At Entelos, I saw first hand that developing mathematical models of disease requires significant investment before you have a simulation platform that you can market to pharma. These PhysioLab models were complicated and required multiple PhD-level engineers and life scientists working elbow to elbow for 6 months to a year or more before we had a mathematical model that could be shown to clients. To leverage internal expertise related to modeling immunology and metabolism, there was an interest in developing a PhysioLab for modeling type 1 diabetes. Towards this aim, the leadership had cultivated a relationship with leaders at the American Diabetes Association (ADA). As further development of the Asthma PhysioLab was put on hold, my team was tasked to define the biological and therapeutic scope for a proposed PhysioLab of human type 1 diabetes. Once defined, we were to present a development plan to a panel of scientific leaders assembled by the ADA. As I dove into the literature, I became increasingly concerned about the feasibility of the task set before us. Ultimately, we developed a plan but we had some concerns that we kept in reserve.
On presentation day, the ADA had brought a stellar cast to our office. Richard Kahn, the Chief Scientific and Medical Officer of the ADA, was accompanied by George Eisenbarth (UColorado), Mark Atkinson (UFlorida), Jeff Bluestone (UCSF), and Diane Mathis (Harvard) to critique our plan. The plan for the meeting was that Entelos leadership would open up with a presentation covering the company’s mission, approach, and highlights of recent successes with pharmaceutical partners. Once they had dazzled them with the company’s brilliance, our technical team would present the development plan. Well the presentation went to plan for about 10 minutes when Jeff Bluestone spoke up. “So I’ve seen this smoke and mirrors presentation before, what are you actually going to do?”
So now attention in the room turned quickly to us, the development team sitting behind all of the Entelos leadership. With a scrambling of the agenda, we jumped into our presentation. While I wasn’t the main presenter, I was there to help answer questions. At some point in the conversation, we kind of pivoted and said that we didn’t think there was enough information in the literature to develop a PhysioLab for human type 1 diabetes. Maybe developing a PhysioLab for a mouse model of the disease might be more appropriate. I said that for instance, we don’t really know how much of a reduction in beta cell mass – the cells that produce insulin – is required for symptomatic onset of the disease. At the time, pretty much every paper on type 1 diabetes stated in their introduction that symptomatic onset of the disease is associated with an 80-95% reduction in beta cell mass and referenced studies by Gepts and others that analyzed pancreatic tissue samples obtained via autopsy.
Type 1 diabetes typically presents in youth. When you look at an infant though, you notice that their proportions – arms, head, torso – are different than an adult. Organs, like the pancreas, are proportionally larger in an infant compared to an adult. The pancreas produces a protein – insulin – into the blood and the concentration of insulin regulates the response in all the other peripheral tissues. It then follows that the pancreas has excess capacity to maintain a given insulin concentration in an infant compared to an adult. This would imply that a greater reduction in beta cell mass is required to occur in an infant compared to an adult for clinical presentation of the disease to occur. When I reanalyzed the data presented in the papers by Gepts and others, a similar trend was observed. The severity in beta cell reduction at onset decreased with age where, on average, a 40% reduction in beta cell mass was sufficient to precipitate clinical symptoms at 20 years of age. Identifying what percentage of beta cells remain at onset strongly influences therapeutic strategies. If 60% remain, then a strategy to sustain and regrow them become important. Alternatively if they are all gone, then exogenous replacement becomes important.
When I presented this data, it seemed that Mark Atkinson was supportive of this idea but George Eisenbarth disagreed. He said that my analysis was not correct as the density of beta cells within the pancreas is also changing. Actually I did account for that in my analysis. While we didn’t close the ADA deal that day in 2003, it did set the stage for the ADA to support a PhysioLab of the NOD mouse model of Type 1 diabetes [1]. As I guess I wasn’t drinking the consulting cool-aid, I wasn’t selected to help develop the Type 1 Diabetes PhysioLab but was reassigned to work on the Type 2 Diabetes PhysioLab developing virtual patients.
Once I joined WVU as a tenure-track faculty, I thought about papers that I could write to bolster by CV. Based on this re-analysis of the Gepts autopsy data, I drafted a manuscript and tried to submit it to the leading journals within the field. First sent it to Diabetes, which rejected it almost immediately. Then I sent it to Diabetologia, who sent it out for review. The reviews came back and were quite negative – essentially the reviewers didn’t support publishing a paper where I reanalyzed these data and came to a different conclusion than the original analysis. Given the dogmatic nature of the reviews associated with these disciplinary journals, it was ultimately published in PLoS One, where acceptance was based on technical competence and not perceived impact. In a review by Mark Atkinson [2], this paper was highlighted as a paper of special interest with the comment: “An interesting meta-analysis that draws to the forefront, the question of, `what percentage of beta cells are destroyed at the symptomatic onset of T1D?’.” By challenging one of the foundations of the field at the time, it’s maybe not surprising how dogmatic the reviewers were. In retrospect, what I didn’t appreciate was how common this behavior is in peer-review [3].
3. Huber J, Inoua S, Kerschbamer R, Konig-Kersting C, Palan S, Smith VL, 2022. “Nobel and novice: Author prominence affects peer review,” Working Paper Series, Social and Economic Sciences 2022-01, Faculty of Social and Economic Sciences, Karl-Franzens-University Graz.
DK Design Blog 