A potential paradigm shift in opioid crisis management: The role of pharmacogenomics

Abstract

Pharmacogenetic investigations into the opioid crisis suggest genetic variation could be a significant cause of opioid-related morbidity and mortality. Variability in opioid system genes, including single nucleotide polymorphisms, manifest after pharmacogenetic testing, as previously invisible risk factors for addiction and overdose. Pharmacodynamic genes regulate opioid-sensitive brain networks and neural reward circuitry. Pharmacokinetic genes expressed in drug metabolic pathways regulate blood levels of active vs. inactive opioid metabolites. Elucidating the complex interplay of genetic variations in pharmacokinetic and pharmacodynamic pathways will shed new light on the addictive and toxic properties of opioids. This narrative review serves to promote understanding of key genetic mechanisms affecting the metabolism and actions of opioids, and to explore causes of the recent surge in opioid-related mortality associated with COVID-19. Personalised treatment plans centred around an individual’s genetic makeup could make opioid-based pain management and opioid use disorder (OUD) treatments safer and more effective at both the individual and system levels.

Keywords:

• Opioids
• pharmacogenetics
• pain management
• overdose
• personalized medicine

Introduction

Opioid overdose has become an escalating epidemic. Opioid overdose deaths have more than quadrupled in the last 18 years, and the rise of synthetic opioids, such as fentanyl has caused a death toll rivalling that of automobile accidents (Cdcgov 2020a, 2021). These deaths cause a social breakdown on a large scale that is most apparent in vulnerable communities with limited access to healthcare resources.

Multifactorial issues like opioid addiction often involve several predisposing factors. Genetic, social, and environmental influences must be analysed to develop more effective measures for the prevention and treatment of opioid addiction. The multifactorial nature of opioid misuse is compounded by current events, such as the COVID-19 pandemic. Circumstances of the pandemic that increase the risk for opioid users include: (1) clinician-patient interaction is suppressed, and in-person contact is central to most successful opioid use management; (2) medical and community support systems are already being stretched to a breaking point; and (3) contagion-related fears may be dissuading treatment-naive users from seeking help for either long term use or acute adverse events (Alexander et al. 2020; Althobaiti et al. 2020; Becker and Fiellin 2020; Green et al. 2020; Khatri and Perrone 2020; Jenkins et al. 2021); (4) the pandemic has resulted in society-wide stress, leading to marked increases in the prevalence of depression (4×, 24 vs. 6%) and anxiety (3×, 25 vs. 8%) (Cdcgov 2020b), which likely lead to increased opioid consumption and reduced resilience against suicidal behaviour and intentional opioid overdose. A cohesive and comprehensive approach to assessing opioid use risk and managing opioid-related therapy is an integral part of improving system-level management of the opioid crisis, particularly in view of pandemic-related challenges and future care-disrupting events.

Clinical studies have shown significant variability in opioid efficacy and response, due in part to genetic factors. The estimated incidence of minor adverse events related to opioid-induced respiratory depression (OIRD) in the post-operative setting is around 30%, with the incidence of serious adverse events related to OIRD estimated to be <1% (Dahan et al. 2010). These statistics are likely to severely underestimate the incidence of these events in less controlled, non-clinical settings, which are the main drivers of the rising overdose death rate. Two individuals could have vastly different opioid use risks without knowing it, with genetically predisposed individuals having a greater risk of respiratory depression and death, even in response to small opioid doses (Dahan et al. 2010; Solhaug and Molden 2017; Singh et al. 2020). Genetic screening for markers of opioid sensitivity could possibly allow for better prediction of OIRD and overdose risk, more effective personalised pain management, as well as improved treatment for OUD in the future (Fudin and Atkinson 2014; Lloyd et al. 2017; Owusu Obeng et al. 2017; Kumar et al. 2019; Margarit et al. 2019; Muriel et al. 2019; Rocco et al. 2019; Pelkowski 2020; Singh et al. 2020; Rodriguez et al. 2021).

Current treatment regimens follow a trial-and-error approach when designing a therapeutic plan (Hassan et al. 2009; Zahari et al. 2016; Singh et al. 2020), and gene-guided personalisation may help avoid suboptimal outcomes, such as relapse, serious side effects, or even death. In the next section, we will review the current knowledge regarding opioidergic physiology to lay the groundwork for the development of novel strategies for treatment.

The opioidergic system

Opioid action

Opioids and opiates are a class of analgesic drugs widely used in a clinical setting for pain management, especially in the post-operative context. Opioid class drugs target both the ascending (sensory) and descending (modulatory) pain pathways to produce analgesia. At the level of the brainstem, opioids act on GABAergic interneurons in the Periaqueductal Grey (PAG) to disinhibit the antinociceptive activity of projection neurons to the rostral ventromedial medulla (RVM) which blocks the ascending transmission of pain signals from the spinal cord (Jalabert et al. 2011; Matsui et al. 2014; Owusu Obeng et al. 2017; Burns et al. 2019). At the level of the spinal cord, opioids act on pain transmission neurons in the dorsal horn, reducing their activity and subsequently the ascending transmission of nociceptive signals (Jalabert et al. 2011; Matsui et al. 2014; Owusu Obeng et al. 2017; Burns et al. 2019).

In addition to their analgesic effects, opioids are known to have euphoric properties which underlie their addictive capacity. As with many addictive stimuli, opioid use activates the dopaminergic reward circuitry in the brain. Opioid drugs target GABAergic interneurons in the rostromedial tegmental nucleus (RMTg) connected to dopaminergic ventral tegmental area (VTA) neurons which project to the nucleus accumbens (NAcc) (Jalabert et al. 2011; Matsui et al. 2014; Owusu Obeng et al. 2017; Burns et al. 2019). By disinhibiting the dopaminergic inputs to the NAcc, opioids reinforce activities and ‘seeking’ behaviours that lead to opioid abuse (Jalabert et al. 2011; Matsui et al. 2014; Owusu Obeng et al. 2017; Burns et al. 2019). This serves as an initial neurobiological basis for opioid addiction.

Tolerance and physical dependence on opioid medications are thought to be distinct phenomena from the addiction itself but may play important roles in potentiating behaviours that contribute to addiction(Williams et al. 2001; Christie 2008; Volkow and McLellan 2016). Chronic use of opioid medication causes and adaptive cell signalling cascades, leading to decreased sensitivity of opioid receptors and neurons to opioid molecules, thought to be tied to tolerance and physical dependence (Williams et al. 2001; Christie 2008; Volkow and McLellan 2016). Such changes are often more susceptible to reversal, and opioid abstinence can lead to a fairly rapid return to baseline opioid reactivity (on the order of days to weeks) (Williams et al. 2001; Christie 2008; Volkow and McLellan 2016). Decreased reactivity to opioidergic stimulation is thought to lead to a ‘dopamine mismatch’ where there is a discrepancy between expected reward based on behavioural and environmental triggers surrounding opioid use and the actual dopaminergic stimulus associated with opioid action in the RMTg (Williams et al. 2001; Christie 2008; Volkow and McLellan 2016). These factors can lead to a perceived increase in nociceptive stimulus and may potentiate emotional symptoms of withdrawal/cravings, potentiating behaviour to seek out increasing opioid doses in what is classically thought of as addiction (Williams et al. 2001; Christie 2008; Volkow and McLellan 2016). These changes are thought to be more persistent and are major factors involved in relapse after abstinence as situational triggers based on memory can persist long after the last opioid use (Williams et al. 2001; Christie 2008; Volkow and McLellan 2016). Other theories suggest that chronic opioid exposure may cause physical shrinkage of dopaminergic neurons through cellular signalling cascades leading to decreased dopaminergic activity (Russo et al. 2007) and that the ‘final common pathway’ of addiction involves neuroplastic changes in projections from the NAcc to areas of the prefrontal cortex resulting in decreased executive control (which would normally play a role in regulating behaviour associated with drug-seeking) and an abnormally elevated excitatory response to drug-related stimuli and reduced response to all other stimuli that would normally be rewarding (Kalivas and Volkow 2005). Current literature suggests that all of these play a role in the addictive process, but the relative contribution of each is yet to be fully elucidated

Both the initial analgesic and euphoric effects of opioid drugs are mediated by opioid receptors present on the cell bodies and axons of neurons in the nociceptive and reward pathways of the nervous system. There are three major types of opioid receptors found in humans: mu (µ) or MOR, delta (δ) or DOR, and kappa (k) or KOR (Jalabert et al. 2011; Crist et al. 2013; Sekhri and Cooney 2017; Burns et al. 2019; Crist et al. 2019; Kumar et al. 2019; Muriel et al. 2019). The MOR is traditionally considered to be the primary target for opioids because of its well-established role in endogenous analgesia, while the DOR is thought to have a very similar function and, in some cases, interact with and regulate MORs (Burns et al. 2019). Most common opioid class drugs have a similar effect on both MORs and DORs, with the major difference being the strength of their affinity for either receptor. There is some evidence suggesting that at the level of the spinal cord MORs and DORs are localised in slightly different areas, allowing them to modulate different components of the afferent pain pathway. DORs are more highly concentrated in layer 1 of the dorsal horn, which is the major synapse point for myelinated Aδ fibres, associated with the sharp, acute pain of noxious stimuli (Burns et al. 2019). MORs are highly localised in the substantia gelatinosa (layer 2) of the dorsal horn which is the major synapse point for unmyelinated C fibres, generally associated with long-lasting, diffuse pain (Burns et al. 2019). The role of KORs is not yet fully understood but they are concentrated in areas related to stress processing, such as the amygdala and prefrontal cortex. They are thought to inhibit dopamine release and their agonists may have dose-dependent anxiogenic effects, suggesting their activation may play a role in the aversive effects of withdrawal and in the emotional experience of pain (Burns et al. 2019; Massaly et al. 2019).

The MOR, DOR, and KOR are encoded by the genes OPRM1, OPRD1, and OPRK1, respectively (Jalabert et al. 2011; Crist et al. 2013; Sekhri and Cooney 2017; Burns et al. 2019; Crist et al. 2019; Kumar et al. 2019; Muriel et al. 2019). Variation in these genes and in the surrounding genetic regions have the potential to affect the structure, function, density, and localisation of opioid receptors in the CNS and subsequently the effect of a given dosage of an opioid in an individual, both in terms of analgesic and addictive properties.

Further downstream in neural opioidergic pathways, the resulting addictive/euphoric effects are mediated by dopamine and serotonin (Burns et al. 2019; Yuferov et al. 2021).

As mentioned before, dopamine released by VTA neurons into the NAcc reinforces behaviours deemed ‘rewarding’ and produces a feeling of euphoria. Dopamine receptors are classified as D1-like and D2-like. D1-like receptors include DRD1 and DRD5. These receptors increase cAMP and activate the intracellular signalling cascade while the D2-like receptors (DRD2, DRD3, and DRD4), inhibit this process (Jalabert et al. 2011; Zhu et al. 2013; Clarke et al. 2014; Jing Li et al. 2018; Burns et al. 2019). Dopamine levels are regulated in the brain by catechol-O-methyltransferase (COMT) through the degradation of catecholamines including dopamine and norepinephrine. Some studies suggest that DRD1 and DRD2 modulate opioid reinforcement, reward, and opioid-induced neuroadaptation (Clarke et al. 2014; Burns et al. 2019).

Serotonin released from the median raphe nucleus acts on dopaminergic centres in the limbic system. Serotonin is known to have numerous effects in the brain, but it is hypothesised that serotonin and dopamine serve reciprocal functions, with dopamine inducing appetitive or seeking behaviours, and serotonin countering these effects by inhibiting dopaminergic activity (Esposito et al. 2008). A major point of serotonin regulation is at the serotonin transporter (SERT), encoded by the gene SLC6A4. This transporter protein serves to remove serotonin from the synaptic cleft and recycle it into storage in the presynaptic neuron (Yuferov et al. 2021). Variation in dopamine receptor genes and serotonin transporter genes may alter an individual’s tendency for addictive behaviours, such as opioid abuse and may provide useful information regarding the individualised risk of opioid prescription for pain management.

Opioid kinetics

The pharmacokinetics of opioid drugs describes how opioids move around the body, including where they can travel while circulating in the blood, how long they can circulate and interact with opioid receptors in the CNS, and how much of a given dose of an opioid drug is active and able to produce classical analgesic and euphoric effects (Meyer and Maurer 2011; Barratt et al. 2014; Fudin and Atkinson 2014; Kuip et al. 2017; Owusu Obeng et al. 2017; Kumar et al. 2019; Saiz-Rodríguez et al. 2019; Singh et al. 2020). All these factors impact the strength of the opioid response experienced by an individual. Regardless of the dosage taken, opioid exposure is determined by how much and for how long the active form of an opioid is allowed to circulate and interact with the CNS (Reynolds et al. 2008; Meyer and Maurer 2011; Barratt et al. 2014; Fudin and Atkinson 2014; Tanaka et al. 2014; Kuip et al. 2017; Owusu Obeng et al. 2017; Kumar et al. 2019; Saiz-Rodríguez et al. 2019; Singh et al. 2020).

The most significant regulator of opioid kinetics is the opioid metabolic pathway. Opioid drugs undergo phase 1 metabolism in the liver by enzymes from the cytochrome (CYP) P450 family (Meyer and Maurer 2011; Owusu Obeng et al. 2017; Saiz-Rodríguez et al. 2019; Singh et al. 2020). The two primary enzymes involved in opioid metabolism are CYP2D6 and CYP3A4/5. CYP2D6 generally ‘activates’ opioid drugs which it metabolises, converting them into forms that have higher affinities for opioid receptors (Reynolds et al. 2008; Samer et al. 2010; Meyer and Maurer 2011; Owusu Obeng et al. 2017; Saiz-Rodríguez et al. 2019; Smith et al. 2019; Singh et al. 2020). CYP3A4/5 generally ‘deactivates’ opioid drugs by converting them into a form that has lower opioid receptor affinity and is more readily filtered from the blood and excreted by the kidneys (Samer et al. 2010; Meyer and Maurer 2011; Naito et al. 2011; Tanaka et al. 2014; Blanco et al. 2016; Owusu Obeng et al. 2017; Saiz-Rodríguez et al. 2019; Singh et al. 2020). Both 2D6 and 3A4/5 represent metabolic pathways for opioid drugs, but they are not equally active. The 3A4/5 enzyme is present in much higher concentrations than many other metabolic enzymes, processing the majority of opioids administered, but the specific ratio of a dose that follows either pathway is dependent on the specific drug being administered (Samer et al. 2010; Meyer and Maurer 2011; Naito et al. 2011; Tanaka et al. 2014; Blanco et al. 2016; Owusu Obeng et al. 2017; Saiz-Rodríguez et al. 2019; Singh et al. 2020). Other CYP enzymes, such as CYP2B6 may play a role in the metabolism of specific opioid drugs, but the clinical impact of polymorphic variants in their genes has been less clearly demonstrated empirically.

Some opioids also undergo phase 2 metabolism via UDP-Glucuronosyltransferase (UGT) enzymes, which play a role in conjugating lipophilic substrates with hydrophilic glucuronic acid to increase water solubility and facilitate renal excretion (Sastre et al. 2015; Blanco et al. 2016; Ning et al. 2019). The most well-known of these enzymes in relation to opioids is UGT2B7, which is key in the metabolism of morphine.

Variation in the genes of opioid metabolic enzymes affects the speed, efficacy, and the number of the enzyme in question (Meyer and Maurer 2011; Owusu Obeng et al. 2017; Saiz-Rodríguez et al. 2019; Singh et al. 2020). Genetic variations which lead to a slower, less effective, or significant reduction in the number of enzymes may impair an individual’s ability to process opioid drugs along a particular metabolic pathway. This would present clinically as a poor metabolizer (PM) phenotype (Crews et al. 2014; Madadi et al. 2018). Analogously, genetic variations leading to faster, more effective, or more enzymes than normal would enhance the specific metabolic pathway, causing an ultra-rapid metabolizer phenotype (UM) (Crews et al. 2014; Madadi et al. 2018; Smith et al. 2019). Individuals with normal metabolic capabilities are classified as extensive metabolizers (EM) and those with reduced capabilities are known as intermediate metabolizers (IM) (Crews et al. 2014; Madadi et al. 2018; Smith et al. 2019).

The clinical significance of each of these phenotypes is dependent on which metabolic pathway is affected. A PM phenotype in the 2D6 pathway (given codeine or oxycodone), or a UM phenotype in the 3A4/5 pathway (given fentanyl) may result in decreased analgesia, and thus lowered efficacy of opioid treatment (Samer et al. 2010; Meyer and Maurer 2011; Tanaka et al. 2014; Owusu Obeng et al. 2017; Saiz-Rodríguez et al. 2019; Smith et al. 2019; Singh et al. 2020). In contrast, a 2D6 UM (given codeine or oxycodone), or a 3A4/5 PM (given fentanyl) may have much more opioid exposure than other patients at a given dosage, putting them at higher risk for respiratory depression, central adverse side effects, and overdose (Samer et al. 2010; Meyer and Maurer 2011; Tanaka et al. 2014; Owusu Obeng et al. 2017; Saiz-Rodríguez et al. 2019; Smith et al. 2019; Singh et al. 2020; Rodriguez et al. 2021). Information about a patient’s metabolism phenotype can help inform pain management strategies, making them safer and more effective for the individual patient.

Pharmacogenetic-clinical connection

For the pursuit of gene-guided therapy in opioid pain management and OUD to yield better outcomes for patients, empirical evidence must demonstrate that variation in opioid system genes alters the safety or efficacy of use for commonly prescribed or abused opioid drugs. A summary of key genes and variants of interest that will be discussed can be found in Table 1.

Table 1. Key opioid system genes and variants of interest.

Gene name	Gene type	Key variants of interest	Associated opioids
CYP3A4	Drug metabolism	rs35599367	Fentanyl, Oxycodone, Buprenorphine
		rs2242480–rs2740574
CYP3A5	Drug metabolism	rs776746	Fentanyl, Oxycodone, Buprenorphine
CYP2D6	Drug metabolism	rs3892097	Codeine, Oxycodone
		rs16947–rs1135840
UGT2B7	Drug metabolism	rs7439366	Morphine, Buprenorphine
OPRM1	Opioid receptor	rs1799971	All opioids
OPRD1	Opioid receptor	rs678849	Buprenorphine
ABCB1	Transport protein	rs1045642	Buprenorphine

Fentanyl

Fentanyl is a synthetic opioid that is estimated to be about 60–100 times more potent than morphine, a consequence of its small molecular size and high lipophilicity which allow it to easily cross the blood–brain barrier (Fukuda et al. 2009; Wu et al. 2009; Zhang et al. 2011; Takashina et al. 2012; Barratt et al. 2014; Tanaka et al. 2014; Yuan et al. 2015; Kuip et al. 2017; Lloyd et al. 2017; Saiz-Rodríguez et al. 2019; Gerhard et al. 2020). Research has found large variability in the required fentanyl dose for analgesia, even in opioid-naive patients (Fukuda et al. 2009; Wu et al. 2009; Zhang et al. 2011; Takashina et al. 2012; Barratt et al. 2014; Tanaka et al. 2014; Yuan et al. 2015; Kuip et al. 2017; Lloyd et al. 2017; Saiz-Rodríguez et al. 2019; Gerhard et al. 2020). This variation may be tied to genetic variation in hepatic cytochrome P450 (CYP) enzymes.

Fentanyl is most extensively metabolised into the inactive metabolite norfentanyl via an N-dealkylation process, performed by CYP3A4 and CYP3A5 (Zhang et al. 2011; Barratt et al. 2014; Kuip et al. 2017; Saiz-Rodríguez et al. 2019; Gerhard et al. 2020). The most studied polymorphism in CYP-mediated fentanyl metabolism is the CYP3A5*3 allele. This produces an inactive copy of the CYP3A5 enzyme which has been shown to reduce fentanyl clearance when compared to the wild-type allele (CYP3A5*1) by 30–50% (Zhang et al. 2011; Takashina et al. 2012; Tanaka et al. 2014). Many researchers have observed an increased incidence of central adverse effects and up to a 2-fold higher normalised plasma concentration of fentanyl in individuals with the *3/*3 genotype when compared to other genotypes (Zhang et al. 2011; Takashina et al. 2012; Tanaka et al. 2014). Interestingly, CYP3A5*3 represents population diversity where allele frequencies range from 0.14 in sub-Saharan Africans to >0.95 in European populations (Ncbi 2021). Furthermore, the occurrence of CYP3A5*3/*3 homozygous genotype, varies in African populations, from as low as 0 to as high as 53% (Bains et al. 2013). Overall, there is significant evidence that the 3A5*3 polymorphism increases exposure to fentanyl.

Other relevant polymorphisms are in the CYP3A4 gene, notably CYP3A4*1G and CYP3A4*22. Patients with the *1G/*1G genotype have a significantly lower fentanyl metabolic rate compared to other genotypes and *1G carriers had significantly lower levels of 3A4 mRNA (Zhang et al. 2011; Yuan et al. 2015). Few investigations have been conducted on the 3A4*22 allele and no significant influence on plasma fentanyl concentrations has been demonstrated (Barratt et al. 2014).

It has been observed that CYP3A inhibitors have a relatively small effect on fentanyl pharmacokinetics compared to other CYP3A substrates (Kuip et al. 2017). It was hypothesised that fentanyl’s high extraction ratio caused this, meaning the clearance rate was most affected by hepatic blood flow (Kuip et al. 2017). However, it was found that levels of fentanyl and its metabolic products in urine did not match the original dose, leading to the suggestion of an undiscovered metabolic path (Kuip et al. 2017). This pathway may also involve a significant and highly variable genetic component which would broaden our understanding of fentanyl pharmacogenetics.

Carfentanil is another synthetic opioid often compared to fentanyl. It is traditionally used as part of tranquiliser medications for large animals and has essentially no indications for human use. It is a very specific MOR agonist and follows the same major metabolic pathway as fentanyl (CYP3A4) but is noted to have a potency 20 times higher than that of fentanyl and 10,000 times more powerful than morphine (Leen and Juurlink 2019). Carfentanil is often cheaper and easier to manufacture than other types of opioids or other drugs that are distributed in a similar form (cocaine, methamphetamine, etc.) so there have been many instances where illicit drug manufacturers lace their products with carfentanil to reduce costs while increasing potency (Leen and Juurlink 2019). It has been suggested that a significant portion of modern drug-related mortality is due to overdose or other adverse reactions to illicit drugs laced with carfentanil (Leen and Juurlink 2019). As the studies on fentanyl largely focus on clinical settings and specific populations (e.g. cancer patients from one ethnic group), future research is necessary to investigate carfentanil specifically and connections between demographic factors (such as socioeconomic status), and illicit use with the demonstrated impact of the polymorphisms discussed.

Codeine/morphine

Codeine and morphine are opiate analgesics widely prescribed for pain control. Codeine is considered a prodrug as most of its analgesic effect results from metabolism into morphine (known to have a 200 times stronger affinity for the MOR) (Caraco et al. 1996; Crews et al. 2014; Madadi et al. 2018). The major pathway governing codeine metabolism into morphine is centred on the hepatic CYP2D6 enzyme (Caraco et al. 1996; Crews et al. 2014; Madadi et al. 2018).

In EMs about 80% of an administered codeine dose is metabolised into an inactive form (Caraco et al. 1996; Crews et al. 2014; Madadi et al. 2018; Smith et al. 2019). A larger percentage is converted to morphine in UM patients which has been shown to induce symptoms of opioid toxicity, such as sleepiness, confusion, and respiratory depression, even when administering a standard dose (Caraco et al. 1996; Crews et al. 2014; Madadi et al. 2018; Smith et al. 2019). PM patients have trouble achieving satisfactory analgesia with regular codeine doses, showing decreased morphine levels compared to EM individuals (Caraco et al. 1996; Crews et al. 2014; Madadi et al. 2018; Smith et al. 2019). Some studies have shown the PM phenotype to be more resistant to gastrointestinal side effects of codeine administration than EM, yet these studies were not able to show significant differences in the incidence of central adverse side effects between PMs and EMs (Caraco et al. 1996; Crews et al. 2014; Madadi et al. 2018; Smith et al. 2019).

Clinical guidelines made by the Clinical Pharmacogenetics Implementation Consortium (CPIC) recommend that patients with either UM 2D6 or PM 2D6 phenotypes avoid codeine and other opioids which follow similar metabolic pathways to avoid the risk of opioid toxicity or insufficient pain relief, respectively (Crews et al. 2014; Madadi et al. 2018; Ruano and Kost 2018; Crews et al. 2021). Alternative analgesics are recommended for pain control in both cases and ethnic variations in percentage frequency are notable. African Americans have a frequency of 5.38% of PM 2D6 genotype while East Asians and European ancestry groups have a prediction frequency of 0.84 and 8.45%, respectively (Gaedigk et al. 2017). Similarly, UM 2D6 genotype varies among African Americans, East Asians, and Europeans (LLerena et al. 2014).

Morphine metabolism is mediated by the activity of UGT2B7. This process produces two major metabolites: morphine-3-glucuronide (M3G) which is an inactive metabolite accounting for about 60% of the original morphine dose, and morphine-6-glucuronide (M6G) which has higher analgesic activity than morphine and accounts for 5–10% of the original dose (Ning et al. 2019). Both metabolites are more hydrophilic than morphine and thus UGT2B7 glucuronidation accounts for about 66% of morphine excretion (Ning et al. 2019). Polymorphisms in UGT2B7 may affect both the activity of the enzyme overall and alter the proportion of M3G and M6G produced (Ning et al. 2019). The most well-studied polymorphism in this gene is C802T (rs7439366), and the variant T allele has been associated with significantly higher pain scores and plasma morphine concentrations in patients given equivalent morphine doses (Ning et al. 2019).

The most well-known SNP linked with the modulation of morphine-induced analgesia is the A118G polymorphism (rs1799971) in the OPRM1 gene. Many studies have shown some association between the G allele variant and the incidence and severity of OUD (Reynolds et al. 2008; Mura et al. 2013; Taqi et al. 2019). Other studies have suggested a role in dysregulation of the hypothalamic-pituitary-adrenal axis, causing a heightened stress response that may predispose individuals to addictive tendencies (Chong et al. 2006). Some biochemical assays have revealed that the 118G variant receptors have a significantly stronger binding affinity for endogenous opioids, such as B-endorphins when compared to wild-type receptors, with no increased affinity for exogenous opioids, like morphine (Wu et al. 2009). It is hypothesised that the mutant allele reduces receptor expression and alters the signal transduction cascade, lowering the efficiency of exogenous opioid signalling (Wu et al. 2009). Some studies have shown statistically significant associations between the 118G allele and reduced analgesic response to opioid administration, demonstrated by elevated pain scores and required doses (Reynolds et al. 2008; Fukuda et al. 2009; Wu et al. 2009; Blanco et al. 2016; Lee et al. 2016; Rocco et al. 2019; Taqi et al. 2019). Some results have even shown the 118G variant to be associated with faster recovery from anaesthesia and a lower occurrence of adverse side effects, which may indicate a protective feature of the variant allele (Wu et al. 2009).

Oxycodone

Oxycodone is a semisynthetic opioid commonly prescribed for controlling post-surgical pain. Its metabolism is mediated by CYP2D6 and CYP3A4/5, which are highly polymorphic in nature (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014; Lloyd et al. 2017).

Noroxycodone is the major circulating metabolite of oxycodone, accounting for ∼80% of the original dose (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014). It is the product of a CYP3A4/5 catalysed reaction and is less active than oxycodone (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014). The CYP2D6 enzyme catalyses the conversion of oxycodone into oxymorphone and noroxycodone into noroxymorphone (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014). Both products of 2D6 metabolism have a higher affinity for the mu-opioid receptor than oxycodone, making them more potent analgesics (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014; Ruano and Kost 2018; Crews et al. 2021). Oxymorphone is the most active metabolite, with a 40-fold higher MOR affinity than oxycodone and a 3-fold higher affinity than morphine (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014).

Genetic studies have found that 2D6 EMs have significantly higher oxymorphone exposure and conversion of oxycodone to oxymorphone than IMs and PMs (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014). The 2D6 PM phenotype was also associated with a prolonged half-life of oxycodone in circulation (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014). Higher 2D6 activity scores have been associated with higher oxymorphone exposure and lower noroxycodone exposure, but significant differences between the metabolism phenotypes have not been shown in pain scores, side effects, or dose efficacy (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014).

Studies of genetic variance in CYP3A4/5 enzymes focussed on the CYP3A5 gene, showing that the trough concentration of noroxycodone and the ratio of noroxycodone to oxycodone were significantly higher wild type allele carriers (3A5*1) compared to non-functional homozygotes (3A5*3) (Samer et al. 2010; Naito et al. 2011). As these studies have relatively small sample sizes, further investigation is necessary to fully understand this relationship.

Pharmacokinetic studies have revealed that inhibition of either enzyme leads to a shift to the alternate metabolic pathway (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014). Inhibition of the 2D6 pathway decreases maximum serum concentration and exposure to oxymorphone/noroxymorphone while increasing exposure to oxycodone and noroxycodone, especially in 2D6 UMs (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014). Inhibition of the 3A pathway causes a shift towards the 2D6 pathway, increasing exposure to oxymorphone while decreasing exposure to noroxycodone and noroxymorphone (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014). The co-administration of drugs that inhibit the CYP3A pathway and oxycodone to patients with CYP2D6 polymorphism, most notably the UM phenotype, may greatly increase the risk of respiratory depression and opioid toxicity (Samer et al. 2010; Naito et al. 2011; Fudin and Atkinson 2014).

Methadone/buprenorphine

Opioid Replacement Therapy (ORT) aims to transition the user from an abused opioid to a clinically controlled regimen of slower-acting opioids (Crist et al. 2013, 2018, 2019). The goal is to achieve moderate analgesia during the transition period away from opioid abuse while avoiding the addictive euphoric spikes associated with faster-acting opioids (Crist et al. 2013, 2018, 2019). Currently, methadone (MET) and buprenorphine (BUP) are the two medications used widely in ORT. With the global rise in this epidemic, medication, such as BUP and MET are now used in combination with counselling, behavioural therapies, and enhanced patient monitoring. Collectively, this approach is known as medication-assisted treatment (MAT) which is now common practice. The outcomes measured with a MAT approach are cognitive, physical, occupational, behavioural. social, and neurological. Studies have found that patients with OUD improved better with MAT than just ORT. For the scope of our study, we will focus on the genetics of ORT medications (Maglione et al. 2018; Maiti et al. 2020).

MET is a mu-opioid receptor (MOR) agonist metabolised slowly by the liver enzyme CYP3A4 into an inactive form (Hassan et al. 2009; Crist et al. 2013, 2018, 2019; Victorri-Vigneau et al. 2019). MET comes in 2 enantiomeric forms (R-MET and S-MET) which are given in a racemic mixture as a dose for ORT. R-MET is significantly more active with regards to opioid effect (analgesia and respiratory depression) (Victorri-Vigneau et al. 2019), while S-MET has a larger effect on voltage-gated potassium channels, which suggests they may play a role in QT prolongation, increasing the risk of cardiac arrhythmias (Victorri-Vigneau et al. 2019). CYP2B6 has been thought to play a major role in MET pharmacogenetics, but recent studies have shown 2B6 to be selective for S-MET, and that polymorphisms in its gene have little effect on the efficacy of MET in ORT with respect to relapse rate or adverse effects, but more research is needed to elucidate their impact on the safety profile of MET, relating to changes in QT length (Victorri-Vigneau et al. 2019).

BUP is a weaker partial agonist of the MOR and kappa-opioid receptor (KOR) when compared to its active metabolite, and an antagonist of the DOR (Crist et al. 2013, 2018, 2019; Blanco et al. 2016). It is metabolised most significantly by CYP3A4 into the active metabolite norbuprenorphine (NBUP) (Huang et al. 2001). NBUP has higher intrinsic activity than BUP. NBUP is a potent agonist for the MOR and DOR and a partial agonist of the KOR (Huang et al. 2001). Intriguingly, NBUP is also noted to produce significantly greater respiratory depression with negligible antinociceptive efficacy (Huang et al. 2001).

This odd behaviour of NBUP is thought to be a product of its high affinity for P-glycoprotein (PGP), encoded by the ABCB1 gene (Huang et al. 2001; Hassan et al. 2009; Liao et al. 2017). This protein is found in many tissues including the blood–brain barrier and acts to pump substances out of these tissues (Hassan et al. 2009; Takashina et al. 2012; Wolking et al. 2015; Liao et al. 2017). It is hypothesised that NBUP has a little analgesic effect because it is rapidly removed from the brain via PGP whereas BUP can interact with the neural pain circuits due to its low affinity for the PGP (Huang et al. 2001; Hassan et al. 2009; Liao et al. 2017). The other relevant substrate, MET, is thought to have a moderate affinity for PGP (Hassan et al. 2009; Zahari et al. 2016).

BUP is also known to be metabolised by UGT2B7 into the inactive metabolite buprenorphine glucuronide (Sastre et al. 2015; Blanco et al. 2016). It has been suggested that the C802T polymorphism increases the enzyme’s affinity for BUP and thus increases its glucuronidation activity (Sastre et al. 2015; Blanco et al. 2016). This would decrease the analgesic efficacy of BUP in patients carrying the T allele, and this has been corroborated by studies showing higher pain scores and increased incidence of severe pain in post-operative T allele carriers being managed on BUP (Sastre et al. 2015; Blanco et al. 2016).

In studies by Crist et. al, it was shown that an intronic (non-coding) variant in OPRD1 (rs678849) can predict the outcome of opioid dependence treatment using either BUP (Suboxone) or MET (Methadone) (Crist et al. 2013, 2018, 2019). The ‘T’ allele of rs678849 appeared to increase the risk of opioid use in patients of Black ancestry being treated with MET while being protected against misuse in BUP patients (Crist et al. 2013, 2018, 2019). This finding is unique in its focus on patients of Black ancestry; however, the sample size is quite small (Crist et al. 2013, 2018, 2019). This highlights the need for a better representation of patients of Black ancestry in genetic studies. These findings suggest that treating patients based on genotype at this locus could help significantly improve opioid treatment efficacy.

Crisis in crisis: The collision of the COVID-19 pandemic and the opioid use epidemic

Though the current pandemic has deeply affected individuals from all walks of life, vulnerable populations, including opioid users, are among the hardest hit. Opioid medication procurement, both illicit and in the form of ORT, is largely dependent on in-person activities and social contact which puts opioid users at a greater risk of contracting and spreading COVID-19 (Jenkins et al. 2021).

While social distancing measures may be able to mitigate this risk, they serve as barriers for vulnerable populations to access ORT and other medical management. Opioid clinics operating at reduced capacity usually increase wait times and may force patients to seek treatment or symptomatic relief from other unregulated sources. During past crises, such as hurricanes (which also limit access to opioid clinics) research has shown <30% of ORT patients in the affected areas were able to obtain sufficient take-home doses of their ORT medications (Green et al. 2020). With COVID-19 such limitations likely apply and are occurring on a global scale.

Limited services available from community health providers severely restrict access to preventative medicine and counselling, while overstretched emergency departments are less able to act as catch-all ‘safety nets’. Telehealth initiatives can bridge some of these gaps, but may not be accessible to all patients, especially those in financial distress or with unstable arrangements for housing and internet access (Khatri and Perrone 2020).

Societal stress associated with disruptions to social interaction, community services, shelters, and food banks may exacerbate concerns regarding food and housing security while weakening individual support networks and community-based resiliency. As mentioned in the introduction, it is estimated that depression has increased 4-fold and anxiety has increased 3-fold in the US since the beginning of the pandemic (Cdcgov 2020b). These trends have been shown in the literature and in COVID-19 statistics to be linked with higher rates of drug use, with an estimated 13.3% of Americans either starting or escalating substance use to mitigate pandemic-related stress (Cdcgov 2020b). Increased depression and anxiety along with social distancing measures may promote more risky opioid use (including using alone), and such use could lead to intentional opioid overdose as rates of suicidal ideation more than doubled (10.7 vs. 4.3%) in 2020 compared to the 2018 level (Cdcgov 2020b).

All these factors coalesce into a ‘perfect storm’ for opioid-related mortality to surge, as forces driving opioid use are high and support systems are severely limited. The CDC reports a nearly 20% increase in the number of overdose deaths compared to the year before the pandemic, and that this represents the highest number of yearly overdose deaths ever recorded (Cdcgov 2020a, 2021). Some areas are seeing increases in emergency medical services (EMS) runs related to opioid overdose, and increased opioid-related deaths upon EMS arrival, while there is a marked decrease in the total number of non-opioid EMS runs (Slavova et al. 2020). Perhaps even more worrying is the significant increase in opioid-related EMS calls wherein the patient refused transport to the emergency department (Slavova et al. 2020). This refusal may reflect a justified fear of COVID-19 infection, however, the resulting aversion to emergency medical resources will likely increase rates of untreated OIRD and overdose deaths.

Now more than ever algorithms are needed to streamline the management of patients with OUD to ensure resources are effectively distributed to those who need them most. Using genetic testing to build opioid use risk profiles for patients can help physicians triage patients for follow-up, support better education for patients regarding their individual risks, and assist in designing more effective treatment strategies to prevent overdose and relapse.

Limitations

Limited research has been done on the pharmacogenetics of opioid use, especially with respect to the impact of genetic factors on the risk profile for specific opioids (like fentanyl or oxycodone). Given that the use of opioid medication is generally tightly controlled (with significant harm risks to study participants), the body of evidence supporting a given polymorphism’s impact can be quite variable. In many cases, such impact is suggested based on other related research, animal models, retrospective analyses, or on prospective studies with small sample sizes, skewed study populations, or open-label designs. Measurement of study outcomes can also be affected by relying on subjective, self-reported pain metrics. Confounding demographic factors, such as sex have not yet been shown to play a significant role in opioid pharmacogenetics (Chartoff and McHugh 2016), but not enough studies have been done on these to rule them out as confounds. Systemic imbalances in study sampling regarding race, socioeconomic status, and sex persist in opioid research. All of these factors can impact the reliability of the results obtained. This paper aims to provide a narrative overview of ongoing research in the field and to serve as a basis to encourage more extensive research in the future. At the same time, we recognise the necessity for critically analysing evidence from the literature to better understand the inherent risk of bias (RoB) in the evidence and to compare the relative strengths and weaknesses. As such we have completed a modified RoB assessment (adapted from RoB2 and ROBINS-I) (Sterne et al. 2016, 2019) for the primary research articles referenced, which can be found in Supplementary Table 1. Each study was rated by the authors on five categories (sample size, confounding, participant selection, measurement of outcomes, and selection of reported results), and globally, on a 4-point scale (low risk, moderate risk, serious risk, critical risk) risk (Sterne et al. 2016, 2019). The majority of studies assessed fell into the low to moderate RoB range, with a few meeting the criteria for serious. Major categories with more serious bias risks in the reviewed papers were confounding and patient selection. The risk of bias with respect to confounding was largely due to study design, with many studies being open-label and non-randomised. The risk of bias related to participant selection was largely due to highly specific sample populations which limits the generalisability of the findings.

Recommendations

• Design a genetic risk assessment panel for opioids using a combination of pharmacokinetic and pharmacodynamic genes to obtain a more comprehensive assessment of individual opioid use risk

• Frequent updates and adjustments to the contents of the panel, to keep pace with the rapidly changing landscape of opioid pharmacogenetic research

• Genetic risk screening for any patients being considered for opioid pain management or opioid replacement therapy and an algorithmic approach for tailoring these treatments, anchored upon individual patient’s genetic risk factors

• Continue to support increased public screening and education initiatives in communities at high risk for OUD to expose hidden risk factors for opioid use and train community members to recognise and respond to possible overdoses

• Increase public awareness and scrutiny regarding the contents of drugs being distributed illicitly and the dangers of fentanyl/carfentanil lacing

Conclusion

Current pharmacogenetic research shows evidence that interindividual differences in opioid efficacy may have important links to polymorphisms in opioid system genes. Widespread genetic screening and investigations on communities vulnerable to opioid abuse may help expose hidden genetic risks related to opioid use and may allow for the design of treatments that mitigate or avoid these risks. Innumerable opioid deaths could be prevented by uncovering and confronting these risks.

Acknowledgements

We would like to thank Mr. Sheraz Cheema for administrative assistance, and for support from the Summer Undergraduate Research Program and the Translational Research Program, both in the Institute of Medical Science, University of Toronto.

Disclosure statement

J.L.K. is a member of the Scientific Advisory Board of Myriad Neuroscience(unpaid) and holds several patents relating to pharmacogenetic tests for psychiatric medications. D.E.J. and A.G.M. have no conflicts of interest to disclose.