Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Metabolomics reveals biomarkers of opioid use disorder

Abstract

Opioid use disorder (OUD) is diagnosed using the qualitative criteria defined by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). Diagnostic biomarkers for OUD do not currently exist. Our study focused on developing objective biological markers to differentiate chronic opiate users with OUD from chronic opiate users without OUD. Using biospecimens from the Golestan Cohort Study, we compared the metabolomics profiles of high opium users who were diagnosed as OUD positive with high opium users who were diagnosed as OUD negative. High opium use was defined as maximum weekly opium usage greater than or equal to the median usage (2.4 g per week), and OUD was defined as having 2 or more DSM-5 criteria in any 12-month period. Among the 218 high opium users in this study, 80 were diagnosed as OUD negative, while 138 were diagnosed as OUD positive. Seven hundred and twelve peaks differentiated high opium users diagnosed as OUD positive from high opium users diagnosed as OUD negative. Stepwise logistic regression modeling of subject characteristics data together with the 712 differentiating peaks revealed a signature that is 95% predictive of an OUD positive diagnosis, a significant (p < 0.0001) improvement over a 63% accurate prediction based on subject characteristic data for these samples. These results suggest that a metabolic profile can be used to predict an OUD positive diagnosis.

Introduction

More than fifty years have passed since Dole and Nyswander described opioid addiction as a metabolic disease, suggesting that opioids disrupt homeostasis to produce drug-seeking behavior in the face of adverse consequences1. An important issue in the addiction is that people exposed to opioids may develop dependence, but not Opioid Use Disorder (OUD)2. OUD is a chronic recurrent disorder that increasingly causes undesirable emotional states by involving the brain’s reward system and could include impaired social functioning3,4.

Despite significant advances in the genetics and neurobiology of addiction as a brain disease, and preliminary studies to discover biomarkers of OUD, validated systemic biomarkers for OUD do not exist5. Differential diagnosis of OUD is obtained through interview or questionnaire to determine if the patients meet the DSM-5 diagnostic criteria. These criteria include impaired control, social impairment, risky use, tolerance, withdrawal, craving, and continued use despite problems. Having at least two of the 11 criteria meets the diagnoses of OUD with the number of criteria met as an indicator of the severity of the OUD6.

Iran is a country with a high rate of opiate use. Opium is the main opiate used7. Our study focused on a random sample of opium users in the Golestan Cohort Study (GCS) in Iran, where more than 8400 individuals (about 17% of the participants) reported chronic opiate use with a median duration of use of 19 years8. 75% of the opium users in the full cohort used a combination of teriak and shireh with only 4 people reporting heroin use8.

We investigated urinary metabolomic profiles to reveal biomarkers that could differentiate high opium users who were diagnosed as OUD positive from high opium users who were diagnosed as OUD negative. Our investigation is important because optimized treatment relies on accurate diagnosis of OUD. The DSM-5, the predominant diagnostic instrument in psychiatry, has known limitations for diagnosis of substance abuse disorders9. Objective biological markers can improve the diagnosis that is currently based on subjective DSM-5 questionnaire. In addition metabolites that are increased or decreased in opium users diagnosed with OUD (compared with opium users not diagnosed with OUD) can be used to determine pathway perturbations, and lead to the identification of druggable or nutritional targets.

Materials and methods

Study population

The details of the GCS (a cohort of over 50,000 adults aged 40–75 living in Golestan Province, Northeast Iran) have been previously published10. The GCS was approved by appropriate ethics committees at Tehran University of Medical Sciences, the US National Cancer Institute (NCI, IRB# 07-C-N120), and the International Agency for Research on Cancer (IARC).

In 2018, a random sample of 451 GCS participants who reported long-term opium use and 92 never-users were recalled. They underwent a detailed interview using modified Persian and Turkman versions of the Section L of WHO Composite International Diagnostic Interview (CIDI, version 2.1) to diagnose lifetime OUD7, based on the Diagnostic and Statistical Manual of Mental Disorders, 5th edition (DSM‐5). The presence of 2 or more of the 11 criteria during any 12-month period of life was defined as lifetime OUD.

Variables considered for adjustment of the logistic regression models (see below) included age at enrollment, gender, tobacco use (current/former/never), BMI, and route of opium use (ingestion/inhalation). OUD diagnosis was the outcome for the models. Alcohol use was not included in the analysis, because it was rare in this population, and only 3.5% of participants reported ever using alcohol.

No subjects participated in drug-related treatment for addiction as part of the GCS. Cohort participants gave non-fasted spot urine samples which were stored at −20 °C until 2015 when they were transferred on dry ice to the NCI Biorepository and stored at −80 °C. Aliquots were then shipped to UNC Chapel Hill.

Sample selection

In this sample selection, we excluded individuals who had discordant reports of their opiate use compared with baseline (baseline users who reported no lifetime opium use at the recall visit and vice versa, n = 24), and those without a urine sample available (n = 8). We also restricted the current analysis to high opium users reporting equal to or more than the median intake (2.4 g per week), to reduce the chance of misclassification. The final sample used in the current study included 138 urine samples from high opium users who were diagnosed as OUD positive, and 80 urine samples from high opium users diagnosed as OUD negative. Urine samples were selected from an additional 80 subjects who reported that they had never used opium.

Untargeted metabolomics via ultra-performance liquid chromatography (UPLC) high-resolution mass spectrometry

Details of the sample preparation, data acquisition, data preprocessing and metabolite identification and annotation are provided in the Supplementary Material Section. In brief, urine samples were prepared according to published methods11, and untargeted metabolomics data were acquired on a Vanquish UHPLC systems coupled with a Q Exactive™ HF-X Hybrid Quadrupole-Orbitrap™ Mass Spectrometer (UPLC-HR-MS; Thermo Fisher Scientific). Data were processed using Progenesis QI (Waters Corporation). Peaks detected by UPLC-HR-MS were identified or annotated. Signals detected on our untargeted platform are matched to an in-house physical standards library that was developed by acquiring data for over 2000 chemical standards run under the same conditions to the study samples. The evidence basis for metabolite identifications and annotations are based on matching to our in-house library physical standards library (Ontology Level, OL), as well as to Public Databases (PD), and are detailed in the supplementary material.

Hypothesis testing

Statistical tests for the normalized peaks in the metabolomics profiles were conducted using a two-tailed t-test with the Satterthwaite correction for unequal variances or the chi-square test. Statistical analyses were conducted using SAS 9.4 (SAS Institute Inc., Cary, NC). In this exploratory metabolomics study, p-values were not adjusted for multiple testing12,13. The nominal p-values are reported for the following comparisons 80 high opium users diagnosed as OUD negative versus 138 high opium users diagnosed as OUD positive.

Logistic regression modeling

Logistic regression was used to model which peaks/metabolites were predictive of a positive OUD diagnosis. Several modeling approaches were used that included all normalized metabolomics peaks, or included subsets of peaks. Stepwise logistic regression procedures (criteria: model entry p < 0.1 and model removal p > 0.05) with standardization of continuous variables, was used for model selection. The Hosmer-Lemeshow goodness-of-fit test was used to assess the final model for adequacy. Receiver operating characteristics (ROC) curve and the area under the curve (AUC) were used to evaluate metabolites as predictors of OUD. Stepwise models were conducted, with and without subject characteristics as potential covariates, using:

  1. (a)

    712 peaks which differentiated high opium users who were diagnosed as OUD positive from high opium users who were diagnosed as OUD negative.

  2. (b)

    40 identified/annotated metabolites that differentiated the high opium users who were diagnosed as OUD positive from high opium users who were diagnosed as OUD negative (26 of these 40 metabolites also differentiated opium users from non-opium users).

  3. (c)

    14 identified/annotated metabolites that were unique to differentiation of high opium users who were diagnosed as OUD positive from high opium users who were diagnosed as OUD negative (but did not also differentiate opium users from non-opium users).

Pathway enrichment: high opium users who were diagnosed as OUD positive from the high opium users who were diagnosed OUD as negative

Pathway enrichment was conducted using the Mummichog software in Metaboanalyst 4.014. All features (m/z) remaining after filtering data were entered together with the p-value that was calculated for the comparison of high opium users who were diagnosed as OUD positive and high opium users who were diagnosed OUD as negative. A p-value cut-off of 0.01 was used to determine the size of the permutation group that the algorithm used for selecting significant features to match for all possible metabolites. A mass accuracy of 3 ppm was used as the threshold for annotations used in identifying candidate pathways. All possible metabolites which were matched by m/z were searched in the human reference metabolic network (hsamfn), and the null distribution of module activities was estimated by using 100 permutations of random lists drawn from the experimental reference feature list. The candidate pathways were based on the similarity of m/z.

Results

Sample characteristics

The subject characteristics for the 218 high Opium users who were diagnosed as OUD positive (138 subjects) or OUD negative (80 subjects) are provided in Table 1. For these study samples, the OUD diagnosis was associated at p < 0.1 with age at the time of enrollment (p = 0.029, OUD positive were 2 years younger than OUD negative), and route of opium exposure (p = 0.054, higher by inhalation than by ingestion), but was not associated with BMI, gender, or tobacco use.

Table 1 Subject characteristics of high opium users diagnosed as OUD positive and high opium users diagnosed as OUD negative.

Metabolic profiles of high opium user diagnosed as OUD positive versus high opium user diagnosed as OUD negative

Over 7714 UPLC-HRMS signals were obtained after data preprocessing. Hypothesis testing and fold change were determined for the normalized peaks in the metabolomics data set for the comparison of (a) high opium users diagnosed as OUD positive vs high opium users diagnosed as OUD negative, and (b) opium users vs non-opium users. Over 700 peaks (712) tested different by t-test (p < 0.10) between high opium users diagnosed as OUD positive versus high opium users diagnosed as OUD negative (Fig. 1). Forty of the 712 peaks were identified or annotated through matching to the in-house physical standards library (Table 2), while additional peaks were annotated using big data analytics (Table S1).

Fig. 1: Over 3700 peaks differentiated high opium users from non-opium users.
figure 1

712 peaks differentiated high opium users diagnosed as OUD positive from high opium users diagnosed as OUD negative. 193 peaks were unique to the differentiation OUD positive versus OUD negative high opium users. Metabolites were identified or annotated using an in-house physical standards library, and peaks were annotated using big data analytics.

Table 2 Signals that differentiated (p < 0.10) high OUD positive opium users from high OUD negative opium users (Fig. 1) matched to 40 metabolites in the in-house physical standards library.

Pathway enrichment

Pathway enrichment was conducted in Metaboanalyst14 using all 7714 peaks. A cut-off for pathway significance (p < 0.01) was used to determine the size of the permutation group that the algorithm used to determine the enrichment between high opium users diagnosed as OUD positive versus high opium users diagnosed as OUD negative. The candidate pathways based on the match of exact mass (<3 ppm) of key metabolites that are included in the known pathway map are provided in Table S2. The distribution plot of the Enrichment Factor versus −log10 (P) is shown in Fig. 2. High opium users diagnosed as OUD positive versus those diagnosed as OUD negative had an enrichment for pathways involving biotin (vitamin B7), folate (vitamin B9), cytochrome P450 metabolism, purine metabolism, keratan sulfate degradation, N-glycan degradation, and R group synthesis. Vitamin absorption, bioavailability, and utilization are known to be impacted by drug addiction15. Cytochrome P450s are involved in the metabolism of opium, and the slow versus fast metabolism has been associated with addiction16. Opioid use has been shown to alter purine metabolism17. Keratan sulfate is a glycosaminoglycan that is at significant levels in central and peripheral nervous systems18. N-glycan is required to express the correctly folded form of the delta-opioid receptor19. R group synthesis is associated with the FAD/FADH2 conversion of fatty acids.

Fig. 2: The distribution plot of the Enrichment Factor versus −log10(P) of the pathway enrichment analysis.
figure 2

Mummichog was used to evaluate the pathway enrichment of all features (m/z) with p < 0.01 based on the t-test for the comparison of high opium users who were diagnosed as OUD positive versus high opium users who were diagnosed OUD as negative was used as the threshold to determine the size of the permutation groups used by the algorithm.

Modeling approach 1

Stepwise logistic regression was used to determine which of the 712 peaks were predictive of an OUD positive diagnosis. First, the area under the ROC curve (AUC) was calculated using the subject characteristics (Table 1) of age at the time of enrollment and route of opium use. This base model resulted in an AUC of 0.625 (Figure S1a). Second, all 712 peaks that differentiated (p < 0.10) the high opium users diagnosed as OUD positive from high opium users diagnosed as OUD negative was modeled without including subject characteristics. This resulted in an AUC of 0.720 which was significantly different (p = 0.042) from the base model. Third, all 712 peaks that differentiated (p < 0.10) the high opium users diagnosed as OUD positive from high opium users diagnosed as OUD negative was modeled with age at the time of enrollment and opium use as covariates. This resulted in an AUC of 0.946, which was significantly increased (p < 0.0001) over the base model. Using this modeling approach, only 16 peaks were selected that were predictive of an OUD positive diagnosis (Table 3). Two of the 16 peaks matched to pterin (OL1) and tryptophan (OL2b) using the in-house physical standards library. Annotations using public databases are provided for 6 additional peaks, while 8 of the peaks remained unknown unknowns.

Table 3 Metabolites predictive of a positive OUD diagnosis in high opium users.

Modeling approach 2

Forty of the 712 peaks that differentiated (p < 0.10) high opium users diagnosed as OUD positive from high opium users diagnosed as OUD negative could be matched to the in-house physical standards library (Table 2, Fig. 1). Major differentiators included metabolites derived from opium use, tobacco use, involved in biopterins and vitamin B9, tryptophan metabolism, acetylation of amino acids, bile acids, fatty acids, and carnitine metabolism. In addition, N-Acetyl-S-(3,4-dihydroxybutyl)-L-cysteine (p = 0.023) and N-Acetyl-S-(2-carbamoylethyl)-L-cysteine (p = 0.083) were lower in urine of high opium users diagnosed as OUD positive vs high opium users diagnosed as OUD negative. N-Acetyl-S-(3,4-dihydroxybutyl)-L-cysteine and N-Acetyl-S-(2-carbamoylethyl)-L-cysteine (p = 0.083) are metabolic products of butadiene20 (BD) and acrylamide21 (AM), respectively. These metabolites have previously been detected in biospecimens from tobacco users at significantly higher levels than non-tobacco users. They are attributed to the metabolism of the parent compounds (AM and BD) that form during the curation process, or on combustion of tobacco22. They could also be formed in the curing and combustion of opium or other plant material.

Stepwise logistic regression using these 40 metabolites resulted in an AUC of 0.76, which was significantly increased (p = 0.0049) over the base model (Figure S1b). Including subject characteristics of age at the time of enrollment and route of opium exposure resulted in an AUC of 0.80, also significantly (p < 0.0001) increased from the base model. Metabolites that were predictive of an OUD positive diagnosis (Table 3) included tryptophan, pterine, sarcosine, N-acetylproline, azelate, octopamine, serine, and nicotine.

Metabolic profiles unique to the OUD positive versus OUD negative diagnosis

Five hundred and nineteen of the 712 peaks that tested different between OUD positive high opium users versus high OUD negative high opium users, also differentiated the opium users from non-opium users (Fig. 1). To provide a focus on only metabolites that are important to the diagnosis of OUD, the 519 signals that were also important to differentiation of opium users from non-opium users were excluded for this analysis. This resulted in 193 peaks unique to the differentiation (p < 0.10) of subjects diagnoses as OUD positive versus those diagnoses as OUD negative. Of these 193 peaks, only 14 peaks that defined the OUD diagnosis matched to the in-house physical standards library. These 14 metabolites are listed in Table 2, while the additional annotated peaks through public databases are provided in Table S1.

Eleven of the 14 peaks that matched to the in-house library that were most important to defining OUD included the following endogenous metabolites: pterine (p = 0.0011), 2,4-dihydroxypterine (p = 0.0695), sarcosine (p = 0.0263), phosphorylcholine (p = 0.0962), 6-carboxyhexonate (p = 0.021), lauroylcarnitine (0.0574), glycocholate (p = 0.0816), 3-methylhistamine (0.087), azelate (p = 0.0713), n-methyl-D-aspartic acid (p = 0.0488), and tryptophan (0.0378).

The biological significance of these 11 endogenous metabolites is summarized:

  1. (a)

    Pterin is part of biopterin and folate. Biopterins are cofactors for aromatic amino acid hydroxylases, which are involved in the synthesis of dopamine, norepinephrine, epinephrine, and serotonin, and trace amines23. The active form of folate (vitamin B9) is tetrahydrofolate which accepts and donates one carbon unit (methyl group). Dihydroxypteridine is involved in folate and riboflavin pathways24.

  2. (b)

    Phosphorylcholine is derived from phosphorylation of choline25, and sarcosine is an intermediate in the metabolism of choline to glycine26.

  3. (c)

    N-Methyl-d-aspartic acid (NMDA) is an agonist at the NMDA receptor and mimics the action of glutamate27, and tryptophan is in the neurotransmitter pathway28. Azelaic acid (AZA) is a competitive inhibitor of tyrosinase in vitro29.

  4. (d)

    Lauroylcarnitine is associated with fatty oxidation disorders involving acyl CoA dehydrogenase deficiency, and carnitine palmitoyltransferase I and II deficiency30. 6-Carboxyhexanoic acid is a medium-chain fatty acid derived from heptanedioic acid and is involved in the gut microbial biosynthesis of biotin31.

  5. (e)

    Glycocholate is a secondary bile acid, produced in the microbial flora of the colonic environment by bacteria32, and is absorbed and recirculated. Bile acids are important for absorption of hydrophobic nutrients, dietary fats and vitamins, and the regulation enzymes involved in cholesterol homeostasis.

  6. (f)

    3-Methylhistamine is a prominent metabolite of histamine, which has a role in allergy, inflammation, gastric acid secretion, and neurotransmission33.

Three metabolites (Table 2) derived from exogenous exposures were also important to the differentiation of the high opium users who were diagnosed as OUD positive from high opium users who were diagnosed as OUD negative. These included mono-isobutyl phthalate (p = 0.0485, +) and mono ethyl hexyl phthalate (p = 0.0852, +), which could arise as metabolic products following ingestion of phthalates that leach from plastics used in inhalation of opium. N-Acetyl-S-(3,4-dihydroxybutyl)-L-cysteine was also a differentiator (p = 0.0228, −), and is presumably derived as a metabolic product of BD intake associated with the curing or combustion of plant matter.

Modeling approach 3

Stepwise logistic regression using the 14 metabolites unique to the differentiation of OUD positive versus OUD negative, together with covariates of age at the time of enrollment and route of opium use resulted in an AUC of 0.751, which was significantly increased (p < 0.0005) over the base model (Figure S1c). Stepwise logistic regression using only the 14 metabolites (with no subject characteristics) resulted in an ACU of 0.706, which was not significantly increased (p = 0.127) over the base model. Results from Model 3 (with or without the covariates of age at time of enrollment and route of opium use) show 5 of the 14 metabolites (pterine, sarcosine, tryptophan, azelate, and N-Acetyl-S-(3,4-dihydroxybutyl)-L-cysteine) as predictive of a positive OUD (Table 3).

Discussion

Our study revealed metabolomics signatures of OUD in a predominantly Turkmen population of chronic high opium users. We provide metabolite identifications and annotations for 712 features detected using untargeted mass spectrometry that are important to the differentiation of high opium users diagnosed as OUD positive from high opium users diagnosed as OUD negative. None of these identifications are known metabolites derived from other drugs of abuse. Pathway enrichment analysis points to a general disruption in vitamin B9 (folate), vitamin B7 (biotin), cytochrome P450, purine, and glycan metabolism, and FAD/FADH2 conversion of fatty acids.

Stepwise logistic regression analysis of these 712 peaks, together with subject characteristics, resulted in 16 candidate peaks that predict 95% of the high opium users who were diagnosed as OUD positive.

Forty of the 712 features which differentiate high opium users who were diagnosed as OUD positive from high opium users who were diagnosed as OUD negative matched to an in-house physical standards library. Models constructed with only these 40 metabolites predicted 80% of the subjects diagnosed as OUD positive, selecting 8 metabolites as predictors. Predictors of an OUD diagnosis in these high opium users included an increase in three endogenous compounds (pterine, sarcosine, and octopamine), a decrease in four endogenous compounds (tryptophan, azelate, N-acetylproline, and serine), and an increase in nicotine. Fourteen metabolites were determined to be unique to OUD diagnosis, after subtracting analytes known to overlap with opium use. Models using these 14 metabolites predicted 75% of the subjects testing OUD positive and replicated pterine, sarcosine, tryptophan, and azelate as metabolite predictors.

Many identified or annotated metabolites that differentiated high opium users who were OUD positive from high opium users who were OUD negative play a significant role in neurotransmitter synthesis and signal transduction34. Tryptophan is the major amino acid precursor of serotonin (5HT). 5HT deficits have been implicated in physical symptoms and emotional dysphoria following withdrawal from opioids35. Alterations in sarcosine, serine, kyneurate, NMDA found in this study are consistent with the observations that glutamatergic signaling is disrupted by opioids36. Sarcosine (methyl-glycine) acts as an NMDA receptor agonist and a glycine receptor agonist37. Serine is converted to D-serine by serine racemace. D-serine acts as co-agonist with glutamate to activate NMDA receptors38. Kyneurate a metabolite of tryptophan metabolized to quinolinic acid acts as a NMDA receptor agonist39. Octopamine is a trace amine that is an agonist of TAAR1 receptors implicated in mediating the actions of drugs of abuse40. Methylhistamine is a histamine receptor (H3) agonist that inhibits the firing of cholinergic neurons in the ventral striatum and decreases dopamine release41.

Limitations of the current study include the (a) use of self-report for the amount of opium consumed, (b) assumption that symptoms over any 12-month period-of-time are accurately recalled, (c) estimated amount/grams of opium may vary within or among regions, and be underestimated42, (d) analyses were not stratified by the route of administration or type of opiate used due to the sample size, and because this paper focuses on a marker of OUD independent of route, (e) that the current study was not powered for multiple testing, and (f) that the metabolomic profiles are not quantitated, or replicated in this cohort or across cohorts.

The year that the individuals met a DSM-5 OUD diagnosis during their history of opium use is unknown because the DSM-5 interview was not conducted at the time of baseline urine collection. These urinary baseline metabolomic profiles presented herein could result from chronic opium use, and/or from inherent individual metabolic differences present prior to the acquisition of OUD.

Chronic use of opiates and opioids without meeting the criteria for DSM-5 OUD is not unique to the Turkman population for opiate use. Some chronic pain patients treated with prescription opioids and chronic users of illicit opioids do not meet the criteria for DSM-5 OUD. This suggests that the approaches used in this study are likely to be generalizable to other cohorts. This is also consistent with other substance use disorders where heavy use does not necessarily imply a substance used disorder43.

Research on biomarkers for OUD and other substance use disorders has focused on neuroimaging (MRI, fMRI, and PET) and EEG studies44. While these biomarkers may eventually be clinically validated in other populations, they will be costly to implement. In contrast, validation of biomarkers in other populations in accessible biological fluids (e.g., urine, blood, saliva) will be less costly, and easier to implement in general medical practice. In addition, these non-invasive biomarkers will be important complements to results from neuroimaging studies.

In conclusion, if the current results are replicated, the identification of peripheral biomarkers for OUD would represent a significant advancement in defining and managing the disease. It would further validate the Dole and Nyswander hypothesis that OUD is a brain disease in which metabolism is disrupted, and would provide biomarkers for OUD that could be used to optimize treatment. In addition, validation of the discovered metabolic perturbations related to vitamins and fatty acids could lead to the development of a nutrient cocktail to test in clinical settings for efficacy to mitigate symptoms that lead to the diagnosis of OUD.

References

  1. Dole, V. P. & Nyswander, M. E. Heroin addiction—a metabolic disease. Arch. Intern. Med. 120, 19–24 (1967).

    CAS  PubMed  Article  Google Scholar 

  2. Volkow, N. D. & Blanco, C. Medications for opioid use disorders: clinical and pharmacological considerations. J. Clin. Invest. 130, 10–13 (2020).

    CAS  PubMed  Article  Google Scholar 

  3. Strang, J. et al. Opioid use disorder. Nat. Rev. Dis. Prim. 6, 1–28 (2020).

    Article  Google Scholar 

  4. Ruffle, J. K. Molecular neurobiology of addiction: what’s all the (Δ) FosB about? Am. J. Drug Alcohol Abus. 40, 428–437 (2014).

    Article  Google Scholar 

  5. Bough, K. J. & Pollock, J. D. Defining substance use disorders: the need for peripheral biomarkers. Trends Mol. Med. 24, 109–120 (2018).

    PubMed  Article  Google Scholar 

  6. Hasin, D. S. et al. DSM-5 criteria for substance use disorders: recommendations and rationale. Am. J. Psychiatry 170, 834–851 (2013).

    PubMed  PubMed Central  Article  Google Scholar 

  7. Amin-Esmaeili, M. et al. Epidemiology of illicit drug use disorders in Iran: prevalence, correlates, comorbidity and service utilization results from the Iranian Mental Health Survey. Addiction 111, 1836–1847 (2016).

    PubMed  Article  Google Scholar 

  8. Khademi, H. et al. Opium use and mortality in Golestan Cohort Study: prospective cohort study of 50,000 adults in Iran. BMJ 344, e2502 (2012).

    PubMed  PubMed Central  Article  Google Scholar 

  9. Gorfinkel, L., Voon, P., Wood, E. & Klimas, J. Diagnosing opioid addiction in people with chronic pain. BMJ 362, k3949. https://doi.org/10.1136/bmj.k3949 (2018).

  10. Pourshams, A. et al. Cohort Profile: The Golestan Cohort Study-a prospective study of oesophageal cancer in northern Iran. Int. J. Epidemiol. 39, 52–59 (2010).

    PubMed  Article  Google Scholar 

  11. Want, E. J. et al. Global metabolic profiling procedures for urine using UPLC–MS. Nat. Protoc. 5, 1005–1018 (2010).

    CAS  PubMed  Article  Google Scholar 

  12. Xi, B., Gu, H. & Baniasadi, H. & Raftery, D. in Mass Spectrometry in Metabolomics 333–353 (Springer, 2014).

  13. Bender, R. & Lange, S. Adjusting for multiple testing—when and how? J. Clin. Epidemiol. 54, 343–349 (2001).

    CAS  PubMed  Article  Google Scholar 

  14. Chong, J. et al. MetaboAnalyst 4.0: towards more transparent and integrative metabolomics analysis. Nucleic Acids Res. 46, W486–W494 (2018).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  15. Grotzkyj-Giorgi, M. Nutrition and addiction-can dietary changes assist with recovery? Drugs Alcohol Today 9, 24 (2009).

    Article  Google Scholar 

  16. Kramlinger, V. M., Rojas, M. A., Kanamori, T. & Guengerich, F. P. Cytochrome P450 3A enzymes catalyze the O6-demethylation of thebaine, a key step in endogenous mammalian morphine biosynthesis. J. Biol. Chem. 290, 20200–20210 (2015).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  17. Mannelli, P. et al. Opioid use affects antioxidant activity and purine metabolism: preliminary results. Hum. Psychopharmacol. 24, 666–675 (2009).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  18. Caterson, B. & Melrose, J. Keratan sulfate, a complex glycosaminoglycan with unique functional capability. Glycobiology 28, 182–206 (2018).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  19. Markkanen, P. M. & Petäjä-Repo, U. E. N-glycan-mediated quality control in the endoplasmic reticulum is required for the expression of correctly folded δ-opioid receptors at the cell surface. J. Biol. Chem. 283, 29086–29098 (2008).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  20. Carmella, S. G. et al. Effects of smoking cessation on eight urinary tobacco carcinogen and toxicant biomarkers. Chem. Res. Toxicol. 22, 734–741 (2009).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  21. Bjellaas, T., Janak, K., Lundanes, E., Kronberg, L. & Becher, G. Determination and quantification of urinary metabolites after dietary exposure to acrylamide. Xenobiotica 35, 1003–1018 (2005).

    CAS  PubMed  Article  Google Scholar 

  22. Chang, C. M. et al. Biomarkers of tobacco exposure: summary of an FDA-sponsored public workshop. Cancer Epidemiol. Biomark. Prev. 26, 291–302 (2017).

    CAS  Article  Google Scholar 

  23. Nagatsu, T. & Ichinose, H. Regulation of pteridine-requiring enzymes by the cofactor tetrahydrobiopterin. Mol. Neurobiol. 19, 79–96 (1999).

    CAS  PubMed  Article  Google Scholar 

  24. Brown, E. G. Ring Nitrogen and Key Biomolecules: The Biochemistry of N-heterocycles (Springer Science & Business Media, 2012).

  25. Jimenez, B., del Peso, L., Montaner, S., Esteve, P. & Lacal, J. C. Generation of phosphorylcholine as an essential event in the activation of Raf-1 and MAP-kinases in growth factors-induced mitogenic stimulation. J. Cell. Biochem. 57, 141–149 (1995).

    CAS  PubMed  Article  Google Scholar 

  26. Ducker, G. S. & Rabinowitz, J. D. One-carbon metabolism in health and disease. Cell Metab. 25, 27–42 (2017).

    CAS  PubMed  Article  Google Scholar 

  27. Paoletti, P. & Neyton, J. NMDA receptor subunits: function and pharmacology. Curr. Opin. Pharm. 7, 39–47 (2007).

    CAS  Article  Google Scholar 

  28. Gheorghe, C. E. et al. Focus on the essentials: tryptophan metabolism and the microbiome-gut-brain axis. Curr. Opin. Pharm. 48, 137–145 (2019).

    CAS  Article  Google Scholar 

  29. Passi, S., Picardo, M., Mingrone, G., Breathnach, A. S. & Nazzaro-Porro, M. Azelaic acid-biochemistry and metabolism. Acta Derm.-venereologica. Supplementum 143, 8–13 (1989).

    CAS  Google Scholar 

  30. Yamada, K. & Taketani, T. Management and diagnosis of mitochondrial fatty acid oxidation disorders: focus on very-long-chain acyl-CoA dehydrogenase deficiency. J. Hum. Genet. 64, 73–85 (2019).

    CAS  PubMed  Article  Google Scholar 

  31. Estrada, P. et al. The pimeloyl-CoA synthetase BioW defines a new fold for adenylate-forming enzymes. Nat. Chem. Biol. 13, 668–674 (2017).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  32. Tanaka, H., Doesburg, K., Iwasaki, T. & Mierau, I. Screening of lactic acid bacteria for bile salt hydrolase activity. J. Dairy Sci. 82, 2530–2535 (1999).

    CAS  PubMed  Article  Google Scholar 

  33. Lieberman, P. The basics of histamine biology. Ann. Allergy, Asthma Immunol. 106, S2–5 (2011).

    Article  Google Scholar 

  34. Dinis-Oliveira, R. J. Metabolism and metabolomics of opiates: a long way of forensic implications to unravel. J. Forensic Leg. Med. 61, 128–140 (2019).

    PubMed  Article  Google Scholar 

  35. Welsch, L., Bailly, J., Darcq, E. & Kieffer, B. L. The negative affect of protracted opioid abstinence: progress and perspectives from rodent models. Biol. Psychiatry 87, 54–63 (2020).

    CAS  PubMed  Article  Google Scholar 

  36. Hearing, M., Graziane, N., Dong, Y. & Thomas, M. J. Opioid and psychostimulant plasticity: targeting overlap in nucleus accumbens glutamate signaling. Trends Pharm. Sci. 39, 276–294 (2018).

    CAS  PubMed  Article  Google Scholar 

  37. Zhang, H. X., Lyons-Warren, A. & Thio, L. L. The glycine transport inhibitor sarcosine is an inhibitory glycine receptor agonist. Neuropharmacology 57, 551–555 (2009).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  38. Wolosker, H. The neurobiology of d-serine signaling. Adv. Pharmacol. 82, 325–348 (2018).

    CAS  PubMed  Article  Google Scholar 

  39. Schwarcz, R., Bruno, J. P., Muchowski, P. J. & Wu, H. Q. Kynurenines in the mammalian brain: when physiology meets pathology. Nat. Rev. Neurosci. 13, 465–477 (2012).

    CAS  PubMed  PubMed Central  Article  Google Scholar 

  40. Liu, J. F. & Li, J. X. TAAR1 in addiction: looking beyond the tip of the iceberg. Front Pharm. 9, 279 (2018).

    Article  CAS  Google Scholar 

  41. Varaschin, R. K. et al. Histamine H3 receptors decrease dopamine release in the ventral striatum by reducing the activity of striatal cholinergic interneurons. Neuroscience 376, 188–203 (2018).

    CAS  PubMed  Article  Google Scholar 

  42. Mohebbi, E. et al. An exploratory study of units of reporting opium usein Iran: implications for epidemiologic studies. Arch. Iran. Med. 22, 541–545 (2019).

    PubMed  Google Scholar 

  43. Degenhardt, L. et al. Agreement between definitions of pharmaceutical opioid use disorders and dependence in people taking opioids for chronic non-cancer pain (POINT): a cohort study. Lancet Psychiatry 2, 314–322 (2015).

    PubMed  Article  Google Scholar 

  44. Moningka, H. et al. Can neuroimaging help combat the opioid epidemic? A systematic review of clinical and pharmacological challenge fMRI studies with recommendations for future research. Neuropsychopharmacology 44, 259–273 (2019).

    PubMed  Article  Google Scholar 

Download references

Acknowledgements

Dr. Blake Rushing and Dr. Krissy Kay contributed to quality control review of figures and tables. Mr. Justin Chandler assisted with literature review, and editorial review. The metabolomics platform was developed with funding in part by the NIH Common Fund Phase 1 Program (1U24DK097193, Sumner PI), and the NIEHS CHEAR program (U2CES026544, Fennell PI). This metabolomics investigation was funded in part through the NIDA Invest Fellowship (Dr. Reza Ghanbari), and the Sumner-Lab. The Golestan Cohort Study was supported in part by Tehran University of Medical Sciences (grant no. 81/15); Cancer Research UK (grant no. C20/A5860); the Intramural Research Program of the NCI, NIH; and various collaborative research agreements with IARC.

Author information

Authors and Affiliations

Authors

Corresponding authors

Correspondence to Reza Malekzadeh or Susan C. J. Sumner.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Disclosure

The views and opinions expressed in this manuscript are those of the authors only and do not necessarily represent the views, official policy or position of the U.S. Department of Health and Human Services or any of its affiliated institutions or agencies.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ghanbari, R., Li, Y., Pathmasiri, W. et al. Metabolomics reveals biomarkers of opioid use disorder. Transl Psychiatry 11, 103 (2021). https://doi.org/10.1038/s41398-021-01228-7

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1038/s41398-021-01228-7

Further reading

Search

Quick links