Phosphatidylethanol (PEth) in Blood as a Marker of Unhealthy Alcohol Use: A Systematic Review with Novel Molecular Insights

The Alcohol Use Disorders Identification Test (AUDIT) and its short form, the AUDIT-C, the main clinical instruments used to identify unhealthy drinking behaviors, are influenced by memory bias and under-reporting. In recent years, phosphatidylethanol (PEth) in blood has emerged as a marker of unhealthy alcohol use. This systematic review aims to investigate the molecular characteristics of PEth and summarize the last ten years of published literature and its use compared to structured questionnaires. A systematic search was performed, adhering to PRISMA guidelines, through “MeSH” and “free-text” protocols in the databases PubMed, SCOPUS, and Web of Science. The inclusion criteria were as follows: PEth was used for detecting unhealthy alcohol consumption in the general population and quantified in blood through liquid chromatography coupled to mass spectrometry, with full texts in the English language. Quality assessment was performed using the JBI critical appraisal checklist. Twelve papers were included (0.79% of total retrieved records), comprising nine cross-sectional studies and three cohort studies. All studies stratified alcohol exposure and quantified PEth 16:0/18:1 through liquid chromatography coupled to mass spectrometry (LC-MS) in liquid blood or dried blood spots (DBS) with lower limits of quantitation (LLOQ) ranging from 1.7 ng/mL to 20 ng/mL. A correlation between blood PEth level and the amount of alcohol ingested in the previous two weeks was generally observed. PEth interpretative cut-offs varied greatly among the included records, ranging from 4.2 ng/mL to 250 ng/mL, with sensitivity and specificity in the ranges of 58–100% and 64–100%, respectively. Although the biomarker seems promising, further research elucidating the variability in PEth formation and degradation, as well as the molecular mechanisms behind that variability, are necessary.


Introduction
Excessive alcohol consumption has been identified as an important risk factor for illness, disability, and mortality, with an increased prevalence of secondary cardiovascular, liver, cancer, and neurological disorders [1,2]. The World Health Organization (WHO) defines a standard drink as 10 g of pure ethanol and harmful consumption as exceeding two standard drinks per day for both men and women [3], although this threshold has not yet been adopted worldwide [4]. Despite the lack of agreement on definitions, it is globally known that the risk of developing alcohol-related diseases is directly proportional to the amount of alcohol consumed [5]. For this reason, it is crucial for all clinicians to gain information on the patient's risk of developing an alcohol use disorder or secondary diseases, as well as for forensic experts to have a suitable tool to assess alcohol consumption for its possible legal ramifications.
In August 2022, one author (LF) performed a systematic literature search via "MeSH" and "free-text" protocols in the PubMed, SCOPUS, and Web of Science databases, with time limits 1 January 2011-1 January 2023. Search terms used for PubMed and Web of Science were as follows: ("phosphatidylethanol" [Supplementary Concept] OR Peth OR phosphatidylethanol) AND (forensic OR legal OR biomarker OR marker OR alcohol abuse OR abstinence OR monitoring). A modified string was used for Scopus: "ALL ((phosphatidylethanol OR peth OR phosphatidylethanol) AND (forensic OR legal OR biomarker OR marker OR alcohol AND abuse OR abstinence OR monitoring))." Subsequently, three authors (LF, AC, and MP) selected papers based on titles and abstracts according to the following inclusion and exclusion criteria.
Inclusion criteria: A. Titles and abstracts available in the English language. B. PEth used for detecting unhealthy alcohol consumption in the general population. C. PEth quantified in liquid human blood or dried blood spots through liquid chromatography coupled to mass spectrometry. D. Full-text available in the English language.
Exclusion criteria: E. Opinion papers, editorials, and narrative reviews without novel data. F.
Papers with data only on specific populations (e.g., pregnant women, HIV-positive individuals, etc.). G. Papers containing only data on the development and validation of analytical methods. H. Papers on postmortem or autopsy cases.
Not meeting at least one of the inclusion criteria A-D or, conversely, meeting one or more of the exclusion criteria E-H was reason for papers' exclusion. In cases of doubtful classification based on title and abstract, the full text was retrieved. Any discrepancy in the paper selection was addressed through collegial discussion among four authors (LF, AC, MP, and GV).
Data extraction from the selected articles was performed independently by four authors (LF, MP, AC, and FT), and two of them (AC and MP) included these in a table. In order to ascertain the correctness of the process and minimize subjective judgment, one author (GV) checked the accuracy of the entire data extraction process. The following items were collected from each study: authors, journal, year, features of the study (type of study, subjects involved, main aim, clinical setting, and inclusion and exclusion criteria), characteristics of the investigated population (numbers of subjects and race/ethnicity), methods for estimating alcohol use, analytical method used for PEth analysis, type of measured PEth and concentration, type of collected sample, other biomarkers used, and main results obtained (sensitivity, specificity, positive predictive value, and negative predictive value). Any discrepancies in the data extraction process were settled by consensus discussion performed by five authors (LF, AC, MP, GC, and GV).
A validity assessment of each included manuscript was performed using the JBI critical appraisal checklist for analytical cross sectional-studies or for cohort studies [26], based on type of study. Cross sectional-studies were evaluated on eight quality items, while cohort studies were evaluated on eleven quality items. For each quality item, we indicated with "Yes", "No", or "Unclear" to indicate the cases in which the data were properly reported, not reported, or not properly reported, respectively. Finally, "Not applicable" refers to those items that are inconsistent with the study in question.

Results and Discussion
As reported in the PRISMA flow-chart (Figure 1), the combined search on the databases PubMed, Web of Science, and Scopus retrieved 2106 records. Of these, 592 were duplicates and thus were removed, resulting in a total 1514 articles evaluated by title and abstract. From the latter, 1356 were excluded because they did not meet criteria A, B, and C. Of the remaining 158 papers, analyzed in full text, 146 were excluded based on criteria D-H. Twelve papers (0.79% of the total) were included in the present review.

Results and Discussion
As reported in the PRISMA flow-chart (Figure 1), the combined search on the databases PubMed, Web of Science, and Scopus retrieved 2106 records. Of these, 592 were duplicates and thus were removed, resulting in a total 1514 articles evaluated by title and abstract. From the la er, 1356 were excluded because they did not meet criteria A, B, and C. Of the remaining 158 papers, analyzed in full text, 146 were excluded based on criteria D-H. Twelve papers (0.79% of the total) were included in the present review. The data extracted from the twelve included papers are presented in detail in the Table S1 of the supplementary material, while the main data are presented in Table 1.
Quality assessment results using the JBI critical appraisal checklist for analytical cross sectional-studies or for cohort studies [25] of the included records are displayed in  The data extracted from the twelve included papers are presented in detail in the Table  S1 of the supplementary material, while the main data are presented in Table 1. "By AUDIT-C score:
Quality assessment results using the JBI critical appraisal checklist for analytical cross sectionalstudies or for cohort studies [25] of the included records are displayed in Figures 2 and 3. In recent years, increasing emphasis has been placed on the detection and treatment of hazardous and harmful drinking disorders, particularly among patients who are seen in primary care se ings [38].
Hazardous drinking is generally defined as a quantity or pa ern of alcohol consumption that places patients at risk for adverse health events, while harmful drinking is defined as alcohol consumption that results in adverse events (e.g., physical or psychological harm). Both hazardous and harmful drinking behaviors are considered "unhealthy" alcohol consumptions [38][39][40].
As defined by the National Institute on Alcohol Abuse and Alcoholism (NIAAA), for women, low-risk drinking is no more than three drinks on any single day and no more than seven drinks per week. For men, it is defined as no more than four drinks on any single day and no more than 14 drinks per week [41].
The Alcohol Use Disorders Identification Test (AUDIT) is currently the only clinical instrument specifically designed to identify hazardous and harmful drinking. It allows the investigation of a patient's alcohol habit through 10 items, returning a numerical value from 0 to 40, with sensitivity of about 90% and specificity of about 80% of detecting an alcohol use disorder [40]. The main limitations of the AUDIT, i.e., the length and time required (about 2-3 min), are partially overcome by the AUDIT-C, which investigates only three items with sensitivity and specificity values of 70% and 90%, respectively [42]. Were the criteria for inclusion in the sample clearly defined?
Were the study subjects and the setting described in detail?
Was the exposure measured in a valid and reliable way?
Were objective, standard criteria used for measurement of the condition?
Were confounding factors identified?
Were strategies to deal with confounding factors stated?
Were the outcomes measured in a valid and reliable way?
Was appropriate statistical analysis used?
Yes Not Not clear Not applicable In the included records, AUDIT has been used alone or in combination with AUDIT-C in five records [28][29][30][31][32], AUDIT-C alone in two [9,33], and AUDIT-QF in one [34]. The remaining records were experimental drinking studies [30], or reconstructed previous alcohol exposure by self-reported alcohol intake [27] or by the alcohol intake questionnaire (AIQ) [37]. Lowery et al., who had to reconstruct a potential alcohol misuse among organ donors from proxy history, used the Uniform Donor Risk Assessment Interview [36].
When risk stratification was performed by AUDIT, healthy alcohol use was generally defined for women as AUDIT < 5 and for men as AUDIT < 8 [28,29,32]. When AUDIT-C or AUDIT-QF were used, the thresholds were <6 [31,32] or <4 [33,34], respectively, for identifying harmless alcohol use.
In contrast, Aboutara et al. [27] included two groups with moderate alcohol intake, Were the two groups similar and recruited from the same population?
Were the exposures measured similarly to assign people to both exposed and unexposed groups?
Was the exposure measured in a valid and reliable way?
Were confounding factors identified?
Were strategies to deal with confounding factors stated?
Were groups/participants free of the outcome at the start of the study (or at the moment of exposure)?
Were the outcomes measured in a valid and reliable way?
Was the follow up time reported and sufficient to be long enough for outcomes to occur?
Was follow up complete, and if not, were the reasons to loss to follow up described and explored?
Were strategies to address incomplete follow up utilized?
Was appropriate statistical analysis used?
Yes Not Not clear Not applicable Figure 3. Cohort studies.
In recent years, increasing emphasis has been placed on the detection and treatment of hazardous and harmful drinking disorders, particularly among patients who are seen in primary care settings [38].
Hazardous drinking is generally defined as a quantity or pattern of alcohol consumption that places patients at risk for adverse health events, while harmful drinking is defined as alcohol consumption that results in adverse events (e.g., physical or psychological harm). Both hazardous and harmful drinking behaviors are considered "unhealthy" alcohol consumptions [38][39][40].
As defined by the National Institute on Alcohol Abuse and Alcoholism (NIAAA), for women, low-risk drinking is no more than three drinks on any single day and no more than seven drinks per week. For men, it is defined as no more than four drinks on any single day and no more than 14 drinks per week [41].
The Alcohol Use Disorders Identification Test (AUDIT) is currently the only clinical instrument specifically designed to identify hazardous and harmful drinking. It allows the investigation of a patient's alcohol habit through 10 items, returning a numerical value from 0 to 40, with sensitivity of about 90% and specificity of about 80% of detecting an alcohol use disorder [40]. The main limitations of the AUDIT, i.e., the length and time required (about 2-3 min), are partially overcome by the AUDIT-C, which investigates only three items with sensitivity and specificity values of 70% and 90%, respectively [42].
In the included records, AUDIT has been used alone or in combination with AUDIT-C in five records [28][29][30][31][32], AUDIT-C alone in two [9,33], and AUDIT-QF in one [34]. The remaining records were experimental drinking studies [30], or reconstructed previous alcohol exposure by self-reported alcohol intake [27] or by the alcohol intake questionnaire (AIQ) [37]. Lowery et al., who had to reconstruct a potential alcohol misuse among organ donors from proxy history, used the Uniform Donor Risk Assessment Interview [36].
When risk stratification was performed by AUDIT, healthy alcohol use was generally defined for women as AUDIT < 5 and for men as AUDIT < 8 [28,29,32]. When AUDIT-C or AUDIT-QF were used, the thresholds were <6 [31,32] or <4 [33,34], respectively, for identifying harmless alcohol use.
In contrast, Aboutara et al. [27] included two groups with moderate alcohol intake, the first one with a weekly alcohol intake exceeding 24 g and the second one with a weekly alcohol intake exceeding 84 g.
It is well-known that PEth synthesis is directly proportional to ethanol exposure and increases with increasing blood ethanol concentration (BAC) [43]; however, it has not yet been elucidated from a molecular point of view why the amount of PEth accumulated in different subjects is different, even if the amount of ethanol ingested is the same and consumed over the same time period [44,45].
Several authors have suggested that a better understanding of the inter-individual variability of PLD activity [44,46] might explain the PEth variability, but molecular evidence on that point is still missing. Another important factor is the concentration of ethanol at the site of PLD, which is influenced by the absorption of ethyl alcohol (e.g., affected by percent body fat, genetically determined alcohol, acetaldehyde dehydrogenase, stomach content, drinking pattern, and drinking rate) and its elimination [47]. Drinking experiments in which volunteers were given standard doses of alcohol showed that PEth forms soon after alcohol consumption and that ethanol absorption is a key-factor influencing PEth formation [44,47]. Recently, a large individual patient data meta-analysis [48] has shown that body mass index (BMI) negatively influences PEth sensitivity for detecting unhealthy alcohol consumptions; indeed, blood alcohol concentration per standard drink is inversely proportional to body weight [49]. Therefore, the higher the BMI, the lower the BAC, and thus the formation of PEth.
There is also uncertainty about the role of other potential factors affecting PEth formation or degradation, such as reactive oxygen species (ROS). Ethanol is metabolized through three enzymatic pathways: alcohol dehydrogenase (ADH), catalase, and the microsomal ethanol-oxidizing system (MEOS). The latter is a multienzyme complex with cytochrome P450 (CYP) and its main isoenzyme, CYP2E1, as the principal elements [50]. The activity of CYP2E1 can be significantly induced by chronic alcohol consumption [51]. It has been demonstrated that, during binge drinking episodes, ethanol is predominately metabolized to acetaldehyde via the MEOS [52,53]; this factor would contribute to the formation of ROS and oxidative stress related to alcohol consumption [54]. The process might also take place outside the liver, given the evidence of the presence of extracellular vesicles containing CYP2E1 in blood, especially in those subjects with alcoholic liver disease [55]. This could be particularly interesting when considering the other constituents of MEOS, such as NADPH-cytochrome P450 reductase and phospholipids [56].
Unfortunately, there is still a lack of evidence of a direct involvement of MEOS in the synthesis or degradation of PEth in human erythrocyte membranes. However, what is clear is that PEth may serve not only as a biomarker of drinking behaviors but also as a key pathological factor that affects cell function due to the changes provoked in the lipid composition of the cell membranes.
All the studies included in this review concluded that PEth is a very promising marker of unhealthy alcohol use; however, great variability emerged regarding the adopted cut-offs and the subsequent measured sensitivity and specificity of the biomarker (see Table S1-Supplementary Material). There is growing consensus to refer to PEth values below 20 ng/mL (0.028 µM) as being compatible with abstinence or minimal alcohol consumption during the weeks prior to sampling and to PEth values above 210 ng/mL (0.300 µM; the so-called "Swedish cut-off") as indicative of chronic excessive alcohol use. On the contrary, there is a lack of consensus on the best cut-off for identifying moderate (but unhealthy) drinking.
In the included records, the interpretative cut-offs varied between 4.2 ng/mL and 250 ng/mL, with the majority of authors proposing a cut-off between 4.2 ng/mL and 67 ng/mL [9,27,28,32,33,[35][36][37] with sensitivity and specificity in the range of 58-100% and 64-100%, respectively. This wide range of cut-offs, along with the fact that PEth can persist at a systemic level and be detected in blood up to two weeks after alcohol ingestion, complicates the interpretation of PEth blood concentrations, as well as the differentiation between chronic excessive drinking and binge drinking behaviors.

Conclusions
Although the marker PEth 16:0/18:1 seems very promising for detecting and classifying unhealthy drinking behavior, further research is necessary to elucidate the variability in PEth formation and degradation, as well as the molecular mechanisms behind that variability. Clearly, cut-off levels should also be further investigated using valid measures of drinking, days since last drink, and other factors potentially influencing the formation and degradation of PEth (i.e., hemoglobin, hematocrit, BMI, drinking pattern and rate, etc.). Optimally, such an investigation of drinking would be in a controlled experimental setting or measured using frequent BrAC, BAC, or wearable biosensors. Combinations of alcohol measures could also be useful. In a clinical setting, where under-reporting of alcohol use is unlikely, AUDIT and self-report of alcohol consumption could be used in combination with PEth with a low cut-off in order to enhance sensitivity and diagnostic accuracy. In a forensic setting, where the subject under examination and is self-reporting, such reports could be unreliable. Other alcohol biomarkers, such as urinary EtG (with the forensic cut-off of 500 ng/mL) and hair EtG (with the 7 pg/mg cut-off for abstinence in a 3 cm long hair sample) might be used in combination with PEth to confirm abstinence or harmless drinking.