Sponsored content brought to you by

Predictive modeling could revolutionize drug manufacturing by helping biopharma organizations achieve right-first-time scale-up,” says Tim Gardner, PhD, Founder and CEO of Riffyn.

Biopharmaceutical process development is complex, time consuming, and expensive. The traditional approach is to test each parameter in separate experiments and combine the results.

But while single-variable trial and error works to an extent, it is inefficient and—Gardner says—a legacy of older technologies. It also misses opportunities to increase process robustness through multi parameter design-space optimization, which is the hallmark
of Quality by Design.

“Traditional approaches have been limited—both technology and training—to siloed viewpoints and one-variable-at-a-time hypothesis testing. Why? Because traditional technology makes it too hard to combine process run parameters, offline analytics, and online or at-line bioreactor data for a complete picture of your system.

“We’re taking tens of hours to weeks to assemble data for a single development batch, “ Gardner says. “So, engineers and scientists do the best they can with an incomplete data set.”

An intelligent approach

But process development doesn’t have to be this way. Technology advances have given industry the tools to create predictive models.

“Intelligent Process Development (IPD) delivers predictive models of process performance and product quality. The models can be physical, statistical, or even hybrid physical/statistical models. We’ve used this approach to achieve right-first-time scale-up and shorten product development time by 50%”, says Gardner.

The IPD approach has already demonstrated its value. In 2017, Riffyn helped Novozymes, a global producer of industrial enzymes and biopharmaceutical ingredients, bring four yeast bioprocessing strains to market twice as fast as any program in their history.

And, Gardener says, IPD can help address common scale-up challenges: “Currently, the transition from small scale to commercial scale production is fraught with error or underperforming batches. Each failed development cycle adds 3–6 months of delays in time-to-market. That’s tens of millions of dollars in real and opportunity cost.

“IPD can deliver on-target performance across scales of production, and thus avoid ‘experimentation at full scale’ which is obscenely costly. The magic is in the data, not the process type or modeling method. That data holds secrets waiting to be discovered—as long as you can integrate it and annotate it with meaningful context.”

A learning process

Firms considering the IPD approach need to embrace innovation in operations, tools, and training. Gardner says, “You need a culture that applauds data-driven and model-based approaches to development. You need the technological tools that make it feasible to combine multidimensional data in minutes, not days or weeks. You need to create the space and incentives to learn and adopt advanced mathematical, process science, and operational approaches.”

Companies may need to invest in technology, Gardner says, as the approach requires data capture systems as well as an IT infrastructure and hub for modeling. Having a flexible analytics architecture that can combine the process map, model, and data to deliver process quality and predictive models is also critical.

The adoption of Intelligent Process Development will also likely involve training, according to Gardner, who says technicians familiar with traditional methods will need to learn new skills.

Organizations have often failed to offer the training their workforces need to harness the power of predictive, multivariate modeling. The incentives were missing to do so. “If your data systems never let you look at integrated, multivariate data, then what’s the incentive to learn the statistical and modeling methods that allow you to harness the power of it? The situation is different now—the tools are there, but the training often is not.”

Manufacturers that adopt Intelligent Process Development must prepare. But according to Gardner, the benefits are worth the effort.

“Certainly, it requires investment and commitment,” Gardner says. “But the investment is tiny compared to the cost of failed product development and missed product opportunity.

“For a start-up of a 50 people, adopting IPD might cost $100,000 –200,000 in technology and training. One failed production batch of Phase I materials could cost the company $1–10 million in lost time and opportunity.”


Take a video tour to learn how Riffyn can make Intelligent Process Development a reality for your R&D team. Riffyn.com/GEN

Previous articleSpike-In Standard Technology for ctDNA-Based NGS Hotspot Assays
Next articleOverview: Manufacturing Adenoviral Vectors at Large Scale