INTRODUCTION
Levels of daily cannabis use among 12th graders are nearing the highest recorded since 1991 (6.4%; Johnston etal., Reference Johnston, Miech, O’Malley, Bachman, Schulenberg and Patrick2020), and 42% of college students and young adults report cannabis use in the last year (Schulenberg etal., Reference Schulenberg, Johnston, O’Malley, Bachman, Miech and Patrick2019). This increase in the frequency of use may be due in part to the emergence of vaping devices, which are exposing teens to unprecedented high potency concentrations of ∆-9-tetrahydrocannabinol (THC, the primary psychoactive constituent of cannabis) in a more accessible fashion. Vaping devices are also crushing the decades of progress in tobacco and nicotine prevention efforts. Johnston etal. (Reference Johnston, Miech, O’Malley, Bachman, Schulenberg and Patrick2020) report that an alarming 35% of high school seniors had vaped nicotine in the last year.
As the brain undergoes dynamic developmental changes throughout adolescence and into the mid-20’s (Giedd etal., Reference Giedd, Raznahan, Alexander-Bloch, Schmitt, Gogtay and Rapoport2015; Gogtay etal., Reference Gogtay, Giedd, Lusk, Hayashi, Greenstein, Vaituzis and Thompson2004), the cognitive implications of high potency cannabis and nicotine use are concerning. THC binds to cannabinoid 1 receptors (CB1) (Schneider, Reference Schneider2008) and nicotine activates nicotinic acetylcholine receptors (nAChRs) (Yuan, Cross, Loughlin, & Leslie, Reference Yuan, Cross, Loughlin and Leslie2015) in the central nervous system. CB1 and nAChRs are both found in cortico-limbic brain regions and modulate processes related to neural circuit development, particularly dopamine pathways. Regular activation of these endogenous neural systems due to frequent use of cannabis and nicotine may alter neurodevelopmental trajectories and behavioral outcomes (Hurd etal., Reference Hurd, Manzoni, Pletnikov, Lee, Bhattacharyya and Melis2019; Yuan etal., Reference Yuan, Cross, Loughlin and Leslie2015). Yet most research aiming to delineate the unique effects of substances on neurodevelopment rely on participant self-report. Evidence suggests that up to 13% of adolescents reported use is discordant with urinalysis results (Akinci, Tarter, & Kirisci, Reference Akinci, Tarter and Kirisci2001), with recommendations to corroborate results with more objective measurements (Harris, Griffin, McCaffrey, & Morral, Reference Harris, Griffin, McCaffrey and Morral2008; Williams & Nowatzki, Reference Williams and Nowatzki2009).
Though cannabis use is associated with a wide range of cognitive deficits in adolescents and young adults (Lisdahl, Wright, Kirchner-Medina, Maple, & Shollenbarger, Reference Lisdahl, Wright, Kirchner-Medina, Maple and Shollenbarger2014), one of the most common findings is in relation to verbal memory (Blest-Hopley, Giampietro, & Bhattacharyya, Reference Blest-Hopley, Giampietro and Bhattacharyya2020). Acutely, THC has a direct impact on verbal learning and memory (Bhattacharyya etal., Reference Bhattacharyya, Morrison, Fusar-Poli, Martin-Santos, Borgwardt, Winton-Brown and McGuire2010; Morgan etal., Reference Morgan, Gardener, Schafer, Swan, Demarchi, Freeman and Curran2012; Morgan, Schafer, Freeman, & Curran, Reference Morgan, Schafer, Freeman and Curran2010). Preliminary longitudinal studies also suggested that cannabis onset is related to poorer verbal memory performance over time (Hanson, Medina, Padula, Tapert, & Brown, Reference Hanson, Medina, Padula, Tapert and Brown2011; Jacobus etal., Reference Jacobus, Squeglia, Infante, Castro, Brumback, Meruelo and Tapert2015; Nguyen-Louie etal., Reference Nguyen-Louie, Castro, Matt, Squeglia, Brumback and Tapert2015). Yet deficits may not be permanent; some studies find improved memory in those who remain abstinent from cannabis for up to a month (Hanson etal., Reference Hanson, Winward, Schweinsburg, Medina, Brown and Tapert2010; Schuster etal., Reference Schuster, Gilman, Schoenfeld, Evenden, Hareli, Ulysse and Evins2018), while others do not (Wallace, Wade, & Lisdahl, In Press). A meta-analysis found cannabis to have a small effect on learning and delayed memory performance (d = −.33 and d = −.26, respectively), while no difference in any other cognitive domain was detected following short term abstinence (Scott etal., Reference Scott, Slomiak, Jones, Rosen, Moore and Gur2018). However, methodological limitations in cannabis research (e.g., self-report bias, varying consumption pattern and product types) may be contributing to variability in study results and therefore, the inclusion of more objective markers of cannabis exposure may help disentangle discrepancies in the research literature (Huestis, Reference Huestis2007; Smith etal., Reference Smith, Alden, Herrold, Roberts, Stern, Jones and Breiter2018).
Measured cannabinoid concentrations are influenced by type of product (flower vs. concentrate), product potency (Fabritius etal., Reference Fabritius, Chtioui, Battistella, Annoni, Dao, Favrat and Giroud2013; Greene, Wiley, Yu, Clowers, & Craft, Reference Greene, Wiley, Yu, Clowers and Craft2018), drug administration route (smoked, vaporized, eaten, dabbed) (Newmeyer etal., Reference Newmeyer, Swortwood, Taylor, Abulseoud, Woodward and Huestis2017), frequency of use, and individual genetics (Hryhorowicz, Walczak, Zakerska-Banaszak, Slomski, & Skrzypczak-Zielinska, Reference Hryhorowicz, Walczak, Zakerska-Banaszak, Slomski and Skrzypczak-Zielinska2018; Stout & Cimino, Reference Stout and Cimino2014). Thus, differences in cannabis products used and consumption rate may influence cognitive outcomes due to changes in THC bioavailability and pharmacokinetics (Sharma, Murthy, & Bharath, Reference Sharma, Murthy and Bharath2012; Spindle etal., Reference Spindle, Cone, Schlienz, Mitchell, Bigelow, Flegel and Vandrey2018, Reference Spindle, Cone, Schlienz, Mitchell, Bigelow, Flegel and Vandrey2019). To illustrate, a within-subject design found significant variability of cannabinoid analyte concentrations in healthy individuals who consumed the same three edible products, each a week apart (Schlienz etal., Reference Schlienz, Cone, Herrmann, Lembeck, Mitchell, Bigelow and Vandrey2018). Utilizing biosamples to measure the THC metabolite (11-nor-9-carboxy-THC or THCCOOH) may bypass this variability to predict more reliable cognitive results. Further, it was suggested (Karschner etal., Reference Karschner, Schwilke, Lowe, Darwin, Herning, Cadet and Huestis2009) that measurable THC concentrations in blood after cessation of cannabis use may explain why some studies find more persistent cognitive decrements, while others do not (e.g., Scott etal., Reference Scott, Slomiak, Jones, Rosen, Moore and Gur2018). Relatedly, more recent best practice recommendations suggest that studies ideally utilize both self-report and THCCOOH concentrations to best understand patterns and profiles of cannabis use (Smith etal., Reference Smith, Alden, Herrold, Roberts, Stern, Jones and Breiter2018). Objective quantified urinary concentrations may yield more interpretative results beyond qualitative urinalysis or self-report, yet the potential utility of this approach for assessing neuropsychological outcomes in cannabis research has been sparingly investigated.
Another complicating factor is the co-use of other substances, particularly nicotine (Ramo, Liu, & Prochaska, Reference Ramo, Liu and Prochaska2012). Acute nicotine smoking was shown to increase verbal memory performance in young adults (Potter, Hammond, Tuffnell, Walker, & Di Forti, Reference Potter, Hammond, Tuffnell, Walker and Di Forti2018) with decrements in performance occurring during withdrawal (Jacobsen etal., Reference Jacobsen, Krystal, Mencl, Westerveld, Frost and Pugh2005). A recent study suggested chronic nicotine use during young adulthood was positively associated with verbal memory performance in females, but not males (Kangiser, Lochner, Thomas, & Lisdahl, Reference Kangiser, Lochner, Thomas and Lisdahl2019), and some find poorer verbal recall associated with greater intensity (cigarettes per day and duration) of nicotine intake (Vajravelu, Gnanadurai, Krishnan, & Ayyavoo, Reference Vajravelu, Gnanadurai, Krishnan and Ayyavoo2015). Adding to the complexity, some also found that nicotine may mask memory deficits in young adult cannabis users (Hindocha, Freeman, Xia, Shaban, & Curran, Reference Hindocha, Freeman, Xia, Shaban and Curran2017; Schuster, Crane, Mermelstein, & Gonzalez, Reference Schuster, Crane, Mermelstein and Gonzalez2015), suggesting interaction between the two substances wherein memory deficits may be most evident among low (or no) levels of nicotine use. In addition, no known studies investigated the relationship between cotinine and verbal memory. Taken together, this establishes a strong need to better assess the influence of nicotine and tobacco product (NTP) use and cotinine concentrations on verbal memory, particularly in young adult cannabis co-users.
For the present study, we aimed to determine if objective urinary markers of cannabis (THCCOOH) and NTP (cotinine) use predict verbal learning and memory performance. Therefore, we hypothesized that current cannabis users would have poorer verbal learning and memory than demographically matched non-users and former users, and that higher urinary THCCOOH and cotinine concentrations would independently associate with poorer performance on learning and memory. Second, THCCOOH has a long urinary detection window in frequent cannabis users following initiation of abstinence (Lowe etal., Reference Lowe, Abraham, Darwin, Herning, Cadet and Huestis2009; Schuster etal., Reference Schuster, Potter, Vandrey, Hareli, Gilman, Schoenfeld and Evins2020). Therefore, we tested if the cannabis-memory relationship was dependent on cannabis intake frequency (frequent vs. occasional group status). Finally, we explored if urine THCCOOH concentrations provided additional information beyond self-reported recency and past-month cumulative cannabis and nicotine use in relation to potential memory deficits.
METHODS
Participants
In total, 105 participants aged 16–22 were included from an ongoing study in San Diego County, California. Two participants were excluded for having incomplete data due to scheduling conflicts and being unable to complete the full study protocol, resulting in a final sample of 103 participants. Participants were recruited via flyers posted physically and electronically at local high schools, community colleges, and four-year universities. The advertisements described a research opportunity for a study on cannabis, nicotine, and brain development. Interested individuals called in to the laboratory phone. They provided verbal informed consent prior to assessing eligibility (or if <18 years old, consent from a parent and verbal assent for the participant; parental participation in the study extended only so far as consenting to their child’s participation). Once consented/assented, participants completed a semi-structured brief interview that took approximately 10 min. Screening questions ensured participants met inclusion/exclusion criteria as listed below, briefly covering prenatal, medical, mental health, and substance use history.
Inclusion Criteria
To ensure inclusion of both cannabis users and controls, recruitment and study procedures included participants who had either used cannabis in the past month or, for controls, not used cannabis in the past month. Sixty-six participants had used cannabis in the past month, and 37 had not used cannabis in the past month. In order to test whether the metabolite-performance relationship was dependent on cannabis user type and consistent with prior definitions from the cannabis toxicology literature (Desrosiers, Lee, etal., Reference Desrosiers, Lee, Concheiro-Guisan, Scheidweiler, Gorelick and Huestis2014; Huestis & Smith, Reference Huestis and Smith2018), cannabis users were divided into a frequent intake group (more days smoking than not, defined as >15 use episodes in the past month, n = 37) and occasional users (>1 and ≤15 use episodes in the past month, n = 29). Participants reported a wide range of past year episodic cannabis use (0–4, 0–15 cannabis use episodes), suggesting some participants were using cannabis multiple times a day in separate use occasions. NTP use was assessed across both cannabis users and controls.
Exclusion Criteria
Exclusion criteria for all participants included: excessive prenatal alcohol (maternal use of >2 drinks per occasion, >4 drinks in a week), tobacco, or drug exposure; premature birth (<34 weeks gestation); other gestational or perinatal complications, including low birth weight (<5 lbs); history of serious medical or neurological problems; head trauma with loss of consciousness >2 min; current or past DSM-5 diagnoses other than cannabis or nicotine use disorder; learning disability; current use of psychotropic medications; non-correctable vision/hearing difficulties; not fluent in English; pregnancy; use of alcohol or cannabis within 12 h of study visit which would indicate potential current intoxication (see toxicology section below) (Dahlgren etal., Reference Dahlgren, Sagar, Smith, Lambros, Kuppe and Gruber2020; Hindocha etal., Reference Hindocha, Freeman, Xia, Shaban and Curran2017; Pope, Gruber, Hudson, Huestis, & Yurgelun-Todd, Reference Pope, Gruber, Hudson, Huestis and Yurgelun-Todd2001).
In addition, all other substance use history was collected. Minimal use of other substances was observed in this sample. Participants reported individual past and recent episodic use of spice, opiates, amphetamines (other than as prescribed), barbiturates, hallucinogens, cocaine, inhalants, benzodiazepines, ecstasy, ketamine, GHB, and PCP. Participants reported an average of 4.4 other drug use episodes in their lifetime (SD = 18.8, range = 0–183) and average of 288.4 days of abstinence from other drug use (SD = 425.4, range = 2–2,555). Drugs used in the past month include cocaine, ADHD medications (not as prescribed), ecstasy, and hallucinogens; no participants were positive for any of these substances on toxicological analysis. Given some participants had used other drugs than cannabis, NTP, and alcohol within the month leading up to study participation, analyses were run both with and without those who had used other drugs in the past month. Findings remained largely consistent in either case; results presented here include all participants, regardless of past month other substance use.
Procedures
After confirming eligibility through screening, participants were brought into the laboratory and completed a 4-h session consisting of cognitive assessment, toxicological analysis, and magnetic resonance imaging (MRI) scan (MRI data to be presented elsewhere). They were asked to remain abstinent from all drug use (other than nicotine) on the day of study participation (see Toxicological section). All participants underwent written informed consent (or consent from a parent if <18 years old and assent from the participant) in accordance with the University of California, San Diego Human Research Protections Program.
MEASURES
Verbal Learning and Memory
Participants were given the Rey Auditory Verbal Learning Test (RAVLT) (Schmidt, Reference Schmidt1996). Participants were read a list of 15 words over five trials and asked to repeat back as many of the words as they could remember after each reading. They were then read a second (distractor) list, with an immediate recall trial. After the distractor list, participants were asked to recall the first list again. Finally, after a 30-min delay, participants were again asked to recall the original list. Variables of interest included raw scores for: initial learning (first trial recall), total learning (sum of trials 1–5 recall), short delay recall, and long delay recall. Each of these variables were included due to the unique aspect of learning and memory that they represent (Strauss, Sherman, & Spreen, Reference Strauss, Sherman and Spreen2006). Trial 1 performance indicates initial learning, a measurement of working memory. Trial 1–5 indicates total learning, revealing initial acquisition of learning. Short delay recall (Trial 6) demonstrates initial retention and consolidation, while long delay recall shows encoding and retrieval.
Substance Use History
Participants completed a modified version of the original Customary Drinking and Drug Use Record (CDDR) (Brown etal., Reference Brown, Lippke, Tapert, Stewart and Vik1998; Jacobus etal., Reference Jacobus, Taylor, Gray, Meredith, Porter, Li and Squeglia2018; Karoly, Schacht, Jacobus, etal., Reference Karoly, Schacht, Jacobus, Meredith, Taylor, Tapert and Squeglia2019; Karoly, Schacht, Meredith, etal., Reference Karoly, Schacht, Meredith, Jacobus, Tapert, Gray and Squeglia2019) to assess past year and lifetime substance use history. Given the complexity of cannabis use and the rise in vaping products, participants were additionally queried on most common forms of substance use (e.g., flower, concentrate, vaping, dabs) and potency, as well as personal history of use (e.g., age of first use, age of onset of regular use). Cannabis and NTP use were measured episodically, allowing for participants to report multiple use episodes in a single day if they were fully separate instances (e.g., right after waking up; after lunch).
Toxicological Assessment
Urine, oral fluid, and breathalyzer samples for alcohol were collected to corroborate self-reported substance use. Oral fluid samples examined THC and other substance use using the Draeger DrugTest® 5000 (cutoff = 5 ng/ml THC). The Draeger DrugTest is one of the most sensitive and effective methods for detecting cannabis and/or other substance use within the past 12 h (Desrosiers, Milman, etal., Reference Desrosiers, Milman, Mendu, Lee, Barnes, Gorelick and Huestis2014; Wille, Samyn, Del Mar Ramirez Fernandez, & De Boeck, Reference Wille, Samyn, Del Mar Ramirez Fernandez and De Boeck2010), ensuring participants are not acutely intoxicated (Desrosiers & Huestis, Reference Desrosiers and Huestis2019; Desrosiers, Milman, etal., Reference Desrosiers, Milman, Mendu, Lee, Barnes, Gorelick and Huestis2014). Redwood Toxicology Laboratory, Santa Rosa, CA, quantified urinary THCCOOH concentrations and normalized these to urinary creatinine, and quantified cotinine concentrations to confirm nicotine use. THCCOOH was confirmed at 5 ng/ml (Redwood Laboratory, 2020), making it even more sensitive than federal workplace guidelines for cannabinoid urine testing (Kulig, Reference Kulig2017). Creatinine-normalized THCCOOH accounted for the individual’s state of hydration and reduced variability (Huestis etal., Reference Huestis, Blount, Milan, Newmeyer, Schroeder and Smith2019; Huestis & Cone, Reference Huestis and Cone1998). THCCOOH, rather than THC, was measured due to the rapid metabolization of THC into THCCOOH once THC is ingested. THCCOOH is also the primary drug analyte tested in urinalysis for cannabinoids (Kulig, Reference Kulig2017). A breathalyzer was used to confirm abstinence from alcohol. Participants were allowed to use NTP adlibitum so as to prevent withdrawal effects; tobacco use recency ranged from 3 min to 3,650 days.
Demographics and Verbal Learning Performance
Demographic characteristics (sex, race, ethnicity, education, maternal education) were considered for inclusion in all analyses. However, as only age differed by group status and no demographic covariates related to RAVLT performance, only age was included as a covariate in analyses. For all group-based analyses, ANOVAs and ANCOVAs examined demographic covariates (i.e., age) and cognitive differences by cannabis group status for comparisons on cognitive measures.
Primary Analyses
Linear regression models examined the influence of quantitative THCCOOH and quantitative cotinine on four subtest measures of RAVLT performance: initial learning (first trial recall), total learning (sum of trials 1–5 recall), short delay recall, and long delay recall subtests. No participants with substance use and RAVLT data were excluded from analyses. Four hierarchical models were run in total, with THCCOOH, cotinine and covariates (i.e., age and past month alcohol use) included in the first step and a product term between urinary THCCOOH and cotinine entered in the second step for all four models to test the interaction between these variables. Data from significant regressions were assessed for outliers using DFBetas with cut-off value size 0.19, and no outliers were found to influence parameter estimates. Primary regressions were further confirmed through use of weight least squares regressions and iteratively reweighted least squares regressions to ensure sensitivity and robustness of results; results were largely invariant. Additional regressions examined whether cannabis user group type (i.e., frequent vs. occasional; frequent = more days using cannabis than not (Desrosiers, Lee, etal., Reference Desrosiers, Lee, Concheiro-Guisan, Scheidweiler, Gorelick and Huestis2014)) moderated the relation between metabolite concentrations and cognitive performance, given cannabis concentrations can have long detection windows for more frequent users despite abstinence (Huestis & Smith, Reference Huestis and Smith2018). For these models, group status (frequent user, occasional user, and control) was included in the first step of the regression with THCCOOH and covariates; two THCCOOH*group interaction terms were included in the second step to reflect dummy coding of frequent vs. control and occasional vs. control interactions. All individuals who were negative on drug screens had 0 imputed for their quantitative THCCOOH and/or cotinine concentrations. This included controls and some self-reported cannabis (n = 14) and NTP users (n = 24) who had metabolite concentrations that were undetectable in urine drug screening.
Secondary Analyses
Exploratory analyses were conducted in the full sample on cognitive variables that demonstrated significant relationships with THCCOOH and/or cotinine concentrations in the primary analysis; only the recency analyses were reduced due to excluding participants who had never used cannabis. Hierarchical regressions were run to test if metabolites predicted cognitive performance above and beyond self-reported use over the past month or self-reported recency of substance use, as studies thus far have not directly compared the predictive utility of objective biological markers of cannabis and nicotine as compared to subjective self-reported use of cannabis and nicotine. Separate models tested self-reported recency and self-reported cumulative intake over the past 30 days, in the event one self-report measurement was a more robust predictor. Covariates (age) and recency or cumulative intake, for each respective model, was entered on step 1 and the corresponding metabolite on step 2 for each verbal learning subtest. Separate models were run to investigate self-reported cannabis and THCCOOH concentrations, and self-reported NTP and cotinine concentrations.
RESULTS
Demographics and Verbal Learning Performance
Mean age of participants was 19 years (SD = 1.6; see Table 1). Cannabis users were significantly older than Controls (F(1,102) = 5.48, p = .02). Cannabis users and Controls did not differ by gender, race, ethnicity, education, maternal education, or by past month NTP or alcohol use episodes (p > .05). Of participants who used cannabis in the past month, relative to low (around <5% THC) or medium (10% THC) flower intake, 46% reported intake of high THC (15%) flower and 20% reported very high THC (>20%) flower intake. Relative to low (around 20% THC) or medium (around 40% THC) concentrate use, 26% reported high potency (60% THC) concentrate intake, while another 33% reported very high (>80% THC) concentrate intake. Participants further reported that 51% mostly to always used high potency flower, while 41% report mostly to always used high potency concentrate.
Table 1. Demographics and substance use characteristics
Notes: sample size (n) indicated for number of participants who endorsed a response when it was not endorsed by all participants with one or both groups (e.g., not all Controls had used cannabis and so did not endorse an age of first use, never having used previously).
As anticipated, tests of initial (Trial 1) recall (F(1,100) = 8.615, p = .004), total learning (F(1,100) = 8.059, p = .005), short delay recall (F(1,100) = 4.985, p = .03) and long delay recall (F(1,98) = 4.308, p = .04) had significantly poorer performance across metrics in all cannabis users as compared to controls (see Table 2 for cognitive performance information). Age was not significantly related to any outcome.
Table 2. Rey auditory verbal learning test (RAVLT) performance
Primary Analyses
Metabolites in Relation to Verbal Learning
Regressions including all participants assessed the predictive utility of urinary THCCOOH and cotinine concentrations on verbal memory performance, controlling for age and past-month alcohol use. Higher concentration of creatinine-normalized THCCOOH was significantly related to lower scores on total learning (ß = −0.196, t = −2.055, p = .043) and long delay recall (ß = −0.232, t = −2.413, p = .018) and is reflected in Figure 1. Higher cotinine concentration was similarly related to fewer recalled words after a short (ß = 0.218, t = −2.251, p = .027) and long delay (ß = −0.195, t = −2.032, p = .045), and is reflected in Figure 2. More past month self-reported alcohol use also was related to poorer total learning (ß = −0.219, t = −2.224, p = .028). The THCCOOH*cotinine interaction term was not significant (ps > .450) in the regression models and, therefore, this term was not retained in the final models in favor of parsimony.
Fig. 1. Scatterplot of significant relationships between creatinine-normalized THCCOOH concentration and (a) total learning and (b) long delay recall. Figures presented represent bivariate relationships, which reflect but do not directly correspond to model results summarized within the text.
Fig. 2. Scatterplot of significant relationships between cotinine concentration and (a) short delay recall and (b) long delay recall. Figures presented represent bivariate relationships, which reflect but do not directly correspond to model results summarized within the text.
Cannabis Use Frequency
Regressions assessed verbal learning and memory performance in relation to THCCOOH concentration and group status (controls, occasional users, or frequent users), and the THCCOOH*group interaction, controlling for age. As there was no THCCOOH*group interaction, the interaction term was not included in the final models for parsimony. Group status was significantly related to initial (Trial 1) learning (ß = −0.259, t = −2.252, p = .027). Follow-up analyses to determine significant differences by group revealed frequent users exhibited significantly lower performance than controls (p = .03) and marginally lower performance than occasional users (p = .055). There was no difference between occasional users and controls.
Secondary Analyses
Recency and Cumulative Cannabis Use
In assessing the utility of THCCOOH concentrations relative to self-reported length of abstinence, self-reported recency since last use of cannabis did not relate to total learning and/or delayed memory. THCCOOH concentrations continued to relate to long delay recall (ß = −0.226, t = −2.016, p = .047, ΔR 2 = .053), controlling for self-reported recency (ß = −0.009, p = .939) and age (ß = −0.059, p = 0.604). Similarly, in assessing the utility of THCCOOH concentrations relative to self-reported cumulative cannabis use in the past month, cumulative use did not relate to total learning and/or delayed recall. Higher THCCOOH concentrations continued to relate to poorer long delay recall (ß = −0.248, t = −2.338, p = .021, ΔR 2 = .054), controlling for self-reported past month use (ß = 0.018, p = .862) and age (ß = −0.026, p = .793).
Recency and Cumulative Nicotine Use
In assessing the utility of nicotine relative to self-reported length of abstinence, self-reported recency of last NTP use was unrelated to short and long delay recall. Cotinine was negatively associated with short delay performance (ß = −0.255, t = −2.089, p = .041, ΔR 2 = .064) controlling for self-reported recency (ß = −0.038, p = .759) and age (ß = −0.090, p = .466). Similarly, past month cumulative NTP use did not predict short and long delay recall. However, in the second step, cotinine concentration was negatively associated with short delay performance (ß = −.224, t = −2.298, p = .024, ΔR 2 = .049), controlling for cumulative NTP use (ß = 0.131, p = .183) and age (ß = −0.90, p = .356).
DISCUSSION
Here we assessed the relation between urinary THCCOOH concentrations, cannabis intake frequency, and memory performance. Consistent with prior results (Blest-Hopley etal., Reference Blest-Hopley, Giampietro and Bhattacharyya2020), cannabis use was related to poorer verbal learning and memory. Specifically, greater urinary THCCOOH concentrations were related to poorer learning and delayed recall performance across both occasional and frequent users of cannabis; and frequent cannabis users in particular demonstrated poorer performance on initial learning and short delay recall. Cotinine, a metabolite of nicotine, was also significantly related to poorer short delay performance. Finally, comparison of self-report metrics (recency of use and cumulative use) and urinary samples on significant cognitive outcomes found that THCCOOH and cotinine concentrations continued to relate to poorer performance, while self-report variables were unrelated.
Results suggest that cannabis metabolite concentrations in urine predict level of performance on a verbal learning and memory task. Others previously showed urinary THCCOOH concentration to be predictive of learning and memory in adults (Owens etal., Reference Owens, McNally, Petker, Amlung, Balodis, Sweet and MacKillop2019; Pope etal., Reference Pope, Gruber, Hudson, Huestis and Yurgelun-Todd2001). In addition, having high THC concentrations (vs. a median-split low concentration) in hair was associated with decrements in memory performance (Morgan etal., Reference Morgan, Gardener, Schafer, Swan, Demarchi, Freeman and Curran2012), and higher THC serum concentration was associated with motor impairment (Bonnet, Borda, Scherbaum, & Specka, Reference Bonnet, Borda, Scherbaum and Specka2015). Though disparate in biological matrices, the cumulative message of these various studies combined with the present results suggest that reliable, objective biological samples may be better predictors of certain cannabis-related cognitive outcomes. This may be due to numerous factors that influence how an individual processes THC, such as dosing, frequency of use, and product used (Musshoff & Madea, Reference Musshoff and Madea2006; Sharma etal., Reference Sharma, Murthy and Bharath2012).
Implications of cannabinoid metabolite concentrations are interesting to consider. THCCOOH in urine is considered a marker of cannabis use, with the known limitation that it can be detected even after a month of monitored abstinence (Goodwin etal., Reference Goodwin, Darwin, Chiang, Shih, Li and Huestis2008; Schuster etal., Reference Schuster, Potter, Vandrey, Hareli, Gilman, Schoenfeld and Evins2020). Notably, the highest concentration of creatinine-normalized THCCOOH is detected usually within a couple days of last use (Goodwin etal., Reference Goodwin, Darwin, Chiang, Shih, Li and Huestis2008). As the small effect of cannabis on learning and memory may only last for the first initial days of abstinence (Scott etal., Reference Scott, Slomiak, Jones, Rosen, Moore and Gur2018), it may be that cognition improves as markers of cannabinoid use decrease (e.g., lower THCCOOH levels, perhaps in conjunction with other factors such as upregulation of cannabinoid activity with recent abstinence (Hirvonen etal., Reference Hirvonen, Goodwin, Li, Terry, Zoghbi, Morse and Innis2012)). Interestingly, self-reported abstinence and recency since last use itself was not related to memory performance. However, this may be due to other important factors that cannot be detected in a broad self-reported variable, such as potency of product used (Fabritius etal., Reference Fabritius, Chtioui, Battistella, Annoni, Dao, Favrat and Giroud2013; Greene etal., Reference Greene, Wiley, Yu, Clowers and Craft2018) or genetics (Hryhorowicz etal., Reference Hryhorowicz, Walczak, Zakerska-Banaszak, Slomski and Skrzypczak-Zielinska2018), which may contribute to biometric data. Clinically, then, measuring objective levels of THCCOOH may be a useful indicator of how much cannabis is potentially influencing memory performance; high levels of THCCOOH would make memory performance suspect. More broadly, results speak to the importance of several days of abstinence from cannabis prior to neuropsychological assessment, to ensure results reflect more enduring brain-behavior relationships during neuropsychological testing, rather than acute cannabinoid-related deficits.
Importantly, a near ubiquitous issue in the substance use literature is reliance on self-report. Adolescent substance use misreporting is associated with factors such as age of onset and mental health (Harris etal., Reference Harris, Griffin, McCaffrey and Morral2008), and one study suggests misreporting is leading to a gross underestimate of how many adolescents actually use cannabis (Murphy & Rosenman, Reference Murphy and Rosenman2019). Accordingly, a family substance use study found urine drug screen results were discordant with 13% of male adolescents substance use self-report (Williams & Nowatzki, Reference Williams and Nowatzki2009). Yet, self-report may still provide some meaningful information. As exhibited in the present results, we did not observe a group (e.g., occasional cannabis use, frequent cannabis use) by THCCOOH interaction on verbal memory performance. Yet, groups representing self-reported frequency of use predicted initial learning and short delay performance—two variables that were not predicted by THCCOOH concentrations across all groups of cannabis users. This may suggest self-reported use and self-reported use patterns also capture other cannabis-related factors (e.g., trait characteristics with similar neurobiological etiology or premorbid functioning) that obscure some direct relationships between cannabis and cognitive outcomes. Importantly, the use of creatinine-normalized analytes is thought to be highly reliable and suggested as appropriate for interpretation of objective cannabis measurement (Huestis etal., Reference Huestis, Blount, Milan, Newmeyer, Schroeder and Smith2019). Further, use of urinary cannabinoid levels is significantly related to carefully quantified (gram-based) measurement of cannabis use (Tomko etal., Reference Tomko, Baker, McClure, Sonne, McRae-Clark, Sherman and Gray2018), suggesting convergent validity of methodologies. Thus we agree with others who suggested biological samples, in addition to self-report, should be sought in all cannabis research (Smith etal., Reference Smith, Alden, Herrold, Roberts, Stern, Jones and Breiter2018).
We found one relation between NTP measures and verbal memory, as a higher cotinine concentration was associated with poorer verbal memory. This is in contrast to other cross-sectional studies of young adult nicotine users (Kangiser etal., Reference Kangiser, Lochner, Thomas and Lisdahl2019). We allowed smoking up to an hour before testing and, therefore, do not believe this to be due to withdrawal effects (Jacobsen etal., Reference Jacobsen, Krystal, Mencl, Westerveld, Frost and Pugh2005). Others found intensity of smoking determines cognitive outcomes in young adults (Vajravelu etal., Reference Vajravelu, Gnanadurai, Krishnan and Ayyavoo2015), suggesting disparate results may be due to young adults smoking at low levels of use, which do not correlate to deficits in cognitive functioning. In addition, users here were not necessarily pure NTP users, but often used alcohol if not cannabis as well. However, follow-up analysis that assessed for interactive relationships between THCCOOH and cotinine concentrations on memory performance revealed no significant results, despite prior work suggesting a moderating influence of NTP on cannabis (Schuster etal., Reference Schuster, Crane, Mermelstein and Gonzalez2015). Therefore, more careful teasing apart of co-use of NTP with other substances by including an NTP-only group is in process for future analyses.
Interestingly, all assessed substances were associated with decrements in learning and/or memory. This is consistent with cannabis (Jacobus etal., Reference Jacobus, Squeglia, Infante, Castro, Brumback, Meruelo and Tapert2015; Solowij etal., Reference Solowij, Jones, Rozman, Davis, Ciarrochi, Heaven and Yucel2011) and alcohol (Spear, Reference Spear2018) literature suggesting memory deficits in regular adolescent and young adult users. Preliminary evidence also suggests greater frequency of nicotine use is associated with cognitive deficits (Vajravelu etal., Reference Vajravelu, Gnanadurai, Krishnan and Ayyavoo2015), and sex may moderate the influence of nicotine on memory (Kangiser etal., Reference Kangiser, Lochner, Thomas and Lisdahl2019). While nicotine may acutely improve cognition (Campos, Serebrisky, & Castaldelli-Maia, Reference Campos, Serebrisky and Castaldelli-Maia2016), and attenuate cannabis-related cognitive decline (Schuster etal., Reference Schuster, Crane, Mermelstein and Gonzalez2015), more recent and chronic use (such as would be measured in cotinine) may be associated with unique cognitive decline in consolidation and retention. Thus, our null interaction results may be unsurprising given there is likely a nuanced relationship between frequency of nicotine use, recency of cannabis use, and learning and memory performance. More prospective research is needed on larger samples to fully delineate patterns of use and co-use in relation to memory function.
This study replicates numerous prior reports of emerging adult cannabis users demonstrating deficits in verbal memory relative to controls, while importantly revealing novel relationships with more reliable, objective measurements of cannabis use through urine toxicology. At the same time, there are limitations to be considered. Due to financial limitations, we did not send the urine sample for quantitative confirmation if the on-site screening was negative, therefore, we used 0 for these samples for their creatinine-normalized THCCOOH concentrations. As aims of this project included increased generalizability of cannabis research, and examine cumulative and recency of use, we chose to enroll controls who had recent and remote use of cannabis as well as other substance use. Future studies may benefit from excluding participants with other substance use so as to more directly isolate the effects of cannabis and NTP use. As a cross-sectional study, we are unable to determine directionality between cannabis use and cognitive functioning. Further, while we carefully selected a relatively small group of theoretically-driven verbal memory subtests a priori, results are preliminary in nature and did not include correction for multiple comparisons, and so replication is important. Longitudinal studies which follow youth before the onset of use will be important for inferring causality in the future (e.g., the Adolescent Brain Cognitive Development Study, abcdstudy.org). Here we focused on a neuropsychological test of one of the most commonly cited cognitive correlates of cannabis use, verbal learning and memory; however, the assessment of use of biometrics across additional cognitive domains is warranted.
In summary, results suggest that young adult cannabis users demonstrate poorer verbal learning and memory versus non-users, and this relationship is well captured through use of urinary creatinine-normalized THCCOOH concentrations and through self-reported frequent cannabis use. Urinary cotinine concentrations similarly related to short delay memory performance. Use of biometrics to test for cannabis exposure may be a means of reliably assessing cannabis-related cognitive decrements, given the vast array of cannabis products and individual use characteristics that influence the pharmacokinetic profile (Huestis, Reference Huestis2007).
ACKNOWLEDGEMENTS
Research reported in this publication was supported by the National Institute on Alcohol Abuse and Alcoholism grant T32 AA13525 (PI: Riley/Tapert to Wade) and National Institute on Drug Abuse grants U01 DA041089 and R21 DA047953, and the California Tobacco-Related Disease Research Grants Program Office of the University of California Grant 580264 (PI: Jacobus).
CONFLICTS OF INTEREST
The authors have nothing to disclose.
ETHICS OF HUMAN SUBJECT PARTICIPATION
This study was completed in accordance with the Declaration of Helsinki and was approved by the Institutional Review Board at the University of California, San Diego.
