Predictive Success Through Lab Testing

The findings, which also showed the platform's ability to predict critical type 1 diabetes molecular "biomarkers," further validate the importance of the new model as a valuable research tool in type 1 diabetes. The software is designed to enable researchers to rapidly streamline laboratory research through the evaluation of alternative scenarios for therapeutic strategies that show the most promise for working in humans.

"Since laboratory studies can cost hundreds of thousands of dollars, and early stage human clinical trials can cost $10 million dollars or more, predicting the right conditions to try is important," said von Herrath.

"We've found that using this in silico (computer analysis) prediction platform can quicken the pace and effectiveness of type 1 diabetes research," he continued. "By allowing us to pre-test our theories in computer models, we can ensure that the more time-intensive and costly process of laboratory testing is focused on the most promising therapeutic strategies, with the greatest chance of success."

The platform, a life sciences company specializing in predictive technologies, has previously been shown to successfully predict various data from published type 1 diabetes experiments. von Herrath's team used a different approach to test the model, asking it to predict the outcome of a hypothetical experiment on nasal insulin dosing frequency in animal models that had not yet been performed. The prediction was then tested in the laboratory, where its results were confirmed.

In addition, he said, the model was able to accurately identify the particular time frame at which key type 1 diabetes "biomarkers" kicked in. Biomarkers are specific cell types or proteins that tell researchers at what point a therapeutic option is working or when it is time to start treatment. In the case of the La Jolla Institute study, the model successfully predicted the onset of biomarkers indicating beta cell protection in the NOD mouse.

"The model accurately predicted that implementing a low frequency nasal insulin dosing regimen in animal models was more beneficial in controlling type 1 diabetes than a high frequency regimen," said von Herrath, noting that the software's prediction of the biomarkers was key in this process. "These results confirmed our hypotheses on the benefits of low-frequency nasal insulin dosing. But even more importantly, the advantage of applying computer modelling in optimizing the therapeutic efficacy of nasal insulin immunotherapy was confirmed."

The platform, developed over two years, simulates autoimmune processes and subsequent destruction of pancreatic beta cells from birth through frank diabetes onset (hyperglycemia).

Specifically von Herrath's team employed the model to investigate the possible mechanisms underlying the effectiveness of nasal insulin therapy, using the B: 9-23 peptide. "The experimental aim was to evaluate the impact of dose, frequency of administration and age at treatment on key molecular mechanisms and optimal therapeutic outcome," he said.

Using parameters input by the scientific team, the model accurately predicted that less frequent doses of nasal insulin, started at an early disease stage, would protect more effectively against beta cell destruction than higher frequency doses in NOD mice.
von Herrath added that the positive results add credence to the idea of creating computer models for analyzing therapeutic interventions in human disease. "These results support the development and application of humanized platforms for the design of clinical trials," he said.; Source: La Jolla Institute for Allergy and Immunology