Dr. Pam Attends International In Vitro Conference in Boston
October 25, 2012
Last month I attended an international conference in Boston, “Developing More Predictive In Vitro Models.” The 2-day meeting brought together researchers from industry and academia from around the world to discuss progress and challenges in the development of various in vitro models, their validation and implementation. I was impressed with the number of different strategies researchers employed to create more sophisticated and complex in vitro models, but also left with some questions: Is enough being done to ensure that we are building the most predictive in vitro models possible? And is the scientific community willing to reduce or replace the use of animals in light of the new methodology if it shows to be more predictive? Below I have highlighted some of the interesting issues discussed at this conference and provided context for their significance. As always, please contact me if you have questions or would like to learn more.
- Dr. Pam Osenkowski, Director of Science Programs
The current process of drug research and development is broken. Reports indicate that it takes about 13 years to get new drugs approved from the time they are discovered, and despite the extensive testing that is performed on the drugs during this time, including a reliance on experiments involving different species of animals, the failure rate is still over 95%. Something is clearly wrong with this picture. The question is: what can be done to fix it?
Where can improvements be made in the discovery pathway so that we can better predict whether a drug will be safe and effective for people? How can compounds that will fail in people be eliminated sooner so that time and money are not wasted? A recent landmark article in Science Translational Medicine entitled “Reengineering Translational Science: The Time is Right” (July 2011) by Director of the National Institutes of Health, Dr. Francis Collins, asked scientists to rethink tradition and change the status quo. Not only did this paper acknowledge the failure of animal models to predict safety in humans, it provided a potential solution to the problem of the 95% failure rate -- the development of more predictive, human relevant, in vitro tests.
Many researchers believe that the development of more sophisticated in vitro assays may hold the key to learning earlier in the drug development process whether or not a compound should move forward. In vitro assays hold the potential of being incredibly powerful tools if relevant cell lines and proper assay endpoints are selected, among other things. Over the last few years, I’ve been impressed with the increasing levels of complexity and sophistication achieved in a number of in vitro models. So needless to say, I was eager to attend the “Developing More Predictive In Vitro Models” conference last month to hear first hand about the experiences scientists are having developing new and improved in vitro models.
As you know, many kinds of in vitro assays are available to address various scientific questions. In vitro assays are often criticized as being too simplistic in that they do not fully recapitulate what happens in the body. Developing an in vitro assay is no simple task for a number of reasons including the large number of variables that one can manipulate during assay development. For instance, should cells be grown on tissue culture plastic dishes, or should the plastic dishes be coated with some protein found in the body? If coated, with which protein and at what concentration? Or should combinations of proteins be used? In which size dish should cells be grown? Each cell culture plate has round wells to grow cells -- some plates can hold 6 large wells, other plates hold 1,536 tiny wells. Does this matter? Which kind of cell is relevant for a particular assay: human cells or animal cells? What level of differentiation is appropriate -- should stem cells be used or more differentiated cells? With which kind of nutrients should the cells be supplied? Will results differ based on which company the reagents are purchased from? What is the outcome scientists should use to determine whether an assay is predictive? Should other endpoints or combinations of them be used? These questions are just the tip of the iceberg, because in science, the devil is in the details.
While it would be no surprise to a scientist that changing any of the variables above would likely influence the outcome of an experiment, I was very impressed with how carefully scientists were learning about and characterizing their individual assays. Much of the data discussed at the meeting detailed the experimental findings of assay development. The level to which these models were characterized was really mind-blowing, as gene expression changes were often monitored, which leads to the generation of a lot of data that needs to be carefully interpreted. Researchers are certainly moving ahead with careful characterization of many new in vitro assays and beginning to carefully test the models as predictive tools.
I was also impressed by the amount of work that is being done to generate more sophisticated 3D models. A tremendous effort is being put forth to construct models which have structure and function that more closely mimic what happens in vivo. Scientists reported improving assay quality by adding features that made the models more in vivo-like. I was encouraged to see scientists openly sharing their successes and failures in a collaborative effort to reach the common endpoint of building predictive models.
But I was disappointed that several important issues were not addressed well that could waste valuable time and resources and undermine efforts to get these in vitro models validated. Many of the models discussed at the meeting relied on animal cell lines when human cell lines could have been used instead. When researchers were questioned on their choice of species for the cell lines used in their models, they did not have a good scientific rationale for choosing an animal cell line over a human cell line. They responded that they’d already spent a lot of time and money on the characterization of the animal cell line. It was generally acknowledged that it would be good to try constructing a similar model with human cells, but I can’t say that anyone seemed particularly driven to make that switch, even when common sense tells us that a human cell line might predict better for humans than an animal cell line. What will happen if scientists continue to invest time and money on developing in vitro models with animal cell lines and they find there is limited predictive value? Someone will surely ask them at that point why a human cell line wasn’t used instead. So why not start now with a more appropriate cell line and cover your bases?
I personally believe that the reduction and replacement of animals in research is an important animal welfare issue and should be a huge incentive for developing predictive in vitro assays. I also presumed that more scientists would share that sentiment. I was discouraged to hear that many scientists seem to be designing these new models just to have another way to test drugs rather than a way that would reduce the use of animals in research. In a session entitled “How Predictive Can We Be?” one audience member directly asked a panel of experts that if they could go back in time with the knowledge that animal models fail to accurately predict what happens in humans, would they still rely on animals models for these purposes? Sadly, the panel did not know how to answer. They agreed that the animal model has problems, but still felt some data could be collected. Scientists are reluctant to move away from animal models, even when they are broken. This is both a scientific and an ethical problem.
This reminds me of an interesting talk at the meeting in which a scientist compared biological modeling with the way astronomers originally studied the solar system. At one time, the earth was thought to be the center of our solar system. However, with this model in mind, it became harder and harder for astronomers to explain data on the orbits of other planets because the data did not seem to fit the model -- because the hypothesis was incorrect. As soon as the model was revised to become sun-centered, all of the planetary orbits were more easily explained. Are animal models like the earth-centered solar system? If and when we move to more human-relevant models, will we make greater progress with more predictive models? Would the data we collect be easier to explain and extrapolate to people?
As a scientist I believe we owe it to ourselves to fully investigate the predictive potential of more human-relevant in vitro models. If scientists want to develop models that are predictive for people, it makes the most sense to invest our research dollars in models that have direct human relevance from the start.