1 of 43

Predicting drug repurposing opportunities �for Alzheimer’s disease prevention �using genome-wide association study dataVenexia Walker

2 of 43

Why use genome-wide association study data?

Methods using genome-wide association study data

    • Mendelian randomization
    • Colocalization

Predicting drug repurposing opportunities for Alzheimer’s Disease prevention

    • Antihypertensive medications
    • Metformin

A note on triangulation of evidence

Future directions

3 of 43

Why use genome-wide association study data?

Methods using genome-wide association study data

    • Mendelian randomization
    • Colocalization

Predicting drug repurposing opportunities for Alzheimer’s Disease prevention

    • Antihypertensive medications
    • Metformin

A note on triangulation of evidence

Future directions

4 of 43

Source: Calcoen et al. (2015) What does it take to produce a breakthrough drug? Nature Reviews Drug Discovery

5 of 43

“We found that, among well-studied indications, the proportion of �drug mechanisms with direct genetic support increases significantly �across the drug development pipeline, from 2.0% at the preclinical stage �to 8.2% among mechanisms for approved drugs, and varies �dramatically among disease areas.”

Source: Nelson et al. (2015) The support of human genetic evidence for approved drug indications. Nature Genetics.

6 of 43

Source: https://www.genome.gov/about-genomics/fact-sheets/DNA-Sequencing-Costs-Data

7 of 43

Source: Lambert et al. (2013) Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer's disease. Nature Genetics.

8 of 43

Source: Elsworth et al. (2020) The MRC IEU OpenGWAS data infrastructure. bioRxiv

9 of 43

Source: Elsworth et al. (2020) The MRC IEU OpenGWAS data infrastructure. bioRxiv

10 of 43

Why use genome-wide association study data?

Methods using genome-wide association study data

    • Mendelian randomization
    • Colocalization

Predicting drug repurposing opportunities for Alzheimer’s Disease prevention

    • Antihypertensive medications
    • Metformin

A note on triangulation of evidence

Future directions

11 of 43

Source: Evans et al. (2015) Mendelian Randomization: New Applications in the Coming Age of Hypothesis-Free Causality. Annual Reviews Genomics Hum Genetics.

12 of 43

Instrument

SNP(s) selected to proxy a drug target

Exposure

Drug target

Outcome

Alzheimer’s disease

(Un)measured confounders

Socioeconomic position, BMI, smoking, …

13 of 43

Source: Hemani et al. (2018) The MR-Base platform supports systematic causal inference across the human phenome. eLife.

 

14 of 43

Source: Walker et al (2017) Mendelian randomization: a novel approach for the prediction of adverse drug events and drug repurposing opportunities. International Journal of Epidemiology.

Strengths of Mendelian randomization in this setting include:

  • Can be used both before and after approval of a drug

  • More robust to non-genetic confounding, including confounding by indication

  • More robust to reverse causality

  • Able to predict combined effects of drugs

  • Aids the distinction of mechanism and biomarker effects

15 of 43

Source: Walker et al (2017) Mendelian randomization: a novel approach for the prediction of adverse drug events and drug repurposing opportunities. International Journal of Epidemiology.

Limitations of Mendelian randomization in this setting include:

  • Rare effects may not be detected

  • Horizontal pleiotropy

  • Estimates are of lifelong exposure

  • Weak instrument bias

  • Subject to availability of appropriate genome-wide association study data

16 of 43

Why use genome-wide association study data?

Methods using genome-wide association study data

    • Mendelian randomization
    • Colocalization

Predicting drug repurposing opportunities for Alzheimer’s Disease prevention

    • Antihypertensive medications
    • Metformin

A note on triangulation of evidence

Future directions

17 of 43

Source: Giambartolomei et al. (2014) Bayesian Test for Colocalisation between Pairs of Genetic Association Studies Using Summary Statistics. PLOS Genetics.

Colocalization quantifies the likelihood of a shared causal variant by calculating the posterior support for five hypotheses:

• H0 : No association with phenotype A or phenotype B

• H1 : Association with phenotype A but not phenotype B

• H2 : Association with phenotype B but not phenotype A

• H3 : Association with phenotypes A and B, different causal variants

• H4 : Association with phenotypes A and B, same causal variant

18 of 43

Source: Giambartolomei et al. (2014) Bayesian Test for Colocalisation between Pairs of Genetic Association Studies Using Summary Statistics. PLOS Genetics.

19 of 43

Why use genome-wide association study data?

Methods using genome-wide association study data

    • Mendelian randomization
    • Colocalization

Predicting drug repurposing opportunities for Alzheimer’s Disease prevention

    • Antihypertensive medications
    • Metformin

A note on triangulation of evidence

Future directions

20 of 43

Step 1

Identify drugs of interest, their protein targets

and corresponding genes

Step 2

Select instruments for each protein target �using eQTL data

Step 3

Validate against the existing indication of the drug and retain SNPs with evidence of an effect

Step 4

Perform two-sample Mendelian randomization analysis of interest

21 of 43

Step 1

Identify drugs of interest, their protein targets

and corresponding genes

Step 2

Select instruments for each protein target �using eQTL data

Step 3

Validate against the existing indication of the drug and retain SNPs with evidence of an effect

Step 4

Perform two-sample Mendelian randomization analysis of interest

22 of 43

Step 1

Identify drugs of interest, their protein targets

and corresponding genes

Step 2

Select instruments for each protein target �using eQTL data

Step 3

Validate against the existing indication of the drug and retain SNPs with evidence of an effect

Step 4

Perform two-sample Mendelian randomization analysis of interest

23 of 43

SNPs to represent each target were obtained by extracting the ‘best SNP’ for the corresponding gene in each tissue from GTEx, where the ‘best SNP’ was defined by GTEx as the variant with the smallest nominal p-value for a variant-gene pair.

Source: GTEx consortium. (2015) Genetic effects on gene expression across human tissues. Nature.

24 of 43

Step 1

Identify drugs of interest, their protein targets

and corresponding genes

Step 2

Select instruments for each protein target �using eQTL data

Step 3

Validate against the existing indication of the drug and retain SNPs with evidence of an effect

Step 4

Perform two-sample Mendelian randomization analysis of interest

25 of 43

Two-sample Mendelian randomization using data from GTEx for the instrument-exposure associations and a UK Biobank GWAS of systolic blood pressure for the instrument-outcome associations.

Source: Walker et al. (2019) Repurposing antihypertensive drugs for the prevention of Alzheimer’s disease: a Mendelian randomization study. International Journal of Epidemiology.

26 of 43

Step 1

Identify drugs of interest, their protein targets

and corresponding genes

Step 2

Select instruments for each protein target �using eQTL data

Step 3

Validate against the existing indication of the drug and retain SNPs with evidence of an effect

Step 4

Perform two-sample Mendelian randomization analysis of interest

27 of 43

Instrument

SNP(s) selected to proxy antihypertensive drug targets

Exposure

Systolic blood pressure

Outcome

Alzheimer’s disease

(Un)measured confounders

Socioeconomic position, BMI, smoking, …

GWAS based on 317,754

UK Biobank participants

IGAP GWAS Stage 1 results

Twelve drug classes identified in the BNF

28 of 43

Source: Walker et al. (2019) Repurposing antihypertensive drugs for the prevention of Alzheimer’s disease: a Mendelian randomization study. International Journal of Epidemiology.

29 of 43

 

Source: Walker et al. (2019) Repurposing antihypertensive drugs for the prevention of Alzheimer’s disease: a Mendelian randomization study. International Journal of Epidemiology.

30 of 43

Sources: Walker et al. (2019) Repurposing antihypertensive drugs for the prevention of Alzheimer’s disease: a Mendelian randomization study. International Journal of Epidemiology. � Gill et al. (2019) Use of Genetic Variants Related to Antihypertensive Drugs to Inform on Efficacy and Side Effects. Hypertension.

31 of 43

Source: Gill et al. (2019) Comparison with randomized controlled trials as a strategy for evaluating instruments in Mendelian randomization. International Journal of Epidemiology.

32 of 43

Why use genome-wide association study data?

Methods using genome-wide association study data

    • Mendelian randomization
    • Colocalization

Predicting drug repurposing opportunities for Alzheimer’s Disease prevention

    • Antihypertensive medications
    • Metformin

A note on triangulation of evidence

Future directions

33 of 43

Source: Zheng et al. (2021) Evaluating the impact of metformin, a multiple targets drugs, on risk of complex diseases: a phenome-wide Mendelian randomisation study. Submitted.

34 of 43

Source: Zheng et al. (2021) Evaluating the impact of metformin, a multiple targets drugs, on risk of complex diseases: a phenome-wide Mendelian randomisation study. Submitted.

35 of 43

Source: Zheng et al. (2021) Evaluating the impact of metformin, a multiple targets drugs, on risk of complex diseases: a phenome-wide Mendelian randomisation study. Submitted.

36 of 43

Source: Zheng et al. (2021) Evaluating the impact of metformin, a multiple targets drugs, on risk of complex diseases: a phenome-wide Mendelian randomisation study. Submitted.

37 of 43

Source: Zheng et al. (2021) Evaluating the impact of metformin, a multiple targets drugs, on risk of complex diseases: a phenome-wide Mendelian randomisation study. Submitted.

38 of 43

Why use genome-wide association study data?

Methods using genome-wide association study data

    • Mendelian randomization
    • Colocalization

Predicting drug repurposing opportunities for Alzheimer’s Disease prevention

    • Antihypertensive medications
    • Metformin

A note on triangulation of evidence

Future directions

39 of 43

“Triangulation is the practice of obtaining more reliable answers to research questions through integrating results from several different approaches, where each approach has different key sources of potential bias that are unrelated to each other. With respect to causal questions in aetiological epidemiology, if the results of different approaches all point to the same conclusion, this strengthens confidence in the finding.”

Source: Lawlor et al. (2016) Triangulation in aetiological epidemiology. International Journal of Epidemiology.

40 of 43

41 of 43

Why use genome-wide association study data?

Methods using genome-wide association study data

    • Mendelian randomization
    • Colocalization

Predicting drug repurposing opportunities for Alzheimer’s Disease prevention

    • Antihypertensive medications
    • Metformin

A note on triangulation of evidence

Future directions

42 of 43

(1) What is the optimal approach for deriving a drug target instrument?

(2) What disease and/or populations are not covered by genome-wide association studies?

(3) How do estimates compare with other forms of evidence, particularly randomized controlled trials, and how can these different forms of evidence be used together?

43 of 43

Predicting drug repurposing opportunities �for Alzheimer’s disease prevention �using genome-wide association study dataVenexia Walker