Improving Translational Research Outcomes for Opioid Use Disorder Treatments

Abstract

Purpose of Review

Pharmacotherapies are the most effective means of reducing the harms associated with opioid use disorder (OUD). Translational research seeking to develop novel medications to treat OUD has been challenging due to the complex etiology of addiction. Preclinical outcome measures are often behavioral, and it is difficult, if not impossible, to fully mirror the various emotional and cognitive processes that motivate opioid use in humans. The goal of the current narrative review was to summarize the translational progression of three potential medications for OUD, which had varying levels of success.

Recent Findings

Memantine, lorcaserin, and lofexidine all showed promise in preclinical studies; however, only lofexidine was able to consistently replicate these findings in human subjects, and receive FDA approval. It was the authors’ objective to use this review to identify areas of needed improvement in translational research for OUD.

Summary

Preclinical studies vary significantly in their ability to forecast effectiveness in clinical trials. Among the various preclinical models, suppression of opioid self-administration appears to have the best predictive validity. As they model a mostly physiological phenomenon, preclinical assessments of opioid withdrawal also appear to have high predictive validity. In our review of the literature, the authors noted numerous examples of clinical trials that were underpowered, lack precision, and proper outcomes. Better-validated preclinical targets and improved design of proof-of-concept human studies should allow investigators to more efficiently develop and test medications for OUD.

Introduction

The opioid crisis was declared a national emergency in 2017, and significant government, non-profit, and industry resources have been invested in medication development for opioid use disorder (OUD). The largest coordinated effort to speed medication development has come from the National Institutes of Health (NIH) Helping to End Addiction Long-term (HEAL) initiative [1]. HEAL objectives for developing OUD treatments include repurposing existing medications for OUD indications and developing novel compounds to prevent or treat OUD. Historically, drug development for OUD (or any substance use disorder) has been notoriously difficult, as preclinical models do not always translate to the human experience of addiction, and there is a relatively high bar for FDA approval of an OUD medication, e.g., the medication is required to demonstrate an increased likelihood of opioid abstinence and/or reduce the severity of opioid withdrawal symptoms relative to placebo.

Translational research on OUD medications is especially challenging because the primary outcomes are often behavioral, and the emotional and cognitive processes that drive human opioid use are impossible to fully recreate in preclinical models. Nonetheless, efforts to improve preclinical evaluation of OUD medications are now emphasizing evidence of sustained decreases in opioid self-administration and selectivity for suppressing opioid vs. natural non-drug rewards [2, 3]. In addition, preclinical mechanistic research on OUD has been especially important in identifying novel and promising avenues for medication development, although it is notable that some new medications might be most effective in certain OUD treatment conditions or in sub-populations of OUD patients [4].

As researchers refine their approach to clinical trials for OUD, it is important to learn from previous translational research outcomes to enhance the successful translation of preclinical findings to the clinic. This review will examine three medication development efforts that had varying levels of success: lorcaserin, an example of a medication that showed some promise in preclinical studies, but ultimately failed in clinical trials; memantine, a medication that yielded mixed results in clinical trials and has not been widely adopted; and lofexidine, a successful effort in medication development that resulted in FDA approval for use in OUD patients. For each medication, we will examine preclinical and clinical evidence of efficacy for various aspects of OUD treatment. The goal of this review is to highlight areas of needed improvement in translational research for OUD, to more efficiently develop therapeutics.

Lorcaserin

Preclinical

Lorcaserin (Belviq, Belviq XR ^®) is a selective 5-HT2C receptor agonist that was approved by the FDA in 2012 for the treatment of obesity but withdrawn in 2020 after being linked to an increased incidence of pancreatic, colorectal, and lung cancers [5, 6]. Serotonin (5-HT) neurotransmission has modulatory control over the limbic-corticostriatal circuitry that is involved in reward and adaptive behaviors that are often dys-regulated in addictive disorders [7]. Before its removal from the market, several preclinical studies evaluated the potential of lorcaserin as a therapeutic in animal models of OUD, including intravenous self-administration, in which the ability of a medication such as lorcaserin to reduce drug-taking behavior (i.e., drug reinforcement) is an indicator of decreased motivation to use drugs, and an important indicator of treatment potential [8]. In a preclinical self-administration study (male, Sprague Dawley), lorcaserin pretreatment (1 mg/kg, intraperitoneal (i.p.)) inhibited intravenous (i.v.) oxycodone intake but did not decrease spontaneous behavior or inactive-lever responding [9]. Decreases in oxycodone self-administration induced by lorcaserin were blocked by a selective 5-HT2C antagonist, SB-242084, indicating that the effects of lorcaserin were mediated via 5-HT2C. In rhesus macaques (n = 3, male), lorcaserin pretreatment (1 mg/kg, intramuscular (i.m.)) produced a flattening of the dose-response function for intravenous heroin reinforcement but did not affect food-maintained responding [10], providing further evidence that the effects of lorcaserin were not solely attributable to non-specific decreases in rates of operant responding. Similarly, lorcaserin (0.32–1 mg/kg, subcutaneous (s.c.)) dose-dependently decreased responding for intravenous remifentanil in another self-administration study in rhesus macaques monkeys (n = 6, male) [11]. Lorcaserin pretreatment has also been shown to decrease rates of reinstatement responding, including cue-induced reinstatement of oxycodone-maintained responding in rats [9], and non-contingent heroin primed reinstatement of heroin-maintained responding in monkeys [10]. Taken together, these studies demonstrate that lorcaserin reduces the reinforcing effects of opioids in self-administration and reinstatement paradigms, providing evidence that selective 5-HT2C agonists could have a therapeutic potential in the treatment of OUD.

In addition to preclinical self-administration studies, several preclinical studies have evaluated lorcaserin in other behavioral assays to explore its potential as a treatment for OUD. An investigation in male Kunming mice employed a model of behavioral sensitization in which heroin was administered twice a day for 3 days (development) and then suspended for 5 days (withdrawal) and subsequently challenged with heroin (expression), lorcaserin (0.5 m/kg, i.p.) suppressed heroin-induced increases in locomotion and decreases in im-mobility in all stages of the procedure. In complementary experiments using a mouse model of antagonist-precipitated withdrawal, lorcaserin (0.5 m/kg, i.p.) suppressed somatic signs of opioid withdrawal in heroin-dependent mice during a naloxone challenge [12]. In a subsequent study, lorcaserin (0.5 m/kg, i.p.) prevented the induction and expression, but not the development, of morphine-induced behavioral sensitization in male Kunming mice [13]. In withdrawal testing, pretreatment with lorcaserin (0.5 m/kg, i.p.) ameliorated naloxone-precipitated withdrawal in morphine-dependent mice, and SB-242084 (a selective antagonist for the 5HT2C receptor) prevented the lorcaserin-mediated suppression of both behavioral sensitization and precipitation of somatic withdrawal signs. Taken together, these studies indicate that lorcaserin suppresses behavioral sensitization and somatic signs of withdrawal in preclinical models of opioid dependence, suggesting that lorcaserin may be viable as a therapeutic for OUD.

In contrast to the preclinical studies supporting the therapeutic potential of lorcaserin for OUD, at least two studies employing opioid- vs. food-choice procedures demonstrated that acute lorcaserin pretreatment produces non-selective decreases in rates of both food and opioid self-administration. A preclinical study by Panlilio and colleagues (2017) (male Sprague-Dawley rats) showed that lorcaserin pretreatment (0.1, 0.3, 1.7, 3 mg/kg, i.p.) decreased both food and remifentanil self-administration, suggesting that lorcaserin produces a non-specific disruption of operant behavior [14]. Another preclinical study using rhesus macaques (n = 1 male, n = 5 female) found that continuous lorcaserin infusion (0.032–0.32 mg/kg/h, i.v.) failed to promote the reallocation of behavior from heroin reinforcement to food reinforcement, and significantly increased heroin choice at the highest dose tested [15•]. Townsend et al. (2020) hypothesized that the differences in outcomes may be accounted for by disparities in the reported endpoints, that is, rates of self-administration in the study by Kohut and Bergman (2018) and behavioral allocation in their own. Overall, preclinical opioid- vs. food-choice procedures provided evidence against the therapeutic potential of lorcaserin as a candidate OUD medication.

Clinical

Several clinical trials (NCT03143543, NCT03169816, NCT03143855) have evaluated the therapeutic potential of lorcaserin for OUD; however, the results of those studies have not been published at the time of writing. Consistent with the findings of Panlilio et al. (2017) and Townsend et al. (2020), a clinical trial (n = 11 males, 1 female) that evaluated lorcaserin’s ability to alter the reinforcing and subjective effects of oxycodone in a 7-week inpatient trial of patients with moderate-to-severe OUD found that lorcaserin at a dose of 10 mg oral (p.o.), twice a day (BID), relative to placebo, failed to selectively decrease oral oxycodone- vs. money-choice and showed a trend to increase heroin-wanting (i.e., heroin craving; Brandt et al. 2020) [16•]. Moreover, results from NCT03143855 showed that BID 10 mg lorcaserin (n = 11) relative to placebo (n = 6) failed to meaningfully decrease subjective responses to oxycodone as measured with a 100-mm visual analog scale. Additionally, 10 mg lorcaserin (n = 12) BID was not effective vs. placebo (n = 7) in increasing the proportion of individuals who were successfully inducted to receive an extended-release naltrexone injection as indicated by results from NCT03169816. Similar to the findings by Brandt and colleagues (2020), two clinical trials on the effects of lorcaserin on the subjective and reinforcing effects of cocaine (NCT02680288, NCT02537873) found that a single dose of 10 mg p.o. lorcaserin failed to selectively decrease intravenous cocaine- vs. money-choice and enhanced some of the positive subjective effects of cocaine—findings that were supported by repeated lorcaserin administration using a preclinical intravenous cocaine- vs. food-choice procedure [17, 18]. One limitation of these clinical studies was that participants were administered a single dose of lorcaserin and therefore it cannot be ruled out that tests incorporating a wider dose range and/or dosing duration of lorcaserin would produce a dose-dependent reallocation of behavior to a non-drug reinforcer. Nevertheless, published clinical data that utilized subjective measures and drug- vs. money-choice procedures to evaluate the therapeutic potential of lorcaserin for OUD suggested that lorcaserin was not effective; these findings are consistent with analogous drug- vs. food-choice procedures in preclinical studies.

Summary

Lorcaserin showed promise as a potential therapeutic for OUD in preclinical studies employing intravenous self-administration and precipitated withdrawal procedures, but ultimately did not show benefit in preclinical studies employing opioid- vs. food-choice procedures or in clinical studies employing opioid- vs. money-choice procedures. The translational agreement of findings from preclinical and clinical choice procedures found among these studies highlights the utility of such endpoints to efficiently evaluate candidate OUD medications. In light of this preclinical-to-clinical concordance, future preclinical evaluations of candidate OUD medications should include opioid- vs. food-choice procedures. Also, to better model clinical usage of OUD medications, preclinical evaluations should also determine whether tolerance to putative therapeutic effects occurs following chronic treatment and whether therapeutic effects are present in both opioid-dependent and non-opioid-dependent subjects [2, 19].

Lastly, special caution should be observed when questions arise about the relative safety of a given drug. Concerns about the carcinogenic potential of lorcaserin were expressed by the FDA as early as 2010 [20]. Lorcaserin (10, 30, and 100 mg/kg) produced neoplastic changes in both male (fibroadenoma, schwannoma, squamous cell carcinoma, adenocarcinoma, fibroadenoma, and astrocytoma) and female rats (mammary adenocarcinoma and fibroadenoma) [21]. In 2013, Arena Pharmaceuticals, the former manufacturer of Belviq ^®, notified the Committee for Medicinal Products for Human Use (CHMP) of the European Medicines Agency (EMA) that it was voluntarily withdrawing its marketing authorization application for Belviq ^® in Europe following concerns by the committee about an increased incidence of tumors reported in preclinical studies [22]. While the carcinogenic potential of lorcaserin has little impact on the viability of 5HT2C receptor agonists as potential OUD treatments, virtually every study evaluating lorcaserin as a candidate medication for OUD was performed after the EMA published the withdrawal notice for Belviq ^®.

Memantine

Preclinical

Memantine (Namenda^®) is a non-competitive NMDA receptor antagonist that was originally developed for diabetes treatment and in 2003 was approved for the treatment of Alzheimer’s disease [23, 24]. Memantine also acts as a non-competitive antagonist at the serotonin 5-HT3 receptor and Alpha-7 nicotinic acetylcholine receptor and is an agonist at the dopamine D2 receptor. In general, memantine is neuroprotective and improves cognitive function [25]; however, the combination of receptor activity suggests that memantine could be a promising therapeutic for several aspects of OUD treatment, with the potential to improve withdrawal, craving, and relapse outcomes [26, 27].

Several preclinical studies have evaluated the potential therapeutic effect of memantine in rodent models of the rewarding effects of opioids (e.g., conditioned place preference), relapse (e.g., reinstatement), and opioid withdrawal (e.g., naloxone-precipitated withdrawal). A study on sensitization to the rewarding effects of morphine found that co-administration of memantine (10 or 20 mg/kg) and morphine, compared to placebo and morphine, blocked biased morphine-conditioned place preference in OF1 mice [28]. Another unbiased conditioned place preference study that compared different doses of both memantine and the NMDA antagonist MK-801 demonstrated that only high doses of memantine (20 and 40 mg/kg) prevented the acquisition of morphine place preference in OF1 mice [29]. Similarly, pretreatment with memantine suppressed morphine self-administration compared with placebo in DBA/2 mice, and this suppression was stronger with pretreatment of memantine compared with two other non-competitive NMDA receptor antagonists [30]. There is evidence that attenuated conditioned place preference is associated with decreased neuroinflammation. For example, a study by Chen and colleagues (2012) demonstrated that 0.2–1 mg/kg memantine compared with placebo attenuated unbiased conditioned place preference for morphine in morphine-dependent Sprague-Dawley rats and that these behavioral results corresponded to decreased serum cytokines (interleukin-6 [IL-6], interleukin-1β [IL-1β]), decreased medial prefrontal cortex (mPFC) and nucleus accumbens (NAc) mRNA expression of IL-6 and IL-1β, and increased brain-derived neurotrophic factor (BDNF) serum levels and mRNA in the mPFC and NAc [31]. Taken together, these studies suggest that memantine reduces the rewarding effects of opioids and could have utility in reducing craving or motivation to take opioids in humans.

There is also preclinical evidence that memantine could improve outcomes for persons undergoing opioid withdrawal. Memantine (5 and 10 mg/kg) dose-dependently attenuated startle response in Sprague-Dawley rats undergoing naloxone-precipitated withdrawal, and to a lesser degree, withdrawal-induced hyperalgesia [32]. Also, memantine (5 and 10 mg/kg) and another non-competitive NMDA receptor antagonist, MK-801, inhibited biased conditioned place aversion compared to placebo in morphine-dependent OF1 mice, but only memantine reduced physical symptoms of naloxone-precipitated withdrawal in these experiments [33]. Both of these studies suggest that memantine might be useful during opioid tapering.

Further evidence of opioid withdrawal suppression, inhibition of the rewarding effects of opioids, and importantly, reinstatement of opioid use following a brief period of abstinence was reported in a series of studies by Popik and colleagues (1997). One study utilized a biased conditioned placed preference paradigm using either morphine or food, in which memantine (7.5 mg/kg) compared with placebo inhibited the acquisition and expression of morphine but not food-induced preference in Wistar rats [34]. In a similar paradigm examining morphine, food, or sexual encounters with female Swiss mice, memantine (7.5 mg/kg) again did not affect food-conditioned mice but did inhibit biased place-preferences for morphine and sexual encounters [35]. Memantine (7.5 mg/kg) compared with placebo also attenuated naloxone-conditioned place aversion [34], further suggesting that memantine might have utility in the treatment of opioid withdrawal. Likewise, this group also demonstrated that memantine (7.5–30 mg/kg) attenuated naloxone-induced withdrawal behaviors in a dose-dependent manner in Swiss mice and that this effect was reversed by administration of the NMDA agonist, glycine [36]. Finally, this group demonstrated that reinstatement of biased morphine-conditioned place preference in C57BL mice could be blocked by memantine (7.5 mg/kg), but not the benzodiazepine, chlordiazepoxide, or the serotonergic psychedelic, lysergic acid diethylamide-25 [37]. A study by a separate group extended this finding by comparing NMDA antagonists, including memantine, with dopamine receptor antagonists and dopamine release inhibitors; only NMDA antagonists blocked morphine-induced reinstatement of biased place preference in OF1 mice [38].

Clinical

There have been several clinical studies that sought to examine the potential benefit of memantine on opioid withdrawal, relapse, and craving outcomes in OUD patients. A trial comparing memantine (30 mg), amitriptyline (75 mg), and placebo for the treatment of protracted withdrawal over 3 weeks found that both memantine and amitriptyline reduced heroin craving, anxiety symptoms, and anhedonia relative to placebo (N = 67: 60 Male, 7 female). Participants taking memantine in this study had significantly lower treatment attrition compared with both amitriptyline and placebo, and lower mean number of side effects compared with amitriptyline [39•]. A smaller human laboratory study of participants with OUD (N = 8) who were maintained on morphine in an inpatient unit examined whether 60 mg oral memantine vs. placebo would reduce withdrawal symptoms in response to a naloxone challenge (.4 mg i.m.). Pretreatment with memantine vs. placebo decreased naloxone-precipitated withdrawal in heroin-dependent, participants (N = 8: 6 male, 2 female) [40]. Another human laboratory study recruited OUD patients (N = 8: 6 male, 2 female) to an 8-week inpatient study to examine whether memantine (30 or 60 mg) would reduce intra-nasal heroin self-administration (0, 12.5, and 50 mg) craving, and subjective effects among detoxified heroin users. Memantine failed to reduce heroin administration but did reduce drug craving in the presence of heroin as well as subjective reports of “high” and “good drug effect,” relative to placebo [41].

A clinical trial (N = 71 completers: 66 male, 5 female) of memantine (30 or 60 mg per day) and placebo as an adjunct therapy to oral naltrexone (100 mg on Mondays and Wednesdays, 150 mg on Fridays) found no significant effect of either active memantine dose (vs. placebo) on measures of treatment retention, heroin use, opioid withdrawal (although participants underwent 5 days of withdrawal before randomization), or craving [42]. A 12-week trial on persons utilizing extended-release naltrexone (XR-NTX; N = 55; 44 male, 11 female) compared 40 mg/day memantine with placebo for treatment retention, weekly opioid use, weekly craving, and weekly opioid withdrawal. Weekly withdrawal ratings were marginally lower in the memantine vs. placebo group in the first 3 weeks of the trial; however, treatment retention was significantly higher in the placebo group [43].

In methadone-maintained patients, a clinical trial (N = 128, 85% male) of low-dose memantine (5 mg daily) demonstrated that memantine (vs. placebo) reduced the required therapeutic dose of methadone, decreased systemic levels of tumor necrosis factor-alpha (TNF-alpha), and increased systemic levels of transforming growth factor (TGF) beta-1 [44]. Another trial in methadone-maintained patients (N = 81: 70 female, 11 male) that measured cognitive performance via the Wisconsin Card Sorting Test before and after 12 weeks of 5 mg memantine or placebo treatment reported that participants in the memantine group demonstrated improved cognitive performance and greater treatment retention [45].

A 13-week trial of young adults (N = 80: 53 male, 27 female) who were maintained on 4–16 mg sublingual buprenorphine/naloxone maintenance, with discontinuation in week 9 of the study, reported that 30 mg memantine was more effective in reducing illicit opioid use during maintenance and after discontinuation use compared with 15 mg memantine or placebo (15 mg memantine had the worst relapse outcomes) [46]. Participants receiving 30 mg memantine also had reduced craving and withdrawal after discontinuation of buprenorphine/naloxone (vs. 15 mg memantine and placebo). However, the study found no group differences in treatment retention.

Summary

Memantine is generally well-tolerated [47] and several clinical trials have reported some benefit of memantine as an adjunct medication for OUD treatment [48, 49]. Still, some trials, especially those recruiting OUD patients who use naltrexone for relapse prevention, have reported no benefit of adjunct memantine [42, 43]. Somewhat more consistent findings include the possible therapeutic benefit of memantine on opioid withdrawal and craving [41, 46], which are common motivators of continued opioid use in many persons with OUD. Despite the efforts into developing memantine for clinical use in OUD patients, it is not FDA-approved for OUD treatment and is largely unused as an off-label medication to treat opioid withdrawal or craving.

Lofexidine

Preclinical

Physical dependence is a normal neurophysiological response to the regular use of opioids. Cessation of opioid use results in a predictable sequelae of physiological and psychological symptoms including yawning, lacrimation, sweating, rhinorrhea, sneezing, irritability, anxiety, chills, muscular and abdominal pains, diarrhea, weakness, and insomnia [50, 51]. Among opioid-dependent individuals, withdrawal is a potent negative reinforcer for continued opioid use and a major challenge to the treatment of OUD. Gradual opioid tapering, also known as detoxification, can reduce the severity of withdrawal symptoms [52]. Opioid withdrawal can also be managed (to some degree) using non-opioid medications such as benzodiazepines and alpha-2-adrenergic agonists [53]. In 2018, the FDA approved the first non-opioid treatment for the management of opioid withdrawal, the central alpha-2-adrenergic receptor agonist lofexidine (Lucemyra^®) [54–56]. The preclinical and clinical data supporting the efficacy of lofexidine represents a successful approach in translational research for OUD medication development.

Levels of norepinephrine and its metabolites are significantly altered with chronic opioid use. Activation of mu-opioid receptors suppresses the release of norepinephrine (particularly in the locus coeruleus and pons). Thus, noradrenergic hyperactivity, when opioid use is discontinued, has been implicated in the sympathetic symptoms of opioid withdrawal [57]. It is generally believed that alpha-2-adrenergic agonists complete a feedback cycle leading to a decrease in sympathetic outflow [58, 59]. The potential of alpha-2-adrenergic agonists like, clonidine, dexmedetomidine, guanfacine, and lofexidine to aid in opioid withdrawal has been noted for over 40 years [60]. Rather than follow a traditional translational trajectory (preclinical research→ clinical trials), animal and human studies, along with off-label use, concurrently provided data on the effectiveness of this class of drugs [61].

Lofexidine began to be viewed as a potential treatment for opioid withdrawal based upon its structural similarity to clonidine, which had been used off-label for this purpose since the 1970s [62, 63]. Research in Charles River rats (N = 30) had also shown that during states of opioid withdrawal, clonidine could reduce noradrenergic hyperactivity in the locus coeruleus [64]. However, in vitro and in vivo animal research suggested that lofexidine had a similar therapeutic benefit to clonidine, but less of an impact on blood pressure, thereby improving the risk/benefit profile [65–67]. Early studies tested the ability of lofexidine (0.04–0.64 mg/kg) and clonidine (0.01–0.16 mg/kg) to reduce symptoms of withdrawal among male Long Evans rats following the discontinuation of chronic morphine infusion [68]. Signs of somatic opioid withdrawal (i.e., body shakes) were dose-dependently reduced by both drugs. This study also identified the ability of both drugs to attenuate the diarrheal effects of naloxone-precipitated withdrawal. The observance of these findings in the presence of the opioid receptor antagonist, naloxone, provided evidence of the non-opioid (i.e., non-narcotic) basis of these treatment effects [69].

Later preclinical studies by Li and colleagues (2000) would examine the molecular basis of lofexidine’s effects on the neural substrates of opioid withdrawal [70]. Lofexidine administration was found to lower c-fos mRNA and fos protein in the locus coeruleus of morphine-dependent male Wistar rats (N = 27). In rats, c-fos responses in these (and other) brain regions have been linked to the motivational, emotional, and memory-related processes of opioid withdrawal [71–73].

Clinical

Early data in healthy participants (N = 6) supported preclinical findings that lofexidine (300 μg) has less hypotensive effects when compared to clonidine (300 μg) [74, 75]. Given its similar pharmacology but superior safety profile, early clinical studies sought to compare the efficacy of lofexidine to a standard of opioid withdrawal treatment, clonidine [76^•]. Several clinical studies at the time demonstrated the effectiveness of lofexidine and/or its equivalence to clonidine among patients (N = 15: all male) withdrawing from methadone [77] and following abrupt discontinuation of chronic methadone or levo-alpha acetylmethadol ((LAAM), N = 30: all male) [78].

The findings from these early clinical studies were confirmed by several randomized, controlled, clinical trials. Lin and colleagues (1997) compared lofexidine (1.6 mg/day maximum) and clonidine (0.6 mg/day maximum) among persons with OUD (N = 80: 65 men, 15 women) completing inpatient detoxification [79]. Carnwath and Hardman (1998) compared similar fixed dosing regimens of lofexidine (0.8–1.6 mg/day) and clonidine (0.3–0.6 mg/day) among participants (N = 50: 70% male) detoxing from 40 mg or less of methadone or equivalent amounts of other opioids [80]. Finally, Khan et al. (1997) utilized a flexible-dose comparison of lofexidine (1.8 mg/day maximum) and clonidine (0.9 mg/day maximum) among persons with OUD undergoing detoxification (N = 28: 19 male, 9 female) [81]. All three trials found an equivalent therapeutic benefit of the two drugs on suppression of opioid withdrawal, but significantly less hypotensive effects with lofexidine treatment. A more recent study reported significantly greater suppression of withdrawal following a fixed titration dosing of lofexidine (1.2, 1.6, 1.2 mg), in comparison to clonidine (0.9, 1.2, 0.9), among detoxing heroin-dependent participants (N = 40: all male) while again showing less hypotensive effects [82]. Conflicting results have been reported when using a different experimental model of opioid withdrawal. Walsh and colleagues (2003) stabilized eight participants (6 male, 2 female) on oral methadone (30 mg/day) and assessed the effects of lofexidine (0.4–1.6) and clonidine (0.1 and 0.2) pretreatment on naloxone-precipitated withdrawal [83]. The investigators concluded that neither medication significantly reduced the subjective discomfort of opioid withdrawal. The aforementioned clinical trials comparing the efficacy of lofexidine to clonidine were excellently summarized in systematic reviews by Kuszmal and colleagues (2003) [84] and Strang and colleagues (1999) [85].

The authors found a single randomized controlled trial that compared the effectiveness of lofexidine to diazepam, a benzodiazepine commonly used as an adjunctive medication during opioid detoxification [86]. Opioid-dependent patients (N = 111: 103 male, 8 female) were randomized to receive a 10-day course of lofexidine (fixed dosing up to 2.2 mg/day) or diazepam (up to 15 mg/day). Study results demonstrated that lofexidine was equally as effective as diazepam in reducing opioid withdrawal and also increased treatment retention. Several trials also compared the effectiveness of lofexidine treatment to dose tapering of opioids. Bearn et al. (1996; N = 86: 80% male) (1998; N = 44: 85% male) and Howells et al. (2002; N = 68: all male) found no difference in retention rates for a 10-day opioid detoxification between lofexidine-treated (fixed dosing 0.6–2.0 mg/day) and methadone-tapered participants [87–89]. Lofexidine (2.4 mg/day) has also been shown to be similarly efficacious to a buprenorphine taper (N = 200: 157 male, 53 female) [90]. Refer to Gish et al. (2010) and Strang et al. (2010) for reviews of these findings [58, 85].

When compared to placebo, the effects of lofexidine on withdrawal are quite robust. A multi-site trial of persons with OUD (N = 68: 59 male, 9 female) undergoing medically supervised opioid detoxification found that lofexidine (3.2 mg/day) outperformed placebo on nearly all measures of opioid withdrawal [91]. Similarly, among patients (N = 264: 200 male, 64 female) underdoing inpatient opioid withdrawal, lofexidine (3.2 mg/day) significantly reduced self-reported and clinician-observed opioid withdrawal, and improved retention rates [92]. These results were replicated in two other clinical trials (lofexidine 2.88 mg/day, N = 602: 71% male; 2.16 mg/day, N = 264: ~ 75% male) [93, 94].

Summary

In 1992, the UK-based Britannia Pharmaceuticals received approval of lofexidine (Britlofex^®) for the management of opioid withdrawal. Britlofex^® has since been used safely in an estimated 290,000 treatments [95]. Despite the early evidence of its efficaciousness, the pathway to FDA approval has not been straightforward in the USA. A lack of insurance coverage for substance use disorder treatment limited lofexidine’s marketability. It was not until the passage of the Patient Protection and Affordable Care Act (PPACA) of 2010 that the USA mandated the coverage of drug treatment services. Much of the funding responsible for the 2000’s reinvigoration of interest in lofexidine came from the National Institute on Drug Abuse (NIDA). Thus, the story of lofexidine reminds us of the importance of pharmacoeconomics in drug development.

Discussion

Thoughts and Suggestions Concerning Medications Development for OUD

Translational science attempts to harness knowledge from basic scientific research into clinical research to create new medications, devices, and diagnostics. Candidate drugs only have an estimated 0.1% chance of becoming FDA-approved medications, as 80–90% of drugs never make it to human trials, and 95% of drugs tested in human trials fail [96, 97]. Animal models are of tremendous value to the process of medication development. However, across drug discovery, the majority of translational failures occur in the failure of clinical studies to replicate the positive findings of preclinical research, or due to unexpected side effects, or poor tolerability [98, 99].

Substance use disorders present a unique problem for medication development. Unlike some conditions, the etiology of SUDs is complex, varies across individuals, and manifests a variable disease course and response to treatment [100, 101]. Given that SUDs can be the result of multiple etiologies, preclinical paradigms most often attempt to model the salient features of SUDs, i.e., testing whether a candidate medication attenuates withdrawal symptoms, or reduces drug-seeking behavior. However, there are many preclinical models with differing levels of construct and predictive validity. The design of human trials may also hamper their ability to replicate the promising findings from animal studies. Finally, there are obstacles to translational medicine that are independent of investigators. Partially based on our review of the medications in this manuscript, this section will briefly review the authors’ opinions as to why the “bench-to-bedside” process often fails medications development for OUD.

Not all preclinical models are created equally. A major cause of failure in clinical trials of OUD pharmacotherapy appears to be a lack of effectiveness that was not predicted in preclinical studies. Various preclinical paradigms have been developed to model the conditioned and unconditioned effects of opioids. These experimental models are meant to reflect the biological, psychological, and social processes responsible for the initiation, and maintenance of drug use. As noted previously, a challenge to preclinical researchers is the complex etiology of SUDs. Preclinical OUD assays often model an individual aspect of a complex disorder (e.g., drug self-administration, extinction, or reinstatement). In order to encapsulate the phenotypic dimensions of addiction, preclinical models should be varied and malleable to address the numerous therapeutic goals needed to manage OUD in humans.

Preclinical OUD models vary in their predictive validity (forecast accuracy), face validity (phenomenological similarity), and construct validity (theoretical rationale). For example, drug self-administration has a high face and construct validity as an assessment of the reinforcing effects of opioids [102]. The same cannot be said for many preclinical addiction outcome measures. Not that these assessments are without merit, they are simply modeling less well-understood aspects of the addictive process, or their clinical models are less analogous [103]. For example, various preclinical models of craving exist, yet, craving is a concept that is challenging to define and the utility of craving in clinical trials has been highly variable [104–106].

It is not surprising that preclinical self-administration models have significant predictive validity in screening candidate medications for the treatment of OUD. Our review of the lorcaserin data suggests that self-administration designs that incorporate a choice procedure between alternative reinforcers (i.e., drug vs. food (preclinical) or drug vs. money (clinical)) may have particular high concordance across the translational spectrum. Thus, comparison of food reward in preclinical studies with monetary reward in clinical studies should be considered in future translational research.

Though self-administration is considered one of the more valid models, there may be ways to improve translational success. Preclinical self-administration is typically considerably shorter (weeks–months) than the duration of human drug use that they model (years–decades). Preclinical self-administration studies may thus be reflective of early drug use and not represent the pathology associated with chronic OUD, which is typically the stage at which most individuals seek treatment. Preclinical and clinical studies of medications that employ self-administration may, therefore, be focused on different stages of addiction. As such, extended-access preclinical models may better mimic the clinical condition.

Human behavioral pharmacology can be a bridge between preclinical studies and clinical treatment trials. As such, human behavioral pharmacology research should mimic preclinical procedures as much as possible (and vice versa) and be correlated to clinical factors including drug use, relapse, and craving in the natural environment. Theorists have also introduced the concept of homological validity, which emphasizes the importance of the adequacy and relatedness of the two species [107]. In this principle, promising data from primates would serve as a better foundation for clinical study than data from zebrafish. Translational research should also not be unidirectional (e.g., rodents→ non-human primates→humans). Given the complex nature of OUD, bidirectional learning (e.g., back-translation) may help clinical researchers design studies that better replicate promising preclinical findings. Meanwhile, better-characterization of the phenotypes and endophenotypes of OUD by clinical scientists will help preclinical researchers identify better-validated preclinical targets and assays [108].

Clinical researchers can also improve the quality of hypotheses before testing them. Consistent promising findings across preclinical OUD models appear to forecast greater chances of clinical success. As such, clinical studies should be based on a significant and varied body of preclinical work (e.g., positive findings using different preclinical outcomes modeling the same feature of addiction). This will also lower the risk that studies are based on irreproducible or false findings, a growing concern among academia and industry [109, 110].

In our review of the literature, we noted numerous examples of clinical trials that were underpowered. Preclinical studies may provide poor effect size estimates for clinical researchers looking to replicate their work. Several clinical trials also lacked precision and had variable definitions of outcomes including the timeframe that outcomes were measured and instruments used to measure outcomes. There has been a wide breadth of pharmacological and other interventional approaches in OUD as summarized in Table 1. It is notable that all of the listed approaches, derived from a search on clinicaltrials.gov, have demonstrated efficacy in various preclinical models of OUD. Finally, an additional issue discovered across clinical studies was a lack of female representation. On average, women made up ~ 18% of participants in clinical trials of the three medications discussed (range: 0–86%); casting doubt on our ability to consider sex as a biological factor in the interpretation of medication effects.

Table 1.

Pending, ongoing and completed clinical trials of various therapies for opioid use disorder

Approach	Class	Hypothesized mechanism of action	# of trials	GCT(s)
Methods
Transcranial electrical stimulation		DA circuitry modulation	3	GCT0103975192
				GCT0103907644
				GCT0102033746
Deep brain stimulation		DA circuitry modulation	6	GCT0101274988
				GCT0103424616
				GCT0103952455
				GCT0102440152
				GCT0103950492
				GCT0104354077
Transcranial magnetic stimulation		DA circuitry modulation	6	GCT0103549065
				GCT0103538444
				GCT0103821337
				GCT0103653169
				GCT0104157062
				GCT0104181515
Low-intensity focal ultrasound		DA circuitry modulation	1	GCT0104197921
Acupuncture		Endogenous opioid modulation	2	GCT0101512433
				GCT0103012373
Pharmacological approaches
Psilocybin	Hallucinogen	5-HT2A agonism; synaptic rewiring	1	GCT0104161066
Atomoxetine	ADHD medication	NE-reuptake inhibition	1	GCT0101863251
Mirtazapine	Antidepressant	Alpha2/5-HT2 antagonist	1	GCT0102541526
Cannabidiol	Ion channels	Multiple targets; ultimately DA modulation	7	GCT0101311778
				GCT0102539823
				GCT0101605539
				GCT0104238754
				GCT0104192370
				GCT0103787628
				GCT0102559167
Opium tincture	Opioid analgesic	Mu agonism	1	GCT0102502175
Suvorexant	Hypnotic	Orexin antagonist	2	GCT0103789214
				GCT0104262193
Oxytocin	Hormone	DA modulation	2	GCT0104051619
				GCT0102028533
Liraglutide	Antidiabetic	GLP-1 agonist	1	GCT0104199728
Facial skin surface lipid pheromone	Pheromone	Unknown	1	GCT0103394911
Pioglitazone	Antidiabetic	PPAR-gamma agonist	1	GCT0101395797
ANS-6636	ALDH-2 inhibitor	DA modulation	2	GCT0103831971
				GCT0104169360
Risperidone/ziprasidone	Mood stabilizer	Decrease DA signaling	1	GCT0100000342
TRV-734	Biased mu-opioid analgesic	Mu agonism	1	GCT0104316559
Tramadol	Opioid analgesic	Mu agonism	1	GCT0100142896
Galantamine	AChe inhibitor	Potentiation of cognitive behavior therapy	1	GCT0103547622
Dynorphin 1–13	Opioid peptide	Mu agonism	1	GCT0100000244
Pregabalin	Analgesic/antiepileptic	GABA modulation	2	GCT0103017430
				GCT0104218240
NT-814	NK-1 antagonist	DA modulation	1	GCT0102692157
Aprepitant	NK-1 antagonist	DA Modulation	1	GCT0100726960

The search on clinicaltrials.gov excluded treatment approaches involving already-approved opioid agonists and antagonists, the drugs discussed in the present review, and cognitive therapies, which cannot be modeled preclinically

U unknown. The link to the full table, including study details, can be found in the supplementary material

Though human studies are very costly, eliminating medication trials that lack rigor may offset the cost of investment [111]. It has been suggested that streamlining and standardizing methodologies across the translational spectrum may be one way to achieve this goal. Successful translation from animal to human studies of medications development requires better communication and collaboration between basic and clinical scientists. However, the methodological intricacies of SUD medication trials make grant proposals that integrate human and animal models extremely complex, which may necessitate special skill and care during scientific review.

For translational research to operate more effectively, there needs to be clear direction and communication between researchers and regulators. For a potential medication to receive FDA approval, it must undergo various phases of human testing (phases 1–3). Late-stage clinical trials are often forced to adopt specific outcome measures, as required by regulators. Thus, these pivotal trials diverge from the preclinical and early-stage clinical trials on which they were based. Another concern, more specific to SUDs, has been the use of drug abstinence as a determining outcome in medication trials. It was not until 2018 that FDA regulations recognized a harm reduction approach concerning the efficacy of medications to treat OUD [112]. This change in policy means that numerous medications may need to be re-evaluated, and opens a pathway forward for drugs with specific efficacy for adjunctive indications such as craving (like memantine).

As noted above with lofexidine, the economics of a drug can also lead to promising drugs stalling along the translational pathway. It is estimated that new drug development can cost up to $2.5 billion [113, 114]. Therefore, significant motivation and investment from the government and the biopharmaceutical industry are needed to distribute the financial burden. Drug repurposing is one way to reduce development costs [115]. Testing pharmacotherapies already approved for other indications is common in OUD medication development (Table 1). However, now that more pharmacotherapeutic OUD indications are attainable, this may prove to be an even more attractive strategy.

Finally, improved translational success may necessitate altering our expectations of what we can achieve from a single medication. It would be surprising if a single compound would have efficacy against all aspects of drug addiction, given its multimodal nature. Indeed, the distinct neural circuits reviewed by Volkow et al. (2016) [116] for acute drug intoxication, acute withdrawal, and maintenance of longer-term drug-seeking behavior as a matter, of course, consist of different drug targets and are therefore susceptible to different mechanistic interventions.

Conclusions

In conclusion, this review highlights only a fraction of the immense efforts of preclinical and clinical scientists to identify and test novel treatments for opioid use disorder. The objective of translational OUD research is to make sure that medications that advance into humans have the highest possible chance of success in terms of both safety and efficacy. However, increases in spending on biomedical research have not resulted in proportional increases in new pharmacotherapies. The problem of non-translatable research is not the fault of a specific group (e.g., preclinical or clinical researchers). A combination of better-validated preclinical targets, improved design of proof-of-principle human studies, and more germane outcome measures in pivotal clinical trials should reduce risks of medication attrition.