Assessing the drinking status of liver transplant patients with alcoholic liver disease


  • John P. Allen,

    Corresponding author
    1. Division of Addictions Research and Treatment, Department of Psychiatry and Behavioral Sciences,, Duke University, Durham, NC
    • Mid-Atlantic Mental Illness Research, Education, and Clinical Center (Veterans Integrated Service Network 6), Durham Veterans Administration Medical Center, Durham, NC
    Search for more papers by this author
  • Friedrich M. Wurst,

    1. Department of Psychiatry and Psychotherapy II, Christian Doppler Hospital, Paracelsus Medical University, Salzburg, Austria
    Search for more papers by this author
  • Natasha Thon,

    1. Department of Psychiatry and Psychotherapy II, Christian Doppler Hospital, Paracelsus Medical University, Salzburg, Austria
    Search for more papers by this author
  • Raye Z. Litten

    1. National Institute on Alcohol Abuse and Alcoholism, Bethesda, MD
    Search for more papers by this author

Address reprint requests to John P. Allen, Ph.D., M.P.A., Veterans Health Affairs, 2009 Carriage Court, Vienna, VA 22181. Telephone: 703-281-0760; E-mail:


The accurate assessment of drinking by patients with alcoholic liver disease is important both before and after liver transplantation. Unfortunately, self-reports by these individuals often underestimate their actual alcohol consumption. Several recently developed biochemical measures can provide additional information on a patient's use of alcohol. This article describes ethyl glucuronide, ethyl sulfate, phosphatidyl ethanol, and carbohydrate-deficient transferrin as biomarkers of drinking and summarizes research dealing with their application in patients with alcohol use disorders who are candidates for or recipients of liver transplantation. The article also offers suggestions for enhancing the reliability of self-report measures of drinking status. Liver Transpl 19:369–376, 2013. © 2013 AASLD.


alcoholic liver disease


carbohydrate-deficient transferrin


ethyl glucuronide


ethyl sulfate




high-performance liquid chromatography


liquid chromatography–mass spectroscopy


liquid chromatography–tandem mass spectroscopy


phosphatidyl ethanol


Timeline Followback.

Debate continues on the prevailing policy that patients with alcoholic liver disease (ALD) must demonstrate at least 6 months of abstinence from alcohol before they become eligible for liver transplantation.[1] Although it is unlikely that this issue will be resolved in the near future, drinking by ALD patients before and after transplantation will remain a significant clinical concern. Twenty to thirty percent of orthotopic liver transplants are performed for patients with end-stage ALD,[2] and alcohol use disorders are the first or second leading cause of liver transplantation.[3, 4]

Among the reasons for monitoring and encouraging abstinence in ALD patients are the stabilization of their medical condition before surgery and the improvement of their prospects for long-term sobriety afterward.[3] Unfortunately, the recurrence of drinking is common even in ALD patients who have committed to abstinence and who have already been selected for liver transplantation.[2, 5, 6] It has also been argued that with consistent abstinence, a candidate for ALD-related transplantation might in fact no longer require a graft.[2] Because the misuse of alcohol costs the individual, his or her family, and our society a tremendous amount and is associated with severe medical, psychological, and social suffering, transplant center staff, like all health care providers, should be concerned about alcohol use in their ALD patients, who constitute a group at demonstrated risk for severe alcohol problems.

Likewise, it is important to monitor alcohol use by ALD patients after transplantation and to intervene or refer patients for alcohol treatment as needed. Avoidance of alcohol would also promote the improvement of a patient's general quality of health and social functioning and would likely enhance compliance with medication and attendance at follow-up medical appointments. At least heavy drinking after liver transplantation has been linked to substantially diminished long-term survival rates (≥5 years) in ALD patients, with the most common causes of death being alcohol-related.[7, 8] Because it is not yet possible to accurately predict which patients will return to heavy drinking, Fuller[9] recommended that all ALD patients be urged to remain alcohol-free after liver transplantation. DiMartini et al.[4] also suggested that the drinking status continue to be monitored over the long term because, although the risk of relapse into abusive drinking appears to decrease over time, it does not cease.

The value of monitoring alcohol use through the measurement of its physical presence in blood, breath, or urine is quite limited because alcohol itself remains in the body for only a period of hours. Residual biomarkers can, however, provide important objective information on a more distant drinking status, which may not be accurately obtained from a patient's verbal report or clinical examination. Traditional alcohol biomarkers are liver function tests [eg, gamma-glutamyltransferase (GGT)]. They are indirect in the sense that, to the extent that excessive use of alcohol damages liver functioning, they may rise. Their use as indicators of drinking itself in pretransplant ALD patients is generally not recommended because they will be expected to often be elevated on account of serious liver damage, whether or not it is related to the current use of alcohol. Traditional biomarkers could provide some indication of the drinking status of ALD patients after transplantation, but they are insensitive to low levels of drinking and generally are elevated only after a long period of heavy drinking.

Biomarkers that are sensitive to recent drinking per se are becoming available. In this article, we describe 4 promising alcohol biomarkers, and we review the results of studies that have employed them in liver transplant patients with ALD. Following this, we offer some general comments and suggestions on the use of self-report measures of drinking.


Carbohydrate-deficient transferrin (CDT) has been approved by the Food and Drug Administration as a biochemical test of alcohol consumption. Transferrin is a glycoprotein synthesized and secreted by the liver. Moderate to heavy drinking (50–80 g of alcohol per day) for several days can decrease the carbohydrate content of transferrin, including sialic acid, galactose, and N-acetylglucosamine; hence, the term CDT is used.[10] After the cessation of drinking, the serum CDT level is reduced to normal values within approximately 2 weeks. CDT has been evaluated as a marker for initial screening as well as relapse.[11] CDT is at least as sensitive as GGT, a traditional marker of alcohol and liver disease. It is, however, far more specific than GGT.[12, 13] CDT appears to be somewhat influenced by smoking, body weight, certain types of liver disease, and female sex.[12]

During the past decade, efforts have been made to measure CDT in a standardized, sensitive manner. In 2007, the International Federation of Clinical Chemistry and Laboratory Medicine[14] issued 2 recommendations: (1) to normalize variations in transferrin levels among individuals (particularly women), CDT should be expressed as the amount of CDT divided by the amount of total transferrin, and (2) the disialotransferrin glycoform of CDT should be measured because this isoform of CDT correlates most highly with alcohol intake.[15]

CDT levels may be elevated in patients with severe liver disease, regardless of their current drinking status.[16-19] In ALD patients, the sensitivity of CDT ranges from 0.46 to 0.73, with the specificity possibly as low as 0.70.[20] In fact, DiMartini et al.[17] found that CDT values correlate quite highly in pretransplant ALD patients with the severity of liver disease as reflected in Child-Pugh scores.

Although CDT appears to be too lacking in specificity to serve as a useful pretransplant indicator of alcohol use in ALD patients, it has shown considerable promise as an indicator of posttransplant drinking. Berlakovich et al.[20] reported CDT's sensitivity to be 0.92 and its specificity to be 0.98 using a criterion of alcohol use result of clinical interviews and questionnaires conducted by a psychologist or other member of the transplant team who had established close clinical relationships and saw patients frequently after transplantation. Heinemann et al.[16] noted that 4 of their posttransplant patients had elevated CDT values, and when they were confronted with these results, 2 acknowledged drinking. Finally, although Staufer's research team[21] found that urinary ethyl glucuronide (EtG) performed considerably better than CDT as an indicator of posttransplant relapse, CDT was nevertheless able to detect some relapses that were missed by EtG.

CDT may be used in combination with other alcohol biomarkers and is used most commonly with GGT. The combination of CDT and GGT increases the sensitivity for identifying drinking problems without significantly compromising specificity.[22] This phenomenon is likely due to the fact that the 2 tests validly identify somewhat nonoverlapping groups of drinking patients. A simple and useful formula for combining the results of CDT and GGT has been offered.[23]


EtG, a promising marker of alcohol intake, is a direct metabolite of alcohol formed as a conjugate of alcohol with glucuronic acid in the liver and is catalyzed by the enzyme uridine diphosphate glucuronyl transferase.[10, 24] EtG is measurable in tissue, blood, hair, and, most commonly, urine. Currently, urinary EtG is measured by liquid chromatography–mass spectroscopy (LC-MS) and liquid chromatography–tandem mass spectroscopy (LC-MS/MS). A faster and less costly method involving an immunoassay has been developed but requires further evaluation.[24] The detection time ranges from hours to up to 4 or 5 days and depends on the amount and time frame of ethanol consumption.[24, 25] Urinary EtG appears to be highly sensitive and specific to alcohol intake. It is so sensitive that EtG may be detected in the urine if there have been significant extraneous exposures to ethanol in foods, over-the-counter cold medications, mouthwash, hand sanitizer gel, or other hygiene products.[25-27] Consideration should be given to this possibility when EtG results are being interpreted. To help guard against the misinterpretation of such unintentional environmental exposures to ethanol, a cutoff of 1 mg/L has been suggested.[28] A potential source of false positives is the storage of urine samples at room temperature because these specimens may contain yeasts that can convert urine glucose into alcohol. This is a particular issue when EtG is used with diabetic patients.[25] False negatives can arise from Escherichia coli hydrolysis of EtG in urinary tract infections,[29] the ingestion of chloral hydrate medications,[30] and urine dilution,[31] the last of which can be corrected by the consideration of the ratio of EtG to creatinine. Results from the World Health Organization/International Society for Biomedical Research on Alcoholism study showed some influence of age, sex, and renal function on EtG levels in urine, whereas race, tobacco use, body mass index, liver cirrhosis, and body water had no significant impact on EtG concentrations in urine.[32] The findings regarding renal and liver function have recently been confirmed.[33]

Ethyl sulfate (EtS) is another minor alcohol metabolite that can be measured simultaneously with EtG in urine by LC-MS/MS. EtG and EtS values tend to be very highly correlated.[34, 35] Measuring both biomarkers may serve to corroborate results, especially because EtS does not seem to be influenced by bacterial degradation.[29, 36, 37]

To assess a longer time span of alcohol intake, the measurement of EtG concentrations in hair has been employed, with 1 cm of hair reflecting approximately a 1-month period. The hair sample required is approximately 200 strands (approximately the diameter of a number 2 pencil) cut as close as possible to the scalp. The determination of EtG in hair can distinguish between chronic excessive alcohol use, moderate alcohol use, and abstinence/very low alcohol intake.[37-40] Morini et al.[37] found that hair EtG was more than twice as sensitive as CDT and was approximately equal to it in specificity in identifying usual alcohol consumption of 60 g or more per day occurring over the past 2 weeks. (This would equate to approximately 5 standard drinks in the United States; a standard drink in the United States is a 12-oz beer, a 5-oz glass of red wine, or a 1.5-oz shot of spirits such as whisky, vodka, brandy, or rum.)

The consensus statement of the Society of Hair Testing[38] defines concentrations greater than 7 pg/mg as strong evidence for regular alcohol intake and EtG concentrations greater than 30 pg/mg as clear evidence for excessive and regular alcohol consumption.[24] Lees et al.[41] found a correlation of 0.42 between total weekly alcohol consumption and hair EtG levels in a sample heterogeneous for drinking behavior that ranged from teetotalers to high-risk drinkers (ie, 50 or more drinks per week). Although increased levels of EtG have been reported in patients with decreased kidney function,[42] ethanol-containing cosmetics have been shown to have no effect on EtG levels in hair samples.[43]

Research has also been conducted on the use of EtG to determine the drinking status of transplant candidates with ALD. On the basis of a single-point-in-time assessment, Webzell et al.[6] observed that although 20% of their patients were positive for urinary EtG or EtS, only 3% self-acknowledged drinking. The authors did not report whether these 3% were also positive for EtG or EtS.

Although none of their patients receiving alcohol treatment before liver transplantation admitted to drinking nor were any of their on-the-spot breath analyses positive for alcohol, Erim et al.[2] reported that half were positive for EtG at some point, and 57% of the recurrent tests for these patients were positive. Furthermore, several patients refused to provide a urine specimen at 1 or more of the monitoring sessions, seemingly because they feared that the results would be positive.

Staufer et al.[21] contrasted the performances of 7 alcohol biomarkers in a sample consisting primarily of liver transplant candidates with ALD but also including some individuals with a history of alcohol abuse who were in follow-up care after liver transplantation. Sixty-four percent of the patients acknowledged recent or chronic consumption of alcohol only after they had been confronted with biomarker testing results indicating drinking. Ninety-three percent of those who were positive for any of the biomarkers were positive for urinary EtG and/or CDT. In half of the patients positive for any alcohol biomarker, EtG was the single biomarker elevated. At a cutoff value of 500 mg/L, the sensitivity of EtG was 0.89, and the specificity was 0.99. Raising the cutoff to 1000 mg/L lowered the sensitivity to 0.75. The authors cautioned that their findings might somewhat overestimate the performance of the biomarkers because some other patients might also have been drinking but not have been identified by any of the biomarkers.

Stewart et al.[44] conducted a study of patients with liver disease who denied drinking and were not suspected of drinking by the hepatologist. The patients were administered the Timeline Followback (TLFB; see the Assessing Alcohol Use by Self-Report section for a description of the measure) by research assistants independently of their health care to determine whether they had been drinking in the past 3 days and in the past 7 days. For the past 3 days, urinary EtG produced a sensitivity of 0.76 with a specificity of 0.93, and urinary EtS yielded a sensitivity of 0.82 with a specificity of 0.86. The corresponding values for identifying patients drinking in the past 7 days were somewhat less favorable. (Approximately one-fourth of the patients drinking during the past 7 days reported being abstinent during the past 3 days.) Discordances between self-reports of drinking and the status according to the biomarkers were not explained by age, sex, ethnicity, or liver disease severity, although the number of subjects was perhaps too low to be able to detect such differences if they existed. The correlation between EtG and EtS was reported to be 0.94, but unfortunately, the researchers did not indicate the strength of the bivariate relationship between positives and negatives on the 2 measures.

Haller et al.[45] collected data on self-reported use of alcohol during the previous 30 days, breath tests for the presence of alcohol, and hair EtG levels in 32 end-stage liver disease patients and 10 end-stage renal disease patients, all of whom had been denied transplantation because of substance abuse in the previous 6 months. Data were collected at the baseline and at 8, 20, and 32 weeks, although all of the data could not be collected from all of the subjects at each time point. Self-reported alcohol use was confirmed by EtG in 85% of the cases. On the other hand, 49% of the patients reporting alcohol abstinence also produced positive hair tests for EtG. The authors determined that 2 negative hair EtG results were indicative of 6 months of alcohol abstinence. If this algorithm is confirmed by future research, it would have considerable applied value in clinical situations in which it is not feasible to do more frequent tests for EtG.


Phosphatidyl ethanol (PEth) is formed in the presence of ethanol via the action of phospholipase D in cell membranes of red blood cells.[46] Although initially PEth was measured with high-performance liquid chromatography (HPLC), in recent years, LC-MS[47] and LC-MS/MS methods have been used. In 2010, the identification of 48 PEth homologues was reported.[48] With the HPLC method, positive results were found after repeated alcohol use exceeding 50 g of ethanol and occurring over the previous 2 to 3 weeks.[49] In a recent drinking experiment in healthy volunteers using the more precise LC-MS/MS method, PEth was found to rise even after a single day of drinking.[50] No sources of false positives for PEth values have yet been found.[51-53] The relationship between alcohol intake and PEth levels is linear.[54-56] Neither liver disease nor hypertension appears to influence the production of PEth.[55]

In vivo formation of PEth has been reported in samples containing ethanol and stored at room temperature or −20°C. It can be assumed that there is a significant advantage in using dried blood spots because capillary blood can be used, transportation and storage are simplified, and the risk of infection with, for example, human immunodeficiency virus and hepatitis C is decreased.[57]

PEth was found to distinguish 3 groups of ALD patients who differed in the level of average daily alcohol consumption (abstinent or less than 1 drink per day, 1–3 drinks per day, and more than 3 drinks per day). In contrast to CDT, the severity of liver disease did not appear to moderate the relationship between PEth values and alcohol consumption. Nor did age or sex influence the association.[55] In a group of healthy nonpregnant women who were 18 to 35 years old, PEth values were linearly associated with alcohol consumption over the preceding 2 weeks and were detectable in 93% of the subjects who had been consuming 2 or more drinks per day.[56] Among patients receiving outpatient treatment for an alcohol problem, Helander et al.[47] computed the correlation between values of PEth and CDT to be 0.62, and importantly, they observed that CDT rarely identified drinking if the PEth level was not also elevated.

Information on the aforementioned biomarkers is summarized in Table 1.

Table 1. Characteristics of Biomarkers of Alcohol Use
BiomarkerBiological MatrixCutoff ValueWindow of AssessmentType of Drinking CharacterizedPossible Sources of False PositivesNotes
  1. a

    The cutoff values are for EtG; EtS cutoffs are approximately one-fifth of these values.

  2. b

    Measured as a percentage of transferrin that is of the disialo isoform.

AlcoholBreath, blood, and urineAny valueSmall number of hoursAcute drinking marker for 1 or more drinksNone from liver diseaseThis is convenient and low-cost, but there is a very short window of assessment.
     Nonbeverage alcohol in common products 
PEthBloodHPLC: 0.22 μmol/L for total PEth2-4 weeksChronic drinking marker for moderately heavy drinking (3 or 4 drinks per day for a period of 2 weeks).None from liver diseaseThis is a good marker of relapse and can be used well with the CDT percentage. The use of dried blood spots facilitates sample collection, storage, and shipping.
  LC/MS-MS: 0.05 μmol/L for PEth homologue 16:0/18.1    
EtG and EtSHair7-30 pg/mg of hairUp to several months (hair)Chronic drinking marker (hair)None from liver diseaseResearch is underway to evaluate these markers in finger nails.
    The amount of drinking needed for an elevation in hair is unknown.  
 Urine100-500 ng/mL of urine (possible environmental exposure to alcohol[58])Few days (urine)Acute drinking marker for 1-2 drinks (urine)Nonbeverage alcohol in common products (urine) 
  1000 ng/mL of urine (environmental exposure to alcohol usually not an issue[58])a  Decreased kidney function (urine) 
CDTbBlood1.7%2-3 weeksChronic drinking marker (moderate to high consumption for 7-10 days)Rare genetic variantThis is a good marker of relapse. CDT may “spike” with less drinking in patients who have had high CDT values in the past that have decreased because of reduced drinking.
     Primary biliary cirrhosis 
     Chronic end-stage liver disease 
     Possibly higher values in women, smokers, and individuals with a low body mass index 


Verbal reports of drinking behavior by ALD candidates for liver transplantation may underestimate their actual drinking, particularly if they believe that there is a policy in place that would delay or deny them a needed liver transplant were it discovered that they had been using alcohol. Although this incentive to dissimulate is not at issue in patients after liver transplantation, research in other clinical settings suggests that biomarkers identify some patients who do not accurately self-report their use of alcohol. Despite obvious problems with solely relying on self-admission of drinking, it is difficult to imagine that ALD patients would not be directly asked about their drinking because verbal reports of drinking are inexpensive and flexible and may identify drinking that is of insufficient quantity or is outside the window of assessment captured by various alcohol biomarkers. Furthermore, information from self-reports of drinking behavior can and should be considered with objective biomarker data. Accurate self-reports of drinking are also useful as a basis for providing information to the clinician on the internal and external stimuli that appear to put the patient at highest risk for drinking and for determining other unique features of the patient's drinking that may have a bearing in developing an alcohol treatment plan.

Research on self-reports of drinking suggests that their accuracy can be improved if a rapport and working alliance with the patient have been achieved. To the extent possible, it is also important to assure the patient of confidentiality. (For example, the pretransplant interview on drinking could be done by an alcoholism specialist who is not associated with the liver transplant decision-making team; this would facilitate needed alcohol treatment for the patient and ensure that evidence of alcohol use obtained in this context would not be used to deny liver transplantation.) The clinician should ask very specific questions about drinking. For example, the interviewer should ask about the types of alcoholic beverages consumed and their volumes rather than simply the total number of drinks because this type of detailed questioning tends to elicit recognition of higher levels of alcohol use than requests for the overall number of drinks. It is also important that the time frame under consideration be clearly stated to the patient, and ideally, this window of assessment should cover a fairly short interval in order to mitigate inaccuracies due to recall difficulties. Patients should be interviewed only when they are alcohol-free as demonstrated by results from an alcohol breathalyzer because individuals under the effects of alcohol tend to underestimate their alcohol consumption.[60] The application of the principles for enhancing the validity of self-reports of drinking behavior by ALD liver transplant recipients was exemplified in a project conducted by DiMartini et al.[61]

Clinical practice suggests the use of the TLFB[61] as the preferred means of collecting self-reported drinking information. The TLFB is a calendar-based method of eliciting information on drinking. The time frame to be covered should also include date-based memory cues, such as events of common significance (eg, weekends and holidays) and events of personal significance to the patient (eg, paydays, out-of-town trips, medical visits, and birthdays). The TLFB can be administered in a verbal interview, in writing, or online via a computer. The period of assessment may vary with the clinical need, but it should be kept in mind that the longer the retrospective period is, the more one might expect the patient's memory to be inaccurate. Depending on the time frame under consideration, the TLFB usually requires 10 to 30 minutes. An instance in which the TLFB might be impracticable, the 4-item quick drinking screen[62] or the first 3 questions of the Alcohol Use Disorders Identification Test[63] should be considered. Unlike the TLFB, these measures will yield only summative information on recent drinking behavior, but the questions have been well standardized and researched.


The results for new biomarkers of alcohol use and the employment of strategies to enhance the accuracy of self-reports of drinking behavior provide liver transplant staff practical means for identifying a return to drinking by patients who have been diagnosed with an alcohol use disorder. The ongoing monitoring of alcohol use is recommended for all ALD patients before and after transplantation. Standardized self-report measures and clinical observation should always be the first step. Patients who deny drinking on a self-report measure but whom the clinician believes may actually be drinking are candidates for monitoring with biomarker measures.

The 4 biomarkers presented in this article extend the window for assessing drinking beyond simple measurements of the physical presence of alcohol in blood, breath, or urine. EtG, EtS, and PEth have a clear advantage over traditional alcohol biomarkers in pretransplant ALD patients because they are not influenced by liver or other organ damage or other common causes of false positives. Measured in urine, EtG and EtS capture drinking behavior up to a few days before testing, with this depending on the amount of alcohol consumed. Although this window of assessment is still rather narrow, if the patient is regularly drinking, EtG and EtS measured in urine may be able to identify it. CDT, PEth, and EtG/EtS (in hair) assess much longer periods. Although CDT typically is not a reliable measure of drinking in ALD patients before transplantation due to liver damage, in posttransplant patients, CDT may be elevated for approximately 2 weeks, and PEth is likely elevated for approximately 3 weeks after a week or 2-week period of moderate to high alcohol consumption (eg, 5 drinks per day on average). CDT and GGT and CDT and PEth are useful combinations of tests in posttransplant patients. Measured in hair, EtG or EtS can assess drinking that has occurred over a period of months.

CDT and urinary EtG or EtS testing is available from commercial labs, and the local laboratory may be willing and able to do this. PEth and hair-based EtG or EtS tests remain somewhat harder to obtain, although some commercial labs now are beginning to perform these tests, and possibly the local laboratory supporting the liver transplant service may be willing to do them.

Although the primary role of alcohol biomarkers is to assess drinking behavior, they may also serve as the basis for providing personalized and objective feedback to patients, educating them on the role of alcohol use in liver disease, and motivating drinking reduction or cessation. Information from the biomarkers may also be used in alcohol treatment efficacy trials and in epidemiological research on patients before and after liver transplantation.


The authors thank Douglas Lewis, president and scientific director of US Drug Testing Laboratories, for his helpful assistance in reviewing the technical content of this article and for his useful comments regarding Table 1.