The Pharmaceutical Industry: Engineering Frustrations

Hedley Rees and Keith Plumb discuss how current methods of drug development are impeding engineers, and suggest a new model to provide patients greater access to medicines

WE (the authors) work mainly in the fields of supply chain management, and engineerng design respectively. In Part 2 of IChemE’s Pharma SIG series, Access to Pharmaceuticals in the 21st Century, we present our personal perspective of the many frustrations that occur for engineers working in the pharmaceutical industry, how this impacts on the access to medicines for patients, and how things could be improved.

Supply chain

Pharma supply chains play a fundamental role in pharma research and development, as has been highlighted by a series of tragic events where adulterated or toxic materials have contaminated the chain to the detriment and even death of patients. In 2007, a blood-thinning agent, heparin, was adulterated due to the product licence holder procuring a toxic substance that had been illegally substituted for the genuine registered material.¹

One of the major reasons for this tragedy was the complex nature of the supply chains. Figure 1 shows the pre-clinical supply chain required to assess the safety of the compound under development.

At this stage, the supply appears to be a relatively simple way to produce the 5–10 kg of active pharmaceutical ingredient (API) for testing in animal models. However, there is already an array of suppliers and service providers involved, often spanning the globe. There has been little attempt to work with producers that can offer a range of processing options that allow starting materials, key intermediates, and API to be produced under the one roof.

In Figure 2 we see the extension (and additional complexity) of the supply chain stages into production of the dosage form and clinical trial kits, to be shipped into transient storage awaiting call-off from investigator sites. As at the pre-clinical stage, production of data (mainly clinical) is the primary focus, rather than preparedness for the commercial supply chain.

When the time comes to submit the regulatory filing to gain approval to sell (manufacturing authorisation approval/new drug application/biologics licence application) – assuming clinical trials have gone to plan – the competent authority mandates the data be submitted using an electronic common technical document (eCTD). Module 3 (Quality) of the eCTD contains the chemistry, manufacturing and controls data, specifying the supply chain in immense detail, not to be deviated from once approval is given. Here we have the nub of the first engineering frustration, the whole of the supply chain has effectively been cast in stone at this part of the development process.

Once the product has become commercialised, the logistical complexity and widely-devolved responsibility for supply-chain activities leaves those responsible for managing this with a daunting task, as illustrated by the list of constraints below:

• Scarce/bespoke raw and starting materials specified.
• Limited sourcing options or sole sourcing.
• Reduced dosage-form options due to poor solubility.
• Registered contractors with insufficient capacity or capability.
• Uncertain and overly variable process yields.
• Supply and quality/technical agreements inadequate to deal with commercial requirements.
• Analytical methods development and transfer not fit-for-purpose.
• Shipping and storage conditions not supported by necessary stability data.
• Inaccurate customs declarations and customs clearance failures.
• Tactical, cost-based contractor relationships.

During the first wave of the Covid-19 pandemic a team of volunteers from IChemE and the ISPE UK Affiliate worked on a consultation response to the UK Government looking at the “UK Science, Research and Technology Capability and Influence in Global Disease Outbreaks”. This response focussed on making the medicines supply chain more resilient.

The response noted many constraints to modifying drug manufacturing processes that mirror the supply-chain issues noted above, including:

• Stringent change control procedures to assess any proposed modification and maintain compliance.
• Plant cleaning can require a substantial amount of time and effort, particularly in multi-purpose process facilities where cleaning is required between products.
• Complicated technical transfer processes for introducing new/different products into a new/multipurpose production plant.
• A reliance on a limited number of machinery suppliers, offering equipment delivery times of more than 6 months.
• The increasing use of single-use equipment has exacerbated the reliance on a limited number of suppliers.
• Specialised equipment, building and services must go through a lengthy qualification exercise before beneficial use.

All of these supply-chain issues come together to make products more expensive than required and reduce the resilience of the overall supply system, potentially reducing access to medicines for many people.

Development and commercialisation

The approach to new product development (NPD) in pharma, as described above, involves an initial stage of discovery research, then handover to a team of developers to handle the registration process and then another handover to the commercial team shortly before approval.

The problem with this process is that it has become ever less productive, as summarised in ‘Eroom’s Law’ which states: “The number of new drugs approved per billion US dollars spent on R&D has halved roughly every nine years since 1950, falling around 80-fold in inflation-adjusted terms.”

Much of the development process focuses on data collection and regulatory submission and the principles of strategic supply chain management have been omitted. It is as if the prime purpose of the whole supply chain – to deliver fit-for-purpose products and services to customers – has been overlooked.

If we look at the development process from a process engineering perspective (see Figure 3) it still shows the importance of the documentation that is required for the common technical document. Having spent so much time (see section on patents) and money on gathering data for the documentation, there is little or no opportunity to refine the process, meaning the validated manufacturing process is frequently complex and highly inefficient.

On top of this the number of viable products identified in basic research, drug discovery and pre-clinical can fall from 10,000 to 5 at phase I clinical trial, and of those, probably only one will achieve regulatory approval. This overall process will cost in the order of £2.6bn (US$3.55bn) per approved product (see Figure 4).

Once the new drug application (NDA) has been submitted based on the validated manufacturing process there is little or no opportunity to make changes to the overall supply chain. This is because changes would require further development work and there is rarely any time available to do this because the “patent clock is ticking” (see below).

Complex manufacturing

Figure 5 shows the overall process for manufacturing a biologics medicine.

The manufacture of traditional small molecule medicines using a chemical synthesis route will be similarly complex. There are opportunities for improving manufacturing efficiency and resilience. The following are some examples:

• Process intensification to simplify and shorten production processes. For example, moving from batch to continuous processing reduces the size and complexity of equipment. Importantly, continuous processes are more responsive to changing product demands. Smaller equipment is also easier to clean, and transfer or replicate in another location.
• Process analytical technologies (PAT), measure the critical process parameters on plant instead of by remote analysis, bringing improvements in cycle time, process robustness and operational efficiency.
• Telescoping is running two sequential process steps as one end-to-end step, ideally eliminating an isolation operation. The simplified step is quicker overall, and removes the ancillary operations of packing, storing, and re-loading an intermediate product, as well as the need for additional isolation equipment with its associated cost of ownership. Telescoping can also reduce the downtime for cleaning.
• Single-use process equipment is employed by some (mainly biological) pharmaceutical processes, and although operationally expensive compared to fixed installations, reduces the requirement for re-sterilising equipment and is more flexible. Mobile equipment is another alternative that is found in both small-scale biological and synthetic processes.
• Modular and system-based techniques allow the facility design, construction and testing to be partially or fully carried out remotely from the manufacturing site, cutting down the installation and qualification durations.

It should also be noted that information technologies (eg data science, machine learning, artificial intelligence) are rapidly-evolving areas that potentially offer huge manufacturing benefits:

• Digital twins (digital replica of physical assets such as process equipment) deliver quicker design, installation, and qualification of new equipment. Digital twins can also assist with process optimisation and setting up virtual factories.
• Robotics are being introduced to make routine operating tasks safer and more compliant.
• Statistical techniques, like chemometrics and statistical process control, characterise production processes and improve understanding of the critical process parameters. This in turn yields greater predictability and less process variability, resulting in more efficient and robust processes.
• Digitalisation can link manufacturing information into the supply chain, enabling real-time optimisation and end-to-end product traceability.
• Virtual and augmented reality is being used to train people to operate specialised pharmaceutical machinery and to monitor production activities.

However, despite all of the techniques that could make a huge difference, it is rarely possible to use them unless they have been built in the pre-clinical stage and included in the eCTD. This all adds further to the engineering frustration.

Outsourcing

Like many other business sectors, the pharmaceutical sector started to outsource its business requirements in the 1980s and this has turned into a flood, with the creation of small drug developers, contract research organisations (CROs) and contract development and manufacturing organisations (CDMOs). Much of the engineering and validation work has also been outsourced.

The Big Pharma companies allowed their engineers to join engineering design companies, or become self-employed freelance consultants, and their scientist to join CROs and CDMOs: thus effectively giving away much of their knowhow and then being obliged to buy it back from the design companies, CROs and CDMOs. Since Big Pharma companies now have few engineers on their payroll, they have to engage freelance engineers to act on their behalf as client representatives on big projects run by the design houses. In effect, both groups of engineers are selling Big Pharma’s previous own knowhow back to them, whichever side of the fence they are sitting on. This is also true for the CROs and CDMOs.

To quote Hedley’s words from “Taming the Big Pharma Monster”: “Pharmaceutical companies (PCs) are dried up prunes compared to the fulsome plums they used to be. They have retrenched into opposite ends of the prescription medicines lifecycle, leaving most of the work of testing, developing, making, storing, moving, and distributing medicines to third parties.”

Figures 6 and 7 show the industry before and after outsourcing.

Patients not patents

As many professionals working in the pharmaceutical industry know, the patent clock is running the show. Somewhere during discovery research, the company registers the molecule with the patent office. This sets a definite date for expiration of the patent-protection umbrella. The patent clock is now ticking (patent is 20 years – remaining patent life is 20 years minus time to develop the drug, hence the clock is ticking during development).

With that clock now ticking, every day wasted is a day that a product with marketing authorisation does not have patent protection. At the end of patent protection, the market share of the owner of the patent can plummet by 80% or more. Each day saved can be literally worth millions. This means there is no time to focus on process improvements, and perhaps more importantly there is no time to focus on patients beyond the safety data demanded by the regulatory authorities. The focus on patents has broken the focus on patients.

The harsh reality is that a yawning gap has emerged between companies developing drugs, and the end users (healthcare professionals and patients). The most contact that drug developers have with healthcare providers is between the developing company’s clinical group (often a contract research organisation) and the clinical trial study investigators who operate within a network of hospitals and clinics signed up for the trials. The communication is pretty much one way; the investigator role is to collect data from recruits on the clinical trial. The developer has to give a brochure telling the investigator everything about the product undergoing the testing. The objective is to collect the clinical data until the end of the study, normally blind to both the investigator and patient for Phase II studies and beyond.

Once patent protection is lost, the company owning the marketing authorisation frequently loses 80% market share and the price of generic alternatives can be less than 10% of the patented product. This means that every day that is spent preventing a product getting to market can cost the company owning the marketing authorisation millions. So here we have yet another engineering frustration caused by the ticking of the patent clock. As engineers, we cannot help to improve medicines for the benefit of patients. We would prefer to take time to solve emerging constraints that could impact all aspects of the future commercial supply chain, rather than press on regardless of impact on the future supply chain.

A new model

In our opinion, the way to fix the many frustrations that have built up in the pharmaceutical industry is to completely remodel the development process to engage closely with the end-users of the new product(s) being developed – specifically healthcare professionals with deep expertise in the medical condition(s) concerned. This focus on end-users is the de facto approach taken by engineers across most industry sectors. In the absence of close communication with end users, it is highly likely that the product output will not fit its intended purpose.

So, our contention is that drug development should be turned on its head – instead of starting with a molecular structure to treat a condition, we begin with the needs of patients, and the healthcare professionals that treat them. This is where an engineering mindset can counterbalance any over-reliance on pure science, at the expense of delivering the desired outcomes in practice.

Figure 8 is a diagrammatic representation of the new model. Note, a voice of customer (VOC) is constructed between end users and the medicine developer, prior to commencing any development work. The VOC is then used to construct prototypes for testing for robustness, with the strongest candidate being selected to undergo commercial development.

The underlying principle is that instead of the current three step process of discovery, development, and production, we move to a two-step process – prototyping and production.

Prototyping is used in almost every other industry, except medicines. In aviation, it is wind tunnels and flight simulators, using the full weight of science, technology, engineering and maths (STEM).

At the core of STEM is integration of all the skills required to bring products to market. ‘TEM’ is the missing ingredient we want to put back in the pot.

The following approach is suggested as a starting point:

• Include all the required disciplines at the beginning of the programme.
• Engage healthcare professionals (HCPs) with deep knowledge of the indication.
• Construct a VOC (patient and HCP).
• Consider therapy and diagnosis together (precision medicine).
• Integrate “discovery research” and “development” as a single function, titled “design”.
• Design develops small-scale prototypes, tested for safety, efficacy, and manufacturability.
• Design make full use of ex vivo methods (eg in vitro, in silico, tissue).
• Prototyping capability established in hospitals by HCPs and manufacturers working together.
• A single group will hold responsibility for pre-clinical, clinical, and commercial supply chains.
• That group has skills of strategic design, management and improvement of supply chains.

In other industries, product development follows a prototype approach, allowing promising candidates to undergo multiple rounds of refinement and optimisation before committing resources to development of a full-scale supply chain. In the case of drug development, the technologies to predict the safety, efficacy and manufacturability of prospective development candidates have moved ahead at great speed, yet the industry still predominantly relies on testing in animal models and humans during multi-stage clinical trials.

We therefore assert that drug development suffers from the lack of application of good biological systems in which to assess the ex vivo performance of a product before it is given to patients. This could include using technologies such as organ-on-a-chip, and isolated cells or tissue, frequently from human donors, increasingly used to assess safety risks ahead of clinical trials.

As a further example, blood samples can be used effectively to map the propensity of therapeutic products to trigger undesirable immune reactions. They also can be used to understand the range of responses across different patient populations and map the influence of specific genetic or pathological background in the observed biological response. This approach may also provide useful data to define meaningful dose ranges for clinical trials.

We believe that such an approach would introduce characteristics that are essential for the drug’s performance, including aspects of the product and process design relevant to the biological activity, safety, stability, half-life, mode of administration, target patient population, or even target product costs, etc.

Also, this should avoid the all-too-common prospect of candidates failing late in development because of issues that should have been averted during the development/product design stages.

In our experience (and also many of our colleagues), critical process parameters are not always well understood when scaling up a new process. A prototyping approach to development should be beneficial in developing a better understanding of these – and allowing flexibility to adjust these as time goes on may also be helpful (rather than having everything rigidly set in stone from the outset).

DESIGN PRODUCTS FOR SUCCESS, BEGINNING WITH THE END IN MIND

Our next point is the importance of the design in the success of a product. It can be argued that Pareto’s law is at play here too: that as much as 80% of a product’s value could be locked in the design stages, which probably consumes less than 20% of total development costs. It is therefore important to get a handle on the important requirements for a product and its potential development risks early on.

With the increase in digitalisation, computer models are being developed to assess the requirements for product safety, and therapeutic efficacy and evaluate potential development risks. Equally important are novel surrogate analytical methods such as low-cost high throughput analytics that can provide valuable information early on in development.

The combination of computer models and surrogate analytical methods are expected to provide valuable help in designing and selecting the best possible drug candidate(s) with the required properties to maximise potential for clinical efficacy and commercial success.

 

CHANGING MANUFACTURING STRATEGIES: PROTOTYPE VERSUS COMMERCIAL PROCESSES

In any other industry, would you define a commercial manufacturing process for a prototype before knowing if it works? Probably not, but this is exactly how current biopharmaceutical development is done, which results in a very high risk.

At present, over 90% of products in clinical development fail to reach registration. This results in hugely inflated costs and is very wasteful. New development approaches that use simpler, more standardised, and flexible manufacturing of prototypes, could allow the development of better, more efficacious, and cheaper drugs, thus making them more available to patients around the world.

We would expect it to reduce the time required to initiate clinical trials, but without compromising quality or safety to patients.

 

NEW APPROACHES TO PRECLINICAL AND CLINICAL VALIDATION

In our opinion, based on seeing so many failures within the pharmaceutical development cycle, it is time to embrace reality and declare openly that humankind simply does not know enough about the biology of many diseases. So, it would be beneficial to change the way of approaching clinical validation. As an example, different biopharmaceuticals targeting the same molecule can elicit completely different, and sometimes opposite, biological effects.

Novel genome editing technologies are now able to generate more refined biological models for disease (cell lines or animal models), simplifying testing and the interpretation of results. To make use of these technologies, there has been an increase in the application of adaptive trials allowing the introduction of modifications in the trials which also allow responses from patients observed in the clinic to be incorporated.

There is also a possibility to use Phase 0 trials which use sub-therapeutic doses or local administration to evaluate adequate safety in cases where a given drug could pose safety risks to patients. The industry also needs to contemplate more the use of alternative approaches, such as multiple branches in trials, to evaluate the performance of multiple alternative candidates when the mode of action is not well understood.

Conclusions

What we describe above are not panaceas, but methods which could help transform the development of drugs from the current hit-and-miss approach to one that is more rational and sustainable.

If these new approaches to development were used, many of problems that beset the manufacture of drugs at the large scale could be eliminated and many of the engineering frustrations removed.

Making safer, more efficacious, probably more targeted drugs, designed with the patient in mind with less of an eye on the need for patent protection should be in everyone’s interest. Particularly if these drugs can be marketed at a lower price and become much more widely available.

Reference

1. After Heparin: Protecting Consumers from the Risks of Substandard and Counterfeit Drugs, https://bit.ly/2XEp3Na 

---

This article is part of an ongoing series on Access to Medicines, developed in collaboration with IChemE’s Pharma Special Interest Group. Read the series here.