Translation: Stroke Flashcards
(34 cards)
Animal experiments
in neuroprotection trials shows a high effectiveness in therapy (20-40% effect size)
What’s the issue of the clinical translation from bench to bedside?
Thousands of animal studies does not work in real humans; for stroke only TPA worked
Steps for drug to be on public
- Discovery: synthesis of certain compounds
- Pre-clinical: disease targeted drug usage and safety
- Phase I: Try general safety of drug on healthy individual (<100) !Not done by pharma, usually have a company dedicated to this step!
- Phase II: First time drug is tested on patients (<200) with a focus on safety check rather than efficacy
- Phase III: Checking whether the drug actually works (several thousands) with a focus on efficacy
- (After 2 successful Phase III) Approval
- Guideline: State that certain drug MUST be used on certain occasions; stable income for pharma
Why is preclinical to approval becoming easier each year?
Because all the basic mechanism drug is being found, so nowadays the discovered drug with smaller effect size needs to be approved as well.
2 Types of translation
Type 1: Innovation (discovery to approval)
Type 2: Implementation (approval to guideline and implementation)
How can a drug be approved but not have an effect?
- Drug may not be reaching the patients due to e.g. lack of guideline knowledge on doctors
- Drug not being taken properly/enough
In successful case, how long does it take from discovery to approval?
10-15 years
What is the only clinically proven pharmacological therapy of acute schemic stroke?
IV Thrombolysis (TPA); which can only applied to few patients There are currently no neuroregeneration therapy for human
Why is translation such a “black box”?
- Complexity
- Low hanging fruits have been picked (easy researches have been done); time to focus on small effect sizez
- Flawed clinical design (due to time window/sensitivity of stroke; median medication is given at 16h after stroke)
- mouse is not human
Limitation of animal models
- We can model what happens after the stroke (e.g. occulusion) but not how it happens
- Due to smaller size, time frame of progress is most likely different (may not be true comparing the tPA)
What is bias
subjective reality informed by one’s preference
What is the issue of “bias” in stroke research
-attrition bias
-sex bias
-bias from low sample size
-winner’s curse
- low external validity
-HARKING
-publication bias
-Low prevalence of methods to prevent bias:
randomisation 40%
blinding 40%
sample size calculation 0
conflict of interest statement 5%
What is attrition bias?
Unexplained “missing” animal on the paper
This is huge bias because excluding one outlier from the already small sample size, the result is most likely be significant
What is a disease that has a high comorbidity with stroke?
Depression
Animal model for depression
e.g. Anhedonia test: checking the wanting of sucrose
What is statistical power
probability of a hypothesis test of finding an effect if there is an effect to be found (20% for neuroscience due to the small sample size); the bias against the null hypothesis
What is the new true?
The true result and the false positives (5%)
The key is to focus on “true” significance but this is impossible information t gain.
In reality, 40-50% of significant result could be false positives
Low sample size bias
mean group size in neuroscience for pre-clinical steps: 8 animals
mean statistical power: 45%
false positives: 50%
overestimation of true effects: 50%
Winner’s curse
Lower power studies will lead to
1) high rate of false positives
2) out of the true significant finding, its effects will be overestimated
HARKING
Hypothesising After the Result are kNown
Low external validity
validity of applying the conclusions of a scientific study outside the context of that study
e. g. for stroke research, it is like “healthy, male twins raised in a super secured room feeding healthy granola” can the result of such study be applied to real life patients?
e. g. SPF fallacy… mouse model for stroke (SPF) has immune system characteristics of a new born baby as they are raised in super secure environments all their life
What is publication bias
aka “file drawer problem”
Only positive results are being published
How can our research finding be more “true”
- Reduce bias
- blinding
- randomisation
- in/exclusion criteria
- report results according to guideline e.g. ARRIVE - Increase power
- check power and achieve at least 80%
- do appropriate sample size calc
- replicate - Use stats sensibly
- do not be deceived by p-value
- think biological significance and effect size
- replicate - Practise open sciences
- preregister (tell the protocol beforehand)
- publish NULL results
- make original data available
Monetary bottlenecks
Phase I-III & implementation is where the big money can be earned
