SPANners: in SPAN 1.2 and 1.3 we used 3 models: a mouse model of obesity induced hyperglycemia, aging mice, and spontaneously hypertensive rats. If you decide you want to use these models in SPAN 2.1, then we need to start feeding mice and ordering aged mice from NIA. Or, you can pick other models. Lets start to discuss. 

We do not know from SPAN 1.0 whether any of the comorbid models brings added value to the predictive ability of the trial. This would be based on subgroup analyses for each treatment, hopefully being done by each site for their treatment. In the absence of that information, the only reasons to use comorbid models is 1) assume that they bring value, 2) make the trial LOOK rigorous to the outside community.

Cenk

Well, I do not know. What I am proposing is to analyze SPAN 1.0 data to see animal model subgroup outcomes with each treatment (agnostic to which treatment). If there is a model (aged, SHR, etc.) in which nothing worked, then perhaps it is not appropriate. Or, if mortality was prohibitive in a particular model (i.e., aged), then perhaps it is not appropriate. There must be something there to guide us. 

If you are interested, we can run an analysis isolating the interaction between treatment and model in SPAN 1.0. 

Our Augusta sites are fine will all those models. In the meantime, we are okay with the HFD-induced obese model first.

The UCSD site is interested in mouse MCAO model with aged mice, non-obese diabetic (NOD) mouse model, or spontaneously hypertensive rat MCAO model. 

The NOD/ShiLtJ strain (commonly called NOD) is a polygenic model for autoimmune type 1 diabetes. Diabetes in NOD mice is characterized by hyperglycemia and insulitis, a leukocytic infiltration of the pancreatic islets. Marked decreases in pancreatic insulin content occur in females at about 12 weeks of age and several weeks later in males. A 2022 phenotyping study found that 86% of females and 48% of males became diabetic by 30 weeks of age; median female incidence was 18 weeks. Immune phenotypes in the NOD background consist of defects in antigen presentation, T lymphocyte repertoire, NK cell function, macrophage cytokine production, wound healing, and C5 complement. These defects make the NOD background a common choice for immunodeficient mouse strains. Diabetes onset data is available.
 

The diabetic B6.BKS(D)-Leprdb/J (Jackson Strain #:000697) mouse line is also interesting for the comorbidity MCAO study. Mice homozygous for the diabetes spontaneous mutation (Leprdb) become identifiably obese around 3 to 4 weeks of age. Elevations of plasma insulin begin at 10 to 14 days and of blood sugar at four to eight weeks. Affected mice are polyphagic, polydipsic, and polyuric. The course of the disease is markedly influenced by genetic background. On the C57BL/6 background there is compensatory hyperplasia of the islet B cells, and continued hyperinsulinemia throughout an 18- to 20-month life span. Wound healing is delayed and metabolic efficiency is increased. Although normal in body weight, blood glucose, and plasma insulin, heterozygotes (Leprdb/+) also have increased metabolic efficiency and can survive a prolonged fast longer than controls. Experiments involving destruction of the ventromedial nucleus of the hypothalamus suggest that Leprdb may cause a defect in the hypothalamus. Steroid sulfotransferase enzymes, aberrantly expressed in diabetic mice, interact with the Leprdb mutation as modifiers of gender differences in obesity-induced diabetes susceptibility. 

I agree with Cenk that a more detailed analysis of the SPAN 1.0 data on which models were informative (or at least told us different information than the young mice- whether this is informative for human trials remains unknown).  I think aging, diabetes, and hypertension are attractive models for SPAN since they model common comorbidities in humans.  We struggled with survival in the aged mice- maybe we need to rethink how to get these mice to survive in SPAN 2.0.  If we go the HFD route again I suggest feeding for 16 weeks since Nirav had data that the phenotype was more pronounced at 16 weeks than at 12 weeks.  

The Iowa group would suggest stay with young animals, mice and rat with the filament model for stage 1. 

Enrique

The UCSD group agrees with Iowa's suggestion of using young mouse and rat filament models for Stage I. 

We should go with young healthy mice and rat in stage 1 (drug vs. vehicle) with filament model. This will give time to analyze more data from SPAN 1.0. Additionally, the reviewers might be interested to see whether intervention works in young healthy mice and rat.  

1) Using different animal models in various stages of SPAN 1 combined with the MAMS design leading to early dropped treatments was a huge problem. Some treatments that dropped out early could have ended above the futility boundary if tested in subsequent animal models. Therefore, we should decide what models to use and use them at all stages while MAMS does its magic.

2) It is not necessarily true that something that does not work in normal young mice will not work in comorbid models. We have no proof of that general principle.

3) We have four months to prepare, so kicking the "which animal model can" down the road is unnecessary if we do the SPAN 1.0 analysis, which should not take long. We are already on it.

4) I am not sure we can afford to consider a new animal model we have never used in the network, such as those offered above for diabetes. I am sure they are valuable and test different aspects of biology. But it will be yet another discovery process as the trial is ongoing and may force us to modify the protocol in midstream, something I do not favor. 

5) Having some labs do some models and not others in SPAN 1 was another problem because of the massive covariance with the site that it created. So I think all sites should do all models.

6) The critical question is what, if any, animal model should be eliminated. Aged animals are high on the list, given the very high mortality, but as Lauren said, perhaps we can figure out how to reduce mortality. 

Agree with Cenk that we should not drop interventions until the end of the study.

Another thought. We can start with rats: normotensive (vehicle and drug) vs. hypertensive  (vehicle and drug). 

I agree that young male and female rats and mice would be a good place to start for SPAN 2.0.  I also think that within a few weeks (I hope) we should have data from each of our analyses and what I think Cenk said they will examine as to whether there were differences in efficacy of treatments across the models.  That is, whether treatments dropped because they lacked efficacy in the co-morbid models or if they dropped because with increased numbers we had refined the treatment effect estimates and they fell below futility estimate.

We will have the results of that analysis for the next SC meeting.

I feel it is better to start with young animals at Stage 1.  It will be easy to order these animals and will cost less. All sites have experience working with young animals.  When the compounds are confirmed to work, we may further test them in other models.  The sites pick up or are assigned to the model they prefer.

I think there may be some misunderstanding as the role of the co-morbid models. The only reason to avoid them would be in the very unlikely event that the results of the young animal models predict exactly the results of the co-morbid models. Marcio is preparing an analysis for each intervention that will show you how each treatment differed in each of the models. One thing we do know is that after decades of failed trials, every drug worked in young animals. So i am not sure it makes sense to repeat the mistakes of the past all over again, unless the MAMS paradigm is so good that you could predict all the results using just young animals. More to follow...

We know the importance of the comorbid models. Do you plan to test all compounds in these models?  We use healthy animals for stroke experiments and the results may be different. However, it is a good tool for initial screening.

Cenk,

this is a rational point, but i do not want expectations that we can use data to choose models. I strongly doubt that we will have the power to show that any one model is superior. On the other hand, there is a literal mountain of recommendations that stroke preclinical research migrate to using older animals with co-morbid conditions. Even if we showed with reasonable power that the results in young animals perfectly recapitulated the findings in co-morbid/aging animals, we would STILL be required to study drugs in non-young animals until/unless one of our drugs proves effective in humans. 

SPANners, i recommend we start off with the SHRs, or another hypertension model, and the diet-induced obese mice. The Aging mice are too brittle, and it is a waste of time to study young animals. 

Comments from others? 

So the NOD mice are a model of Type 1, but not Type 2 diabetes. I think acquired, e.g. diet induced, are more representative of what humans face. 

Is this the same as the so called db/db mice? 

db/db mice are used to model phase 1 to 3 of diabetes type II and obesity. Mice that are homozygous for the diabetes spontaneous mutation (Leprdb) demonstrate morbid obesity, chronic hyperglycemia, pancreatic beta cell atrophy and come to be hypoinsulinemic.

Non-obese diabetic or NOD mice are used as an animal model for type 1 diabetes. Diabetes develops in NOD mice as a result of insulitis. NOD mice will develop spontaneous diabetes when left in a sterile environment. The incidence of spontaneous diabetes in the NOD mouse is 60–80% in females and 20–30% in males. 

I like the idea of starting with the SHRs. In this case, should we monitor the BP to understand the drug effect on pressure?

Pat,

I am not sure what you mean by "... i do not want expectations that we can use data to choose models." This is not about what you or I want. Any valid information that gives us insight is better than no information when deciding what models to include. We have analyzed the efficacy of treatments in each animal model in SPAN 1. The SC has not yet seen our analyses and the interesting findings. I suggest we start by reviewing them at the SC. I am not suggesting that I have the magic bullet, but dismissing an opportunity to get some insight would be unwise.

I would also like to reiterate my earlier comment:

In SPAN 1, it was a mistake to use MAMS to drop interventions at each stage while using different animal models in different stages. The reason is very simple. An intervention that drops out based on data from only a subset of models is not given a chance in other models where it may shine and surpass others in overall efficacy. So we either use all decided-upon models in all stages equitably, or we test all interventions throughout the trial without dropping any. Tough decision, I know. But "Errare humanum est, sed in errare perseverare diabolicum." Seneca.

Thanks,

C

Stage 1: I still prefer including young, healthy rats, but I am okay going with SHR rats and high-fat diet-fed mice for 20 weeks. 

If none of the five interventions work in young animals, then will provide a strong rationale for future NIH studies to exclude young animals. 

The patients with comorbidities are treated with other related drugs, such as drugs for controlling glucose and hypertension. Will we also consider using these drugs when we assess the efficacies of new compounds in these stroke models?  The experiment results depend on the action mechanism of new compounds. We may see different results when they are tested in these models.

The project includes stages 1-4. The cost will be much higher if we skip young animals and start with the stroke comorbidities. 

I suggest using young animals first so we do not need to test all compounds in late phases. If the decision is made to go with the stroke comorbidities in stage 1, we are OK to go with it. 

The pilot study is completed now. Should we have a post-pilot experiment to refine the filament model? We found that the filament sizes in the SOP did not work well in the pilot study. Should we consider selecting large-size or longer silicone-coated filaments for the stage 1 study or keep CCA occluded during ischemia in the mouse model (not a clinically relevant model)?

Many of you have made the point that it is unwise to change models over the stages of SPAN 2.0. I agree. I would hope you could agree to a few animals and use them in the first 3 stages. We have tentatively planned on a clot model for Stage 2.4, but that is open to further discussion. 

With respect to data from 1.0, again i would caution that we may or may not be able to glean much. the sample size was designed to allow the MAMS to work. We did not power the study to address the drug effects in the individual models. 

Hi, I agree with the power caution in interpreting the model data but still think it has the potential to be informative.  Other things we learned from SPAN 1.0 will also be informative- the high mortality in the aging and the moderate phenotype of the HFD mice.  

We are in our weekly Yale SPAN group meeting discussing models and propose the following groups.  We like the idea of including multiple models right from Stage 1 (incl young) and suggest Stages 1-2 can randomize all the models except for embolic model across the sites.  Then the embolic model is begun at Stage 3 or 4.  

1- aging, but we need to figure out how to get the mice to survive better

2- HFD, fed for at least 16 weeks

3- SHR

4- consider mice with induced HTN- eg 8%NaCl chow plus L-NAME in drinking water- anyone have experience?

5- young rats and mice- even if it is to be able to test whether young are informative

6- embolic model in rats (later stage)

I hope we could discuss models with or without CCAO:

Case 1:             MCAO + CCAO = 70-80% reduction of MCAO blood flow

                        MCAO – CCAO = 50-60% reduction of MCAO blood flow

Case 2:             MCAO + CCAO = 70-80% reduction of MCAO blood flow

                        MCAO – CCAO = 10-20% reduction of MCAO blood flow

Conclusion:  MCAO + CCAO produces more consistent focal brain ischemia.

I hope we could discuss models with or without CCAO:

Case 1:             MCAO + CCAO = 70-80% reduction of MCAO blood flow

                        MCAO – CCAO = 50-60% reduction of MCAO blood flow

Case 2:             MCAO + CCAO = 70-80% reduction of MCAO blood flow

                        MCAO – CCAO = 10-20% reduction of MCAO blood flow

Conclusion:  MCAO + CCAO produces more consistent focal brain ischemia.

Span 2.1 would include all 4 models: young mice (YM), Aging mice (AG), High fat diet induced obese mice(OB), and SHR

Please comment and try to commit to all 4 or which 3 or which 2 you woud rather dol.

Bingren,

This is good data and i think we all agree that CCA closed reduces CBF more than CCA open. Let's settle on which co-morbidity models to use, and then turn to this quesiotn. 

My initial ranking of the animal stroke model choices are hypertension > diabetes > aging > young health animals, because about 75% stroke patients are with hypertension, 20% stroke patients with diabetes, about 85% stroke patients are older than 60 years old, and 10-15% stroke patients are healthy young adults.  

I agree to all 4 models to be tested concurrently.  I think the MAMS design would work such that we are increasing our power with each stage, and by taking these interim looks we can address any issues like high mortality.  However, say there is very high mortality in one treatment paradigm (eg drug/placebo requiring multiple iv injections in the aging mice) could we drop that one model/treatment pair but continue the model in the other treatments or would we have to drop for all?

Learning a new model would take much work in our short time. I agree that we have 4 (5, if young healthy mice and rats are counted as separate groups) good models that were very sensible in SPAN 1.0. As you may remember from the data I showed during the steering meeting, the young health mouse was the least sensitive model on the day 28 corner test, and it was the mouse model with the higher sample size. With that said, I summarized some of my thoughts:

1. It would be great if all models were tested at the same stage.

2. If both young rats and mice are included, we have 5 model groups. Is it feasible for all 5 models to be tested at the same? Or, do we need to drop something?

3. I also agree with what was discussed at the steering meeting: stroke happens in young people. However, since it happens in a smaller frequency. From a statistical standpoint, is it feasible that Young healthy mice + rats be 1/4 of the sample size?

4. Last, if we include young healthy mice, I suggest matching their age with the high-fat diet model. The age gap only adds more variability when analyzing the data.

My vote would be for Aging mice (AG), High fat diet induced obese mice (OB), and SHR. I will send our preliminary analyses of treatment x model interaction by email. We did not have much time to go beyond the first slide.

My vote would be to start with 1) high-fat  diet induced obese mice (20 weeks on high-fat diet, starting at 8 weeks old) and chow-fed diet for the same age total age: 7 months 

2) Hypertensive rats 

3) Aging mice

We vote for two rat models (SHR and young, healthy rats) and two mouse models (aged and high-fat diet-induced obese mice).  If we would like to use the model more relevant to patients, the drugs for hypertension and hyperlipidemia should be added to the experiments. 

Our vote is for (1) aging mice- perhaps start at 15 months to see if helps survival, (2) HFD for 16 weeks, (3) younger healthy mice age matched for the HFD cohort- 20 weeks, and (4) SHR.

Our site is interested in (1) high-fat diet-induced obese mice (OB), (2) SHR (3) Aging mice (AG)

We have a transcranial MCA ligation model, which we used in aged mice, and the survival rate was almost 100%.  Is it possible to use it in stage 1?  

We need to consider the ischemic stroke model we used in SPAN. Did we have sufficient ischemia? We typically use young mice and rats to test new drugs. The results did show us if the drugs worked or not. That is also what most of the other groups did. 

We vote for young rats, SHR rats, high-fat diet mice, and aged mice. However, we will do what you assign.