Cynthia A. Challener is a contributing editor to Pharmaceutical Technology.

Cynthia A. Challener

Future and media concept | Image Credit: ©peshkova - stock.adobe.com

Use of process analytical technology (PAT) and advanced digital solutions such as artificial intelligence (AI) and machine learning (ML) is a crucial component of the move to Pharma 4.0. Online/inline/at-line PAT enables real-time monitoring of processes during development and manufacturing. That information can then be applied to support process control, leading to more consistent performance. When PAT is integrated with AI/ML algorithms, it is possible to develop predictive models that enable improved control, support more effective decision-making, and reduce risks, leading to optimized operations, enhanced customer experiences, and better financial outcomes, according to Edita Botonjic-Sehic, head of process analytics, data engineering and data science at ReciBioPharm.

Several PAT tools are particularly well-suited for integration with AI and ML algorithms. At the top of the list are soft sensors, which integrate a software layer based on a mathematical model that is employed to monitor process states, observes Stacy Shollenberger, senior manager, process analytical technology with MilliporeSigma.

Analysis of the results is complex, however, and requires a high level of expertise to identify, correlate, and quantify the signal and use it for predictive analysis, according to Kaschif Ahmed, principal data scientist with ReciBioPharm. “The best integration solution is one that has an extreme amount of flexibility for measurement and analysis and can easily be validated without the constant need for calibration,” he adds.

AI/ML models can be directly integrated inside the PAT tool, providing a data-driven means to estimate or predict parameters that are otherwise difficult to measure or cannot be measured directly, says Moritz von Stosch, chief innovation officer of DataHow.

Shollenberger points to spectroscopic sensors including Raman, nuclear magnetic resonance (NMR), and near-infrared (NIR) technologies, which offer near-instantaneous measurement of key metabolites and nutrients (e.g., lactate, glucose). However, the spectra generated by these techniques require interpretation using chemometric models leveraging AI/ML algorithms tailored for chemical analysis. Mass spectrometry, she notes, can also greatly benefit from AI/ML interpretation of the produced spectra.

High-performance liquid chromatography (HPLC) and ultra-high performance liquid chromatography (UHPLC), meanwhile, can also be coupled to online analyzers that rely on models and used to monitor product purity in near real-time, according to Shollenberger.

Importantly, in comparison to standard linear regression approaches, ML can increase prediction accuracy when nonlinearities arise (e.g., due to superimpositions in the spectra), says von Stosch. Fusion of data from different sensors is also possible, enabling capture of the overall state of the process better and reducing the impact of measurement noise. “By fusing sensor information with predictions coming from process models, the performance of such approaches can be extended, (e.g., in the case of low signal-to-noise ratios),” he remarks.

Depending on the monitoring scope, Shollenberger notes that hybrid models combining AI/ML with mechanistic understanding can be better suited for real-time monitoring than AI/ML models alone because the former generally better capture the complexity of bioprocesses, leading to more reliable, interpretable, and generalizable simulation results. Upstream processes in particular stand to gain the most from hybrid models, she says, as they are characterized by the high degree of unpredictability inherent in living cells.

There are applications for AI/ML outside of soft sensors as well. For example, van Stosch notes that the advent of deep learning, a specific ML method, has also made real-time-based image processing possible. “Today, ML methods are used to count cells or particles, classify different types of cells or particles, inspect vials for release, and other applications,” von Stosch comments.

PAT tools integrated with AI/ML-driven models can be used in many different applications across modalities, the drug development and manufacturing cycle, and unit operations. “Such integrated tools are suitable for all types of processes, including those that produce small molecules and biotherapeutics and for both upstream and downstream unit operations at process development and manufacturing scale,” states Botonjic-Sehic.

“Real-time operation and immediate decision-making using PAT tools and digital twins will increase throughput and reduce failures during process development and manufacturing,” adds Ahmed. For these and other reasons, Botonjic-Sehic notes that the use of PAT and digital infrastructure is supported by regulators, with some guidelines available to help the industry move in this direction.

Real-time process optimization relying on process models is one specific activity that can particularly benefit from AI/ML at the upstream level, according to Shollenberger. “Integration of AI/ML/hybrid models with PAT in process development helps to accelerate process understanding, including the impact of different critical process parameters (CPPs) on product critical quality attributes (CQAs), and facilitates the extrapolation of process knowledge across different scales,” she says.

Process development activities can greatly benefit from more continuous tracking of the process state and the ability to make decisions during or after the run, rather than having to wait for the analytics for several hours or days, agrees von Stosch. “In addition to accelerating process development, PAT tools with integrated AI/ML models can render on-the-fly process optimization possible,” he states.

Such knowledge management is also crucial for a smooth process scale-up and transfer from development to pilot and manufacturing scales, all while ensuring stable and effective performance, Shollenberger contends.

As examples, Shollenberger highlights the use of multivariate statistical process control methods within a quality-by-design approach to process development to maintain CPPs and CQAs within the identified design space and new AI/ML design-of-experiment (DoE) tools such as the Bayesian DoE to reduce experimental effort and development costs by providing experiments that are the most informative for characterizing processes. Finally, she points to the use of digital twins for predicting process performance at production scale by mirroring and extrapolating insights gained from smaller-scale experiments. “This approach allows for a more accurate representation of how processes will behave in full-scale production, ultimately streamlining process development and enhancing overall efficiency,” Shollenberger comments.

In addition to supporting the scale-up from process development to production by learning to adapt scale-dependent process parameters and optimizing their configurations, AI/ML/hybrid models supporting real-time or near real-time monitoring and control can greatly improve process efficiency, reliability, yield, and product quality, says Shollenberger. PAT and digitally driven infrastructure can, agrees Botonjic-Sehic, reduce uncertainty by providing insights into process conditions through simulators enhanced with integrated AI/ML and PAT real-time data. “Real-time release testing will also be enabled by AI/ML with PAT due to the ability to achieve rapid quality verification,” she notes.

Improved responsiveness and better decision-making

The integration of AI, ML, and hybrid models with PAT tools offers numerous benefits that extend beyond simple process monitoring. Having an infrastructure that includes novel PAT tools, ML models, and automation allows researchers to identify their targets of interest faster and more accurately during process development and manufacturing, according to Ahmed. “Having predictive models available for each step in the process enhances process understanding and ultimately allows businesses to deliver drugs to market more quickly with greater quality while also reducing waste and cost,” he states.

Increased (near) real-time insight into the evolution of processes makes several actions possible, adds von Stosch. He highlights the ability to take corrective actions either manually or via closed-loop control, make more informed decisions about when to harvest or release material for further processing, introduce changes that afford greater insights into the process, and ultimately increase process capability.

Specifically, according to Shollenberger, chemometric models can interpret spectroscopic data and predict in near real-time process variables of interest. Advanced control techniques based on these predictive models can anticipate future deviations, enabling timely interventions and process adjustment. “It is the combination of real-time deviation monitoring and process optimization that enhances process understanding and control, leading to better quality, increased speed, and greater flexibility in the manufacturing process,” she concludes.

The potential presented by integrating AI/ML with PAT technologies is limited to some degree by several challenges to successful implementation of these solutions. Complexity, connectivity, and data availability and quality are the top issues.

“The complexity of model building and maintenance, particularly in the highly regulated biopharmaceutical environment, necessitates extensive qualification and validation efforts,” says Shollenberger. Identifying the right PAT tools that are suitable for the right level of analysis, developing soft sensors specifically targeting each CQA in the process, and building in the analytics and logic to enable process control must be achieved first, according to Botonjic-Sehic. Appropriate automation solutions that support collection and use of real-time information to predict process performance and identify failures as they occur must also be developed. Then once deployed, validation, compliance and data integrity components must be well understood before a drug is processed for release, she comments.

In addition, Shollenberger notes that executing models in a real-time process control environment involves managing time latencies and ensuring seamless connectivity and full compatibility between sensors, other equipment, software, and all interfaces.

Effective training of AI/ML models, meanwhile, requires substantial volumes of experimental data to ensure reliable predictions, placing a considerable burden on operators to generate these datasets. “ML data quantity and quality requirements are rarely met today as current datasets often consist of infrequent data points, such as daily measurements as PATs are mostly offline, limiting the availability of high-frequency historical data for model training,” Shollenberger observes. Hybrid models, she says, are therefore a good alternative for reducing the data quantity required for model training.

On a positive note, Botonjic-Sehic emphasizes that once digital models are developed using historical run data, the robustness and validation is integrated into the process, resulting in faster manufacturing with better quality while decreasing the cost of operation.

The implementation of PAT solutions and AI/ML-enabled PAT solutions in the pharmaceutical industry at production scale is more prevalent in the small-molecule segment. “The chemical nature and synthesis of small molecules are relatively straightforward, and well-established manufacturing technologies dominate the production environment. In the face of intense generic competition and cost pressures, however, drug makers are increasingly seeking innovative, speedy, and cost-efficient production approaches, and AI/ML-enabled PAT solutions have attracted significant interest,” Shollenberger explains.

Biotherapeutics producers are moving toward the use of integrated AI/ML/PAT, but at a slower pace due to the greater complexity of biomolecules, according to Ahmed. While there is significant potential for AI to enhance process understanding and control, enabling informed decisions in the future, today human involvement remains essential to address the current limitations of AI, Shollenberger adds. “Human expertise is still necessary to interpret analytical results accurately and to develop robust models that will leverage AI support in the future,” she says. In fact, she believes collaboration between human insight and AI capabilities will be crucial for advancing biomolecule production.

One example already under development is ReciBioPharm’s novel process for RNA-based vaccines and therapeutics, which it is developing in collaboration with the Massachusetts Institute of Technology using grant money from FDA. The continuous GMP process is extensively enabled by real-time analytical technologies leveraging AI/ML models, including new fit-for-purpose solutions. It is expected to be operational in mid-2025, according to Aaron Cowley, CSO, ReciBioPharm.

DataHow, meanwhile, released a spectral data analytics module within DataHowLab, its advanced process data analytics solution that deploys AI/ML with hybrid models (1).

Integration of AI/ML within PAT tools will, according to Ahmed, eventually make real-time data analysis more autonomous and ultimately support real-time release and thus a paradigm shift in pharma manufacturing.

Indeed, Shollenberger believes significant advances in PAT capabilities will be achieved in the coming years, with integration of AI and ML representing a crucial next step toward the envisioned “facility of the future”.

In the short term, she expects the push for cost reduction and environmental sustainability to drive the miniaturization and cross-unit factorization of automated PAT devices, making these advanced PAT solutions more accessible and cost-effective. In addition, use of calibration-free or limited-calibration PATs could help to streamline the integration of these technologies into existing processes with lower cost.

Other advances will occur in knowledge management, as it is vital for collecting and structuring knowledge coming from different devices and PATs. “Scale agnostic, cross-process, cross cell line, and cross-proteins knowledge management is essential for enabling the application of AI and ML across various processes and steps,” Shollenberger states.

Once these advances are achieved, it will be possible to deploy integrated AI and ML solutions that facilitate near real-time quality assurance, process optimization, and continuous improvement. “These advances will help ensure that final products consistently meet regulatory and quality standards, enhancing the efficiency and reliability of pharmaceutical manufacturing,” observes Shollenberger. She cautions, however, that integration of AI and ML models into GMP processes introduces regulatory challenges related to data security, algorithm transparency, and the validation of AI-driven processes against existing industry standards, all of which must be addressed to ensure their effective use while maintaining compliance with regulatory requirements.

Cynthia A. Challener, PhD, is a contributing editor to 

®
Vol. 48, No. 12
December 2024
Pages: 16–18

When referring to this article, please cite it as Challener, C.A. Integrating Advanced Technologies for Pharma Analysis. 

Articles

Automation technologies provide real potential to improve responsiveness and decision-making in drug development.

Integrating Advanced Technologies for Pharma Analysis

New chemical combination capsule drug on the colorful hand drawing chemical structure formula | Image Credit: ©HYUNGKEUN - stock.adobe.com

High-throughput screening (HTS) enables the rapid testing of numerous compounds against biological targets. By automating and miniaturizing assays, HTS significantly accelerates the identification of potential drug candidates or optimal process conditions and can be applied throughout the entire development path of a potential drug, from discovery to process development. In the discovery phase, HTS is one of several hit-identification (Hit-ID) paradigms and involves the use of sensitive and miniaturized methods for screening large libraries of small molecules in a practical and efficient manner. During process development, HTS helps researchers rapidly establish an optimal synthetic route and conditions for each process step that reliably produce sufficient material quickly.

The most widely used HTS technologies for identification of small-molecule API candidates include fluorescence-based, luminescence-based, and absorbance-based assays, according to Sachin Mahale, research leader, Discovery Sciences at Charles River. He adds that methods for detection of unlabeled biomolecules, such as those based on mass spectrometry, have also become more widely used at an HTS-scale in recent years, allowing the screening of compounds in both biochemical and cellular settings and thus expanding the breadth of targets to those that would otherwise be unscreenable. High-content imaging and automated electrophysiology are additional techniques providing multi-parametric cellular data that are becoming more widespread in high-throughput formats for discovery research, Mahale observes.

“Considering the vast number of technologies, each one of them serving a different purpose to address the scientific question, naturally different attributes, including binding affinities, reaction kinetics, enzymatic activity, functional activity, target engagement, selectivity, and specificity, are often analyzed. These attributes are critical because they provide early insights into the potential efficacy and selectivity of drug candidates, which is essential for optimizing pharmacological profiles,” says Mahale. Given the wide range of technologies available, he also emphasizes the importance of selecting a fit-for-purpose technology to address each biological question.

Numerous advantages for process development

During the early stages of candidate development, the synthetic pathway is still flexible, making HTS particularly valuable, according to Jens Schmidt, associate director of manufacturing, science, and technology at Lonza. “HTS can help process development scientists rapidly evaluate the feasibility of different synthetic routes or optimal chemical combinations, such as solvents, catalysts, and bases,” he explains. As the project advances through clinical phases, the flexibility in process steps decreases, but HTS remains useful for expanding the understanding of the process, such as its reaction kinetics.

Overall, using HTS during process development can, contends Schmidt, “help in the optimization and simplification of established workflows, integration of emerging technologies such as photochemistry and electrochemistry, and leveraging of rapidly evolving computational tools like artificial intelligence (AI) and machine learning (ML).”

One of the key advances in HTS for drug discovery and development is the adoption of a wider array of state-of-the-art analytical techniques for this application. The most notable technologies introduced recently as high-throughput platforms include affinity selection mass spectrometry (ASMS)-based screening platforms such as self-assembled monolayer desorption ionization (SAMDI), target protein degradation (TPD), and molecular glue platforms; and CRISPR (clustered regularly interspaced short palindromic repeats)-based functional screening platforms, notes Luke Alderwick, senior group leader, Discovery Sciences with Charles River.

ASMS screening is increasingly used to discover small molecules that engage a specific target, says Alderwick. “Highly advanced ASMS platforms that use rapid workflows to quickly generate high-quality data are getting traction as HTS methods. The platform is amenable to a broad spectrum of targets, including proteins, complexes, and oligonucleotides such as RNA, and can be a leading assay to initiate drug discovery programs,” he comments. Unlike traditional label-free approaches, which are often slow, restrictive, and not optimal for large screening campaigns, Alderwick adds that new solutions such as Charles River’s SAMDI ASMS platform allow efficient investigation of protein-protein interactions, TPD, molecular glues, RNA binders, etc. in a high-throughput format.

CRISPR-based functional screening, meanwhile, is employed to elucidate the biological pathways involved in disease processes, according to Alderwick. CRISPR editing of genes for knock-out and knock-in experiments have greatly advanced assay designs for cell-based phenotypic HTS approaches, he explains. In addition, by selectively tagging proteins of interest, CRISPR is advancing the understanding of target engagement and functional effects of drug treatments. “By screening small molecules alongside CRISPR-modified cells, researchers can better understand drug-target interactions at a genomic level, enabling the selection of candidates with higher precision,” says Alderwick.

Overall, Alderwick summarizes, these various innovations have improved the scalability, specificity, and throughput of HTS, enabling researchers to screen more diverse compound libraries and discover novel molecular mechanisms of action.

Automation and miniaturization technologies having an impact

Some of the biggest advances improving and expanding HTS for drug discovery and development relate to automation and miniaturization of laboratory systems. “Automation is transforming the landscape of API development and manufacturing,” Schmidt states.

Automated liquid-handling robots, says Mahale, have become faster, more accurate, and capable of working with extremely low volumes of reagents, enabling more cost-effective screening. “Advances in pipetting precision, multi-plate handling, and integration with other systems (like incubators, centrifuges, and imagers) have significantly reduced human error and improved reproducibility. Automated workflows for complex cell assays and readouts based on imaging analysis, meanwhile, allow multiple assays to be performed with minimal consumable cost and keeping sustainability in mind. Use of multimode plate readers in the automation workflow also allows for multiplex assays to maximize data collection,” he details.

By automating process development tasks, researchers can significantly accelerate the identification of optimal synthetic pathways and fine-tune processes for maximum efficiency, according to Schmidt. “This translates to faster development times, reduced errors, and more consistent API production. For drug developers, this means reduced risk and uncertainty in API development, enabling a greater focus on clinical trials and regulatory approvals,” he contends.

More specifically, automation accelerates workflows by allowing researchers to complete multiple tasks at once. “This kind of parallelization is especially helpful for processes like catalyst screening, ligand screening, or solvent selection. Automation also allows chemists to conduct experiments in high-pressure and high-temperature conditions, which are difficult to complete manually. With these features, automation makes experimentation in the lab more targeted and efficient, saving valuable time, energy, and cost,” Schmidt notes.

In addition to advances in automation, innovation in miniaturization technologies is helping to improve HTS performance. “By shrinking assay volumes, miniaturized platforms like microfluidics and nanodispensing have allowed for higher throughput while reducing reagent consumption,” Mahale remarks. Access to smaller-volume assays, he says, is especially useful in fragment-based drug discovery and the screening of complex biological samples, as it allows for cost-effective experiments with even larger libraries.

Data analysis and sharing solutions making a difference

In addition to innovation in the analytical methods and laboratory equipment used for HTS, significant advances have also occurred in the software arena, most notably the integration of AI and ML for data analysis and improvements in cloud-based platforms for collaborative research.

New AI algorithms can, observes Schmidt, analyze data from high-content screening systems, detecting complex patterns and trends that would otherwise be challenging for humans to identify. AI/ML algorithms can identify patterns and predict the activity of small-molecule candidates, even when the data are noisy or incomplete, Mahale adds. In addition to significantly speeding up the analysis of screening results and identifying promising leads faster and with higher accuracy, integration of AI into HTS processes can also improve assay optimization, he comments. Another key advantage of AI-driven HTS, according to Mahale, is its ability to adapt to new data in real time when compared to traditional HTS relying on pre-determined conditions.

“These advances in AI and ML have been pivotal in accelerating drug discovery by improving both the quality of data and the efficiency of experimental processes. In the early stages of development, meanwhile, using these advanced solutions provides a holistic view that streamlines route scouting and accelerates the transition from research and development to robust manufacturing processes,” Schmidt says. As an example, he points to Lonza’s AI Route Scouting Service, which integrates proprietary commercial data with cutting-edge synthesis planning technology. “This innovation allows for more accurate and comprehensive analysis of cellular phenotypes, providing deeper insights into the efficacy and safety of drug candidates,” he notes.

Cloud technology, meanwhile, has enhanced the ability to store, share, and analyze HTS data across multiple sites in real time, says Mahale. “The result is improved collaboration between research teams globally, reduced data redundancy, and increased application of machine learning models on large datasets,” he explains.

Combined, Mahale believes the various software innovations incorporated into HTS systems are enabling the handling of more complex datasets, improving prediction accuracy and streamlining the drug discovery pipeline.

Implementing HTS for drug discovery and development is not a simple matter. The systems are both complex and costly. Fortunately, minimizing these issues has also been a focus of technology providers and end users.

Schmidt points to significant advancements in standardization and simplification of workflows as helping to accelerate HTE experimentation timelines and reduce costs. Examples include the development of standardized protocols and data formats, which has simplified the implementation and operation of HTE assays, specifically improving their accessibility and accuracy. He also notes that software tools powered by AI have made it easier for researchers to design, execute, and analyze HTE experiments.

Another effective approach, according to Alderwick, is to leverage tailored compound libraries such as Charles River’s Lead-Like Compound Library, which include compounds with lead-like properties and diversity and excludes problem chemotypes. The company’s Fast-Track HTS offerings leverages this library to provide rapid hit discovery and lead identification while maintaining high-quality data outputs, he adds. “With a fixed-cost and streamlined program, this service is ideal for proof-of-concept studies and suits research groups or companies with limited funding because it allows for high-quality, early-stage screening without the need for extensive internal resources or infrastructure,” Alderwick says.

Alderwick also notes that modular systems in development offer the potential to customize HTS systems to the specific needs of a lab, which will further lower the initial capital investment required. “All of these advances are making HTS more cost-effective, scalable, and user-friendly, reducing barriers to entry for smaller research organizations,” he concludes.

Despite the many remarkable advances made in HTS technologies for drug discovery and process development, significant challenges impacting both the efficiency of identifying optimal small-molecule candidates and processes, and their downstream success in clinical trials, remain.

One important issue, according to Schmidt, is the need to analyze large HTE datasets that require significant computational resources. “This can be a bottleneck for many drug developers, which is why partnering with a contract development and manufacturing organization (CDMO) like Lonza can be helpful. CDMOs have the tools and technology, such as AI algorithms, to help drug developers analyze data and progress through development stages quickly,” he says.

There are also challenges to handling very small amounts of solids, Schmidt notes. “Although this may seem like a minor technical detail, understanding and accepting these limitations is crucial,” he says. One potential solution highlighted by Schmidt is the use of solid supports such as ChemBeads (MilliporeSigma), small, chemical-coated glass beads created specifically for use in HTS applications.

A primary challenge in HTS for drug discovery is identifying assays that are physiologically relevant to the disease state despite the need for simplification to ensure scalability and robustness, according to Mahale. Model systems do not always translate well. In addition, biochemical assays, while simpler, often fail to capture the nuances of cellular contexts, such as the impact of membrane permeability, metabolite production, or protein-protein interactions, while cell-based assays are closer to physiological reality but are more complex, expensive, and challenging to miniaturize.

“The complexity of human diseases, particularly multifactorial diseases like cancer or neurodegenerative disorders, makes it difficult to develop assays that can recapitulate the true pathophysiology of the disease,” Alderwick observes. “Poor assay choice,” he continues, “leads to the identification of hits that might show promise in early testing but later fail due to lack of efficacy or off-target effects in more complex disease models or clinical settings. The incorrect assay not only risks identifying false positives but also misses out on potential therapeutic candidates that could have better real-world efficacy. This limitation significantly contributes to the high attrition rates in drug development.”

Another challenge is the development of a robust screening cascade (primary, secondary, and tertiary screenings) that helps to identify, confirm, validate, and refine viable candidates and reduce the attrition rate during later stages of drug development. Further assays are then employed during lead optimization to assess toxicity, selectivity, functional activity, bioavailability, and pharmacokinetics for further hit validation. “Failures in designing an appropriate HTS cascade occur when these different stages are not sufficiently de-risked,” contends Alderwick.

While incorporating more predictive absorption, distribution, metabolism, excretion, and toxicity screens early in the process has helped de-risk candidate selection to some degree, the challenge still lies in balancing the need for throughput and speed with the inclusion of more complex, multi-parametric data, according to Alderwick. “Early prediction of 

 efficacy and safety in HTS remains a weak point, contributing to high failure rates in the transition from hit-to-lead and lead-to-clinical candidate stages,” he concludes.

Finally, Mahale notes that the selection of chemically diverse and high-quality libraries for HTS is pivotal in the success of small-molecule candidate identification, because even with the best assays, if the chemical library lacks diversity or contains unsuitable and non-drug-like compounds, the likelihood of finding optimal candidates diminishes. “Many historical compound libraries are biased toward certain chemical scaffolds that have been repeatedly screened over decades, thus reducing the chances of discovering novel small molecules with new mechanisms of action,” he observes. Charles River, Mahale adds, has tackled this issue by developing an expertly curated small-molecule compound library that covers a wide range of structural and functional diversity, thereby maximizing the probability of finding useful hits.

Several emerging strategies aim to improve the efficiency, reliability, and outcomes of HTS by leveraging advancements in technology, biology, and chemistry, according to Alderwick. Advances in organoid and 3D-cell culture technologies are helping to bridge the gap between simple 

 assays and complex human disease biology, and thus address the need for more physiologically relevant assays, he observes. “These models better mimic the physiological conditions of tissues and organs, providing a more relevant system for screening potential drug candidates,” Alderwick explains. Organoids derived from patient tissues, for example, can offer insights into how drugs might work in specific disease contexts, improving the predictive power of assays.

To shrink the gap between biochemical, cell-free assays and phenotypic, cell-based screens and thus better understand the on-target activity of identified hit molecules, including binding assays such as for target engagement and protein-protein interaction analysis in the HTS screening cascade, would greatly advance the ability to select optimal compounds for further development, believes Alderwick. He also notes that the use of advanced cellular models, such as CRISPR/Cas (

-associated protein) knockouts and induced pluripotent stem cells, for early screening is likely to improve HTS outcomes.

Continued advances in automation and the use of AI and ML will further enable HTS platforms to handle more complex assays with higher precision, reproducibility, and throughput, according to Mahale. “Fully automated HTS systems can reduce human error, improve the consistency of assay conditions, and allow for more sophisticated workflows that balance throughput with data quality, while incorporation of AI/ML algorithms can reduce the rates of false readouts, making HTS more reliable and reducing the time spent following up on false positives,” he says.

Ultimately, concludes Mahale, these emerging solutions aim to reduce the high attrition rates in drug discovery by improving the relevance and quality of the screening process, optimizing candidate selection, and expanding the chemical diversity available for exploration. “While challenges remain, these developments promise to significantly enhance the utility of HTS in the drug development pipeline,” he contends.

Looking beyond these strategies, Alderwick is excited about the trend toward integration of multi-omics approaches into HTS, which he says represent a major leap forward in the ability to identify optimal small-molecule API candidates. “Multi-omics approaches leverage data from various molecular levels, including genomics, proteomics, transcriptomics, metabolomics, and even epigenomics, providing a more comprehensive and holistic understanding of how small molecules interact with entire biological networks and pathways rather than isolated targets,” he explains. “Complex diseases such as cancer, neurodegenerative disorders, and metabolic syndromes often involve multiple, interconnected biological systems, and thus multi-omics approaches have a greater likelihood of identifying effective new drug candidates,” he states.

In today’s fast-paced drug manufacturing environment, HTS has become an indispensable tool. However, Schmidt emphasizes that advances in technology are not sufficient by themselves to enhance the performance of these platforms. “Looking ahead, we can expect continued improvements in both reliability and simplicity, but the key to success lies in having a motivated, dedicated, and technically skilled team that is ready to embrace challenges and open to adopting innovative techniques,” he insists. 

Cynthia A. Challener, PhD, is a contributing editor for 


Vol. 48, No. 11
November 2024
Pages: 12–16

When referring to this article, please cite it as Challener, C. Accelerating Discovery and Development with Advances in High-Throughput Screening. 

Automation, miniaturization, and new software algorithms are improving throughput and accuracy.

Accelerating Discovery and Development with Advances in High-Throughput Screening

pills recycle | Image credit: ©Zafirah - stock.adobe.com

Improving the sustainability of small-molecule API manufacturing is essential to ensuring the continued success of the pharmaceutical industry. For most APIs, however, evaluating the global warming potential along the whole value chain or lifecycle is difficult. The greatest contributor to the carbon footprint of API manufacturing operations is the use of solvents, reagents, and precious metal catalysts, for which significant restrictions on recycling and reuse can exist. The major argument against reuse and recycling is based on product and application safety, according to Michael Nonnenmacher, director global business development differentiating technologies with Evonik.

The ability to recycle and/or reuse these materials, however, enables measurable resource savings and carbon footprint reduction. “For multi-step syntheses, it is likely that hundreds of kilos of raw materials are used to produce a single kilogram of final product. It is still common practice to incinerate most of the waste generated from these raw materials, oftentimes resulting in tons of carbon dioxide (CO2) per kilo product. The possibility to recover some of these materials or replace them with greener alternatives would enable a tremendous reduction of the process carbon footprint and improve the overall sustainability of these operations,” Nonnenmacher states.

Solvent and catalyst recycling offers several significant advantages. In contrast to incineration or other forms of disposal, which generate substantial amounts of CO

, catalyst and solvent recycling allow repurposing in the same process or in different applications with minimal energy expenditure, according to Niklaus Künzle, senior director, global head of process technology and innovation for small molecules at Lonza. In addition to the reduced direct CO

 emissions, this approach also necessitates less production of raw materials, resulting in further substantial savings in both energy and CO

Solvent recovery is the most practiced recycling solution in established large-scale processes. “For API manufacturing, the use of recycled solvent is mostly restricted to the same process and a limited number of cycles, and a thorough risk assessment to mitigate potential impurity formation and accumulation in the process is needed,” Nonnenmacher says. “Since the discovery of nitrosamine impurities, which are potent carcinogens, stemming from solvent degradation, recovery is under a lot of scrutiny, but it should still be considered in cases where it will result in no process risk,” he adds. In cases where it is not appropriate to use recycled solvent in the API production process, Nonnenmacher observes that it is possible to find other fine-chemical applications where they are suitable.

Recycling of expensive palladium, platinum, rhodium, and other precious metal catalysts helps not only to reduce costs, but the carbon footprint associated with the mining and refining operations necessary to obtain the metals, notes Nonnenmacher. While solvent recycling is often performed in-house, the transformation of used catalysts back into active species is typically performed externally by specialized companies, according to Künzle.

In addition, although filtration of heterogeneous catalysts from process wastes is typically easy, separation of dissolved homogeneous catalysts from oftentimes dilute solutions requires more specialized technologies, comments Nonnenmacher. “Evonik has established an efficient process for the concentration of precious metal-containing organic waste streams using membrane cross filtration. This approach enables the sustainable recovery of the metal at the refinery, and thus contributes to resource saving and carbon footprint reduction,” he says.

The best strategy for realizing the greatest benefits from recycling during API manufacturing is to, according to Nonnenmacher, apply the principles of green chemistry early on during process development. “In this manner, it is possible to implement eco-friendly and more sustainable raw material and API manufacturing processes,” he notes. Key considerations highlighted by Nonnenmacher include avoiding the use of large volumes of especially hazardous organic solvents that cannot be recycled, applying process intensification approaches, and controlling the atom economy for waste reduction.

“Beyond these approaches, the introduction of recycling concepts or reuse options into processes right from the beginning can facilitate acceptance by regulatory authorities,” Nonnenmacher contends. Performance of process and product life-cycle analyses to generate important sustainability metrics such as process mass intensity and E-factor values when a candidate is in clinical development can, he says, help pharmaceutical firms and API makers visualize and understand potential adverse steps in the synthesis early enough, allowing manufacturers to implement more sustainable solutions.

Of course, one of the best approaches to improving sustainability is to avoid the use of problematic processes and materials. “Ideally, the most elegant solution is to minimize or completely eliminate the use of solvents,” Künzle states. Lonza therefore strives to implement process improvements in collaboration with drug developers during the early development phase. “These process improvements aim to reduce the number of used solvents, decrease dilution, and enhance the process yield. Selection of process solvents is also optimized by considering solvents and mixtures that are feasible to recover from process waste streams,” he explains.

Recycling compounds from a complex mixture requires efficient separation techniques, such as rectification and membrane separation. Continuous liquid-liquid or liquid-solid separation may also be employed, according to Nonnenmacher. The adsorption to scavengers or filter aids can be efficient for catalysts as well. Immobilizing catalysts such as enzymes by attaching them to a solid substrate also facilitates recovery and reuse, he says.

For solvents specifically, distillation, extraction, adsorption, and membrane separation techniques, according to Künzle. “Often,” he notes, “multiple technologies must be combined to ensure the purity required for recycling these materials back into API processes.”

Nonnenmacher also emphasizes that both increasing cost constraints and sustainability ambitions are driving the development of more efficient processing techniques for separation, as well as more selective processes that result in waste mixtures that are less difficult to separate.

Recycling, whether of solvents or catalysts, is often not an easy task given the complex solutions used in API manufacturing. “Extracting valuable materials from complex process mixtures can be a significant challenge, especially when aiming to restore them to a high-purity starting quality with minimal energy consumption and at minimal cost,” says Künzle.

Another major challenge for implementing recycling solutions at both clinical and commercial scale is that recycling is usually not attractive enough for small campaigns, according to Nonnenmacher. “It is considered a given that waste is sent to incineration and, therefore, oftentimes from a cost and effort perspective, recycling strategies are not a priority,” he adds. In addition, especially for smaller API manufacturing sites that are not fully self-sustained, there are significant logistical constraints given that recycling requires significant site infrastructure. “Even if waste streams are shipped offsite for recovery, a minimum buffer capacity separate from other wastes needs to be available,” Nonnenmacher observes.

Another factor that impacts recycling opportunities is regulatory considerations. The recycling of solvents in API manufacturing may come with potential impurity accumulation and is therefore scrutinized by regulating authorities, especially when considering environmental- and health-critical solvents, according to Nonnenmacher.

For example, Nonnenmacher points to the decomposition of dimethyl formamide (DMF) to dimethylamine and the potential for subsequent formation and accumulation of the carcinogen N-nitrosodimethylamine, which makes DMF recycling almost prohibitive.

When developing a solvent recycling process, notes Künzle, it is therefore crucial to ensure that no by-products accumulate in the solvents, which involves extensive laboratory and pilot testing and continuous analysis. In addition, whenever recycling of less critical solvents is possible from a risk perspective, Nonnemacher once again emphasizes the importance of establishing any recycling concepts during early development and into validation and using a defined, limited number of cycles before final disposal. “As part of the process, the recycling concept needs to be filed and accepted by the authorities,” he notes.

A risk-based approach to recycling is also required for materials other than solvents. “In most cases, waste streams from API manufacturing must be incinerated due to the potential risk of pharmaceuticals or other contaminants leaking into the environment,” observes Nonnenmacher. However, he does comment that recycling of metals that in principle undergo complete reset to their original state is handled outside of the process and not questioned by regulatory agencies.

Regardless of the material or technology, the best strategy is to integrate recycling considerations into regulatory filing activities from the very beginning, Künzle concludes.

Recognition of the importance of increasing the sustainability of pharmaceutical manufacturing operations is driving investment in many new technologies and solutions, including those for recycling. “Current and future innovation in many fields will contribute to the sustainability of API manufacturing and recycling of process components. Those innovations may come from established fields like from physical separations, or from research areas less obviously related, such as more in-depth process development supported by artificial intelligence resulting in more efficient production methods and cleaner off-streams, which would be easier to recycle,” Nonnenmacher comments.

Lonza, for instance, is working to enhance its entire infrastructure to enable the realization of additional recycling solutions, according to Künzle. One example is the expansion of pervaporation capacities in Switzerland, which allow the company to process challenging mixtures with minimal energy expenditure. Lonza is also applying new oxidation and enzymatic technologies for wastewater treatment that involve significantly reduced energy consumption.

Evonik, meanwhile, is leveraging fundamentally different approaches to how pharmaceutical chemistry is performed. One example is the Chemistry in Water platform the company is developing and using for the manufacture of active ingredients. “This technology allows significant reduction of organic solvent consumption, oftentimes combined with offering superior reaction performance, such as in the field of cross-coupling reactions,” Nonnenmacher says.

®
Vol. 48, No. 10
October 2024
Pages: 18–20, 33

When referring to this article, please cite it as Challener, C.A. Improving the Sustainability of API Manufacturing with Recycling Technologies. 

Sustainability of small-molecule API manufacturing ensures continued success.

Improving the Sustainability of API Manufacturing with Recycling Technologies

Antibody colored by heavy and light chains against black background; glycosylated immunoglobulin in pink | Image Credit: ©Eric H Cheung-Stock.Adobe.com

Glycan modifications of recombinant proteins and antibodies can impact their function and stability and thus the safety and efficacy of drug products based on these molecules. Understanding these important critical quality attributes (CQAs) is essential, and therefore, regulatory bodies expect a comprehensive understanding of glycans for submission.

This article was published in the October 2024 print issue of 

Comprehension necessitates overcoming inherent challenges such as positional isomers and heterogeneity, as well as method-induced artifacts that influence sensitivity and specificity, according to Paula Orens, senior global market development manager, Biopharma, at SCIEX.

Multispecifics, including bi-, tri-, and tetraspecific antibodies, present additional challenges to glycan analysis beyond those of traditional monoclonal antibodies (mAbs) because the complexity of these molecules is much greater.

“Site-specific and intact mass analysis may become more challenging for increasingly complex molecules, particularly when the number of N-glycosylation sites, post translational modifications, and potential degradation products increases,” observes Oscar Potter, biological chemistries R&D manager in the Chemistries and Supplies Division at Agilent.

Structurally, multispecific antibodies are a combination of different antibody domains that allow the multispecific to bind at multiple sites. “The various domains will have their own glycosylation patterns (integral to efficacy, safety, and stability) and their own impurities, all of which need to be characterized comprehensively, effectively multiplying the glycosylation analysis complexity by the number of binding sites,” Orens explains. She adds that as the number of binding sites increases, so too does the risk of intra-protein cross talk.

Multispecifics can also include candidates in which an antibody-portion is combined with some other protein-based component. Chimeric proteins, where the domains are not only antibody in origin, are complicated to analyze due to the individual properties of each component, according to Orens. “For example,” she says, “fusion proteins are typically heavily glycosylated with N- and O-linked glycans to improve biological activity, properly stabilize the structure, or improve solubility.”

“Overall,” Orens concludes, “while the complexity of analysis is dependent on the types of domains used in a multispecific, it is certain that the analysis of these molecules will be more complicated than that of a mAb.”

Consequences of incorrect glycosylation analysis

The greater complexity of multispecifics raises the stakes for achieving correct glycan analysis. “Not only can the glycosylation patterns in multispecifics impact the stability, pharmacokinetics, off-target effects, and effector functions such as complement activation and antibody-dependent cellular cytotoxicity, conceivably the additional functions that multispecific antibodies are capable of might result in unique biological consequences from incorrect glycosylation,” Potter comments.

In addition, because glycosylation is a critical and highly regulated quality attribute of all antibodies, misunderstanding or incorrectly characterizing glycosylation patterns could lead to a variety of consequences, depending on the development stage of the candidate, notes Orens. She highlights a couple of examples. “At the hit-identification stage, a mis-characterized but suitable candidate could be eliminated due to poor efficacy or stability. At later development stages, meanwhile, a drug metabolism and pharmacokinetics study could exhibit faster-than-expected clearance or undesired concentration in the target tissue. An incorrect analysis could also result in promotion of a misunderstood candidate through to late-stage toxicology studies, causing unnecessary investment in time and resources for the project only to be shut down later than expected.”

There are several different techniques used for the glycan analysis of antibody-based candidates. “Fundamentally, the same glycan analysis approaches are used for multispecific antibodies as mAbs, most commonly liquid chromatography (LC), capillary electrophoresis (CE), and mass spectrometry (MS). Each analytical approach offers a plethora of techniques which provide complementary or alternative details about the glycosylation profiles,” Orens observes. She does note, though, that the inherent complexity of the molecule will impart different challenges on each approach.

Typical analytical methods for glycosylation analysis of multispecifics are similar to those used for more traditional therapeutic mAbs, agrees Potter. “Released glycan analysis methods generate a glycan profile, usually based on separation by CE or LC. Intact LC/MS analysis can provide limited information about the major glycoforms, while LC/MS of protease-digested samples can give site-specific information for molecules with more than one glycosylation site,” he outlines.

Released glycan analysis relies on high-resolution separation of fluorescently labeled glycans by hydrophilic interaction liquid chromatography (HILIC) or CE. HILIC is used in conjunction with fluorescence or mass detection to separate and quantify individual glycan species over long gradients, according to Orens. Similarly, CE systems can offer comparable quantitation details with an alternative separation mode, often with faster separation. “These approaches are often used as monitoring techniques to ensure batch consistency during manufacturing,” she says.

Released glycan analysis once suffered from tedious sample prep, according to Potter, but protocols have been greatly simplified in recent years through rapid-release and instant fluorescent labeling sample prep kits. He observes that HILIC methods enable the most exact assignment of glycan structures, either using retention time reference standards or retention libraries, or by hyphenating with MS detection. “HILIC separations can often distinguish between isomeric structures, including potentially harmful alpha-galactose (α-Gal) structures masquerading as beneficial galactosylated complex glycans,” he says.

Measuring the intact mass of multispecific antibodies via LC/MS does not require any sample prep, which can be advantageous. A well-optimized method run on a high-resolution instrument such as a time-of-flight (TOF)-based system is needed, however, Potter observes.

Orens adds that TOF systems can be used as the detection method for several techniques to help identify and localize glycans, as well as quantify, resolve isomers, and structurally elucidate the different species.

Potter cautions, though, that while very convenient, measuring the intact mass may only reveal the sum of masses of the two or more glycans on each molecule, which may confound assignment of exact glycan structures. Distinguishing between isomers is also not possible.

The most information can be obtained by first digesting the multispecific with a protease such as IdeS (immunoglobulin G-degrading enzyme of 

) or trypsin into smaller peptides, according to Potter. “Digestion allows for site-specific glycosylation analysis by LC/MS, which may be particularly useful when distinguishing between glycosylation in the fragment antigen-binding (Fab) and fragment crystallizable (Fc) regions, the latter of which greatly impacts the Fc effector functions of the antibody,” Potter contends.

Orens further notes that when peptide mapping is performed using a MS system with specialized fragmentation capabilities such as electron activated dissociation (EAD), it is possible to identify unknown modifications or impurities, then localize the species to a specific domain or amino acid of the protein. “EAD,” she says, “is integral to this process, as it differentiates between isomers and maps specific amino acids, and thus provides a comprehensive understanding of the protein with its modifications.”

Ultimately, Orens emphasizes, the appropriate analytical approach is entirely dependent on the purpose of the analysis.

Given the importance of accurate glycan analysis for multispecifics, the best strategy for ensuring success according to both Orens and Potter is to use a combination of available methods.

“While using top down, sub-unit, or peptide mapping methods and state-of-the-art MS systems enables comprehensive glycan analysis of multispecifics since it provides the most information in a single step, including molecular weight, domain/site localization, species identity, and species quantity, most organizations employ multiple analytical approaches to support their characterization efforts and obtain secondary, confirmatory evidence,” says Orens.

“The best way to ensure accurate and comprehensive characterization,” agrees Potter, “is to use more than one orthogonal method. Released glycan analysis, intact mass LC/MS, and LC/MS after protease digestion needn’t be mutually exclusive—they can be complementary,” he concludes.


October 2024
Vol. 36, No. 9
Pages: 28-30

When referring to this article, please cite it as Challener, C.A. Glycosylation Analysis for Multispecifics. 

The best strategy is to use a combination of complementary methods.

Glycosylation Analysis for Multispecifics

The potential benefits afforded by molecular modeling when applied to formulation development are, in fact, numerous. Molecular modeling tools, while initially applied to enhance drug-discovery efforts, are increasingly used to inform other aspects of drug development. Formulation development, which traditionally has been performed using physical experimentation in a time-consuming trial-and-error approach, is one area that can particularly benefit from 

“While the use of molecular modeling tools has historically been used extensively in the drug discovery setting, application to formulation development has been slowly evolving,” says Sanjay Konagurthu, senior director of science and innovation in the pharma services business of Thermo Fisher Scientific. Molecular modeling, he observes, can accelerate many aspects of formulation development.

“Compared with trial-and-error approaches, the application and intentional use of the scientific method is a superior approach to formulation development,” notes Nathan Bennette, director and scientific advisor, pharmaceutical development at Catalent. “In this context,” he contends, “molecular modeling tools are extremely useful to increase the efficiency of the scientific process, from problem-statement definition to experiment design and through the development of a formulation strategy.”

For instance, to tackle the challenge of poorly soluble compounds for oral drug delivery, Thermo Fisher Scientific has developed artificial intelligence/machine learning (AI/ML) models specifically to identify the most effective combinations of solubility-enhancement technologies and formulations in early drug development, according to Konagurthu.

Molecular modeling can play a crucial role in various aspects of formulation development through its ability to offer valuable insights on molecular behaviors and its predictive capabilities. “One of the primary applications is the visualization and understanding of physical and molecular processes which yields deeper intuition of the underlying mechanisms,” says Bennette. “This mechanistic understanding allows for the generation of rational hypotheses and more informed decision-making,” he adds.

Molecular modeling tools also enable formulators to explore and evaluate various formulation options, according to Bennette. “Such models allow scientists to simulate scenarios, analyze the performance of different formulations, and predict outcomes, which aids in the selection and optimization of formulations against the target product profile and critical performance attributes,” he explains.

Several different modeling tools can be applied during formulation development, from molecular mechanics to advanced quantum mechanical (QM) calculations. Examples highlighted by Konagurthu include density functional theory (DFT), molecular dynamics (MD) simulations, quantitative structure activity relationship (QSAR) models, and statistical techniques. These tools are leveraged within Thermo Fisher Scientific’s AI/ML models to aid in formulation development, including the company’s proprietary Quadrant 2 platform for addressing formulation challenges associated with solubility and bioavailability.

Bennette points to physiologically- based biopharmaceutics modeling (PBBM) for understanding absorption and the Flory-Huggins and newer, more sophisticated, perturbed chain statistical associating fluid (PC-SAFT) models for estimating the miscibility of solid solutions and phase boundaries in dynamic systems.

Once the rate limiting step for absorption of a drug substance for an oral formulation is understood using PBBM modeling, these parameters can be used to estimate the impacts of different formulation approaches, such as the use of bioavailability-enhancing technology, before any laboratory testing is needed, says Bennette. “If, for example, the rate limiting step is identified as low solubility, then an amorphous solid dispersion (ASD) might be selected as an enabling technology to improve bioavailability,” he notes.

The Flory-Huggins or PC-SAFT models can then be incorporated to identify formulations with suitable physical and chemical stability and provide insight into manufacturing conditions, according to Bennette. Understanding the miscibility of solid solutions is key to successful development of stable and metastable ASD formulations. For formulations in which the composition and thermal history change throughout the manufacturing process, the PC-SAFT model is of particular use.

Once a formulation is selected, Bennette comments that modeling tools such as FreeThink’s ASAPprime software can be used to predict the shelf-life of the product, including the impact of packaging, through its application of fundamental Arrhenius rate laws.

“Such a comprehensive modeling approach to formulation development ensures a higher probability of clinical and eventual commercial product success,” Bennette states.

Gaining a fundamental understanding of drug–drug and drug–excipient interactions is critical to developing stable formulations that can perform and do not pose manufacturing challenges. Modeling tools are beginning to be used to understand chemical stability and excipient compatibility, according to Konagurthu. “The use of these tools for both small and large molecules has been evolving and increasing with the advent of high-performance computing,” he notes.

A key current challenge to using modeling for predicting drug–drug and drug–excipient interactions is the need to ensure the models are validated using real-world, experimental data. Often, however, only limited and sparse data sets are available given the relatively new use of modeling in this application, observes Konagurthu. “Lack of access to sufficient, high-quality data can hinder the prediction capacity of such models,” he says.

Using modeling to predict molecular interactions is an evolving and growing field, though. “With advances in software, hardware, and AI/ML techniques, we anticipate this technology will continue to be an exciting and expanding area that could result in better outcomes and accessibility of life-saving therapies for the patients who need them the most,” Konagurthu states. “It is important that scientists embrace these new technologies that can help them overcome long-standing challenges in the pursuit of more effective treatments. We believe that these molecular modeling tools hold great potential for the pharmaceutical industry and patients around the world,” he continues.

Thermo Fisher Scientific, for example, has developed some of these tools for tackling formulation challenges with ASDs.

Enabling time, cost, and materials savings

The potential benefits afforded by molecular modeling when applied to formulation development are, indeed, numerous. “Mechanistic models enable visualization of molecular processes, formation of rational hypotheses, and interpretation of data against the backdrop of fundamental kinetic and thermodynamic principles,” comments Bennette. “Predictive models—both physics- and statistics-based—are useful for defining specific experimental conditions, rationally selecting formulation candidates, and informing key risk factors for formulations and processes,” he says.

“Modeling tools thus make the whole process of formulation development more efficient by enabling formulators to use a rational, mechanistic approach to defining the key problem statements, developing hypotheses, and interpreting data,” Bennette remarks. “Within the framework of these models, developers can ‘work smarter, not harder’ and avoid time- and resource-intensive trial-and-error approaches,” he emphasizes. “Indeed, by leveraging predictive molecular modeling tools that enable in silico formulation development, significant savings in time, costs, and materials can be saved, especially for expensive active pharmaceutical ingredients,” Konagurthu concludes.

®
Vol. 48, No. 9
September 2024
Pages: 15–16

When referring to this article, please cite it as Challener, C.A. Molecular Modeling in Formulation Development. 

Insights into molecular behaviors and predictive capabilities are bringing numerous benefits.

Molecular Modeling in Formulation Development

Decreasing the cost and time of drug development while advancing differentiated therapies with improved efficacy and safety is a priority for drug makers. Automation of activities across all aspects of drug development and manufacturing can help achieve these goals. Indeed, “automation can help the API industry with improved safety, productivity, product quality, and optimized costs, making it an essential tool for the industry to remain competitive and meet regulatory requirements,” states Raghavendar Rao Morthala, chief scientific officer, Navin Molecular. These benefits can be realized at all stages of development as well as across development and manufacturing.

Many applications for automation in API development and manufacturing

Automation in many forms has, in fact, been used for API development and manufacturing for many decades, according to Morthala. He points to examples such as electronic lab notebooks and electronic batch manufacturing records, statistical software for design-of-experiment studies employed in process optimization efforts, control systems for API manufacture in both batch and continuous mode, and inventory management and material movement solutions.

“Automation is used extensively in API synthesis,” emphasizes David Ford, senior director of chemistry at Snapdragon Chemistry, a Cambrex company. “While batch manufacturing processes are primarily executed under direction of a human following a batch record, automation is used for temperature and flow control (reagent dosing), as well as for safety interlocks. Continuous flow manufacturing processes use automation more extensively, as automation can be used for reliable start-up and shut-down of the reactor, clean-in-place operations to recover from a clog, and to divert reactor effluent to waste if the quality is expected to be poor,” he explains.

The above forms of automation, Ford says, make work more convenient, but an operator is still required to be present to initiate operations and complete the batch record. “In contrast, automation in the R&D lab can be used to run processes completely without operator intervention,” he observes. In this context, Ford explains, whether batch or flow processes, all steps including reagent charges, sampling, on-line or off-line liquid chromatography, and storage of product for further analysis can be automated.

Such automation tools can also be coupled with statistically powered experimental design or auto-optimization algorithms. “This type of automation can accelerate tasks that can be labor-intensive, such as process optimization and process characterization. Automation can, in addition, free a process development chemist or engineer to research more broadly in the literature or execute more high-level analysis of the data,” Ford comments.

In addition to accelerating physical experiments, literature search engines are available that continuously update comprehensive databases of identified and potential chemical substances, chemical transformations, and newer methodologies. “Using these automated systems, research scientists can quickly identify desired organic transformations and optimum reaction conditions, as well as gain insights into reactant and product properties and raw material sourcing options,” says Morthala.

It is important, cautions Jens Schmidt, process expert and high-throughput experimentation (HTE) manager at Lonza, to apply phase-appropriate automation techniques to ensure highly effective use of the technology across the entire lifecycle of a project. In the initial phase of a project, according to Schmidt, automation is particularly potent and extensively utilized in process development, commonly in the form of HTE.

“This technique facilitates the miniaturization and parallelization of chemical reactions, allowing for the rapid testing of a broad spectrum of reagents, solvents, catalysts, and conditions. It enables the assessment of various routes, thereby increasing the likelihood of identifying optimal conditions early in the project’s timeline,” he explains.

For example, Ford highlights the use of automation for screening of reagents, catalysts, or solvents to find beneficial combinations and automated workflows for measuring solubility and partitioning in liquid-liquid extraction to investigate isolation strategies for intermediates and APIs. “Such a screening approach can enable identification of useful combinations that would escape attention if experiments are performed manually because a much larger number of combinations can be screened in an automated context,” he notes.

As a project matures, HTE continues to play a crucial role, albeit in a slightly different capacity. The main goal is to aid in process optimization and characterization, thereby enhancing process robustness, notes Schmidt. He adds that automation during this phase also encompasses the standardization and automated execution of individual experiments, with the objective of delivering reliable and high-quality data.

“Upon reaching the commercial stage, automated experiment execution or HTE supports a diverse range of activities, including the optimization of established processes and the enhancement of process understanding,” Schmidt comments. Automation also contributes to the attainment of sustainability goals, the securing of supply chains, and the resolution of unforeseen issues. Of course, automation of large-scale manufacturing is a significant aspect that warrants consideration due to the numerous advantages it affords.

“Automation is transforming the landscape of API development and manufacturing. By automating tasks like route exploration, reaction optimization, and data analysis, researchers can significantly accelerate the identification of optimal synthetic pathways and fine-tune processes for maximum efficiency. These benefits translate to faster development times, reduced errors, and more consistent API production. For drug developers, that means reduced risk and uncertainty in API development, enabling a greater focus on clinical trials and regulatory approvals,” Schmidt concludes.

Using automation in synthetic route development

The concept of automated multicriteria evaluation of new process routes, combined with route development, is based on data mining, according to Morthala. “There is an increasing prevalence of chemical processes being developed based on automated experiments that make use of data science, which can provide multiple numeric criteria for scientists to reach a balanced decision on the suitability of a novel synthetic route,” he says.

Information generated via data mining can be used for chemical route development and evaluation, including planning of hypothetical synthetic routes, Morthala contends. Such efforts can take into consideration not only process conditions, but environmental indicators, such as energy consumption, E-factor values, solvent scores, reaction reliabilities, and route efficiencies allowing for multi-criteria environmental sustainability evaluations, he observes. “This type of automated workflow is based on deep data mining and algorithmic decision-support tools to enable multi-criteria decision making,” he says.

Small-molecule APIs are becoming more and more complex, and the resulting longer synthetic pathways pose challenges for process chemists to develop efficient API manufacturing processes. Overcoming these challenges is crucial to avoid delays in investigational new drug applications and expenses in transitioning new drug candidates from early development to commercial production, according to Simon Wagschal, associate director of advanced chemistry technologies, Lonza Small Molecules. “As the industry faces new drug candidates requiring over 20 synthetic steps on average, AI [artificial intelligence]-enabled predictive route design technologies have become essential for creating efficient retrosynthetic routes,” he says.

Indeed, AI-enabled API development tools are already being used by several R&D companies for the identification of optimized synthetic routes, according to Raghavendar Rao Morthala, chief scientific officer, Navin Molecular. “An AI platform has the potential to design and synthesize APIs in a matter of weeks or months, compared to the years it can take using conventional methods,” he says.

Currently, however, the application of AI to API synthetic route design is in its early stages, contends David Ford, senior director of chemistry at Snapdragon Chemistry, a Cambrex company. Synthetic route design by AI is used primarily at the brainstorming stage of route development, he says, with identified options narrowed down by humans. “Given the importance of the choice of a route or routes and the quantity of resources used to pursue them, the role of the synthetic chemist remains central in route design,” he states.

Wagschal adds that AI tools have traditionally been used for early-phase discovery, often generating routes not suitable for commercial-scale production.

Other exciting applications for AI in process development, Ford says, lie in its ability to manage knowledge. “Recent advances in large language models have made it possible to index and explore diverse sets of documents like weekly updates, PowerPoint presentations, technical reports, batch records, and campaign summaries,” he explains. These AI models allow for the content to be searched semantically, based on the meaning of the search, instead of by keywords.

“Given the volume of information trapped in these documents, it is very likely that work is forgotten, and hard lessons then need to be re-learned,” Ford observes. “We see great opportunity in helping researchers make connections between their work and the work that has been done in the past,” he notes. Generative AI models may also prove useful for preparing first drafts of templated documents like standard operating procedures, qualification protocols, and batch records, according to Ford.

Lonza, meanwhile, is one company not deterred by the limitations of early AI programs developed for scouting of synthetic routes for small-molecule APIs. “The introduction of computer-aided synthesis planning (CASP) technologies with route prediction capabilities has significantly influenced the route selection process,” says Wagschal.

Nearly all CASP tools employ a form of machine learning called Monte Carlo Tree Search to rapidly prioritize computationally predicted synthetic routes, Wagschal explains. This heuristic search method evaluates the likelihood of reaction success at each step of the proposed pathway, from starting materials through intermediates to the final API. The prioritized routes are then reviewed and refined by process chemists. “This approach, guided by our experts, has increased the number of potential strategies and led to more efficient API synthesis routes,” he states.

To overcome the limitations of early AI-based models for route development with respect to commercial practicality, Lonza has combined leading AI-powered predictive route design models with proprietary supply-chain data, including raw material costs, availability, and detailed supply chain intelligence, according to Wagschal. “This integrated technology allows process chemists to quickly identify not just the shortest pathways, but also the most commercially viable ones by considering real-world supply chain constraints. These advanced AI solutions provide a holistic view that streamlines route scouting and accelerates the transition from research and development to robust manufacturing processes, resulting in significant time and cost savings in developing complex APIs,” Wagschal concludes.

A holistic approach to automation can provide benefits at all stages of development and manufacturing.