1. Using Biomarkers to Measure Alcohol Use
Sarah Cook
London School of Hygiene & Tropical Medicine
Using biomarkers in research on health workshop 20th February 2015

2. Alcohol Consumption
Alcohol consumption is associated with many negative effects including physical, psychological and social problems
Alcohol research needs valid measures of alcohol consumption

3. Measurement of Alcohol Consumption
Population level data on sales, tax and production
Self-reported from survey data
Volume of ethanol
Drinking pattern
Drinking behaviours (e.g. drunkenness, hangover)

4. Self-reported volume of ethanol
Quantity- Frequency
Usual amount X Frequency X Drink strength
(number of drinks) (drinking days) (ethanol content)
Graduated- Frequency
Recent recall

5. Problems with measuring alcohol Consumption
Measurement error
Difficult to remember
Difficulty in calculating volumes
What is a “Standard drink”?
Shared containers
Variation in strength of drinks
Social desirability

6. Alcohol Biomarkers
Using self-reported data on alcohol use is problematic
More objective measures of alcohol use are needed
Alcohol consumption has an effect on a wide range of biological parameters

7. Alcohol Biomarkers Blood Urine Hair Liver enzymes (GGT, AST, ALT)
Carbohydrate deficient transferrin (CDT)
High Density Lipoprotein cholesterol (HDL)
Mean corpuscular volume (MCV)
Urine
5-Hydroxtryptophol
Hair
Ethyl glucuronide

8. Liver Enzymes Gamma glutamyl transferase (GGT)
Aspartate Transanimase (AST)
Alanine Transanimase (ALT)

9. GGT Serum GGT is most commonly used alcohol biomarker
Raised with chronic heavy drinking (on average g ethanol/day for several weeks) but this can be influenced by other factors (age, gender).
More sensitive to alcohol than other liver enzymes but still generally low
Not specific – raised by many other factors
10-25 UK alcohol units/day

10. Carbohydrate Deficient Transferrin (CDT)
Variant of serum glycoprotein transferrin produced in the liver
Increased with chronic heavy drinking (60-80g ethanol daily over at least 2 weeks)
Several factors effect relationship between alcohol and CDT e.g. gender, smoking, body mass index
More specific than GGT but similar sensitivity
Not strongly correlated wit GGT so they could be used in combination.
Not raised in non alcoholic liver disease but raised with primary biliary cirrhosis, chronic acute hepatitis, severely decompensated liver disease

11. Limitations of alcohol biomarkers
There are lots of potential alcohol biomarkers but no gold standard measure has been found
Sensitivity and specificity generally low
Only raised with heavy, sustained drinking
Relationship between alcohol intake and biomarker may be influenced by other factors (interactions)
Difficult to interpret raised levels in terms of actual volume consumed, drinking pattern

12. Are alcohol biomarkers useful?

13. Association between alcohol use and cardiovascular risk factors in Russian Men: An Example of the Use of Alcohol Biomarkers

14. Study Design Data from the Izhevsk Family Study
Cross-sectional survey of men aged living in the city of Izhevsk, Russia (2008-9)
Men and a proxy completed a questionnaire which included very detailed questions on alcohol consumption
Attended a health check which included a blood test
Sample size N=978 men

15. Outcome: Cardiovascular risk factors
Hypertension (binary)
Serum lipids
High density lipoprotein cholesterol (mmol/litres)
Low Density lipoprotein cholesterol (mmol/litres)
Hypertension mean SBP>139mm hg or mean DBP>89 mm Hg or taking prescribed anti-hypertensive medication

16. Alcohol measures: self-reported spirit intake
Usual volume of spirits
0.76
1.00
Spirit Intake
Maximum volume of spirits
0.49
Frequency of drinking spirits

17. Alcohol measures- proxy reported acute dysfunctional drinking
Frequency of hangover
0.84
0.93
Frequency of excessive drunkenness
Acute alcohol-related dysfunction
0.84
Frequency of sleeping in clothes because of drunkenness
0.85
Frequency of failing family or personal obligations because of drinking

18. Alcohol measures- alcohol biomarkers
Log Gamma-glutamyl transferase (GGT)
Log Carbohydrate Deficient Transferrin (CDT)

19. Alcohol and Hypertension

20. Adjusted for age,education, level of amenities, marital status, employment status, smoking, physical activity and body mass index
Ors for continuous variables
Adjusted for age, education, level of amenities, marital status, employment status, smoking, physical activity and body mass index

21. Alcohol and Hypertension
Alcohol Use Measure
Odds ratio* (95% CI)
Spirit intake
1.27 (1.08, 1.49)
Acute alcohol-related dysfunction
1.33 (1.14, 1.56)
GGT
2.26 (1.79, 2.87)
CDT
1.74 (1.43, 2.13)
Odds ratios refer to change per unit increase in latent variables/biomarkers
*Adjusted for age, education, level of amenities, marital status, employment status, smoking, physical activity and body mass index

22. Alcohol use and HDL-C

24. Alcohol and HDL Alcohol Use Measure Coefficient* (95% CI)
Spirit intake
0.18 (0.15, 0.20)
Acute alcohol-related dysfunction
0.16 (0.13, 0.19)
GGT
0.20 (0.18, 0.23)
CDT
0.31 (0.29, 0.33)
*Adjusted for age, education, level of amenities, marital status, employment status, smoking, physical activity and body mass index

25. Alcohol Use and LDL-C

27. Alcohol and LDL-C Alcohol Use Measure Coefficient* (95% CI)
Spirit intake
-0.11 (-0.17, -0.05)
Acute alcohol-related dysfunction
-0.12 (-0.19, -0.06)
GGT
0.04 (-0.02, 0.11)
CDT
-0.16 (-0.23, -0.08)
*Adjusted for age, education, level of amenities, marital status, employment status, smoking, physical activity and body mass index

28. Why does GGT look different?
GGT is raised for many reasons (not just alcohol)
These include age, obesity, smoking, use of certain medications, non alcoholic liver disease and diabetes
GGT associated with cardiovascular disease even in non-drinkers
Is it an appropriate biomarker for the question?

29. Conclusions
Alcohol biomarkers are not a simple solution to the issues of how to measure alcohol use
BUT they can be useful alongside data on alcohol use from other sources to improve overall picture/provide validation
Caution is needed when your biomarker is raised for many reasons as you may see associations not due to the factor of interest
The best way to use biomarkers in alcohol research is still a question to explore

30. Acknowledgements
David Leon
Bianca De Stavola