Development of computational models of bone formation and resorption to predict changes in bone in preclinical intervention studies.
Lead Research Organisation:
University of Sheffield
Department Name: Human Metabolism
Abstract
We want to develop computer programs to test new therapies for diseases of bone to reduce animal testing and accelerate research progress. Musculoskeletal disease including arthritis, osteoarthritis and osteoporosis are the most prevalent cause of work-related disease affecting the UK and account for one third of GP consultations. Patients affected by these conditions have often reduced mobility, leading to isolation and poor quality of life. Despite excellent progress in the care of these conditions there are still high numbers of people, who do not respond to these treatments very well. New cures are therefore required for these patients.
Any new drug needs to be tested on animals before it can be used in people. This is not only to see if the drugs are safe but also to understand whether they are efficacious. When doing this, groups of mice are needed to test each individual dose or time of administration and the mice have to be killed before the effect of the drug on bone can be determined. This can take many experiments, and is very laborious, expensive and time consuming. More importantly, it may require the use of many animals as several conditions need to be tested, such as how much drug needs to be administered to see a benefit, for how long and how often. Another important problem is that sometimes it is difficult to know how well the mouse will predict what will happen in patients. This is because we have never been able to compare them properly. To do that, we need to take pieces of bone for analysis. We can take bone samples in mice but this is less straightforward in patients because even taking extremely small samples can be an unpleasant procedure and if done for research purposes can only be practically done in very small numbers of volunteers.
This project aims to overcome these problems by creating computer programs able to predict how a drug may work, how much is needed and for how long. Similar programs are being created to predict responses in patients, so comparisons between human and mice can be made. In this way when a new drug will have to be tested, we can use the computer program first and then determine the best dose and time of administration to be tested only in a small group of animals to verify the accuracy of the computer prediction. To create the computer program we will collect data first by taking a drug which is known to build bone well and is already used in patients affected by osteoporosis; giving it to mice at different doses and for different times and then analysing the effects on bone mass, and strength in two ways. One in the traditional way of looking at several different time points, and one uses a new piece of equipment, called an in vivo microcomputed tomography (microCT) scanner, which is capable of looking at the bones in the same mouse over a period of time, without the need to kill it. This equipment is very similar to the DEXA scan used in patients. We will compare the new method with the traditional one to ensure the new method performs in the same way or better than the traditional one. The data will be used to generate an equation which will model how bone is formed so that when a new drug needs to be tested we will be able to use this equation to predict how good this will be. By adopting both the new equipment - in vivo microCT and this equation we will reduce the number of mice used, and the data derived will be more precise because they have been measured in the same mouse. Moreover, we can build similar programs in humans and compare the two understanding better when mice will predict patients' response.
Any new drug needs to be tested on animals before it can be used in people. This is not only to see if the drugs are safe but also to understand whether they are efficacious. When doing this, groups of mice are needed to test each individual dose or time of administration and the mice have to be killed before the effect of the drug on bone can be determined. This can take many experiments, and is very laborious, expensive and time consuming. More importantly, it may require the use of many animals as several conditions need to be tested, such as how much drug needs to be administered to see a benefit, for how long and how often. Another important problem is that sometimes it is difficult to know how well the mouse will predict what will happen in patients. This is because we have never been able to compare them properly. To do that, we need to take pieces of bone for analysis. We can take bone samples in mice but this is less straightforward in patients because even taking extremely small samples can be an unpleasant procedure and if done for research purposes can only be practically done in very small numbers of volunteers.
This project aims to overcome these problems by creating computer programs able to predict how a drug may work, how much is needed and for how long. Similar programs are being created to predict responses in patients, so comparisons between human and mice can be made. In this way when a new drug will have to be tested, we can use the computer program first and then determine the best dose and time of administration to be tested only in a small group of animals to verify the accuracy of the computer prediction. To create the computer program we will collect data first by taking a drug which is known to build bone well and is already used in patients affected by osteoporosis; giving it to mice at different doses and for different times and then analysing the effects on bone mass, and strength in two ways. One in the traditional way of looking at several different time points, and one uses a new piece of equipment, called an in vivo microcomputed tomography (microCT) scanner, which is capable of looking at the bones in the same mouse over a period of time, without the need to kill it. This equipment is very similar to the DEXA scan used in patients. We will compare the new method with the traditional one to ensure the new method performs in the same way or better than the traditional one. The data will be used to generate an equation which will model how bone is formed so that when a new drug needs to be tested we will be able to use this equation to predict how good this will be. By adopting both the new equipment - in vivo microCT and this equation we will reduce the number of mice used, and the data derived will be more precise because they have been measured in the same mouse. Moreover, we can build similar programs in humans and compare the two understanding better when mice will predict patients' response.
Technical Summary
We will develop new computational models capable of predicting the extent and anatomical location of bone formation and/or resorption following bone anabolic interventions. We will use models of osteoporosis as proof of concept but the models developed will be applicable to any given intervention of genetic, metabolic, surgical, pharmacological, or functional nature. We will use a combination of in vitro data and data collected by the non-destructive in vivo micro computed tomography (microCT). This will allow the accurate three-dimensional measurement of bone tissue morphometry changes over time due to the intervention under investigation in the same region of the skeleton of the same animal.
Once validated, this quantitative information will be fed into a computer model that will determine the relationship between each specific intervention, the time, and the bone remodelling each point of the bone. These models will then be used to make predictions on how the tissue morphology in a region of the mouse skeleton would change over a given time because of a given intervention. The accuracy of these predictions will be confirmed at the next time points, by simply comparing the computer predictions to the data generated by the in vivo microCT scanning.
The way in which candidate interventions are approved for human trials involves animal use for proof of principle studies in physiological models and uses large numbers of animals to generate statistically robust data at each time point. This technology will replace most of this in vivo testing and will require reduced number of animals to test at the chosen stratification due to the higher accuracy of the serial in vivo imaging technique. Moreover it will generate better quality data and allow a better understanding of the advantages and limitations of mouse models in testing interventions compared to human subjects.
Once validated, this quantitative information will be fed into a computer model that will determine the relationship between each specific intervention, the time, and the bone remodelling each point of the bone. These models will then be used to make predictions on how the tissue morphology in a region of the mouse skeleton would change over a given time because of a given intervention. The accuracy of these predictions will be confirmed at the next time points, by simply comparing the computer predictions to the data generated by the in vivo microCT scanning.
The way in which candidate interventions are approved for human trials involves animal use for proof of principle studies in physiological models and uses large numbers of animals to generate statistically robust data at each time point. This technology will replace most of this in vivo testing and will require reduced number of animals to test at the chosen stratification due to the higher accuracy of the serial in vivo imaging technique. Moreover it will generate better quality data and allow a better understanding of the advantages and limitations of mouse models in testing interventions compared to human subjects.
Planned Impact
Musculoskeletal disease including arthritis, osteoarthritis and osteoporosis have been identified by the WHO among the most costly because of the long-term cost and support they require with expenses on average of three per cent of total GDP in developed countries. With an increasing ageing population the size of the burden of musculoskeletal disorders is projected to increase relative to other conditions. Therefore there is an urgent need of improving our understanding of these disorders and find new therapeutic strategies in this area. This project will
1. Impact on animal welfare as there will be a drastic reduction in the numbers of animals necessary to achieve robust statistical power in experiments that are essential in terms of proof of principle and regulatory approval. We will reduce the number of animals of over 95% by using longitudinal non-invasive assessment of bones in the same animals by in vivo microCT and a) improving accuracy of measurements and reduction in variability during measurements in the range of 6-10 fold; b) allowing measurements at multiple time point in the same animal.
2. Develop computational models which will replace many of the experimental procedures (including the more invasive ones, which cause discomfort to the animals) used to determine the correct dose (or type of implant) and time of administration by computationally testing those conditions and verifying only the best conditions in a small number of animals.
In our bone analysis laboratory in 2011 we have analysed approximately bones from 750 animals to determine kinetics of bone formation and resorption. This is a facility which caters for all investigators within the Mellanby Centre (24 investigators) and their collaborators in UK and abroad (last year alone we have supported over 10 collaborators and a number of industrial partners including AstraZeneca, Acceleron, Novartis and Biocomposites). In UK the number of procedures in the musculoskeletal discipline varies between 39,000 to 44,000 per year between 2008 and 2010 according to the data provided by the Statistics of Scientific Procedure in Living Animals. This includes procedures for bone, muscle and tendon. When considering the number of project licences granted in 2010 and in the first 8 months of 2011 by the Home Office, 75% are for projects related to bone, suggesting that there are around 30,000 procedures in relation to bone scientific research. These figures suggest that the majority of these animals could be spared if our project was successful and the techniques were disseminated.
3. Generate better quality data. Present dynamic histomorphometry approaches allows to measures bone formation rates but not temporal rates of bone resorption. Moreover they are operator dependent and involve their subjective judgement and allow reliable detection of changes of parameter such as BV/TV in the region of 30-50%. In contrast this new technology development overcomes these limitations and allows the detection of more subtle changes (for example in the region of 5-10% for BV/TV)
4. Give more accurate prediction when translating from the preclinical model to the human cohort as we will be able to address the very important question of how relevant are the observations we make on mice for human health. We will be able to account at least for the most evident differences (e.g. anatomo-functional, dimensional, metabolic rates, different ageing process), and using predictive models we can simulate how those parameters change for a human host, bringing a considerable improvement over the current state of the art.
1. Impact on animal welfare as there will be a drastic reduction in the numbers of animals necessary to achieve robust statistical power in experiments that are essential in terms of proof of principle and regulatory approval. We will reduce the number of animals of over 95% by using longitudinal non-invasive assessment of bones in the same animals by in vivo microCT and a) improving accuracy of measurements and reduction in variability during measurements in the range of 6-10 fold; b) allowing measurements at multiple time point in the same animal.
2. Develop computational models which will replace many of the experimental procedures (including the more invasive ones, which cause discomfort to the animals) used to determine the correct dose (or type of implant) and time of administration by computationally testing those conditions and verifying only the best conditions in a small number of animals.
In our bone analysis laboratory in 2011 we have analysed approximately bones from 750 animals to determine kinetics of bone formation and resorption. This is a facility which caters for all investigators within the Mellanby Centre (24 investigators) and their collaborators in UK and abroad (last year alone we have supported over 10 collaborators and a number of industrial partners including AstraZeneca, Acceleron, Novartis and Biocomposites). In UK the number of procedures in the musculoskeletal discipline varies between 39,000 to 44,000 per year between 2008 and 2010 according to the data provided by the Statistics of Scientific Procedure in Living Animals. This includes procedures for bone, muscle and tendon. When considering the number of project licences granted in 2010 and in the first 8 months of 2011 by the Home Office, 75% are for projects related to bone, suggesting that there are around 30,000 procedures in relation to bone scientific research. These figures suggest that the majority of these animals could be spared if our project was successful and the techniques were disseminated.
3. Generate better quality data. Present dynamic histomorphometry approaches allows to measures bone formation rates but not temporal rates of bone resorption. Moreover they are operator dependent and involve their subjective judgement and allow reliable detection of changes of parameter such as BV/TV in the region of 30-50%. In contrast this new technology development overcomes these limitations and allows the detection of more subtle changes (for example in the region of 5-10% for BV/TV)
4. Give more accurate prediction when translating from the preclinical model to the human cohort as we will be able to address the very important question of how relevant are the observations we make on mice for human health. We will be able to account at least for the most evident differences (e.g. anatomo-functional, dimensional, metabolic rates, different ageing process), and using predictive models we can simulate how those parameters change for a human host, bringing a considerable improvement over the current state of the art.
