The Pharmacogenetics of Warfarin: Assessing quality of studies in systematic review

1
The Pharmacogenetics of Warfarin Assessing
quality of studies in systematic review

• Andrea L Jorgensen
• Paula R Williamson Munir Pirmohamed

2
Outline

• What is pharmacogenetics ?
• Quality assessment of genetic association
  studies
• Warfarin an example
• Extra issues in pharmacogenetic studies
• Conclusions

3
Genetic association studies

• Study to investigate association between
  variation in an individuals DNA and a
  particular outcome e.g. disease
• Outcome can be response to a drug
  pharmacogenetics
• Pharmacogenetics a study of the influence of
  genetic variation on drug response in patients by
  correlating genetic variation with a drugs
  efficacy or toxicity

4
Problems with replication

• Major advancements in knowledge of genetic
  variation (Human Genome Project/HapMap)
• Improved genotyping technologies
• Increased opportunities for evaluating role of
  genetic variation in human disease/inter-individua
  l variation in response to drugs
• Even so.role of genetic association studies
  remains controversial
• Few novel insights so far
• Failure to confirm initial positive findings
• Lack of power (small sample size)
• Differences in biological factors
• Quality issues also contribute e.g study
  design analysis interpretation

5
Quality under question

• Meta-analysis of many smaller, some severely
  underpowered studies advised
• Biological factors addressed (at least partly)
  in meta-analysis e.g. meta-regression
• Quality assessment essential element of
  systematic review
• Evaluate appropriateness of combining results
  in meta-analysis
• Insight into reliability of results
• Demonstration of importance warfarin
  systematic review

6
What is warfarin ?

• Anticoagulant of choice in UK
• Efficacy well demonstrated in many RCTs and
  MAs. However
• large inter-individual variability in
  maintenance dose required to achieve therapeutic
  anticoagulation from 0.5mg/day to over 10mg/day
• Several environmental factors identified
• Not enough to explain genetic variants must play
  a part too
• 5 studies
• CYP2C9 VKORC1 environmental factors
  50 variation
• Varying populations and environmental factors
• Currently undertaking systematic review of
  papers investigating association between CYP2C9
  variants VKORC1 variants or both

7
VKORC1 Studies

• 44 investigating CYP2C9 alone
• 3 investigating at VKORC1 alone
• 17 investigating both
• Focus on 20 VKORC1 as example. Sample size

8
Results from assessing quality Design Issues

• Type of study
• Retrospective 80
• Prospective 20
• How were genetic variants investigated chosen ?
• Re-sequencing/random sample of patients 45
• All known variants from genetic databases 15
• Previous genetic association studies 30
• Not stated 10

9
Results from assessing quality analysis issues

• Was genotyping method described ?
• Yes 100
• No 0
• What quality control measures were described ?
• Reference to paper where genotyping assay
  reliability assessed 10
• Re-genotyping all samples/random subset of
  samples 20
• None 70
• Was missing data acknowledged ?
• Yes 151
• No 852

• Patients where genotypes not available excluded
  from the analyses
• For 2 of these studies genotype frequencies
  clearly did not add up to give sample size. In a
  further 9 the genotype frequencies added up to
  give sample size.

10
Pharmacogenetics extra questions

• What are additional issues relevant to
  pharmacogenetics studies ?
• Compliance (Analysis issue)
• Main focus drug x genotype interaction
• More important than in drug RCT where focus
  typically pragmatic trying to understand true
  impact on biological process
• How much of drug being taken may well be key
  factor
• Not measured in any of the VKORC1 papers
• Choice of outcome (Design issue)
• Trying to capture efficacy and/or toxicity
• Larger scope for heterogeneity in outcome
  definition
• Increased chance of outcome reporting bias

11
Choice of outcome Results

• Primary outcomes investigated
• How was stable dose outcome defined ?
• Same dose INR in therapeutic range for 3
  months 20
• Same dose INR in therapeutic range for 3
  consecutive clinic visits (some specify minimum
  period) 15
• Other definitions (same dose with INR in range
  for varying periods, from 3 weeks to 6
  months)
  30
• Not defined
  15
• Not measured 20

12
Conclusions

• Quality concerns reflected in warfarin papers
• Extra issues relevant to pharmacogenetics
  also poor
• Underpowered studies with high risk of bias
• Meta-analysis very useful tool in
  pharmacogenetics however
• essential to first assess quality of included
  studies
• Undertake meta-analysis in light of findings
  from assessment
• Must improve quality of future studies e.g.
  warfarin so we can rely on results