Participants

We recruited chronic cannabis users who were seeking treatment at the Substance Use Disorder Program, Federal University of São Paulo. To assess study eligibility, telephone screening interviews were administered to all potential participants. After enrolment in the study, participants were interviewed to provide a complete medical and psychiatric history. Although abstinence was encouraged, it was not required. All participants participate in a day-long interview in order to start treatment at UNIAD. The neuropsychological assessment was performed on the first day of the treatment.

The following inclusion criteria were used for chronic cannabis users: males and females, aged 18 to 55 years, with DSM-IV cannabis misuse or dependence (cannabis as the misuse substance of choice) as determined by Composite International Diagnostic Interview (CIDI). ^{Reference Robins, Wing, Wittchen, Helzer, Babor and Burke21} Exclusion criteria included current history of other DSM-IV Axis I disorders, with the exception of nicotine-related disorders as determined by CIDI; current use of psychoactive medications, history of head trauma with loss of consciousness for more than 5 min, intellectual disability or estimated IQ below 80, and non-correctable vision or hearing impairment.

Individuals in the control group qualified for this study if they were between the ages of 18 and 55, did not use any psychoactive substance, did not have a history of head trauma, and had no lifetime diagnosis of Axis I DSM-IV disorders. They had not used cannabis at all in the last 3 months and had been exposed no more than five times in their lifetime. The study protocol was approved by the local institutional review board. All participants were required to sign a written informed consent, in accordance with the Federal University of São Paulo review board.

Instruments

A detailed assessment was administered at enrolment consisting of a semi-directed protocol measuring psychosocial functioning, medical history and history of substance use. We used the modified Time Line Follow Back calendar (TLFB) ^{Reference Sobel and Sobel22} to obtain detailed information regarding type, amount, and frequency of substance exposure. Composite International Diagnostic Interviews were used to assess DSM-IV Axis I diagnosis. Urine analyses for tetrahydrocannabinol detection were performed on all participants on the same day as the interview to evaluate the reliability of what they reported in the interview. Tests were not performed to detect other substances.

The neuropsychological battery consisted of the following.

1. (a) Stroop Test – Victoria Version, composed of three cards, each one containing six rows of four items. ^{Reference Spreen and Strauss23} The first card contains 24 rectangles filled with colours (green, pink, blue and brown). The participant should name as quickly as possible the printed colours. The second card has nouns written in coloured words and the participant is required to name as quickly as possible the colour of the words, disregarding their verbal content (for example the participant is expected to respond green when the word ‘each’, written in green, is presented). On the third card, the coloured stimuli are colour names, so that the printed colour never corresponds to the colour name (for example ‘Brown’ is written in blue ink). This later task, thus, requires the individual to inhibit an automatic reading response and to produce a more effortful colour-naming response by exerting an inhibitory control.

2. (b) Wisconsin Card Sorting Test – version 64 (planning, executive functioning, conceptualisation, perseveration, and ability to form abstract concepts). ^{Reference Heaton, Chelune, Talley, Kay and Curtis24}

3. (c) Frontal Assessment Battery – FAB, a short neuropsychological instrument aimed at assessing executive functions. ^{Reference Dubois, Slachevsky, Litvan and Pillon25} The FAB consists of six subtests, each exploring functions related to frontal functioning: conceptualisation (similarities task), mental flexibility (phonological fluency task), motor programming (by means of Luria's motor series), sensitivity to interference (conflicting instructions task), inhibitory control (go/no- go task) and environmental autonomy (evaluation of prehension behaviour).

4. (d) Vocabulary and block design (Wechsler Adult Intelligence Scale – Revised (WAIS-R)) for IQ estimation. ^{Reference Wechsler26} Vocabulary is a WAIS-R subtest that evaluates the degree to which one has learned, been able to comprehend and verbally express vocabulary; there are 35 words. Block design is a subtest that measures visuospatial and motor skills. In the test the participant has to take blocks that have all white sides, all red sides, and red and white sides and arrange them according to a pattern. They are timed on this task and compared to a normative sample. Estimates of full-scale IQ were made using vocabulary and block design. ^{Reference Spreen and Strauss27}

Statistical analyses

For primary analysis, we divided the sample of chronic cannabis users into two groups: the early-onset group, who initiated use of cannabis before the age of 15; and the late-onset group, who initiated use of cannabis at the age of 15 or later. The mean age of first use was 16.6 (s.d. = 3.5) years. However, we chose 15 years as the cut-off in order to stress the impact of cannabis exposure in early adolescence as opposed to late adolescence. This was possible because of the large sample size (104 participants) and wide range of age of first use (8 to 30 years).

We compared the demographic characteristics of both the early- and late-onset groups with those of the controls using Student's t-test (for variables with normal distribution) or Mann–Whitney U-test. We used Student's t-test to compare age of first use, age starting daily use, pattern of use, years of daily use, estimation of life consumption, and days of abstinence in early-onset and late-onset groups.

Comparisons involving neuropsychological measures were performed using generalised linear model analysis of variance (ANOVA). In order to minimise the problem of multiple comparisons in neuropsychological analyses, we decided to select only the more relevant measures according to the literature for each test. For example, for the WCST, we limited analyses to two measures (number of correct categories and perseverative errors). The alpha level was set at P<0.05 (two-tailed) for all analyses.

Main findings

This study evaluated the neurocognitive functioning of early- and late-onset chronic cannabis users and controls. The early-onset, but not the late-onset, group performed poorly on selected executive tasks compared with controls. In addition, the early-onset group performed poorly on the FAB compared with both the late-onset and the control groups. Our findings indicate that early-onset chronic cannabis users are cognitively impaired compared with controls, suggesting that early exposure to cannabis is associated with more adverse effects on the brain. These results are in line with previous studies examining cognitive effects associated with early cannabis exposure. ^{Reference Ehrenreich, Rinn, Kunert, Moeller, Poser and Schilling28–Reference Pope, Gruber, Hudson, Huestis and Yurgelun-Todd30}

Fig. 1 Wisconsin Card Sorting Test number of categories in early-onset, late-onset and control groups.

Error bars: 95% confidence intervals.

Fig. 2 Wisconsin Card Sorting Test perseverative errors in early-onset, late-onset and control groups.

Error bars: 95% confidence intervals.

Fig. 3 Stroop Test (noun–colour) time in seconds in early-onset, late-onset and control groups.

Error bars: 95% confidence intervals.

Fig. 4 Frontal Assessment Battery (FAB) total score in early-onset, late-onset and control groups.

Error bars: 95% confidence intervals.

Implications of early cannabis use

In the past few decades, a steady increase in people seeking treatment for cannabis-related problems has been registered in Western countries, with an associated decrease in the age of first cannabis use. ^{Reference Degenhardt, Chiu, Sampson, Kessler, Anthony and Angermeyer31} Adolescence is a period in which the brain appears to be particularly vulnerable to the neurotoxic effects of cannabis. Mounting evidence has associated cannabis use with psychiatric morbidity, such as depression and psychosis, and the strength of this association appears to be highly influenced by age at first exposure. ^{Reference Stefanis, Delespaul, Henquet, Bakoula, Stefanis and Van Os32}

Cannabis exposure during adolescence may potentially be associated with long-lasting neuronal consequences. Cross-sectional and longitudinal structural brain imaging studies have revealed that the brain before age 15 is under a complex process of biological maturation, especially in the prefrontal cortex, which has peak grey matter volumes at age 11, which is then followed by rapid grey matter loss in late adolescence. ^{Reference Gogtay, Giedd, Lusk, Hayashi, Greenstein and Vaituzis33} During adolescence, rapid changes in neural organisation occur, including a reduction in the number of synapses in cortical and subcortical structures as well as changes in neurotransmitter and receptor levels. ^{Reference Teicher, Andersen and Hostetter34} It has been further suggested that these changes in central nervous system organisation may make the brain uniquely vulnerable to insults by substances of misuse. ^{Reference Slawecki and Ehlers35} Early cannabis use was found to be strongly associated with misuse and dependence during adolescence; thus, it may be both more ‘addicting’ and cause greater effects. ^{Reference Ehlers, Slutske, Gilder and Lau36} Ehrenreich et al evaluated a broad spectrum of attentional functions in 99 marijuana users, with a computerised battery. ^{Reference Ehrenreich, Rinn, Kunert, Moeller, Poser and Schilling28} They studied 48 participants exposed to cannabis before 16 years of age, 51 who first used cannabis after the age of 16, and 49 controls. Associations involving visual attention measures and age of first exposure to cannabis were found, suggesting that exposure to marijuana during adolescence may lead to persistent attentional deficits.

Taken together, those findings suggest that exposure to exogenous cannabinoids in critical periods of brain development, including early adolescence, may be more neurotoxic. In the postnatal brain, endocannabinoids act as retrograde messengers to regulate the function of many synapses. Furthermore, several maturational events appear to depend on the modulatory influence of the endocannabinoid system in these critical periods. ^{Reference Fernandez-Ruiz, Berrendero, Hernandez and Ramos37} Dysregulation of these fine-tuned developmental processes may underlie both cognitive impairments and psychopathology associated with early exposure to cannabis use. ^{Reference Jager and Ramsey19} Alternatively, the present findings may be accounted for by previous cognitive impairments in the early-onset group, which would make individuals in this group more prone to exposure to substances at an early age. This hypothesis, however, appears not to be supported by the present findings since the three groups were matched for estimated IQ and a measure of premorbid functioning assessed by vocabulary (the WAIS-R).

The early-onset group showed a particularly poor performance in the FAB, and revealed deficits in comparison with both the late-onset and the control groups. The FAB is a short neuropsychological battery aimed at assessing conceptualisation, mental flexibility, programming, sensitivity to interference, inhibitory control and environmental autonomy. Its results have been shown to significantly correlate with medial and dorsolateral prefrontal activity as measured by neurofunctional brain imaging studies. ^{Reference Guedj, Allali, Goetz, Le Ber, Volteau and Lacomblez38,Reference Slachevsky, Villalpando, Sarazin, Hahn-Barma, Pillon and Dubois39} Performance on the six subtests of the FAB provides a composite global score that evaluates the severity of the dysexecutive syndrome affecting both cognition and motor function. ^{Reference Dubois, Slachevsky, Litvan and Pillon25} In addition, the poor performance of the early-onset group on the Stroop test and perseverative responding (as measured by the WCST) suggest that exposure to cannabis in early adolescence may lead to lower mental flexibility. This could explain partially why chronic users find it difficult to maintain treatment and to discontinue substance use despite evident physical and psychological consequences.

Aharonovich et al suggest that mild cognitive impairments negatively affect retention in out-patient cognitive–behavioural therapy treatment for cocaine dependence. ^{Reference Aharonovich, Hasin, Brooks, Liu, Bisaga and Nunes40} Cognitive deficits, particularly in executive functioning, have been associated with poorer adherence to treatment in several studies. ^{Reference Lundqvist41,Reference Hall and Solowij42} In this sense, executive deficits appear to negatively affect treatment outcome and may represent a potentially important aspect for designing specific or target-orientated cognitive strategies for treatment of substance use disorders, especially for those who initiated use at an early age. ^{Reference Cunha and Novaes43}

We did not find differences in executive functioning performance when comparing the late-onset and control groups. Another important issue is whether starting cannabis use at a later age is inoffensive to cognition. It is possible that participants who start to use cannabis at a later age may use different neural networks and compensate for their deficits more than individuals who initiated the use of cannabis at an earlier stage of brain development. Chang et al showed that the brain network might be altered in some cannabis users but without an apparent change in cognition. ^{Reference Chang, Yakupov, Cloak and Ernst44} The long-term chronic effects of marijuana on the altered brain network may be reversible with prolonged abstinence. A plausible explanation for this is that the brain, in response to chronic cannabis use, appears to undergo neuroadaptation by accessing compensatory processes in order to maintain usual cognitive performance. ^{Reference Ponto45} Future studies should focus on neurocognitive tasks including decision-making; this is a core part of the executive dysfunction now understood to be one of the hallmark features of addiction that also includes impaired impulse dysregulation.

Limitations

The findings from the present study should be interpreted with caution while considering the following limitations. First, the cross-sectional study design does not allow us to make any conclusions about causality. Some executive deficits are explicitly found, but it is not possible to establish whether they are the consequences of cannabis use or were present before first use. Second, the present study evaluated cannabis users in a period of early abstinence. At least in part, the results may be related to substance residues or withdrawal symptoms. ^{Reference Bolla, Eldreth, Matochik and Cadet46} This effect was minimised because both the cannabis user groups (early and late onset) were matched for pattern of use, years of daily use, life consumption and days of abstinence. Third, these two groups were predominantly composed of males whereas the control group had equal numbers of males and females. However, the early- and late-onset groups were matched for gender and the early-onset group performed poorly on the FAB compared with both the late-onset and control groups. Yet, the results did not change when only males were analysed.

In conclusion, our findings suggest that early-onset, but not late-onset, chronic cannabis users exhibit executive deficits. Although the underlying mechanisms are not fully understood, cannabis exposure at an early age may have greater deleterious effects on neurocognitive functioning. Future longitudinal studies examining adolescents who initiated cannabis use recently may be especially informative