Enduring Effects of Adolescent Drug Exposure

While adolescent substance use may contribute to the risk for psychosis (Di Forti et al., 2009), drug use during this vulnerable developmental period also enhances vulnerability for continued substance use. Some of the commonly used substances (e.g., tobacco, alcohol and cannabis) in patients with schizophrenia are associated with increased risk for future substance use if use is initiated in adolescence (Chadwick et al., 2013; Kandel and Kandel, 2014; Levine et al., 2011). While the neurodevelopmental consequences of drug exposure during adolescence are not fully understood, previous epidemiological and pre-clinical investigations provide valuable clues regarding the long-term changes in addiction vulnerability arising from this exposure.

Tobacco use usually begins during adolescence, and this exposure is associated with increased risk for the development of cocaine, marijuana, heroin and alcohol use in adulthood (Grant et al., 2004; Strong et al., 2016). The causal direction of these associations is not entirely known, primarily due to a potential confound that needs to be considered in the context of the “gateway drug” hypothesis: that the development of another drug use after early exposure to a drug could reflect a common or shared liability for drug use and that this liability, rather than the exposure to the first drug, could increase the risk of progression to the use of another drug. The causal direction is further complicated since animal studies of the long-term effects of adolescent nicotine exposure on future drug use have produced mixed results (Alajaji et al., 2016; Pomfrey et al., 2015; Spear, 2016). However, exposure to nicotine during adolescence may produce long-term epigenetic changes that may increase susceptibility to the initiation and continued use of tobacco and other substances (Kandel and Kandel, 2014).

Alcohol use also begins in adolescence (Swendsen et al., 2012), and this early exposure predisposes users to future risk for an alcohol use disorder as well as other cognitive and reward-related dysfunctions (Nguyen-Louie et al., 2015). While the interaction between adolescent alcohol exposure and risk for schizophrenia has not been explicitly studied in patients, a recent study suggested that alcohol use in adolescence predicted future co-occurring mental health and poly-substance use disorders (Salom et al., 2015). A substantial body of preclinical literature has also suggested that alcohol exposure in adolescence promotes suboptimal decision-making, amplifies the incentive value of reward-predictive cues and produces long-lasting changes in alcohol’s activation of brain reward circuitry (Boutros et al., 2015; Clark et al., 2012; Liu and Crews, 2015; McClory and Spear, 2014; Spoelder et al., 2015; Vetreno and Crews, 2015).

Cannabis use during adolescence increases the risk for life-time cannabis dependence from 9% to 16%, and even higher to 25–50% among daily adolescent users (Volkow et al., 2014). While it is debated whether cannabis serves as a “gateway” drug, associations between adolescent cannabis use and use of illicit and novel psychoactive substances have been reported (Agrawal et al., 2004a, b), and longitudinal studies have suggested that regular or heavy cannabis use is associated with an increased risk of using other illicit drugs (Khan et al., 2013; Lynskey et al.,2003). Moreover, these findings have been supported by preclinical studies suggesting increased vulnerability for future drug use in rats exposed to cannabinoids during adolescence (e.g., increased intake of heroin, morphine, cocaine and WIN-55212 [a synthetic cannabinoid agonist] (Chadwick et al., 2013; Dow-Edwards and Izenwasser, 2012; Ellgren et al., 2007; Rodriguez-Arias et al., 2016)). While the enduring effects of adolescent drug exposure on future risk of co-occurring schizophrenia and substance use disorder is important, it needs to be considered in the context of genetic predisposition or susceptibility that might be related to the risk for schizophrenia.

Genetic underpinnings of schizophrenia and co-occurring substance use disorder

Based on epidemiological data and family studies, it appears clear that genetic factors significantly influence susceptibility to schizophrenia. The concordance rate for schizophrenia in monozygotic twins is 40–50% (Cariaga-Martinez et al., 2016; Schwab and Wildenauer, 2013). Genetic factors are also thought to play a role in the susceptibility to develop schizophrenia and a co-occurring substance use disorder. Indeed, polygenic risk scores for schizophrenia are also associated with cannabis use, cocaine use, nicotine use, and severe alcohol use (Carey et al., 2016).

It appears that gene x environment interactions generate risk of schizophrenia and co-occurring substance use disorders. Three distinct genes, encoding brain-derived neurotrophic factor (BDNF), catechol-O-methyltransferase (COMT), and protein kinase B (AKT), have garnered the most attention for their relationship with both schizophrenia and substance use. While polymorphisms of BDNF (rs6265 and rs103411), involved in synaptic plasticity and the activity of dopamine (Guillin et al., 2001; Hartmann et al., 2001), are not associated with alcohol dependence alone, they are strongly associated with schizophrenia and co-occurring alcohol dependence, suggesting that these BDNF variants may be important in the co-occurrence of these disease states (Cheah et al., 2014).

A particular allele of COMT, which influences catecholamine metabolism, has been implicated in schizophrenia (see Apud and Weinberger, 2007 for review). Specifically, the COMT Val/Val allele, associated with reduced dopamine function in the prefrontal cortex (PFC), results in increased risk for developing endophenotypes related to schizophrenia, but not necessarily the development of the disease (Ira et al., 2013). Environmental factors, however, can interact with COMT genotypes to impact the risk of psychosis development. Caspi and colleagues demonstrated a significant interaction between the COMT Val/Val allele and adolescent (but not adult) cannabis use to predict adult psychosis (2005), suggesting that two risk factors, a particular COMT polymorphism and adolescent cannabis use, together can increase the risk of psychosis. This finding, however, has not been replicated by subsequent investigations (Kantrowitz et al., 2009). Moreover, since BDNF and COMT are implicated in a number of genome wide associations studies of other psychiatric disorders, the associations between these genes and schizophrenia and co-occurring substance use disorder could arise as a result of pleiotropy (an effect of gene on multiple traits resulting in an effect on the outcome, but not via the behavior/trait displaying a significant association), and, thus, must be interpreted with caution.

AKT, a serine/threonine protein kinase involved in protein synthesis, synaptic plasticity and cell proliferation, (Scheid and Woodgett, 2001), is reduced in post-mortem brains of patients with schizophrenia (Emamian et al., 2004; Kalkman, 2006; Swiech et al., 2008). Interestingly, one study found an interaction between AKT polymorphisms and cannabis use in the risk of developing psychosis (Di Forti et al., 2012). Specifically, AKT C/C (but not T/T) allele carriers were 7 times more likely to develop psychosis if they used cannabis on a daily basis. Moreover, van Winkel and colleagues showed that patients with schizophrenia who used cannabis regularly had poorer cognitive performance if they carried the AKT C/C allele when compared to T/T allele carriers (van Winkel et al., 2011). Like BDNF and COMT polymorphisms, it may be that the AKT gene confers risk for schizophrenia development when environmental risk factors, specifically cannabis use, are also present. However, such associations, requiring the presence of genetic risk and developmental substance use, do not rule out the possibility of a shared susceptibility or reverse causality, and require further investigation to assess the causal directions in complex disorders such as schizophrenia.

An interesting finding from the Psychiatric Genomics Consortium 2 genome wide association study showed that a variant in the nicotinic acetylcholine receptor CHRNA5-A3-B4 gene cluster (previously associated with heaviness of smoking in smokers in the Tobacco Genetics Consortium study (2010)) reached genome-wide significance (Schizophrenia Working Group of the Psychiatric Genomics, 2014), suggesting shared genetic architecture between schizophrenia and cigarette smoking. This is further corroborated by a study in this special issue by Hartz and colleagues showing that a statistically significant genetic correlation exists between schizophrenia and various smoking phenotypes (e.g., cigarettes per day, nicotine dependence) (Hartz et al., 2017). Moreover, a recent pre-clinical study suggests the contribution of genetic variation within CHRNA5 may mediate both schizophrenia and nicotine use through a shared hypofrontality phenotype (Koukouli et al., 2017), making it important to study the circuit-based correlates of these co-occurring disorders.

Recent studies have also used a mendelian randomization approach to begin to assess the causal direction of the effects of substance use on schizophrenia. While the impact of initiation of cannabis and cigarette smoking on the risk for schizophrenia has been suggested by this technique (Gage et al., 2016b; Gage and Munafo, 2015a, b; Vaucher et al., 2017), a recently published finding also suggests that there is stronger evidence supporting the hypothesis that schizophrenia genetic risk predicts cannabis initiation (Gage et al., 2016b). This is consistent with the increased risk for substance use in non-psychotic relatives of patients with schizophrenia (Smith et al., 2008; Stone et al., 2001) and suggests that even though the substance use arises prior to the onset of psychotic symptoms, there may be pre-existing vulnerabilities to initiating substance use associated with the risk for schizophrenia.

Brain Reward Circuit Dysfunction as a missing link?

One manifestation of the genetic susceptibility to schizophrenia may be dysfunction within brain circuits involved in reward and motivation (specifically the mesocorticolimbic dopamine circuitry) that may drive both the initiation and continued use of substances. Reward processing in healthy subjects is related to dopaminergic activity in the ventral striatum (Boehme et al., 2015; Deserno et al., 2015; Schlagenhauf et al., 2013), which may be impaired in patients. Thompson et al., (2013) found that patients with schizophrenia and a co-occurring substance use disorder have decreased striatal dopamine release; the authors suggested that this decreased release might be due to the substance use history in these patients. Importantly, the authors also suggested that in such patients, there might be a hypersensitive dopamine system (as observed via the changes in positive symptoms in response to amphetamine); we propose that this hypersensitivity may further contribute to their vulnerability toward substance use.

Abnormal reward processing has been studied in patients with schizophrenia using functional magnetic resonance imaging. In medication-naive patients, a reduction of activity has been described in the ventral striatum during reward anticipation, and with reward outcomes (Esslinger et al., 2012; Juckel et al., 2006; Nielsen et al., 2012; Schlagenhauf et al., 2009). Abnormal reward processing in patients is also known to correlate with symptoms of the disease, e.g., severity of positive symptoms has been correlated with reduced ventral striatal activity during reward anticipation (Esslinger et al., 2012; Nielsen et al., 2012; Simon et al., 2015), with reduced prefrontal cortical activity between favorable and unfavorable outcomes (Schlagenhauf et al., 2009), and with over-activation of the midbrain to neutral cues (Romaniuk et al., 2010). Moreover, negative symptoms have been correlated with reduced ventral striatal activity in reward anticipation (Juckel et al., 2006; Simon et al., 2010) and decreased response to reward in both putamen (Waltz et al., 2009) and ventral striatum (Gradin et al., 2013). Interestingly, a recent study also assessed neural responses to cigarette smoking cues in smokers with schizophrenia, and found that patients with schizophrenia showed greater activation of a region within the brain reward circuit (i.e., ventromedial prefrontal cortex) in response to smoking cues compared to smokers without a psychiatric comorbidity (Potvin et al., 2016), suggesting that drug rewards may produce amplified responses in these patients.

In addition to the studies noted above related to task based brain activity, a few recent studies have demonstrated reduced resting-state functional activity in patients with co-occurring nicotine addiction between insula and anterior cingulate cortex (Moran et al., 2013; Moran et al., 2012), and in patients with co-occurring cannabis use disorder and schizophrenia between nucleus accumbens and frontal cortical regions of the brain reward circuit (Fischer et al., 2014). The Fischer study also demonstrated that smoked cannabis or oral THC partially ameliorated the hypoconnectivity, suggesting that cannabis use may be an effort by patients to correct dysfunctional brain reward circuitry (Fischer et al., 2014), consistent with both the reward deficiency syndrome and primary addiction hypotheses.

Developmental dysfunction of the hippocampus and frontal cortex may form the mechanistic underpinnings of this brain reward circuit dysfunction, as proposed by Chambers et al (2001). The hippocampus not only plays a key role in the modulation of dopaminergic activity within the nucleus accumbens, but also mediates the integration of information in the nucleus accumbens. Although the nucleus accumbens receives diverse synaptic input (Finch, 1996), the hippocampus ultimately drives these neurons to respond (O’Donnell and Grace, 1995). The hippocampus has also been shown to induce neuroplasticity in the nucleus accumbens (Mulder et al., 1998), providing more evidence that abnormal hippocampal development and activity can have both long term and immediate effects in accumbal development and information processing. Developmental abnormalities in both the hippocampus and prefrontal cortex may lead to overall dopaminergic hypersensitivity of the mesolimbic system (Weinberger and Lipska, 1995), whereby reduced inhibitory control of the nucleus accumbens (Weinberger and Lipska, 1995) can lead to disinhibition of accumbal outputs to thalamocortical loops driving behavior associated with motivational states (Lavin and Grace, 1994). This disinhibition may result in impaired reward learning such as the persistent interpretation of rewards as novel or the inability to tune out irrelevant stimuli (Gray et al., 1992). These developmental abnormalities in the interplay between the hippocampus, prefrontal cortex and nucleus accumbens may produce motivationaldeficits consistent with long-term substance abuse (without the prior drug exposure), while facilitating the reinforcing effects of substances, consistent with the dopaminergic hypersensitivity observed in dual diagnosis patients (Thompson et al., 2013).

Suggestions that developmental dysfunction of the hippocampus leads to a reward circuit dysfunction gets further validation in a rodent model of preadolescent hippocampal lesioning – the neonatal ventral hippocampal lesion rat (NVHL rat), in which normal hippocampal signaling (particularly to the frontal cortex and the nucleus accumbens) is altered during development (Lipska and Weinberger, 2000; Tseng et al., 2009). NVHL rats not only exhibit a schizophrenia-like phenotype, but they also have a dysregulated reward circuit. Importantly, they also demonstrate increased consumption of (and increased behavioral sensitization to) substances commonly used by patients, e.g., cocaine (Chambers et al., 2013; Chambers et al., 2010; Chambers and Taylor, 2004), nicotine (Berg and Chambers, 2008; Berg et al., 2014), methamphetamine (Brady et al., 2008) and alcohol (Berg et al., 2011; Conroy et al., 2007; Jeanblanc et al., 2015). Data from research on this animal model suggest that brain reward circuit dysfunctions increase the vulnerability for substance use in patients with schizophrenia, which combined with potentially accentuated reinforcing properties of the substances, may make patients with schizophrenia especially vulnerable to initiating substance use. While the developmental time-course of these deficits are not known, the epidemiological evidence regarding higher rates of substance use in adolescents prior to the onset of schizophrenia (as well as in non-psychotic first degree relatives (Smith et al., 2008; Stone et al., 2001) suggests that these dysfunctions may predate the onset of psychosis.

Effects of antipsychotics on substance use and reward

An important related issue concerns the effects of antipsychotic medication on substance use in this population. Most antipsychotics do not lessen substance use in patients with schizophrenia, with one exception – clozapine has been shown in many (albeit still preliminary) studies to do so across a variety of substances (e.g., alcohol, cannabis, tobacco) (Brunette et al., 2011; Brunette et al., 2006; Brunette et al., 2008; Buckley et al., 1999; Drake et al., 2000; George et al., 1995; Lee et al., 1998; Wu et al., 2013; Zimmet et al., 2000). Related to this, Mesholam-Gately et al., (2014) have observed, in a population of substance using patients with schizophrenia that in comparison with other antipsychotics, clozapine strengthens the hedonic experience associated with olfactory stimuli. Moreover, Machielsen et al, (2014), assessing the effects of clozapine versus risperidone in patients with schizophrenia and cannabis use disorder, suggested that clozapine may be more efficacious at reducing cannabis use due to its ability to modulate attentional bias to drug cues and its actions on reducing activation within the brain reward circuit. Green et al. (1999) have hypothesized that clozapine’s unusual effects may relate to its weak dopamine D2 receptor blockade, coupled with its ability to potentiate the action of norepinephrine in the brain, which together may ameliorate the brain reward circuit dysfunction in these patients.

We would like to thank Emily Kirk for her assistance with manuscript editing.

ROLE OF FUNDING SOURCE

This work was supported in part by grants from the National Institute of Drug Abuse (AIG; DA032533 and DA034699), from the National Center for Advancing Translational Science (AIG; NCATS UL1TR001086) and by a Canadian Institute of Health Research Fellowship (JYK).