Alcohol & Alcoholism Vol. 44, No. 5, pp. 464–467, 2009
Advance Access publication 17 June 2009
ASSESSMENT AND DETECTION
Preliminary Evaluation of Phosphatidylethanol and Alcohol Consumption in Patients with Liver Disease
Scott H. Stewart1,2,∗, Adrian Reuben3, Walter A. Brzezinski2, David G. Koch3, Jan Basile4, Patrick K. Randall1and
Peter M. Miller1
1Center for Drug and Alcohol Programs, Medical University of South Carolina, Charleston, SC 29425, USA,2Division of General Internal Medicine, Medical
University of South Carolina, Charleston, SC 29425, USA,3Liver Disease and Liver Transplant Service, Medical University of South Carolina, Charleston, SC
29425, USA and4Department of Medicine, Ralph H. Johnson VA Medical Center, Charleston, SC 29425, USA
∗Corresponding author: Center for Drug and Alcohol Programs, PO Box 250861, 67 President Street, Charleston, SC 29425, USA. Tel: +1-843-792-5226;
Fax: +1-843-792-7353; E-mail: firstname.lastname@example.org
(Received 12 February 2009; first review notified 21 April 2009; in revised form 4 May 2009; accepted 26 May 2009; advance access publication 17 June 2009)
Abstract — Aims: The goal of this preliminary study was to evaluate the relationship between blood phosphatidylethanol (PEth) and
recent drinking in patients with liver disease and hypertension. Methods: Twenty-one patients with liver disease and 21 patients with
essential hypertension were recruited at an academic medical center. Alcohol consumption was estimated using validated self-report
methods, and blood PEth was measured by HPLC-MS/MS at a contracted laboratory. Nonparametric comparisons were made between
abstainers/light drinkers, moderate drinkers consuming between 1 and 3 drinks per day, and those drinking above this level. Regression
methods were used to estimate the effects of liver disease, gender, and age on the relationship between PEth and alcohol use, and
to estimate the strength of the linear relationship between PEth and drinking. Results: PEth differed significantly between the three
drinking groups (P < 0.001). The relationship between PEth and alcohol did not differ between hypertension and liver disease patients
(P = 0.696), nor by gender and age. While there was substantial variability between subjects in the PEth concentration given a
similar level of reported drinking, the amount of ethanol consumed was strongly associated with the PEth concentration (P < 0.001).
Conclusion: Results support PEth measurement by HPLC-MS/MS as a promising marker of past 1- to 2-week moderate to heavy
alcohol consumption in patients with and without liver disease. PEth appears useful for differentiating abstinence or light drinking from
moderate to heavy consumption, but may have limited utility for differentiating moderate from heavy alcohol use.
Chronic heavy drinking is a direct cause of many medical con-
ditions due to toxic effects of ethanol or its metabolism (Saitz,
ical and research settings mainly relies on self-report (National
Institute on Alcohol Abuse and Alcoholism, 2007). While very
useful, self-report methods are in general subject to potential
which may be magnified in patients with alcohol-associated
illnesses. Such individuals may wish to conceal their heavy
drinking or prior unsuccessful attempts to control their drink-
ing due to associated stigma, and may suffer from cognitive
impairment due to chronic alcohol abuse or other factors.
Biomarkers of alcohol consumption may aid detection and
treatment efforts in patients with alcohol-related conditions
ethanol exposure. However, biomarkers typically do not have
ticularly with liver disease, where other factors besides alco-
hol can influence markers such as gamma glutamyltransferase,
aminotransferases and red blood cell mean corpuscular volume
(Conigrave et al., 2003). The only FDA-approved marker for
heavy drinking is the percent carbohydrate-deficient transfer-
rin (Anton, 2001), but this requires ∼60 g of ethanol per day
(i.e. 4–5 standard drinks) to become elevated, has a sensitiv-
ity of 60–70%, and loses its otherwise very high specificity
in advanced liver disease (Heinemann et al., 1998; DiMartini
oxidative product of ethanol elimination that is formed from
phosphatidylcholine and ethanol via the action of phospholi-
pase D (Gustavsson, 1995), which can occur extra-hepatically
such as in human erythrocytes (Butikofer et al., 1993). Given
this mechanism of formation, the relationship between PEth
and drinking may be less influenced by liver function rela-
tive to other biomarkers. Circulating PEth is found primarily
in the red blood cell fraction (Varga et al., 2000), has a de-
tection window of ∼1–3 weeks following drinking cessation
measured using an evaporative light scattering detector assay
(Hartmann et al., 2007). This PEth assay has been used, for ex-
ample, to detect unreported drinking in emergency department
patients (Kip et al., 2008). An alternative mass spectrometer-
based assay may be more sensitive for detecting low PEth
concentrations (Gunnarsson et al., 1998) and may allow for
the detection of even moderate consumption. As such, the lat-
ter assay may reveal less heavy but still harmful drinking in
patients with alcohol-associated disorders, and may also pre-
dict the amount of alcohol consumed. However, prior to use
as a marker of alcohol consumption in clinical care or clinical
research, the relationship between PEth and drinking must be
characterized in relevant populations that show a wide range
of recent alcohol exposure. This manuscript describes our pi-
lot testing of PEth in patients with two conditions that can
be caused or aggravated by frequent moderate to heavy drink-
ing, namely liver disease and hypertension. Importantly, liver
impairment, by influencing the major oxidative mechanisms
for ethanol elimination that occur primarily in the liver, may
modify the association between the amount of alcohol con-
sumed and PEth. Since hypertension should not modify this
C ?The Author 2009. Published by Oxford University Press on behalf of the Medical Council on Alcohol. All rights reserved
Evaluation of PEth465
association, hypertension patients were also viewed as a con-
be related to liver function.
Subjects and recruitment strategy
We recruited 21 consecutively eligible and consenting patients
with liver disease from inpatient and outpatient hepatology
services, aiming to include a minimum of 10 subjects with re-
cent drinking. The main exclusion criterion was cognitive dys-
function of sufficient severity to preclude informed consent.
The clinically diagnosed etiology, alcohol or otherwise, was
confirmed by a review of the medical record with the attend-
ing hepatologists who were responsible for the liver disease
patients, and the severity of liver disease was estimated us-
ing the Model for End-Stage Liver Disease (MELD) (Kamath
et al., 2001). The MELD score is calculated from the inter-
national normalized ratio (INR), creatinine and total bilirubin,
and is used in the United States for prioritizing liver trans-
plantation. In order to compare the PEth–drinking relationship
in a group without liver disease and to assess the potential of
PEth as a screen for alcohol-related hypertension, we recruited
an identical number of hypertension patients. Because differ-
entiating abstinence from light to moderate alcohol use is not
of concern in hypertension, this sample was limited to current
drinkers, defined as those who reported having at least one
drink per week. Potential subjects were asked to participate in
a study that included blood tests and a survey about their al-
cohol drinking. Those who provided signed informed consent
were enrolled and interviewed, and received $25 compensation
following completion. The protocols were approved by the
university Institutional Review Board (federal-wide assurance
number for protection of human subjects 00001888).
Alcohol consumption data
Alcohol consumption for the month prior to enrollment was
estimated by personnel with clinical alcohol research experi-
ence using a timeline follow-back method (Sobell and Sobell,
1992). This approach uses typical drinking habits and memory
cues to obtain daily drinking estimates. The estimated quan-
tity of alcoholic beverage consumed in US fluid ounces was
converted to milliliters, and grams of ethanol were calculated
from the ethanol content of the beverages and the density of
ethanol (i.e., 0.79 g/ml). In presenting the results of this study,
we defined one standard drink as 14 g of ethanol (equivalent,
for example, to a 12-oz beer containing 5% ethanol, or a 1.5-oz
serving of 80 proof spirits).
Collection and measurement of blood PEth
Ten mL of whole blood was drawn into a tube containing
100 mg sodium fluoride and 20 mg potassium oxalate.
Samples were shipped fresh or following a maximum of
2 weeks storage at −20◦C to a contracted laboratory (US
Drug Testing Laboratories, Des Plaines, IL, USA) that was
masked to the status of the study subjects. The HPLC-MS/MS
assay detected the isomers 1-palmitoyl-2-oleoyl-sn-glycero-
3-phosphatidylethanol and 1-oleoyl-2-palmitoyl-sn-glycero-3-
phosphatidylethanol, which included ∼40% of total PEth
(Gunnarsson et al., 1998). The within-batch coefficient of vari-
ation for the assay was 9%, the between-batch coefficient of
variation 13% and the statistical limit of quantification 19.8
ng/ml.A testwas considered positive at20 ng/ml or higher (i.e.
above the assay cutoff at which PEth can reliably be detected
over background noise).
Data analysis included non-parametric methods for PEth com-
parison between subjects reporting abstinence or an average
of less than one drink per day (n = 17), those reporting an
average of one to three drinks per day (n = 14) and those re-
porting more than three drinks per day during the prior 2 weeks
(n = 11). Pairwise comparisons were made without adjustment
for multiple comparisons due to the preliminary nature of this
work. Regression methods were also used to further describe
the relationship between PEth and drinking, and to explore
for any modification of PEth synthesis by diagnosis (i.e., liver
disease versus hypertension), gender and age (dichotomized at
the median of 52). The PEth concentration served as the de-
pendent variable, and estimated grams of alcohol consumed
in the 2 weeks preceding measurement was the main indepen-
dian) and gender to explore modification of the PEth–alcohol
relationship by these factors. The distribution of PEth included
a number of censored results (i.e. PEth was not quantifiable
between 0 and 20 ng/ml). Because of this distribution, we used
tobit rather than linear regression, which includes adjustment
for the unknown distribution of PEth below the limit of quan-
titation (Austin et al., 2000).
Descriptive characteristics of the subjects are listed in Table 1.
The overall median age was 52, about two-thirds were male
and 70% had quantifiable PEth. For the liver disease patients,
two-thirds had a diagnosis of alcoholic liver disease, and the
MELD scores indicate inclusion of patients with both mildly
and severely impaired liver function. Box plots of the PEth re-
sults for subjects in each drinking category are shown in Fig. 1.
There were 17 subjects in the abstinent to less than one drink
jects), 14 in the group averaging between one and three drinks
per day (4 liver disease subjects and 10 hypertension subjects)
and 11 in the group averaging more than three drinks per day
(6 liver disease subjects and 5 hypertension subjects).
Despite the inclusion of two alcoholic liver disease sub-
jects who reported at least 2 months of abstinence but had
quantifiable PEth (specifically, 44 ng/ml and 427 ng/ml), there
was a statistically significant difference in PEth findings be-
tween the three groups (Kruskal–Wallis P < 0.001, see Fig. 1).
Pairwise comparisons demonstrated significant differences be-
consuming one to three drinks per day (P = 0.002), those con-
suming less than one drink per day versus those consuming
more than three drinks per day (P < 0.001) and those consum-
ing one to three drinks per day versus those consuming more
than three drinks per day (P = 0.020). Spearman’s correlations
between alcohol consumption and PEth were 0.76 (P < 0.001)
466 Stewart et al.
Table 1. Subject characteristics
Characteristic Liver disease patients (n = 21)
165 (0–726; 0–2816)
63 (0–431; 0–3710)
Hypertension patients (n = 21)
374 (141–539; 45–1602)
86 (28–206; 0–1016)
Median age (range)
Non-Hispanic white (%)
Current drinkers (%)
Median grams ethanol in past 2 weeks (interquartile range; range)
Quantifiable Peth (%)
Median PEth (ng/ml) (interquartile range; range)
Median MELD score (range)a
Clinical diagnosis of alcoholic liver diseaseb(%)
aMELD indicates the Model for End-Stage Liver Disease severity score.
bThe 14 patients with alcoholic liver disease include 5 patients with co-morbid chronic Hepatitis C.
Fig. 1. Box plot of PEth concentrations by drinking category (tails indicate
the range, horizontal lines indicate the median and interquartile range and solid
squares indicate the mean). The asterisk in the plot for the heaviest drinking
group indicates that one outlying PEth value of 3710 ng/ml was truncated to
0.001) and each pairwise comparison (P < 0.05) were statistically significant.
for liver disease subjects and 0.66 (P =0.001) for hypertension
subjects. Consistent with these similar correlations, the Tobit
regression also suggested that the relationship between drink-
ing and PEth concentration was comparable in the liver disease
and hypertension subjects (P-value for interaction = 0.696),
as well as similar for men and women (P = 0.221), and unaf-
fected by age (P = 0.591). There was a statistically significant
association between PEth and grams of ethanol consumed in
the prior 2 weeks (regression slope = 0.81, P < 0.001). Re-
gression results were repeated without an outlying observation
having a PEth of 3710 ng/ml, resulting in a regression slope =
0.54 (P < 0.001).
This preliminary study suggested that blood PEth predicted
alcohol consumption in patients with liver disease and
hypertension, and that the effect of alcohol consumption on the
PEth concentration is not dependent on liver function. Higher
there was substantial variability between individuals. This sug-
gests that a single PEth measurement is useful for detecting
recent drinking but is a crude estimate of the amount of alcohol
Early phase validation studies primarily involving heavy
drinkers in detoxification facilities have mainly utilized a dif-
ferent PEth assay (Hansson et al., 1997; Aradottir et al., 2006;
Hartmann et al., 2007). With an evaporative light-scattering
detection system, PEth has been ∼95% sensitive and 100%
specific in differentiating very heavy drinkers from known
abstainers and social drinkers. Our results suggest that PEth
quantification using the mass spectrometer-based assay will
detect even moderate consumption in patients with potentially
alcohol-associated disorders. However, due to the apparent
variability in PEth synthesis between individuals, specificity
may be limited for differentiating moderate from heavy drink-
ing. Regarding specificity for any ethanol ingestion, we did
find quantifiable PEth in two self-reported abstainers with al-
coholic liver disease. It is possible that these represented false
positive PEth results. However, since ethanol is required for
PEth synthesis, these cases likely represent an underestimation
of actual alcohol ingestion and thus biomarker detection of
recent drinking that was missed by detailed self-report meth-
ods. Supporting this, no non-alcoholic liver disease patients
who reported abstinence had detectable PEth, and the lowest
reported consumption resulting in PEth detection among the
hypertension subjects was equivalent to about one drink per
The main strengths of this report are the evaluation of a
novel alcohol consumption biomarker in medical patients with
conditions that can be caused by alcohol consumption, the
comparison between patients with and without liver disease,
and the representation of both light to moderate and heavier
drinkers. In situations where highly objective estimates of al-
cohol drinking are particularly desirable (e.g. liver transplant
evaluations, clinical research protocols, confirmation of alco-
hol use in both acute and chronic alcohol-related illness), PEth
may be useful in determining the presence or absence of at
least moderate consumption. PEth appears particularly useful
in patients with liver disease, in whom many alcohol biomark-
ers (e.g. aminotransferases, gamma-glutamyltransferase, mean
red blood cell volume, percent carbohydrate-deficient trans-
ferrin) may be elevated due to the liver disease itself
Evaluation of PEth 467 Download full-text
(Conigrave et al., 2003). PEth also appears sensitive for mod-
erate drinking, with many of the aforementioned markers only
responding to heavy drinking. Conversely, unlike some other
imal drinking or unintentional ethanol ingestion will likely not
result in detectable PEth (Palmer, 2009). Given the preliminary
nature of this work, future studies will need to confirm these
findings in larger samples, evaluate the effects of drinking pat-
on PEth concentration and identify sources of variation in PEth
other than recent consumption, such as a history of chronic
heavy drinking (Varga and Alling, 2002). For widespread clin-
ical use, it will also be important to develop and validate assays
that will be practical for typical clinical laboratories. The re-
cent development of monoclonal antibodies to PEth suggests
that this will be feasible (Nissinen et al., 2008).
In summary, this pilot study and previously published re-
search suggests that PEth is a highly promising marker of alco-
ment. Relative to many other biomarkers, the modest drinking
threshold required for mass spectrometry detection, the half-
lifeof∼4days,and theprobable validitydespite impairedliver
ing drinking in patients with alcohol-associated conditions.
Acknowledgements — This work was supported by a Career Development Award from the
National Institute on Alcohol Abuse and Alcoholism (K23AA014188).
Allen JP. (2003) Use of biomarkers of heavy drinking in health care
practice. Mil Med 168:364–7.
Anton RF. (2001) Carbohydrate-deficient transferrin for detection and
monitoring of sustained heavy drinking. What have we learned?
Where do we go from here? Alcohol 25:185–8.
Aradottir S, Asanovska G, Gjerss S et al. (2006) Phosphatidylethanol
(PEth) concentrations in blood are correlated to reported alcohol
intake in alcohol-dependent patients. Alcohol Alcohol 41:431–7.
Austin PC, Escobar M, Kopec JA. (2000) The use of the Tobit model
for analyzing measures of health status. Qual Life Res 9:901–10.
drinking outcomes in clinical trials. J Stud Alcohol 61:55–63.
Butikofer P, Yee MC, Schott MA et al. (1993) Generation of phos-
phatidic acid during calcium-loading of human erythrocytes. Evi-
dence for a phosphatidylcholine-hydrolyzing phospholipase D. Eur
J Biochem 213:367–75.
Conigrave KM, Davies P, Haber P et al. (2003) Traditional markers
of excessive alcohol use. Addiction 98(Suppl 2):31–43.
Del Boca FK, Darkes J. (2003) The validity of self-reports of alco-
hol consumption: state of the science and challenges for research.
Addiction 98(Suppl 2):1–12.
ferrin in abstaining patients with end-stage liver disease. Alcohol
Clin Exp Res 25:1729–33.
Gunnarsson T, Karlsson A, Hansson P et al. (1998) Determination
of phosphatidylethanol in blood from alcoholic males using high-
performance liquid chromatography and evaporative light scatter-
ing or electrospray mass spectrometric detection. J Chromatogr B
Biomed Sci Appl 705:243–9.
Gustavsson L. (1995) Phosphatidylethanol formation: specific effects
of ethanol mediated via phospholipase D. Alcohol Alcohol 30:391–
Hansson P, Caron M, Johnson G et al. (1997) Blood phos-
phatidylethanol as a marker of alcohol abuse: levels in alcoholic
males during withdrawal. Alcohol Clin Exp Res 21:108–10.
transpeptidase, mean corpuscular volume and carbohydrate-
deficient transferrin. Addict Biol 12:81–4.
stage liver disease before and after liver transplantation. Alcohol
Clin Exp Res 22:1806–12.
Kamath PS, Wiesner RH, Malinchoc M et al. (2001) A model to
predict survival in patients with end-stage liver disease. Hepatology
Kip MJ, Spies CD, Neumann T et al. (2008) The usefulness of direct
ethanol metabolites in assessing alcohol intake in nonintoxicated
male patients in an emergency room setting. Alcohol Clin Exp Res
National Institute on Alcohol Abuse and Alcoholism. (2007) Helping
US Department of Health and Human Services.
Nissinen AE, Makela SM, Vuoristo JT et al. (2008) Immunologi-
cal detection of in vitro formed phosphatidylethanol—an alcohol
biomarker—with monoclonal antibodies. Alcohol Clin Exp Res
Palmer RB. (2009) A review of the use of ethyl glucuronide as a
marker for ethanol consumption in forensic and clinical medicine.
Semin Diagn Pathol 26:18–27.
Saitz R. (2003) Overview of medical and surgical complications. In
Graham AW, Schultz TK, Mayo-Smith MF, Ries RK, Wilford
BB (eds). Principles of Addiction Medicine. Chevy Chase, MD:
American Society of Addiction Medicine, 1027–52.
Sobell L, Sobell M. (1992) Timeline follow-back: a technique for as-
sessing self-reported ethanol consumption. In Litten RZ, Allen JP
ical Methods. Totowa, NJ: Humana Press, 41–72.
Varga A, Alling C. (2002) Formation of phosphatidylethanol in vitro
in red blood cells from healthy volunteers and chronic alcoholics.
J Lab Clin Med 140:79–83.
Varga A, Hansson P, Johnson G et al. (2000) Normalization rate and
cellular localization of phosphatidylethanol in whole blood from
chronic alcoholics. Clin Chim Acta 299:141–50.