chroma.2013.07.099.
32. ICH quality guidelines, Q8(R2), https://www.ich.org/page/quality-guidelines (accessed December 2024).
33. S. Chen and A. Kord, J. Chromatogr., A, 2009, 1216, 6204–6209,  DOI:10.1016/j.chroma.2009.06.084.
34. USP 〈621〉 Chromatography. https://www.usp.org/(accessed February 2025).
35. M. Farhang, A. R. Akbarzadeh, M. Rabbani and A. M. Ghadiri, Polyhedron, 2022, 227, 116124,  DOI:10.1016/j.poly.2022.116124.
36. H. Steinberg and D. L. Hunter, Ind. Eng. Chem., 1957, 49, 174–181,  DOI:10.1021/ie50566a023.
37. P. Jackson, et al. , Anal. Chem., 2019, 91, 2577–2585,  DOI:10.1021/acs.analchem.8b04596.

---