Content uploaded by Mina Kwon
Author content
All content in this area was uploaded by Mina Kwon on Sep 14, 2023
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
1
Abstract 1
Background and Objectives: Use of psychotropic substances in childhood has been 2
associated with both impulsivity and other manifestations of poor executive function as 3
well as escalation over time to use of progressively stronger substances. However, how 4
this relationship may start in earlier childhood has not been well explored. Here, we 5
investigated the neurobehavioral correlates of daily caffeinated soda consumption in 6
preadolescent children and examined whether caffeinated soda intake is associated with a 7
higher risk of subsequent alcohol initiation. Methods: Using Adolescent Brain Cognitive 8
Development study data (N=2,092), we first investigated cross-sectional relationships 9
between frequent caffeinated soda intake and well-known risk factors of substance 10
misuse: impaired working memory, high impulsivity, and aberrant reward processing. We 11
then examined whether caffeinated soda intake at baseline predicts more alcohol sipping 12
at 12 months follow-up using a machine learning algorithm. Results: Daily consumption 13
of caffeinated soda was cross-sectionally associated with neurobehavioral risk factors for 14
substance misuse such as higher impulsivity scores and lower working memory 15
performance. Furthermore, caffeinated soda intake predicted a 2.04 times greater 16
likelihood of alcohol sipping after 12 months, even after controlling for rates of baseline 17
alcohol sipping rates. Conclusions: These findings suggest that previous linkages 18
between caffeine and substance use in adolescence also extend to younger initiation, and 19
may stem from core neurocognitive features thought conducive to substance initiation. 20
Keywords: caffeinated soda, alcohol sipping, risk factors of substance use, impulsivity, 21
working memory, ABCD study 22
23
2
Soft drinks are commonly consumed even by children, and a vast majority of 1
sodas contain caffeine (Temple, 2009, 2018). Moreover, caffeinated soda typically 2
contains sweeteners such as high-fructose corn syrup, which can affect neurocognitive 3
function and cause physical side effects, such as through effects on microbiome (Ettinger, 4
2022). Not surprisingly, the consequences of excessive consumption of both sugars and 5
caffeine have been well documented (Ooi et al., 2022; Porciúncula et al., 2013; Temple et 6
al., 2017), including a strong association between caffeinated beverage consumption in 7
adolescence and future substance use (Arria et al., 2011; Barrense‐Dias et al., 2016; 8
Kristjansson et al., 2018; Leal & Jackson, 2019; Marmorstein, 2018; Miyake & 9
Marmorstein, 2015). In prospective studies tracking the effects of substance use, the 10
percentage of regular energy drink users who became alcohol or marijuana users after 1–11
2 years was approximately five times higher than that of non-energy drink users (Leal & 12
Jackson, 2019; Marmorstein, 2018; Miyake & Marmorstein, 2015). Others have shown 13
that coffee or energy drink consumption in adolescents or young adults significantly 14
predicts future substance use, such as tobacco and alcohol use (Arria et al., 2011; 15
Barrense‐Dias et al., 2016; Kristjansson et al., 2018, 2022; Marmorstein, 2018). 16
In light of the potential for a problematic progression from caffeinated beverages 17
to harder substance use, there is a critical need to investigate aspects of this progression 18
as early as possible. As adolescence is the most common period for initiating substance 19
use and an earlier onset of substance use predicts greater addiction severity (Chambers et 20
al., 2003; Jordan & Andersen, 2017), examining these relationships in preadolescence is 21
crucial in that caffeinated soda intake in childhood could provide useful predictive 22
3
information on future substance use. For the preadolescence children, caffeinated soda is 1
the most preferred and accessible mode of caffeine intake (Temple, 2009, 2018). 2
However, few studies have directly examined the effects of frequent caffeinated 3
beverage consumption in preadolescent children, in light of their lower rates of 4
consumption (0.4% for coffee and < 0.1% for energy drinks in 9–10-year-old children) 5
(Lisdahl et al., 2021). Most of the previous studies examining the association between 6
caffeine consumption and later substance use have focused primarily on adolescents who 7
drink multiple caffeinated beverages on a daily basis (Temple, 2009, 2018). Moreover, 8
while a few existing studies have examined behavioral risk factors associated with 9
caffeinated soda intake, they have not obtained neural assessments of such risk factors 10
(James et al., 2011; Miyake & Marmorstein, 2015; Solnick & Hemenway, 2013; Suglia et 11
al., 2013). 12
Here, we addressed the unanswered question of whether frequent consumption of 13
caffeinated soda in preadolescent children indicates a higher risk of future alcohol 14
experimentation, using data from the Adolescent Brain Cognitive Development (ABCD) 15
Study (Bjork et al., 2017). In addition, we wished to explore the potential 16
neurobehavioral mechanisms of such relationships. In light of previous linkages between 17
activity in neurocircuits germane to motivation and inhibition (as detected from 18
functional magnetic resonance imaging (fMRI)) and substance use in adolescents (Lees et 19
al., 2021), we examined the relationship between daily caffeinated soda intake and both 20
behavioral and neuroimaging markers of neurobehavioral risk factors for substance use 21
disorders (SUDs) in children. These include impaired working memory (WM), high 22
impulsivity, and altered mesolimbic reward processing (Fig. 1A). These are the three 23
4
primary neurocircuit functions targeted in the ABCD study due to their significant 1
implications for addiction (see Casey et al. (2018) for more detail). We applied a machine 2
learning approach (a least absolute shrinkage and selection operator (LASSO) regression; 3
Tibshirani, 1996) to the measures so that we could identify multivariate risk factors for 4
SUDs associated with daily caffeinated soda intake and select features that are highly 5
associated with caffeinated soda intake (Volkow et al., 2015). Using the same approach, 6
we then examined whether caffeinated soda intake could predict future alcohol sipping, 7
as (1) alcohol sipping has been reported to predict future alcohol abuse (Jackson et al., 8
2015; Watts et al., 2021), and (2) alcohol sipping is the most common gateway behavior 9
toward other substances of abuse (Barry et al., 2016). Based on the previous literature 10
suggesting that caffeinated beverage consumption predicts future substance use (Arria et 11
al., 2011; Barrense‐Dias et al., 2016; Leal & Jackson, 2019; Marmorstein, 2018; Miyake 12
& Marmorstein, 2015), we hypothesized that daily caffeinated soda intake in the baseline 13
ABCD assessment (children age 9-10) would predict future alcohol sipping at the year 1 14
follow up (Fig. 1B). Taken together, we aimed to elucidate the potential underlying risks 15
of frequent caffeinated soda intake during childhood. 16
17
Methods 18
Participants 19
The ABCD study collected data from 11,878 children aged 9–10 years. The 20
participants were recruited via school systems from 21 different sites in the United States. 21
The Institutional Review Board (IRB) at the University of California, San Diego, 22
approved all the research protocol of the ABCD study (Auchter et al., 2018). All 23
5
participants provided written assent, and their parents or guardians provided written 1
consent (Auchter et al., 2018). More information about the recruitment and study design 2
is available in Garavan et al. (2018). Further details of the demographic, physical, and 3
mental health assessments are described in Barch et al. (2018). 4
Out of 11,878 children from ABCD release 2.0, we excluded those with any 5
missing data in the measures including caffeinated soda intake, neurobehavioral risk 6
factors for SUDs, future alcohol sipping, and confounding variables (see Measures). We 7
also excluded outliers based on the measures of caffeinated soda intake with the cutoff of 8
larger/smaller than the mean ±5 standard deviations. Of these, 147 reported to drink more 9
than 7 cans of caffeinated soda per week (daily-drinkers) and 1,945 reported not drinking 10
a single can (non-drinkers) in the past 6 months. Therefore, we included 2,092 11
participants for the main analyses (see Fig. S1 for a flowchart of the selection process). 12
To examine the potential for selection bias, we compared the baseline characteristics 13
among children who were included and excluded in the analyses (Table S1). 14
15
Measures 16
Caffeinated soda intake 17
Caffeinated soda intake was assessed by self-report of the participants in 18
response to the question, “How many drinks of the following beverages have you had per 19
week in the past 6 months? – soda with caffeine (Mountain Dew, Jolt, Coke, Pepsi, Dr. 20
Pepper, Barq’s root beer)”. The participants who reported consuming more than 7 cans 21
per week are allocated to daily-drinking group (N=147), and those who reported 22
consuming 0 can per week were allocated to non-drinking groups (N=1,945). See Table 23
6
S2 for a comparison of the daily-drinkers and non-drinkers on the following behavioral 1
measures, and see Table S3 for a comparison of the neural measures. 2
3
Neurobehavioral risk factors for SUDs 4
As shown in Fig. 1A, we focused on three well-known neurobehavioral risk 5
factors for SUDs; WM, impulsivity, and reward processing. These constructs were 6
collectively measured by self-report surveys, behavioral tasks, and fMRI, as described 7
below. Spatial regions of interest (ROIs) for each task were restricted a priori to the 8
canonical activations initially reported using the ABCD samples (Casey et al., 2018). The 9
curated data used were based on the Destrieux atlas (Destrieux et al., 2010). 10
To obtain behavioral measures of WM, we used the List Sorting Working 11
Memory Test (list sorting test; Tulsky et al., 2013) and Dimensional Change Card sort 12
Test (card sort test; Tulsky et al., 2013), as well response accuracy during the 2-back 13
condition of the emotional N-back (EN-back) task (Cohen et al., 2015) (see 14
Supplementary materials). The EN-back task was conducted inside the MRI scanner. 15
The contrast of "2-back versus 0-back" was used for the fMRI analysis, and the ROIs 16
were selected based on activation maxima of the contrast in the initial subset of 17
participants (Casey et al., 2018): These were the rostral middle frontal gyrus (MFG), 18
caudal MFG, inferior frontal gyrus (IFG; pars triangularis and pars orbitalis), lateral 19
orbitofrontal cortex (OFC), superior parietal lobule and inferior parietal lobule (IPL) in 20
the frontoparietal network, and the caudate nucleus, putamen, nucleus accumbens (NAc), 21
rostral anterior cingulate cortex (ACC), caudal ACC, thalamus proper, ventral 22
diencephalon, amygdala, and hippocampus in the fronto-thalamic network. 23
7
Due to the multi-faceted construct of impulsivity, wherein laboratory and self-1
report assessments of impulsivity are thought to capture different components (Sharma et 2
al., 2014), we analyzed each of self-report trait-like impulsivity as well as rapid-response 3
and decision-based impulsivity. Trait impulsivity was measured using the short form 4
Urgency–Premeditation–Perseverance–Sensation Seeking–Positive Urgency (UPPS-P) 5
impulsive behavior scale for children (the 20-item short version for youths; Barch et al., 6
2018) and a parental report of the ABCD Youth Behavioral Inhibition System/ 7
Behavioral Activation System (BIS/BAS; Carver & White, 1994). The BAS is related to 8
goal-directed efforts, such as motor activation in response to an impending reward, while 9
the BIS is engaged when inhibition toward a goal occurs, such as the avoidance of 10
punishment (Carver & White, 1994). The existing literature suggests that excessive 11
behavioral inhibition is associated with depression and anxiety, while excessive 12
behavioral activation is associated with impulsive behaviors, compulsive behaviors, 13
substance misuse, and aggression (Newman et al., 2005). To identify the neurocircuit 14
correlates of impulsivity, participants completed the stop signal task (SST; Logan et al., 15
1984) during fMRI. The SST measures impulsivity related to impulse control or response 16
inhibition when performing an action. The contrast of "correct stop versus correct go" 17
was used for the fMRI analysis. ROIs were the lateral prefrontal cortex (rostral MFG, 18
pars orbitalis in IFG, and lateral OFC), rostral and caudal ACC, and striatum (caudate, 19
putamen, and NAc). These areas have been associated with impulsivity and impulse 20
control (Aron et al., 2014; Casey et al., 2018). See Supplementary materials for more 21
details of the SST. 22
8
Reward processing is closely linked to impulsivity, as impulsive people often 1
show immediate reward-seeking behavior (Zuckerman, 2001). For our offline metric of 2
reward sensitivity, we used the Cash Choice Task (Luciana et al., 2018), which assesses 3
willingness to delay gratification (see Supplementary materials). Previous studies have 4
suggested that people with SUDs exhibit hyper-responsiveness of mesolimbic 5
motivational neurocircuitry toward drug-related cues (Bechara et al., 2019) while 6
showing blunted responsiveness toward cues for a non-drug reward (e.g., monetary 7
reward) (Luijten et al., 2017). The monetary incentive delay (MID) task (Knutson et al., 8
2000) is widely used to measure the neural correlates of the anticipation of monetary 9
rewards and losses. The contrast of “reward versus neutral” at the cue onset, which 10
reflects reward anticipation, was used for the fMRI analysis, using the ROIs in the ventral 11
striatum (i.e., NAc), rostral ACC, and medial OFC in the medial prefrontal cortex, which 12
play a key role in reward processing, particularly in reward anticipation (Bartra et al., 13
2013). See Supplementary materials for more details about the MID task. 14
15
Future alcohol sipping 16
Alcohol sipping was measured by self-report from the participants using the iSay 17
Sip Inventory, which was performed once a year and asked only in children who had 18
heard of alcohol (see Lisdahl et al. (2018) for more details). From the year 1 data (data 19
release 3.0), we used a binary response to a single question asking if the participants had 20
sipped alcohol outside of a religious ceremony. We considered those who had not heard 21
of alcohol as having yet to sip alcohol (i.e., no alcohol sipping). 22
23
9
Confounding variables 1
In light of the potential for socioeconomic and structural factors to influence 2
health behaviors, we included the following variables as confounding variables in the 3
data analysis: socioeconomic status (SES), family history of substance use, parental 4
monitoring, sleep deprivation, externalizing/internalizing symptomatology, data 5
collection sites, and type of MRI scanner. See Supplementary materials for more detail. 6
7
Analysis 8
Caffeinated soda intake and risk factors for SUDs 9
To identify a multivariate profile of risk factors for caffeinated soda 10
consumption, we performed a binomial LASSO regression analysis (Tibshirani, 1996), a 11
machine learning algorithm, to distinguish the daily soda-drinking group (N=147) from 12
the non-drinking group (N=1,945). We used all of the risk factors and confounding 13
variables as input (i.e., predictors) in the LASSO-based prediction model to classify each 14
individual into the daily-drinking group (coded as "1") or non-drinking group (coded as 15
"0). More specifically, candidate predictors included every measure for the three 16
cognitive factors as well as other control variables. As LASSO regression offers feature 17
selection (Tibshirani, 1996), we aimed to identify features that could differentiate the 18
daily-drinking group from the non-drinking group. See Supplementary materials for 19
more details on LASSO regression. 20
21
22
23
10
Caffeinated soda intake and future alcohol sipping 1
We performed a binomial LASSO regression analysis to test the association 2
between daily caffeinated soda intake and alcohol sipping. We used all possible variables, 3
including the three primary neurobehavioral risk factors (WM, impulsivity, and reward 4
processing) along with the confounding variables as candidate predictors collected at 5
baseline. Using these inputs (i.e., predictors), the LASSO-based model was used to 6
classify individuals who reported alcohol sipping after 12 months (coded as "1"; N=90) 7
and those who reported no alcohol sipping (coded as "0"; N=2,002). We aimed to select 8
predictors that could distinguish individuals who reported alcohol sipping from those who 9
did not. All of the procedures, except for the predictors and dependent variables, are 10
identical to those in the former analysis (see "Methods - Caffeinated soda intake and 11
risk factors for SUDs"). Lastly, we performed a chi-square test to compare the ratio of 12
alcohol sipping after 12 months between the two groups. Furthermore, we estimated the 13
risk ratio of the daily-drinking group by dividing the cumulative incidence of alcohol 14
sipping in the daily-drinking group by the cumulative incidence in the non-drinking 15
group. 16
To rigorously examine the association between caffeinated soda intake and future 17
alcohol sipping, we conducted three additional analyses. First, we used continuous 18
measure for caffeinated soda intake instead of using the categorical measure of daily- and 19
non-drinkers. Using data from 4,517 participants (see Fig. S1), we carried out the same 20
binomial LASSO regression analysis mentioned earlier (Fig. S3). Second, we only used a 21
behavioral measures of risk factors in predicting future alcohol sipping, to include as 22
many participants as possible. Using data from 8,939 participants (see Fig. S1), we again 23
11
performed the binomial LASSO regression analysis (Fig. S4). Lastly, using the sample of 1
8,939 participants, we used hierarchical logistic regression analysis to confirm whether 2
continuous measures of caffeinated soda intake at baseline can predict alcohol sipping 3
after 12 months (Table S4). For more details, please refer to the Supplementary 4
materials. 5
6
Results 7
Caffeinated soda intake and neurobehavioral risk factors for SUDs 8
Our first question was whether daily caffeinated soda intake is associated with 9
well-known neurobehavioral risk factors for SUDs. Fig. 2A shows the multivariate 10
profiles from binomial LASSO regression analysis distinguishing the daily-drinking 11
group from the non-drinking group. Family history of drug use and low parental 12
education were the two strongest predictors of daily consumption, along with sex (being 13
male), lack of sleep, low family income, race (being African American), high body mass 14
index (BMI), high externalizing behaviors, and low parental monitoring. Fig. 2B shows a 15
receiver-operation characteristic (ROC) curve and its mean area under the curve (AUC) 16
for classification of the daily-drinking and non-drinking groups. The mean AUC values 17
were 0.80 and 0.72 for the training and test sets, respectively. 18
Among our a priori candidate neurobehavioral risk factors for SUDs, high 19
impulsivity measured using the BAS was most strongly associated with daily caffeinated 20
soda intake (Fig. 2C). A higher UPPS-P score was also related to daily intake. 21
Additionally, hypoactivation in the caudal ACC during the SST predicted the 22
12
classification of the daily-drinking group. Among the WM measures, poor performance 1
on the list sorting test and 0-back performance in the EN-back Task predicted daily intake 2
of caffeinated soda. In addition, hypoactivation of the pars orbitalis of the IFG and 3
greater activation of the IPL by working memory demands during the EN-back task (2-4
back vs. 0-back) also predicted daily caffeinated soda intake (Fig. 2C). However, 5
variables related to reward processing were not found to be predictors of daily-drinking 6
group. 7
These results support that high impulsivity and low WM are significantly 8
associated with daily caffeinated soda consumption. Along with some other demographic 9
factors (family history of drug use, male sex, low SES, low parental monitoring, high 10
externalizing behaviors, less sleep, and high BMI), the two neurobehavioral risk factors 11
for SUDs distinguished the daily-drinking group from the non-drinking group. 12
13
Caffeinated soda intake and future alcohol sipping 14
To address our second question of whether daily caffeinated soda intake predicts 15
future alcohol sipping, we conducted binomial LASSO regression predicting alcohol 16
sipping after 12 months, using all of the neurobehavioral risk factors and confounding 17
variables collected at baseline. As shown in Fig. 3A, daily caffeinated soda intake was 18
identified as one of the predictors that distinguished alcohol sipping after 12 months 19
(mean estimate of coefficients=0.122, 95% confidence interval (CI)=[0.010, 0.235]), 20
even after including alcohol sipping at baseline as a predictor. The mean AUC values of 21
the predictive model were 0.90 and 0.72 for the training and test sets, respectively (Fig. 22
3B). Moreover, the ratio of alcohol sipping after 12 months was twice as high in the 23
13
daily-drinking group compared to the non-drinking group (daily-drinking group: 0.082% 1
(12 out of 147 children); non-drinking group: 0.040% (78 out of 1,945 children); X2 = 2
5.802, p = 0.016) (Fig. 3C). This finding suggests that the daily-drinking group had 2.04 3
times the risk of sipping alcohol compared to the non-drinking group. Our results were 4
further supported by the binomial LASSO regression analyses with larger samples of 5
4,529 participants (Fig. S4) and 8,939 participants (Fig. S5), as well as hierarchical 6
logistic regression analysis (Table S4) which can be found in the Supplementary 7
materials. 8
9
Discussion 10
In this study, we investigated whether caffeinated soda intake in preadolescent 11
childhood is associated with a higher risk of alcohol experimentation in the future. Using 12
the large dataset from the ABCD study, we first clarified cross-sectional relationships 13
between daily caffeinated soda intake and well-known neurobehavioral risk factors for 14
SUDs at study baseline, then evaluated whether caffeinated soda intake at baseline was 15
predictive of alcohol sipping after 12 months. Our findings suggest that frequent 16
consumption of caffeinated soda in children is closely related to previously-established 17
neurobehavioral risk factors for SUDs and can predict future alcohol sipping. 18
Our machine learning approach partially supported our hypothesized association 19
between caffeinated soda intake and the neurobehavioral risk factors for SUDs. Notably, 20
high impulsivity scores and low WM performance each singly distinguished daily 21
caffeinated soda drinkers from non-drinkers even after considering the effects of other 22
14
confounding factors (Fig. 2). However, there was no significant association observed for 1
reward processing. 2
In relation to impulsivity, we found higher self-reported impulsivity in the daily-3
drinking group based on the UPPS-P and BAS scores along with altered activation in the 4
ACC, a brain region implicated in cognitive control and impulsivity (Kerns et al., 2004; 5
Shenhav et al., 2016). Reduced activities in the ACC during response inhibition or 6
behavior-monitoring are commonly reported in children with attention deficit 7
hyperactivity disorder (ADHD) (Cortese et al., 2012; Hart et al., 2013; McTeague et al., 8
2017) and individuals with SUDs (Luijten et al., 2014; Nestor et al., 2011; Yücel et al., 9
2007). Thus, hypoactivation of the ACC during response inhibition in daily-drinking 10
group further seems to strengthen the association between daily soda consumption and 11
elevated levels of impulsivity. 12
We also found WM impairments in the daily-drinking group on the list sorting 13
test and the 0-back accuracy in the EN-back task, accompanied by hypoactivation in the 14
IFG and hyperactivation in the IPL by working memory demands during the EN-back 15
task. Prior studies have shown that greater activation in the prefrontal cortex is related to 16
greater WM capacity (Casey et al., 2018; Owen et al., 2005), and increased activation in 17
the IPL is associated with higher WM load (Baldo & Dronkers, 2006; Veltman et al., 18
2003). Taken together, these findings strongly suggest an association between daily-soda 19
consumption and WM deficits. 20
In contrast, we did not find a significant association between neurocognitive 21
measures of reward processing and daily caffeinated soda intake. Aberrant reward 22
processing is a commonly observed neurocognitive feature in addiction (Zeng et al., 23
15
2023), with individuals either displaying increased or decreased sensitivity to reward 1
(Berridge & Robinson, 2016; Blum et al., 2000; Demidenko et al., 2020; Robinson & 2
Berridge, 1993). Altered reward processing has been also reported in children with a high 3
risk of alcohol use, such as those with a family history of alcohol use problems (Bjork et 4
al., 2008; Martz et al., 2022). One possibility for the lack of association is that the effect 5
of reward processing variables may have been masked by the strong effect of family 6
history of drug use, since our study controlled for the familial risk of alcohol and drug 7
use. Additionally, it is worth considering that monetary rewards might have differential 8
effects compared to drug rewards (Nestor & Ersche, 2023), which could explain the 9
absence of significant associations. Consistent with our findings, a recent study 10
comparing reward processing of alcohol dependent patients, first-degree relatives, and 11
healthy controls could not find any significant group differences both in monetary reward 12
and loss anticipation when controlling for age (Musial et al., 2023). Thus, further research 13
is required to reconcile the mix findings related to reward processing in youth with a risk 14
of addiction, perhaps by studying more narrowly defined subgroups and by examining 15
both drug and non-drug rewards. 16
After examining the link between the neurobehavioral risk factors and 17
caffeinated soda intake, we showed that frequent consumption of caffeinated soda 18
predicted alcohol sipping after 12 months using LASSO regression (Fig. 3). Even after 19
controlling for other well-established variables for alcohol sipping, including baseline 20
alcohol sipping, caffeinated soda intake remained predictive of future alcohol sipping 21
(Fig. 3A). Other survived predictors of future sipping include predictors for daily 22
caffeinated soda intake itself (Fig. 2A), higher behavioral impulsivity score (i.e., UPPS-23
16
P) and BMI, and hypoactivation of the IFG during the EN-back task. Conversely, reduced 1
activity in the NAc and greater activity in the amygdala during the EN-back task, and 2
reduced activity of the medial OFC during the MID task were associated with the 3
participants who experienced alcohol sipping but did not survive as predictors for daily 4
caffeinated soda intake. Higher family income was associated with the participants who 5
experienced alcohol sipping, while it was associated with the participants who drink 6
caffeinated soda daily. Interestingly, predictors of caffeinated soda intake were mostly 7
similar to the risk factors for SUDs (Jordan & Andersen, 2017), suggesting that 8
caffeinated soda intake during childhood and SUDs share similar neurobehavioral 9
vulnerabilities. Note that the majority of the cohort was substance-naïve (Lisdahl et al., 10
2018), and we evaluated alcohol sipping measures instead of substantial self-11
administration of alcohol. Thus, it would be informative to track the predictive ability of 12
caffeinated soda and alcohol use and investigate changes in the direction of the predictors 13
as the children get older. 14
Our findings suggest that caffeinated soda in children is predictive of substance 15
use in the near future. The longitudinal associations between the use of more benign 16
psychotropic substances early in life and the use of “harder” substances later in 17
adolescence or emerging adulthood have been characterized for decades (Kandel, 1975), 18
and have been attributed to two competing (but not mutually-exclusive) theories. The 19
“gateway hypothesis” (Kandel & Yamaguchi, 2002) generally implies that exposure to 20
the earlier-used substance itself, such as nicotine or cannabis, induces a toxicological 21
effect on brain which renders the individual more sensitive to reinforcing effects of 22
harder drugs. Evidence for this theory is supported by controlled animal model 23
17
intervention studies (e.g., (Collins & Izenwasser, 2004)). Animal studies on the effects of 1
caffeine intake on increasing later alcohol consumption (Hou et al., 2016; Kunin et al., 2
2000; SanMiguel et al., 2019) supports that the gateway hypothesis could also be applied 3
to the association between caffeine and alcohol. Thus, one possible explanation of higher 4
alcohol sipping rate of the daily-drinking group in our study is that the substances 5
contained in caffeinated soda (caffeine and sugar) may have induced neurophysiological 6
effects and reinforced regular soda drinkers to try alcohol after 12 months. 7
Conversely, the “common liability” hypothesis posits that the progression from 8
softer to harder substances results primarily from the intersection of a genetically-9
regulated under-controlled temperament with progressively expanding access to harder 10
substances with advancement into young adulthood (Vanyukov et al., 2012). Cross-11
sectional observations supported the common liability hypothesis, as children and 12
adolescents with disruptive behavior disorders such as conduct disorder, which is 13
strongly linked to SUDs, showed deficits in executive function (e.g., impaired impulse 14
control) (Matthys et al., 2013). Longitudinal studies also support the common liability 15
hypothesis (Debenham et al., 2021), wherein impaired impulse control and high 16
sensation-seeking in young adolescents are predictive of increased drinking over time 17
(Castellanos‐Ryan et al., 2011). Similarly, caffeinated soda intake itself may not directly 18
cause children to initiate substance use, but the drinking behavior of the beverages could 19
indicate high impulsivity, which may be linked to a high risk of initiating substance use 20
in the near future, consistent with the common liability hypothesis (Vanyukov et al., 21
2012). Because caffeinated soda contains two addictive substances, sugar and caffeine, it 22
is somewhat natural for children to prefer and seek the beverages (Temple, 2009). While 23
18
this taste preference could lead to a seeking behavior and habitual consumption, previous 1
studies have shown impulsivity as one of the most common traits of SUDs (Cyders et al., 2
2009; Magid & Colder, 2007). Therefore, the behavior of frequently consuming 3
caffeinated soda could indicate a high risk of initiating substance use in the future, due to 4
the common risk factors (e.g., high impulsivity) between the two behaviors, although 5
further research is needed to disentangle this complex relationship. 6
To our knowledge, this is the first study to investigate the direct link between 7
caffeinated soda intake in childhood and the risk of substance use. The results are 8
consistent with studies on caffeine consumption during adolescence and its association 9
with future substance misuse (Barrense‐Dias et al., 2016; Leal & Jackson, 2019), 10
supporting a higher risk of caffeinated soda consumption in childhood, particularly 11
regarding vulnerability to future substance misuse. Such information is invaluable, as 12
caffeinated soda is incomparably the most common medium for caffeine consumption in 13
childhood, and the risk of substance misuse should be detected before adolescence, the 14
most common period of substance use onset. 15
Our results have important implications for public health recommendations, as 16
our study provides novel insight into the neurobehavioral correlates of caffeinated soda 17
consumption in children, which has rarely been evaluated. At the same time, a few 18
limitations of our research and future directions for further investigations should be 19
discussed. First, we encountered a substantial number of samples with missing data, 20
which led to their exclusion from the analyses (Fig. S1). We found significant differences 21
between included and excluded samples in terms of variables such as family income and 22
parental education (Table S1). As a result, there is a possibility that the excluded data is 23
19
missing not at random, potentially influencing our findings. Although we supported the 1
robustness of our main results by applying statistical methods that could control for other 2
confounding variables and also by examining the findings using larger samples through 3
additional analyses, future studies could benefit from employing imputation methods 4
(Saragosa-Harris et al., 2022; Woods et al., 2023) or other techniques to investigate the 5
potential impact of missing data on the results. Second, we did not perform functional 6
connectivity analyses or multivoxel pattern analyses, which may have provided additional 7
insight into the effects of caffeinated soda intake. Third, as the 9–10-year-old children in 8
this study had not yet started other substances except for alcohol, such as tobacco or 9
marijuana, future work using the longitudinal 10-year follow-up data of the ABCD study 10
should examine whether frequent consumption of caffeinated soda is associated with 11
alcohol or other substance misuse. Fourth, we acknowledge that multiple variables other 12
than caffeinated soda intake may mediate the relationship between neurobehavioral risk 13
factors and future alcohol use; thus, extensive investigation of how caffeinated soda 14
intake interacts with other SUD risk/protective factors is needed in the future. In addition, 15
the effects of acute caffeine intake could have influenced the task performance of the 16
soda drinking groups (Graczyk et al., 2018). Thus, future studies investigating the 17
longitudinal effects of caffeinated soda intake should control for the acute caffeine 18
consumption of children. Lastly, the ABCD dataset included only a small set of measures 19
of food/drink consumption. We relied on a self-report measure consisting of a single item 20
to assess caffeinated soda intake. For a more comprehensive understanding of the risks 21
associated with caffeinated soda intake, it is critical for future research to integrate more 22
objective and detailed assessments of soda consumption. Moreover, it is essential to 23
20
differentiate the effects of caffeine and sweeteners by including more diverse measures of 1
caffeine or sugar consumption. 2
In conclusion, our results revealed the potential risks of caffeinated soda 3
consumption in children by investigating the associations between caffeinated soda 4
consumption and risk factors for SUDs and examining the ability of caffeinated soda 5
consumption to predict future alcohol sipping, using the large ABCD dataset. While 6
previous research on the side effects of caffeinated soda consumption has been limited to 7
negative physical consequences, the present results strongly suggest that caffeinated soda 8
drinking in children is also associated with altered neurobehavioral function and can 9
predict alcohol sipping after 12 months. Our study further suggests a strong need to 10
develop evidence-based recommendations for caffeinated soda consumption in minors 11
(Temple, 2018), as there is no consensus on a safe dose of caffeinated soda in children, 12
and some children are at higher risk of adverse events from frequent caffeinated soda 13
intake. Further clarification on the causal relationships and neuro-developmental 14
evidence are needed to determine whether caffeinated soda is a warning sign for future 15
substance misuse and whether it induces neurobehavioral impairments in children. 16
17
Acknowledgements 18
We thank the families and children who participated, and continue to participate, in the 19
ABCD study, as well as staff at the study sites who are involved in data collection and 20
curating the data release. Data used in the preparation of this article were obtained from 21
the Adolescent Brain Cognitive Development (ABCD) Study (https://abcdstudy.org), 22
held in the NIMH Data Archive (NDA). This is a multisite, longitudinal study designed 23
21
to recruit more than 10,000 children age 9-10 and follow them over 10 years into early 1
adulthood. The ABCD Study® is supported by the National Institutes of Health and 2
additional federal partners under award numbers U01DA041048, U01DA050989, 3
U01DA051016, U01DA041022, U01DA051018, U01DA051037, U01DA050987, 4
U01DA041174, U01DA041106, U01DA041117, U01DA041028, U01DA041134, 5
U01DA050988, U01DA051039, U01DA041156, U01DA041025, U01DA041120, 6
U01DA051038, U01DA041148, U01DA041093, U01DA041089, U24DA041123, 7
U24DA041147. A full list of supporters is available at https://abcdstudy.org/federal-8
partners.html. A listing of participating sites and a complete listing of the study 9
investigators can be found at https://abcdstudy.org/consortium_members/. ABCD 10
consortium investigators designed and implemented the study and/or provided data but 11
did not necessarily participate in analysis or writing of this report. This manuscript 12
reflects the views of the authors and may not reflect the opinions or views of the NIH or 13
ABCD consortium investigators. The ABCD data repository grows and changes over 14
time. The ABCD data used in this report came from [NIMH Data Archive Digital Object 15
Identifier (10.15154/1503209)]. DOIs can be found at 16
http://dx.doi.org/10.15154/1503209. 17
18
Declaration of interest 19
The authors declare no competing interests. 20
22
References 1
2
3
Aron, A. R., Robbins, T. W., & Poldrack, R. A. (2014). Inhibition and the right inferior 4
frontal cortex: one decade on. Trends in Cognitive Sciences, 18(4), 177–185. 5
https://doi.org/10.1016/j.tics.2013.12.003 6
Arria, A. M., Caldeira, K. M., Kasperski, S. J., Vincent, K. B., Griffiths, R. R., & 7
O’Grady, K. E. (2011). Energy Drink Consumption and Increased Risk for Alcohol 8
Dependence. Alcoholism: Clinical and Experimental Research, 35(2), 365–375. 9
https://doi.org/10.1111/j.1530-0277.2010.01352.x 10
Auchter, A. M., Mejia, M. H., Heyser, C. J., Shilling, P. D., Jernigan, T. L., Brown, S. A., 11
Tapert, S. F., & Dowling, G. J. (2018). A Description of the ABCD Organizational 12
Structure and Communication Framework. Developmental Cognitive Neuroscience, 13
32, 8–15. https://doi.org/10.1016/j.dcn.2018.04.003 14
Baldo, J. V., & Dronkers, N. F. (2006). The Role of Inferior Parietal and Inferior Frontal 15
Cortex in Working Memory. Neuropsychology, 20(5), 529–538. 16
https://doi.org/10.1037/0894-4105.20.5.529 17
Barch, D. M., Albaugh, M. D., Avenevoli, S., Chang, L., Clark, D. B., Glantz, M. D., 18
Hudziak, J. J., Jernigan, T. L., Tapert, S. F., Yurgelun-Todd, D., Alia-Klein, N., 19
Potter, A. S., Paulus, M. P., Prouty, D., Zucker, R. A., & Sher, K. J. (2018). 20
Demographic, physical and mental health assessments in the adolescent brain and 21
cognitive development study: Rationale and description. Developmental Cognitive 22
Neuroscience, 32(Biol. Psychol. 124 2017), 55–66. 23
https://doi.org/10.1016/j.dcn.2017.10.010 24
Barrense‐Dias, Y., Berchtold, A., Akre, C., & Surís, J. (2016). Consuming energy drinks 25
at the age of 14 predicted legal and illegal substance use at 16. Acta Paediatrica, 26
105(11), 1361–1368. https://doi.org/10.1111/apa.13543 27
Barry, A. E., King, J., Sears, C., Harville, C., Bondoc, I., & Joseph, K. (2016). 28
Prioritizing Alcohol Prevention: Establishing Alcohol as the Gateway Drug and 29
Linking Age of First Drink With Illicit Drug Use. Journal of School Health, 86(1), 30
31–38. https://doi.org/10.1111/josh.12351 31
Bartra, O., McGuire, J. T., & Kable, J. W. (2013). The valuation system: A coordinate-32
based meta-analysis of BOLD fMRI experiments examining neural correlates of 33
subjective value. NeuroImage, 76, 412–427. 34
https://doi.org/10.1016/j.neuroimage.2013.02.063 35
23
Bechara, A., Berridge, K. C., Bickel, W. K., Morón, J. A., Williams, S. B., & Stein, J. S. 1
(2019). A Neurobehavioral Approach to Addiction: Implications for the Opioid 2
Epidemic and the Psychology of Addiction. Psychological Science in the Public 3
Interest, 20(2), 96–127. https://doi.org/10.1177/1529100619860513 4
Berridge, K. C., & Robinson, T. E. (2016). Liking, Wanting, and the Incentive-5
Sensitization Theory of Addiction. American Psychologist, 71(8), 670–679. 6
https://doi.org/10.1037/amp0000059 7
Bjork, J. M., Knutson, B., & Hommer, D. W. (2008). Incentive‐elicited striatal activation 8
in adolescent children of alcoholics. Addiction, 103(8), 1308–1319. 9
https://doi.org/10.1111/j.1360-0443.2008.02250.x 10
Bjork, J. M., Straub, L. K., Provost, R. G., & Neale, M. C. (2017). The ABCD Study of 11
Neurodevelopment: Identifying Neurocircuit Targets for Prevention and Treatment of 12
Adolescent Substance Abuse. Current Treatment Options in Psychiatry, 4(2), 196–13
209. https://doi.org/10.1007/s40501-017-0108-y 14
Blum, K., Braverman, E. R., Holder, J. M., Lubar, J. F., Monastra, V. J., Miller, D., 15
Lubar, J. O., Chen, T. J. H., & Comings, D. E. (2000). The Reward Deficiency 16
Syndrome: A Biogenetic Model for the Diagnosis and Treatment of Impulsive, 17
Addictive and Compulsive Behaviors. Journal of Psychoactive Drugs, 32(sup1), 1–18
112. https://doi.org/10.1080/02791072.2000.10736099 19
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and 20
affective responses to impending reward and punishment: The BIS/BAS Scales. 21
Journal of Personality and Social Psychology, 67(2), 319. 22
https://doi.org/10.1037/0022-3514.67.2.319 23
Casey, B. J., Cannonier, T., Conley, M. I., Cohen, A. O., Barch, D. M., Heitzeg, M. M., 24
Soules, M. E., Teslovich, T., Dellarco, D. V., Garavan, H., Orr, C. A., Wager, T. D., 25
Banich, M. T., Speer, N. K., Sutherland, M. T., Riedel, M. C., Dick, A. S., Bjork, J. 26
M., Thomas, K. M., … Workgroup, the A. I. A. (2018). The Adolescent Brain 27
Cognitive Development (ABCD) study: Imaging acquisition across 21 sites. 28
Developmental Cognitive Neuroscience, 32(Biol. Psychiatry 69 2011), 43–54. 29
https://doi.org/10.1016/j.dcn.2018.03.001 30
Castellanos‐Ryan, N., Rubia, K., & Conrod, P. J. (2011). Response Inhibition and 31
Reward Response Bias Mediate the Predictive Relationships Between Impulsivity and 32
Sensation Seeking and Common and Unique Variance in Conduct Disorder and 33
Substance Misuse. Alcoholism: Clinical and Experimental Research, 35(1), 140–155. 34
https://doi.org/10.1111/j.1530-0277.2010.01331.x 35
Chambers, R. A., Taylor, J. R., & Potenza, M. N. (2003). Developmental Neurocircuitry 36
of Motivation in Adolescence: A Critical Period of Addiction Vulnerability. American 37
Journal of Psychiatry, 160(6), 1041–1052. https://doi.org/10.1176/appi.ajp.160.6.1041 38
24
Collins, S. L., & Izenwasser, S. (2004). Chronic nicotine differentially alters cocaine-1
induced locomotor activity in adolescent vs. adult male and female rats. 2
Neuropharmacology, 46(3), 349–362. 3
https://doi.org/10.1016/j.neuropharm.2003.09.024 4
Cortese, S., Kelly, C., Chabernaud, C., Proal, E., Martino, A. D., Milham, M. P., & 5
Castellanos, F. X. (2012). Toward Systems Neuroscience of ADHD: A Meta-Analysis 6
of 55 fMRI Studies. American Journal of Psychiatry, 169(10), 1038–1055. 7
https://doi.org/10.1176/appi.ajp.2012.11101521 8
Cyders, M. A., Flory, K., Rainer, S., & Smith, G. T. (2009). The role of personality 9
dispositions to risky behavior in predicting first‐year college drinking. Addiction, 10
104(2), 193–202. https://doi.org/10.1111/j.1360-0443.2008.02434.x 11
Debenham, J., Birrell, L., Champion, K., Lees, B., Yücel, M., & Newton, N. (2021). 12
Neuropsychological and neurophysiological predictors and consequences of cannabis 13
and illicit substance use during neurodevelopment: a systematic review of longitudinal 14
studies. The Lancet Child & Adolescent Health, 5(8), 589–604. 15
https://doi.org/10.1016/s2352-4642(21)00051-1 16
Demidenko, M. I., Huntley, E. D., Jahn, A., Thomason, M. E., Monk, C. S., & Keating, 17
D. P. (2020). Cortical and subcortical response to the anticipation of reward in high 18
and average/low risk-taking adolescents. Developmental Cognitive Neuroscience, 44, 19
100798. https://doi.org/10.1016/j.dcn.2020.100798 20
Destrieux, C., Fischl, B., Dale, A., & Halgren, E. (2010). Automatic parcellation of 21
human cortical gyri and sulci using standard anatomical nomenclature. NeuroImage, 22
53(1), 1–15. https://doi.org/10.1016/j.neuroimage.2010.06.010 23
Ettinger, S. (2022). Diet, Gut Microbiome, and Cognitive Decline. Current Nutrition 24
Reports, 1–10. https://doi.org/10.1007/s13668-022-00435-y 25
Garavan, H., Bartsch, H., Conway, K., Decastro, A., Goldstein, R., Heeringa, S., 26
Jernigan, T., Potter, A., Thompson, W., & Zahs, D. (2018). Recruiting the ABCD 27
Sample: Design Considerations and Procedures. Developmental Cognitive 28
Neuroscience, 32(Journal of Studies on Alcohol and Drugs 76 2015), 16–22. 29
https://doi.org/10.1016/j.dcn.2018.04.004 30
Graczyk, A. M., Ziegler, A. M., Bendlin, A., Sion, T., Vattana, K., & Temple, J. L. 31
(2018). Effects of Caffeine Administration on Reaction Time, Attention, and 32
Inhibitory Control in Children and Adolescents. Journal of Cognitive Enhancement, 33
2(3), 276–286. https://doi.org/10.1007/s41465-018-0074-3 34
Hart, H., Radua, J., Nakao, T., Mataix-Cols, D., & Rubia, K. (2013). Meta-analysis of 35
Functional Magnetic Resonance Imaging Studies of Inhibition and Attention in 36
Attention-deficit/Hyperactivity Disorder: Exploring Task-Specific, Stimulant 37
25
Medication, and Age Effects. JAMA Psychiatry, 70(2), 185–198. 1
https://doi.org/10.1001/jamapsychiatry.2013.277 2
Hou, F., Lesmes, L. A., Kim, W., Gu, H., Pitt, M. A., Myung, J. I., & Lu, Z.-L. (2016). 3
Evaluating the performance of the quick CSF method in detecting contrast sensitivity 4
function changes. Journal of Vision, 16(6), 18. https://doi.org/10.1167/16.6.18 5
Jackson, K. M., Barnett, N. P., Colby, S. M., & Rogers, M. L. (2015). The Prospective 6
Association Between Sipping Alcohol by the Sixth Grade and Later Substance Use. 7
Journal of Studies on Alcohol and Drugs, 76(2), 212–221. 8
https://doi.org/10.15288/jsad.2015.76.212 9
James, J. E., Kristjánsson, Á. L., & Sigfúsdóttir, I. D. (2011). Adolescent substance use, 10
sleep, and academic achievement: Evidence of harm due to caffeine. Journal of 11
Adolescence, 34(4), 665–673. https://doi.org/10.1016/j.adolescence.2010.09.006 12
Jordan, C. J., & Andersen, S. L. (2017). Sensitive periods of substance abuse: Early risk 13
for the transition to dependence. Developmental Cognitive Neuroscience, 25, 29–44. 14
https://doi.org/10.1016/j.dcn.2016.10.004 15
Kandel, D. (1975). Stages in adolescent involvement in drug use. Science, 190(4217), 16
912–914. https://doi.org/10.1126/science.1188374 17
Kandel, D. B., & Yamaguchi, K. (2002). Stages and Pathways of Drug Involvement. 65–18
89. https://doi.org/10.1017/cbo9780511499777.006 19
Kerns, J. G., Cohen, J. D., III, A. W. M., Cho, R. Y., Stenger, V. A., & Carter, C. S. 20
(2004). Anterior Cingulate Conflict Monitoring and Adjustments in Control. Science, 21
303(5660), 1023–1026. https://doi.org/10.1126/science.1089910 22
Knutson, B., Westdorp, A., Kaiser, E., & Hommer, D. (2000). FMRI Visualization of 23
Brain Activity during a Monetary Incentive Delay Task. NeuroImage, 12(1), 20–27. 24
https://doi.org/10.1006/nimg.2000.0593 25
Kristjansson, A. L., Kogan, S. M., Mann, M. J., Smith, M. L., Juliano, L. M., Lilly, C. L., 26
& James, J. E. (2018). Does early exposure to caffeine promote smoking and alcohol 27
use behavior? A prospective analysis of middle school students. Addiction, 113(9), 28
1706–1713. https://doi.org/10.1111/add.14261 29
Kristjansson, A. L., Mann, M. J., Smith, M. L., Kogan, S. M., Lilly, C. L., & James, J. E. 30
(2022). Caffeine consumption and onset of alcohol use among early adolescents. 31
Preventive Medicine, 163, 107208. https://doi.org/10.1016/j.ypmed.2022.107208 32
Kunin, D., Gaskin, S., Rogan, F., Smith, B. R., & Amit, Z. (2000). Caffeine promotes 33
ethanol drinking in rats Examination using a limited-access free choice paradigm. 34
Alcohol, 21(3), 271–277. https://doi.org/10.1016/s0741-8329(00)00101-4 35
26
Leal, W. E., & Jackson, D. B. (2019). The role of energy drink consumption in the 1
intention to initiate marijuana use among adolescents. Addictive Behaviors, 2
93(Postgraduate Medicine 127 3 2015), 240–245. 3
https://doi.org/10.1016/j.addbeh.2019.02.008 4
Lees, B., Garcia, A. M., Debenham, J., Kirkland, A. E., Bryant, B. E., Mewton, L., & 5
Squeglia, L. M. (2021). Promising vulnerability markers of substance use and misuse: 6
A review of human neurobehavioral studies. Neuropharmacology, 187, 108500. 7
https://doi.org/10.1016/j.neuropharm.2021.108500 8
Lisdahl, K. M., Sher, K. J., Conway, K. P., Gonzalez, R., Ewing, S. W. F., Nixon, S. J., 9
Tapert, S., Bartsch, H., Goldstein, R. Z., & Heitzeg, M. (2018). Adolescent brain 10
cognitive development (ABCD) study: Overview of substance use assessment 11
methods. Developmental Cognitive Neuroscience, 32, 80–96. 12
https://doi.org/10.1016/j.dcn.2018.02.007 13
Lisdahl, K. M., Tapert, S., Sher, K. J., Gonzalez, R., Nixon, S. J., Ewing, S. W. F., 14
Conway, K. P., Wallace, A., Sullivan, R., Hatcher, K., Kaiver, C., Thompson, W., 15
Reuter, C., Bartsch, H., Wade, N. E., Jacobus, J., Albaugh, M. D., Allgaier, N., 16
Anokhin, A. P., … Consortium, A. (2021). Substance use patterns in 9-10 year olds: 17
Baseline findings from the adolescent brain cognitive development (ABCD) study. 18
Drug and Alcohol Dependence, 108946. 19
https://doi.org/10.1016/j.drugalcdep.2021.108946 20
Logan, G. D., Cowan, W. B., & Davis, K. A. (1984). On the ability to inhibit simple and 21
choice reaction time responses: A model and a method. Journal of Experimental 22
Psychology: Human Perception and Performance, 10(2), 276–291. 23
https://doi.org/10.1037/0096-1523.10.2.276 24
Luciana, M., Bjork, J. M., Nagel, B., Barch, D. M., Gonzalez, R., Nixon, S. J., & Banich, 25
M. T. (2018). Adolescent Neurocognitive Development and Impacts of Substance 26
Use: Overview of the Adolescent Brain Cognitive Development (ABCD) Baseline 27
Neurocognition Battery. Developmental Cognitive Neuroscience, 32, 67–79. 28
https://doi.org/10.1016/j.dcn.2018.02.006 29
Luijten, M., Machielsen, M. W. J., Veltman, D. J., Hester, R., Haan, L. de, & Franken, I. 30
H. A. (2014). Systematic review of ERP and fMRI studies investigating inhibitory 31
control and error processing in people with substance dependence and behavioural 32
addictions. Journal of Psychiatry and Neuroscience, 39(3), 149–169. 33
https://doi.org/10.1503/jpn.130052 34
Luijten, M., Schellekens, A. F., Kühn, S., Machielse, M. W. J., & Sescousse, G. (2017). 35
Disruption of Reward Processing in Addiction : An Image-Based Meta-analysis of 36
Functional Magnetic Resonance Imaging Studies. JAMA Psychiatry, 74(4), 387. 37
https://doi.org/10.1001/jamapsychiatry.2016.3084 38
27
Magid, V., & Colder, C. R. (2007). The UPPS Impulsive Behavior Scale: Factor structure 1
and associations with college drinking. Personality and Individual Differences, 43(7), 2
1927–1937. https://doi.org/10.1016/j.paid.2007.06.013 3
Marmorstein, N. R. (2018). Investigating associations between caffeinated beverage 4
consumption and later alcohol consumption among early adolescents. Addictive 5
Behaviors, 90(Psychopharmacology 203 2009), 362–368. 6
https://doi.org/10.1016/j.addbeh.2018.11.033 7
Martz, M. E., Hardee, J. E., Cope, L. M., McCurry, K. L., Soules, M., Zucker, R. A., & 8
Heitzeg, M. M. (2022). Nucleus Accumbens Response to Reward among Children 9
with a Family History of Alcohol Use Problems: Convergent Findings from the 10
ABCD Study® and Michigan Longitudinal Study. Brain Sciences, 12(7), 913. 11
https://doi.org/10.3390/brainsci12070913 12
Matthys, W., Vanderschuren, L. J. M. J., & Schutter, D. J. L. G. (2013). The 13
neurobiology of oppositional defiant disorder and conduct disorder: Altered 14
functioning in three mental domains. Development and Psychopathology, 25(1), 193–15
207. https://doi.org/10.1017/s0954579412000272 16
McTeague, L. M., Huemer, J., Carreon, D. M., Jiang, Y., Eickhoff, S. B., & Etkin, A. 17
(2017). Identification of Common Neural Circuit Disruptions in Cognitive Control 18
Across Psychiatric Disorders. American Journal of Psychiatry, 174(7), 676–685. 19
https://doi.org/10.1176/appi.ajp.2017.16040400 20
Miyake, E. R., & Marmorstein, N. R. (2015). Energy drink consumption and later alcohol 21
use among early adolescents. Addictive Behaviors, 43, 60–65. 22
https://doi.org/10.1016/j.addbeh.2014.12.009 23
Musial, M. P. M., Beck, A., Rosenthal, A., Charlet, K., Bach, P., Kiefer, F., Vollstädt-24
Klein, S., Walter, H., Heinz, A., & Rothkirch, M. (2023). Reward Processing in 25
Alcohol-Dependent Patients and First-Degree Relatives: Functional Brain Activity 26
During Anticipation of Monetary Gains and Losses. Biological Psychiatry, 93(6), 27
546–557. https://doi.org/10.1016/j.biopsych.2022.05.024 28
Nestor, L. J., & Ersche, K. D. (2023). Abnormal Brain Networks Related to Drug and 29
Nondrug Reward Anticipation and Outcome Processing in Stimulant Use Disorder: A 30
Functional Connectomics Approach. Biological Psychiatry: Cognitive Neuroscience 31
and Neuroimaging, 8(5), 560–571. https://doi.org/10.1016/j.bpsc.2022.08.014 32
Nestor, L. J., Ghahremani, D. G., Monterosso, J., & London, E. D. (2011). Prefrontal 33
hypoactivation during cognitive control in early abstinent methamphetamine-34
dependent subjects. Psychiatry Research: Neuroimaging, 194(3), 287–295. 35
https://doi.org/10.1016/j.pscychresns.2011.04.010 36
28
Newman, J. P., MacCoon, D. G., Vaughn, L. J., & Sadeh, N. (2005). Validating a 1
Distinction Between Primary and Secondary Psychopathy With Measures of Gray’s 2
BIS and BAS Constructs. Journal of Abnormal Psychology, 114(2), 319–323. 3
https://doi.org/10.1037/0021-843x.114.2.319 4
Ooi, J. Y., Wolfenden, L., Sutherland, R., Nathan, N., Oldmeadow, C., Mclaughlin, M., 5
Barnes, C., Hall, A., Vanderlee, L., & Yoong, S. L. (2022). A Systematic Review of 6
the Recent Consumption Levels of Sugar-Sweetened Beverages in Children and 7
Adolescents From the World Health Organization Regions With High Dietary–8
Related Burden of Disease. Asia Pacific Journal of Public Health, 34(1), 11–24. 9
https://doi.org/10.1177/10105395211014642 10
Owen, A. M., McMillan, K. M., Laird, A. R., & Bullmore, E. (2005). N‐back working 11
memory paradigm: A meta‐analysis of normative functional neuroimaging studies. 12
Human Brain Mapping, 25(1), 46–59. https://doi.org/10.1002/hbm.20131 13
Porciúncula, L. O., Sallaberry, C., Mioranzza, S., Botton, P. H. S., & Rosemberg, D. B. 14
(2013). The Janus face of caffeine. Neurochemistry International, 63(6), 594–609. 15
https://doi.org/10.1016/j.neuint.2013.09.009 16
Robinson, T. E., & Berridge, K. C. (1993). The neural basis of drug craving: An 17
incentive-sensitization theory of addiction. Brain Research Reviews, 18(3), 247–291. 18
https://doi.org/10.1016/0165-0173(93)90013-p 19
SanMiguel, N., López-Cruz, L., Müller, C. E., Salamone, J. D., & Correa, M. (2019). 20
Caffeine modulates voluntary alcohol intake in mice depending on the access 21
conditions: Involvement of adenosine receptors and the role of individual differences. 22
Pharmacology Biochemistry and Behavior, 186, 172789. 23
https://doi.org/10.1016/j.pbb.2019.172789 24
Saragosa-Harris, N. M., Chaku, N., MacSweeney, N., Williamson, V. G., Scheuplein, M., 25
Feola, B., Cardenas-Iniguez, C., Demir-Lira, E., McNeilly, E. A., Huffman, L. G., 26
Whitmore, L., Michalska, K. J., Damme, K. S., Rakesh, D., & Mills, K. L. (2022). A 27
practical guide for researchers and reviewers using the ABCD Study and other large 28
longitudinal datasets. Developmental Cognitive Neuroscience, 55(Am. J. Public 29
Health 100 Suppl 1 2010), 101115. https://doi.org/10.1016/j.dcn.2022.101115 30
Sharma, L., Markon, K. E., & Clark, L. A. (2014). Toward a Theory of Distinct Types of 31
“Impulsive” Behaviors: A Meta-Analysis of Self-Report and Behavioral Measures. 32
Psychological Bulletin, 140(2), 374–408. https://doi.org/10.1037/a0034418 33
Shenhav, A., Cohen, J. D., & Botvinick, M. M. (2016). Dorsal anterior cingulate cortex 34
and the value of control. Nature Neuroscience, 19(10), 1286–1291. 35
https://doi.org/10.1038/nn.4384 36
29
Solnick, S. J., & Hemenway, D. (2013). Soft drinks, aggression and suicidal behaviour in 1
US high school students. International Journal of Injury Control and Safety 2
Promotion, 21(3), 266–273. https://doi.org/10.1080/17457300.2013.815631 3
Suglia, S. F., Solnick, S., & Hemenway, D. (2013). Soft Drinks Consumption Is 4
Associated with Behavior Problems in 5-Year-Olds. The Journal of Pediatrics, 5
163(5), 1323–1328. https://doi.org/10.1016/j.jpeds.2013.06.023 6
Temple, J. L. (2009). Caffeine use in children: What we know, what we have left to learn, 7
and why we should worry. Neuroscience & Biobehavioral Reviews, 33(6), 793–806. 8
https://doi.org/10.1016/j.neubiorev.2009.01.001 9
Temple, J. L. (2018). Trends, Safety, and Recommendations For Caffeine Use in 10
Children and Adolescents. Journal of the American Academy of Child & Adolescent 11
Psychiatry, 58(1), 36–45. https://doi.org/10.1016/j.jaac.2018.06.030 12
Temple, J. L., Bernard, C., Lipshultz, S. E., Czachor, J. D., Westphal, J. A., & Mestre, M. 13
A. (2017). The Safety of Ingested Caffeine: A Comprehensive Review. Frontiers in 14
Psychiatry, 8, 80. https://doi.org/10.3389/fpsyt.2017.00080 15
Tibshirani, R. (1996). Regression Shrinkage and Selection Via the Lasso. Journal of the 16
Royal Statistical Society: Series B (Methodological), 58(1), 267–288. 17
https://doi.org/10.1111/j.2517-6161.1996.tb02080.x 18
Tulsky, D. S., Carlozzi, N. E., Chevalier, N., Espy, K. A., Beaumont, J. L., & Mungas, D. 19
(2013). V. NIH Toolbox Cognition Battery (CB): measuring working memory. 20
Monographs of the Society for Research in Child Development, 78(4), 70–87. 21
https://doi.org/10.1111/mono.12035 22
Vanyukov, M. M., Tarter, R. E., Kirillova, G. P., Kirisci, L., Reynolds, M. D., Kreek, M. 23
J., Conway, K. P., Maher, B. S., Iacono, W. G., Bierut, L., Neale, M. C., Clark, D. B., 24
& Ridenour, T. A. (2012). Common liability to addiction and “gateway hypothesis”: 25
Theoretical, empirical and evolutionary perspective. Drug and Alcohol Dependence, 26
123, S3–S17. https://doi.org/10.1016/j.drugalcdep.2011.12.018 27
Veltman, D. J., Rombouts, S. A. R. B., & Dolan, R. J. (2003). Maintenance versus 28
manipulation in verbal working memory revisited: an fMRI study. NeuroImage, 18(2), 29
247–256. https://doi.org/10.1016/s1053-8119(02)00049-6 30
Volkow, N. D., Koob, G., & Baler, R. (2015). Biomarkers in Substance Use Disorders. 31
ACS Chemical Neuroscience, 6(4), 522–525. 32
https://doi.org/10.1021/acschemneuro.5b00067 33
Watts, A. L., Wood, P. K., Jackson, K. M., Lisdahl, K. M., Heitzeg, M. M., Gonzalez, R., 34
Tapert, S. F., Barch, D. M., & Sher, K. J. (2021). Incipient alcohol use in childhood: 35
Early alcohol sipping and its relations with psychopathology and personality. 36
30
Development and Psychopathology, 33(4), 1338–1350. 1
https://doi.org/10.1017/s0954579420000541 2
Woods, A. D., Gerasimova, D., Dusen, B. V., Nissen, J., Bainter, S., Uzdavines, A., 3
Davis‐Kean, P. E., Halvorson, M., King, K. M., Logan, J. A. R., Xu, M., Vasilev, M. 4
R., Clay, J. M., Moreau, D., Joyal‐Desmarais, K., Cruz, R. A., Brown, D. M. Y., 5
Schmidt, K., & Elsherif, M. M. (2023). Best practices for addressing missing data 6
through multiple imputation. Infant and Child Development. 7
https://doi.org/10.1002/icd.2407 8
Yücel, M., Lubman, D. I., Harrison, B. J., Fornito, A., Allen, N. B., Wellard, R. M., 9
Roffel, K., Clarke, K., Wood, S. J., Forman, S. D., & Pantelis, C. (2007). A combined 10
spectroscopic and functional MRI investigation of the dorsal anterior cingulate region 11
in opiate addiction. Molecular Psychiatry, 12(7), 691–702. 12
https://doi.org/10.1038/sj.mp.4001955 13
Zeng, J., You, L., Yang, F., Luo, Y., Yu, S., Yan, J., Liu, M., & Yang, X. (2023). A meta‐14
analysis of the neural substrates of monetary reward anticipation and outcome in 15
alcohol use disorder. Human Brain Mapping, 44(7), 2841–2861. 16
https://doi.org/10.1002/hbm.26249 17
Zuckerman, M. (2001). International Encyclopedia of the Social & Behavioral Sciences. 18
Developmental Psychology: Personality Development: Motivational/Emotional 19
Psychology: Overview, Classification, and Methods: Personality Psychology and Self: 20
Personality: Risk: Article Titles: S, 13905–13908. https://doi.org/10.1016/b0-08-21
043076-7/01772-1 22
23
24
31
1
2
Figure 1. An overview of the analytical approach 3
4
A. The assessments used to capture the three main categories of neurobehavioral risk 5
factors: working memory, impulsivity, and reward processing. Working memory was 6
measured from the performance of list sorting test, card sort test, and 2-back in the 7
emotional N-Back Task (EN-back) performed inside a functional magnetic resonance 8
imaging (fMRI) scanner. The fMRI data from the EN-back task were analyzed based on 9
the contrast of 2-back versus 0-back, using regions in the frontoparietal and fronto-10
thalamic network as the regions of interest (ROIs). Impulsivity was measured by 11
Urgency-Premeditation-Perseverance-Sensation Seeking-Positive Urgency (UPPS-P), 12
and Behavioral Inhibition System/ Behavioral Activation System (BIS/BAS), and the 13
stop signal reaction time (SSRT) during the stop signal task (SST) performed inside the 14
fMRI scanner. The fMRI data from the SST were analyzed based on the contrast of 15
correct stop versus correct go in lateral prefrontal cortex, anterior cingulate cortex (ACC), 16
and striatum ROIs. Reward processing was measured behaviorally by the cash choice 17
task, and also by comparing the success rate of reward versus neutral conditions during 18
the monetary incentive delay (MID) task performed inside the fMRI scanner. The fMRI 19
data from the MID task were analyzed based on the contrast of reward anticipation versus 20
neutral anticipation, using regions in the ventral striatum and medial frontal cortex as the 21
ROIs. The ROIs were selected based on the Destrieux atlas (Destrieux et al., 2010). See 22
"Methods" for more details on the variables. B. Diagram of the research aims. First, the 23
associations between the risk factors for substance use disorders (SUDs) and caffeinated 24
soda intake were examined cross-sectionally using the baseline ABCD data. Then, we 25
assessed whether caffeinated soda intake would predict alcohol sipping after 12 months 26
while controlling other risk factors of SUDs. 27
28
32
1
2
3
Figure 2. Results of binomial LASSO regression predicting daily caffeinated soda intake 4
(daily-drinking group vs. non-drinking group) 5
6
A. Estimates of coefficients of the survived variables (x axis: standardized coefficient 7
estimates; y axis: predictor). The site variables were excluded for clarification (see 8
Supplementary Fig. S2 for results including all variables). Each dot indicates mean of 9
the coefficient, positive coefficient in red and negative coefficient in green. Each error 10
bar indicates 95% confidence interval. B. Distribution of the area under the curve (AUC) 11
values (left) and a representative receiver-operation characteristic (ROC) curve (right) for 12
the training and test datasets. C. Regions of interest (ROIs) identified as having 13
33
significant estimates of coefficients during the stop signal task (SST) and emotional N-1
Back Task (EN-back). 2
Abbreviations. LASSO, least absolute shrinkage and selection operator; BMI, body mass 3
index; BAS, Behavioral Activation System; UPPS-P, Urgency-Premeditation-4
Perseverance-Sensation Seeking-Positive Urgency; IPL, Inferior Parietal Lobule; IFG, 5
Inferior Frontal Gyrus; ACC, Anterior Cingulate Cortex. 6
7
8
9
34
1
Figure 3. Results of binomial LASSO regression predicting alcohol sipping after 12 2
months 3
A. Estimates of coefficients of the survived variables (x axis: coefficient estimates; y axis: 4
predictor). The site variables were excluded for clarification (see Supplementary Fig. S3 5
for results including all variables). Each dot indicates mean of the coefficient, positive 6
coefficient in red and negative coefficient in green. Each error bar indicates 95% 7
confidence interval. B. Distribution of the area under the curve (AUC) values (left) and a 8
representative receiver-operation characteristic (ROC) curve (right) for the training and test 9
datasets. C. Rate of alcohol sipping at 12-month follow-up in each group. The asterisk 10
indicates significance from the chi-square test (p < 0.05). Abbreviations. LASSO, least 11
absolute shrinkage and selection operator; BMI, body mass index; UPPS-P, Urgency-12
Premeditation-Perseverance-Sensation Seeking-Positive Urgency; EN-Back, emotional N-13
Back Task; MID, Monetary Incentive Delay Task; NAc, Nucleus Accumbens; OFC, 14
Orbitofrontal cortex. 15
16
