In today’s fast-paced world of scientific discovery, cutting-edge technology is revolutionizing the way we approach complex biological problems. One such innovation is in silico modeling, a method that harnesses the power of computer simulations to model biological systems. But what exactly is in silico modeling? How does it work, and why is it becoming a cornerstone in research and development?
To find out more, we sat down with Premier’s Senior Vice President and Global Head, Biometrics, Abie Ekangaki, Ph. D., for an engaging conversation about this innovative approach to drug development.
Abie, for those who might be less than savvy in in silico methodologies, can you describe what in silico modeling is?
In the simplest terms, you can think of in silico as a way to generate or produce something using computer modeling or computer simulation, and that’s it. It’s worth noting that one can apply in silico modeling in a broad array of situations, and it simply involves the use of mathematical or computational models that can replicate actual real-world phenomena.
A unique benefit of in silico predictive modeling of the molecular dynamics and system biology, for instance, is that it can eventually yield a digital rendition of an entity that mimics the biological conditions associated with a particular disease. In other words, it can yield a digital rendition of a patient having a specific disease. This is the area where in silico methods can truly be groundbreaking. And I believe that as in silico modeling and simulation methods evolve, it will eventually transform the future of drug development.
Is in silico modeling a new concept in clinical research? If not, what is driving the recent interest and focus in the industry?
It’s important to understand firstly that in silico modeling and simulation is just one application of machine learning techniques. And machine learning methods in general represent only a sliver of the much broader and more sophisticated world of artificial intelligence (AI). We know that for years, AI has been transforming industries by automating tasks, enhancing decision making, and creating new value.
The drug development space has been slower and more conservative in its adoption of these approaches, particularly when it comes to securing regulatory agency endorsement of these methodologies for driving regulatory decisions. We’re seeing recent trends, however, where both the FDA and EMA are more open to leveraging in silico modeling and simulation techniques in clinical trials for drug approval. In fact, in a 2020 press announcement, the FDA commissioner at the time, Scott Gottlieb, shared that in silico methods are recognized by the FDA as a useful tool for streamlining randomized control trials and advancing personalized treatment.
The following year, FDA’s internal modeling and simulation group released a report which outlined its successes and opportunities in modeling and simulation. The report revealed the extent to which CDER, the Center for Drug Evaluation and Research, had already been actively implementing these techniques in PK modeling, Bayesian modeling, and in disease modeling, among other areas. All of this demonstrates FDA’s recognition of, and commitment to, the inevitable role of complex modeling and simulation techniques, which will continue to be the case in the drug development process. And I believe this is what contributes to the growing interest around in silico modeling and simulation in the industry.
Are there particular therapeutic areas or indications where in silico modeling is preferred?
There is currently no clear guidance from FDA around restricting in silico modeling and simulation approaches to specific therapeutic areas. For now, we know anecdotally that it depends on the area of application of these in silico techniques.
For example, in silico methods used for PK/PD modeling are generally disease or therapeutic area agnostic. However, when using these methods for disease mechanistic modeling, which is a unique approach that leads to generating synthetic patients for use as control arms in clinical trials, the FDA is generally more likely to favor in silico approaches for rare and ultra rare diseases, particularly where there are no available treatments or perhaps little available real-world data that can adequately serve as a control arm for efficacy evaluation. In these cases, it may also be unethical to assign patients to a placebo, like in pediatric glioblastoma trials.
In contrast, it’s most unlikely that the FDA would accept in silico synthetic controls in situations where there is no justifiable clinical reason why human subjects cannot be issued a placebo or active control in a clinical trial, such as in pain or diabetes studies. As a general rule of thumb, I always recommend that sponsors engage the FDA in discussions of trial design options for their proposed study, especially when more complex in silico methods are being proposed.
There is a notion that in silico trial designs can lead to early conditional approval of a new drug. Can you outline how this is achieved?
A strategic benefit of in silico modeling and simulation methods is that they can circumvent or condense key stages of the drug development process that would result in significant reduction in the development time and cost. There are several success stories where applications of in silico methods have received FDA endorsement to progress towards a conditional approval. In fact, computational modeling and simulation has been incorporated into pre-market approval applications for medical devices as far back as 2002. In 2023, a Hong Kong-based AI drug discovery company called Insilico Medicine succeeded in securing orphan drug designation from the FDA for a small molecule developed for idiopathic pulmonary fibrosis. Excitingly, the company developed their drug candidate purely using in silico AI modeling and simulation.
Another recent example is the COVID vaccine trials of 2021. In silico modeling and simulation was applied in the rush for a vaccine by conducting virtual trials to assess the efficacy prospects of proposed vaccine candidates. These computational methods directly led to the selection of the optimal treatment option or optimal vaccine candidate that was then moved forward for confirmatory evaluations in the single pivotal human trials that the likes of Pfizer and Moderna conducted. Without doubt, the implementation of complex in silico modeling and simulation techniques for those vaccine trials had a transformational impact in precipitating the drug development process and achieving vaccine approval within one year.
Can you expand on the appetite for in silico modeling amongst regulatory bodies? Is this method generally accepted or is there work to do on that front?
Both the FDA and EMA have encouraged the implementation of complex modeling and simulation techniques in support of the design, analysis and reporting of clinical trials. In fact, in December 2020 the FDA published its regulatory guidance on complex innovative designs for drugs and biologics. This guidance has since set the framework for how sponsors may obtain FDA feedback on technical issues related to modeling and simulation in their trials. And this guidance has also made quite clear the type of quantitative and qualitative information that support these approaches and which sponsors should submit for FDA review.
Four years later, in silico modeling and simulation applications have become even more sophisticated, largely due to advancements in technology and computational power, and are delving into areas like generating synthetic patients, digital twins, etc. In my view, more work is needed by regulatory agencies to update current guidance that addresses regulatory expectations for the increased sophistication of the modeling and simulation techniques.
However, we can see that work is ongoing to strengthen and refine expectations. Earlier this year, the Avicenna Alliance, an international consortium of more than 130 experts in the industry, academia, and regulatory agencies, including 13 FDA modeling and simulation working group members, released a comprehensive publication, which they titled Toward Good Simulation Practice. This publication offers the first comprehensive compilation of good practices for ensuring the quality and reliability of computational modeling and simulation in the assessment of medical products.
How do you foresee the clinical trial industry evolving as in silico modeling becomes more widely adopted?
In a funny way, I still remember 1999 when there was stupendously wild speculation globally about how the world will suddenly change once we enter the new millennium. But what did we find? All the anxiety was completely unwarranted. I think something similar will happen here. In the short-term, I don’t foresee any dramatic transformation in how clinical trials are conducted. Rather, I expect to see gradual enhancements and efficiencies across the drug development process over the next decade, with innovative modeling and simulation approaches playing an important part in that.
I expect that for some diseases, in silico modeling will help reduce the burden of requiring human subjects, or they may even circumvent the need for traditional early phase human trials in some instances. I also believe these approaches will help boost the speed of bringing drugs to market, particularly for certain types of diseases that are better positioned to satisfy regulatory requirements for more innovative approaches.
There is much more work needed for developing technical skills, expertise, and for building more sophisticated computational systems that are readily accessible and used for highly complex dynamic AI modeling and simulation. We are entering an inflection point today, but it’s incumbent on those of us who are knowledgeable in this space to raise awareness of what’s possible with innovative in silico methods, and to help enlighten investigators and sponsors on how regulatory bodies have embraced such innovation.
You’ve been a biostatistician for over 30 years and have seen lots of changes in the space. What excites you most about innovative methodologies like in silico?
I will unabashedly modify a famous quote by saying this, “the beauty of progress lies not in reaching the end, but in the continual refinement of our tools and understanding.” Having said that, when I think about where we are today and compare it to when I first became a professional biostatistician in 1989, there’s no doubt in my mind that we are in an exciting era of innovative statistical methodologies, where the methods we bring to the table can have major impact in bringing needed medicines to patients faster. In fact, one can arguably say that today we’re standing at the threshold of new truths where uncertainty is not a limitation, but instead an invitation to discovery. So yes, these are indeed exciting times to be a biostatistician or a data scientist, but I would say more importantly, these are exciting times in the drug development space.
___
To hear more of Abie’s perspectives and expertise around in silico modeling and simulation, book a meeting with him here.
Webinar 
Perspectives Blog 