Evidence-Based Medicine (1-5) Flashcards
What is the discovery timeline for evidence based medicine?
Basic science → clinical research → clinical medicine
Basic science → laboratory discovery and preclinical research
→ initial drug discovery, animal model analysis
Clinical research → human clinical trials (2 - discovery and validation run by separate organisations)
→ requires application and regulatory body approval
→ results undergo Cochrane and NICE review to review all the data and compare the new therapy to old
Clinical practice → Cochrane + NICE review determine if released to public
→ monitoring during practice to see true side effects, release mass amounts
What does the preclinical research for evidence / experimental based medicine aim to discover?
Basic science → does the drug work in cell lines
→ e.g. skin fibroblasts from SMA patients, look for effects of drug
Animal models → does the drug work in animals (a system)
→ use a mouse model to see how much you can give someone before seeing negative side effects (toxicology studies toxic dose TD and lethal dose LD)
→ look for side effects
What do clinical trials aim to discover?
Does the drug work in humans / patients - is it beneficial?
→ multiple repeat trials to remove bias
What is the purpose of regulatory body reviews?
Review evidence from clinical trials
→ impartial meta analysis of multiple trials
→ does the drug offer improved clinical management?
What is spinal muscular atrophy (SMA)?
Disease caused by loss through apoptosis of the anterior horn cells in the spinal cord
→ motor neurones that connect spinal cord to skeletal muscle (leave by ventral route)
Lose innovation of muscle - progressive atrophy of skeletal muscle
→ limbs - can’t walk, wheel chair required
→ intercostal - death
Second most common cause of infant mortality, carrier frequent ~1 in 34 in the UK, 1 in 10,000 births
How can SMA be typed based on clinical presentation and age of onset?
Type I (severe, Werdnig-Hoffmann disease) → onset 0-6 months, never sit, die <2 years
Type II (intermediate) → onset 7-18 months, sit never stand, die >2 years
Type III (mild, Kugelberg-Welander disease)
→ onset > 18 months, stand and walk, die in adulthood
What are the genetics of SMA?
Autosomal recessive (need to inherit copies from both mum and dad)
Caused by mutations to survival motor neurone SMN gene
→ homologous loss of SMN1 gene (located Chr 5q13: unstable genomic region)
SMN1: anestral gene
SMN2: inverted duplication of SMN1
→ every species (if express RNA) has a copy of the SMN gene
→ recent evolutionary change in humans/apes - SMN1 gene copied, duplicated and inserted into genome creating SMN2 (multiple gene copies - pseudogenes)
→ only species where SMA is an issue e.g. in mice no SMN gene is embryonically lethal
What is the difference between SMN1 and SMN2?
SMN1: ancestral gene
SMN2: inverted duplication (5 SNP)
→ only differ by 11 nucleotides
C-T transition (nucleotide 6; exon 7) is the important difference
→ when C is present produces an exon splicing enhancer = correct splicing
→ when T present = alternate splicing
SMN1 produces 100% functional SMN protein
SMN2 produces 90% alternatively spliced protein (unstable) - delta 7 truncated protein deleted
What happens when you lose SMN1 gene?
Only SMN2 produces SMN protein → huge reduction in functional SMN protein
If you delete both SMN1 and 2 → embryonically lethal
Up to 8 copies of SMN2 on our chromosomes
→ copy number of SMN2 affects severity e.g. 1 copy produces severe type I, 3+ produces more SMN protein, less symptoms and no longer life limiting = type III
How is alternative splicing generally caused?
Weak polypyrimidine tracts → strong if ~15 pyrimidine residues
→ weak when 2-5 purines in it - means U2AF (binds the 3’ splice site and anchors U2 to branch site) will only bind transiently and can’t anchor U2
→ U2AF then binds nearest downstream Py
Intron 6 has a weak Py in both SMN1 and SM2 RNA → but in SMN1 its compensated for
How does SMN1 compensate for its weak polypyrimidine tract?
SMN1 and 2 have a weak Py tract → U2AF only binds transiently
SMN1 has an exon splicing enhancer to compensate in exon 7 (first nucleotide is the C which is different in SMN2)
→ the splicing enhancer recruits ASF2 (splicing enhancer) and exon 7 is recognised = correct splicing
How does the C-T transition in SMN2 gene lead to alternative splicing?
C to T transition in the gene (U in RNA)
→ U binds to ASF2 (splicing enhancer) weakly
Higher affinity for hnRNPA1 = splicing inhibiter
→ prevents U2AF and thus U2 from binding - masks start of exon
Exon 7 not recognised → exon 6 splices to exon 8
→ delta 7 product = unstable when translated
What is an antisense oligonucleotide (ASO)?
Small single ‘complementary’ RNA that bind / block cis elements
→ block / inhibit alternative splicing
How does Sprinraza promote correct splicing?
Spinraza physically blocks ISS in SMN2 intron 7
U1 has to bind at start of exon 7 → blocked by hnRNPA1
Spinraza blocks ISS → hnRNPA1 can’t bind
Forces 7-8 slicing which then forces 6-7 splicing
→ only RNA retained where 7-8 splicing produces 6-7 splicing knock on effect
→ no knock on effect = degraded
How was Spinraza shown to work in animals?
Mice humanised → give human SMN2, knock out own SMN
Results showed same findings as cell lines
What questions do you need to get a clinical trial approved?
Small compound (drug) → does it work (treat)? is there a biomarker to track outcomes? what is the mechanism? what is the therapeutic dose? what is the lethal dose? what are the side effects? does it improve current treatments? better than gold standard?
Devise (diagnostic test) → does it work (diagnose)? what is the sensitivity? what is the specificity? what is the PPV/NPV? does it compare to current diagnostic tests? would more people be diagnosed? does early diagnoses improve outcomes?
Why is the mechanism of drugs needed for clinical trials now?
Thalidomide scandal (affected embryo development - more than 10,000 babies born with physical abnormalities
→ now we need to know exactly how it works, used to be fine to just be based off animal testing
→ lead to stricter regulations
What basic data needs to be generated from mouse models for clinical trial approval?
Establish PD/PK profile
PD → pharmacodynamics - best route of delivery, how much makes it to the target organ? stability in the blood (degraded?)?
PK → pharmacokinetics - how quickly is the drug cleared by the liver / kidneys?
Identify target organs and assess reversibility of toxic responses
Establish max.tolerate dose (MTD) / establish no observed adverse effect level (NOAEL) in animals
Estimate starting dose for human studies
Evaluate reprotox / genitor / carcinogenicity
FDA (US) will review basic research to ensure drugs are safe for clinical trials and approve clinical trial protocols
What are the phases of clinical trials?
Phase 1 → testing of safety, generally with a few healthy volunteers
Phase 2 → testing of effectivity and toxicity (looking to see does drug work?)
→ generally with ~100 patients (non-blinded, not randomised)
Phase 3 → testing of efficacy
→ new drug against gold standard treatment and/or placebo (blinded)
Phase 4 → post-marketing surveillance
→ clinical use after registration
Why is an active placebo used sometimes instead of a sugar pill?
With a sugar pill its easy to tell you have the placebo as there are no side effects
→ introduces systemic bias
Active placebo → gives side effects (can’t unblind based on symptoms)
What are randomised controlled trials (RCT)?
Assigns treatment and control at random
Blind RCT → clinical knows patient doesn’t (rare)
Double-blinded RCT → clinician and patient don’t know (trial managers know)
What are the FDA pre-trial requirements?
- Toxicology / PD/PK / LD / TD
- NOAEL - to determine safe start dose in humans
- Clinical trial protocol - start dose, no. 0f participants (stat. sig data?), power calculation (ensures data relevant and useful)
- Safety protocols - how is drug administered, checks for adverse effects - how would you address them
- Biomarkers - how will the outcomes be measured to show improvements
- Statistical approaches - what are the trial outcomes, how are they measured and when unblinded, what starts will be run to compare test vs control arms
What biomarker is used for multiple myeloma?
Free light chain (FLC)
→ antibodies have 2LC and 2HC, for every Ig released by plasma cell you release 4 FLC
→ powerful inflammatory mediators
→ measure B cell regulation
What can the outcomes of clinical trails be?
Survival / death → easiest to measure, defined clear event
Disease status (chronic diseases) → progression free survival - how long did it take for the disease to get worse
→ relapse free survival - how long did the drug work for (in remission)
Measured with biomarkers → anything you can measure that tracks / monitors disease
When were the Spinraza clinical trials approved?
FDA approved Spinraza clinical trials in 2011 (same year as mouse work published - important for collaboration)
First trial → type II / III patients (not type I as really ill and susceptible)
Second trial → type I patients (phase I-III)
What contributes to making Spinraza expensive?
Delivery method - injection into spinal cord
→ need operative theatre, surgical team, anaesthetist
What did the phase 1 trial of Spinraza (2016) in children with SMA demonstrate?
Muscle function improved 29-85 days post injection at 9mg
→ thus moved onto infants
What is the biomarker used for SMA clinical trials?
Hammersmith functional test → measurement of motor function
What did the Spinraza clinical trials (2017) in infants with SMA demonstrate?
There was a statistical benefit to Spinraza (hazard ratio 0.53, p value 0.005
Probability of event-free survival ~30% better
→ looking at whether they died or required permanent assisted ventilation - lost so many motor neurone that lost control of intercostal muscles
Overall survival ~40% with control died, ~20% with drug died → not perfect as not everyone responds well
Trial halted - immoral to keep giving kids the placebo due to positive effects
→ FDA agreed unethical to not treat placebo arm, orphan drug approval given
What is orphan drug approval?
FDA can award orphan drug approval if a therapy is the only potential drug before phase III trials have been performed
How are clinical trials generally measured?
Event Analysis → relative risk and logistic regression
Time to Event Analysis → Kaplan Meier and cox regression
What is relative risk/odds ratio?
Both are a comparison of risk occurring in a test group v a control group (event analysis)
What is hazard ratio?
A modified relative risk/odds ratio that is dependant on time (time to event analysis)
What is logistic regression?
Measures variables (biomarkers) at the start and end of a trial
→ doesn’t tell you when, but tells you if they had an event during the trial
→ e.g. odds ratio at start and end
What is Cox regression?
Measures variables (biomarkers) at defined time points e.g. calculate OR every month
→ tells you how quickly you’re likely to progress
→ average OR = OR for a specific time interval = Hazard ratio - predicts the chance of progression at the next timepoint
How do you calculate risk ratio?
RR = (A/(A+B)) / (C/C+D))
→ risk of an event in the test arm compared to the risk of an event in the control arm
3 = 200% increase in the risk
1 = baseline - no increase in the risk
0.5 = 50% decrease in the risk
Below 1 = fewer events in the test arm
→ what you want for treatments
Always cohort specific → the estimated RR from a study will not be the real RR - outcome of an experiment not the population level
What are confidence intervals?
Provide an estimate of cohort variability → e.g. how can you trust OR/RR/HR is accurate
→ if you repeat in a large group high chance of getting same risk
An estimate of RR from a clinical trial has imprecision → confidence intervals indicate the precision of the estimate
→ usually 95% - interval contains the true value within probability of 95%
→ confidence intervals large when cohort is small / smaller in larger trials
If CI cross 1 threshold → can’t trust relative risk (i.e. drug only sometimes beneficial) - displaying too much cohort variability
