Part of this work has been presented as a poster at the 1st International Symposium “Neurodevelopment and CNS vulnerability” in Erlangen, Germany in September 2018.
Yearly, nearly 2.5 million deaths worldwide are attributable to alcohol use, in addition to other consequences of alcohol-related diseases and injuries (World Health Organization: Global status report on alcohol and health). Various risk factors and mechanisms have been suggested to play a role in the development and maintenance of alcohol-use disorders [1
]. Genetic components account for 50%–60% according to twin, adoption, and family studies [2
]. Prenatal hyperandrogenization [3
] may partially be responsible for the two-fold higher prevalence of alcohol dependence in males compared to females. Disturbances in sphingolipid metabolism have been identified in human studies and animal models of psychiatric disorders [5
], including major and mild depression [6
], which share a high comorbidity with alcohol dependence.
Sphingolipids play an increasingly recognized role in neuronal function in the brain, not only by serving as a membrane component to form a physical barrier. They also influence the local composition of the plasma membrane, the localization and activity of proteins and, thus receptor-mediated signaling, in addition to their own actions as ligands [9
]. Current research specifically focuses on enzymes crucial for maintaining the balance between the pro-apoptotic ceramide and its anti-apoptotic metabolite sphingosine 1-phosphate (the so-called “rheostat”) in the context of various physiological and pathophysiological conditions [10
]. Sphingomyelinases and ceramidases are involved, for example, in ceramide-mediated signal transduction required for apoptosis, differentiation, and other cellular (including inflammatory) responses, in intracellular cholesterol trafficking and metabolism, as well as in lysosomal degradation of sphingomyelin and ceramide [11
]. Two further pathways, de novo biosynthesis and the salvage pathway, lead to the generation of ceramide.
Acid sphingomyelinase (ASM, EC 184.108.40.206), encoded by the gene SMPD1
, catalyzes the hydrolysis of the abundant membrane lipid sphingomyelin into ceramide and phosphorylcholine. Altered enzyme activities have been noted in a variety of common human diseases [12
]. In alcohol dependence, levels of both the lysosomal [13
] and secretory form (S-ASM) of the enzyme are increased, and decrease gradually during withdrawal treatment in male, as well as female, patients [14
]. Consequently, plasma glycerophospholipid and sphingolipid species are also dysregulated in alcohol-dependent patients [16
]. In ethanol-fed mice, tissue ASM activity is increased [17
]. To our knowledge, there are no further published studies on the influence of the exposure of animals or cultured cells to ethanol on peripheral or culture supernatant S-ASM levels, except for one report. No alteration of serum S-ASM activity was detected in both transgenic mice overexpressing ASM and wildtype mice in a two-bottle free-choice drinking paradigm with a gender-balanced design [19
]. Stimulation of the neutral sphingomyelinase by ethanol likewise contributes to alterations in the sphingomyelin/ceramide balance [20
Hepatotoxicity is a major consequence of alcohol misuse. Of note, liver damage could also result from components of alcoholic beverages beyond their ethanol content. Both in experimental models of chronic ethanol-induced steatohepatitis, and patients with severe chronic alcohol-related liver disease, the immunoreactivity and ceramide content are increased [22
]. Interestingly, an accumulation of ceramide and elevated levels of S-ASM have also been found in non-alcoholic fatty liver disease [23
]. ASM knockout mice are resistant to alcohol-mediated fatty liver and cell death [24
]. Inhibition of ASM by imipramine blocked the ethanol-induced ASM activation and ceramide generation, resulting in amelioration of hepatic steatosis in ethanol-fed mice [18
]. Likewise, treatment of ethanol-fed rats with antioxidants, for example, N
-acetylcysteine, reduced the severity of chronic alcohol-related steatohepatitis, possibly attributable to the observed decreased expression of inflammatory mediators, reduced acid sphingomyelinase activity, and lowered ceramide load [25
]. The role of ASM likely involves sensitization of hepatocytes to the cytotoxic effects of TNFalpha [24
] and regulation of autophagy [26
The lack of reliable markers of alcohol consumption is a major obstacle to the diagnosis and treatment of alcohol dependence. Interviews and subjective questionnaires have their limitations, particularly because subjects are known to downplay the extent of their drinking behavior. Direct measurement of alcohol concentration in the breath, blood, or urine does not provide information more than a few hours beyond the most recent consumption of alcohol [27
], or it requires special equipment with a high cost (ethyl glucuronide, ethyl sulfate, phosphatidylethanol) [28
]. Currently available indirect biochemical markers, including carbohydrate-deficient transferrin (CDT, a form of the serum iron-carrying protein transferrin, with altered carbohydrate composition), mean corpuscular volume of erythrocytes (MCV), as well as the liver enzymes gamma-glutamyl transferase (GGT), alanine aminotransferase (ALT, also glutamic-pyruvic transaminase GPT), and aspartate aminotransferase (AST, also glutamic-oxaloacetic transaminase GOT), react to steady and significant alcohol intake over weeks or months, but suffer from relatively low sensitivity and specificity, and an uncertain time window of detection [28
Furthermore, there is a need for reliable predictors of relapse after withdrawal treatment, which is a common problem in alcohol dependence, resulting in a rate of up to 85% in the absence of further support after the initial detoxification phase [30
]. A number of known risk factors have been identified, albeit with limited accuracy and a high cost and time investment, which restricts their clinical applications for the identification of patients at risk of relapse and for individualized treatment [31
In our large and sex-balanced cohort of alcohol-dependent patients and matched healthy controls, we aimed at characterizing the readily quantifiable activity of peripheral S-ASM with respect to the phenotype and known biomarkers of alcohol dependence, with particular emphasis on liver parameters. Moreover, we evaluated the diagnostic performance of this enzyme in discriminating between patients and controls, and predicting relapse as assessed as alcohol-related readmissions to the hospital.
We have confirmed both the previously described increased activity of S-ASM in alcohol-dependent patients, and the decrease in S-ASM activity during detoxification treatment in a large and sex-balanced cohort [14
]. However, in a smaller previous study consisting predominantly of males, S-ASM activity in patients was 3-fold higher compared to healthy controls, and it declined in every single individual over 7–10 days of withdrawal treatment by 52% of the initial value, on average [14
]. In another small mixed gender study without controls, S-ASM activity fell by 15%–20% for females to 22%–29% for males during 2–7 days of treatment. A possible explanation for the smaller effects, in this study, could be our time window for inclusion during early abstinence (i.e., 24 to 72 h after the last consumption of alcohol), which could already be too late to detect the high initial drop in ASM activity.
While the increase in S-ASM activity in alcohol-dependent patients has been replicated, its origin remains elusive. A wide variety of cells have been demonstrated to secrete substantial amounts of this Zn2+
-dependent enzyme, including human vascular endothelial cells, macrophages, and platelets [12
], resulting in detectable levels not only in the blood, but also in cerebrospinal fluid [39
]. In mice fed on an atherogenic diet containing saturated fats and cholesterol, an increased macrophage secretion seemed to be responsible for the elevated S-ASM activity [40
]. The enzyme might be released into the bloodstream when cells are injured by ethanol, or other components of alcoholic beverages or as a response to systemic changes induced by these factors. ASM is a key regulator of ceramide-dependent signaling pathways, and it can be induced by cellular stress resulting from inflammation or infection [12
]. Mechanisms of ASM activation by ethanol could involve post-translational, as well as transcriptional effects [41
On the other hand, alcohol-dependent patients could carry risk factors for endogenously higher ASM levels. Genetically determined ASM activity is already known to influence the susceptibility for common human diseases, such as allergy [42
]. However, because known single nucleotide polymorphisms or variations in the repeat number within the special signal peptide negatively affect ASM activity [12
], it is rather unlikely that a frequent, but so far undetected, variant within the SMPD1
gene would predispose carriers to developing alcohol dependence, and be the cause of higher S-ASM levels. On the other hand, processes associated with posttranslational modifications, and regulation that modulate ASM trafficking, maturation, or secretion [45
], as well as those leading to degradation of the enzyme, could permanently or temporarily be altered in patients. Additionally, gene variants encoding proteins further upstream in the lipid synthesis pathway, such as in SERINC2
, could alter ASM levels. This gene was identified as a top-ranked risk gene for alcohol dependence [46
], and the encoded protein incorporates serine into membranes, facilitating the synthesis of phosphatidylserine and sphingolipids [48
]. Moreover, SMPD1
] has been reported to influence ASM activity and be altered in major depression [50
]. However, it has not yet been analyzed in alcohol dependence.
Remarkably, GGT values of a considerable proportion of healthy controls (17% of males, 7% of females) were above the normal reference range of the analyzing laboratory, with the upper limit of 60 U/L for males and 40 U/L for females. Moreover, there is evidence from data on blood pressure, pulse rate, relative body weight, and serum insulin, that call for an even lower upper limit of 10 U/L, compared to the 28 U/L limit at the time of the investigation [51
]. Consumption of small amounts of alcoholic beverages as “social” drinking, that is not yet detected by the CAGE or AUDIT questionnaires, but is probably the main cause of elevated GGT activity which reflects a low level of liver damage, seems to also be associated with higher S-ASM levels in control subjects. However, the influence of chronic low-dose alcohol exposure appears to be different from the impact of acutely high levels, suggested by the different correlations between GGT and S-ASM after subdividing the healthy control sample according to the presence of binge-drinking episodes. These associations of S-ASM activity with liver enzyme activities warrant further investigation.
We detected a significant correlation between S-ASM and each patient’s alcohol levels at admission. A greater effect was observed in the male subsample, but no relationship was found with the predominantly consumed type of alcoholic beverage. It is also noteworthy that there were different correlation strengths for S-ASM and various classical biomarkers of alcohol-dependence. While the association was very strong for liver enzymes and CDT, it was much weaker for the well-accepted indicator, MCV. Similarly, significant correlations between ASM and the indicators of hepatic injury (GGT, ALT, and AST) have also been described for patients with a hepatitis C virus infection where ASM showed a high discriminative power [23
]. There could be a common link between ASM and hepatotoxicity that involves endoplasmic reticulum stress and cholesterol loading of mitochondria [52
] that are highly abundant in hepatocytes, possibly via mechanisms of transcriptional regulation recently identified for mitochondrial defects in lysosomal storage disorders, like Niemann–Pick disease caused by a genetic defect in the ASM encoding gene (Yambire K.F. et al. preprint under revision).
Withdrawal symptoms, assessed during the follow-up visit using the CIWA-Ar scale, were significantly related to S-ASM levels, and this correlation was stronger in females than in males. On the other hand, the strong and highly significant positive relationship, between the sub-item tremor during detoxification and S-ASM, was most prominent in males. Interestingly, when beta-endorphin levels from the same cohort were analyzed with respect to withdrawal severity, the female subgroup contributed mostly to the correlation of the sub-item score for impaired concentration with higher initial beta-endorphin levels, as well as a stronger decline during withdrawal [53
]. These sex-specific effects once more emphasize the importance of separate analyses for males and females in the field of alcohol addiction.
Although there is a need for very cautious interpretation due to the clearly different mechanisms at work, there is an intriguing association between tremor and ASM in a very different disorder—Parkinson’s disease—with tremor being one of the primary early symptoms. Following the initial identification of the rare p.L302P mutation in SMPD1
as a strong risk factor for Parkinson’s disease in Ashkenazi Jews [54
], two additional SMPD1
founder mutations were identified in this population [55
]. A pathogenic mechanism for Parkinson’s disease has been hypothesized that involves alterations of the autophagy–lysosome pathway based on additional genetic factors encoding lysosomal enzymes [56
Biomarkers for alcohol consumption, that are more reliable than self-reports and physiological assessments, are essential not only for diagnosis and treatment of alcohol-related disorders, but also for epidemiological studies of the health effects of alcohol itself, or of other exposure events with alcohol as a cofactor. No gold standard is yet available, and commonly applied biochemical markers are far from ideal with respect to their discriminatory power, as indicated by AUC values ranging from 0.21 to 0.67 [29
]. While new approaches, including protein markers [29
], DNA methylation patterns [57
], aldehyde-induced DNA, and protein adducts [58
], and even neuroimaging [59
] enhance the detection of heavy drinkers, there is still room for improvement, for example, by utilizing novel biomarkers or by developing of a composite score for excessive alcohol use screening [61
]. In our study population, S-ASM activity alone did not perform better than any of the analyzed markers (Table 2
, AUC = 0.77 compared to AUC ≥ 0.80 for conventional biomarkers in our cohort), and it also did not appear to be beneficial when this information was added to a combined model. However, S-ASM elevation might respond to a different threshold of alcohol consumption and/or span a different time window than CDT, MCV, or GGT. So far, only the gradual decrease during the short period of withdrawal treatment of up to 10 days has been described, characterized by a reduction ranging from 6% to 52% of the initial value, in this and previous studies [14
]. The rate of decline would be relevant in monitoring for relapse. The response time of S-ASM to different amounts and types of alcohol intake has not been investigated yet.
There was no indication that S-ASM levels are suitable to predict 24-month alcohol relapse, although the collected data appeared valid, as suggested by the observed and known strong predictive value of the number of previous withdrawal treatments. Further predictive factors have also been identified in this NOAH cohort. For example, a single nucleotide variant in OPRM1
, which encodes the mu opioid receptor binding the endogenous ligand beta-endorphin, was associated with an increased risk of more and earlier alcohol-related hospital readmissions [53
]. Moreover, a higher body mass index in male patients and higher craving scores [62
], as well as clinical Cloninger and Lesch typology classifications [63
] are suggested as easily accessible risk factors and promising tools. The failure of S-ASM to serve as a useful predictor of alcohol relapse, however, does not imply that other components and enzymes of the sphingolipid pathway could not serve as biomarkers. Analysis of their activities or of the serum sphingolipid profile has, thus, some potential, given the relationships of S-ASM with the alcohol biomarkers observed in this study.
Although offering promising results, our study has some limitations beyond the limited sample size. The group of healthy controls was not abstinent and might not be representative because they were recruited from a largely academic environment. They also differ from patients in parameters such as BMI, smoking, and certainly nutrition, including supply with vitamins, which could have an additional influence on the observed effects. Many patients had to be excluded during the screening process and, thus, the generalization of the patient data might also be limited. The frequency of relapse is certainly underestimated, and could contribute to the observed lack of an effect of ASM because we relied on medical records for readmission to the two study centers. We, therefore, may have missed patients treated at other centers, or who did not seek out medical advice at all. Our data need to be interpreted with caution because they do not reflect causal relationships, but are instead associational. Future studies should investigate the potential causation of these findings. Some aspects warrant verification in cell culture or animal models. Due to the explorative nature of the study, we have not corrected the p-values for multiple testing. As such, some strong and nominally highly significant associations would survive strict corrections, but still require independent verification in larger cohorts of mixed sexes, if possible.