Cohort Profile: The Saguenay Youth Study (SYS)

Abstract

The Saguenay Youth Study (SYS) is a two-generational study of adolescents and their parents (n = 1029 adolescents and 962 parents) aimed at investigating the aetiology, early stages and trans-generational trajectories of common cardiometabolic and brain diseases. The ultimate goal of this study is to identify effective means for increasing healthy life expectancy. The cohort was recruited from the genetic founder population of the Saguenay Lac St Jean region of Quebec, Canada. The participants underwent extensive (15-h) phenotyping, including an hour-long recording of beat-by-beat blood pressure, magnetic resonance imaging of the brain and abdomen, and serum lipidomic profiling with LC-ESI-MS. All participants have been genome-wide genotyped (with ∼ 8 M imputed single nucleotide polymorphisms) and a subset of them (144 adolescents and their 288 parents) has been genome-wide epityped (whole blood DNA, Infinium HumanMethylation450K BeadChip). These assessments are complemented by a detailed evaluation of each participant in a number of domains, including cognition, mental health and substance use, diet, physical activity and sleep, and family environment. The data collection took place during 2003–12 in adolescents (full) and their parents (partial), and during 2012–15 in parents (full). All data are available upon request.

Why was the cohort set up?

The Saguenay Youth Study (SYS; [http://www.saguenay-youth-study.org]) is a population-based study of adolescents and their middle-aged parents. It is aimed at investigating the aetiology, early stages and trans-generational trajectories of common cardiometabolic and brain diseases. The ultimate goal is to identify effective means for increasing healthy life expectancy.¹

The design of the SYS was motivated by the following biomedical considerations: (i) many common cardiometabolic and brain diseases originate in utero; (ii) they involve interactions between adverse environments and vulnerability genes; (iii) many of these diseases emerge during adolescence and become established during middle-aged adulthood; and (iv) most of them are multi-systemic, affecting both the brain and the rest of the body.

Our main methodological considerations were: (i) genetics can be used to uncover aetiology and mechanistic pathways; (ii) emergence and trans-generational trajectories of disease phenotypes can be monitored through high- fidelity ‘intermediary’ (pre-clinical) phenotypes; (iii) multi- system (cardiovascular, metabolic and cognitive) and multi-level (environment, tissues and molecules) assessments of each participant are necessary to understand how these systems and levels interact as part of an integrated whole—the human body; (iv) studies of complex genetic traits, such as common cardiometabolic and brain diseases, benefit from reduced genetic and environmental heterogeneity; and (v) disease risk and early disease processes can be tagged by easily assessable and highly predictive genetic, epigenetic and molecular biomarkers.

Finally, the SYS was designed as a retrospective study of long-term outcomes of prenatal exposure to maternal cigarette smoking. The choice of this particular prenatal adversity reflected high prevalence of maternal cigarette smoking during pregnancy^2,3and the reports of long-term cardiometabolic^4–6and behavioural^7–10abnormalities in the exposed offspring. Maternal cigarette smoking during pregnancy was common; in Canada and the USA, for example, close to 40% of pregnant women smoked in the 60s and 70s. Over 40 years later, the proportion of pregnant women smoking during pregnancy is still high (12–16%). Importantly, this number has not been declining in recent years;^2,3 it is even higher (> 20%) in vulnerable groups, such as pregnant teenage girls and women with low education attainment.^3,7 Thus, a large proportion of the population has been and continues to be exposed prenatally to maternal cigarette smoking.

Who is in the cohort?

The SYS cohort includes 1029 adolescents and their 962 parents. The cohort was recruited via adolescents attending high schools in the Saguenay–Lac-Saint-Jean region of Quebec, Canada. The region is home to the largest genetic founder population in North America.^11–14 Both maternal and paternal grandparents of the adolescents were required to be of French-Canadian ancestry and born in the region; as such, all adolescents and their parents are of a single ethnicity [European (French) ancestry]. Half of the adolescents were exposed prenatally to maternal cigarette smoking. The cohort is family based (481 families), including only adolescents who have one or more siblings of similar age (12 to 18 years) and both biological parents of the French-Canadian origin born in the region. The data collection occurred in two waves. Wave 1 involved the recruitment and ‘complete assessment’ of all 1029 adolescents, as well as a ‘partial assessment’ of 962 parents. Wave 2 involved the ‘complete assessment’ of an available (and willing) subset of the parents (n = 664). Characteristics of the SYS cohort are provided in Tables 1 and 2, and in Tables S1–S4 (available as Supplementary data at IJE online).

Table 1.

Basic characteristics of participants

	Adolescents	Parents
	Wave 1	Wave 1	Wave 2
Measure		Count, mean ± SD or proportion	Count, mean ± SD or proportion
Number of participants	1029	962	664
Number of families	481	481	382
Age (years)	15.0 ± 1.8	43.3 ± 4.6	49.2 ± 5.0
Sex (%, males/females)	48/52	50/50	45/55
Household CAD income (%)
≤ $20,000	13	13	13
$30,000-$40,000	19	19	20
$50,000-$60,000	24	24	24
$70,000-$80,000	20	20	20
≥ $85,000	24	24	23
Education (%)
No high school	0	1	< 1
Some high school	1029	16	15
High school	0	51	53
College degree	0	19	19
Bachelor’s	0	9	8
Master’s or doctorate	0	3	4
Unknown	0	< 1	< 1

Table 2.

Wave 1 ‘complete’ assessment of adolescents (n = 1029): phenotyping domains and tools

Domain	Tool	Phenotypes
Brain	MRI	Global and regional volumes; cortical surface and thickness; MTR
Cognition	6-hour battery	PIQ and VIQ; memory; executive functioning, phonological and motor skills; social cognition
Mental health	DPS, GRIP	Epidemiological diagnoses; symptom counts
Substance use	GRIP	Cigarette smoking, cannabis, alcohol use, drug experimentation (age of initiation, past 30 days, binge drinking)
Personality	NEO-PI-R	Neuroticism, extroversion, openness, agreeableness, conscientiousness
Cardiovascular	Finometer	Beat-by-beat blood pressure, heart rate, stroke volume, total peripheral resistance at rest and in response to physical and mental challenges (52-min protocol)
Autonomic balance	Power spectral analysis	Low- and high-frequency powers of inter-beat interval and low-frequency power of blood pressure; sympathetic and parasympathetic tone
Body composition	Anthropometry, MRI, bioimpedance	Height, weight, circumferences, skinfolds; subcutaneous and visceral fat and muscle volumes; fat and muscle mass
Glucose/lipid metabolism	Blood	Glucose, insulin, cholesterol, HDL-cholesterol, triglycerides, leptin, C-reactive protein
Lipidomicsa	Blood (LC-ESI-MS)	∼ 700 lipid species
Hormones	Blood	Testosterone, estrogen, cortisol
Hormones	Saliva	Cortisol (before and after mental stress)
Genetic variation	Blood DNA	Illumina Human610-Quad BeadChip and HumanOmniExpress BeadChip; a total of 7 746 837 typed and imputed SNPs
Epigenetic variationb	Blood DNA	Infinium HumanMethylation450K BeadChip (> 485 000 CpGs)
Sexual maturation	PDS	Stages of pubertal development (Tanner stages)
Lifestyle	Lerner	Sleep, physical activity, extracurricular activities, sexuality, academic/vocational aspirations
Diet	24-hour food recall	Energy and nutrient intake
Family environment	FamEnvi	Stressful life events, financial difficulties, SES (family income, parental education)

MTR, magnetization transfer ratio; DPS, DISC Predictive Scales; GRIP, Groupe de Recherche sur l’Inadaptation Psychosociale, adolescent self-assessment of mental health and substance use developed for the SYS by J. R. Séguin based on validated National Longitudinal Survey of Children and Youth (NLSCY) and Quebec Longitudinal Study of Child Development (QLSCD)³¹ protocols; Lerner, adolescent self-assessment developed by Richard Lerner; PIQ, performance intelligence quotient; VIQ, verbal intelligence quotient; SES, socioeconomic status; PDS, Puberty Development Scale; HDL, high-density lipoprotein; CRP, C-reactive protein; LC-ESI-MS, liquid-chromatography electrospray-ionization mass spectrometry; Family Environment, questionnaire on family environment developed by the SYS team; NEO-PI, Neuroticism, Extraversion, Openness Personality Inventory.¹⁵

^aIn progress.

^bAssessed in a subset of 144 adolescents.

Recruitment

Wave 1 took place over a 10-year period (2003–12). Recruitment was conducted via high schools. During the 10-year period, our team made 28 visits to schools, contacting a total of 27 190 students (18 127 families). Of the 18 127 families, 5570 (33%) sent a response card; of these, 3269 families (59%) indicated their interest in the study and 2301 families (41%) declined further participation. Based on the inclusion (e.g. maternal smoking during pregnancy, two or more siblings per family) and exclusion [e.g. magnetic resonance imaging (MRI) contraindications] criteria,¹⁵a total of 1801 families (55% of the interested families) were eligible to participate in the study; a research nurse determined the eligibility via a structured telephone interview with the mother. From these, 481 families participated in the study. ‘Exposed’ adolescents were recruited first; we defined exposure as maternal cigarette smoking of at least 1 cigarette/day during the second trimester of gestation. The ‘non-exposed’ adolescents were matched to the ‘exposed’ adolescents by the level of maternal education and by the high school attended. Maternal cigarette smoking before and during pregnancy was ascertained retrospectively by a research nurse who conducted a structured telephone interview with the mother. We found ‘good’ agreement between the exposure status noted in the medical records at the time of pregnancy and the maternal report during the telephone interview (Kappa statistics = 0.69 ± 0.04; assessed in the first 260 SYS participants). Siblings were concordant for the exposure status in the majority of families (446/481 families; 93%).

Wave 2 took place during a 3-year period between 2012 and 2015. We contacted all 962 parents by mail and telephone. A total of 664 were interested in participating and underwent the complete assessment. The remaining 282 parents declined our invitation to participate and 16 parents were deceased. Parents who participated vs those who declined did not differ by age (P = 0.87), household income (P = 0.40) or education (P = 0.73), but they did differ by sex [the proportion of men was lower in the participating (46% men, 54% women) vs declining (57% men, 43% women) groups of parents, P = 0.001].

Further follow-up of both adolescents and parents (as young and ageing adults) is planned.

What has been measured?

The data collection took place in two waves. Wave 1 (2003–12) involved the recruitment and complete assessment of all 1029 adolescents (Table 1), as well as a partial assessment of all 962 parents (Table 2). Wave 2 (2012–15) involved the complete assessment of 664 parents (Table 1). Linkage to health registries is in progress.

Wave 1: complete assessment of adolescents

The ‘complete’ assessment of adolescents took place over several sessions (∼ 15 h in total) and included a number of domains (Table 2). Each adolescent provided a fasting (morning) blood sample. The key features of the phenotyping protocol were: (1) MRI of the brain and abdomen; (ii) 1-h cardiovascular assessment; and (iii) 6-h cognitive evaluation.

MRI of the brain and abdomen

(Figure 1): brain MRI was used to assess structural and volumetric features of the brain. Two types of magnetic resonance images were acquired: (i) T1-weighted images, which are used for deriving a number of anatomical features (e.g. global and regional volumes of grey and white matter, cortical thickness and surface area, and normalized MR intensities); and (ii) magnetization-transfer ratio (MTR), which provides insights into micro-structural properties of white matter, such as the relative content of myelin and axons. Abdominal MRI was used to quantify visceral and subcutaneous fat using a semi-automated method.¹⁶

Figure 1.

Magnetic resonance imaging of the brain (A) and abdomen (B). (A) T1-weighted images are used for deriving a number of anatomical features (e.g. global and regional volumes of grey and white matter, cortical thickness and surface area); and images of the magnetization-transfer ratio (MTR) provide insights into micro-structural properties of white matter, such as the relative content of myelin and axons (B); a set of heavily T1-weighted images are acquired during a single breath-hold and these are used to measure volumes of the kidneys (middle) and to segment the subcutaneous and visceral fat, respectively. A transverse slice (position at the level of the umbilicus) shows the native and segmented images (bottom); on the segmented image, subcutaneous and visceral fat are shown, respectively, in grey and white.

Cardiovascular assessment

(Figure 2) involved a 52-min protocol during which beat-by-beat blood pressure (BP) was monitored at rest and in response to simple physical and mental challenges mimicking daily-life activities, using Finometer (FNS Finapres, Amsterdam, The Netherlands). The Finometer monitors continuously finger blood flow¹⁷ and, by the reconstruction and level-correction of the finger blood-flow waveform, it derives beat-by-beat brachial systolic and diastolic BPs, as well as a number of haemodynamic parameters including inter-beat interval, stroke volume, cardiac output, ventricular ejection time, peripheral resistance, aortic impedance and aortic compliance. The Finometer has been approved as a reliable device for tracking BP in adults and children older than 6 years^18,19 and the precision of BP measurement by this device meets the American Association for the Advancement of Medical Instruments (AAAMI) requirements.^20,21

Figure 2.

Blood pressure and underlying haemodynamic parameters. Unadjusted 1-min means and standard errors of the mean of systolic blood pressure (A). Diastolic blood pressure (B), heart rate (C), stroke volume (D), cardiac output (E) and total peripheral resistance (F) are shown for girls and boys during a 52-min cardiovascular protocol that includes postural and mental challenges. Reprinted with permission from Syme et al.Arch Pediatr Adolesc Med 2009;163:818-25.

The beat-by-beat recordings of diastolic BP and inter-beat interval were analysed with power spectral analysis (PSA)²²to assess cardiovascular autonomic function, a key component of hypertension aetiology.^23–25 Low-frequency power of diastolic BP is considered a proxy of sympathetic modulation of vasomotor tone,^26–28 and high-frequency power of inter-beat interval is considered a proxy of parasympathetic modulation of cardiac chronotropic function.²⁹

Cognitive assessment

(Figure 3A) was carried out by trained psychometricians in two 3-h sessions (separated by a lunch break) on the same day. This assessment consisted of two types of tests: (i) standardized psychological instruments; and (ii) domain-specific tests. The standardized tests included the Wechsler Intelligence Scale for Children WISC-III (except Mazes), Woodcock–Johnson Achievement subtests (reading comprehension and arithmetic and a spelling test) and Children’s Memory Scale (Dot Locations and Stories). Domain-specific tests included tests of executive functions (self-ordered pointing, word fluency, resistance to interference), phonological skills and frequency-modulation auditory threshold, delayed auditory feedback, phonological learning, fine motor skills and motor coordination, number sense, and emotion and motivation (morphed facial expressions, repeated failure test and a gambling task).

Figure 3.

Visualization of the cognition (A) and DPS-based symptom (B) matrices as networks. (A) The cognition variables (n = 63; age-adjusted) are represented as nodes and connected by an edge if the correlation exceeds r = 0.3. In the electronic version, green (red) edges indicate positive (negative) correlations. Line thickness corresponds to strength of correlation, i.e. thicker lines represent stronger correlations. In the electronic version, colour of nodes represents the cognitive instrument used to derive a particular measure. (B) Visualization of the DPS-based symptom matrices as networks. The symptom variables (n = 98; age-adjusted) are represented as nodes and connected by an edge if the correlation exceeds r = 0.1. In the electronic version, green (red) edges indicate a positive (negative) correlation. Line thickness corresponds to strength of correlation, i.e. thicker lines represent stronger correlations. In the electronic version, colour of nodes represents the different DPS-based diagnostic categories [DPS, Diagnostic Schedule for Children (DISC) Predictive Scales].

Mental health and substance use

(Figure 3B) were assessed using the Diagnostic Interview Schedule for Children (DISC) Predictive Scales (DPS);³⁰ it contains questions about 98 symptoms (18 diagnostic categories, such as generalized anxiety disorder) the adolescent may have experienced in the past year. The DPS contains also questions about problems associated with use of alcohol (four questions), marijuana (three questions) and other substances (eight questions). In addition, we administered another questionnaire (Groupe de Recherche sur l’Inadaptation Psychosociale, GRIP) focusing on anti-social behaviour (e.g. stealing, fighting), hyperactivity and inattention, and anxiety and depression, as well as smoking, drinking (including binge drinking) and the use of other illicit substances,³¹ as well as a questionnaire focusing on the key components of positive youth development.³²

Wave 1: partial assessment of parents

The ‘partial’ assessment protocol of parents took place over two sessions; telephone interview with the mother, and a home visit. A research nurse conducted both sessions. During the telephone interview, which was always conducted with the mother, information on her life habits during pregnancies and on the medical history of her children was acquired. During the home visit, mothers (99% of families) filled out a series of questionnaires about the family environment, and each parent answered questions about her/his mental health (anxiety and depression) and substance use; the latter included questions about cigarette smoking, alcohol use and drug experimentation throughout their life (including current habits); and the presence of antisocial behaviour (at present and during her/his adolescence; Table 3). Parents also provided a convenience blood sample for genetic analyses, and self-reported weight and height.

Table 3.

Wave 1 ‘partial’ assessment of parents (n = 962): phenotyping domains and tools

Domain	Tool	Phenotypes
Family environment	FamEnvi	Stressful life events, financial difficulties, SES (family income, parental education)
Mental health	GRIP adult	Symptom counts (depression, anxiety, antisocial behaviour)
Substance use	GRIP adult	Cigarette smoking, alcohol use, drug experimentation
Medical history	Medical questionnaire	Personal and family history of cancer, hypertension, diabetes, heart disease, lipid disease, psychiatric disorders
Genetic variation	Blood DNA	Illumina Human610-Quad BeadChip and HumanOmniExpress BeadChip; a total of 7 746 837 typed and imputed SNPs
Epigenetic variationa	Blood DNA	Infinium HumanMethylation450K BeadChip (> 485 000 CpGs)

FamEnvi, questionnaire on family environment developed by the SYS team; GRIP Adult, self-assessment of mental health and substance use, as adapted by J. R. Séguin at the Groupe de Recherche sur l’Inadaptation Psychosociale of the University of Montreal.

^aAssessed in a subset of 288 parents

Wave 2: complete assessment of parents

The ‘complete’ assessment of parents (Table 4) was similar to the ‘complete’ assessment of adolescents (Table 2). The following additions and adjustments have been made to the brain MRI protocol, cognitive evaluation and assessment of mental health and substance use.

Table 4.

Wave 2 ‘complete’ assessment of parents (n = 664): phenotyping domains and tools

Domain	Tool	Phenotypes
Brain	MRI	Global and regional volumes; cortical surface and thickness; white-matter hyperintensities; magnetization transfer ratio; diffusion tensor imaging (DTI); resting state functional MRI
Cognition	Cambridge Brain Sciences Platform	Executive functioning; attention; learning and memory; reasoning; spatial skills
Mental health	MINI International Neuropsychiatric Interview; Mental Health and Addiction Questionnaire; ASR; CES-D; Family History Screen	Depression, anxiety, attention-deficit hyperactive disorder, antisocial personality disorder; post-traumatic stress disorder; obsessive compulsive disorder; alcohol and substance dependence, bulimia, anorexia; family history of psychiatric disorders
Substance use and addiction	Mental Health and Addiction Questionnaire; YFAS; FNDS; AUDIT; SRE; ESPAD; IAT; OGS	Cigarette smoking, alcohol and drug use, gambling, internet addition, food addiction
Personality	NEO-FFI	Neuroticism, extroversion, openness, agreeableness, conscientiousness
Cardiovascular	Finometer	Beat-by-beat blood pressure, heart rate, stroke volume, total peripheral resistance at rest and in response to physical and mental challenges (52-min protocol)
Autonomic balance	Power spectral analysis	Low- and high-frequency powers of inter-beat interval and low-frequency power of blood pressure; sympathetic and parasympathetic tone
Body composition	Anthropometry, MRI, bioimpedance	Height, weight, circumferences, skinfolds; subcutaneous and visceral fat and muscle volumes; fat and muscle mass
Lung function	Spirometer	Forced vital capacity, forced expiratory volume
Glucose lipid metabolism	Blood	Lipid profile (TG, TC, HDL-C, LDL-C), glucose, insulin, free fatty acids, glycerol, C-reactive protein
Lipidomicsa	Blood, LC-ESI-MS	700 lipid species
Diet	24-h food recall	Energy and nutrient intake
Medical history	Medical Questionnaire	Personal and family history of: cancer, hypertension, diabetes, heart disease, lipid disease, psychiatric disorders, addiction; reproductive and sexual health; medications
Lifestyle	Life Experiences Questionnaire; PBI; Hand Preference	Family characteristics; education; socioeconomic status; physical activity; sexual activity; parental style; hand laterality
Sleep	PSQI; ESS	Sleep quality, latency, duration, efficiency and disturbances; daytime sleepiness

MRI, magnetic resonance imaging; MINI, International Neuropsychiatric Interview³⁴; Mental Health and Addiction Questionnaire adapted from the Ontario Health Study and the Wave-1 questionnaire developed by the SYS team; ASR, Adult Self Report³⁵; CES-D, Center for Epidemiology Studies Depression Scale³⁶; YFAS, Yale Food Addiction Scale⁸⁰; FNDS, Fagerström’s Nicotine Dependence Scale⁸¹; AUDIT, Alcohol Use Disorder Identification Test⁸²; SRE, Subjective Response to Ethanol⁸³; ESPAD, European School Survey Project on Alcohol and Other Drugs⁸⁴; IAT, Internet Addiction Test⁸⁵; SOGS, South Oaks Gambling Screen⁸⁶; NEO-FFI, NEO-Five Factor Inventory⁸⁷; Life Experiences Questionnaire adapted from the OHS www.ontariohealthstudy.ca and the Wave-1 questionnaire developed by the SYS team; PBI, Parental Bonding Instrument⁸⁸; Hand Preference adapted from⁸⁹; PSQI, Pittsburgh Sleep Quality Index⁹⁰; ESS, Epsworth Sleepiness Scale⁹¹; Medical Questionnaire adapted from the Ontario Health Study [www.ontariohealthstudy.ca]; TG, triglycerides; TC, total cholesterol; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; CRP, C-reactive protein; LC-ESI-MS, liquid-chromatography electrospray- ionization mass spectrometry.

^aIn progress.

MRI protocol

(Table 4) In addition to T1-weighted images and MTR, the MRI protocol includes diffusion-weighted imaging (DTI) and resting-state functional MRI.

Cognitive performance

was assessed using the Cambridge Brain Sciences platform [www.cambridgebrainscience.com],³³ a computer-based battery of 12 tests designed to assess executive functioning, reasoning, working memory and visual-spatial skills. Participants completed the cognitive battery under the supervision of study personnel. The 12 tasks include: colour-word remapping, spatial planning, self-ordered search, paired associates learning, digit span, spatial span, visuospatial working memory, interlocking polygons, feature match, odd one out, grammatical reasoning and spatial rotation (Table 4).

Mental health

was assessed with the MINI International Neuropsychiatric Interview.³⁴ Adult Self Report³⁵ and Center for Epidemiology Studies Depression Scale³⁶(CES-D). Cigarette smoking, alcohol and drug use, gambling, internet addiction and food addiction were assessed with a number of standard questionnaires (Table 4).

Blood sample-derived assessments

Genome-wide genotyping and epityping was carried out with DNA from peripheral blood cells using the adolescent and parent blood samples collected during Wave 1. Targeted lipidomics profiling was conducted with fasting sera collected from adolescents during Wave 1 and parents during Wave 2.

Genome-wide genotyping

was carried out with the Human610-Quad and HumanOmniExpress BeadChips (Illumina, San Diego, CA). Following genotype imputation, a total of 7 746 837 typed and imputed SNPs are available for analysis.

Genome-wide epityping

was performed using the Infinium HumanMethylation450K BeadChip (Illumina, San Diego, CA) in 144 adolescents and 288 parents.

Serum lipidomics

targeted lipidomics profiling is currently conducted using fasting serum samples in all adolescents and parents. Liquid chromatography, electrospray ionization mass spectrometry (LC-ESI-MS) are used to identify and quantify serum glycerophospholipid species within the 440-640 Da range.

What has been found?

The SYS is one the largest adolescent cohorts with MRI of the brain and abdominal fat, as well as detailed cardiovascular, cognitive and behavioural assessments. Since the first publication in 2008, the SYS has provided data for a total of 49 original papers (List of publications in Supplement, available at IJE online). The main findings relate to questions asking whether: (i) higher visceral adiposity impacts adversely on cardiometabolic health, brain structure and cognition already during adolescence; ii) early life modifiers of the brain-reward system are involved in obesogenic eating and illicit drug use; and (iii) early (pre-natal and early postnatal) and late (adolescent) environments shape the brain development to influence brain reserve in adulthood.

Does visceral adiposity impact adversely on cardiometabolic health, brain structure and cognition already during adolescence?

In general, the SYS data support this possibility, but they also indicate that some of the relationships between visceral fat and cardiometabolic/brain health may vary by sex, or may be limited to certain (genetically defined) types of obesity. One of the first SYS studies showed that accumulation of visceral fat is associated with higher rates of the metabolic syndrome already in adolescence.⁴¹ The study also showed that visceral fat is associated with dyslipidaemia and insulin resistance in both sexes, but with elevated BP in males only.^41,42 This male-specific visceral fat/BP relationship is in part explained by a functional variant of the androgen receptor gene (AR), which was associated with higher visceral fat and BP in males but not females.⁴³ We also showed that vasomotor sympathoactivation may be one of the mechanisms linking visceral fat to BP in males who carry the risk AR variant.⁴³

A genome-wide association study carried out in the SYS adolescents found two new (PAX5 and MRPS22) and several previously identified loci of obesity, and it showed that only some of these obesity loci are also associated with higher BP. For example, both FTO and MC4R (two best-established loci of obesity⁴⁴) were associated with higher body fat, but only FTO was also associated with higher BP.⁴⁵ These results indicate that, although obesity is a well-established risk factor of hypertension,⁴⁶ not every pathway enhancing body-fat accumulation results in BP elevation.⁴⁵

Mid-life obesity has been recognized as a major risk factor of all-cause dementia.⁴⁰ In the SYS, we showed that obesity was associated with lower executive functioning already during adolescence, and that this association was specific to fat stored viscerally rather than elsewhere in the body.⁴⁷ We also showed that visceral fat and fat deposited elsewhere in the body were independently associated with structural properties of the adolescent brain, including signal intensity in white matter, white matter/grey matter signal contrast, and magnetization transfer ratio in both white and grey matter.⁴⁸ These relationships may reflect adiposity-related variations in phospholipid composition of brain lipids.

Early life modifiers of the brain-reward system contribute obesogenic eating and illicit drug use

diets rich in fat are obesogenic.^49,50 Fat is highly palatable, and dietary preference for fat is a behaviour regulated in part by reward-related mechanisms that process the hedonic properties of food independently of the body’s energy status.^51,52 Such mechanisms overlap with those mediating the addictive properties of drugs of abuse.⁵¹ Our findings in the SYS suggest that these mechanisms may be at play with regard to the risk for obesity and drug experimentation in the adolescents who were prenatally exposed to maternal cigarette smoking. We showed that prenatal exposure to maternal cigarette smoking is associated with higher body adiposity,⁵³ dietary preference for fat⁵⁴ and drug experimentation,^55,56 as well as with structural variations in brain regions processing reward.^54,55,57 In a genome-wide association study, we also showed that dietary preference for fat (and body adiposity) is associated with genetic variation in the opioid receptor mu 1 gene (OPRM1).⁵⁸ Finally, we have demonstrated that prenatal smoke exposure is associated with modifications of DNA methylation that persist in the postnatal life of the exposed offspring into adolescence,⁵⁹ and that some of these modifications are present in OPRM1 and may ‘silence’ the protective (fat intake-lowering) allele of this gene.⁶⁰

Early (prenatal) and late (adolescent) environments shape the brain to influence brain reserve in adulthood

By design, the SYS focuses on two periods of brain development, namely the prenatal and early postnatal periods (exposure to maternal smoking and breastfeeding) and adolescence (sex hormones, stress, substance use).

The prenatal and early postnatal periods are characterized by a rapid growth: the human brain reaches ∼ 420 cm³ in volume (∼ 36% of adult values) at birth, 855 cm³ (72% of adult) by the end of the first year, and 983 cm³ (83% of adult) at the end of the second year.⁶¹ We have shown that prenatal exposure to maternal cigarette smoking is associated with lower cortical expansion in girls with a particular genetic variant (in KCTD8), possibly by influencing apoptosis of neuronal progenitors during embryonic development.⁶² We also observed more subtle relationships between prenatal smoke exposure and the size of the corpus callosum⁶³as well as the thickness of the cerebral cortex.⁵⁷ Finally, we have discovered that FTO, the best-established gene of obesity, may modulate the relationship between the overall brain size and adiposity; this might arise from the effect of FTO during early embryonic development of ectoderm and mesoderm.⁶⁴ Altogether, these findings point to the importance of early environment, genes and their interactions in shaping various structural properties of the developing human brain, thus contributing to the person’s brain reserve.

The second period of development studied in the SYS—namely adolescence—is characterized by modifications of the structural properties of the brain grey and white matter, often in a sex-specific manner. It is likely that some of such sex differences in the adolescent brain underlie those in psychopathology in general,⁶⁵ and in the emergence of many psychiatric disorders during this developmental period in particular.⁶⁶ For example, we know that schizophrenia begins earlier in men than women.⁶⁷ In our work, we have focused on two environments that show striking variations during adolescence: sex hormones and substance use. With regard to sex hormones, we have described large testosterone-related increases in the volume of white matter during male adolescence; in boys, these variations differed as a function of genetic variations in AR.⁶⁸ Based on the divergent trajectories in the white-matter volumes and MTR signals, we speculated that these testosterone-related changes reflect radial growth of axons, i.e. their diameter.^68–70 We have confirmed this hypothesis in experimental animals (rats),⁷¹ and provided additional evidence suggesting that testosterone influences axonal transport,⁷¹ a biological process essential for both the delivery of neurotransmitter-related organelles to the synapse and axonal growth.⁷² With regard to substance use, we have observed a negative relationship between the extent of drug experimentation and the thickness of the orbitofrontal cortex in adolescents with prenatal smoke exposure.⁵⁵ In a more recent investigation that combined data from over 1500 adolescents, including the SYS, we showed that cannabis use before age 16 is associated with a thinner cortex (across the entire cortical mantle) but only in male adolescents with high polygenic risk score for schizophrenia.⁷³ On a regional basis, the group differences between cannabis users and non-users varied as a function of regional variations in the expression of the cannabinoid receptor 1 gene (CNR1).⁷³ Overall, the above findings reinforce a well-accepted view that brain maturation continues beyond the early postnatal years. Thus, adolescence represents another developmental period shaping the brain reserve through a variety of inter-twined influences of sex hormones, cardiometabolic factors and initial substance use.

In addition to research driven primarily by the SYS (described above), the SYS contributed results to five multi-cohort studies investigating genetic underpinnings of brain structure (ENIGMA^74,75), smoking, depression and anxiety (CARTA⁷⁶), facial morphology⁷⁷ and body adiposity.⁷⁸.

What are the main strengths and weaknesses?

The SYS cohort has a number of strengths, including: (i) multi-generational design that makes it possible to examine two critical periods of life—adolescence when common chronic diseases of the brain and body emerge, and middle adulthood when these diseases become entrenched and comorbidities develop; (ii) extensive, multi-level assessment of cardiometabolic and brain health in the same individual; and (iii) genome-wide evaluation of genetic and epigenetic variations. At present, the main weaknesses are the lack of follow-up data and modest sample size.

Can I get hold of the data?

Data are available upon request addressed to Dr Zdenka Pausova [zdenka.pausova@sickkids.ca] and Dr Tomas Paus [tpaus@research.baycrest.org]. Further details about the protocol can be found at [http://www.saguenay-youth-study.org/].

Supplementary Data

Supplementary data are available at IJE online.

Funding

The Saguenay Youth Study has been funded by the Canadian Institutes of Health Research (T.P., Z.P.), Heart and Stroke Foundation of Canada (Z.P.) and the Canadian Foundation for Innovation (Z.P.).

Key Messages

• Visceral adiposity associates adversely with cardiometabolic health, brain structure and cognition already during adolescence.

• Early life modifiers of the brain-reward system relate to both obesogenic eating and illicit drug use.

• Early (prenatal and postnatal) and late (adolescent) environments shape the brain to influence brain reserve in adulthood.